Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Appendix A Contributed Manuscripts A1 ANIMAL MIGRATION AND INFECTIOUS DISEASE RISK1 Sonia Altizer,2 Rebecca Bartel,2 and Barbara A. Han2 Abstract Animal migrations are often spectacular, and migratory species har- bor zoonotic pathogens of importance to humans. Animal migrations are expected to enhance the global spread of pathogens and facilitate cross- species transmission. This does happen, but new research has also shown that migration allows hosts to escape from infected habitats, reduces disease levels when infected animals do not migrate successfully, and may lead to the evolution of less-virulent pathogens. Migratory demands can also reduce immune function, with consequences for host susceptibility and mortality. Studies of pathogen dynamics in migratory species and how these will re- spond to global change are urgently needed to predict future disease risks for wildlife and humans alike. 1ââOriginally printed as Altizer et al. 2011. Animal migration and infectious disease risk. Science 331(6015):296-302. Reprinted with permission from AAAS. 2ââOdum School of Ecology, University of Georgia, Athens, GA 30602, USA. 111
112 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Billions of animals from groups as diverse as mammals, birds, fish, and insects undertake regular long-distance movements each year to track seasonal changes in resources and habitats (Dingle, 1996). The most dramatic migrations, such as those by monarch butterflies (Figure A1-1), gray whales, and some shore- birds and dragonflies (Figure A1-2), span entire continents or hemispheres, can take several months to complete, and are accompanied by high energetic demands and extreme physiological changes. The ultimate cause of these seasonal migra- tions remains debated; most hypotheses focus on avoidance of food scarcity, seeking physiologically optimal climates, and avoiding predation during periods FIGURE A1-1â Monarch butterflies (Danaus plexippus), shown here at a wintering site in central Mexico, undertake one of the longest distance two-way migrations of any insect species worldwide. Monarchs are commonly infected by a debilitating protozoan parasite that can lower the flight ability of migrating butterflies.
APPENDIX A 113 FIGURE A1-2âRepresentative migratory species, including migration distances and routes, known parasites and pathogens, and major threats to species persistence. Infec- tious diseases have been examined in the context of migration for some, but not all, of these species. Supporting references and photo credits are provided in the supporting online material (SOM) text.
114 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS of reproduction [e.g., (McKinnon et al., 2010)]. Contemporary studies of migra- tion have uncovered mechanisms of animal navigation, energy budgets, resource use, and phenological responses to environmental change; migratory species have also been recognized for their potential to connect geographically distant habitats and transfer large amounts of biomass and nutrients between ecosystems [reviewed in (Bowlin et al., 2010)]. These studies illustrate the profound ecologi- cal and evolutionary consequences of migratory journeys for animal populations on a global scale. Owing to their long-distance movements and exposure to diverse habitats, migratory animals have far-reaching implications for the emergence and spread of infectious diseases. Importantly, most previous work on the role of host move- ment in infectious disease dynamics has focused on spatially localized or random dispersal. For example, dispersal events give rise to traveling waves of infection in pathogens such as raccoon rabies (Russell et al., 2005), influenza in humans (Viboud et al., 2006), and nuclear polyhedrosis viruses in insects (White et al., 2000). In the context of metapopulations, limited amounts of host movement could actually prevent host extinction in the face of a debilitating pathogen and allow host resistance genes to spread (Carlsson-Granner and Thrall, 2002; Gog et al., 2002). From a different perspective, case studies of species invasions dem- onstrate that one-time transfers of even a few individuals can transport pathogens long distances and introduce them to novel habitats (Daszak et al., 2000). Yet relatively few studies have examined how regular, directed mass movements that characterize seasonal migration affect the transmission and evolution of patho- gens within host populations and the response of migratory species to infection risks. In this article, we review the consequences of long-distance migrations for the ecological dynamics of hostâpathogen interactions and outline key challenges for future work. Ecological processes linked with migration can increase or de- crease the between-host transmission of pathogens, depending on host migratory behavior and pathogen traits (Figure A1-3). Moreover, new work shows that for some species, the energetic demands of migration compromise host immunity, possibly increasing susceptibility to infection and intensifying the impacts of dis- ease. Importantly, many migratory species are at risk of future declines because of habitat loss and exploitation, and animal migrations are shifting with ongoing anthropogenic change (Wilcove, 2008). Thus, understanding how human activi- ties that alter migratory patterns influence wildlifeâpathogen dynamics is urgently needed to help guide conservation and management of migratory species and mitigate future risks from infectious disease. What Goes Around Comes Around: Pathogen Exposure and Spatial Spread An oft-cited but poorly supported assumption is that long-distance move- ments of migrating animals can enhance the geographic spread of pathogens,
APPENDIX A 115 FIGURE A1-3â Points along a general annual migratory cycle where key processes can increase (red text) or decrease (blue text) pathogen exposure or transmission. Behavioral mechanisms such as migratory escape and migratory culling could reduce overall pathogen prevalence. As animals travel to distant geographic locations, the use of multiple habitat types including stopover sites, breeding areas, and wintering grounds can increase trans- mission as a result of host aggregations and exposure to multihost pathogens. This might be especially true for migratory staging areas where animals stop to rest and refuel. High energetic demands of spring and fall migration can also result in immunomodulation, pos- sibly leading to immune suppression and secondary infections. [Photo credits (clockwise): J. Goldstein, B. McCord, A. Friedlaender, and R. Hall]
116 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS including zoonotic pathogens important for human health such as Ebola virus in bats, avian influenza viruses in waterfowl and shorebirds, and Lyme disease and West Nile virus (WNV) in songbirds. For example, WNV initially spread in North America along a major corridor for migrating birds and rapidly expanded from its point of origin in New York City along the Atlantic seaboard from 1999 to 2000 (Rappole et al., 2000). Although experimental work concluded that pas- serine birds in migratory condition were competent hosts for WNV and could serve as effective dispersal agents (Owen et al., 2006), evidence to show that this expansion resulted from movements of migratory birds remains equivocal. For the zoonotic pathogen Ebola virus, a recent study points to the coincident timing of an annual influx of migratory fruit bats in the Democratic Republic of Congo and the start of human Ebola outbreaks in local villages during 2007 (Leroy et al., 2009). In central Kazakhstan, saiga antelopes (Saiga tatarica) become infected with gastrointestinal nematodes (Marshallagia) during the course of seasonal migration by grazing on pastures used by domesticated sheep earlier in the sea- son. As migration continues, saiga carry and transmit Marshallagia to northern sheep populations, leading to pulses of infection that coincide with annual saiga migrations (Morgan et al., 2007). The potential for serious disease risks for human and livestock health has raised alarm about the role of migratory species in moving infectious agents to distant locations. Yet surprisingly few examples of long-distance pathogen dispersal by migrating animals have been clearly documented in the published literature, and some studies indicate that migrants might be unfairly blamed for transporting pathogens. As a case in point, wild waterfowl (Anseriformes) and shorebirds (Charadriiformes) represent the major natural reservoirs for diverse strains of avian influenza virus (AIV) worldwide, including the highly pathogenic (HP) H5N1 subtype that can lethally infect humans (Olsen et al., 2006). Although many of these migratory birds can become infected by HP H5N1, recent work incorporating what is known about viral shedding period, host migration phe- nology, and the geographical distribution of viral subtypes suggests that most wild birds are unlikely to spread HP H5N1 long distances (e.g., between Asia and the Americas) as previously suspected [e.g., (Krauss et al., 2007; Takekawa et al., 2010)]. Central to the question of how far any host species can transport a pathogen are the concepts of pathogen virulence and host tolerance to infection. Specifically, virulence refers to the damage that parasites inflict on their hosts, and tolerance refers to the hostâs ability to withstand infection without suffering major fitness costs. Thus, hostâparasite species or genotype combinations associ- ated with very low virulence or high tolerance should be the most promising can- didates for long-distance movement of pathogen strains, a simple prediction that could be explored within migratory species or using cross-species comparisons. Beyond their potential role in pathogen spatial spread, a handful of studies suggest that migratory species themselves encounter a broader range of pathogens from diverse environments throughout their annual cycle compared with species
APPENDIX A 117 residing in the same area year-round (Figure A1-3). One field study showed that songbird species migrating from Europe became infected by strains of vector- borne blood parasites originating from tropical bird species at overwintering sites in Africa (Waldenström et al., 2002), in addition to the suite of parasite strains transmitted at their summer breeding grounds. The authors posited that winter exposure to parasites in tropical locations is a significant cost of migration, be- cause resident species wintering in northern latitudes encounter fewer parasite strains and do not experience year-round transmission. Similarly, the number of parasite species per host was positively related to distances flown by migratory waterfowl (Figuerola and Green, 2000), indicating that migrating animals could become exposed to parasites through encounters with different host species and habitat types. Although some animals undertake nonstop migrations, most migratory spe- cies use stopover points along the migration route to rest and feed. These stop- over points usually occur frequently along a journey, although some species like shorebirds fly thousands of kilometers between only a handful of staging areas (Dingle, 1996). Refueling locations are often shared by multiple species, and the high local densities and high species diversity can increase both within- and between-species transmission of pathogens. In one of the most striking examples of this phenomenon, shorebirds such as sanderlings (Calidris alba), ruddy turn- stones (Arenaria interpres; Figure A1-2), and red knots (Calidris canutus), which migrate annually between Arctic breeding grounds and South American wintering sites, congregate to feed in massive numbers in the Delaware Bay and the Bay of Fundy to rebuild fat reserves, leading to upwards of 1.5 million birds intermin- gling, at densities of over 200 birds per square meter (Krauss et al., 2010). This phenomenon creates an ecological hotspot at Delaware Bay, where the prevalence of AIV is 17 times greater than at any other surveillance site worldwide (Krauss et al., 2010). Leaving Parasites Behind: Migration as a Way of Lowering Infection Risk Although greater exposure to parasites and pathogens can pose a significant cost of long-distance migration, for some animal species, long-distance migration will reduce infection risk by at least two nonexclusive processes (Figure A1-3). First, if prolonged use of habitats allows parasites with environmental transmis- sion modes to accumulate (i.e., those parasites with infectious stages that can persist outside of hosts, such as many helminths, ectoparasites, and microbial pathogens with fecal-oral transmission), migration will allow animals to escape from contaminated habitats [i.e., âmigratory escapeâ (Loehle, 1995)]. Between intervals of habitat use, unfavorable conditions (such as harsh winters and a lack of hosts) could eliminate most parasites, resulting in hosts returning to these habitats after a long absence to encounter largely disease-free conditions (Loehle, 1995). Empirical support for migratory escape comes from a few well-studied
118 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS hostâparasite interactions, including research on reindeer (Rangifer tarandus), which showed that the abundance of warble flies (Hypoderma tarandi) was nega- tively correlated with the distance migrated to summer pastures from reindeer calving grounds (the main larval shedding area in early spring) (Folstad et al., 1991). This observation prompted researchers to suggest that the reindeersâ an- nual postcalving migration reduces warble fly transmission by allowing animals to leave behind areas where large numbers of larvae have been shed (and where adult flies will later emerge). It is worth noting that escape will be less successful from pathogens with long-lived infectious stages that persist between periods of host absence or pathogens that cause chronic or life-long infections. Long-distance migration can also lower pathogen prevalence by removing infected animals from the population [i.e., âmigratory cullingâ (Bradley and Altizer, 2005)]. In this scenario, diseased animals suffering from the negative consequences of infection are less likely to migrate long distances owing to the combined physiological demands of migration and infection. Work on the migra- tory fall armyworm moth (Spodoptera frugiperda) suggested that insects infected by an ectoparasitic nematode (Noctuidonema guyanense) had reduced migratory ability because few to no parasites were detected in moths recolonizing sites as they returned north (Simmons and Rogers, 1991). More recent work on Bewickâs swans (Cygnus columbianus bewickii) showed that infection by low-pathogenic avian influenza (LPAI) viruses delayed migration over a month and reduced the travel distances of infected birds compared with those of healthy individuals (van Gils et al., 2007). However, a study of AIV in white fronted geese did not find any difference in distances migrated between infected and uninfected birds (Kleijn et al., 2010), suggesting that, not surprisingly, some species are better able to tolerate infections during long journeys and raising the possibility that migra- tion could select for greater tolerance to infections in some hosts due to the high fitness costs of attempting migration with a debilitating pathogen. Whether the net effects of migration will increase or decrease prevalence depends in large part on the mode of parasite transmission and the level of host specificity, both of which will affect opportunities for cross-species transmission at staging and stopover sites. Parasites that decline in response to host migra- tion may include specialist pathogens, as well as those with transmission stages that can build up in the environment, pathogens transmitted by biting vectors or intermediate hosts, or for which transmission occurs mainly from adults to juveniles during the breeding season (e.g., Box A1-1). Conversely, migrating hosts could experience higher pressure from generalist parasites if opportunities for cross-species transmission are high at stopover areas or wintering grounds or from specialist pathogens if transmission increases with dense host aggre- gations that accompany mass migrations. Importantly, effects of migration on pathogen dynamics within host populations should translate to large differences in prevalence across host populations with different migratory strategies. Over the past few years, we have focused on monarch butterflies (Danaus plexippus)
APPENDIX A 119 as a model system to study the effects of migration on hostâpathogen interac- tions (Box A1-1) and found that both migratory culling and migratory escape can cause spatiotemporal variation in prevalence within populations and extreme differences in prevalence among populations with different migratory strategies. However, we are not aware of intraspecific comparisons of prevalence between migratory and nonmigratory populations for other animal species. Immune Defense Balanced Against the Demands of Migration In addition to ecological mechanisms affecting between-host transmission, the physiological stress and energetic demands of migration can alter the out- come of infection within individuals through interactions with the hostâs im- mune system (Figure A1-2). More generally, because several immune pathways in both vertebrates and invertebrates are known to be costly (Eraud et al., 2005; Schmid-Hempel, 2005), seasonal demands such as premigratory fattening or strenuous activity will likely lower the resource pool available for mounting an immune response (Weber and Stilianakis, 2007). In anticipation of migration, for example, some animals accrue up to 50% of their lean body mass in fat, increase muscle mass, and atrophy organs that are not essential during migration (Dingle, 1996). Thus, before migration, animals might adjust components of their immune response to a desired level (i.e., immunomodulation), or the energetic demands of migration could reduce the efficacy of some immune pathways (i.e., immunosuppression). To date, the effects of long-distance migration on immune defenses have been best studied in birds. In a rare study of immune changes in wild individu- als during migration, field observations of three species of thrushes showed that migrating birds had lower baseline measures for several components of innate immunity (including leukocyte and lymphocyte counts), and exhibited lower fat reserves and higher energetic stress, relative to individuals measured outside of the migratory season (Owen and Moore, 2006). Captive experiments with Swainsonâs thrushes (Catharus ustulatus; Figure A1-2) later demonstrated that cell-mediated immunity was suppressed with the onset of migratory restlessness (the agitated behavior of birds that would normally precede their migratory depar- ture) (Owen and Moore, 2008a), suggesting that predictable changes in immunity occur in preparation for long-distance flight. In this species, the energetic costs of migration can intensify seasonal immune changes: Migrating thrushes that arrived at stopover sites in poorest condition had the lowest counts of immune cells (Owen and Moore, 2008b). The extent of altered immunity before and during migration is likely to be both species and resource dependent and will further depend on the specific im- mune pathway measured. Red knots, for example, exhibited no change in either antibody production or cell-mediated immunity after long flights in a wind tunnel, a result that argues against migration-mediated immunosuppression (Hasselquist
120 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS BOX A1-1 Lessons from a Model System: Monarch Migration Drives Large-Scale Variation in Parasite Prevalence During the past 10 years, we studied monarch butterflies (Danaus plexippus) and a protozoan parasite (Ophryocystis elektroscirrha) (top-right images) for the effects of seasonal migration on hostâpathogen dynamics. Monarchs in eastern North America (A) migrate up to 2,500 km each fall from as far north as Canada to wintering sites in Central Mexico (Brower and Malcolm, 1991). Monarchs in western North America (B) migrate shorter distances to winter along the coast of California (Nagano et al., 1993). Monarchs also form nonmigratory populations et al., 2007). Another study of captive red knots revealed no declines in costly immune defenses during the annual periods of mass gain (Buehler et al., 2008); however, animals in this study had constant access to high-quality food, which might have negated energetic trade-offs between immune investment and weight gain. Interestingly, barn swallows (Hirundo rustica) in better physical condition showed greater measures of cellular immunity during migration, cleared ecto- parasites and blood parasites more effectively, and arrived earlier at breeding grounds than birds with poor energy reserves (Møller et al., 2004). These studies suggest that animals in robust condition or with access to resources might toler- ate long journeys without significant immunocompromise. Studies of migratory species to date also emphasize the need for a more detailed understanding of the
APPENDIX A 121 that breed year-round in southern Florida (C), Hawaiâi, the Caribbean Islands, and Central and South America (Ackery and Vane-Wright, 1984). Because monarchs are abundant and widespread and can be studied easily both in the wild and in captivity, field and experimental studies can explore effects of annual migra- tions on hostâpathogen ecology and evolution. A recent continent-scale analysis showed that parasite prevalence increased throughout the monarchsâ breeding season, with highest prevalence among adults associated with more intense habitat use and longer residency in eastern North America, consistent with the idea of migratory escape (bottom-right graph) (Bartel et al., 2010). Experiments showed that monarchs infected with O. elektroscirrha flew shorter distances and with reduced flight speeds, and field studies showed parasite prevalence de- creased as monarchs moved southward during their fall migrations (Bartel et al., 2010; Folstad et al., 1991), consistent with the idea of migratory culling. Parasite prevalence was also highest among butterflies sampled at the end of the breed- ing season than for those that reached their overwintering sites in Mexico (bot- tom right graph). Collectively, these processes have likely generated the striking differences in parasite prevalence reported among wild monarch populations with different migratory behaviors (bottom-left graph) (Altizer et al., 2000). Laboratory studies also showed that parasite isolates from the longest-distance migratory population in eastern North America (A) were less virulent than isolates from short-distance (B) and nonmigratory (C) populations (de Roode and Altizer, 2010; Altizer, 2001), suggesting that longer migration distances cull monarchs carrying virulent parasite genotypes. Work on this model system illustrates how multiple mechanisms can operate at different points along a migratory cycle and highlights the role that migration plays in keeping populations healthy. Monarch migrations are now considered an endangered phenomenon (Brower and Malcolm, 1991) due to deforestation of overwintering grounds, loss of critical breeding habitats, and climate-related shifts in migration phenology. If climate warming extends monarch breeding seasons into fall and winter months, migrations may eventually cease altogether. Evidence to date indicates that the loss of migration in response to mild winters and year-round resources could prolong exposure to parasites, elevate infection prevalence, and favor more virulent parasite genotypes. Images reproduced from (Bartel et al., 2010; Altizer et al., 2000). [Photos by S. Altizer] mechanisms linking nutrient intake and metabolic activity to innate and adaptive immune measures, a step that is essential to predicting how different immune pathways will respond to physiological changes that occur before and during long-distance migrations. Perhaps most importantly, immune changes that accompany long-distance migration could lead to a relapse of prior infections and more severe disease fol- lowing exposure to new pathogens, increasing the likelihood of migratory culling and lowering the probability of spatial spread. This possibility was investigated for Lyme disease in redwings (Turdus iliacus) (Gylfe et al., 2000). Consistent with results showing negative effects of migratory status on immunity, migratory restlessness alone was sufficient to reactivate latent Borrelia infections in captive
122 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS birds. Thus, the demands of migration could ultimately lead to more severe in- fections and greater removal of infected hosts. Together, these results point to a role for migration-mediated immune changes in the dynamics of other wildlife pathogens, including zoonotic agents such as WNV (Owen et al., 2006) and bat- transmitted corona and rabies viruses (Li et al., 2005; Messenger et al., 2002). Effects of Anthropogenic Change and Climate Changes to the ecology of migratory species in the past century (Figure A1-2) could have enormous impacts on pathogen spread in wildlife and livestock, as well as altering human exposure to zoonotic infections. As one example, habitat loss caused by urbanization or agricultural expansion can eliminate stopover sites and result in higher densities of animals that use fewer remaining sites along the migration route (Wilcove, 2008). Although the resulting impacts on infectious diseases remain speculative, dense aggregations of animals at dwindling stopover sites might create ecological hot spots for pathogen transmission among wildlife species, as illustrated in the case of AIV in migrating shorebirds at Delaware Bay (Krauss et al., 2007). Moreover, continuing human encroachment on stopover habitats increases the likelihood of contact and spillover infection from wildlife reservoir hosts to humans and domesticated species. For some animal species, physical barriers such as fences (terrestrial species) or hydroelectric dams (aquatic species) impede migration (Berger et al., 2008), leaving animals to choose between navigating a narrow migratory corridor or forming nonmigratory populations. Consequently, pathogen prevalence could increase when animals stop migrating and become confined to smaller habitats, if parasite infectious stages build up with more intense use of a given habitat. Attempts to control cattle exposure to brucellosis from bison (Bison bison) and elk (Cervus elaphus) in the Greater Yellowstone Ecosystem illustrate these risks. Due to the potential threat of Brucella transmission from bison to cattle, bison are routinely culled if they leave the confines of Yellowstone National Park (Bienen and Tabor, 2006). Elk migration is less restricted, but there is evidence that supplemental feeding areas encourage the formation of dense nonmigratory populations that support higher prevalence of brucellosis, with 10 to 30% sero- prevalence in animals at the feeding grounds compared with 2 to 3% seropreva- lence in unfed elk ranging the park (Cross et al., 2010). High population densities in elk also correlate with higher gastrointestinal parasite loads at feeding grounds (Hines et al., 2007), suggesting that high densities of nonmigrating hosts lead to increasing intraspecific transmission of multiple parasites. More generally, human activities that discourage long-distance animal move- ments and encourage the formation of local year-round populations can cause the emergence of zoonotic pathogens in humans. For example, human-meditated environmental changes facilitated outbreaks of two zoonotic paramyxoviruses
APPENDIX A 123 carried by flying foxes (Pteropus fruit bats; Figure A1-2): These animals are highly mobile and seasonally nomadic in response to local food availability (Daszak et al., 2006). Anthropogenic changes such as deforestation and agricul- tural production likely influenced the emergence of lethal Nipah and Hendra virus outbreaks in humans in Australia and Malaysia in two key ways: by resource supplementation and habitat alteration limiting migratory behaviors of fruit bats and by facilitating close contact with domesticated virus-amplifying hosts (pigs and horses). In Malaysia, resident flying foxes foraging on fruit trees on or near pig farms transmitted Nipah virus to pigs, probably via urine or partially con- sumed fruit with subsequent spread from pigs to humans [(Daszak et al., 2006) and references therein]. Human activities are also thought to increase the risk of Hendra virus outbreaks in Australia by driving flying foxes from formerly forested areas into urban and suburban areas (Plowright et al., 2008), where they form dense nonmigratory colonies that aggregate in public gardens containing abundant food sources. In marine systems, aquaculture increases exposure to parasites in wild fish species, particularly in salmonids. Migration normally protects wild juvenile salmon from marine parasites in coastal waters by spatially separating them from infected wild adults offshore (KrkoÅ¡ek et al., 2007), but densely populated salmon farms place farmed fish enclosures adjacent to wild salmon migratory corridors, increasing the transmission of parasitic sea lice (Lepeophtheirus salmonis) to wild juveniles returning to sea (KrkoÅ¡ek et al., 2007). Finally, climate change will alter infectious disease dynamics in some migra- tory species (Harvell et al., 2009). To survive, many migratory species must re- spond to climate changes by shifting migratory routes and phenology in response to temperature and the availability of key resources (i.e., flowering plants, insects) [e.g., (Saino et al., 2010)]. It is possible that changes in the timing of migration could disrupt the synchronicity of host and parasite life cycles, much in the way that ecological mismatch in migration timing or altered migratory routes could impact whether suitable food and habitat are available when migrants arrive. For example, the spawning periodicity of whale barnacles in calving lagoons of gray whales is a classic example of a parasite synchronizing its reproduction to overlap with a hostâs migratory cycle (Rice and Wolman, 1971). If the tim- ing of whale migrations and barnacle reproduction shift in response to different environmental cues, this could result in reduced infections over time. On the other hand, altered migration routes might facilitate contact between otherwise geographically separated host species, leading to novel pathogen introductions and increasing disease risks for some wildlife species (Harvell et al., 2009). One example of this phenomenon involves outbreaks of phocine distemper virus in harbor seals (Phoca vitulina) in the North Sea, which was likely introduced by harp seals (Pagophilus groenlandicus) migrating beyond their normal range and contacting harbor seal populations (Jensen et al., 2002). Moreover, if climate
124 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS warming extends hostsâ breeding seasons, migrations may cease altogether, with year-round resident populations replacing migratory ones (Box A1-1), leading to greater pathogen prevalence through a loss of migratory culling and escape. Outlook and Future Challenges Understanding the mechanisms by which long-distance movements affect hostâpathogen systems offers exciting challenges for future work, especially in the context of global change and evolutionary dynamics. In terms of basic re- search, there remains a great need to identify conditions under which migration will increase host exposure to infectious agents versus cases where migration can reduce transmission, with the ultimate goal of predicting the net outcomes for host species where multiple mechanisms operate on the same or different pathogens (e.g., Box A1-1). To that end, mechanistic models are needed to examine how migration affects infectious disease dynamics and to explore the relevance of possible mechanisms. Such models must combine within-season processes (including host reproduction, overwintering survival, and pathogen transmission) with between-season migration (Figure A1-4). For example, to examine the importance of environmental transmission for the dynamics of LPAI in North American birds, Breban et al. (Breben et al., 2009) modeled a waterfowl population migrating between two geographically distant sites, with transmission dynamics occurring at both breeding and wintering grounds. Similarly, models describing interconnected networks of metapopulations could be useful in inves- tigating disease dynamics between habitats linked through seasonal migrations (Keeling et al., 2010). Although currently uncommon in the literature, epidemio- logical models can also be extended to capture mechanisms such as migratory culling and migratory escape and to include multiple infectious agents to explore questions of coinfection and multihost transmission dynamics (Figure A1-4). One outstanding question is whether parasites can increase the migratory propensity of their hosts by favoring the evolution of migratory behaviors. Long- distance migration has previously been hypothesized to reduce predation risks for ungulates and birds, with the general rationale being that the survival costs of mi- gration should be outweighed by fitness benefits associated with reproduction. In support of this idea, field studies of wolf predation on North American elk at their summer breeding grounds (Hebblewhite and Merrill, 2007) and nest predation on migrating songbirds (McKinnon et al., 2010) showed that animals traveling farthest experienced the lowest predation risk. Similar observational studies could ask how the prevalence, intensity, virulence, and diversity of key parasites change with migratory distances traveled. To that end, comparing infection dynamics between migratory and nonmigratory populations of the same species offers a powerful test of both pattern and process (e.g., Box A1-1), although research- ers will need to keep in mind that climate differences (e.g., milder climates for habitats used by nonmigratory populations) could confound some comparisons.
APPENDIX A 125 FIGURE A1-4âA compartmental model illustrating infectious disease dynamics (S-I model) in a migratory host population moving between geographically distinct breeding and overwintering habitats. Susceptible hosts (S) in the breeding grounds are born (v), die (μb) because of background mortality, and become infected at a rate, β. Infected hosts (I) suffer disease-induced mortality (αb). Different fractions of susceptible (xb) and in- fected hosts (yb) survive migration from the breeding habitat and arrive successfully at an overwintering habitat at some rate (δb). Here, natural (μw) and disease-induced mortality (αw) are both influenced by a different set of environmental conditions that characterize wintering grounds. The fraction of hosts surviving the spring migration the following year (xwδw, ywδw) will return to the breeding grounds to reproduce. A simple model like this can be readily modified to accommodate different parasite species and their transmission modes, host recovery, host age structure, and cross-species transmission. Modeling approaches are also needed to explore how seasonal migration might respond evolutionarily to parasite-driven pressures, similar to other studies that examined effects of within-site competition, costs of dispersal, and variation in habitat quality on random dispersal strategies (McPeek and Holt, 1992). Another question related to host evolution is whether the combined demands of migration and disease risk could select for greater or lower investment in resistance or immunity. Field and laboratory studies have already documented between-season changes in immune investment, suggesting that some migratory species suppress specific immune responses before or during migration (Owen
126 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS and Moore, 2006). The reduction in investment in immune defense could be an adaptive response to lower risks from certain parasites in migratory species (beyond issues related to energetic trade-offs) and might affect adaptive im- munity (shown to be costly for many vertebrate species) more strongly than in- nate defenses. Over longer time scales, long-distance migration could select for greater levels of innate immunity in migratory species or populations, especially if migrating animals encounter more diverse parasite assemblages (Møller and Erritzøe, 1998). With this in mind, comparisons of adaptive and innate immune defense and resistance to specific pathogens between migratory and nonmigra- tory populations represent a challenge for future work that could be especially tractable with invertebrate systems (Altizer, 2001). Pathogens might also respond to migration-mediated selection, with ecologi- cal pressures arising from migration leading to divergence in virulence. There is some evidence to show that less-virulent strains circulate in migratory popula- tions than in resident populations. The negative correlation between virulence and host migration distance, illustrated in the monarch system (Box A1-1), highlights the troubling possibility that pathogens infecting other migratory species could become more virulent if migrations decline. Moreover, dwindling migrations might affect host life history by altering pathogen virulence in once-migratory hosts. For example, a theoretical study showed that even moderate increases in virulence can change host breeding phenology to stimulate hosts to develop more quickly and breed earlier before they have a chance to become heavily infected (Restif et al., 2004). The recent facial tumor disease devastating Tasmanian devil populations provides a striking empirical example of high disease-induced mor- tality shifting host reproductive strategy from an iteroparous to a semelparous pattern through precocious sexual maturity in young devils (Jones et al., 2008). Although the hosts in this example are nonmigratory, they illustrate how virulent pathogens can generate longer-term fecundity costs beyond their direct impacts on host survival. Studying the migratory process in any wildlife species poses exceptional logistical challenges, in part because distances separating multiple habitats can sometimes span thousands of kilometers, making sampling for infection or im- munity intractable for field researchers. One problem is that historically, large numbers of animals have been sampled and marked at migratory staging areas, but for many species their subsequent whereabouts remain unknown (Webster et al., 2002). Tracking animals over long time periods and across vast distances has become more feasible with technological innovations such as radar and satel- lite telemetry for larger animals and ultra-light geolocators, stable isotopes, and radio tags to record or infer the movements of smaller animals (Robinson et al., 2010). Furthermore, physiological measurements such as heart rate, wing beat frequency, and blood metabolites can be obtained remotely for some species, enabling scientists to examine how infection status influences movement rates and the energetic costs of migration (Robinson et al., 2010).
APPENDIX A 127 Interdisciplinary studies to connect the fields of migration biology and in- fectious disease ecology are still in the early stages, and there are many excit- ing research opportunities to examine how infection dynamics relate to animal physiology, evolution, behavior, and environmental variation across the annual migratory cycle. Most evidence comes from studies of avian-pathogen systems, especially viruses. Although this is not surprising given the relevance of patho- gens such as avian influenza and WNV to human health, there remains a great need to explore other systems. Good places to start would be to make connections between disease and migration for species such as sea turtles, wildebeest, bats, dragonflies, and whales (Figure A1-2). Parasite infections and movement ecology in species in each of these groups have been well studied separately but not yet bridged. Taking a broad view of diverse host life histories and parasite trans- mission modes will allow future studies to identify ecological generalities and system-specific complexities that govern the mechanistic relationships between host movement behavior and infectious disease dynamics. Acknowledgments For helpful discussion and comments, we thank J. Antonovics, A. Davis, A. Dobson, V. Ezenwa, R. Hall, C. Lebarbenchon, A. Park, L. Ries, P. Rohani, P. R. Stephens, D. Streicker, and the Altizer/Ezenwa lab groups at the University of Georgia. This work was supported by an NSF grant (DEB-0643831) to S.A., a Ruth L. Kirschstein National Research Service Award through the NIH to R.B., an NSF Bioinformatics Postdoctoral Fellowship to B.A.H., and a National Center for Ecological Analysis and Synthesis working group on Migration Dynamics organized by S.A., L. Ries, and K. Oberhauser. References H. Dingle, Migration: The Biology of Life on the Move (Oxford Univ. Press, Oxford, 1996). L. McKinnon et al., Science 327, 326 (2010). M. S. Bowlin et al., Integr. Comp. Biol. 50, 261 (2010). C. A. Russell, D. L. Smith, J. E. Childs, L. A. Real, PLoS Biol. 3, e88 (2005). C. Viboud et al., Science 312, 447 (2006); 10.1126/science.1125237. A. White, A. D. Watt, R. S. Hails, S. E. Hartley, Oikos 89, 137 (2000). U. Carlsson-Graner, P. H. Thrall, Oikos 97, 97 (2002). J. Gog, R. Woodroffe, J. Swinton, Proc. R. Soc. London Ser. B 269, 671 (2002). P. Daszak, A. A. Cunningham, A. D. Hyatt, Science 287, 443 (2000). D. S. Wilcove, No Way Home: The Decline of the Worldâs Great Animal Migrations (Island Press, Washington, DC, 2008). J. H. Rappole, S. R. Derrickson, Z. Hubálek, Emerg. Infect. Dis. 6, 319 (2000). J. Owen et al., EcoHealth 3, 79 (2006). E. M. Leroy et al., Vector Borne Zoonotic Dis. 9, 723 (2009). E. R. Morgan, G. F. Medley, P. R. Torgerson, B. S. Shaikenov, E. J. Milner-Gulland, Ecol. Modell. 200, 511 (2007). B. Olsen et al., Science 312, 384 (2006).
128 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS S. Krauss et al., PLoS Pathog. 3, e167 (2007). J. Y. Takekawa et al., Avian Dis. 54 (suppl.), 466 (2010). J. Waldenström, S. Bensch, S. Kiboi, D. Hasselquist, U. Ottosson, Mol. Ecol. 11, 1545 (2002). J. Figuerola, A. J. Green, Evol. Ecol. 14, 143 (2000). S. Krauss et al., Proc. R. Soc. London Ser. B 277, 3373 (2010). C. Loehle, Ecology 76, 326 (1995). I. Folstad, A. C. Nilssen, O. Halvorsen, J. Andersen, Can. J. Zool. 69, 2423 (1991). C. Bradley, S. Altizer, Ecol. Lett. 8, 290 (2005). A. M. Simmons, C. E. Rogers, J. Entomol. Sci. 26, 136 (1991). J. A. van Gils et al., PLoS One 2, e184 (2007). D. Kleijn et al., Proc. R. Soc. London Ser. B 277, 2041 (2010). P. Schmid-Hempel, Annu. Rev. Entomol. 50, 529 (2005). C. Eraud, O. Duriez, O. Chastel, B. Faivre, Funct. Ecol. 19, 110 (2005). T. P. Weber, N. I. Stilianakis, Emerg. Infect. Dis. 13, 1139 (2007). J. C. Owen, F. R. Moore, Condor 108, 389 (2006). J. C. Owen, F. R. Moore, J. Ethol. 26, 383 (2008a). J. C. Owen, F. R. Moore, Can. J. Zool. 86, 638 (2008b). D. Hasselquist, A. Lindström, S. Jenni-Eiermann, A. Koolhaas, T. Piersma, J. Exp. Biol. 210, 1123 (2007). D. M. Buehler, T. Piersma, K. Matson, B. I. Tieleman, Am. Nat. 172, 783 (2008). A. P. Møller, F. de Lope, N. Saino, Ecology 85, 206 (2004). A. Gylfe, S. Bergström, J. Lundström, B. Olsen, Nature 403, 724 (2000). S. L. Messenger, J. S. Smith, C. E. Rupprecht, Clin. Infect. Dis. 35, 738 (2002). W. Li et al., Science 310, 676 (2005); 10.1126/science.1118391. J. Berger, J. K. Young, K. M. Berger, PLoS Biol. 6, e165 (2008). L. Bienen, G. Tabor, Front. Ecol. Environ 4, 319 (2006). P. C. Cross et al., Ecol. Appl. 20, 278 (2010). A. M. Hines, V. O. Ezenwa, P. Cross, J. D. Rogerson, Vet. Parasitol. 148, 350 (2007). P. Daszak et al., in Disease Ecology: Community Structure and Pathogen Dynamics, S. K. Collinge, C. Ray, Eds. (Oxford Univ. Press, Oxford, 2006), pp. 186â201. R. K. Plowright et al., Proc. R. Soc. London Ser. B 275, 861 (2008). D. Harvell, S. Altizer, I. M. Cattadori, L. Harrington, E. Weil, Ecology 90, 912 (2009). N. Saino et al., Proc. R. Soc. London Ser. B; published online 22 September 2010 (10.1098/ rspb.2010.1778). D. W. Rice, A. A. Wolman, The Life History and Ecology of the Gray Whale (Eschrichtius robustus) (American Society of Mammalogists Special Publication 3, Stillwater, OK, 1971). T. Jensen et al., Science 297, 209 (2002). R. Breban, J. M. Drake, D. E. Stallknecht, P. Rohani, PLoS Comput. Biol. 5, e1000346 (2009). M. J. Keeling, L. Danon, M. C. Vernon, T. A. House, Proc. Natl. Acad. Sci. U.S.A. 107, 8866 (2010). M. Hebblewhite, E. H. Merrill, Oecologia 152, 377 (2007). M. A. McPeek, R. D. Holt, Am. Nat. 140, 1010 (1992). A. P. Møller, J. Erritzøe, Evol. Ecol. 12, 945 (1998). S. M. Altizer, Evol. Ecol. Res. 3, 1 (2001). O. Restif, M. E. Hochberg, J. C. Koella, J. Evol. Biol. 14, 967 (2004). M. E. Jones et al., Proc. Natl. Acad. Sci. U.S.A. 105, 10023 (2008). M. S. Webster et al., Trends Ecol. Evol. 17, 76 (2002). W. D. Robinson et al., Front. Ecol. Environ 8, 354 (2010). L. P. Brower, S. B. Malcolm, Am. Zool. 31, 265 (1991). C. D. Nagano et al., in Biology and Conservation of the Monarch Butterfly, S. B. Malcolm, M. P. Zalucki, Eds. (Natural History Museum of Los Angeles County, Los Angeles, CA, 1993), pp. 217â232.
APPENDIX A 129 P. R. Ackery, R. I. Vane-Wright, Milkweed Butterflies: Their Cladistics and Biology (Cornell Univ. Press, Ithaca, NY, 1984). R. A. Bartel et al., Ecology; published online 19 July 2010 (10.1890/10-0489.1). S. M. Altizer, K. Oberhauser, L. P. Brower, Ecol. Entomol. 25, 125 (2000). J. C. de Roode, S. Altizer, Evolution 64, 502 (2010). A2 CLIMATE CHANGE AND INFECTIOUS DISEASES: FROM EVIDENCE TO A PREDICTIVE FRAMEWORK3 Sonia Altizer,4 Richard S. Ostfeld,5 Pieter T. J. Johnson,6 Susan Kutz,7 and C. Drew Harvell8 Abstract Scientists have long predicted large-scale responses of infectious diseases to climate change, giving rise to a polarizing debate, especially concerning human pathogens for which socioeconomic drivers and control measures can limit the detection of climate-mediated changes. Climate change has already increased the occurrence of diseases in some natural and agricul- tural systems, but in many cases, outcomes depend on the form of climate change and details of the hostâpathogen system. In this review, we highlight research progress and gaps that have emerged during the past decade and develop a predictive framework that integrates knowledge from ecophysiol- ogy and community ecology with modeling approaches. Future work must continue to anticipate and monitor pathogen biodiversity and disease trends in natural ecosystems and identify opportunities to mitigate the impacts of climate-driven disease emergence. 3ââOriginally printed as Altizer et al. 2013. Climate change and infectious diseases: From evidence to a predictive framework. Science 341(6145):514-519. Reprinted with permission from the AAAS. 4ââOdum School of Ecology, University of Georgia, Athens, GA 30602, USA. 5ââCary Institute of Ecosystem Studies, 2801 Sharon Turnpike, or Post Office Box AB, Millbrook, NY 12545-0129, USA. 6ââEcology and Evolutionary Biology, N122, CB334, University of Colorado, Boulder, CO 80309- 0334, USA. 7ââDepartment of Ecosystem and Public Health, Faculty of Veterinary Medicine, University of Calgary, and Canadian Cooperative Wildlife Health Centre, Alberta Node, 3280 Hospital Drive NW, Calgary, Alberta T2N 4Z6, Canada. 8ââEcology and Evolutionary Biology, E321 Corson Hall, Cornell University, Ithaca, NY 14853, USA.
130 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS The life cycles and transmission of many infectious agentsâincluding those causing disease in humans, agricultural systems, and free-living animals and plantsâare inextricably tied to climate (Garrett et al., 2013; Harvell et al., 2002). Over the past decade, climate warming has already caused profound and often complex changes in the prevalence or severity of some infectious diseases (Figure A2-1) (Baker-Austin et al., 2013; Burge et al., 2014; Garrett et al., 2013; FIGURE A2-1â Animalâparasite interactions for which field or experimental studies have linked climate change to altered disease risk. (A) Black-legged ticks, Ixodes scapularis, vectors of Lyme disease, attached to the ears of a white-footed mouse, Peromyscus leuco- pus, show greater synchrony in larval and nymphal feeding in response to milder climates, leading to more rapid Lyme transmission. (B) Caribbean coral (Diploria labyrinthiformis) was affected by loss of symbionts, white plague disease, and ciliate infection during the 2010 warm thermal anomaly in Curaçao. (C) Malformed leopard frog (Lithobates pipiens) as a result of infection by the cercarial stage (inset) of the multihost trematode R. ondatrae; warming causes nonlinear changes in both host and parasite that lead to marked shifts in the timing of interactions. (D) Infection of monarchs (D. plexippus) by the protozoan O. elektroscirrha (inset) increases in parts of the United States where monarchs breed year-round as a result of the establishment of exotic milkweed species and milder winter climates. (E) Infection risk with O. gruehneri (inset shows eggs and larvae) the common abomasal nematode of caribou and reindeer (R. tarandus), may be reduced during the hottest part of the Arctic summer as a result of warming, which leads to two annual trans- mission peaks rather than one (e.g., Figure A2-3C). Photo credits (A to E): J. Brunner, E. Weil, D. Herasimtschuk, S. Altizer, P. Davis, S. Kutz.
APPENDIX A 131 Harvell et al., 2009). For human diseases, vector-control, antimicrobial treat- ments, and infrastructural changes can dampen or mask climate effects. Wild- life and plant diseases are generally less influenced by these control measures, making the climate signal easier to detect (Harvell et al., 2009). For example, although the effects of climate warming on the dynamics of human malaria are debated, climate warming is consistently shown to increase the intensity and/or latitudinal and altitudinal range of avian malaria in wild birds (Garamszegi, 2011; Zamora-Vilchis et al., 2012). Predicting the consequences of climate change for infectious disease severity and distributions remains a persistent challenge surrounded by much controversy, particularly for vector-borne infections of humans [boxes S1 and S2 (available as supplementary materials on Science Online)]. Work using climate-based envelope models has predicted that modest climate-induced range expansions of human malaria in some areas will be offset by range contractions in other locations (Rogers and Randolph, 2000). An alternative approach, based on mechanistic models of physiological and demographic processes of vectors and pathogens (Ruiz-Moreno et al., 2012), predicts large geographic range expansions of hu- man malaria into higher latitudes (Martens et al., 1995). Both approaches have their limitations (Garrett et al., 2013), and the challenge remains to accurately capture the contributions of multiple, interacting, and often nonlinear underly- ing responses of host, pathogen, and vector to climate. This challenge is further exacerbated by variation in the climate responses among hostâpathogen systems arising from different life history characteristics and thermal niches (Molnár et al., 2013). A decade ago, Harvell et al. (2002) reviewed the potential for infectious diseases to increase with climate warming. Since then, the frequency of studies examining climateâdisease interactions has continued to increase (Figure A2-2), producing clear evidence that changes in mean temperature or climate vari- ability can alter disease risk. Some of the best examples of climate responses of infectious diseases to date are from ectothermic hosts and from parasites with environmental transmission stages that can persist outside the host (Figure A2-1). Indeed, first principles suggest that the rates of replication, development, and transmission of these pathogens should depend more strongly on temperature relative to other hostâpathogen interactions. The next challenges require integrat- ing theoretical, observational, and experimental approaches to better predict the direction and magnitude of changes in disease risk. Identifying the contribution of other environmental variables, such as precipitation, humidity, and climate variability remains a challenge (Paijmans et al., 2009; Raffel et al., 2013). Here, we review the individual, community, and landscape-level mechanisms behind climate-induced changes in infectious disease risk and illustrate how a quantitative, ecophysiological framework can predict the response of different hostâpathogen relations to climate warming. We mainly focus on changes in temperature, which have been more thoroughly explored both empirically and
132 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A2-2â Rising interest in climateâdisease interactions. Research focused on asso- ciations between infectious disease and climate change has increased steadily over the past 20 years. After correcting for total research interest in climate change (open symbols) or infectious disease (closed symbols), the frequency of papers referencing a climateâdisease link in the title has nearly doubled over this period, based on long-term publication trends following a Web of Science search of article titles (1990 to 2012). Search criteria and statistical analyses are provided in the supplementary materials, and the total number of climate changeâinfectious disease papers identified by our search criteria ranged from 5 to 117 publications per year. theoretically, relative to other environmental variables. We consider impacts of climate change on human diseases and on pathogens affecting species of con- servation or economic concern, including agroecosystems [box S3 (available as supplementary materials on Science Online)]. A crucial need remains for long- term ecological studies that examine the consequences of climate-disease inter- actions for entire communities and ecosystems, as well as for efforts that couple effective disease forecasting models with mitigation and solutions. Ecophysiology of HostâPathogen Interactions More than a century of research has firmly established that temperature and other climatic variables strongly affect the physiology and demography of free- living and parasitic species [e.g., (Walther et al., 2002)], with effects on behavior, development, fecundity, and mortality (Parmesan and Yohe, 2003). Because these
APPENDIX A 133 effects can be nonlinear and sometimes conflicting, such as warmer temperatures accelerating invertebrate development but reducing life span, a central challenge has been to identify the net outcomes for fitness (Harvell et al., 2002). For infec- tious diseases, this challenge is compounded by the interactions between at least two speciesâa host and a pathogenâand often vectors or intermediate hosts, which make the cumulative influence of climate on disease outcomes elusive [e.g., (Lafferty, 2009; Rohr et al., 2011)]. Immune defenses are physiological processes crucial for predicting changes in disease dynamics. Warmer temperatures can increase immune enzyme ac- tivity and bacterial resistance for insects, such as the cricket Gryllus texensis (Adamo and Lovett, 2011). Positive effects of temperature on parasite growth and replication, however, might outweigh greater immune function of the host. In gorgonian corals, for example, warmer temperatures increase cellular and hu- moral defenses (Mydlarz et al., 2006), but because coral pathogens also replicate faster under these conditions, disease outbreaks have coincided with warmer sea temperatures in the Caribbean (Figure A2-1) (Burge et al., 2014; Harvell et al., 2009). Warm temperatures also can lower host immunity; for example, melanization and phagocytic cell activity in mosquitoes are depressed at higher temperatures (Murdock et al., 2012). In addition, increased climate variability can interfere with host immunity, as illustrated by decreased frog resistance to the chytrid fungus Batrachochytrium dendrobatidis (Bd) in response to temperature fluctuations (Raffel et al., 2013). Even though Bd grows best in culture at cooler temperatures, which suggests that warming should reduce disease, incorporating variability-induced changes in host resistance suggests a more complex relation between climate change and Bd-induced amphibian declines (Rohr and Raffel, 2010). These issues are important for predicting the immunological efficiency of ectotherms outside of their typical climate envelope. One promising approach for predicting how hostâpathogen interactions re- spond to climate warming involves infusing epidemiological models with re- lations derived from the metabolic theory of ecology (MTE). This approach circumvents the need for detailed species-specific development and survival parameters by using established relations between metabolism, ambient tempera- ture, and body size to predict responses to climate warming (Brown et al., 2004). One breakthrough study (Molnár et al., 2013) used MTE coupled with traditional hostâparasite transmission models to examine how changes in seasonal and annual temperature affected the basic reproduction number (R0) of strongylid nematodes with direct life cycles and transmission stages that are shed into the environment. By casting R0 in terms of temperature-induced trade-offs between parasite development and mortality, this approach facilitated both general predic- tions about how infection patterns change with warming and, when parameterized for Ostertagia gruehneri, a nematode of caribou and reindeer (Figure A2-1), specific projections that corresponded with the observed temperature dependence of parasite stages. Moreover, this model predicted a shift from one to two peaks
134 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS in nematode transmission each year under warming conditions (Figure A2-3C), a result consistent with field observations (Hoar et al., 2012; Molnár et al., 2013). In some cases, ecophysiological approaches must consider multiple host species and parasite developmental stages that could show differential sensitivity to warming. Such differential responses can complicate prediction of net effects, especially for ectothermic hosts with more pronounced responses to tempera- ture. For instance, because both infectivity of a trematode parasite (Ribeiroia ondatrae) and defenses of an amphibian host (Pseudacris regilla) increase with FIGURE A2-3âTheoretical underpinnings and categorization of disease responses to climate change.
APPENDIX A 135 temperature; maximal pathology (limb malformations) (Figure A2-1) occurs at intermediate temperatures (Paull et al., 2012). Other work showed that the virulence of both a coral fungus (Aspergillus sydowii) and protozoan (Aplanochy- trium sp.) increased with temperature, probably because pathogen development rate continued to increase in a temperature range where coral defenses were less potent (Burge et al., 2013). Thus, the ideal approach will be an iterative one that combines metabolic and epidemiological modeling to predict general responses and to identify knowledge gaps, followed by application of models to specific hostâpathogen interactions. Pathogen responses to climate change depend on thermal tolerance relative to current and projected conditions across an annual cycle. (A) Gaussian curves relating temperature to a metric of disease risk suggest symmetrical temperature zones over which warming will increase and decrease transmission, whereas left-skewing [a common response for many terrestrial ectotherms, including arthropod vectors (Deutsch et al., 2008)] indicates greater potential for pathogen transmission to increase with warming [box S2 (available as supplementary materials on Science Online)]. Bold arrows represent geographic gradients that span cool, warm, and hot mean temperatures, which indicate that the net effect of warming (at point of arrows) depends on whether temperatures grow to ex- ceed the optimum temperature (Topt) for disease transmission. Projected changes in disease will further depend on the starting temperature relative to Topt, the magnitude of warming, measurement error, adaptation, and acclimation. (B) Pathogens at their northern or altitudinal limits might show range expansion and nonlinear shifts in their life cycle in response to warmer temperatures (red) rela- tive to baseline (blue). For example, a shift from 2- to 1-year cycles of transmis- sion has occurred for the muskox lungworm (Kutz et al., 2009). This outcome could generate sporadic disease emergence in a naïve population (if extremes in temperature allow only occasional invasion and/or establishment), or could gradually increase prevalence and establishment. (C) At the low-latitude or low- altitude extent of a pathogenâs range, where temperature increases could exceed the pathogenâs thermal optimum, transmission might be reduced, or we might see the emergence of a bimodal pattern whereby R0 peaks both early and late in the season, but decreases during the midsummer [as in the case of the arctic O. gruehneriâreindeer example (Molnár et al., 2013)]. In (B) and (C), the lower blue line represents R0 = 1, above which the pathogen can increase; values above the pink line represent severe disease problems owing to a higher peak of R0 and a greater duration of time during which R0 > 1. Community Ecology, Biodiversity, and Climate Change Hostâpathogen interactions are embedded in diverse communities, with cli- mate change likely leading to the loss of some hostâpathogen interactions and the gain of novel species pairings. In some cases, pathogen extinction and the loss of
136 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS endemic parasites could follow from climate change, potentially reducing disease or conversely releasing more pathogenic organisms from competition. In other cases, multiple pathogens can put entire host communities at risk of extinction. Although ecosystems of low biodiversity, such as occur in polar regions, can be particularly sensitive to emerging parasitic diseases (Kutz et al., 2009), most knowledge of community-wide responses stems from tropical marine systems. For example, the wider Caribbean region is a âdisease hot spotâ characterized by the rapid, warming-induced emergence of multiple new pathogens that have caused precipitous coral declines with ecosystem-wide repercussions (Rogers and Muller, 2012; Ruiz-Moreno et al., 2012). Impacts of climate-induced changes in disease can be especially large when the host is a dominant or keystone species. For example, near extinction of the once-dominant, herbivorous abalone (genus Haliotis) by warming-driven rickettsial disease caused pervasive community shifts across multiple trophic levels (Burge et al., 2014). Similarly, seagrass (Zostera marina) declines caused by infection with the protist Labyrinthula zos- terae, which correlates positively with warming, have degraded nursery habitats for fish and migratory waterfowl and caused the extinction of the eelgrass limpet (Hughes et al., 2002). Microbial communities, which are often part of the extended phenotype of host defenses, are also likely to respond to climate changes. For instance, warm- ing sea-surface temperatures in coral reefs can inhibit the growth of antibiotic- producing bacteria, sometimes causing microbial communities to shift from mutualistic to pathogenic (Ritchie, 2006). In agroecosystems, higher tempera- tures can suppress entomopathogenic fungi and antibiotic production by bacte- rial mutualists in plants (Humair et al., 2009). Warming also underlies bacterial shifts from endosymbiotic to lytic within host amoebas that live in human nasal passages, increasing the potential risk of respiratory disease (Corsaro and Greub, 2006). Thus, effects of warmer temperatures on the diversity and function of commensal or mutualist microbes could promote pathogen growth and pest outbreaks. From a broader perspective, biodiversity loss is a well-established con- sequence of climate change (Jetz et al., 2007; Parmesan and Yohe, 2003) and can have its own impact on infectious diseases. For many diseases of humans, wildlife, and plants, biodiversity loss at local or regional scales can increase rates of pathogen transmission (Cardinale et al., 2012; Johnson and Hoverman, 2012; Keesing et al., 2010). This pattern can result from several mechanisms, includ- ing the loss of the dilution effect (Johnson and Hoverman, 2012). For example, lower parasite diversity could allow more pathogenic species to proliferate when endemic and competing parasites are lost from a system (Johnson and Hoverman, 2012). Climate warming can also weaken biotic regulation of disease vectors by inhibiting their predators (Hobbelen et al., 2013) and competitors (Farjana et al., 2012). Interactions between biodiversity and infectious disease underscore the need to put climateâdisease interactions into the broader context of other forms
APPENDIX A 137 of global change, such as land-use change and habitat loss, when extending pre- dictions from focused hostâpathogen interactions to larger spatial and taxonomic scales. Shifts in Behavior, Movement, and Phenology of Hosts and Parasites Changes in climate are already affecting the phenology of interactions be- tween plants and pollinators, predators and prey, and plants and herbivores (Parmesan and Yohe, 2003). Climate-induced shifts in phenology and species movements (Chen et al., 2011) will likely affect disease dynamics. Many species are already moving toward higher elevations or latitudes (Hickling et al., 2006), and an open question is whether these shifts could disrupt established interactions or bring novel groups of hosts and pathogens into contact (Morgan et al., 2004). For instance, the range expansion of the Asian tiger mosquito (Aedes albopictus) across Europe and the Americas has created the potential for novel viral diseases such as Chikungunya to invade (Ruiz-Moreno et al., 2012); this pathogen is already expanding in geographic range, and a recent outbreak in Europe empha- sizes the need for surveillance and preparedness. Along eastern North America, warming sea temperatures and changes in host resistance facilitated a northward shift of two oyster diseases into previously unexposed populations (Burge et al., 2014). Migratory species in particular can be sensitive to climate change (Hickling et al., 2006), with the routes and timing of some speciesâ migrations already shift- ing with climate warming (Parmesan and Yohe, 2003). Long-distance migrations can lower parasite transmission by allowing hosts to escape pathogens that accu- mulate in the environment or by strenuous journeys that cull sick animals (Altizer et al., 2011). In some cases, milder winters can allow previously migratory host populations to persist year-round in temperate regions (Bradshaw and Holzapfel, 2007); this residency fosters the build-up of environmental transmission stages, and mild winters further enhance parasite over-winter survival (Garrett et al., 2013). A case study of monarch butterflies (Danaus plexippus) and the protozoan parasite Ophyrocystis elektroscirrha (Figure A2-1) provides support for climate- warming shifts in migration and disease. Monarchs typically leave their northern breeding grounds in early fall and fly to Mexican wintering sites. Milder winters, combined with increased planting of exotic host plants, now allow monarch populations to breed year-round in parts of the United States (Howard et al., 2010). Relative to migratory monarchs, winter-breeding monarchs suffer from higher rates of infection (Altizer et al., 2011). Similarly, migration is considered an important parasite avoidance strategy for barren-ground caribou (Hoar et al., 2012), but the loss of sea ice with climate warming will likely inhibit migrations and prevent them from seasonally escaping parasites (Post et al., 2013). Thus, diminishing migration behaviors among animals that use seasonal habitats can increase the transmission of infectious diseases.
138 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Changes in the timing of vector life stages and feeding behavior can also arise from interactions between climate and photoperiod. For several tick-borne infections (Figure A2-1), pathogens are sequentially transmitted from infected vertebrate hosts to naïve larval tick vectors, and from infected nymphal ticks to naïve vertebrate hosts. Asynchrony in the seasonal activity of larval and nymphal ticks can delay transmission and select for less virulent strains of the Lyme bacterium Borrelia burgdorferi (Kurtenbach et al., 2006), whereas synchrony allows for more rapid transmission and the persistence of virulent strains. In the case of tick-borne encephalitis (TBE), viral transmission occurs directly between cofeeding ticks; thus, viral maintenance requires synchronous larval and nymphal feeding (Randolph et al., 1999). Because synchrony of larval and nymphal ticks characterizes milder winter climates, climate change could increase tick syn- chrony and the transmission and virulence of several tick-borne infections. Changes in the timing of shedding or development of environmental trans- mission stages could result from climate warming. Some parasites could experi- ence earlier hatching, exposure to hosts earlier in the season, and encounters with earlier (and often more sensitive) life stages of hosts. For example, a long-term data set of lake plankton showed that warming shifted fungal prevalence patterns in diatom hosts from acute epidemics to chronic persistence, in part because of faster transmission and more widespread host population suppression under warmer temperatures (Ibelings et al., 2011). In contrast, Brown and Rohani (Brown and Rohani, 2012) argued for the opposite outcome with respect to avian influenza (AI) in reservoir bird hosts. Climate-driven mismatch in the timing of bird migration and their prey resources (e.g., horseshoe crab eggs) amplified vari- ability in epidemiological outcomes: Although mismatch increased the likelihood of AI extinction, infection prevalence and spillover potential both increased in cases where the virus persisted. Plasticity in parasite traits could allow parasites with environmental trans- mission stages to respond more rapidly to climate warming. For example, arrested development (hypobiosis) of the nematode O. gruehneri within its caribou host is a plastic trait more commonly expressed in areas with harsher winters as com- pared with milder climates (Hoar et al., 2012). This arrested state prevents wasted reproductive effort for the parasites, because eggs produced in late summer in colder regions are unlikely to develop to infective-stage larvae by fall. Ultimately, plasticity in life history traits could speed parasite responses to changing envi- ronments and allow parasites to deal with climate instabilities (e.g., a series of severe winters interspersed by mild), relative to the case where selection must act on genetically variable traits (Moritz and Agudo, 2013). For example, if climate warming extends the transmission season for O. gruehneri on tundra, a rapid decrease in the frequency of nematode hypobiosis could shorten the life cycle and increase infection rates.
APPENDIX A 139 Consequences for Conservation and Human Health Climate change is already contributing to species extinctions, both directly and through interactions with infectious disease (Thomas et al., 2004). Roughly one-third of all coral species and the sustainability of coral reef ecosystems are threatened by human activities, including climate warming and infectious dis- eases (Burge et al., 2014). In contrast to tropical marine systems, the Arctic is a less diverse and minimally redundant system that is warming at least twice as fast as the global average (International Panel on Climate Change, 2007) and si- multaneously experiencing drastic landscape changes from an expanding human footprint. Altered transmission dynamics of parasites, poleward range expansion of hosts and parasites, and disease emergence coincident with climate warming or extremes have all been reported in the Arctic (Kutz et al., 2009; Laaksonen et al., 2010). Together, these phenomena are altering hostâparasite dynamics and causing endemic Arctic speciesâunable to compete or adapt rapidly enoughâto decline (Gilg et al., 2012). Changes in wildlife health can also compromise the livelihoods and health of indigenous people who depend on wildlife for food and cultural activities (Meakin and Kurvitz, 2009). In humans, exposure to diarrheal diseases has been linked to warmer tem- peratures and heavy rainfall (Pascual et al., 2002). Human infections of cholera, typically acquired through ingestion of contaminated water (in developing coun- tries) or undercooked seafood (in the developed world), affect millions of people annually with a high case-fatality rate. Coastal Vibrio infections are associated with zooplankton blooms, warmer water, and severe storms (Baker-Austin et al., 2013). For example, in the Baltic Sea, long-term warming and temperature anomalies have been linked to increased disease from Vibrio vulnificus, which was first reported in 1994 along the German coast after an unusually warm sum- mer (Baker-Austin et al., 2013). Long-term sea surface warming can increase the geographic range, concentration, and seasonal duration of Vibrio infections, as seen in coastal Chile, Israel, and the U.S. Pacific Northwest. Modeling ap- proaches indicate that Vibrio illnesses from the Baltic region could increase nearly twofold for every 1°C increase in annual maximum water temperature (Baker-Austin et al., 2013). Human mosquito-borne diseases, such as malaria and dengue fever, are fre- quently proposed as cases where vector and disease expansion into the temperate zone could follow from climate warming (Mills et al., 2010). However, some researchers have argued that ranges will shift with warming, rather than expand, and that the best predictors of infection risk are economic and social factors, especially poverty (Lafferty, 2009; Randolph, 2010). Controversy has also arisen over which climatic variables are most important in delimiting the distributions of these diseases [boxes S1 and S2 (available as supplementary material on Science Online)]. Detecting impacts of climate change on human vector-borne diseases remains difficult, in part, because active mitigations, such as vector-control, an- timicrobials, and improved infrastructure, can complicate detection of a climate
140 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS signal. Several unresolved issues include identifying conditions under which climate warming will cause range expansions versus contractions, understanding the impact of increasing variability in precipitation, and determining the addi- tional economic costs associated with increased disease risk caused by warming. Ultimately, the societal implications of climate-driven shifts in diseases of humans, crops, and natural systems will demand solutions and mitigation, including early-warning programs. Recently, a forecasting system linking global coupled ocean-atmosphere climate models to malaria risk in Botswana allowed anomalously high risk to be predicted and anticipatory mitigations to be initiated (Thomson et al., 2006). Forecasting is well established in crop disease manage- ment and leads to improved timing of pesticide application and deployment of planting strategies to lower disease risk [box S3 (available as supplementary material on Science Online)]. Modeling efforts to better predict crop loss events are also tied to improved insurance returns against losses (Garrett et al., 2013). Similarly, accurate forecasting programs for coral bleaching have become a main- stay of marine climate resilience programs (Eakin et al., 2010) and are leading to the development of coral disease forecasting algorithms (Maynard et al., 2011). Appropriate management actions under outbreak conditions include reef closures to reduce stress and transmission, culling of diseased parts of some colonies, and increased surveillance (Beeden et al., 2012). In the ocean, efforts are also under way to increase the resilience of marine ecosystems to disease, including devel- oping no-fishing zones and reducing land-based pollution that can introduce new pathogens (Burge et al., 2014). Outlook and Future Challenges Climate change will continue to limit the transmission of some pathogens and create opportunities for others. To improve predictions and responses we need to deepen our understanding of mechanistic factors. Although the initial climatic drivers to be explored should be temperature variables (both mean and variability), because the data are available and we understand the mechanisms at work, future work must concurrently explore the effects of precipitation, relative humidity, and extreme events. In particular, models are needed that combine the principles of ecophysiology and MTE (Brown et al., 2004) with epidemiologi- cal response variables, such as R0 or outbreak size, and that are designed to ac- commodate distinct pathogen types (e.g., vector-borne, directly transmitted, or complex life cycle) and host types (ectotherm versus endotherm) (Molnár et al., 2013). These models should be applied, by using climate-change projections, to evaluate how broad classes of pathogens might respond to climate change. Building from this foundation, the next step is to extend such general models to specific pathogens of concern for human health, food supply, or wildlife conser- vation, which will require empirical parameterization, with attention to the on- the-ground conditions. Modeling efforts should be integrated with experiments to
APPENDIX A 141 test model predictions under realistic conditions, and with retrospective studies to detect the âfingerprintâ of climate-induced changes in infection. Scientists still know relatively little about the conditions under which evolu- tion will shape host and pathogen responses to climate change. Although evo- lutionary change in response to climate warming has been reported for some free-living animals and plants, the evidence remains limited (Moritz and Agudo, 2013). Even less is known about how climate change will drive hostâpathogen evolution. Corals have multiple levels of adaptation to intense selection by ther- mal stress that could also affect resistance to pathogens, including symbiont shuf- fling of both algae and bacteria, and natural selection against thermally intolerant individuals (Howells et al., 2011). In oysters (Crassostrea virginica), warming might have contributed to increased resistance to the protozoan multinucleated sphere X (MSX) disease (Ford and Bushek, 2012), but climate variability might also slow the evolution of oyster resistance (Powell et al., 2012). In cases where increased rates of transmission follow from warming, selection could favor higher pathogen virulence, although examples are now unknown. A persistent challenge involves the ability to detect changes in disease risk with climate across different systems. In the oceans, for example, impacts of disease on sessile hosts like corals, abalones, and oysters are readily apparent, and for terrestrial systems, clear impacts are seen for plant diseases and some wildlife-helminth interactions. But for highly mobile species and many human diseases, detecting signals of climate change remains problematic. For these less tractable systems, long-term ecological studies that examine the past distribu- tions of pathogens, important hosts, and severity of diseases are indispensable. Permanent repositories of intact physical specimens, as well as their DNA, can provide records of diversity that will be critical resources as new methodologies become available (Fernandez-Triana et al., 2011; Hoberg, 2010). Moreover, new technologies can detect variability in physiological processes and gene expression and can improve climate projections from global circulation models. Sophisti- cated experimental designs conducted under appropriate ranges of environmental conditions and retrospective studies to identify past climatic effects on disease (Burge et al., 2014; Hoverman et al., 2013) will help advance predictive power. An additional key challenge is predicting the impacts of climateâdisease interactions for human societies and gauging how these compare with other components of climate change, such as the loss of arable land. By affecting food yields and nutrition, water quality and quantity, social disorder, population displacement, and conflict, past climate changes have long influenced the burden of infectious disease in many human societies (McMichael, 2012; Wheeler and von Braun, 2013). Predicting the regions where humans and natural systems are most vulnerable to pressures from infectious disease and how these pressures will translate to changes in global health and security constitute critical research pri- orities (Myers and Patz, 2009). Building a mechanistic understanding of climateâ disease interactions will allow public health interventions to be proactive and will
142 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS facilitate effective responses to new or expanding health threats. Surveillance programs capable of detecting pathogen or disease emergence are essential and, in many instances, predicting and detecting local-scale impacts might be more important than predicting global-scale changes. To this end, the value of engag- ing local communities in disease surveillance is increasingly recognized, with the goal of advancing science on climateâdisease linkages for practical solutions to protecting human and wildlife health. Acknowledgments This work was supported in part by an NSF grant (DEB-0643831) to S.A., a Fellowship from the David and Lucile Packard Foundation and NSF grant (IOS-1121529) to P.T.J.J., an NSF Research Coordination Network grant on the Ecology of Infectious Marine Diseases, NSF Ecology and Evolution of Infec- tious Diseases grant (OCE-1215977) to C.D.H., and by the Atkinson Center for a Sustainable Future at Cornell University. S.K. thanks the Natural Sciences and Engineering Council of Canada, the Nasivvik Centre for Inuit Health; the govern- ments of the Northwest Territories, Nunavut, and Yukon; and the government of Canada International Polar Year Program. References C. D. Harvell, C. E. Mitchell, J. R. Ward, S. Altizer, A. P. Dobson, R. S. Ostfeld, M. D. Samuel, Cli- mate warming and disease risks for terrestrial and marine biota. Science 296, 2158â2162 (2002). K. Garrett, A. D. M. Dobson, J. Kroschel, B. Natarajan, S. Orlandini, H. E. Z. Tonnang, C. Valdivia, The effects of climate variability and the color of weather time series on agricultural diseases and pests, and on decisions for their management. Agric. For. Meteorol. 170, 216â227 (2013). C. Baker-Austin et al., Emerging Vibrio risk at high latitudes in response to ocean warming. Nat. Clim. Change 3, 73 (2013). D. Harvell, S. Altizer, I. M. Cattadori, L. Harrington, E. Weil, Climate change and wildlife diseases: When does the host matter the most? Ecology 90, 912â920 (2009). 10.1890/08 C. Burge et al., Climate change influences on marine infectious diseases: implications for manage- ment and society. Annu. Rev. Mar. Sci. 6, (2014). 10.1146/annurev-marine-010213 L. Z. Garamszegi, Climate change increases the risk of malaria in birds. Glob. Change Biol. 17, 1751â1759 (2011). 10.1111/j.1365 I. Zamora-Vilchis, S. E. Williams, C. N. Johnson, Environmental temperature affects prevalence of blood parasites of birds on an elevation gradient: Implications for disease in a warming climate. PLoS ONE 7, e39208 (2012). D. J. Rogers, S. E. Randolph, The global spread of malaria in a future, warmer world. Science 289, 1763â1766 (2000). 10976072 D. Ruiz-Moreno, I. S. Vargas, K. E. Olson, L. C. Harrington, Modeling dynamic introduction of Chikungunya virus in the United States. PLoS Negl. Trop. Dis. 6, e1918 (2012). 10.1371/ journal.pntd.0001918 W. J. Martens, L. W. Niessen, J. Rotmans, T. H. Jetten, A. J. McMichael, Potential impact of global climate change on malaria risk. Environ. Health Perspect. 103, 458â464 (1995). P. K. Molnár, S. J. Kutz, B. M. Hoar, A. P. Dobson, Metabolic approaches to understanding climate change impacts on seasonal host-macroparasite dynamics. Ecol. Lett. 16, 9â21 (2013).
APPENDIX A 143 K. P. Paaijmans, A. F. Read, M. B. Thomas, Understanding the link between malaria risk and climate. Proc. Natl. Acad. Sci. U.S.A. 106, 13844â13849 (2009). Abstract/FREE Full Text T. R. Raffel, J. M. Romansic, N. T. Halstead, T. A. McMahon, M. D. Venesky, J. R. Rohr, Disease and thermal acclimation in a more variable and unpredictable climate. Nat. Clim. Change 3, 146â151 (2013). G. R. Walther, E. Post, P. Convey, A. Menzel, C. Parmesan, T. J. Beebee, J. M. Fromentin, O. Hoegh- Guldberg, F. Bairlein, Ecological responses to recent climate change. Nature 416, 389â395 (2002). C. Parmesan, G. Yohe, A globally coherent fingerprint of climate change impacts across natural systems. Nature 421, 37â42 (2003). K. D. Lafferty, Calling for an ecological approach to studying climate change and infectious diseases. Ecology 90, 932â933 (2009). 10.1890/08 J. R. Rohr, A. P. Dobson, P. T. Johnson, A. M. Kilpatrick, S. H. Paull, T. R. Raffel, D. Ruiz-Moreno, M. B. Thomas, Frontiers in climate change-disease research. Trends Ecol. Evol. 26, 270â277 (2011). S. A. Adamo, M. M. Lovett, Some like it hot: The effects of climate change on reproduction, immune function and disease resistance in the cricket Gryllus texensis. J. Exp. Biol. 214, 1997â2004 (2011). L. D. Mydlarz, L. E. Jones, C. D. Harvell, Innate immunity, environmental drivers, and disease ecol- ogy of marine and freshwater invertebrates. Annu. Rev. Ecol. Evol. Syst. 37, 251â288 (2006). C. C. Murdock, K. P. Paaijmans, A. S. Bell, J. G. King, J. F. Hillyer, A. F. Read, M. B. Thomas, Complex effects of temperature on mosquito immune function. Proc. Biol. Sci. 279, 3357â3366 (2012). J. R. Rohr, T. R. Raffel, Linking global climate and temperature variability to widespread amphibian declines putatively caused by disease. Proc. Natl. Acad. Sci. U.S.A. 107, 8269â8274 (2010). J. H. Brown, J. F. Gillooly, A. P. Allen, V. M. Savage, G. B. West, Toward a metabolic theory of ecology. Ecology 85, 1771â1789 (2004). 10.1890/03 B. M. Hoar, K. Ruckstuhl, S. Kutz, Development and availability of the free-living stages of Osterta- gia gruehneri, an abomasal parasite of barrenground caribou (Rangifer tarandus groenlandicus), on the Canadian tundra. Parasitology 139, 1093â1100 (2012). S. H. Paull, B. E. LaFonte, P. T. J. Johnson, Temperature-driven shifts in a host-parasite interaction drive nonlinear changes in disease risk. Glob. Change Biol. 18, 3558â3567 (2012). C. A. Burge, C. J. Kim, J. M. Lyles, C. D. Harvell, The ecology of marine opportunists. Microb. Ecol. 65, 869â879 (2013). S. J. Kutz, E. J. Jenkins, A. M. Veitch, J. Ducrocq, L. Polley, B. Elkin, S. Lair, The Arctic as a model for anticipating, preventing, and mitigating climate change impacts on host-parasite interactions. Vet. Parasitol. 163, 217â228 (2009). C. Rogers, E. Muller, Bleaching, disease and recovery in the threatened scleractinian coral Acropora palmata in St. John, US Virgin Islands: 2003â2010. Coral Reefs 31, 807â819 (2012). D. Ruiz-Moreno, B. L. Willis, A. C. Page, E. Weil, A. Cróquer, B. Vargas-Angel, A. G. Jordan-Garza, E. Jordán-Dahlgren, L. Raymundo, C. D. Harvell, Global coral disease prevalence associated with sea temperature anomalies and local factors. Dis. Aquat. Organ. 100, 249â261 (2012). J. E. Hughes, L. A. Deegan, J. C. Wyda, M. J. Weaver, A. Wright, The effects of eelgrass habitat loss on estuarine fish communities of southern New England. Estuaries 25, 235â249 (2002). K. B. Ritchie, Regulation of microbial populations by coral surface mucus and mucus-associated bacteria. Mar. Ecol. Prog. Ser. 322, 1â14 (2006). B. Humair, N. González, D. Mossialos, C. Reimmann, D. Haas, Temperature-responsive sensing regulates biocontrol factor expression in Pseudomonas fluorescens CHA0. ISME J. 3, 955â965 (2009). D. Corsaro, G. Greub, Pathogenic potential of novel Chlamydiae and diagnostic approaches to in- fections due to these obligate intracellular bacteria. Clin. Microbiol. Rev. 19, 283â297 (2006). 10.1128/CMR.19.2.283
144 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS W. Jetz, D. S. Wilcove, A. P. Dobson, Projected impacts of climate and land-use change on the global diversity of birds. PLoS Biol. 5, e157 (2007). B. J. Cardinale, J. E. Duffy, A. Gonzalez, D. U. Hooper, C. Perrings, P. Venail, A. Narwani, G. M. Mace, D. Tilman, D. A. Wardle, A. P. Kinzig, G. C. Daily, M. Loreau, J. B. Grace, A. Lar- igauderie, D. S. Srivastava, S. Naeem, Biodiversity loss and its impact on humanity. Nature 486, 59â67 (2012). P. T. J. Johnson, J. T. Hoverman, Parasite diversity and coinfection determine pathogen infection success and host fitness. Proc. Natl. Acad. Sci. U.S.A. 109, 9006â9011 (2012). F. Keesing, L. K. Belden, P. Daszak, A. Dobson, C. D. Harvell, R. D. Holt, P. Hudson, A. Jolles, K. E. Jones, C. E. Mitchell, S. S. Myers, T. Bogich, R. S. Ostfeld, Impacts of biodiversity on the emergence and transmission of infectious diseases. Nature 468, 647â652 (2010). P. H. Hobbelen, M. D. Samuel, D. Foote, L. Tango, D. A. LaPointe, Modeling the impacts of global warming on predation and biotic resistance: Mosquitoes, damselflies and avian malaria in Ha- waii. Theor. Ecol. 6, 31â44 (2013). T. Farjana, N. Tuno, Y. Higa, Effects of temperature and diet on development and interspecies competition in Aedes aegypti and Aedes albopictus. Med. Vet. Entomol. 26, 210â217 (2012). I.-C. Chen, J. K. Hill, R. Ohlemüller, D. B. Roy, C. D. Thomas, Rapid range shifts of species associ- ated with high levels of climate warming. Science 333, 1024â1026 (2011). R. Hickling, D. B. Roy, J. K. Hill, R. Fox, C. D. Thomas, The distributions of a wide range of taxo- nomic groups are expanding polewards. Glob. Change Biol. 12, 450â455 (2006). 10.1111/j.1365 E. R. Morgan, E. J. Milner-Gulland, P. R. Torgerson, G. F. Medley, Ruminating on complexity: Mac- roparasites of wildlife and livestock. Trends Ecol. Evol. 19, 181â188 (2004). S. Altizer, R. Bartel, B. A. Han, Animal migration and infectious disease risk. Science 331, 296â302 (2011). W. E. Bradshaw, C. M. Holzapfel, Evolution of animal photoperiodism. Annu. Rev. Ecol. Evol. Syst. 38, 1â25 (2007). E. Howard, H. Aschen, A. K. Davis, Citizen science observations of monarch butterfly overwintering in the southern United States. Psyche (Camb. Mass.) 2010, 1â6 (2010). E. Post, U. S. Bhatt, C. M. Bitz, J. F. Brodie, T. L. Fulton, M. Hebblewhite, J. Kerby, S. J. Kutz, Ian Stirling, D. A. Walker, Ecological consequences of sea ice decline. Science 341, 519â524 (2013). K. Kurtenbach, K. Hanincová, J. I. Tsao, G. Margos, D. Fish, N. H. Ogden, Fundamental processes in the evolutionary ecology of Lyme borreliosis. Nat. Rev. Microbiol. 4, 660â669 (2006). S. E. Randolph, D. Miklisová, J. Lysy, D. J. Rogers, M. Labuda, Incidence from coincidence: Patterns of tick infestations on rodents facilitate transmission of tick-borne encephalitis virus. Parasitol- ogy 118, 177â186 (1999). B. W. Ibelings, A. S. Gsell, W. M. Mooij, E.. Van Donk, S. Van Den Wyngaert, L. N. De Senerpont Domis, Chytrid infections and diatom spring blooms: Paradoxical effects of climate warming on fungal epidemics in lakes. Freshw. Biol. 56, 754â766 (2011). 10.1111/j.1365 V. L. Brown, P. Rohani, The consequences of climate change at an avian influenza âhotspotâ. Biol. Lett. 8, 1036â1039 (2012). B. M. Hoar, A. G. Eberhardt, S. J. Kutz, Obligate larval inhibition of Ostertagia gruehneri in Rangifer tarandus? Causes and consequences in an Arctic system. Parasitology 139, 1339â1345 (2012). C. Moritz, R. Agudo, The future of the species under climate change: Resilience or decline? Science 341, 504â508 (2013). C. D. Thomas, A. Cameron, R. E. Green, M. Bakkenes, L. J. Beaumont, Y. C. Collingham, B. F. Erasmus, M. F. De Siqueira, A. Grainger, L. Hannah, L. Hughes, B. Huntley, A. S. Van Jaars- veld, G. F. Midgley, L. Miles, M. A. Ortega-Huerta, A. Townsend Peterson, O. L. Phillips, S. E. Williams, Extinction risk from climate change. Nature 427, 145â148 (2004). Intergovernmental Panel on Climate Change, Summary for Policymakers. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (Cambridge Univ. Press, New York, 2007).
APPENDIX A 145 S. Laaksonen, J. Pusenius, J. Kumpula, A. Venäläinen, R. Kortet, A. Oksanen, E. Hoberg, Climate change promotes the emergence of serious disease outbreaks of filarioid nematodes. EcoHealth 7, 7â13 (2010). O. Gilg, K. M. Kovacs, J. Aars, J. Fort, G. Gauthier, D. Grémillet, R. A. Ims, H. Meltofte, J. Moreau, E. Post, N. M. Schmidt, G. Yannic, L. Bollache, Climate change and the ecology and evolution of Arctic vertebrates. Ann. N. Y. Acad. Sci. 1249, 166â190 (2012). 10.1111/j.1749 S. Meakin, T. Kurvits, Assessing the Impacts of Climate Change on Food Security in the Canadian Arctic (GRID-Arendal, Arendal, Norway, 2009). M. Pascual, M. J. Bouma, A. P. Dobson, Cholera and climate: Revisiting the quantitative evidence. Microbes Infect. 4, 237â245 (2002). 10.1016/S1286-4579(01)01533 J. N. Mills, K. L. Gage, A. S. Khan, Potential influence of climate change on vector-borne and zoo- notic diseases: A review and proposed research plan. Environ. Health Perspect. 118, 1507â1514 (2010). S. E. Randolph, To what extent has climate change contributed to the recent epidemiology of tick- borne diseases? Vet. Parasitol. 167, 92â94 (2010). M. C. Thomson, F. J. Doblas-Reyes, S. J. Mason, R. Hagedorn, S. J. Connor, T. Phindela, A. P. Morse, T. N. Palmer, Malaria early warnings based on seasonal climate forecasts from multi-model ensembles. Nature 439, 576â579 (2006). C. M. Eakin, J. A. Morgan, S. F. Heron, T. B. Smith, G. Liu, L. Alvarez-Filip, B. Baca, E. Bartels, C. Bastidas, C. Bouchon, M. Brandt, A. W. Bruckner, L. Bunkley-Williams, A. Cameron, B. D. Causey, M. Chiappone, T. R. Christensen, M. J. Crabbe, O. Day, E. de la Guardia, G. DÃaz-Pulido, D. DiResta, D. L. Gil-Agudelo, D. S. Gilliam, R. N. Ginsburg, S. Gore, H. M. Guzmán, J. C. Hendee, E. A. Hernández-Delgado, E. Husain, C. F. Jeffrey, R. J. Jones, E. Jordán-Dahlgren, L. S. Kaufman, D. I. Kline, P. A. Kramer, J. C. Lang, D. Lirman, J. Mallela, C. Manfrino, J. P. Maréchal, K. Marks, J. Mihaly, W. J. Miller, E. M. Mueller, E. M. Muller, C. A. Orozco Toro, H. A. Oxenford, D. Ponce-Taylor, N. Quinn, K. B. Ritchie, S. RodrÃguez, A. R. RamÃrez, S. Romano, J. F. Samhouri, J. A. Sánchez, G. P. Schmahl, B. V. Shank, W. J. Skirving, S. C. Steiner, E. Villamizar, S. M. Walsh, C. Walter, E. Weil, E. H. Williams, K. W. Roberson, Y. Yusuf, Caribbean corals in crisis: Record thermal stress, bleaching, and mortality in 2005. PLoS ONE 5, e13969 (2010). J. Maynard, K. R. N. Anthony, C. D. Harvell, M. A. Burgman, R. Beeden, H. Sweatman, S. F. Heron, J. B. Lamb, B. L. Willis, Predicting outbreaks of a climate-driven coral disease in the Great Barrier Reef. Coral Reefs 30, 485â495 (2011). 10.1007/s00338-010-0708 R. Beeden, J. A. Maynard, P. A. Marshall, S. F. Heron, B. L. Willis, A framework for responding to coral disease outbreaks that facilitates adaptive management. Environ. Manage. 49, 1â13 (2012). E. Howells, V. H. Beltran, N. W. Larsen, L. K. Bay, B. L. Willis, M. J. H. van Oppen, Coral thermal tolerance shaped by local adaptation of photosymbionts. Nat. Clim. Change 2, 116â120 (2011). S. E. Ford, D. Bushek, Development of resistance to an introduced marine pathogen by a native host. J. Mar. Res. 70, 205â223 (2012). E. N. Powell, J. M. Klinck, X. Guo, E. E. Hofmann, S. E. Ford, D. Bushek, Can oysters Crassostrea virginica develop resistance to dermo disease in the field: The impediment posed by climate cycles. J. Mar. Res. 70, 309 (2012). E. P. Hoberg, Invasive processes, mosaics and the structure of helminth parasite faunas. Rev. Sci. Tech. 29, 255â272 (2010). J. Fernandez-Triana, M. A. Smith, C. Boudreault, H. Goulet, P. D. Hebert, A. C. Smith, R. Roughley, A poorly known high-latitude parasitoid wasp community: Unexpected diversity and dramatic changes through time. PLoS ONE 6, e23719 (2011). 10.1371/journal.pone.0023719 J. T. Hoverman, S. H. Paull, P. T. J. Johnson, in Climate Vulnerability: Understanding and Addressing Threats to Essential Resources, P. Roger, Ed. (Academic Press, Oxford, 2013), vol. 4, pp. 61â70. A. J. McMichael, Insights from past millennia into climatic impacts on human health and survival. Proc. Natl. Acad. Sci. U.S.A. 109, 4730â4737 (2012).
146 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS T. Wheeler, J. von Braun, Climate change impacts on global food security. Science 341, 508â513 (2013). 10.1126/science. S. S. Myers, J. A. Patz, Emerging threats to human health from global environmental change. Annu. Rev. Environ. Resour. 34, 223â252 (2009). C. A. Deutsch, J. J. Tewksbury, R. B. Huey, K. S. Sheldon, C. K. Ghalambor, D. C. Haak, P. R. Martin, Impacts of climate warming on terrestrial ectotherms across latitude. Proc. Natl. Acad. Sci. U.S.A. 105, 6668â6672 (2008). A3 MIGRATION, CIVIL CONFLICT, MASS GATHERING EVENTS, AND DISEASE Chris Beyrer9,10 and James Wren Tracy9 Introduction Human agency can drive infectious disease establishment, adaptation, and spread, which can subsequently have profound impacts on the health of individu- als, communities, and populations. Civil conflicts and the complex humanitarian emergencies they generate are widespread, common, and may increase in con- text of current global environmental change (Hsiang et al., 2013). Conflict, civil disruption, and the implicit migration that comes with both can compromise our ability to understand, track, respond, and mitigate infectious disease threats mak- ing their impact on human health even more difficult to address. With increased migration and mobility of peoples, a concurrent increase in exposure to multiple infectious diseases can occur. Populations mixing from the movement of individuals, groups, and sometimes whole communities can allow for a greater mixing of infectious diseases and heightened vulnerability to those diseases. Work by our group in Eastern Burma documented much higher rates of childhood and adult malaria, water-borne diarrheal diseases, childhood malnutri- tion, and land mine injuries among displaced populations in civil conflict zones than among stable and nondisplaced communities (Richards et al., 2009). More than just increasing the exposure to infectious diseases, migration can allow for greater acquisition and transmission of the diseases. Studies performed in South Africa, Kenya, Guinea-Bissau, and Nepal show an increased odds of HIV acquisition and infection among migratory groups, including rural to urban migration or migration out of the country (Beyrer et al., 2006). Migratory peoples 9ââCenterfor Public Health and Human Rights, Johns Hopkins University. 10ââDepartment of Epidemiology, Johns Hopkins Bloomberg School of Public Health, 615 N. Wolfe Street, E 7152, Baltimore, MD, 21205.
APPENDIX A 147 seldom receive proper health care, particularly undocumented migrants who have left home countries. They can experience treatment delays and gaps, barriers to access and care, and lack many protective commodities such as bed nets, water filters, and condoms that would decrease further exposures. Given a lack of treatment and access to trained health care workers, morbid- ity and mortality can increase among migratory and mobile populations. Addi- tional limited access to essential medications can increase disease severity and the likelihood of onward transmissionâsustaining cholera outbreaks, for example, as has occurred in Zimbabwe and Haiti among displaced populations (Piarroux et al., 2011; Sollom et al., 2009). The increase in infectious disease exposure and transmission within migra- tory groups does not only affect those within the group, but it can also affect those with whom the group comes into contact (Beyrer and Lee, 2008). A study performed in China showed that cities with a higher number of immigrants per 1,000 people also had a greater incidence of STDs (Tucker et al., 2005). There- fore, in order to protect the health of the displaced peoples and those they come into contact with, the underlying rights of these mobile groups and their access to adequate care must be protected and preserved. Often, particularly in the context of civil conflict, this does not happen. The flawed response to Cyclone Nargis, an enormous cyclone, which struck Burma/Myanmar, helps illustrate these issuesâ and demonstrates how climate change and human agency can interact in complex and challenging pathwaysâextracting heavy tolls on vulnerable populations. The conflict in Côte dâIvoire illustrates another challengeâthe loss of health care workers in conflict and our subsequent diminished capacity to both understand and address the health impacts of conflicts on populations. Conflict and Complex Humanitarian Emergencies Humanitarian emergencies can arise from many possible causes, but one of the more common causes, conflict, creates very complex problems. Conflict leads to displaced and marginalized people, as do many humanitarian crises. The political and social unrest that accompanies conflict is what makes the associated humanitarian issues much more difficult to right. Threats to humanitarian assis- tance are much more likely if conflict is ongoing, and the deliberate politicization of aid by forces in conflict is an increasing reality which can undermine responses and expose relief workers and beneficiaries to violence and intimidation, a feature of relief efforts in Sudan, DR Congo, and Burma (Lischer, 2006). The First Ivorian Civil War erupted in Côte dâIvoire during 2002 with attacks by rebel forces on the government. These rebels took hold of the northern regions of the country, while the government maintained its claim to the southern regions, making the central region of the country a barrier zone. The conflict continued until 2007, despite numerous attempts at peace throughout the ensuing 4 years.
148 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Before the First Ivorian Civil War, around 2001, a sizeable number of health care staff worked around the country. In the north, there were 38 doctors and 257 nurses; in the central region, there were 127 doctors and 471 nurses; and, in the west, there were 69 doctors and 310 nurses. After the conflict started and had been going on for a few years, around 2004, these numbers changed dramatically. In the north, there were 2 doctors and 82 nurses; in the central region, there were 3 doctors and 67 nurses; and, in the west, there were 6 doctors and 42 nurses (Betsi et al., 2006). Access to care then arose as a major problem for everyone living within the borders of Côte dâIvoire. While the health staff dwindled, the prevalence of STIs markedly rose. Baseline measurements around 2002 showed that 24,636 people in Côte dâIvoire had been infected, making the prevalence risk at that time 10.1 per 1,000 people. Around 2004, the infection rate increased. Measurements taken showed that 29,688 people now lived with an STI, making the new prevalence risk 21.5 per 1,000 people. Within just a few years, not forgetting the conflict that started in 2002, the prevalence had doubled. This increase in STIs is not rare in conflict situations. With decreased access to and use of reproductive health services, the normalization of sexual predation and violence, and increased population mixing, among others, the increase is hard to combat (Mills et al., 2006). The First Ivorian Civil War not only left the country with very little health care infrastructure, but it also started a massive spread of STIs, making many health issues much more complicated. While the issues here discussed may be somewhat specific to Côte dâIvoire, the complex nature of the humanitarian cri- sis, causing rapid displacement, is something shared by all conflicts. The current strife in Syria, with over 100,000 dead and over 2 million refugees, shows that the problems are indeed not isolatable. Natural disasters add additional challenges to these health threats. According to the Brookings-Bern Report (Brookings-Bern Project on Internal Displacement, 2008), the human rights of disaster victims are often not taken into account and include: · Unequal access to assistance · Discrimination in aid provision · Enforced relocation · Sexual and gender-based violence · Loss of documentation · Recruitment of children into fighting forces · Unsafe or involuntary return or resettlement · Property restitution These problems are additional to the many consequences of a natural disaster felt by its victims. The tsunamis, hurricanes, and earthquakes, which hit parts of Asia and the Americas in 2004/2005, highlighted the multiple human rights
APPENDIX A 149 challenges victims of disasters may face, but the 2008 Cyclone Nargis and the response of the Myanmar government best shows the overwhelming problem of human rights within the context of conflict and natural disasters. Case Study: Cyclone Nargis and Burma/Myanmar In May of 2008, Cyclone Nargis hit the southwest corner of Myanmar and sent a massive storm surge into the Irrawaddy Delta. At least 146,000 died, 2.4 million were displaced, and 700,000 homes were destroyed in the wake of this enormous storm. The cyclone washed over some 5,000 km, and radically altered the geography of the Irrawaddy Delta itself. Much of what was rice fields and farmlands before the storm is now open water. As a consequence, 60 percent of Burmaâs rice crop was obliterated. Myanmar is no stranger to civil conflict. At the time of Cyclone Nargisâ landfall, a military dictatorship or junta, the State Peace and Development Coun- cil (SPDC), headed by Senior General Than Shwe, held power. Cyclone Nargis and the response of the Myanmar government to international aid revealed what many had known for decades: The regime of Senior General Than Shwe was incompetent, corrupt, and focused on political survival. On the third day after Cyclone Nargis hit, May 5, the BBC reported a death toll of 351 and that the toll was likely higher. In Labutta, a southwestern town- ship, 75 percent of buildings were said to have collapsed. MRTV, a state-owned television station, reported that 222 people were dead in Irrawaddy and 19 were dead in Rangoon. No official cleanup crews existed, and the Burmese embassy in Thailand closed for a (Thai) holiday. A Rangoon trishaw driver questioned, to the Associated Press, âWhere are all the uniformed people who are always ready to beat civilians?â On day 4, the official death toll rose to 3,394, with 2,879 missing, which was later increased to 22,000 dead and 41,000 missing. Reports arose of looting in Rangoon from a lack of food and clean water. The European Union called the aftermath a âmassive disaster . . . with destruction [of some communities] close to 100 percent.â International aid forces began to assemble as it became clear that the loss of life was enormous, and the response by the ruling junta clearly inadequate. UN Secretary General Ban Ki Moon put a UN Disaster Assessment & Coordina- tion team on standby in Thailand to assist the Burmese government as soon as necessary. The United States released $250,000 of cyclone aid funds and also has a disaster relief team on standby, awaiting permission. The WHO had âof- ficers [who were] on the ground and ready for rapid assessment, surveillance, and mobilization,â including medical teams. The only thing holding all of these groups back was permission from the Burmese government to supply visas and allow the aid to enter the country. Unfortunately, as Jean Maurice Ripert, French
150 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS ambassador to the UN, noted, they were ânot able to [deliver aid] because they [wouldnât] give visas to humanitarian workers.â On day 6, many top leaders in the government disappeared. No responses to world leadersâ condolence messages came in. Most importantly the government did not answer Ban Ki Moonâs calls to discuss aid restrictions. Images of monks helping in cyclone cleanup and relief were banned, while only aid from the SPDC aired on state television. The government asked for direct donations of cash and supplies. All international aid workers still awaited visas. On day 8, the junta proceeded with its long planned constitutional referen- dum, evicting all storm refugees from any polling places. The UN and the U.S. government, among others, had strongly urged the junta not to proceed with the referendum, and to focus on the relief effort. But the generals proceeded, and reported greater than 94 percent voter participation. To no oneâs surprise, the referendum passed overwhelmingly. The SPDC rejected international monitors and barred international relief from the delta. The government went out of their way to ensure that all aid went through them. By day 10, the death toll had risen to 31,938 dead with 29,770 missing. It was not until the 16th day after the cyclone that Than Shwe visited show camps. All of the relief supplies still sat perfectly wrapped and unopened. A Myanmar newspaper reported that âthe government took prompt action to carry out the relief and rehabilitation work after the stormâ despite the differing report from the UN stating that only 20 percent of survivors had received some rudimen- tary aid. The Burmese regime then requested $11.7 billion for rehabilitation and reconstruction with no needs assessment, saying that the first phase of emergency relief was over and that they were moving into the rebuilding phase. Finally, on the 21st day after Cyclone Nargis hit, Than Shwe met with Ban Ki Moon, who had personally come to the country to break the block on assistance, and agreed to allow in aid workers. While it is certainly difficult to ascribe the power and scale of Cyclone Nar- gis to climate change, land use patterns and environmental destruction did likely play more measureable roles in the stormâs impact and loss of life. The Irrawaddy Delta is a very large, low-lying coastal marsh region, once protected from the open sea by dense mangrove forests. Under British rule in the nineteenth cen- tury, the delta was drained, and a century of intensive rice paddy cultivation and population in-migration followed. By the time Nargis hit, the delta was a densely populated region producing more than 50 percent of Burmaâs wet rice crop, and the protective mangroves and coastal marshes had been decimated. This exposed rural and remote coastal communities to the full force of the storm, and many communities were washed over in the first, massive storm surge. Military misrule limited the humanitarian response to this natural disaster, but climate change and land use patterns exposed communities and led to enormous losses of life.
APPENDIX A 151 Instability Bias In times of conflict, diseases and health problems do not subside. In fact, as we have discussed, the opposite is true. Research of all kinds, including health research and disease surveillance, however, can markedly decline when conflict arises. This problem, which we have characterized, is known as instability bias and makes it difficult to assess health outcomes related to conflict. During the rule of Mobutu Sese Seko from 1965 to 1997, the Democratic Republic of the Congo or Zaire, as it was known at the time, faced corruption, state violence, and internal conflict. Zaire was also an epicenter of the emergence of HIV/AIDS, and a key country in early efforts to investigate and understand this newly emerging human pathogen. New HIV/AIDS studies in DR Congo peaked from 1986â1988 at 16 studies per year and then started to decline (Fig- ure A3-1). New malaria studies also peaked from 1986â1988 (Figure A3-2). In 1994, Mobutu ordered that international collaborative research stop (Beyrer and Pizer, 2007). Seventeen peer-reviewed publications on HIV/AIDS in the Democratic Re- public of the Congo came out in 1990. In 2002, none were published (Beyrer and Pizer, 2007). Of course, the issue of HIV/AIDS had not been resolved in DRC. Instead, the political unrest within the country halted the research. The example of research in the Democratic Republic of the Congo shows not only the power that conflict can have on controlling the amount of research FIGURE A3-1 Causal loop diagram of Cyclone Nargis. The causal loop diagram illus- trates the relationship between climate change, international and national governance, and conflict in Myanmar in the aftermath of Cyclone Nargis in 2008. SOURCE: Naples, 2011.
152 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A3-2âBibliometric analysis of HIV publications, Democratic Republic of Congo, 1982â2004. SOURCE: Beyrer and Pizer, 2007. produced, but also the power of the ruling class. Researchers must utilize cre- ative methods such as expanding community engagement practices (Amon et al., 2012). Otherwise, research on issues such as infectious diseases will continue to be sparse on conflict zones. Ways Forward According to an ICE case study, âThe Intergovernmental Panel on Climate Change (IPCC) predicts an increase in extreme weather events such as tropical cyclones in the Southeast Asian region. Cyclone Nargis, which struck Myanmar on May 2, 2008, illustrates the potential for extreme weather events to contribute to conflictâ (Naples, 2011). In this same case study, the ICE proposed a frame- work that outlines how human agency in the form of climate change can lead to natural disasters and the outcome of conflict (Figure A3-3). Taking into account what we know about the impact of conflict on the spread of infectious disease and what we have learned from Cyclone Nargis and Myanmar, we must begin to rec- ognize human agency and its interactions with global infectious disease threats. To better address this researchers can partner with grassroots organizations and human rights groups in country and internationally. More importantly, part- nering with those we seek to serve facing these complex and overlapping threats and again expanding community engagement practices can provide opportunities for more effective health efforts in conflict zones (Amon et al., 2012). Migration, civil conflicts, and climate change are all likely to be more com- mon, and to interact with the well-being of communities and populations in the
APPENDIX A 153 FIGURE A3-3â Malaria studies initiated, Democratic Republic of Congo, 1980â2004. SOURCE: Beyrer and Pizer, 2007. years and decades to come. Relief efforts must be prepared for complex crises, and new approaches to delivery of relief will likely be required to address these emerging threats. References Amon, J. J., S. D. Baral, C. Beyrer, and N. Kass. 2012. Human rights research and ethics review: protecting individuals or protecting the state? PLoS Medicine 9(10):e1001325. Betsi, N. A., B. G. Koudou, G. Cisse, A. B. Tschannen, A. M. Pignol, Y. Ouattara, Z. Madougou, M. Tanner, and J. Utzinger. 2006. Effect of an armed conflict on human resources and health systems in Cote dâIvoire: prevention of and care for people with HIV/AIDS. AIDS Care 18(4):356-365. Beyrer, C., and T. Lee. 2008. Responding to infectious diseases in Burma and her border regions. Conflict and Health 2(1):2. Beyrer, C., and H. Pizer. 2007. Civil conflict and health information: The Democratic Republic of Congo. In Public Health & Human Rights: Evidence-Based Approaches. Baltimore: Johns Hopkins University Press. Beyrer, C., S. Baral, and J. Zenilman. 2006. Holmes et al., eds. STDs, HIV/AIDS, and migrant popula- tions. In Sexually transmitted diseases 4th edn. Seattle: McGraw-Hill Professional. Brookings-Bern Project on Internal Displacement. 2008. Human Rights and Natural Disasters: Op- erational Guidelines and Field Manual on Human Rights Protection in Situations of Natural Disaster http://www.refworld.org/docid/49a2b8f72.html (accessed April 1, 2014). Hsiang, S. M., M. Burke, and E. Miguel. 2013. Quantifying the influence of climate on human con- flict. Science 341(6151): DOI: 10.1126/science.1235367
154 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Lischer, S. K. 2006. Dangerous sanctuaries: Refugee camps, civil war, and the dilemmas of humani- tarian aid. Ithaca, N.Y: Cornell University Press. Mills, E. J., S. Singh, B. D. Nelson, and J. B. Nachega. 2006. The impact of conflict on HIV/AIDS in sub-Saharan Africa. Int J STD AIDS 17(11):713-717. Naples, E. 2011. ICE Case Study 249: Cyclone Nargis, Climate and Conflict. http://www1.american. edu/ted/ice/cyclone-nargis.htm (accessed April 1, 2014). Piarroux, R., R. Barrais, B. Faucher, R. Haus, M. Piarroux, J. Gaudart, R. Magloire, and D. Raoult. 2011. Understanding the cholera epidemic, Haiti. Emerging Infectious Diseases 17(7):1161-1168. Richards, A. K., K. Banek, L. C. Mullany, C. I. Lee, L. Smith, E. K. S. Oo, and T. J. Lee. 2009. Cross- border malaria control for internally displaced persons: observational results from a pilot pro- gramme in eastern Burma/Myanmar. Tropical Medicine & International Health 14(5):512-521. Sollom, R. C. Beyrer, D. Sanders, and A. F. Donaghue. 2009. Health in ruins: A man-made crisis in Zimbabwe. An emergency report by Physicians for Human Rights. Tucker, J. D., G. E. Henderson, T. F. Wang, Y. Y. Huang, W. Parish, S. M. Pan, X. S. Chen, and M. S. Cohen. 2005. Surplus men, sex work, and the spread of HIV in China. AIDS 19(6):539-547. A4 THE IMPORTANCE OF MOVEMENT IN ENVIRONMENTAL CHANGE AND INFECTIOUS DISEASE Nita Bharti11 Abstract Global environmental changes directly impact human movement and mobility, which in turn drive infectious disease dynamics and pathogen transmission. In addition to establishing the importance of movement in disease dynamics, characterizing the mechanistic relationship between envi- ronment, host behavior, and pathogen transmission is becoming increasingly necessary. Environmental systems are diverging from previous patterns while continuing to mediate individual and group movements as well as the complex interactions between population dynamics and disease dynamics. Introduction Movement and Disease The effects of environmental changes on infectious diseases are most often discussed in their direct links to wildlife diseases and vector borne pathogens. However, a closer look into the complexity of infectious disease systems reveals 11ââCenter for Infectious Disease Dynamics, Pennsylvania State University, University Park, PA 16802.
APPENDIX A 155 that environmentally driven host movements are a critical element in infectious disease dynamics. Movement and mobility are known to be important underly- ing mechanisms driving the spatiotemporal dynamics of infectious diseases, both within and between populations (Altizer et al., 2011; Bharti et al., 2010; Bradley and Altizer, 2005; Gray et al., 2009; Loehle, 1995; Morgan et al., 2007; Tatem et al., 2009, 2012; Viboud et al., 2006). Examples of environmentally mediated movements include movement motivated by food and water security and seasonal migration patterns. The links between host movement and infectious diseases have been studied in the context of animal migrations (Altizer et al., 2011). Data show that long- distance mass animal migrations can either reduce disease via âmigratory escapeâ (Loehle, 1995) or âmigratory cullingâ (Bradley and Altizer, 2005) or increase disease by creating new or high density contacts at stopovers and destinations (Morgan et al., 2007). In addition to recognizing the importance of mobility, these examples from animal migrations illustrate why it is necessary to develop a mechanistic understanding of the relationship between movement and contact patterns as environmental changes occur. Despite establishing the importance of host movement in infectious disease dynamics (Bharti et al., 2010; Gray et al., 2009; Tatem et al., 2009, 2012; Viboud et al., 2006), many aspects of mobility remain difficult to measure, particularly in humans. Epidemiologically important patterns of movement can remain un- known or poorly understood outside of local knowledge. As a result, it can be challenging to incorporate these movements into public health planning and disease prediction efforts. Improved knowledge and quantification of movement patterns would help in planning and implementing more effective public health interventions (Camargo et al., 2000). We investigate the role of seasonal human movement in pathogen transmis- sion, disease incidence, and immunization programs. Specifically, we investigate measles dynamics in three cities in the West African nation Niger from 1995 to 2005 as an example of an environmentally driven migration impacting pathogen transmission and control in urban areas. We intentionally use an inherently simple disease, measles, to understand the complex dynamics of populations and disease. Population Dynamics and Disease Measles is a strongly immunizing, directly transmitted human disease with no vector, no animal reservoir, and no direct interaction with the environment. Recorded cases of measles in prevaccination industrialized nations are among the richest disease data sets (Fine and Clarkson, 1982). These data demonstrate the highly regular annual and multiennial cycles of measles incidence (Anderson and May, 1991; Cliff et al., 1993). Detailed demographic and measles case records clearly linked population dynamics to disease dynamics; births replenished the susceptibles in the population (Grenfell et al., 2002), the birth rate determined
156 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS the periodicity of outbreaks (Bartlett, 1957), and the aggregation of susceptibles increased contact rates raised transmission rates and triggered measles outbreaks (Fine and Clarkson, 1982). In prevaccination Europe and England this aggrega- tion took place in classrooms, and measles outbreaks were seasonally forced onto school terms with a mean age of infection around 5 years (Fine and Clarkson, 1982). After a vaccine was developed, mathematical analyses showed that the most effective time to vaccinate the population was during the troughs of infection (London and Yorke, 1973; Yorke et al., 1979). This reduced the density of sus- ceptibles prior to the start of school terms, preventing epidemics from taking off. This strategy was successful and became conventional practice in measles immu- nization. Throughout large regions of the world, similar preventative vaccination campaigns targeted during the troughs of infection were extremely effective in achieving and maintaining high levels of coverage and interrupting local chains of transmission (Cliff et al., 1993). Many high-income nations have maintained greater than 90 percent vaccination coverage for decades, locally eliminating measles infections during these periods (Cliff et al., 1993). In particular, the Pan American Health Organization (PAHO) has been widely commended for implementing a highly successful vaccination strategy that fo- cused on age-specific routine vaccinations and catch-up campaigns at regular intervals to keep immunization levels consistently high (Castillo-Chavez et al., 2011). PAHOâs strategy largely eliminated measles in the Americas and was her- alded as an example that would pave the way for measles eradication. The African Health Observatory (AHO) adopted a similar strategy for the African regionâs measles control initiative. However, the program has not been as succcessful as it was in the Americas, and measles outbreaks continued to cause morbidity and mortality across the continent (Simons et al., 2012). So why was a strategy highly effective in one place yet failed to produce similar results in another? In contrast to the American region, the African region had a completely different geography along with higher levels of diversity and population movements that were not considered in PAHOâs vaccination strategy but were likely major contributors to its significantly reduced efficacy in the African region. Measles in Niger Today, measles continues to persist across many areas of the globe, but nowhere more than Asia and Africa (The case of measles, 2011; Simons et al., 2012), particularly in places with high birth rates (Bongaarts and Caterline, 2012). Throughout the past decades, Niger put forth significant public health efforts to reduce the national burden of measles. In addition to routine immuniza- tions and catch-up campaigns, the Ministry of Health maintained detailed records of measles cases and vaccine coverage. Despite significant investments, measles epidemics persisted, with the biggest cities at risk for particularly large outbreaks (Ferrari et al., 2008).
APPENDIX A 157 Nigerâs measles outbreaks are strongly seasonal, occurring only during the annual dry season. Although the magnitude of outbreaks can vary greatly be- tween years, the timing is extremely consistent (Ferrari et al., 2008). We focus on measles epidemics in the three largest cities in NigerâNiamey, Maradi, and Zinder (Figure A4-1A)âwhere the seasonal forcing in transmission is stronger than previously observed anywhere else, including prevaccination cities (Ferrari et al., 2010). The strong seasonal forcing causes the outbreaks to subside; epi- demics are not self-limiting due to an exhaustion of susceptible individuals, as is often the case with measles (Cliff et al., 1993; Ferrari et al., 2008, 2010; Grenfell and Bolker, 1998). Despite frequent recurrences, measles is not endemic in Niger, often disap- pearing completely during the rainy season, even from the largest cities (Bartlett, 1957; Bjornstad and Grenfell, 2008; Ferrari et al., 2008; Grenfell and Bolker, 1998). The very high birth rates rapidly replenish the supply of susceptibles, creating the potential for frequent or large outbreaks in the absence of high vac- cination levels (Bartlett, 1957; Ferrari et al., 2008; Grenfell et al., 2002). The median age of measles infection in Niger is around 2 years, which is too young for transmission to be focused in schools (Ferrari et al., 2008). The observed seasonal outbreaks of measles in Niger had also been noted in other parts of the region. The underlying reason, though definitively unknown, was hypothesized to be the result of agricultural labor migrations (Ferrari et al., 2008). During the rainy season in this highly agricultural economy, it is not un- common to disperse to rural areas for agricultural work and then aggregate in ur- ban areas during the dry season (Faulkingham and Thorbahn, 1975; Rain, 1999). Measuring Movement and Interpreting Its Role Epidemiologically important movements within a familiar context may be relatively easy to detect; in western societies this may include weekday work commutes (Viboud et al., 2006), long-distance travel (Tatem et al., 2012), and travel around major holidays. It can be much more problematic to identify movement patterns found only outside of oneâs own culture, including different livelihoods and the embedded movement patterns. In this case, inhabitants of western societies may be unfamiliar with nomadic pastoralism (Dyson-Hudson and Dyson-Hudson, 1980) or cyclical regional migration (Byerlee, 1974). In some cases, these types of movements are studied in an ethnographic context with relatively small sample sizes. Careful ethnographic research has detailed seasonal movement patterns for agricultural work in Niger (Faulkingham and Thorbahn, 1975; Rain, 1999). After detection, these movements must be measured and, to explain the observed city-level measles dynamics, the seasonal movements of small groups must be scaled to match large urban areas. However, extracting large-scale data from small-scale samples or scaling down movement data from large data sets, such as a national census, is simply not possible.
158 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Fortunately, advances in technology and methodology have improved our abilities to measure movement patterns at high spatiotemporal resolution. Vari- ous aspects of human presence can be captured and traced by satellite imagery (Elvidge et al., 1997, 2009; Sutton et al., 1997, 2001), including changes in urban population density and spread (Bharti et al., 2011). Remote Measures of Changing Urban Populations To assess the relationship between changes in population density and measles epidemics, it was necessary to quantify urban populations across seasons. These data could not be extracted from existing ethnographic studies and census data, as mentioned earlier. Other existing data sources, such as composite satellite imagery, often present annually or biannually aggregated information. Commer- cial flight records illustrate high-temporal-resolution movements but track only long-distance movements. Lastly, there are no functioning railways in Niger; movement occurs along roads, but seasonal measurements of road use and traffic are not recorded. To quantify seasonally varying high-resolution spatial changes in popula- tions in these three cities in Niger, we developed a method using noncompos- ited serial nighttime light satellite images. These images capture anthropogenic visible light at night during low moon and cloud-free conditions (Bharti et al., 2011; Elvidge et al., 1997). We created an annual signature of brightness values for each city using 155 images, concurrent with the time period of measles data collection (Bharti et al., 2011; see supporting online materials 1 for method de- tails). We found a consistent, pronounced dip in brightness in each city during the rainy season and a peak during the dry season (Figure A4-1B-D), illustrating that population fluctuations were strongly correlated to the measles transmission curve specific to each city (Figure A4-1B-D) (Bharti et al., 2011; Sutton, 1997). To look more closely at the spatial relationship between brightness and measles cases, we focused on the three communes within the city of Niamey (Figure A4-2A, inset). Daily case records at the commune level from a 2004 out- break showed that the epidemic appeared earliest in the two largest communes, where 90 percent of the cases in the city occurred, and appeared last with the fewest cases in the smallest commune (Bharti et al., 2011; Dubray et al., 2006) (Figure A4-2A). The brightness curves for these three communes displayed a similar pattern: the two largest communes increased and peaked earlier with very high brightness values, and the third commune increased and peaked later with a relatively lower brightness value (Figure A4-2B) (Bharti et al., 2011). A Dynamic Model In simple theoretical models, migration may be thought of as a source of infected individuals. The dynamics of measles epidemics in the cities of Niger
APPENDIX A 159 FIGURE A4-1âMeasles transmission rates and brightness for three cities in Niger (adapted from Bharti et al., in prep). Top left: map of Africa, Niger shaded; top right: map of Niger showing three largest cities and national health districts. B. Niamey. Left: annual brightness pattern against day of year shown in open circles. Color corresponds to time (blue to red = January to December); estimated measles transmission rates for biweekly time steps shown in circles connected by dark lines. Right: estimated transmission rates against brightness values; colors correspond to time of year as on left. C. Same as B for Maradi. D. Same as B for Zinder.
160 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A4-2â Measles and brightness in the communes of Niamey (adapted from Bharti et al., 2011). A. Inset: map of city of Niamey showing pixels of the city color-coded by commune. Formal boundaries of each commune are shown with black outlines. Time se- ries of weekly reported measles cases for Niameyâs 2003â2004 outbreak by commune. B. Time series of brightness values, colors by commune as in A. Red arrow indicates onset of measles epidemic in Niamey. C. Left: commune 1, right: commune 2. Points show daily reported measles cases, shading gives central 95 percent of predicted measles incidence from 25,000 model simulations from nighttime lights-informed model (red), no immigra- tion model (blue), and constant immigration model (gray). The x-axis spans the duration of the epidemic: day 307 of 2003 to day 153 of 2004. suggested that migration was also an important source of susceptible individu- als as well as a critical driver of changes in population density and contact rates (Bharti et al., 2011; Ferrari et al., 2008). To determine whether the satellite-derived quantified changes in population density could drive the seasonal forcing of the observed measles epidemics in each commune of Niamey, we adapted an SIR model that included fluctuations in the total population size. Conventional SIR models of measles have included seasonal forcing in the transmission rate, a single rate that encompasses (1) the
APPENDIX A 161 probability of a contact event occurring between an infected and a susceptible person, and (2) the probability of a transmission event occurring, given such a contact (Begon et al., 2002). We separated these two probabilities and assessed them independently. In the communes of Niger, the per capita rate of contacts between susceptible and infected individuals is unlikely to change across sea- sons. Instead, the number of contact events increases with the overall population density. So while more contacts occur during the dry season, the proportional number of contacts between susceptible and infected individuals does not change. Secondly, the probability of transmission, given a contact between a susceptible and infectious individual, does not change between seasons. Measles outbreaks consistently occur with host aggregation and have been historically observed across all seasons (Cliff et al., 1993). Thus, instead of including seasonality in the transmission term, we allowed the total population size to change with the derivative of the brightness curve for each commune (Bharti et al., 2011). In addition to the model with migration informed by commune-level changes in brightness, we fit two additional models for comparison. For each commune, we also fit (1) an SIR model with a constant migration rate, and (2) a model with no migration (static population size) to the daily measles case reports. The model results showed that for communes 1 and 2, where 90 percent of the measles cases within the city occurred, the brightness-informed changes in population size were required to produce the correct rate of increase and decrease of cases as well as the timing and height of the peak of cases (Figure A4-2C) (Bharti et al., 2011). Although commune 3 had far fewer cases, the model with brightness-informed population fluctuations was better than the other two at replicating the correct timing of the peak as well as the increase and decrease in cases (Bharti et al., 2011). The model illustrates that in order to reach the observed height at the peak of each communeâs epidemic, susceptible individuals must be added over the course of the epidemicâs increase. If the population starts with the necessary number of susceptible individuals to sustain the epidemic, the number of cases increases far too quickly and the peak appears much sooner than observed before declining exceptionally more rapidly than that of the recorded outbreaks. In the absence of population fluctuations, the epidemic trajectory in each of Niameyâs three com- munes would look very different (Bharti et al., 2011). Vaccination Often overlooked, migration has a strong impact on health care and immu- nization programs. In several instances, movement and mobility have been de- finitively identified not only as the underlying drivers of spatiotemporal epidemic patterns, but also as important disregarded elements in public health interven- tions. Within São Paulo, Camargo et al. (2000) showed that risk factors related to a 1997 measles epidemic included migration from other states and ruralâurban
162 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS migration within the state and determined that movement should be considered when planning a measles vaccination strategy. Relocating also increased the risk that a child would miss a vaccination for polio in India, Angola, and Pakistan (Unicef, 2013). Perhaps most specifically, seasonal migration in Niger was identi- fied as a high-risk factor for children lacking measles vaccination in a 1990â1991 outbreak in Niamey (Malfait et al., 1994). In a place like Niamey, population fluctuations are not only strongly seasonal and pronounced, they are also the mechanism underlying measles outbreaks. This means that the troughs of infection align with troughs in population size and, contrary to conventional wisdom, may not be the most effective time to vaccinate the population. Practiced in Niger, this strategy would vaccinate fewer individuals than would be present in urban areas during other times of the year. This also means that individuals from hard to reach or remote locations are not opportunistically vaccinated when they are easily accessible in urban areas. Planning successful intervention strategies relies heavily on understanding local patterns of mobility. Instead of characterizing populations as static entities that can be described with relatively constant values of size and density, we may benefit from considering them to be more fluid with changing membership. It is possible to use fluctuations in populations as opportunities to immunize or provide access to health care to groups and individuals who might otherwise be difficult to reach. Understanding population fluctuations is also important in estimating the population size at the time of an intervention so that the correct number of vaccine doses can be provided. Previous research has illustrated the merits of regional coordination in infec- tious disease interventions in this area due to common transnational movement patterns (Bharti et al., 2010, 2012). When looking specifically at urban regions, this is an even more valid argument. Seasonal or ruralâurban movements are not always contained within a stateâs or nationâs borders, and regionally coordi- nated vaccination efforts will reduce the gaps in coverage created by population movements. Conclusion Though perhaps unintentionally, we often consider populations to be rela- tively stable and static in size and density. We know with certainty that this is not only an overly simplistic representation of human populations, it also overlooks the massive impact that movement has on health. This perspective inadvertently inhibits our ability to understand geographically varying important underlying mechanisms of pathogen transmission and epidemic spread as well as access to health care. Understanding the relationship between human movement patterns and dis- eases has presented unique challenges. Although known to be central in disease transmission and spatiotemporal patterns of disease dynamics, epidemiologically
APPENDIX A 163 important patterns of movement can be difficult to identify and measure. In- terdisciplinary research and technological and methodological advances have made immense progress towards enhancing our understanding of movement and mobility in the context of the environment and health. Mobility traces from cell phones (Bengtsson et al., 2011; Gonzalez et al., 2008; Tatem et al., 2009), satel- lite imagery (Bharti et al., 2011; Checchi and Grundy, 2012), and high-resolution aerial photography as well as ground-truthing some of these proxy measures (Min et al., 2013) have already greatly advanced the methods and data behind under- standing populations and their movements across a wide range of geographic areas, environmental settings, and health concerns. Measuring the many aspects of mobility and interpreting their prevalence across spatiotemporal scales is a difficult task, but it is a necessary step towards reducing disease and informing intervention strategies. References Altizer, S., R. Bartel, and B. A. Han. 2011. Animal migrations and infectious disease risk. Science 331:296-302. Anderson, R. M., and R. M. May. 1991. Infectious diseases of humans: Dynamics and control, Oxford science publications. Oxford; New York: Oxford University Press. Bartlett, M. S. 1957. Measles periodicity and community size. Journal of the Royal Statistical Society Series A-General 120(1):48-70. Begon, M., M. Bennett, R. G. Bowers, N. P. French, S. M. Hazel, and J. Turner. 2002. A clarification of transmission terms in host-microparasite models: Numbers, densities and areas. Epidemiology and Infection 129(1):147-153. Bengtsson, L., X. Lu, A. Thorson, R. Garfield, and J. von Schreeb. 2011. Improved response to disasters and outbreaks by tracking population movements with mobile phone network data: A post-earthquake geospatial study in Haiti. PLoS Medicine 8(8):9. Bharti, N., A. Djibo, M. J. Ferrari, R. F. Grais, A. Tatem, C. McCabe, O. N. Bjornstad, and B. Grenfell. 2010. Measles hotspots and epidemiological connectivity. Epidemiology and Infection 138(09):1308-1316. Bharti, N., A. J. Tatem, M. J. Ferrari, R. F. Grais, A. Djibo, and B. T. Grenfell. 2011. Explaining sea- sonal fluctuations of measles in Niger using nighttime lights imagery. Science 334:1424-1427. Bharti, N., H. Broutin, R. F. Grais, M. J. Ferrari, A. Djibo, A. J. Tatem, and B. T. Grenfell. 2012. Spatial dynamics of meningococcal meningitis in Niger: Observed patterns in comparison with measles. Epidemiology and Infection 140(8):1356-1365. Bjornstad, O. N., and B. T. Grenfell. 2008. Hazards, spatial transmission and timing of outbreaks in epidemic metapopulations. Environmental and Ecological Statistics, 15:265-277. Bongaarts, J., and J. Caterline. 2012. Fertility transition: Is Sub-Saharan Africa different? Population and Development Review 38(s1):153-168. Bradley, C. A., and S. Altizer. 2005. Parasites hinder monarch butterfly flight: Implications for disease spread in migratory hosts. Ecology Letters 8:290-300. Byerlee, D. 1974. Rural-urban migration in Africa: Theory, policy and research implications. Inter- national Migration Review 8(4):543-566. Camargo, M. C. C., J. C. De Moraes, V. A. U. F. Souza, M. R. Matos, and C. S. Pannuti. 2000. Predic- tors related to the occurrence of a measles epidemic in the city of Sao Paulo in 1997. Revista Panamericana de Salud Pública 7(6):359-365. The case of measles. 2011. News feature: MacMillan Publishers Limited.
164 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Castillo-Chavez, C., C. R. Matus, B. Flannery, M. C., G. Tambini, and J. K. Andrus. 2011. The Americas: Paving the road toward global measles eradication. Journal of Infectious Diseases 204(Suppl 1):S270-S278. Checchi, F., and C. Grundy. 2012. Satellite imagery for rapid estimation of displaced populations: A validation and feasability study. Final project report. Cliff, A. D., P. Haggett, and M. Smallman-Raynor. 1993. Measles: An historical geography of a ma- jor human viral disease from global expansion to local retreat, 1840-1990. Oxford [England]; Cambridge, MA: Blackwell. Dubray, C., A. Gervelmeyer, A. Djibo, I. Jeanne, F. Fermon, M. H. Soulier, R. F. Grais, and P. J. Guerin. 2006. Late vaccination reinforcement during a measles epidemic in Niamey, Niger (2003-2004). Vaccine 24(18):3984-3989. Dyson-Hudson, R., and N. Dyson-Hudson. 1980. Nomadic pastoralism. Annual Review of Anthropol- ogy 9:15-61. Elvidge, C. D., K. E. Baugh, E. A. Kihn, H. W. Kroehl, and E. R. Davis. 1997. Mapping city lights with nighttime data from the DMSP Operational Linescan System. Photogrammetric Engineer- ing and Remote Sensing 63(6):727:734. Elvidge, C. D., P. C. Sutton, T. Ghosh, B. T. Tuttle, K. E. Baugh, B. Bhaduri, and E. Bright. 2009. A global poverty map derived from satellite data. Computers & Geosciences 35(8):1652-1660. Faulkingham, R. H., and P. F. Thorbahn. 1975. Population dynamics and drought: A village in Niger. Population Studies 29(3):463-477. Ferrari, M. J., R. F. Grais, N. Bharti, A. J. K. Conlan, O. N. Bjornstad, L. J. Wolfson, P. J. Guerin, A. Djibo, and B. T. Grenfell. 2008. The dynamics of measles in sub-Saharan Africa. Nature 451:679-684. Ferrari, M. J., R. F. Grais, A. Djibo, N. Bharti, C. N. Bjornstad, and B. Grenfell. 2010. Rural-urban gradient in seasonal forcing of measles transmission in Niger. Proceedings of the Royal Society B-Biological Sciences 277(1695):2775-2782. Fine, P. E. M., and J. A. Clarkson. 1982. Measles in England and WalesâI: An analysis of factors underlying seasonal patterns. International Journal of Epidemiology 11(1):5-14. Gonzalez, M. C., C. A. Hidalgo, and A. L. Barabasi. 2008. Understanding individual human mobility patterns. Nature 453:779-782. Gray, R. R., A. J. Tatem, S. Lamers, W. Hou, O. Laeyendecker, D. Serwadda, N. Sewankambo, R. H. Gray, W. Wawer, T. C. Quinn, M. M. Goodenow, and M. Salemi. 2009. Spatial phylodynamics of HIV-1 epidemic emergence in east Africa. AIDS 23(14):F9-F17. Grenfell, B. T., and B. M. Bolker. 1998. Cities and villages: Infection hierarchies in a measles meta- population. Ecology Letters 1(1):63-70. Grenfell, B. T., O. N. Bjornstad, and B. F. Finkenstadt. 2002. Dynamics of measles epidemics: Scaling noise, determinism, and predictability with the TSIR model. Ecological Monographs 72(2):185-202. Loehle, C. 1995. Social barriers to pathogen transmission in wild animal populations. Ecology 76(2):326-335. London, W. P., and J. A. Yorke. 1973. Recurrent outbreaks of measles, chickenpox and mumps. I. Seasonal variation in contact rates. American Journal of Epidemiology 98:453-468. Malfait, P., I. M. Jataou, M. C. Jollet, A. Margot, A. C. Debenoist, and A. Moren. 1994. Measles epidemic in the urban-community of Niamey - Transmission patterns, vaccine efficacy and im- munization strategies, Niger, 1990 to 1991. Pediatric Infectious Disease Journal 13(1):38-45. Min, B., K. M. Gaba, O. F. Sarr, and A. Agalassou. 2013. Detection of rural electrification in Af- rica using DMSP-OLS night lights imagery. International Journal of Remote Sensing 34(22): 8118-8141. Morgan, E. R., G. F. Medley, P. R. Torgerson, B. S. Shaikenov, and E. J. Milner-Gulland. 2007. Parasite transmission in a migratory multiple host system. Ecological Modelling 200:511-520. Rain, D. 1999. Eaters of the dry season: Circular labor migration in the West African Sahel. Boulder, CO: Westview Press.
APPENDIX A 165 Simons, E., M. J. Ferrari, J. Fricks, K. Wannemuehler, A. Anand, A. Burton, and P. Strebel. 2012. Assesment of the 2010 global measles mortality reduction goal: Results from a model of surveil- lance data. Lancet 379(9832):2173-2178. Sutton, P. 1997. Modeling population density with night-time satellite imagery and GIS. Computers, Environment and Urban Systems 21(3-4):227-244. Sutton, P., D. Roberts, C. D. Elvidge, and H. Meij. 1997. A comparison of nighttime satellite imagery and population density of the continental United States. Photogrammetric Engineering and Remote Sensing 63(11):1303-1313. Sutton, P., D. Roberts, C. D. Elvidge, and K. E. Baugh. 2001. Census from heaven: An estimate of the global population using nighttime satellite imagery. International Journal of Remote Sens- ing 22(16):3061-3076. Tatem, A. J., Y. Qui, D. L. Smith, O. Sabot, A. S. Ali, and B. Moonen. 2009. The use of mobile phone data for the estimation of the travel patterns and imported P. falciparum rates among Zanzibar residents. Malaria Journal 8(1):12. Tatem, A. J., Z. Huang, A. Das, Q. Qi, J. Roth, and Y. Qui. 2012. Air travel and vector-borne disease movement. Parasitology 139(14):1816-1830. Unicef. 2013. Polio Info. http://polioinfo.org/ (accessed October 24, 2013). Viboud, C., O. N. Bjornstad, D. L. Smith, L. Simonsen, M. A. Miller, and B. T. Grenfell. 2006. Syn- chrony, waves, and spatial hierarchies in the spread of influenza. Science 312(5772):447-451. Yorke, J. A., N. Nathanson, G. Pianigiani, and J. Martin. 1979. Seasonality and the requirements for perpetuation and eradication of viruses in popluations. American Journal of Epidemiology 109(2):103-123. A5 TOWARD A COUNTY-LEVEL MAP OF TUBERCULOSIS RATES IN THE U.S.12 David Scales,13 John S. Brownstein,13 Kamran Khan,14 and Martin S. Cetron15 Introduction Active tuberculosis (TB) is a reportable communicable disease in all 50 states, but nationwide, county-level data are not released publicly. The CDCâs On- line Tuberculosis Information System (OTIS) provides public surveillance data only by state. Owing to an agreement with the states, the CDC cannot publicly 12ââReprinted from American Journal of Preventive Medicine, 46(5), Scales et al. Toward a county- level map of tuberculosis rates in the U.S. Pp. e49-e51, Copyright 2014, with permission from Elesevier. 13ââChildrenâs Hospital Informatics Program, Boston Childrenâs Hospital, Harvard Medical School, Boston, Massachusetts. 14ââDepartment of Medicine, Division of Infectious Diseases, University of Toronto, Toronto, On- tario, Canada. 15ââDivision of Global Migration and Quarantine, U.S. Centers for Disease Control and Prevention, Atlanta, Georgia.
166 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS release TB data at the county level, precluding the development of publicly avail- able, county-level maps of TB cases and incidence rates. The lack of a more granular nationwide data set has limited the study of TB trends and socioeconomic risk factors to states (Holtgrave and Crosby, 2004), Metropolitan Statistical Areas (Greenwood and Warriner, 2011), or census tracts within a single state (Myers et al., 2006). A nationwide county-level data set of TB rates provides opportunities to examine TB-related trends across multiple states, metropolitan areas, and across counties with similar demographic char- acteristics, such as the number of people deemed to be at high risk (Cain et al., 2008). Methods TB statistics were generated after extracting publicly available data from state health department websites and requesting public but unpublished county- level data from state TB programs. States providing TB data assented to their use and presentation. The data set, metadata, and sources are published on an interactive map with downloadable data at healthmap.org/tb. TB incidence rates were calculated using 5-year county-level case counts with corresponding (2006â2010) population estimates from the American Com- munity Survey (ACS). Specifically, the total county-level case counts for 2006â 2010 were divided by five to obtain the average number of cases per year. This average was divided by the average population in the county in those 5 years, and finally multiplied by 100,000 to calculate the incidence rate (cases/ 100,000). Therefore, âratesâ reported in Table A5-1 and Figure A5-1 represent average an- nual incidence during the 5-year period. Counties were cross-classified by four U.S. regions (Midwest, Northeast, South, and West) and by urban/rural classification (metropolitan [urban area of â¥50,000]; micropolitan [urban area of 10,000â49,999]; and rural), according to the Office of Management and Budget classifications. ANOVA was performed to assess differences in means across these 12 cross-classifications. ANOVAs were examined using Welch two-sample t-tests with Bonferroni adjustment for 42 comparisons (α = 0.0012); significant comparisons were defined as p < 0.0012. Maps were created using ESRI ArcMap, version 10.3, and statistical analyses were performed using R, version 2.14.2. Data were available on a year-by-year basis for 2,892 (92.0%) counties; supra-county health district level for 161 (5.1%) counties; and only multi-year aggregated data for 90 counties (2.9%). Collectively, these data enabled the cre- ation of a U.S. map depicting 5-year average TB incidence rates (Figure A5-1) and a corresponding data set of 3,006 counties for analysis. Henceforth, we use the term âcountyâ to refer collectively to counties; county-equivalents (e.g., bor- oughs); and health districts.
TABLE A5-1â Comparison of Average Annual TB Rates of U.S. Counties and Regions by Urban (Rural/Micropolitan/ Metropolitan) Classification, 2006â2010a Number of counties, county equivalents, and health districts Median annual TB rates per 100,000 by county and urban classification Region Rural Micropolitan Metropolitan Total Region Rural Micropolitan Metropolitan All Classes Midwest 486 218 259 963 Midwest 0 0.805 0.95 0.33 Northeast 41 53 123 217 Northeast 0.52 0.90 1.78 1.10 South 594 297 570 1,461 South 1.67 2.49 2.305 2.16 West 165 82 118 365 West 0 1.28 2.21 1.22 Total 1,286 650 1,070 3,006 All regions 0 1.38 1.78 1.28 Mean annual TB rates per 100,000 by county and urban classification, Annual TB rates per 100,000 by region and urban classification, 2006â2010, M (SD) 2006â2010 Region Rural Micropolitan Metropolitan All Classes Region Rural Micropolitan Metropolitan All Classes Midwest 1.01 (3.06) 1.30 (2.28) 1.26 (1.64) 1.14 (2.57) Midwest 0.95 1.17 2.70 2.32 Northeast 0.68 (0.82) 0.94 (0.67) 2.49 (2.57) 1.77 (2.16) Northeast 0.63 0.99 4.50 4.14 South 3.07 (6.21) 3.72 (5.86) 2.90 (2.70) 3.14 (5.06) South 3.09 3.43 4.60 4.33 West 1.70 (3.95) 1.81 (1.88) 3.33 (3.34) 2.25 (3.46) West 2.33 1.95 5.64 5.31 All regions 2.04 (4.93) 2.44 (4.40) 2.50 (2.66) 2.29 (4.14) All regions 2.20 2.27 4.48 4.11 Significant difference of M pairs from Table 1Cb Between regions National: South versus all other regions (p < 0.0001); Midwest versus all other regions (p < 0.001) Within regions Rural: South versus all other regions (p < 0.001) Northeast: Metro versus rural or micro (p < 0.0001) Micro: South versus all other regions (p < 0.0001); Northeast versus West (p < 0.001) West: Metro versus rural or micro (p < 0.001) Metro: Midwest versus all other regions (p < 0.0001) aRegion and metropolitan/micropolitan classifications follow Office of Management and Budget definitions. Rural counties are those not defined as either metropolitan or micropolitan. bSignificant according to Welch two-sample t-test and Bonferroni adjustment, where α = 0.0012. Number of comparisons = 42. TB, tuberculosis 167
168 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A5-1â Average annual tuberculosis rate per 100,000 population, 2006â2010, by county tuberculosis data from publicly available sources. Population estimates from U.S. Census American Community Survey, 2006â2010. Results More than 600 counties have TB rates above the 2011 national rate of 3.4 cases per 100,000 people (Miramontes et al., 2012). The top 15 counties exceeded a rate of 20 cases per 100,000 (range = 20.9â120.3 cases); nine of these were rural, and eight were in the Southern region. TB case rates were generally high- est in U.S. metropolitan areas; the South had the highest mean and median rates among U.S. regions (Table A5-1). Only the Northeast and West had statistically different mean rates when metropolitan counties were compared with micropoli- tan and rural means. Discussion A publicly available, county-level TB data set enables analysis of TB rates (per 100,000) at the sub-state level. Although TB rates in the U.S. are expected to be high in urban areas (Oren et al., 2011) that have large at-risk foreign-born populations (Liu et al., 2009), certain rural areas also have high TB rates, par- ticularly in Southern states.
APPENDIX A 169 Publicly available county-level TB data can assist TB surveillance and con- trol efforts. TB âhotspotsâ that cross state borders can be identified. Socio- economic variables can now be tested to identify nationwide trends in at-risk populations for targeted prevention efforts. Thus, we encourage all states to publish county-level TB data online. Additional demographic information distinguishing cases by birth country will help researchers and public health officials understand emerging TB trends. Although TB data should be interpreted within a local context, these data will facilitate more efficient identification of locales where high rates of TB cross state lines, facilitate collaboration between states to jointly target those areas, and al- low health departments to discern regional and nationwide trends. Acknowledgements We gratefully acknowledge the state and county public health departments that contributed to this project by posting data online (Alaska, Alabama, Ari- zona, California, Colorado, Connecticut, District of Columbia, Delaware, Florida, Georgia, Idaho, Illinois, Indiana, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Missouri, Mississippi, Montana, North Carolina, Nebraska, New Hampshire, New Jersey, New Mexico, Nevada, New York, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Washington, Wisconsin, West Virginia, and Wyoming) or providing data via e-mail (Arkansas, Hawaii, Iowa, Kansas, Massachusetts, Maine, North Dakota, and Vermont). We thank Rachel Chorney for her work on the healthmap.org/tb website, for which no direct compensation was received. DS had full access to all study data and takes responsibility for its integrity and the accuracy of the data analysis. Dr. Brownstein is supported by grant R01 LM010812-04 from the National Li- brary of Medicine. This study was funded by the CDC. Dr. Cetron from the Division of Global Migration and Quarantine at the CDC participated as a full scientific collaborator in the investigation; however, the find- ings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC. No financial disclosures were reported by the authors of this paper.
170 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS References Cain KP, Benoit SR, Winston CA, Mac Kenzie WR. Tuberculosis among foreign-born persons in the U.S. J Am Med Assoc 2008;300(4):405â12. Greenwood MJ, Warriner WR. Immigrants and the spread of tuberculosis in the U.S.: a hidden cost of immigration. Popul Res Policy Rev 2011;30:839â59. Holtgrave DR, Crosby RA. Social determinants of tuberculosis case rates in the U.S. Am J Prev Med 2004;26(2):159â62. Liu Y, Weinberg MS, Ortega LS, Painter JA, Maloney SA. Overseas screening for tuberculosis in U.S.-bound immigrants and refugees. N Engl J Med 2009;360(23):2406â15. Miramontes R, Pratt R, Price SF, Jeffries C, Navin TR, Oramasionwu GE. Trends in tuberculosisâ U.S., 2011. MMWR Morb Mortal Wkly Rep 2012;61(11):181â5. Myers WP, Westenhouse JL, Flood J, Riley LW. An ecological study of tuberculosis transmission in California. Am J Public Health 2006;96(4):685â90. Oren E, Winston CA, Pratt R, Robison VA, Narita M. Epidemiology of urban tuberculosis in the U.S., 2000â2007. Am J Public Health 2011; 101(7):1256â63. A6 ASSESSING THE ORIGIN OF AND POTENTIAL FOR INTERNATIONAL SPREAD OF CHIKUNGUNYA VIRUS FROM THE CARIBBEAN16 Kamran Khan,17 Isaac Bogoch,18 John S. Brownstein,19 Jennifer Miniota,20 Adrian Nicolucci,19 Wei Hu,19 Elaine O. Nsoesie,21 Martin Cetron,22 Maria Isabella Creatore,19 Matthew German,19 and Annelies Wilder-Smith23 16ââOriginally printed as âAssessing the Origin of and Potential for International Spread of Chikun- gunya Virus from the Caribbean.â PLoS Currents Outbreaks. 2014 Jun 6. Edition 1. doi: 10.1371/ currents.outbreaks.2134a0a7bf37fd8d388181539fea2da5. 17ââDepartment of Medicine, Division of Infectious Diseases, University of Toronto, Toronto, Canada. 18ââDepartment of Medicine, Division of Infectious Diseases, University of Toronto, Toronto, Can- ada; University Health Network, Divisions of Internal Medicine and Infectious Diseases, Toronto, Canada. 19ââBoston Childrenâs Hospital, Harvard Medical School, Boston, Massachusetts, USA. 20ââLi Ka Shing Knowledge Institute, St. Michaelâs Hospital, Toronto, Canada. 21ââChildrenâs Hospital Informatics Program, Boston Childrenâs Hospital, Boston, Massachusetts, USA; Department of Pediatrics, Harvard Medical School, Boston, Massachusetts, USA; Network Dynamics and Simulation Science Laboratory, Virginia Bioinformatics Institute, Virginia Tech, Blacksburg, Virginia, USA. 22ââDivision of Global Migration and Quarantine, Centers for Disease Control and Prevention, Atlanta, USA; Departments of Medicine and Epidemiology, Emory University School of Medicine and Rollins School of Public Health, Atlanta, USA. 23ââLee Kong Chian School of Medicine, Nanyang Technological University, Singapore; Institute of Public Health, University of Heidelberg, Germany.
APPENDIX A 171 Abstract Background: For the first time, an outbreak of chikungunya has been reported in the Americas. Locally acquired infections have been confirmed in fourteen Caribbean countries and dependent territories, Guyana and French Guiana, in which a large number of North American travelers vacation. Should some travelers become infected with chikungunya virus, they could potentially introduce it into the United States, where there are competent Aedes mosquito vectors, with the possibility of local transmission. Methods: We analyzed historical data on airline travelers departing areas of the Caribbean and South America, where locally acquired cases of chikungunya have been confirmed as of May 12th, 2014. The final destina- tions of travelers departing these areas between May and July 2012 were determined and overlaid on maps of the reported distribution of Aedes aeygpti and albopictus mosquitoes in the United States, to identify potential areas at risk of autochthonous transmission. Results: The United States alone accounted for 52.1% of the final des- tinations of all international travelers departing chikungunya indigenous areas of the Caribbean between May and July 2012. Cities in the United States with the highest volume of air travelers were New York City, Miami and San Juan (Puerto Rico). Miami and San Juan were high travel-volume cities where Aedes aeygpti or albopictus are reported and where climatic conditions could be suitable for autochthonous transmission. Conclusion: The rapidly evolving outbreak of chikungunya in the Carib- bean poses a growing risk to countries and areas linked by air travel, includ- ing the United States where competent Aedes mosquitoes exist. The risk of chikungunya importation into the United States may be elevated following key travel periods in the spring, when large numbers of North American travelers typically vacation in the Caribbean. Introduction Chikungunya virus is a mosquito-transmitted alphavirus endemic to sub- Saharan Africa and South and East Asia. In recent years, chikungunya has been appearing outside of its endemic zone as a result of increasing international travel (Enserink, 2007; Tomasello and Schlagenhauf, 2013). Concurrently, the geo- graphic ranges of Aedes aeygpti and albopictusâthe primary vectors for chikun- gunya virusâhave been expanding, a phenomenon thought to be a consequence of climate change and globalization (Reiter et al., 2006). The combination of international travel by potentially infected persons and the increasing geographic availability of competent vectors has set the stage for the introduction and spread
172 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS of Chikungunya to previously unaffected areas. In recent years, autochthonous transmission of chikungunya has occurred in non-endemic areas such as the 2007 outbreak in Italy and 2010 outbreak in France, and most recently, in multiple Caribbean Islands where competent Aedes mosquitoes exist (Tomasello and Schlagenhauf, 2013). The geographic dispersion of chikungunya virus may occur in instances where susceptible travelers in endemic areas are bitten by infected female Aedes mosquitoes (Powers and Logue, 2007). After the typical incubation period of 3-7 days (range 2-12 days), infected individuals become viremic (Borgherini et al., 2007; Sissoko et al., 2008). Among those who develop illness, common symptoms include fever, headache, rash, and severe symmetrical polyarthralgia. The potential for an infected individual to then transmit chikungunya virus to a susceptible Aedes mosquito is greatest during the first 2-6 days of illness, during the viremic phase (Appassakij et al., 2013). For the first time in the Americas, chikungunya was reported among non- travelers on the Caribbean island of St. Martin in December 2013 (CDC, 2013). Since then, locally acquired cases have been reported in multiple countries and territories in the region for a total count of over 4,000 probable or confirmed cases, raising concerns that this virus could spread into and within neighboring areas, including parts of the United States (Gibney et al., 2011; Reiskind et al., 2008). Every year, large numbers of North American tourists vacation in the Carib- bean during spring and summer months. After returning home, these individuals could potentially introduce chikungunya virus into areas where the conditions necessary for autochthonous transmission exist. We used a novel approach com- bining a number of datasets related to travel routes, volumes of travelers, historic temperature data and zoonotic distribution of Aedes mosquitoes in order to model the recent outbreak in the Caribbean and the risk of spread to other countries via international travel. Due to the large travel volume between the Caribbean and the U.S. we conducted an analysis to determine the vulnerability of U.S. cities and states to the importation of chikungunya virus and subsequent local transmission due to favorable environmental conditions. Methods We accessed anonymized, worldwide, passenger-level flight itinerary data for 2012 from the International Air Transport Association (IATA). The IATA dataset represents an estimated 93% of the worldâs commercial air traffic at the passenger level. Flight itinerary data includes information on the airport where the traveler initiated their trip, and where relevant, connecting flights leading up to their final destination. Using this dataset, we first analyzed the origins of all air travelers depart- ing chikungunya endemic areas of the world (as defined by the U.S. Centers
APPENDIX A 173 for Disease Control and Prevention, [2014a]) that had final destinations in the Caribbean region (as defined by the United Nations [2014]) during the period from October to December 2012 (to assess potential origins of chikungunya virus introduction into the Caribbean in December 2013). Next, we analyzed the final international destinations of all travelers (be- tween May and July 2012) departing areas of the Caribbean where locally ac- quired cases of chikungunya have been confirmed as of May 12th, 2014 (i.e. Aruba, Anguilla, Antigua, British Virgin Islands, Dominica, Dominican Republic, French Guiana, Guadeloupe, Haiti, Martinique, St. Barthelemy, St. Kitts and Nevis, and St. Martin, Sint Maarten, St. Vincent and the Grenadines). We then calculated the volume of travelers departing these indigenous areas of the Caribbean between May and July 2012 with and their countries of final destination. We also calculated city-level volumes of travelers with final destina- tions in North America. These monthly city-level travel data were mapped and overlaid with the geographic extents of Aedes aeygpti and Aedes albopictus mosquitoes across the United States (CDC, 2014b). We then determined the aver- age monthly temperature of each state between May and July using 60 years of historical data (WeatherBase, 2014). While there are many unknowns regarding the climatic conditions necessary for Aedes aeygpti and albopictus mosquitoes to transmit chikungunya virus (Ruiz-Moreno et al., 2012), an average temperature of 20° Celsius was identified as an important threshold in the 2007 chikungunya outbreak in Italy (Charrel et al., 2008; Fischer et al., 2013; Tilston et al., 2009). Results While the specific origin of the Caribbean chikungunya epidemic is not precisely known, we found that five countries were the source of 84.4% of all international air travelers departing chikungunya endemic areas of the world with final destinations in the Caribbean region between the months of October and December 2012. These countries included South Africa (4,348 travelers; 23.4% of all travelers from chikungunya endemic areas of the world), India (4,012 travelers; 21.6%), China (2,561 travelers; 13.8%), Philippines (2,555 travelers; 13.7%) and the French territory of Réunion (2,218 travelers; 11.9%). With respect to the possibility of receiving an imported case via international air travel, the final destinations of travelers departing areas of the Caribbean where locally acquired cases of chikungunya have been confirmed (as of May 12th, 2014), over the three-month period from May to July 2012 are shown in Table A6-1. Three countries represented the final destinations of 70.0% of all travelers worldwide. The United States, including Puerto Rico, had the strongest links through international air travel (1,071,658 travelers; 52.1% of the global total), followed by France (298,921 travelers; 14.5%), and the Netherlands An- tilles, not including Sint Maarten (68,604 travelers; 3.3%). By comparison, ten cities represented the final destinations of 49.0% of all travelers. These included
174 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS TABLE A6-1â Leading Destination Countries for Travelers Departing Chikungunya Indigenous Areas of the Caribbean Country Traveler Volume* Global Total (%) Cumulative Total (%) United Statesâ 1,071,658 52.2 52.2 France 298,921 14.5 66.7 Netherland Antilles 68,604 3.3 70.0 Canada 64,736 3.2 73.2 Spain 55,329 2.7 75.9 Venezuela 42,774 2.1 78.0 Germany 36,984 1.8 79.8 United Kingdom 28,480 1.4 81.1 Italy 27,159 1.3 82.4 St. Lucia 24,102 1.2 83.6 Panama 23,576 1.2 84.8 *Between May and July 2012. â Includes Puerto Rico. New York (283,224 travelers; 13.8% of the global total), Paris (240,204 travelers; 11.7%), Miami (161,430 travelers; 7.8%), San Juan, Puerto Rico (80,571 travel- ers; 3.9%), Curacao (48,594 travelers; 2.4%), Fort Lauderdale (45,076 travelers; 2.2%), Madrid (41,286 travelers; 2.0%), Boston (40,829 travelers; 1.9%), Toronto (36,162 travelers; 1.7%), and Caracas (29,973 travelers; 1.4%). Discussion Global forces from climate change to surging worldwide air travel are con- tributing to the globalization of vector-borne diseases such as West Nile virus, dengue and chikungunya (Fischer et al., 2013; Greer et al., 2008; Sutherst, 2004; Tatem et al., 2006). In December 2013, chikungunya virus was identified for the first time in the Americas, where it has since caused over four thousand lo- cally acquired cases across numerous Caribbean islands in addition to the South American nations of Guiana and French Guiana. While the origins of chikungu- nya introduction in the Caribbean are not precisely known, molecular diagnos- tics have determined that the strain currently circulating in the region belongs to the subtype CHIKV-JC2012 and closely resembles a strain found in China, the Philippines and Micronesia (Laniciotti and Valadere, 2014). Our analysis suggests that five chikungunya endemic countries account for the vast majority of international air travel into the Caribbean region in the months leading up to the first reported cases, with China and the Philippines accounting for 27.5% of all such travelers. However, the probability of importation into the Caribbean is a function not only of travel volumes but also of chikungunya incidence in the origin countries. Our analyses indicate that the United States is the final destination of over half of all travelers departing chikungunya indigenous areas of the Caribbean,
APPENDIX A 175 followed by France, which accounts for almost 15% of all travelers. The United States has never reported local transmission of chikungunya virus, despite the presence of Aedes aeygpti and albopictus mosquitoes across the southeastern region of the country, while autochthonous transmission of chikungunya has pre- viously been documented in southeastern France in 2010, where Aedes albopictus is known to exist (Vega-Rua et al., 2013). Furthermore, many North American travelers vacationing in the Caribbean will return to areas of the United States where the climate may be suitable for autochthonous transmission. We found that New York City, Miami and San Juan are the leading U.S. des- tination cities of travelers from chikungunya indigenous areas of the Caribbean between May and July. Healthcare providers in these locations should familiarize themselves with the clinical presentation of chikungunya, which overlaps sig- nificantly with dengue fever. The early detection of chikungunya is particularly important in areas such as San Juan, Miami, and Charlotte where competent mos- quito vectors could become infected through bites of viremic travelers (Reiskind et al., 2008). Symptomatic individuals with suspected or confirmed chikungunya infection should take special measures to avoid mosquito bites in the week fol- lowing the onset of their illness (when viremia is greatest) to decrease the poten- tial for autochthonous spread. Although there are many unknowns about the biology of Aedes mosquitoes and the specific climatic conditions that would support autochthonous transmis- sion, warmer weather is thought to shorten the interval between the time when an Aedes mosquito is infected by a viremic patient and when that mosquito can transmit the virus to another susceptible human host (i.e. the extrinsic incubation period) (Liu-Helmersson et al., 2014). Since the 2013-2014 winter season has been unseasonably cool across many parts of the United States, this could favor longer extrinsic periods, and consequently a lower probability of viral transmis- sion from vector to human host. Although belonging to a different strain from the one currently circulating in the Caribbean, of potential concern is the chikun- gunya E1-A226V mutation identified during the 2005-2006 Réunion epidemic, which facilitated more efficient transmission specifically in Aedes albopictus mosquitoes (Schuffenecker et al., 2006; Tsetsarkin et al., 2007). This mutation was subsequently imported to Italy, and has since appeared in China and Papua New Guinea (Bordi et al., 2008; Horwood et al., 2013; Schuffenecker et al., 2006; Wu et al., 2013). However, this mutation does not appear to dominate in the major chikungunya outbreaks that occurred in India 2006-2010 (Kumar et al., 2014). Our analysis has several important limitations. First, we are relying on ac- curate identification of indigenous chikungunya cases in the Caribbean region to conduct our analyses of population movements through air travel. Some countries in the Caribbean may have limited infectious disease surveillance capacity, par- ticularly for a newly emerging pathogen such as chikungunya. Our transportation analysis was also limited to commercial air travel despite the fact that many indi- viduals vacationing in the Caribbean may travel on cruise ships or other means of
176 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS transport. This limitation would presumably lead to an underestimate of travelers arriving in U.S. port cities that face the Caribbean islands, though the length of travel by sea may exclude them spreading disease further. Similarly, we ana- lyzed commercial air travel data from 2012, which may not reflect forthcoming patterns of travel in 2014. While we found a highly consistent seasonal pattern of travel between the United States and chikungunya indigenous areas of the Caribbean in earlier years (analyses not shown), travel behaviors this year could be influenced by evolving news of chikungunya in the media. We also relied on accurate vector surveillance data for Aedes aeygpti and albopictus to identify areas at risk of potential autochthonous transmission. While such vector surveil- lance has limitations, we used contemporary data reported by the U.S. Centers for Disease Control and Prevention as of January 2014 (CDC, 2014b). Finally, the environmental factors necessary to support autochthonous transmission of chikungunya are complex and influenced not only by the type of vector, but also chikungunya virus characteristics. The climatic conditions required for efficient viral transmission are still under investigation; however, it is likely that warmer temperatures are more favorable. Therefore climatic conditions that evolve over the next several months will likely play a significant role in either hindering or supporting autochthonous transmission of chikungunya. At a time when locally acquired cases of dengue (also transmitted by Aedes aeygpti and albopictus mosquitoes) have recently been reported in southern regions of the United States (Adalja et al., 2012; Bouri et al., 2012; Effler et al., 2005; Radke et al., 2012; Ramos et al., 2008), our findings highlight the risk for introduction and potential autochthonous transmission of chikungunya virus in selected areas of the country. The effectiveness and efficiency of interventions to mitigate these risks could be optimized through a combination of public educa- tion, early detection by medical providers, and the strategic use of public health resources in areas of greatest risk. Author Contributions Kamran Khan and Isaac Bogoch jointly developed the design of the study, oversaw the completion of all analyses, and produced the first draft of the manu- script. Jennifer Miniota, Wei Hu, and Adrian Nicolucci conducted reviews of the literature, performed all transportation and spatial analyses, created figures and cartograms, and edited the final version of the manuscript. John Brownstein contributed epidemiological data pertaining to chikungunya in the Caribbean and made significant content contributions and edits to the final manuscript. Marisa Creatore, Martin Cetron and Annelies Wilder-Smith made significant content contributions to the initial draft of the manuscript and edits to the final draft of the manuscript.
FIGURE A6-1â Volume of travelers from chikungunya indigenous areas of the Caribbean* to the United States and Canada in Mayâ . *As of May 12th 2014. 177 â Using historic air travel data from May 2012.
178 FIGURE A6-2â Volume of travelers from chikungunya indigenous areas of the Caribbean* to the United States and Canada in Juneâ . *As of May 12th 2014, â Using historic air travel data from June 2012.
FIGURE A6-3â Volume of travelers from chikungunya indigenous areas of the Caribbean* to the United States and Canada in Julyâ . â *As of May 12th 2014. 179 â â Using historic air travel data from July 2012.
180 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Funding Statement This study was funded by the Canadian Institutes of Health Research. The funders did not influence the content of this manuscript nor the decision to submit it for publication. References Adalja AA, Sell TK, Bouri N, Franco C. Lessons learned during dengue outbreaks in the United States, 2001-2011. Emerg Infect Dis. 2012 Apr;18(4):608-14. PubMed PMID:22469195. Appassakij H, Khuntikij P, Kemapunmanus M, Wutthanarungsan R, Silpapojakul K. Viremic profiles in asymptomatic and symptomatic chikungunya fever: a blood transfusion threat? Transfusion. 2013 Oct;53(10 Pt 2):2567-74. PubMed PMID:23176378. Bordi L, Carletti F, Castilletti C, Chiappini R, Sambri V, Cavrini F, Ippolito G, Di Caro A, Capobianchi MR. Presence of the A226V mutation in autochthonous and imported Italian chikungunya virus strains. Clin Infect Dis. 2008 Aug 1;47(3):428-9. PubMed PMID:18605910. Borgherini G, Poubeau P, Staikowsky F, Lory M, Le Moullec N, Becquart JP, Wengling C, Michault A, Paganin F. Outbreak of chikungunya on Reunion Island: early clinical and laboratory features in 157 adult patients. Clin Infect Dis. 2007 Jun 1;44(11):1401-7. PubMed PMID:17479933. Bouri N, Sell TK, Franco C, Adalja AA, Henderson DA, Hynes NA. Return of epidemic dengue in the United States: implications for the public health practitioner. Public Health Rep. 2012 May- Jun;127(3):259-66. PubMed PMID:22547856. Charrel RN, de Lamballerie X, Raoult D. Seasonality of mosquitoes and chikungunya in Italy. Lancet Infect Dis. 2008 Jan;8(1):5-6. PubMed PMID:18156081. Effler PV, Pang L, Kitsutani P, Vorndam V, Nakata M, Ayers T, Elm J, Tom T, Reiter P, Rigau-Perez JG, Hayes JM, Mills K, Napier M, Clark GG, Gubler DJ. Dengue fever, Hawaii, 2001-2002. Emerg Infect Dis. 2005 May;11(5):742-9. PubMed PMID:15890132. Enserink M. Infectious diseases. Chikungunya: no longer a third world disease. Science. 2007 Dec 21;318(5858):1860-1. PubMed PMID:18096785. Fischer D, Thomas SM, Suk JE, Sudre B, Hess A, Tjaden NB, Beierkuhnlein C, Semenza JC. Climate change effects on Chikungunya transmission in Europe: geospatial analysis of vectorâs climatic suitability and virusâ temperature requirements. Int J Health Geogr. 2013 Nov 12;12:51. PubMed PMID:24219507. Gibney KB, Fischer M, Prince HE, Kramer LD, St George K, Kosoy OL, Laven JJ, Staples JE. Chikungunya fever in the United States: a fifteen year review of cases. Clin Infect Dis. 2011 Mar 1;52(5):e121-6. PubMed PMID:21242326. Greer A, Ng V, Fisman D. Climate change and infectious diseases in North America: the road ahead. CMAJ. 2008 Mar 11;178(6):715-22. PubMed PMID:18332386. Horwood P, Bande G, Dagina R, Guillaumot L, Aaskov J, Pavlin B. The threat of chikungunya in Oceania. Western Pac Surveill Response J. 2013 Apr-Jun;4(2):8-10. PubMed PMID:24015365. Kumar A, Mamidi P, Das I, Nayak TK, Kumar S, Chhatai J, Chattopadhyay S, Suryawanshi AR, Chattopadhyay S. A novel 2006 Indian outbreak strain of Chikungunya virus exhibits different pattern of infection as compared to prototype strain. PLoS One. 2014;9(1):e85714. PubMed PMID:24465661. Lanciotti RS, Valadere AM. Transcontinental movement of Asian genotype chikungunya virus. Emerging Infectious Diseases 2014;20(8): http://dx.doi.org/10.3201/eid2008.140268 Liu-Helmersson J, Stenlund H, Wilder-Smith A, Rocklov J. Effects of diurnal temperature variations on global dengue epidemic potential. PLoS ONE 2014;In Print.
APPENDIX A 181 Powers AM, Logue CH. Changing patterns of chikungunya virus: re-emergence of a zoonotic arbo- virus. J Gen Virol. 2007 Sep;88(Pt 9):2363-77. PubMed PMID:17698645. Radke EG, Gregory CJ, Kintziger KW, Sauber-Schatz EK, Hunsperger EA, Gallagher GR, Barber JM, Biggerstaff BJ, Stanek DR, Tomashek KM, Blackmore CG. Dengue outbreak in Key West, Florida, USA, 2009. Emerg Infect Dis. 2012 Jan;18(1):135-7. PubMed PMID:22257471. Ramos MM, Mohammed H, Zielinski-Gutierrez E, Hayden MH, Lopez JL, Fournier M, Trujillo AR, Burton R, Brunkard JM, Anaya-Lopez L, Banicki AA, Morales PK, Smith B, Muñoz JL, Waterman SH. Epidemic dengue and dengue hemorrhagic fever at the Texas-Mexico border: results of a household-based seroepidemiologic survey, December 2005. Am J Trop Med Hyg. 2008 Mar;78(3):364-9. PubMed PMID:18337327. Reiskind MH, Pesko K, Westbrook CJ, Mores CN. Susceptibility of Florida mosquitoes to infection with chikungunya virus. Am J Trop Med Hyg. 2008 Mar;78(3):422-5. PubMed PMID:18337338. Reiter P, Fontenille D, Paupy C. Aedes albopictus as an epidemic vector of chikungunya virus: an- other emerging problem? Lancet Infect Dis. 2006 Aug;6(8):463-4. PubMed PMID:16870524. Ruiz-Moreno D, Vargas IS, Olson KE, Harrington LC. Modeling dynamic introduction of Chi- kungunya virus in the United States. PLoS Negl Trop Dis. 2012;6(11):e1918. PubMed PMID:23209859. Schuffenecker I, Iteman I, Michault A, Murri S, Frangeul L, Vaney MC, Lavenir R, Pardigon N, Reynes JM, Pettinelli F, Biscornet L, Diancourt L, Michel S, Duquerroy S, Guigon G, Frenkiel MP, Bréhin AC, Cubito N, Desprès P, Kunst F, Rey FA, Zeller H, Brisse S. Genome microevolu- tion of chikungunya viruses causing the Indian Ocean outbreak. PLoS Med. 2006 Jul;3(7):e263. PubMed PMID:16700631. Sissoko D, Moendandze A, Malvy D, Giry C, Ezzedine K, Solet JL, Pierre V. Seroprevalence and risk factors of chikungunya virus infection in Mayotte, Indian Ocean, 2005-2006: a population-based survey. PLoS One. 2008 Aug 26;3(8):e3066. PubMed PMID:18725980. Sutherst RW. Global change and human vulnerability to vector-borne diseases. Clin Microbiol Rev. 2004 Jan;17(1):136-73. PubMed PMID:14726459. Tatem AJ, Hay SI, Rogers DJ. Global traffic and disease vector dispersal. Proc Natl Acad Sci U S A. 2006 Apr 18;103(16):6242-7. PubMed PMID:16606847. Tilston N, Skelly C, Weinstein P. Pan-European Chikungunya surveillance: designing risk stratified surveillance zones. Int J Health Geogr. 2009 Oct 31;8:61. PubMed PMID:19878588. Tomasello D, Schlagenhauf P. Chikungunya and dengue autochthonous cases in Europe, 2007-2012. Travel Med Infect Dis. 2013 Sep-Oct;11(5):274-84. PubMed PMID:23962447. Tsetsarkin KA, Vanlandingham DL, McGee CE, Higgs S. A single mutation in chikungunya virus affects vector specificity and epidemic potential. PLoS Pathog. 2007 Dec;3(12):e201. PubMed PMID:18069894. United Nations. Composition of macro geographical (continental) regions, geographical sub-regions, and selected economic and other groupings. United Nations Statistical Division, 2014 US Centers for Diease Control and Prevention. The Yellow Book: CDC Health Information for In- ternational Travel. US Centers for Diease Control and Prevention, 2014a ââââ. Chikungunya: Information for vector control programs. US Centers for Diease Control and Prevention, 2014b US Centers for Diease Control and Prevention. First reports of Chikungunya in Western Hemisphere. US Centers for Diease Control and Prevention, 2013 Vega-Rua A, Zouache K, Caro V, Diancourt L, Delaunay P, Grandadam M, Failloux AB. High ef- ficiency of temperate Aedes albopictus to transmit chikungunya and dengue viruses in the Southeast of France. PLoS One. 2013;8(3):e59716. PubMed PMID:23527259. WeatherBase. United States of America - Weather Averages. WeatherBase, 2014 Wu D, Zhang Y, Zhouhui Q, Kou J, Liang W, Zhang H, Monagin C, Zhang Q, Li W, Zhong H, He J, Li H, Cai S, Ke C, Lin J. Chikungunya virus with E1-A226V mutation causing two outbreaks in 2010, Guangdong, China. Virol J. 2013 Jun 2;10:174. PubMed PMID:23725047.
182 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS A7 EIGHT CRITICAL QUESTIONS FOR PANDEMIC PREDICTION Toph Allen,24 Kris Murray,24 Kevin J. Olival,24 and Peter Daszak24 Introduction Like hurricanes or earthquakes, pandemics are rare events that can be ex- tremely devastating, causing substantial mortality and economic damages. How- ever, unlike hurricanes or earthquakes, efforts to identify where pandemics are most likely to originate, and to intervene and preempt their impact, are in their nascence. Here, we review recent advances in disease ecology, virology, and biogeography that move the field towards these goals and pose a series of critical questions that must be addressed to sufficiently improve our predictive capacity. This provides a framework for pandemic prediction that may allow us to better allocate our global resources to mitigate this threat. Because the majority of recent pandemics are zoonotic in origin, most in- volving wildlife reservoirs, we consider this group specifically. The emergence of pandemic zoonoses reflects a complex interplay of socioeconomic, ecological, and biological factors and can be thought of as a three-stage process (Morse et al., 2012). Initially, pathogens with pandemic potential exist only in their natural reservoirs. In the first stage, pre-emergence, our encroachment into a reservoirâs natural habitat, often related to changing land use, may bring these pathogens into contact with livestock or humans or otherwise alter the ecological system in which it and its host exist. In the second stage, localized emergence, initial transmission to humans occurs, directly from a wildlife host or via domesticated animals. Some of these events may involve small chains of person-to-person transmission. When a pathogen achieves sustained person-to-person transmis- sion, the right confluence of circumstances can lead to pandemic emergence, ultimately with large outbreaks propelled internationally by the movement of people and disease vectors. Each of these stages is itself driven by a plethora of socioeconomic, eco- logical, and biological factors (e.g., change in land use, migration, agricultural intensification) that alter pathogen dynamics and expose human populations to increasing risk of zoonotic disease emergence, amplification, and spread. It fol- lows that to predict and pre-empt pandemics, we must improve our understanding of how these factors drive increased risk of each stage of the pandemic process (Morse et al., 2012). The complexity of these processes is daunting, but the in- terplay of ecology, demography, virology, and biology provides a wide range of new tools and approaches that can be used in pandemic prediction and prevention. 24ââEcoHealth Alliance, 460 West 34th Street, New York, USA.
APPENDIX A 183 These include strategies to analyze prior outbreaks, model future trends in pan- demic drivers, conduct targeted surveillance in wildlife and human populations, and probe the depth of the zoonotic âpoolâ from which novel EIDs arise. Here we review some of these by posing eight critical questions for pandemic prediction. Eight Critical Questions for Pandemic Prediction Are Emerging Infectious Diseases (EIDs) Really on the Rise? The literature on emerging infectious diseases, and concern among policy makers and the public, has grown substantially in recent years (IOM, 1992, 2003). Does this reflect a public health threat that is also growing, or is this trend driven by increased surveillance, or simply better reporting of outbreaks as they occur? To test this, we expanded and updated a database of all known emerging infectious disease, first collated by Mark Woolhouseâs group (Taylor et al., 2001). We focused on âEID events,â which we defined as âthe first temporal emergence of a pathogen in a human population . . . related to the increase in distribution, increase in incidence or increase in virulence or other factor which led to that pathogen being classed as an emerging diseaseâ (Jones et al., 2008). For each event, we collected data on location, time, and host and/or vector, as well as on associated ecological, biological, and sociodemographic drivers of disease emergence, and performed a number of temporal and spatial regression analyses. Our analyses showed that the number of EID events has increased over time, peaking in the 1980â1990 decade. This peak was associated with increased sus- ceptibility to infection due to the HIV/AIDS pandemic. Like Taylor et al. (2001), we found that zoonoses comprised the majority of EID events (60.3 percent), and that almost 71.8 percent of zoonotic EIDs were from wildlife (43.3 percent of all EID events). Furthermore, zoonoses from wildlife were increasing as a propor- tion of all EID eventsâin the last decade analyzed (1990â2000), 52.0 percent of EID events were zoonoses with known a wildlife origin. We attempted to correct for increasing infectious disease reporting effort over time by including in our regression model the number of articles published in the Journal of Infectious Diseases (which gives a crude measure of research effort for infectious diseases generally, not just EIDs) for each decade as an offset. Controlling for reporting effort gave further support to the conclusions that EID events are becoming more common, that zoonoses comprise the majority of EID events, and that zoonoses are rising significantly faster as a proportion of all EID events. Are There Predictable Patterns to Disease Emergence? The first step in predicting a biological phenomenon is to look for patterns that underlie previous events. This approach underpins hurricane forecasting and the identification of earthquake zones, and is a logical strategy for pandemic
184 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS prediction. Both hurricane forecasting and the identification of earthquake zones look to the underlying drivers of these phenomena to identify patterns. We used a similar approach for disease emergence, focusing on the hypothesized drivers of zoonotic disease emergence. We assigned geographic coordinates to EID events and, using a logistic regression, tested associations between subsets of EID events and a small selection of hypothesized drivers. We found that drug-resistant and vector-borne pathogens, and zoonoses with wildlife and non-wildlife origins, differed in their global patterning and in their associations with different drivers. In particular, all categories of EID events were strongly associated with human population density, which we have suggested â. . . supports previous hypotheses that disease emergence is largely a product of anthropogenic and demographic changes. . . .â Human population growthâtaken as a broad proxy for change in socio-economic factorsâpredicts zoonoses from non-wildlife and the emergence of drug-resistant pathogens. However, zoonoses from wildlife are alone in their association with wildlife host species richnessâpatterns of wildlife diversity. The overall predicted risk from different categories was differentially distributed across the globe. For instance, wildlife zoonoses and vector-borne pathogens were more likely to have originated in lower-latitude, developing countries (Jones et al., 2008) (Figure A7-1). Describing these patterns provides the first step towards pandemic prediction: predictive models exist of future trends in socio- economic and demographic drivers, and may be used to derive predictive models of the future trends in disease emergence. The analyses of Jones et al. (2008) show that EID emergence is driven by socioeconomic as well as biological factors, but they are somewhat preliminary, and substantial gaps remain. For example, what aspects of human population density drive disease emergence? Is it anthropogenic environmental changes (e.g., road building, deforestation, land use change)? Is it increased contact with wildlife, or the perturbation of pathogen transmission dynamics in wildlife? Or do dense human populations simply provide an âamplification zoneâ that allows more frequent recognition of new EIDs otherwise lost to our analyses? Efforts to tease apart the mechanisms underlying these patterns will involve ecological, vi- rological, and biological disciplines collaborating in exciting new ways (Murray and Daszak, 2013). Finally, it is interesting to note that the Jones et al. (2008) models leave 85 percent of the variation in global patterns of disease emergence unexplained. This emphasizes the magnitude of the problem, sets the bar for fu- ture studies, and highlights that efforts to gradually improve the modelâs power need to be prioritized if we are to accurately predict the next pandemic. Where Will the Next Pandemic Originate? There are significant geopolitical and logistical constraints to pandemic prevention. Newly emerged pathogens often originate in remote areas that are difficult to access, and in resource-constrained countries that cannot afford to
FIGURE A7-1â Map of relative risk of a zoonotic disease of wildlife origin emerging in people. Because almost all prior pandemics, and the majority of emerging infectious diseases, are zoonotic in origin, with the majority of these having a wildlife host, this map acts as a potential basis for future targeted surveillance and the pre-empting of potential pandemics. 185
186 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS systematically identify novel pathogens in their early stages of emergence. Once emerging diseases become pandemic, the large number of cases and wide geo- graphic distribution make response programs costly and complicated by geopo- litical issues. Given the finite global capacity for pandemic preparedness, and a limited global budget, can we reconfigure where our resources are spent, based on a scientific understanding of where novel diseases emerge and where our current effort is lacking in relation? To this end, our previous âEID hotspotsâ analysis attempts to correct for bias caused by this unequal distribution of surveillance resources and to make recommendations about where surveillance should be increased in response to predicted disease emergence risk. We can draw two conclusions from this work. First, reporting effort significantly influences where we observe EID events. This implies that EID events of a similar scale are occurring, unobserved, in locations with weaker disease reporting infrastructure. Second, reporting infrastructure is stronger in developed countries, in northern latitudes, whereas wildlife zoonoses more commonly emerge in lower latitudes, in countries with weaker reporting effort. The implications are that our resources to rapidly identify novel EIDs poorly match their likely occurrence, and that this can be remedied by improving infrastructure in EID hotspot developing countries to identify pathogens spilling over from wildlife into people. It is important to note that our analysis, while suggestive, is preliminary. Reporting effort is likely more collinear with population density than a country- level measure can show, and the Journal of Infectious Diseases may not be the most accurate measure of where infectious disease reporting is strongest. Con- structing higher-fidelity maps of infectious disease reporting effort would allow us to better correct for the lens through which we view disease emergence and identify areas with the greatest need for increased surveillance. Furthermore, the EID hotspot maps are relevant only at large spatial scales. New approaches are needed to identify where, within a region, country, or landscape, the highest risk of a new disease originating exists. One approach is to conduct targeted surveil- lance efforts at specific wildlifeâhuman interfaces such as people living in remote villages close to forests in EID hotspots, or people engaged in hunting bushmeat, producing livestock, selling live animals in markets, or butchering them in abat- toirs or restaurants. Better analysis of the spatial distribution and relative risk of these interfaces is likely to be a productive research line. Finally, with the grow- ing availability of âbig data,â increasing ease by which it can be manipulated and analyzed, and new models that predict future trends in the underlying drivers of EIDs, hotspot models will become more rigorous, accurate, and based on con- crete hypotheses about biological mechanisms. How Many Unknown Pathogens Are There? The perfect pandemic prevention program would prevent spillover of patho- gens from wildlife to human hosts before they have the opportunity of infecting
APPENDIX A 187 people, amplifying their transmission, and becoming pandemic. This approach is theoretically possible. If we target surveillance of wildlife to EID hotspot coun- tries and conduct pathogen discovery in these species, we can identify pathogens with pandemic potential before they emerge and target prevention efforts to block their spillover. This is the basis for a number of new programs, includ- ing the USAID Emerging Pandemic Threat (EPT) program (Morse et al., 2012) and research programs that target pathogen discovery in bats and other zoonotic disease reservoirs (Drexler et al., 2012; Marsh et al., 2012; Wacharapluesadee et al., 2013). However, even when we have narrowed down interfaces and locales of inter- est, two significant challenges remain. Firstly, the diversity of unknown pathogens may be so high that it is not cost effective to identify them all. Indeed, until recently there was no systematic attempt to predict the unknown viral diversity in any single species, let alone all wildlife. Using samples collected and tested through the USAID EPT PREDICT program, we have recently published the first attempt at a strategy to estimate unknown viral diversity. We did this using incidence-based species richness estimators, which have their origin in the âmark- recaptureâ modeling approach used by conservation biologists to estimate the density of rare animals in a patch of land. In this method, animals are captured, tagged, and released, and the number of recaptures of tagged individuals relative to the number of untagged individuals gives a way to statistically predict the total number of individuals in a region. For pathogen discovery, we repeatedly sampled a large population of Pteropus giganteus, a bat species known to carry zoonotic viruses, collecting high-quality samples from around 2,000 unique indi- vidual bats. We then used degenerate viral family-level primers (12,793 separate consensus PCR assays) to discover 55 viruses from nine viral families known to harbor zoonoses (Anthony et al., 2013). We then used statistical approaches to estimate the total viral richness of these nine families in this single species. Our analysis suggests that this bat species harbors 58 viruses (i.e., 3 not yet discov- ered) in these viral families, and if this is extrapolated simplistically to all 5,517 known mammal species, we estimate that there are at least 320,000 mammalian viruses awaiting discovery in these nine viral families. This is a large number, but using the PREDICT program costs of field and lab work, we estimate the cost to uncover 100 percent of virodiversity in this critical group of wildlife reservoirs to be $6.8 billion, and to uncover 85 percent of virodiversity to be only $1.4 billion, considering the exponentially diminishing returns of continued sampling. The latter figure is less than the cost of a single SARS-scale pandemic and, if spread over a decade, a small portion of current global pandemic prevention spending. Which Wildlife Species Harbor the Next Pandemic Pathogen? The second challenge to wildlife pathogen discovery as a pandemic pre- vention strategy is knowing which wildlife are the highest-risk reservoirs (i.e., which species to sample so that we can maximize the discovery of pathogens
188 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS with zoonotic, and pandemic, potential). Species differ in the composition of their viral diversity and in the propensity of those pathogens to infect people, but the genetic, behavioral, and ecological rules that underpin these relationships are poorly understood (Bogich et al., 2012b). A recent analysis of the literature found that sampling effort, IUCN threat status, and population genetic structure of bat species were the best predictors of how many viral species they harbored, inde- pendent of their phylogenetic relationships (Turmelle and Olival, 2009). Among mammal groups, rodents and bats host a particularly large number of zoonotic pathogens: Rodents have a larger diversity, while bats host more per species (Luis et al., 2013). Within bat and rodent species, those with greater sympatry (range overlap) with other related species host more viral diversity, and bats with smaller litters, greater longevity, and more litters per year tended to host more zoonoses. These are tantalizing glimpses of ecological and evolutionary patterns that likely drive viral speciation and zoonotic risk, and may ultimately inform which species we target for viral discovery. However, there is much more to learn. For example, a logical assumption is that viruses are more able to infect more closely related species, due to the sharing of host cell receptors, for example. Thus, mammals are the source of the majority of zoonotic EIDs (Jones et al., 2008; Taylor et al., 2001) and across all mammal-virus associations, more closely related mammals are more likely to share virus species (Bogich et al., 2012b). However, when two unrelated species have extensive, intimate contact over long periods of time (e.g., humans and domesticated mammals), does this phylogenetic rule still hold? If we continue to expand the wildlife trade, bringing more diverse animals from differ- ent regions into close contact with people, will we see pathogens emerging that would normally have difficulty successfully infecting humans? Can We Predict the Pandemic Potential of a Newly Discovered Pathogen? With targeted improvements in public health infrastructure and surveil- lance for pathogen discovery, we can increase our odds of catching a zoonotic outbreak in its nascence or discovering novel pathogens of pandemic potential. But will we be able to identify which ones, out of the hundreds of thousands of new species of virus to be discovered in wildlife, will be able to infect humans? With most of these potential zoonoses being identified by only a short RNA or DNA sequence, is there a logical strategy to identify their potential pandemicity? Identifying which novel pathogens in a wildlife species are most likely able to infect, replicate in, cause cycles of human-to-human infection, and then amplify into pandemics remains one of the biggest challenges to pandemic prevention. Morse et al. (2012) reviewed some of the known factors that affect whether a particular virus can infect a species and what gaps remain. In some pathogens, receptor specificity and other biological characteristics may be used to predict host range and potential pathogenicity to humans. However, animal models, hu- man cell cultures, and similar methods cannot empirically validate a pathogenâs
APPENDIX A 189 capacity to infect humans. Some characteristics that may yield improvements in our predictive ability include the effects of host relatedness, relatedness of a virus to known human viruses, host range and evolutionary capacity, and predictive capacity of virulence in humans (some pathogens can infect humans but cause no disease, whereas others cause severe illness) (Morse et al., 2012). As we work towards a better understanding of these factors, we can use a few simple heuristics to prioritize certain pathogens. Certainly, if a pathogen exists at a zoonotic interface, and if there has been documented human infection, it should be prioritized. Pathogens that cause small chains of human-to-human infection with a basic reproductive number (R0) approaching or higher than 1 should be considered âprime epidemics in waiting,â as small evolutionary changes could boost their transmissibility and enable them to cause epidemics. In fact, though none of the models outlined above can tell us exactly how dangerous a pathogen is, they all contribute valuable information to a risk assess- ment. Whether a pathogen exists at an interface of interest, how closely related it is to known human pathogens, how closely related to humans its reservoir host is, and various viral traits all convey information about a particular pathogen. Future work may involve testing the zoonotic potential of wildlife pathogens by sequencing receptor binding domains, producing pseudo-type viruses with these proteins expressed, and conducting binding assays, in vitro culture assays, and ultimately animal infections with transgenic animals that express human cell sur- face receptors. This work has already shown the capacity to identify high-priority potential zoonoses for SARS-like viruses in bats, which bind to human, civet, and bat ACE2 (Ge et al., 2013). Can We Predict How, and to Where, a New EID Will Spread? The emergence of triple reassortant A/H1N1 influenza in 2009 highlighted how rapidly diseases can spread once they have achieved capacity for effective human-to-human transmission. Targeting these diseases may be effective if we can accurately predict their likely pattern of spread out of a region and strategi- cally allocate resources to respond. Analyses of travel and trade data have shown that predicting spread is relatively straightforward and can provide accurate estimates of spread and case numbers when applied to prior outbreaks (e.g., of SARS (Hufnagel et al., 2004) and A/H1N1 influenza (Hosseini et al., 2010)). This approach has been used to analyze recent historical spread of vectors through shipping trade, and their likely routes of spread via air travel (Tatem, 2009; Tatem et al., 2006a,b). It has also been used to predict the spread of ongoing emergence events such as the MERS-CoV outbreak in Saudi Arabia (Khan et al., 2010). It has particular relevance in zoonotic disease spread when patterns of wildlife mi- gration and trade are implicated, and where policy can be rapidly set to prevent importation. This approach has been used to examine the likely cause of past spreading events for A/H5N1 influenza and to predict and set policy for its likely
190 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS route of introduction to the New World (Kilpatrick et al., 2006a). Finally, it has been used effectively in Hawaii and the Galapagos Islands to allocate resources to reduce the risk of West Nile virus introduction via the most likely pathway of mosquitoes transported via air travel (Kilpatrick, 2011; Kilpatrick et al., 2004, 2006b). As in all predictive models, their rigor improves as the quality of data on travel and trade pathways and volumes, on biological characteristics of patho- gen and host, and on the human contact networks that allow transmission also improves. For example, the capacity and willingness of countries to identify and report outbreaks early are critical to make accurate predictions about spread, once a pandemic has begun. Analyses of the spread of the 2009 H1N1 influenza showed that two key factors influenced the pandemicâs arrival timeâa countryâs global accessibility through air travel, and the percentage of GDP per capita spent on health care (a proxy for testing and reporting capacity) (Hosseini et al., 2010). Again, gaps in this approach remain, including the need for a better understanding of the role of intra-country human movement in disease spread. Newly available datasets on road infrastructure, migration, and human network connectivity will increasingly illuminate this area. Can We Eventually Stop Pandemics from Emerging? The new approaches described above to identifying novel pathogens in emerging disease hotspots, and predicting their pandemic potential and likely spread, have likely improved our global pandemic preparedness. But what prog- ress has been made in using this approach to prevent pandemics? One significant shift is in the way pandemic prevention programs are funded and managed. Traditionally, outbreak threats were dealt with by state and national agencies, the World Health Organization, and field laboratory networks funded through these programs. The emergence of H5N1 influenza via small-scale outbreaks, which suggested chronic persistence in backyard poultry farms, led to calls for a âsystems approachâ to the pandemic prevention (Bogich et al., 2012a), and a cross-sectorial âOne Healthâ collaboration of animal health, public health, and environmental agencies (FAO et al., 2008; Karesh, 2009; Zinsstag et al., 2011). International development agencies, which had been trending towards specialized programs to target specific infectious diseases, are now actively involved in this systems approach to pandemic prevention. This involves funding for crucial in- frastructure investments required for pandemic prevention, and a specific focus on collaborative One Health programs (Bogich et al., 2012a). With most EID events occurring in regions that are under-resourced in public health capacities, disease- based programs for AIDS, malaria, TB, and polio do not address the underlying flaws in public health systems that predispose locations to outbreaks of emerging infectious diseases (Standley and Bogich, 2013). Standley and Bogich (2013) propose an âecohealthâ approach, addressing destructive land use change and biodiversity loss in places like China, Brazil, and India. This approach defines how we can deal with pandemics as distinct
APPENDIX A 191 from dealing with hurricanes or earthquakes: by identifying and mitigating the underlying causes, particularly anthropogenic activities that promote pathogen spillover, amplification, and spread. Strategies include programs that educate and promote alternatives to high pandemic risk behavior like the trading, butchering, and consumption of wild animals, or the comingling of livestock and wildlife on farms. They also include more fundamental approaches that address large-scale anthropogenic changes. For example, 43 percent of past EID events are attribut- able to land use change and agricultural changes, including extractive industries (timber/logging, oil and gas, mining, and plantations). The economic impact of EIDs from land use change is estimated to be $10â40 billion over the next 10 years, which could be considered a potential liability to extractive industries. Industrialized mining and plantation operations in EID hotspot countries are likely to be on the front line of disease outbreaks, and are often under pressure to improve their environmental impacts. Programs that better quantify the risk of novel pathogens to these industries, and the economic damages they might entail, may become valuable in mitigating their impact on global health, conservation, and the environment (Murray and Daszak, 2013). The two-fold approach of treating emerging pandemics as targets for interna- tional development programs and as byproducts of economic activity is relatively new and suggests that long-term solutions to their emergence can be found. A future without pandemics may be possible, but only with the very best interdisci- plinary science, ambitious approaches to risk prediction, and bold strategies taken by industry and development agencies to ensure against them. References Anthony, S. J., J. H. Epstein, K. A. Murray, I. Navarrete-Macias, C. M. Zambrana-Torrelio, A. Solovyov, R. Ojeda-Flores, N. C. Arrigo, A. Islam, S. Ali Khan, P. Hosseini, T. L. Bogich, K. J. Olival, M. D. Sanchez-Leon, W. B. Karesh, T. Goldstein, S. P. Luby, S. S. Morse, J. A. Mazet, P. Daszak, and W. I. Lipkin. 2013. A strategy to estimate unknown viral diversity in mammals. Mbio 4(5). Bogich, T. L., R. Chunara, D. Scales, E. Chan, L. C. Pinheiro, A. A. Chmura, D. Carroll, P. Daszak, and J. S. Brownstein. 2012a. Preventing pandemics via international development: A systems approach. PLoS Medicine 9(12). Bogich, T. L., K. J. Olival, P. R. Hosseini, E. Loh, S. Funk, I. L. Brito, J. H. Epstein, J. S. Brownstein, D. O. Joly, M. A. Levy, K. E. Jones, S. S. Morse, A. A. Aguirre, W. B. Karesh, J. A. K. Mazet, and P. Daszak. 2012b. Using mathematical models in a unified approach to predicting the next emerging infectious disease. In New directions in conservation medicine, edited by A. A. Aguirre, R. S. Ostfeld, and P. Daszak. New York: Oxford University Press. Pp. 607-618. Drexler, J. F., V. M. Corman, M. A. Müller, G. D. Maganga, P. Vallo, T. Binger, F. Gloza-Rausch, A. Rasche, S. Yordanov, and A. Seebens. 2012. Bats host major mammalian paramyxoviruses. Nature Communications 3:796. FAO, OIE, WHO, UNSIC, UNCF, and WHO. 2008. Contributing to One World, One Health: A strategic framework for reducing risks of infectious diseases at the animal-human-ecosystems interface. Rome: Food and Agriculture Organization; World Organisation for Animal Health; World Health Organization; United Nations System Influenza Coordinator; United Nations Childrenâs Fund; World Bank.
192 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Ge, X.-Y., J.-L. Li, X.-L. Yang, A. A. Chmura, G. Zhu, J. H. Epstein, J. K. Mazet, B. Hu, W. Zhang, C. Peng, Y.-J. Zhang, C.-M. Luo, B. Tan, N. Wang, Y. Zhu, G. Crameri, S.-Y. Zhang, L.-F. Wang, P. Daszak, and Z.-L. Shi. 2013. Isolation and characterization of a bat SARS-like Coronavirus that uses the ACE2 receptor. Nature 503:535-538. Hosseini, P., S. H. Sokolow, K. J. Vandegrift, A. M. Kilpatrick, and P. Daszak. 2010. Predictive power of air travel and socio-economic data for early pandemic spread. PLoS ONE 5(9):e12763. Hufnagel, L., D. Brockmann, and T. Geisel. 2004. Forecast and control of epidemics in a global- ized world. Proceedings of the National Academy of Sciences of the United States of America 101(42):15124-15129. IOM (Institute of Medicine). 1992. Emerging infections: Microbial threats to health in the United States. Washington, DC: National Academy Press. IOM. 2003. Microbial threats to health: Emergence, detection, and response. Washington, DC: The National Academies Press. Jones, K. E., N. Patel, M. Levy, A. Storeygard, D. Balk, J. L. Gittleman, and P. Daszak. 2008. Global trends in emerging infectious diseases. Nature 451:990-994. Karesh, W. B. 2009. One worldâOne health. Clinical Medicine 9(3):259-260. Khan, K., Z. A. Memish, A. Chabbra, J. Liauw, W. Hu, D. A. Janes, J. Sears, J. Arino, M. Macdonald, F. Calderon, P. Raposo, C. Heidebrecht, J. Wang, A. Chan, J. Brownstein, and M. Gardam. 2010. Global public health implications of a mass gathering in Mecca, Saudi Arabia, during the midst of an influenza pandemic. Journal of Travel Medicine 17(2):75-81. Kilpatrick, A. M. 2011. Globalization, land use, and the invasion of West Nile virus. Science 334(6054):323-327. Kilpatrick, A. M., Y. Gluzberg, J. Burgett, and P. Daszak. 2004. A quantitative risk assessment of the pathways by which West Nile virus could reach Hawaii. Ecohealth 1:205-209. Kilpatrick, A. M., A. A. Chmura, D. W. Gibbons, R. C. Fleischer, P. P. Marra, and P. Daszak. 2006a. Predicting the global spread of H5N1 avian influenza. Proceedings of the National Academy of Sciences of the United States of America 103:19368-19373. Kilpatrick, A. M., P. Daszak, S. J. Goodman, H. Rogg, L. D. Kramer, V. Cedeno, and A. A. Cunningham. 2006b. Predicting pathogen introduction: West Nile virus spread to Galapagos. Conservation Biology 20(4):1224-1231. Luis, A. D., D. T. S. Hayman, T. J. OâShea, P. M. Cryan, A. T. Gilbert, J. R. C. Pulliam, J. N. Mills, M. E. Timonin, C. K. R. Willis, A. A. Cunningham, A. R. Fooks, C. E. Rupprecht, J. L. N. Wood, and C. T. Webb. 2013. A comparison of bats and rodents as reservoirs of zoonotic viruses: Are bats special? Proceedings of the Royal Society B-Biological Sciences 280(1756). Marsh, G. A., C. de Jong, J. A. Barr, M. Tachedjian, C. Smith, D. Middleton, M. Yu, S. Todd, A. J. Foord, V. Haring, J. Payne, R. Robinson, I. Broz, G. Crameri, H. E. Field, and L. F. Wang. 2012. Cedar virus: A novel henipavirus isolated from Australian bats. PLoS Pathogens 8(8). Morse, S. S., J. A. K. Mazet, M. Woolhouse, C. R. Parrish, D. Carroll, W. B. Karesh, C. Zambrana- Torrelio, W. I. Lipkin, and P. Daszak. 2012. Prediction and prevention of the next pandemic zoonosis. Lancet 380:1956-1965. Murray, K. A., and P. Daszak. 2013. Human ecology in pathogenic landscapes: Two hypotheses on how land use change drives viral emergence. Current Opinion in Virology 3(1):79-83. Standley, C. J., and T. L. Bogich. 2013. International development, emerging diseases, and eco-health. Ecohealth 10:1-3. Tatem, A. J. 2009. The worldwide airline network and the dispersal of exotic species: 2007-2010. Ecography 34:94-102. Tatem, A. J., S. I. Hay, and D. J. Rogers. 2006a. Global traffic and disease vector dispersal. Proceed- ings of the National Academy of Sciences of the United States of America 103(16):6242-6247. Tatem, A. J., D. J. Rogers, and S. I. Hay. 2006b. Estimating the malaria risk of African mosquito movement by air travel. Malaria Journal 5. Taylor, L. H., S. M. Latham, and M. E. J. Woolhouse. 2001. Risk factors for human disease emer- gence. Philosophical Transactions of the Royal Society B-Biological Sciences 356:983-989.
APPENDIX A 193 Turmelle, A. S., and K. J. Olival. 2009. Correlates of viral richness in bats (Order Chiroptera). Eco- health 6(4):522-539. Wacharapluesadee, S., C. Sintunawa, T. Kaewpom, K. Khongnomnan, K. J. Olival, J. H. Epstein, A. Rodpan, P. Sangsri, N. Intarut, A. Chindamporn, K. Suksawa, and T. Hemachudha. 2013. Iden- tification of group C betacoronavirus from bat guano fertilizer, Thailand. Emerging Infectious Diseases [Internet](August 2013). Zinsstag, J., E. Schelling, D. Waltner-Toews, and M. Tanner. 2011. From âone medicineâ to âone healthâ and systemic approaches to health and well-being. Preventive Veterinary Medicine 101(3-4):148-156. A8 MISCONCEPTIONS AND EMERGING PATHOGENS: WHAT CAN MATHEMATICAL MODELS TELL US? Andrew Dobson25 The last 25 years have seen a renaissance in the use of mathematical models in epidemiology; much of this is largely due to the influence of An- derson and May and their colleagues (Anderson and May, 1992; Grenfell and Dobson, 1994, 1995). The transformation came about as the models they developed were based upon empirical assumptions. This allowed the whole discipline to move from an overt fascination with mathematical elegance, to embrace data and become the pragmatic powerhouse that is at the center of quantitative insight to any modern epidemiological problem. At first glance, this creates problems for the use of these models in studies of emerging dis- eases, as almost by definition, there will be no data prior to emergence. None- theless, all of the recent major studies of disease emergence have quickly led to the almost obligatory use of mathematical models in infectious disease biology. A nice index of this was the chance remark by the editor of one major journal during a recent influenza outbreak, âHalf the world is wor- ried about this new pathogenâwhile weâre facing an epidemic of submitted papers, all claiming to have produced the definitive predictive model for it!â In this short overview, I will take a brief personal and idiosyncratic review of the key ways in which mathematical models have been used, misused, or could potentially be used to provide insights into the dynamics of emerging pathogens. I will offer no specific recommendations or recipes for the âbest wayâ to use models to understand pathogen emergence. This is partly because different model structures will provide different insights to different pathogens; moreover, each new emergence usually leads to the development of new mathematical tricks, techniques, and approaches that 25ââDepartment of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544.
194 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS provide powerful new tools for the current crisis and often retrospective insights into older emergences. Dynamics of Initial Cross-Over A huge number of pathogens are circulating in all free-living species of animals and plants. One of the most profound testimonies to the shortsighted- ness of scientific exploration is that we know neither how many other species share the planet with us, nor how many are pathogens or parasites of the more apparent and better classified free-living species (Dobson et al., 2008). The most conservative estimate is that 50 percent of species are parasitic, but it could be significantly higher, potentially larger than 90 percent. Although a huge num- ber of pathogens could potentially colonize humans (or domestic livestock and crops), only a relatively small proportion seem to have done so. Although search- ing for âthe next pandemic virusâ has achieved the momentum of a well-oiled government job-creation scheme (a curious European phenomenon that may be unfamiliar to USA readers!), I suspect that a large proportion of pathogens that might jump the species barrier to humans may already have attempted this leap and have failed the test. The simple logic here is humans have explored most of the terrestrial parts of the planet and exposed themselves to a multitude of insect bites, scratches by plants, and samplings of local cuisine; this suggests there are very few pathogens that one of us has not yet been exposed to. Yet it would be foolish to be apathetic. A pathogen that failed to establish in the past might get a second chance in the future if the host it contacts has different susceptibility and moves further or contacts more people while infectious. Conservatively, there are likely to be a couple hundred other pathogens out there that could create a new pandemic, but our attempts to locate them on virological fishing expeditions do a poor job of differentiating between exotic minnows and efficient pike. So my principle focus here will be to assemble the known structures of a mathematical framework that needs to be applied if we are to quantify zoonotic disease emer- gence of a pathogen and our immediate response to it. A recent review explains the mathematical logic of epidemic dynamics at the humanâanimal interface (Lloyd-Smith et al., 2009). The classification of epi- demic potential is based on the relative magnitude of R0, the basic reproductive number of the pathogen, or more formally the number of secondary cases that an initial infected individual creates in a wholly susceptible population, before âcase zeroâ either recovers or succumbs to infection. Lloyd-Smith et al. base their classification on an earlier review by Wolfe et al. (2007) that classified pathogens along a five-point spectrum with those that are exclusive to wildlife as type I, while those that are exclusive to humans as type V. Most zoonotic pathogens can be arranged along the spectrum from II to IV based upon their affinity for sus- tained transmission in the novel human hosts and the associated pathology. Type II pathogens are those that can cause primary infections in humans, but humans
APPENDIX A 195 are unable to transmit the pathogen on to other humans; classic examples would be Brucella abortus and West Nile virus. Type III pathogens occur when the primary infections are able to infect a number of secondary hosts, but these stut- tering chains of transmission quickly fade out. Classic examples here would be Nipah and Hendra virus that are endemic in Pteropid fruit bats, and humans either acquire infection either directly or indirectly from livestock (pigs and horse,s re- spectively) (Plowright et al., 2011; Pulliam et al., 2011). The most worrying type of zoonotic disease is that in type IV, where the primary infection can give rise to self-sustaining chains of infection. Classic examples here are plague (Yersinia pestis) and pandemic influenza. Lloyd-Smith et al. (2009) point out that each level of classification corresponds to a different range of values of R0; thus type IV (and V) will have R0 > 1, type III will have R0 < 1, and type II, R0 = 0. All of this has made estimation of R0 a central part of any emerging disease outbreak. Before briefly considering ways in which R0 might be measured, it is worth noting that the vast majority of pathogens will be type I. There will then be a smaller number of type II, even less of type III, and so on. This does suggest a future and potentially more focused line of inquiry for virus hunters. Is there any underlying taxonomic signal in the R0 values for zoonotic pathogens? Are there some families of viruses with high proportions of type III and type IV zoonotic pathogens, and are the relative proportions of these different types of pathogen similar or very different in different taxa of viruses, bacteria, fungi, and so on? Putting together a database to address these questions might well provide a more quantitative framework of where to look for potential future pandemic pathogens. It should also provide important background information on how hosts infected with similar pathogens have been diagnosed and treated. All of this assumes a significant degree of phylogenetic inertia in the ways in which the pathology of taxonomically similar pathogens are expressed. R0 or Not? Most mathematical models for emergent disease start with the development of an expression for R0, the basic reproductive number of the disease. It played a central role in the development of the initial response to the SARS epidemic (Dye and Gay, 2003; Lipsitch et al., 2003; McLean et al., 2005). There is a sig- nificant volume of mathematical literature on R0 and many examples of the use of this concept (Diekmann and Heesterbeek, 2000; Diekmann et al., 1990). Most mathematical epidemiologists were delighted that R0 played a central role in the movie Pandemic, and it was a testimony to the elegance of the concept that it resonated well with audiences. There are two approaches to estimating R0: one is classified by the publica- tion feeding frenzy that follows the appearance of data describing the course of an epidemic outbreak; a range of increasingly sophisticated statistical methods are used here, and they increasingly prove central to guiding the public health
196 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS response. The second approach involves deriving algebraic expressions for R 0 in the absence of any epidemiological data. These âmathodsâ are potentially highly informative for identifying the weakest links in the transmission cycle and then determining methods of control that can break these weak links in the chain of transmission. The simple threshold condition (R0 > 1) is useful for defining the absolute conditions for whether a pathogen will establish, while also pointing towards the level of control needed to contain an outbreak. Nonetheless, there are some important clarifications about the magnitude of R0 that strongly determine what happens once an outbreak begins to take off. The first of these insights deals with the relationship between the magnitude of R0 and persistence time of the epidemic. If R0 is significantly larger than 1, anywhere above 4, and host popu- lation is relatively restricted, then the epidemic may rise quickly, but will con- comitantly run out of new susceptible host, and the epidemic will quickly burn itself out as chains of transmission are broken. This consistently happens with measles in small towns and villages (Keeling and Grenfell, 1997) and with Ebola and Nipah virus outbreaks. In contrast, when R0 is a little bit larger than unity, but less than 2, then outbreaks can persist for longer. An interesting example of this is illustrated in theoretical work recreating distemper outbreaks in different carnivore species in Serengeti National Park (Craft et al., 2008). Species such as jackals with large populations exhibit sharp epidemics of short durations; in contrast, outbreaks persist for much longer in hosts with lower abundance, particularly if they are split into relatively isolated social groups (such as lions). The pathogen then causes an outbreak in each social group, but then more slowly jumps between social groups, or between species, and despite having a lower R 0, persists for much longer. The first (and most eloquent) demonstration of this is found in the work of Swinton and colleagues on the outbreaks of distemper in seal population in the North Sea (Swinton, 1998; Swinton et al., 1998). Theoretical models for persistence sharply illustrated that persistence time increases hugely as populations are subdivided into social groups whose rate of contact is always lower than rates of contact within group (Figure A8-1). This effectively lowers R0 for populations of identical size, but hugely increases pathogen persistence. Anderson and May proposed that this sort of mechanism was likely central to the initial emergence of HIV with the virus entering the human population in small, weakly coupled villages in Africa, none of which could support a sustained outbreak, but each of which was weakly coupled to an unexposed village that could keep the chain of infection intact (Anderson and May, 1986). Eventually the pathogen was passed into hosts who had contact with much larger and more actively interacting community, and the epidemic was detected in the United States and other Western countries. The epidemic of AIDs that emerged in the 1980s in major Western cities contrasts sharply with the previous half century of HIV in rural Africa, where dynamics were most likely characterized by low
APPENDIX A 197 FIGURE A8-1â The expected persistence time of a pathogen that infects its hosts for 2 weeks and is infectious for the second of those weeks in populations of different sizes. The four different graphs compare populations that are divided into social groups of different sizes (left GS = 4; right GS = 10); the different lines compare persistence in a well-mixed population with no groups with ones where the rate of between-group contacts is 5 percent, 20 percent, 50 percent, and 90 percent of the rate of within-group contacts. The upper two graphs are for groups arranged in a linear sequence (along a coastline) and that only contact the groups on either side of them; in the bottom two figures the groups tessellate a plane, so each group is in contact with at least four other groups SOURCES: McCallum and Dobson, 2006; Swinton, 1998. Reprinted with permission of Cambridge University Press. R0 slowly moving the pathogen through a sequence of weakly connected and relatively isolated villages somewhere in the forests of central Africa. A related effect that operates in a similar fashion has been proposed by Pulliam and colleagues (Pulliam et al., 2007, 2011). They observe that many immunity-inducing pathogens may facilitate emergence by initially creating an outbreak that creates an intermediate level of herd immunity in the novel host population that reduces the pool of susceptible hosts to a level where a second crossing over of the pathogen from the reservoir creates an outbreak with an R0 closer to unity, and thus one that will persist for longer and be more able to spread
198 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS to the new population. They suggest this was a dominant factor in the emergence of Nipah virus in Malaysia and show that many other pathogens have epidemic demography that would also allow them to establish in this fashion (Pulliam et al., 2007). When they plot the transmission and virulence parameters of a number of pathogens into a graph that determines emergence potential they find that many emergent pathogens have these characteristics (Figure A8-2). FIGURE A8-2â(A) Deterministic prediction of the parameter ranges where epidemic enhancement may be observed. The range of parameter values (grey) for a population size of N = 50,000 and the initial condition (S0, I0) = (N â 1, 1) which demonstrate the behavior of an initial epidemic which dies out (there exists t > 0 such that It < 1) followed by persistence upon reintroduction (I* > 1), depending on the level of population turnover between pathogen extinction and reintroduction. R0 is the basic reproductive number of the pathogen in a naïve host population; Ï is the duration of infectiousness relative to the average duration of immunity. Stars represent parameter values taken from the literature for a variety of common and emerging infectious diseases. Note that the x-axis is shown on a log scale. (B) Enhancement of epidemic duration for diseases in human populations. Epidemic duration in a population of N = 50,000 individuals for a variety of human patho- gens as a function of population immunity at introduction. Solid lines show the median duration in disease generations for 1,000 simulation runs at each level of initial population immunity; dashed lines show quartiles. Each pathogen shows some level of enhancement of epidemic duration with increased immunity except pertussis. Enhancement of epidemic size is not observed for these pathogens for N = 50,000. SOURCE: Pulliam et al., 2007.
APPENDIX A 199 Incubation and Infectious Period Two parameters of any model for R0 are central to our ability to control the initial emergence of a novel pathogen. Ironically we tend to worry more about the transmissibility of the pathogen, which is always the hardest thing to estimate, than we do about these other two equally vital parameters: incubation and infectious period. Key insights into the importance of these parameters come from work comparing SARS, influenza, and HIV (Fraser et al., 2004). This work shows that although R0 is fundamental in determining the level of intervention, even pathogens with low R0 can cause huge epidemics if they have a long silent incubation period during which transmission occurs without any apparent symp- toms. The classic example of this is HIV, which has caused the largest epidemic in human history since the plague epidemics of thirteenth-century Europe. This contrasts sharply with SARS where symptoms of infection appear almost simul- taneously to ability to infectiousness; this made it relatively easy to contain SARs through a combination of isolation of infectious individuals and contact tracing (Figure A8-3). This problem with long âsilentâ incubation periods is particularly worrying from the current vogue for virus hunting. I think that if the approaches currently used to detect emerging pathogens were retrospectively applied to the HIV virus, they would dismiss it as a mild and innocuous pathogen of limited concern. This is primarily because the incubation period of the virus is as long as the life ex- pectancy of most primate species used in laboratory research (Anderson, 1991). If injected into humans, we would only see an initial rise in virus abundance that was quickly knocked back by the hostâs immune system. Although an astute clinician might detect the virusâs rapid mutation rate, I suspect that the absence of any symptoms in the first 5 to 10 years postinfection would lead to the virus being dismissed as a hazard. From a simple mathematical perspective this makes me much more concerned about viruses with long silent incubation periods (and the opportunities to infect thousands of people) than it does about highly virulent viruses whose violent symptoms make for powerful movies, but also for ready detection, isolation, and the development of a rapid response. There Be Dragons! Maps are powerful tools that have multiple uses in biology and epidemiol- ogy. For example, they have been widely used in conservation biology to identify areas of unusually high biological diversity in areas with relatively low land val- ues, or to identify areas with unusually high extinction rates (Bibby et al., 1992; Conroy and Noon, 1996; Dobson et al., 1997). Conservation biology is essentially a complementary discipline to infectious disease biology; one discipline seeks to drive organisms to extinction, the other hopes to bring back rare species from the brink. Both disciplines have benefitted from the huge rise of geographical information systems (GIS) that allows maps to be readily created from detailed
200 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A8-3â (A) Parameter estimates. Plausible ranges for the key parameters R0 and θ for four viral infections of public concern are shown as shaded regions. The size of the shaded area reflects the uncertainties in the parameter estimates. The areas are color-coded to match the assumed variance values for β(Ï) and S(Ï) of Fig. 1 in Fraser et al. (2004) appropriate for each disease, for reasons that are apparent in Fig. 3 in Fraser et al. (2004). (B) Criteria for outbreak control. Each curve represents a different scenario, consisting of a combination of interventions and a choice of parameters. For each scenario, if a given in- fectious agent is below the R0âθ curve, the outbreak is always controlled eventually. Above the curve, additional control measures (e.g., movement restrictions) would be required to control spread. Black lines correspond to isolating symptomatic individuals only. Colored lines correspond to the addition of immediate tracing and quarantining of all contacts of isolated symptomatic individuals. The black (isolation only) line is independent of distri- butional assumptions made (low or high variance), whereas the colored (isolation + contact tracing) lines match the variance assumptions made in Fig. 1 in Fraser et al. (2004) (red = high variance; blue = low variance). The efficacy of isolation of symptomatic individuals is 100% in Panel B-1, 90% in Panel B-2, and 75% in Panel B-3. Contact tracing and isolation is always assumed 100% effective in the scenarios in which it is implemented (colored lines). Curves are calculated using a mathematical model of outbreak spread incorporating quarantining and contact tracing. SOURCE: Fraser et al., 2004.
APPENDIX A 201 geographical data and more sparse biological and epidemiological surveys. Maps also have a very distinguished history of use in epidemiology. This stretches from John Snowâs original map identifying the Broad Street pump as the most likely source of cholera in London, through the path-breaking work of Cliff, Haggett, and Ord (Cliff and Haggett, 1988; Cliff and Ord, 1981): their studies of the history of measles in Iceland, combined with the work of Bartlett (1957, 1960, 1966), paved the way for our current deep understanding of the dynamics of SIR pathogens (Anderson and May, 1983, 1985, 1990; Bjornstad et al., 2002; Ferrari et al., 2008; Grenfell and Anderson, 1985; Grenfell et al., 2001). The great power of maps is that politicians and government decision makers have an instinctive understanding of maps (they use them to plan their vacations and political or military campaigns); in contrast, they seem much more wary of mathematical models, or even graphs (although they have teams of policy mak- ers that happily abuse these models to plan economies and election campaigns!). The central problem with maps is apparent in some of the oldest maps; when there was limited or no information for an underexplored region there was a tendency to assume âThere be dragons.â This creates a historical precedent to use maps to identify the location of unknown scary monsters such as emerging pathogens or endangered species that have not been seen for some time. More disconcertingly, it means that we tend to forget that the data that underlie maps need to be verified and tested against an epidemiological model that provides some mechanism to explain the observed geographical patterns of incidence. Sometimes this is done (see Cliff and Haggett, 1988; Cliff and Ord, 1981). All too frequently the map is presented as a predictive tool, when all it is really presenting is a rather undigested mass of data detectable in the literature. When these maps are used as a predictive tool, many quantitative disease modelers become nervous. These fears could be allayed by some relatively simple math- ematical or statistical tests of the mapâs utility. The simplest approach to seeing if a map of emerging disease hot spots has any predictive value would be to take the first half of the historical data used to make the map and see if it has any predictive âskillâ in reproducing the latter half of the observed data (an approach widely used by climatologists). I suspect that these approaches would exhibit high skill in predicting the locations of large urban areas with major medical facilities that consistently detect new antibiotic resistant strains of bacteria. The approach will be less powerful at detecting areas for new unknown (viral) dis- eases (which may or not emerge), as this will reflect where people have decided to work on a hunch that this will be a hot spot, or because it was one in the past. I was very amused when a colleague told me that he had received funding for his emerging disease work because a hot spots map had identified his site as a likely hot spot. This was because he was one of the few people to have published a previous epidemiological survey from within this broad geographical location! Most quantitative ecologists working on emerging pathogens are exception- ally skeptical about maps produced that purport to predict hot spots of disease
202 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS emergence. This skepticism is well justified by the very limited ability of these maps to predict anything other than antibiotic-resistant strains of pathogens that tend to emerge in around Western cities where drugs are widely used and there are well-funded medical schools focused on detecting these strains. This form of prediction is essentially a self-fulfilling fantasy! A Role for Climate? There is increasing interest in the role that climate change will play in the emergence of new pathogens. At the risk of being hypocritical, I am going to use a couple of maps to shape the discussion about how climate change will interact with other aspects of global change to affect pathogens. Here it is important to explicitly acknowledge that climate change is only one component of global change. While all sane scientists not in the deep pockets of oil industry now acknowledge that anthropologically driven climate change is a real effect that is increasingly influencing the Earthâs climate, predicting how this will influence patterns of infectious disease dynamics and outbreaks will not be straightforward (Rodó et al., 2013), not least the influence of climate change may be masked by other aspects of global change, particularly in the parts of the world where most people are going to be living over the next 100 years. The work of Jetz et al. (2007) on how geographic distributions of all the worldâs bird species are likely to change over the next 100 years is instructive here. Jetz et al. (2007) base their analysis on the Millennium Ecosystem Assess- ments of land use change under four different scenarios. The work is based on detailed forecasts for climate, land use change, human population growth, and agricultural expansion over the next 100 years (Alcamo et al., 2005; Reid et al., 2005). These scenarios were then applied to the current distribution data for each of the worldâs bird species. Birds were chosen as we have better data for birds than for any other taxa. The impact of land use change was applied to the geo- graphical range of each species, and this was used to quantify the proportional loss of habitat for all bird species under complementary drivers of anthropologi- cal land use change (agriculture and urbanization) and climate change. Figures A8-4 and A8-5 present map projections and latitudinal cross sections that result from these analyses. The results generalize for other taxa and so provide impor- tant implications for pathogens, as well as for their nonhuman reservoir species. The figure illustrates that although climate change will dominate the future of polar regions, the impacts of land use change will hugely mask any climate change signature in the tropics and temperate regions. As the vast majority of birds (and mammals, plants, insects, etc.) live in the tropics, then it is going to be very hard to detect a climate change signal when we try and predict any aspect of the future of these systems.
FIGURE A8-4â Geographic patterns and projected impact of environmental change. (A, B) Patterns of change in land cover due to land use and climate change by 2100. This represents the summed, current-day occurrence of qualifying species across a 0.5° grid. Patterns are given for the environmentally proactive âAdapting Mosaicâ scenario, and the environmentally reactive âOrder from Strengthâ scenario. Maps are in equal-area cylindrical projection. 203 SOURCE: Jetz et al., 2007.
204 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A8-5â Environmental change, avian biogeography, and loss in range size. Pro- jected latitudinal pattern in type of global environmental change, geographic range size, species richness, and the resulting loss in geographic range size (8,750 bird species, 1° bands of latitude). Climate (cyan, on top and semitransparent) and land use (red) changes between now and 2100 are evaluated for two scenarios: on the left, âAdapting Mosaicâ (A, C), and on the right, âOrder from Strengthâ (B, D). Top (A, B): Total area transformed (area plot, lighter color indicates overlap) and average (± SE) current geographic range size of species per latitudinal band (point and line plot); Bottom (C, D): Average pro- portional loss of range size (area plot, lighter color indicating overlap) and total number of bird species whose range currently overlaps at each latitudinal band (point and line plot). Whereas climate change leads to a significant net change of habitat in the polar and temperate regions, the small numbers of bird species that live there on average have very large geographic ranges. Thus, proportional contractions in range size there are much smaller than for the vast majority of bird species that live in the tropics and experience significant reductions in their smaller range sizes due to land use change. The outcome is many species with significant range reduction in the tropics and subtropics, because of the coincidence of habitat conversion with areas of high species richness. This is particularly the case in the environmentally reactive âOrder from Strengthâ scenario, where large areas of land are converted to agriculture. SOURCE: Jetz et al., 2007.
APPENDIX A 205 There are three insights that I want to make from these figures: 1. This does not mean that climate change is not important; it means we need to understand how climate interacts with other aspects of global change. 2. In particular, if we want to understand how climate change will affect disease dynamics then we should expand studies of disease dynamics in the Arctic as these systems have a much stronger climate signal and many less confounding effects. Work undertaken here is already providing important insights that will eventually help interpret what will eventually happen in the temperate and tropical regions. 3. Our biggest worry about emerging pathogens in the tropics will come from land use change modifying the natural habitats of wild reservoir spe- cies living in these regions and the increasingly large human population that interact with them. R0, Biodiversity, and Dilution Effects The principle scientific justification for virus hunting in the tropics is that these regions contain the highest levels of biological diversity and hence more species should equate with more undiscovered pathogens. This in turn has led to some dubious estimates of the number of undiscovered viral species that assume all host species harbor the same number of pathogen species (Anthony et al., 2013). The logic of this approach assumes some of the methodology (and none of the rigor) of previous attempts to estimate global insect diversity by taking the numbers of insects associated with a small number of host trees and multiplying these numbers up by the known number of tree species (Erwin, 1982; Gaston, 1991, 1994; Hodkinson, 1992). Future attempts to estimate viral diversity would benefit hugely from the adoption of the methodology employed in these earlier entomological studies. In particular it should also be realized that host population size, density, and spatial distribution will all play a crucial role in determining the diversity of microbial pathogens harbored by any host species, and it is highly likely that rare species will host lower pathogen diversity than more common species. Rare hosts and hosts of low abundance create significant challenges for pathogens who adapt to these constraints by either reducing their virulence, so as to reduce the chance the hosts die between encounters, or increase their efficiency of transmission to ensure it occurs on the rare occasions that hosts encounter each other. These constraints mean the pathogens of rare species with low abun- dance are most likely to be STDs; there is little chance for anything else to be transmitted or maintained in the host population (Altizer et al., 2003; Lockhart et al., 1996). Common hosts are likely to harbor a greater diversity of pathogens, particularly if they live in large social groups (Ryan et al., 2013). All of which
206 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS suggests that virus hunters who head for the tropics to look for undiscovered viruses in rare species are significantly, scientifically deluded. An alternative perspective on biodiversity considers the role it may play in buffering pathogen emergence and reducing the potential for the emergence of novel pathogens. Disease ecologists call these phenomena âthe dilution effectâ (Dobson et al., 2006; Hudson et al., 1995; Keesing et al., 2006; Schmidt and Ostfeld, 2001), and there is an intense debate about the role they play in buffer- ing disease outbreaks (Lafferty and Wood, 2013; Ostfeld and Keesing, 2013; Randolph et al., 2012). Dilution effects can only occur when a pathogen uses multiple species of hosts. When one or more of these host species is able to withstand infection with the pathogen, but fails to transmit it efficiently to other hosts, it effectively creates a dilution of transmission rates and slows the rate of epidemic spread. Dilution effects are likely to be most efficient for vector-borne diseases than for directly transmitted pathogens (although evidence does suggest they are important for directly transmitted pathogens with frequency-dependent transmission such as Hanta virus). Dilution effects are also likely to be stronger for mosquito-borne pathogens than for tick-borne pathogens, as the abundance of the vectors is independent of host abundance for mosquitoes, but not for ticks (Dobson, 2009). Ironically, the best studied example of the dilution effect comes from work on tick-transmitted Lyme disease (LoGuidice et al., 2003; Ostfeld and Keesing, 2000; Ostfeld and LoGiudice, 2003). From the perspective of pathogen emergence, we simply do not know whether these effects are strong enough to buffer rates of disease emergence. These is some correlative evidence that supports the case that they might be operating (Bonds et al., 2012; Ezenwa et al., 2006; Roche et al., 2012), and if this is the case then it presents a powerful argument for finding ways to conserve species diversity as agriculture and land use intensifies. Subsequent Evolution The emerging pathogens that cause me to lose most sleep are those caused by the evolution of resistance to the drugs and antibiotics we have used over the last 50 years (Cohen, 1992; Palumbi, 2001). These are pathogens that we know have caused significant mortality to humans and domestic livestock in the past. We know they have no difficulty establishing and multiplying in their host popu- lations. People now in their mid-50s have benefited from their absence for most of their lives; yet, it is likely that at least half of us will acquire them late in life (probably in a hospital or while travelling on public transport), and they will be our terminal interaction with a pathogen. The mathematics of drug resistance has been studied from a variety of per- spectives. One of the earliest and simplest insights comes from May and Dobson who show simply that the rate at which drug (or pesticide/insecticide) resistance evolves is mainly determined by the log of the basic reproductive rate of the or- ganism evolving resistance (May and Dobson, 1986). This explains very simply
APPENDIX A 207 why mosquitoes and bacteria quickly evolve resistance; in contrast, birds of prey were unlikely to ever evolve resistance to egg-shell thinning: if it takes a hundred generations to evolve resistance, then what takes months for bacteria requires centuries for a bird of prey. There is really only one pathogen that has emerged recently where no attempt has been made to eradicate the pathogen and its evolution has been studied beyond the first few generations of cases. Work on Mycoplasma gallisepticum (MG) in the North American house finch provides a number of important insights that are likely to generalize to other emergent pathogens should they escape detection and control at the initial stages of emergence. MG emerged from domestic fowl into wild finches in the fall of 1993. House finches were the most conspicuous hosts as their pathology of MG is characterized by pronounced swelling of the eyes, which reduces their ability to locate food and likely increases their susceptibility to predation (Dhondt et al., 1998; Dobson, 2013). Through a large Citizen Sci- ence network established by the Laboratory of Ornithology in Cornell, the spread and impact of the disease has been monitored for the last 20 years (Hosseini et al., 2006). The pathogen has now spread across the entire United States infecting house finches both in their introduced range (eastern United States) and in their native range (western United States) as well as up to 30 different species of birds. The eastern population has declined by around 60 percent (around half a billion birds), less significant declines are observed in the west (Dhondt et al., 2006; Hochachka and Dhondt, 2000). Detailed laboratory, field, and genetic studies illustrate that the pathogen has evolved continuously since emerging, and this evolution has both increased and decreased virulence depending on the condition determining selection. This evolution can happen quite quickly and is reversible (Dhondt et al., 2005; Hawley et al., 2010, 2013). There is essentially no differ- ence in the behavior of identical strains of the pathogen in inbred eastern birds and outbred, more genetically diverse western birds; essentially, the hosts show no real evidence for genetic resistance to the pathogen. This is not surprising as their age-structured population has only been exposed to less than 10 generations of selection; the pathogen has likely had several thousand generations of selection in this time. All of which should give pause for thought to those who think or preach that we can readily breed or otherwise genetically modify hosts for disease resistance. Even for small rapidly breeding hosts like passerine birds, the asym- metry in host and pathogen demographic rates will always allow the pathogen to evolve at rates that make the hosts genetic response essentially inert. The situa- tion is even more asymmetrical when we consider humans and emerging viruses. Conclusions There is increasing evidence that pathogens play a significant role in deter- mining the economic well-being of most of the worldâs nations (Bonds et al., 2012); this occurs through their direct effects on the size and efficiency of the labor force. The traditional economic argument that the wealth of nations is
208 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS determined solely by governance is now seen as a deeply flawed and biased argument (Acemoglu et al., 2000, 2003; McArthur and Sachs, 2001). Although the economic impact of recent disease outbreaks is frequently cited as a central reason to increase funding for research on emerging pathogens, I suspect that the continued impact of older diseases such as malaria and the neglected tropical diseases has a larger annual effect on the global economy than any of the recent emerging disease threats. Ultimately we need more funds to study the dynamics of both emerging and endemic pathogens and their hosts. Disease is as important as governance in driving national economiesâbiodiversity may play a role in buffering pathogens. Pathogens are a large component of natural ecosystemsâperhaps as much as 90 percent of biodiversity is parasitic on free-living species. Pathogens emerge when we disturb natural ecosystemsâbut, we have as much chance of a new pathogen emerging in our own back yards as we have of something else emerging from the tropics. We need to think more deeply about the population dynamics of pathogen emergence, and step back a little from the romance of fishing for tropi- cal viruses with microchips. If we ask the simple question âWould the tropical virus hunters have identified HIV?â I suspect the long incubation period with no initial pathology would have led them to dismiss it as inert and innocuous and to miss it entirely. Ultimately pathogens emerge and cause problems for humans, domestic live- stock, and other wildlife species because we have disturbed their natural habitat in ways that modifies their transmission rates. This suggest that developing a better mathematical understanding of the dynamics of food webs and the role that parasites play in these large complex nonlinear systems will provide alternative insights into the way in which pathogens emerge (Dobson et al., 2009; Hudson et al., 2006; Lafferty et al., 2008). Similar mathematical models will also be needed to understand how immune systems function and how the brain commu- nicates with the nervous and endocrine systems. These mathematical understand- ings of the function of complex systems are easily as important to the future of human health as is the current focus on genomic understanding. Indeed, in the absence of the understanding about how the parts coded for by genes interact together, we are simply dealing with the ânatural historyâ of these systems at a heroically tiny, but essentially disconnected, biological scale. Acknowledgements APDâs research is sponsored by the NSF/NIH Ecology of Infectious Dis- ease Program and a grant from the McDonnell Foundation for Studies of Com- plex Systems. An initial draft of this paper was prepared for discussion at the ANTIGONE workshop on interspecies barriers and zoonotic disease emergence in Toledo, Spain, September 2014. All of the work described above benefited from discussions with colleagues at this workshop and in the Ecology and Evolu- tion of Infectious Disease Group at Princeton.
APPENDIX A 209 References Acemoglu, D., S. Johnson, and J. A. Robinson. 2000. The colonial origins of comparative develop- ment: An empirical investigation. Cambridge, MA: National Bureau of Economic Research. Acemoglu, D., S. Johnson, and J. Robinson. 2003. Disease and development in historical perspective. Journal of the European Economic Association 1(2-3):397-405. Alcamo, J., D. v. Vuuren, C. Ringler, J. Alder, E. M. Bennett, D. Lodge, T. Masui, T. Morita, M. Rosegrant, O. Sala, K. Schulze, and M. B. Zurek. 2005. Methodology for developing the MA scenarios. In Ecosystems and human well-being. Vol. 2, Millennium Ecosystem Assessment, edited by S. R. Carpenter, P. L. Pingali, E. M. Bennett, and M. B. Zurek. Washington, DC: Island Press. Pp. 145-172. Altizer, S., C. L. Nunn, P. Thrall, J. M. Gittleman, J. Antonovics, A. A. Cunningham, A. P. Dobson, V. Ezenwa, K. E. Jones, A. B. Pedersen, M. Poss, and J. R. C. Pulliam. 2003. Social organiza- tion and parasite risk in mammals: Integrating theory and empirical studies. Annual Review of Ecology, Evolution, and Systematics. 34:517-547. Anderson, R. M. 1991. Populations and infectious diseases: Ecology or epidemiology? Journal of Animal Ecology 60:1-50. Anderson, R. M., and R. M. May. 1983. Vaccination against rubella and measles: Quantitative inves- tigations of different policies. Journal of Hygiene (Cambridge) 90:259-325. Anderson, R. M., and R. M. May. 1985. Age-related changes in the rate of disease transmission: Impli- cations for the design of vaccination programmes. Journal of Hygiene (Cambridge) 94:365-436. Anderson, R. M., and R. M. May. 1986. The invasion, persistence and spread of infectious diseases within animal and plant communities. Philosophical Transactions of the Royal Society of Lon- don Series B-Biological Sciences 314:533-570. Anderson, R. M., and R. M. May. 1990. Immunisation and herd immunity. Lancet 335:641-645. Anderson, R. M., and R. May. 1992. Infectious diseases of humans. New York: Oxford University Press. Anthony, S. J., J. H. Epstein, K. A. Murray, I. Navarrete-Macias, C. M. Zambrana-Torrelio, A. Solovyov, R. Ojeda-Flores, N. C. Arrigo, A. Islam, S. Ali Khan, P. Hosseini, T. L. Bogich, K. J. Olival, M. D. Sanchez-Leon, W. B. Karesh, T. Goldstein, S. P. Luby, S. S. Morse, J. A. Mazet, P. Daszak, and W. I. Lipkin. 2013. A strategy to estimate unknown viral diversity in mammals. MBio 4(5):e00598-00513. Bartlett, M. S. 1957. Measles periodicity and community size. Journal of the Royal Statistical Society Series A 120:48-70. Bartlett, M. S. 1960. The critical community size for measles in the U.S. Journal of the Royal Statisti- cal Society Series A 123:37-44. Bartlett, M. S. 1966. An introduction to stochastic processes, with special reference to methods and applications. Cambridge, UK: Cambridge University Press. Bibby, C. J., N. J. Collar, M. J. Crosby, M. F. Heath, C. Imboden, T. H. Johnson, A. J. Long, A. J. Stattersfield, and S. J. Thirgood. 1992. Putting biodiversity on the map: Priority areas for global conservation. Cambridge, UK: International Council for Bird Preservation. Bjornstad, O. N., B. F. Finkenstadt, and B. T. Grenfell. 2002. Dynamics of measles epidemics: Es- timating scaling of transmission rates using a time series SIR model. Ecological Monographs 72(2):169-184. Bonds, M. H., A. P. Dobson, and D. C. Keenan. 2012. Disease ecology, biodiversity, and the latitu- dinal gradient in income. PLoS Biology 10(12):e1001456. Cliff, A. D., and P. Haggett. 1988. Atlas of disease distributions: Analytic approaches to epidemio- logic data. Oxford, UK: Basil Blackwell. Cliff, A. D., and J. K. Ord. 1981. Spatial processes: Models and applications. London, UK: Pion. Cohen, M. L. 1992. Epidemiology of drug resistance: Implications for a post-antimicrobial era. Sci- ence 257:1050-1055.
210 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Conroy, M. J., and B. R. Noon. 1996. Mapping of species richness for conservation of biological diversity: Conceptual and methodological issues. Ecological Applications 6(3):763-773. Craft, M. E., P. L. Hawthorne, C. Packer, and A. P. Dobson. 2008. Dynamics of a multihost pathogen in a carnivore community. Journal of Animal Ecology (in press, online early). Dhondt, A. A., D. L. Tessaglia, and R. L. Slothower. 1998. Epidemic mycoplasmal conjunctivitis in house finches from Eastern United States. Journal of Wildlife Diseases 34(2):265-280. Dhondt, A. A., S. Altizer, E. G. Cooch, A. K. Davis, A. P. Dobson, M. J. L. Driscoll, B. K. Hartup, D. M. Hawley, W. M. Hochachka, P. R. Hosseini, C. S. Jennelle, G. V. Kollias, D. H. Ley, E. C. H. Swarthout, and K. V. Sydenstricker. 2005. Dynamics of a novel pathogen in an avian host: Mycoplasmal conjunctivitis in house finches. Biotropica 94(1):77-93. Dhondt, A. A., A. V. Badyaev, A. P. Dobson, D. M. Hawley, M. J. L. Driscoll, W. M. Hochachka, and D. H. Ley. 2006. Dynamics of mycoplasmal conjunctivitis in the native and introduced range of the host. Ecohealth 3(2):95-102. Diekmann, O., and J. A. P. Heesterbeek. 2000. Mathematical epidemiology of infectious diseases: Model building, analysis and interpretation. Hoboken, NJ: John Wiley and Sons. Diekmann, O., J. A. P. Heesterbeek, and J. A. J. Metz. 1990. On the definition and the computation of the basic reproduction ratio R0 in models for infectious-diseases in heterogeneous populations. Journal of Mathematical Biology 28:365-382. Dobson, A. P. 2009. Climate variability, global change, immunity, and the dynamics of infectious diseases. Ecology 90(4):920-927. Dobson, A. P. 2013. The eye of the finch. The Scientist, August 1. Dobson, A. P., J. P. RodrÃguez, W. M. Roberts, and D. S. Wilcove. 1997. Geographical distribution of endangered species in the United States. Science 275:550-553. Dobson, A. P., I. Cattadori, R. D. Holt, R. S. Ostfeld, F. Keesing, K. Krichbaum, J. R. Rohr, S. E. Perkins, and P. J. Hudson. 2006. Sacred cows and sympathetic squirrels: The importance of biological diversity to human health. PLoS Medicine 3(6):e231. Dobson, A. P., K. D. Lafferty, A. M. Kuris, R. F. Hechinger, and W. Jetz. 2008. Homage to Linnaeus: How many parasites? How many hosts? Proceedings of the National Academy of Sciences of the United States of America 105(August 12):11482-11489. Dobson, A. P., S. Allesina, K. Lafferty, and M. Pascual. 2009. The assembly, collapse and resto- ration of food webs. Philosophical Transactions of the Royal Society of London Series B 364(1524):1803-1806. Dye, C., and N. Gay. 2003. Modeling the SARS epidemic. Science Express (10.1126/ science.1086925):1-2. Erwin, T. L. 1982. Tropical forests: Their richness in Coleoptera and other arthropod species. Cole- optera Bulletin 36:74-82. Ezenwa, V. O., M. S. Godsey, R. J. King, and S. C. Gupthill. 2006. Avian diversity and West Nile virus: Testing associations between biodiversity and infectious disease risk. Proceedings of the Royal Society B 273:109-117. Ferrari, M. J., R. F. Grais, N. Bharti, A. J. K. Conlan, O. N. Bjornstad, L. J. Wolfson, P. J. Guerin, A. Djibo, and B. T. Grenfell. 2008. The dynamics of measles in sub-Saharan Africa. Nature 451:679-684. Fraser, C., S. Riley, R. M. Anderson, and N. M. Ferguson. 2004. Factors that make an infectious disease outbreak controllable. Proceedings of the National Academy of Sciences of the United States of America 101:6146-6151. Gaston, K. J. 1991. The magnitude of global insect species richness. Conservation Biology 5:283. Gaston, K. J. 1994. Spatial patterns of species description: How is our knowledge of the global insect fauna growing? Biological Conservation 67(1):37-40. Grenfell, B. T., and R. M. Anderson. 1985. The estimation of age-related rates of infection from case notifications and serological data. Journal of Hygiene (Cambridge) 95:419-436. Grenfell, B. T., and A. P. Dobson. 1994. Infectious diseases of wildlife. Edited by B. T. Grenfell and A. P. Dobson. Cambridge, UK: Cambridge University Press.
APPENDIX A 211 Grenfell, B. T., and A. P. Dobson. 1995. Ecology of Infectious Diseases in Natural Populations. Cambridge: Cambridge University Press. Grenfell, B. T., O. N. Bjornstad, and J. Kappey. 2001. Travelling waves and spatial hierarchies in measles epidemics. Nature 414(6865):716-723. Hawley, D. M., K. V. Dhondt, A. P. Dobson, J. L. Grodio, W. M. Hochachka, D. H. Ley, E. E. Osnas, K. A. Schat, and A. A. Dhondt. 2010. Common garden experiment reveals pathogen isolate but no host genetic diversity effect on the dynamics of an emerging wildlife disease. Journal of Evolutionary Biology 23(8):1680-1688. Hawley, D. M., E. E. Osnas, A. P. Dobson, W. M. Hochachka, D. H. Ley, and A. A. Dhondt. 2013. Parallel patterns of increased virulence in a recently emerged wildlife pathogen. PLoS Biology 11(5):e1001570. Hochachka, W. M., and A. A. DHondt. 2000. Density dependent decline of host abundance resulting froma new infectious disease. Proceeding of the National Academy of Sciences of the United States of America 97(10):5303-5306. Hodkinson, I. D. 1992. Global insect diversity revisited. Journal of Tropical Ecology 8:505-508. Hosseini, P., A. A. Dhondt, and A. P. Dobson. 2006. Spatial spread of an emerging infectious disease: Conjunctivitis in house finches. Ecology 87(12):3037-3046. Hudson, P. J., R. Norman, M. K. Laurenson, D. Newborn, M. Gaunt, L. Jones, H. Reid, E. Gould, R. Bowers, and A. Dobson. 1995. Persistence and transmission of tick-borne viruses: Ixodes ricinus and louping-ill virus in red grouse populations. Parasitology 111:S49-S58. Hudson, P. J., A. P. Dobson, and K. D. Lafferty. 2006. Is a healthy ecosystem one that is rich in parasites? Trends in Ecology and Evolution 21(7):381-385. Jetz, W., D. Wilcove, and A. P. Dobson. 2007. Projected impacts of climate and land-use change on the global diversity of birds. PLoS Biology 5(6):e157. doi:110.1371/journal.pbio.0050157. Keeling, M. J., and B. T. Grenfell. 1997. Disease extinction and community size: Modeling the per- sistence of measles. Science 275:65-67. Keesing, F., R. D. Holt, and R. S. Ostfeld. 2006. Effects of species diversity on disease risk. Ecology Letters 9(4):485-498. Lafferty, K. D., and C. L. Wood. 2013. Itâs a myth that protection against disease is a strong and general service of biodiversity conservation: Response to Ostfeld and Keesing. Conservation Biology 14:722-728. Lafferty, K. D., S. Allesina, M. Arim, C. J. Briggs, G. De Leo, A. P. Dobson, J. A. Dunne, P. T. J. Johnson, A. M. Kuris, D. J. Marcogliese, N. D. Martinez, J. Memmott, P. A. Marquet, J. P. McLaughlin, E. A. Mordecai, M. Pascual, R. Poulin, and D. W. Thieltges. 2008. Parasites in food webs: The ultimate missing links. Ecology Letters 11(6):533-546. Lipsitch, M., T. Cohen, B. Cooper, J. M. Robins, S. Ma, L. James, G. Gopalakrishna, S. K. Chew, C. C. Tan, M. H. Samore, D. Fisman, and M. Murray. 2003. Transmission dynamics and control of severe acute respiratory syndrome. Science 300(5627):1966-1970. Lloyd-Smith, J. O., D. George, K. M. Pepin, V. E. Pitzer, J. R. C., J. R. C. Pulliam, A. P. Dobson, P. J. Hudson, and B. T. Grenfell. 2009. Epidemic dynamics at the human-animal interface. Science 326(5958):1362-1367. Lockhart, A. B., P. H. Thrall, and J. Antonovics. 1996. Sexually transmitted diseases in animals: Ecological and evolutionary implications. Biological Reviews 71(3):415-471. LoGuidice, K., R. S. Ostfeld, K. A. Schmidt, and F. Keesing. 2003. The ecology of infectious disease: Effects of host diversity and community composition on Lyme disease risk. Proceedings of the National Academy of Sciences of the United States of America 100(2):567-571. May, R. M., and A. P. Dobson. 1986. Population dynamics and the rate of evolution of pesticide re- sistance. In Pesticide resistance management (from an International Symposium, Washington, D.C., 27-29 November 1984). Washington, DC: National Academy Press. Pp. 170-193. McArthur, J. W., and J. D. Sachs. 2001. Institutions and geography: Comment on Acemoglu, Johnson and Robinson (2000). Cambridge, MA: National Bureau of Economic Research.
212 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS McCallum, H., and A. Dobson. 2006. Disease and connectivity. In Connectivity conservation, ed- ited by K. R. Crooks and M. A. Sanjayan. Cambridge, UK: Cambridge University Press. Pp. 479-501. McLean, A. R., R. M. May, J. Pattison, and R. A. Weiss. 2005. SARS. A case study in emerging infec- tions. Oxford, UK: Oxford University Press. Ostfeld, R. S., and F. Keesing. 2000. Biodiversity and disease risk: The case of Lyme disease. Con- servation Biology 14:722-728. Ostfeld, R. S., and F. Keesing. 2013. Straw men donât get Lyme disease: Response to Wood and Lafferty. Trends in Ecology and Evolution 28(9):502-503. Ostfeld, R. O., and K. L. LoGiudice. 2003. Community disassembly, biodiversity loss, and the erosion of an ecosystem service. Ecology 84:1421-1427. Palumbi, S. R. 2001. EvolutionâHumans as the worldâs greatest evolutionary force. Science 293(5536):1786-1790. Plowright, R. K., P. Foley, H. E. Field, A. P. Dobson, J. E. Foley, P. Eby, and P. Daszak. 2011. Urban habituation, ecological connectivity and epidemic dampening: The emergence of Hendra virus from flying foxes (Pteropus spp.). Proceedings of the Royal Society B: Biological Sciences 278:3703-3712, Pulliam, J. R., J. G. Dushoff, S. A. Levin, and A. P. Dobson. 2007. Epidemic enhancement in partially immune populations. PLoS ONE 2(1):e165. Pulliam, J. R. C., J. H. Epstein, J. Dushoff, S. A. Rahman, M. Bunning, A. A. Jamaluddin, A. D. Hyatt, H. E. Field, A. P. Dobson, and P. Daszak. 2011. Agricultural intensification, priming for persistence and the emergence of Nipah virus: A lethal bat-borne zoonosis. Journal of the Royal Society Interface 9(66):89-101. Randolph, S., A. Dobson, B. Allan, H. Dutra, L. Goessling, K. Barnett, J. Chase, R. Marquis, G. Pang, and G. Storch. 2012. Pangloss revisited: A critique of the dilution effect and the biodiversity- buffers-disease paradigm. Parasitology-Cambridge 139(7):847. Reid, W. V., H. A. Mooney, A. Cropper, D. Capistrano, S. R. Carpenter, K. Chopra, P. Dasgupta, T. Dietz, A. K. Duriappah, R. Hassan, R. Kasperson, R. Leemans, R. M. May, A. J. McMichael, P. L. Pingali, C. Samper, R. Scholes, R. Watson, A. H. Zakri, Z. Shidong, N. J. Ash, E. M. Bennett, P. Kumar, M. J. Lee, C. Raudsepp-Hearne, H. Simons, J. Thonell, and M. B. Zurek. 2005. Ecosystems and human well-being. Synthesis, Millennium Ecosystem Assessment. Wash- ington, DC: Island Press. Roche, B., A. P. Dobson, J. F. Guegan, and P. Rohani. 2012. Linking community and disease ecology: The impact of biodiversity on pathogen transmission. Philosophical Transactions of the Royal Society B-Biological Sciences 367(1604):2807-2813. Rodó, X., M. Pascual, F. J. Doblas-Reyes, A. Gershunov, D. A. Stone, F. Giorgi, P. J. Hudson, J. Kinter, M.-Ã. RodrÃguez-Arias, N. Ch. Stenseth, D. Alonso, J. Garcia, and A. P. Dobson. 2013. Climate Change and Infectious diseases: Can we meet the needs for better prediction? Climate Change 118:625-640. Ryan, S. J., J. H. Jones, and A. P. Dobson. 2013. Interactions between social structure, demography, and transmission determine disease persistence in primates. PLoS ONE 8(10):e76863. Schmidt, K. A., and R. S. Ostfeld. 2001. Biodiversity and the dilution effect in disease ecology. Ecology 82:609-619. Swinton, J. 1998. Extinction times and phase transitions for spatially structured closed epidemics. Bulletin of Mathematical Biology 60(2):215-230. Swinton, J., J. Harwood, B. T. Grenfell, and C. A. Gilligan. 1998. Persistence thresholds for phocine distemper virus infection in harbour seal Phoca vitulina metapopulations. Journal of Animal Ecology 67(1):54-68. Wolfe, N. D., C. P. Dunavan, and J. Diamond. 2007. Origins of major human infectious diseases. Nature 447:279-283.
APPENDIX A 213 A9 ENVIRONMENTAL CHANGE AND INFECTIOUS DISEASE: HOW NEW ROADS AFFECT THE TRANSMISSION OF DIARRHEAL PATHOGENS IN RURAL ECUADOR26 Joseph N. S. Eisenberg,27 William Cevallos,28 Karina Ponce,27 Karen Levy,29 Sarah J. Bates,30 James C. Scott,30 Alan Hubbard,30 Nadia Vieira,28 Pablo Endara,28 Mauricio Espinel,28 Gabriel Trueba,28 Lee W. Riley,30 and James Trostle31 Abstract Environmental change plays a large role in the emergence of infectious disease. The construction of a new road in a previously roadless area of northern coastal Ecuador provides a valuable natural experiment to examine how changes in the social and natural environment, mediated by road con- struction, affect the epidemiology of diarrheal diseases. Twenty-one villages were randomly selected to capture the full distribution of village population size and distance from a main road (remoteness), and these were compared with the major population center of the region, Borbón, that lies on the road. Estimates of enteric pathogen infection rates were obtained from case- control studies at the village level. Higher rates of infection were found in nonremote vs. remote villages [pathogenic Escherichia coli: odds ratio (OR) = 8.4, confidence interval (CI) 1.6, 43.5; rotavirus: OR = 4.0, CI 1.3, 12.1; and Giardia: OR = 1.9, CI 1.3, 2.7]. Higher rates of all-cause diarrhea were found in Borbón compared with the 21 villages (RR = 2.0, CI 1.5, 2.8), as well as when comparing nonremote and remote villages (OR = 2.7, CI 1.5, 26ââReprinted with permission from the Proceedings of the National Academy of Sciences of the United States of America. Originally printed as Eisenberg et al. 2006. Environmental change and infectious disease: how new roads affect the transmission of diarrheal pathogens in rural Ecuador. Pro- ceedings of the National Academy of Sciences of the United States of America 103(51):19460-19465. 27ââSchool of Public Health, University of Michigan, Ann Arbor, MI 48104. 28ââSchool of Public Health, University of California, Berkeley, CA 94720. 29ââDepartment of Environmental Science, Policy, and Management, University of California, Berke- ley, CA 94720. 30ââUniversidad San Francisco de Quito, Quito, Ecuador. 31ââDepartment of Anthropology, Trinity College, Hartford, CT 06106. ââNotes: Author contributions: J.N.S.E., M.E., G.T., L.W.R., and J.T. designed research; W.C., K.P., K.L., S.J.B., N.V., P.E., and J.T. performed research; J.N.S.E., J.C.S., and A.H. analyzed data; and J.N.S.E., K.L., and J.T. wrote the paper. ââThe authors declare no conflict of interest. ââAbbreviations: ââCI: confidence interval; OR: odds ratio.
214 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS 4.8). Social network data collected in parallel offered a causal link between remoteness and disease. The significant and consistent trends across viral, bacterial, and protozoan pathogens suggest the importance of considering a broad range of pathogens with differing epidemiological patterns when as- sessing the environmental impact of new roads. This study provides insight into the initial health impacts that roads have on communities and into the social and environmental processes that create these impacts. The more public health scientists learn about infectious disease processes, the more they can implicate environmental changes in the recent emergence or reemergence of infectious diseases (Colwell et al., 1998; Morse, 1995; Patz et al., 2000). Given the increasing number of emerging pathogens recently identified, there is an urgent need to understand how environmental change influences disease burden. Such changes are potentially more visible in places where they have been caused by human activity, such as construction of dams, pipelines, and roads. Anthropogenic environmental changes that cause populations to move and settle in new ways can provide the opportunity to observe the relationship between environmental change and disease transmission. Where such environ- mental changes are unevenly distributed across a region, thereby producing the conditions of a natural experiment, these relationships can be observed easily and systematically. The construction of a new road in a previously roadless area in northern coastal Ecuador provides just such a natural experiment to examine how changes in the social and natural environment, mediated by road construction, affect the epidemiology of diarrheal diseases. Various studies have examined the impact of road construction on disease incidence (Birley, 1995). For example, the building of the TransAmazon Highway was associated with an increase in malaria (Ault, 1989; Coimbra, 1988). These in- creases in incidence were attributed to the presence of water pools created by road construction practices. More recently, a study in the Peruvian Amazon indicated that mosquito biting rates are significantly higher in areas that have undergone deforestation and development associated with road development (Vittor et al., 2006). Analogously, a study in India measured a higher prevalence of dengue vec- tors along major highways than elsewhere (Dutta et al., 1998). Studies in Uganda suggest that the main road linking Kenya to Kampala has higher proportions of HIV-positive women working in bars and HIV-positive truck drivers than does the surrounding area (Carswell, 1987). In general, transportation changes mobility and circulation of humans, which can affect the incidence of sexually transmitted diseases (Panos Institute, 1988), as well as health-care-seeking behavior (Airey, 1991, 1992). As opposed to sexually transmitted diseases, fecalâoral pathogens can survive outside of the human host and therefore will behave differently under environmental changes. Some studies have suggested that remote villages sepa- rated by large distances are less able to sustain transmission of certain fecalâoral pathogens, such as amoebas and rotavirus (Black, 1975; Gilman et al., 1976;
APPENDIX A 215 Gunnlaugsson et al., 1989). The impact that environmental changes from road construction have on these diarrheal diseases remains largely unexplored and un- known, despite the fact that diarrheal diseases remain a major cause of mortality among infants and children under 5 years of age (WHO/UNICEF, 2004). In 1996 the Ecuadorian government began a road construction project to link the southern Colombian border with the Ecuadorian coast. A two-lane asphalt highway was completed in 2001, spanning 100 km across the southern end of the Chocó rainforest near the Pacific Ocean. Secondary roads continue to be built, linking additional villages to the paved road (Figure A9-1). These roads provide a faster and cheaper mode of transportation compared with rivers. The extent to which roads influence communities should be measured by their proximity in time and distance to a given village (e.g., remoteness) and not merely by their presence or absence. To examine the impact of remoteness on diarrheal disease we implemented a hierarchical design that collects data by village to obtain information about the region, and by individual to obtain information about potential confounding factors that may bias the analysis. Roads influence disease transmission through a variety of mechanisms. For example, road proximity can increase in- and out- migration rates causing multiple demographic changes in the age, racial, and socioeconomic profile. These rapid and complex changes can reduce social con- nectedness within a community, which may in turn reduce a communityâs ability to maintain good sanitation and hygiene conditions. Road proximity can also affect short-term travel patterns, thereby increasing the potential for the introduc- tion of new pathogen strains into communities. In addition to diarrheal symptoms, three specific marker pathogens (Esch- erichia coli, rotavirus, and Giardia) were followed, each with a distinct epidemi- ology. Both pathogenic E. coli and rotavirus are responsible for a large proportion of diarrhea mortality and severe morbidity throughout the developing world, whereas Giardia, also a major cause of diarrhea, is more pervasive, resulting in higher infection rates (Blaser et al., 2002). Taken together, these three pathogens represent the primary pathways (food, water, and person-to-person) for transmis- sion of diarrhea. Results Table A9-1 presents community characteristics, with two methods for char- acterizing location: remoteness of a community relative to the town Borbón, and river basin in which a community resides. The least remote community has a remoteness value of 0.012, and the most remote village has a remoteness value of 0.198. Close villages were defined as those with a remoteness value of < 0.03; medium villages were defined as those with a remoteness value between 0.03 and 0.13; and remote villages were defined as those with a remoteness value > 0.13. These classifications are also represented in the regional map (Figure A9-1).
216 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A9-1âMap of study region. The 21 villages are categorized by river basin (Santiago, Cayapas, Onzole, Bajo Borbón, and road) and by remoteness (close, medium, and far).
APPENDIX A 217 TABLE A9-1â Community Characteristics Remoteness Remoteness Village Population size metric category River basin 1 284 0.012 Close Road 2 731 0.015 Close Road 3 78 0.022 Close Cayapas 4 482 0.027 Close Road 5 156 0.040 Medium Santiago 6 55 0.040 Medium Bajo Borbón 7 138 0.040 Medium Bajo Borbón 8 72 0.049 Medium Road 9 90 0.049 Medium Santiago 10 60 0.061 Medium Onzole 11 86 0.080 Medium Onzole 12 110 0.113 Medium Cayapas 13 135 0.122 Medium Santiago 14 83 0.140 Far Onzole 15 300 0.152 Far Santiago 16 228 0.155 Far Santiago 17 79 0.158 Far Cayapas 18 268 0.165 Far Cayapas 19 28 0.173 Far Onzole 20 443 0.190 Far Onzole 21 130 0.198 Far Cayapas Borbón 864 0 Remoteness is a measure of the time and cost of travel to Borbón. Roads provide cheaper and faster access to Borbón, and therefore remoteness is a measure of the proximity to the road. Note that the population of Borbón is the sample size enrolled in the study, rather than the size of the entire population (â 5,000). Village population size ranged from 28 to 731, and the random sample of 200 houses in Borbón resulted in 864 individuals, or â20% of the population. A total of 298 cases of diarrhea were identified in the communities during the three case-control cycles, and 44 cases were identified in Borbón during the one case-control cycle (Table A9-2). In addition, a total of 845 and 125 controls were sampled from the communities and Borbón, respectively. Crude prevalence estimates are shown in Table A9-3 for diarrhea and infection by both case status and remoteness category. The crude prevalence estimates for diarrhea [RR = 2.0, 95% confidence interval (CI) 1.5, 2.8] and pathogenic E. coli (RR = 16.0, 95% CI 13.2, 19.2) were significantly higher in Borbón compared with those in other communities (Table A9-4). These large differences between infection prevalence in Borbón vs. the community are seen in both cases and controls (Table A9-3). We found no evidence that crude prevalence estimates for rotavirus and Giardia varied between Borbón and the other 21 communities.
218 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS TABLE A9-2â Number of Cases and Controls by Remoteness No. of Remoteness No. of collection No. of category villages Population days No. of cases controls Remote 8 1,669 45 112 317 Medium 9 895 45 91 248 Close 4 1,592 45 95 280 Community* 21 4,156 45 298 845 Borbón 1 867 15 44 125 For communities other than Borbón, figures are the sum from three 15-day case-control studies across all 21 study villages between August 2003 and February 2006. Borbón figures are from one 15-day case-control study in July 2005. *Total from all 21 villages (sum of remote, medium and close villages). Adjusting for age of individuals, community population size, and sanitation level, the prevalence of infection was significantly higher in villages closer to or along a road compared with those communities far from the road for pathogenic E. coli [odds ratio (OR) = 3.9, 95% CI 1.1, 13.6], rotavirus (OR = 4.1, 95% CI 2.0, 8.4), and Giardia (OR = 1.6, 95% CI 1.0, 2.4); the same was true for all- cause diarrhea (OR = 1.8, 95% CI 1.2, 2.6) (Table A9-5). Precipitation was not included in the final model because its P value was > 0.2. These overall infection trends were largely driven by the controls, as evident from the crude prevalence estimates in Table A9-3 that are stratified by case status. Although the crude diar- rhea prevalence values show no trend as a function of remoteness, the adjusted risk estimates comparing both remote and medium as well as remote and close were significant, after adjusting for the population size and sanitation level of each community (Table A9-5). To test for a trend, remoteness was modeled as a continuous variable. The relative risk of infection associated with a decrease in remoteness from the far- thest to the closest village was significant for all infections: pathogenic E. coli (OR = 8.4, 95% CI 1.6, 43.5), rotavirus (OR = 4.0, 95% CI 1.3, 12.1), and Giar- dia (OR = 1.9, 95% CI 1.3, 2.7). For all-cause diarrhea the relative risk was also significant (OR = 2.7, 95% CI 1.5, 4.8) (Table A9-5). Discussion We observed strong trends in infection rates and all-cause diarrhea in villages across a gradient of remoteness for our marker pathogens even after adjusting for population size, sanitation, and precipitation. This result suggests that villages farther from the road have lower infection rates than villages closer to the road. This relationship between infection and road proximity is also seen in Borbón, the only community directly connected to both the primary road and all of the major rivers that serve the region. We observed significantly higher rates of E.
TABLE A9-3â Crude Infection Prevalence by Case Status and Remoteness (prevalence per 100 persons) Overall infection prevalence, Asymptomatic infection prevalence, Symptomatic infection prevalence, Diarrhea infections/100 infections/100 infections/100 Remoteness prevalence, category cases/100 E. coli Rotavirus Giardia E. coli Rotavirus Giardia E. coli Rotavirus Giardia Remote 2.6 1.0 2.7 16.7 0.6 2.2 15.8 0.4 0.6 0.9 Medium 4.6 3.1 3.6 16.6 2.3 2.7 15.2 0.5 0.9 1.5 Close 2.2 3.9 6.7 23.2 3.0 6.2 22.4 0.1 0.5 0.8 Community 2.8 2.4 4.5 19.4 1.9 4 18.4 0.3 0.6 0.9 Borbón 5.6 22.5 3.6 19.5 20.7 2.3 17.6 1.7 1.2 1.9 For communities other than Borbón estimates are based on the average of three 15-day case-control studies across all 21-study villages. Borbón estimates are based on one 15-day case-control study. Overall infection prevalence is based on a weighted average of infection in cases and controls. Prevalence estimates are based on a 15-day period prevalence. 219
220 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS TABLE A9-4â Comparison of Infection Prevalence in Communities vs. Borbón Community, cases/100 Borbón, cases/100 Relative risk (95% CI) E. coli 1.6 22.5 16.0 (13.2, 19.2) Rotavirus 4.5 3.6 0.8 (0.6, 1.2) Giardia 19.4 19.5 1.0 (0.9, 1.2) Diarrhea 2.8 5.6 2.0 (1.5, 2.8) For communities other than Borbón estimates are based on the average of three 15-day case-control studies across all 21 study villages. Borbón estimates are based on one 15-day case-control study. Pathogen prevalence is based on infection (a weighted average of cases and controls). Relative risk is the prevalence risk ratio (the risk of illness or infection in Borbón relative to the communities). coli and all-cause diarrhea in Borbón than in the other 21 study communities. These health differences have policy significance given that both pathogenic E. coli and rotavirus are major causes of mortality and severe morbidity in children. These data were collected across three river basins during three visits to each town over 2 years, minimizing the chance that unmeasured localized events either temporally or spatially confounded the risk estimates. We found no statistical relationship between diarrhea or infection rates and time period or river basin. Any unmeasured confounding would have had to continue over the 2-year study period or had to occur across the three river basins. Explaining the causes of the trends discussed here requires understanding the ecological and social impacts of roads. One common purpose (and consequence) of a new road is increased logging. Deforestation causes major changes in water- shed characteristics and local climate, both of which can affect the transmission of enteric pathogens (Curriero et al., 2001). Perhaps more important than ecologi- cal processes, social processes facilitated by roads such as migration, creation of new communities, and increased density of existing communities can affect pathogen transmission. Changes in community social structures often create or are accompanied by inadequate infrastructure, which affects hygiene and sanita- tion levels, and in turn the likelihood of transmission of enteric pathogens. Roads TABLE A9-5â Infection as a Function of Remoteness OR (95% CI) E. coli Rotavirus Giardia Diarrhea Remote 1.00 1.00 1.00 1.00 Medium 3.0 (0.8, 11.9) 1.3 (0.5, 3.2) 1.2 (0.7, 2.0) 1.8 (1.1, 3.0) Close 3.9 (1.1, 13.6) 4.1 (2.0, 8.4) 1.6 (1.0, 2.4) 1.8 (1.2, 2.6) Continuous 8.4 (1.6, 43.5) 4.0 (1.3, 12.1) 1.9 (1.3, 2.7) 2.7 (1.5, 4.8) OR of infection/disease for individuals in communities that are classified as close or medium from Borbón as compared with those communities that are classified as far (remote). The continuous mea- sure is the OR comparing the farthest with the closest using a continuous measure of remoteness. Esti- mates were adjusted for age of individual, population size of village, and community-level sanitation.
APPENDIX A 221 can also increase flows of consumer goods such as processed food, material goods, and medicines and may also provide communities with increased access to health care, health facilities, and health information. By determining the transmission potential of the causal factors associated with new roads, we can better interpret the observed trends in infection rates across our study region. The propensity of a pathogen to persist within a com- munity is characterized by the reproductive number Ro, defined as the average number of infections caused by an infectious individual in a completely suscep- tible population (Anderson and May, 1991). For directly transmitted diseases, Ro is a function of (i) contact rate among others within or outside the community, (ii) infectivity (the probability of infection given a contact), and (iii) duration of the infectious period. For enteric pathogens that can persist in the environment, Ro is also a function of a pathogenâs viability outside the human host and its ability to move to a new susceptible one. The consistent and strong trends observed in these data across viral, bacterial, and protozoan pathogens suggest that Ro for many enteric pathogens is lower for remote villages compared with nonremote villages; i.e., these remote communities are less able to sustain transmission of pathogens. The trends in infection rates that we observed are partially explained by the effect of social connectedness on the risk of transmission of many pathogens. Figure A9-2 shows a causal diagram that illustrates how demographic changes, measured by rates of in- and out-migration for a community, and contact outside of village, measured by short-term travel of people in and out of a community, might increase levels of infection or disease for fecalâoral pathogens. Localized migration facilitated by roads can lead to a community whose residents have few social connections, which is one measure of social capital (Bebbington and Perreault, 1999). Previous studies have shown that communities with more social capital tend to be successful in creating adequate water and sanitation infrastruc- ture because they tend to know one another, are accustomed to working together, and share social norms (Grootaert and van Bastelaer, 2002; Isham and Kahkonen, 1998; Watson et al., 1997). On the other causal pathway, road proximity can increase the contact that individuals within a village have with those outside the village, increasing the rate of introduction of pathogens. FIGURE A9-2â Causal diagram linking proximity of the road to increases in infection and diarrheal disease.
222 FIGURE A9-3â Relationship between social factors and remoteness. (A) Movement outside of community, measured by the percentage leaving the village during the past week (linear fit R2 = 0.25, P ⤠0.05). (B) The social connectedness within a community, as measured by the number of villagers a given individual spent time with during the past week (linear fit R2 = 0.50, P ⤠0.05).
APPENDIX A 223 Our study villages show some evidence of these hypothesized relationships among demographic characteristics, social connectedness, and movement of people. Village data suggest that connectedness, as measured by the average number of individuals a given person spends time with (social network degree), is positively associated with remoteness (Figure A9-3B). Additionally, villages closer to the road have increased movement of people (Figure A9-3A), which provides opportunities for pathogen incursion. The slope of the line reflects the strength of the relationship: twice as many connections exist in the most remote village compared with the least remote. Likewise, 28% of the remote villagers said they had left the village in the last week, compared with 48% of the least remote villagers. Pathogen-specific outcomes provide additional insight into the relationship between remoteness and transmission. Observed trends were strongest for E. coli, followed by rotavirus and then Giardia. This differential can be partially explained by the biological and environmental factors that govern transmission dynamics and level of Ro; e.g., pathogen infectivity, as measured by infectious inoculum, shedding rates, and environmental persistence, as measured by the ability of the pathogen to remain viable in the environment, all directly affect Ro. Infectivity data suggest that Giardia, with a low ID50 (the inoculum at which 50% of exposed subjects are infected) and long shedding duration, and rotavirus, with a low ID50 and high shedding rates, are more infectious (Carter 2005; Regli et al., 1991; Teunis et al., 1986) than diarrhea-causing E. coli (Dupont et al., 1971, 1989; Feachem, 1983; Haas et al., 1999; Karch et al., 1995; Teunis et al., 1986). Diarrheagenic E. coli species tend to persist in the environment for shorter peri- ods of time than either Giardia or rotavirus (Carter, 2005; deRegnier et al., 1989; Enriquez et al., 1995; Estes, 1991; McFeters et al, 1974; Raphael et al., 1985). The above observations on both infectivity and environmental persistence suggest that Giardia is able to maintain transmission within the more remote vil- lages despite limited outside social contact and higher levels of social connected- ness. Likewise, E. coli would be less able to maintain transmission, and rotavirus would lie somewhere in between. The significant difference in E. coli infection rates between Borbón and the other communities and the lack of difference in Giardia infection rates are consistent with this hypothesis. The significant and consistent trends across viral, bacterial, and protozoan pathogens suggest the importance of considering a broad range of health out- comes when assessing environmental impact. Each of our marker pathogens has a different epidemiology that is affected by environmental changes in different ways. A stratified analysis that looks across pathogen types, and not just at a broader disease category like diarrhea, allows for a more sensitive measure of change and can elucidate more specific interventions to alleviate these environ- mental impacts. We propose this design as a general model that can be used to examine anthropogenic environmental determinants of health in other places.
224 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS A number of issues require further examination. In this regional analysis we compare remote and nonremote villages at a given point in time. Investigating changes in incidence compared with changes in remoteness over time may pro- vide additional causal information about how road development affects disease, because the time scale of these social changes may take years or decades, and the details are complex and poorly understood. In addition, molecular analysis of pathogens could elucidate transmission patterns across the landscape, and data on human migration patterns might provide information on causal linkages between roads and diarrheal disease. To substantiate the causal diagram shown in Figure A9-2, better measures of social capital and its relation to water and sanitation are needed. Gathering information on other health outcomes such as nutrition and vectorborne and sexually transmitted disease would also provide the opportunity to broaden our examination of causal linkages between road development and disease, because these are likely to vary for different etiologies. Environmental effects are often both geographically widespread and tempo- rally extended and therefore can be difficult to correlate with disease outcomes. The ability to observe change requires a study design and analysis that involve data collection within a systems-level framework. The natural experiment created by road construction in this region, combined with the regional design, allows these relationships to be studied. When associations between exposure and out- come are placed in the broader context of processes in which they occur (Figure A9-2), one can examine the causal linkages between environmental change and disease at a systems level. When international agencies like the World Bank make decisions about whether to invest or how best to proceed in large-scale infrastructure projects, their impact assessments have begun to pay attention to variables associated with environmental, social, and health factors (World Bank, 1997). Although the World Bank now includes human health as a component of the environmental impact of road construction (Tsunakawa and Hoban, 1997), few studies of the health effects of roads exist, particularly with respect to infectious disease trans- mission (see www.who.int/hia/examples/en). This analysis provides insight into the interactions between roads, the social and environmental processes that they affect, and the resulting impacts on the health of human communities. These com- plex causal pathways suggest that efforts to mitigate the negative effects of roads should consider a larger range of their short- and long-term health implications. Materials and Methods Study Population and Selection Process The study area is located in the northern Ecuadorian province of Esmeraldas in the canton Eloy Alfaro, which comprises â150 villages. Villages are located
APPENDIX A 225 along three rivers, the RÃo Cayapas, RÃo Santiago, and RÃo Onzole, all draining toward the town Borbón, the main population center of the region. Borbón, with â5,000 inhabitants, is distinct from the other communities along the river. It has a higher population density but nonetheless maintains an underdeveloped infra- structure for its size, with untreated sewage, rudimentary solid waste management systems, and minimal water and sanitation services that vary in quality between households. The communities outside Borbón, on the other hand, are smaller in size and density. Their water is primarily obtained from rivers and consumed untreated, although rainwater is used intermittently, and a few communities have wells or receive piped water from surface sources. Sanitation facilities are of varying quality, although they generally would be classified as unimproved by World Health Organization criteria; flush toilets are uncommon. The region is primarily populated by Afro-Ecuadorians, with a smaller proportion represented by Chachis, an indigenous group that mostly resides in more remote villages. There are an increasing number of mestizos (people of mixed origin) moving into villages close to or on the road. More details on the region can be accessed elsewhere (Rival, 2003; Sierra, 1998, 1999; Whitten, 1965, 1974). The construction of the road from Borbón westward to the coast was com- pleted in 1996. The portion of the road connecting Borbón eastward to the upper reaches of the Andes was completed in 2003. Secondary and tertiary dirt roads off of this two-lane asphalt highway are continually being built, mostly for log- ging. At the time the data were collected, both the primary and secondary roads reached 15% of the 150 villages in the canton. All villages in the region were categorized based on their geographic location relative to Borbón. A sample of 21 villages was selected by using block random- ization to ensure that villages throughout the study region were represented. At the beginning of the study, four of the 21 study villages were connected to the road. All households within each village were recruited. In Borbón, a random sample of 200 households (of â1,000) was selected for inclusion in the study. Consent was obtained at both the village and household level. Institutional review board committees at the University of California (Berkeley), Trinity College, and Universidad San Francisco de Quito approved all protocols. Study Design Between July 2003 and May 2005 each enrolled village was visited three times on a rotating basis. Each visit lasted 15 days, during which all cases of diar- rhea were identified by visiting each household every morning. For each occur- rence of diarrhea two controls were randomly sampled from the same community and one control was sampled from the case household. One 15-day case-control study was conducted in Borbón in July 2005.
226 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Cases were defined as an individual who had three or more loose stools in a 24-h period. Controls were defined as someone with no signs of diarrhea in the past 6 days. Microbiologic Analyses Every morning during the 15-day period field staff members visited each household to find cases of diarrhea and collect stool samples from cases and controls. The samples were tested for rotavirus, pathogenic E. coli, and Giardia. All stools were stored on ice and processed within 48 h. In the field an EIA kit was used to identify rotavirus (RIDA Quick Rotavirus; R-Biopharm, Darmstadt, Germany). An aliquot of stool was preserved in liquid nitrogen and tested for Giardia in Quito with an ELISA kit (RIDASCREEN Giardia; R-Biopharm). For bacterial analysis, stool was plated directly onto MacConkey and XLD agar. All lactose-negative isolates were analyzed for urease and oxidase, and with API 20 E (bioMèrieux, Marey lâEtoile, France) to speciate the bacterial isolates. Lactose- negative isolates that were identified as either Shigella or E. coli, along with five randomly chosen lactose-positive isolates, were further analyzed by PCR. Pathotype-specific primers were used to diagnose the following: enterotoxigenic E. coli (ETEC), enteropathogenic E. coli (EPEC), enteroinvasive E. coli (EIEC), and Shigella spp., as reported previously (Tornieporth et al., 1995). The primers amplified the bfp gene of EPEC, the LT and STa genes of enterotoxigenic E. coli, and the ipaH gene of EIEC and Shigella spp. The specific procedure is discussed elsewhere (Tornieporth et al., 1995). Both a positive and negative control were used in each gel run. A positive control for each pathotype was provided by Lee Riley (University of California, Berkeley). A K12 E. coli strain was used for the negative control. In the following analysis Shigella and E. coli cases were grouped together. Demographic and Socioeconomic Survey To determine individualsâ movements and social interactions, we adminis- tered demographic and sociometric surveys to all study participants. The surveys included questions regarding travel to and from the community, as well as social contacts outside the individualâs household during the previous week. The degree of social connectedness for each individual was defined as the number of names provided to the interviewer in response to the question, âwho did you spend time with in your community, other than household members, during the past week?â plus the number of times that individual was nominated by others within the community (Bell, 1999; Scott, 2000). The surveys were developed after extensive anthropological observations to obtain regionally appropriate phrasing of ques- tions. They were translated and back-translated to ensure accuracy. Interviewers
APPENDIX A 227 were trained together to ensure uniformity. All data were entered into Access (Microsoft, Redmond, WA). Standard quality control procedures were conducted, including examining the data for logical errors and double entry of 10% of the surveys. The surveys were administered once to each study participant, with an average of 82% coverage per village. To cover all study villages, half of the sur- veys were administered in the summer of 2003 and half in the summer of 2004. Statistical Analyses For each village, travel time and total cost of travel to Borbón were recorded by field staff members. For each village i, rank of remoteness, Ri, was then calculated by summing normalized values of time, Ti, and cost, Ci. Specifically, Ti Ci Ri = + â 21 Tj â 21C j j j Because the metric is the result of two values standardized to a [0,1] scale, the possible range of Ri is from 0 (the town Borbón itself) to 2 (the theoretical farthest community from Borbón). Villages were classified into three groups based on their remoteness metric: close, medium, and far from Borbón. Community prevalence of infection for each village was calculated by ag- gregating data from all three case-control cycles and weighting cases and controls appropriately; i.e., we assumed that all cases were identified during the 15-day surveys and that the controls were a random sample of those without diarrhea. Specifically, the population prevalence of pathogen i in community j was esti- mated as follows: w1 j I casesij w2 j I controlsij Pij = + w1 j + w2 j N casesi w1 j + w2 j N controlsi where Icasesij and Icontrolsij are the number of individuals in which pathogen i was isolated in the cases and controls, respectively, Ncasesi and Ncontrolsi are the number of cases and controls, respectively, w1 = the inverse of the proportion of cases tested for the particular pathogen (this weight is equal to one when diarrhea is the outcome variable), and w2 = (total population â no. of cases identified)/ (no. of controls). To estimate the change in risk of infection/disease by remoteness we used a logistic regression model, parameterizing remoteness in two different ways: (i) as a continuous variable (distance between the closest village, which is ad- jacent to Borbón, to the farthest among the study villages) and (ii) as a pair of
228 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS categorical indicator variables (âcloseâ and âmedium,â with âfarâ considered baseline). Models included one individual-level variable (age of participant at time of case control visit) as well as the following community-level variables: sanitation level (percentage of individuals who stated that they used improved sanitation, i.e., latrines or septic tanks), population size, and average 30-day rainfall (using data from the 15 days before and 15 days during the case-control study). For all analyses, we derived the statistical inference using robust estimates of the standard errors from a generalized estimating equation approach (Zeger et al., 1988). This approach accounts for residual correlation of the outcomes of individuals within the same villages and provides inference that is not sensitive to model misspecification. The relatively low prevalence of diarrhea in this popu- lation permitted us to estimate relative risk with the prevalence odds ratio from our logistic model (Jewell, 2004). All analysis was conducted by using STATA version 8 (Stata, College Station, TX). Acknowledgments We thank the Ecologia, Desarrollo, Salud, y Sociedad (EcoDESS) proj- ect field team for their invaluable contribution collecting the data. This study was supported by National Institute of Allergy and Infectious Diseases Grant R01-AI050038. References Airey T (1991) Transport Rev 11:273â290. Airey T (1992) Soc Sci Med 34:1135â1146. Anderson RM, May R (1991) Infectious Diseases of Humans: Dynamics and Control (Oxford Univ Press, New York). Ault SK (1989) in Demography and Vector-Borne Diseases, ed Service MW (CDC, Boca Raton, FL), pp 283â301. Bebbington A, Perreault T (1999) Econ Geogr 75:395â418. Bell DC (1999) Social Networks 21:1â21. Birley MH (1995) The Health Impact Assessment of Development Projects (HMSO, London). Black FL (1975) Science 187:515â518. Blaser MJ, Smith PD, Greenberg HB, Rivdin JI, Guerrant RL (2002) Infections of the Gastrointestinal Tract (Lippincott Williams & Wilkins, Philadelphia), 2nd Ed. Carswell JW (1987) AIDS 1:223â227. Carter MJ (2005) J Appl Microbiol 98:1354â1380. Coimbra CEA (1988) Hum Org 47:254â260. Colwell RR, Epstein PR, Gubler D, Maynard N, McMichael AJ, Patz JA, Sack RB, Shope R (1998) Science 279:968â969. Curriero FC, Patz JA, Rose JB, Lele S (2001) Am J Public Health 91:1194â1199. deRegnier DP, Cole L, Schupp DG, Erlandsen SL (1989) Appl Environ Microbiol 55:1223â1229. DuPont HL, Formal SB, Hornick RB, Snyder MJ, Libonati JP, Sheahan DG, LaBrec EH, Kalas JP (1971) N Engl J Med 285:1â9. DuPont HL, Levine MM, Hornick RB, Formal SB (1989) J Infect Dis 159:1126â1128.
APPENDIX A 229 Dutta P, Khan SA, Sharma CK, Doloi P, Hazarika NC, Mahanta J (1998) Southeast Asian J Trop Med Public Health 29:173â176. Enriquez CE, Hurst CJ, Gerba CP (1995) Water Res 29:2548â2553. Estes MK (1991) in Fundamental Virology, eds Fields BN, Knipe DM (Raven, New York), pp 619â644. Feachem RG (1983) Sanitation and Disease: Health Aspects of Excreta and Wastewater Management (Wiley, Chichester, UK). Gilman RH, Davis C, Gan E, Bolton M (1976) Am J Trop Med Hyg 25:663â666. Grootaert C, van Bastelaer T, eds (2002) Understanding and Measuring Social Capital: A Multidis- ciplinary Tool for Practitioners (World Bank, Washington, DC). Gunnlaugsson G, Smedman L, da Silva MC, Grandien M, Zetterstrom R (1989) Acta Paediatr Scand 78:62â66. Haas CN, Rose JB, Gerba CP (1999) Quantitative Microbial Risk Assessment (Wiley, New York). Isham J, Kahkonen (1998) The Institutional Determinents of the Impact of Community-Based Water Services: Evidence from Sri Lanka and India (World Bank, Washington, DC). Jewell NP (2004) Statistics for Epidemiology (Chapman and Hall/CRC, Boca Raton, FL). Karch H, Russmann H, Schmidt H, Schwarzkopf A, Heesemann J (1995) J Clin Microbiol 33: 1602â1605. McFeters GA, Bissonnette GK, Jezeski JJ, Thomson CA, Stuart DG (1974) Appl Microbiol 27: 823â829. Morse SS (1995) Emerg Infect Dis 1:7â15. Panos Institute (1988) AIDS and the Third World (Panos Institute, London). Patz JA, Graczyk TK, Geller N, Vittor AY (2000) Int J Parasitol 30:1395â1405. Raphael RA, Sattar SA, Springthorpe VS (1985) Can J Microbiol 31:124â128. Regli S, Rose JB, Haas CN, Gerba CP (1991) J Am Water Work Assn 83:76â84. Rival L (2003) Oxford Dev Stud 31:479â501. Scott J (2000) Social Network Analysis: A Handbook (SAGE, London). Sierra R (1999) Environ Conserv 26:136â145. Sierra R, Stallings J (1998) Hum Ecol 26:135â161. Teunis PFM, van der Heijden OG, van der Giessen JWB, Havelaar AH (1986) The DoseâResponse Relation in Human Volunteers for Gastrointestinal Pathogens (National Institute of Public Health and the Environment, Bilthoven, The Netherlands). Tornieporth NG, John J, Salgado K, de Jesus P, Latham E, Melo MC, Gunzburg ST, Riley LW (1995) J Clin Microbiol 33:1371â1374. Tsunakawa K, Hoban C, eds (1997) Roads and the Environment: A Handbook (World Bank, Wash- ington, DC), Technical Paper no. 376. Vittor AY, Gilman RH, Tielsch J, Glass G, Shields T, Lozano WS, Pinedo-Cancino V, Patz JA (2006) Am J Trop Med Hyg 74:3â11. Watson G, Jagannathan V, Gelting R, Beteta H (1997) in User Organizations for Sustainable Water Services, eds Subramanian A, Jagannathan V, Meinzen-Dick R (World Bank, Washington, DC), Technical Paper no. 354. Whitten NE (1965) Class, Kinship, and Power in an Ecuadorian Town: The Negroes of San Lorenzo (Stanford Univ Press, Stanford, CA). Whitten NE (1974) Black Frontiersmen: A South American Case (Schenkman, Cambridge, MA). World Bank (1997) Health Aspects of Environmental Impact Assessment: Environmental Assessment Sourcebook Update 18 (World Bank, Washington, DC). World Health Organization and United Nations Childrenâs Fund (2004) (WHO/UNICEF, Geneva), p 36. Zeger SL, Liang KY, Albert PS (1988) Biometrics 44:1049â1060.
230 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS A10 IN-ROADS TO THE SPREAD OF ANTIBIOTIC RESISTANCE: REGIONAL PATTERNS OF MICROBIAL TRANSMISSION IN NORTHERN COASTAL ECUADOR32 Joseph N. S. Eisenberg,33 Jason Goldstick,33 William Cevallos,34 Gabriel Trueba,34 Karen Levy,35 James Scott,36 Bethany Percha,33 Rosana Segovia,34 Karina Ponce,34 Alan Hubbard,37 Carl Marrs,33 Betsy Foxman,33 David L. Smith,38 and James Trostle39 Abstract The evolution of antibiotic resistance (AR) increases treatment cost and probability of failure, threatening human health worldwide. The relative im- portance of individual antibiotic use, environmental transmission, and rates of introduction of resistant bacteria in explaining community AR patterns is poorly understood. Evaluating their relative importance requires studying a region where they vary. The construction of a new road in a previously roadless area of northern coastal Ecuador provides a valuable natural ex- periment to study how changes in the social and natural environment affect the epidemiology of resistant Escherichia coli. We conducted seven bi-annual 15-day surveys of AR between 2003 and 2008 in 21 villages. Resistance to both ampicillin and sulphamethoxazole was the most frequently observed profile, based on antibiogram tests of seven antibiotics from 2210 samples. The prevalence of enteric bacteria with this resistance pair in the less remote communities was 80 percent higher than in more remote communities (OR = 1.8 [1.3, 2.3]). This pattern could not be explained with data on individual antibiotic use. We used a transmission model to help explain this observed discrepancy. The model analysis suggests that both transmission and the rate of introduction of resistant bacteria into communities may contribute to the observed regional scale AR patterns, and that village-level antibiotic 32ââReprinted with permission from The Royal Society. Originally Printed as Eisenberg et al. 2011. In-roads to the spread of antibiotic resistance: Regional patterns of microbial transmission in northern coastal Ecuador. Journal of the Royal Society: Interface 9(70):1029-1039. 33ââSchool of Public Health, University of Michigan, Ann Arbor, MI, USA. 34ââDepartment of Microbiology, Universidad San Francisco de Quito, Quito, Ecuador. 35ââRollins School of Public Health, Emory University, Atlanta, GA, USA. 36ââDepartment of Mathematics and Statistics, Colby College, Waterville, ME, USA. 37ââSchool of Public Health, University of California, Berkeley, CA, USA. 38ââDepartment of Biology and Emerging Pathogens Institute, University of Florida, Gainesville, FL, USA. 39ââDepartment of Anthropology, Trinity College, Hartford, CT, USA.
APPENDIX A 231 use rate determines which of these two factors predominate. While usually conceived as a main effect on individual risk, antibiotic use rate is revealed in this analysis as an effect modifier with regard to community-level risk of resistance. Introduction Antibiotic resistance (AR) threatens human health worldwide (Hogberg et al., 2010). As resistant bacteria spread, and failure of antibiotics in the clini- cal setting increases in frequency, infections require more expensive drugs and are more likely to be associated with serious morbidity and/or mortality (IOM, 2003). The cost of these failures exceeds billions of dollars annually in the United States (Foster, 2010). That the evolution of AR is influenced by individual antibiotic use in human and veterinary medicine is well known (Collignon et al., 2009; Love et al., 2011), and programmes aimed at limiting the spread of resistant bacteria often focus on restricting antibiotic use and/or choosing therapeutic options that minimize selection for resistance (Drusano, 2003). Yet, resistance mechanisms are often complex, suggesting that resistant bacteria are not likely to arise by antibiotic selection pressure over the course of treatment alone, and in many cases, the genes that confer resistance must have been acquired by colonizing bacteria or shared among bacteria on mobile genetic elements (MacLean et al., 2010). The emphasis on evolution of AR during treatment ignores the role of acqui- sition of resistant bacteria via other transmission routes, such as environmental pathways and human contact patterns. The relative role of these different factors in determining the prevalence of AR within and across communities has not been studied, however, and in general, little is known about the spread of resistant bacteria in community settings. The relationship between the total antibiotic use and the rate of AR spread among individuals in a population is an important but unresolved question, as is the role of broader ecological processes in spreading resistant bacteria among animals and humans (Smith et al., 2002a, 2006). Study- ing population-level processes shifts the emphasis from individual use to overall antibiotic use rates and the number of other people who carry resistant bacteria (Bonten et al., 1998). Transmission models are important tools to study such system-level population processes. Mathematical models of infection transmission have been used throughout the twentieth century to help understand the epidemiology of infectious diseases (Anderson and May, 1991). These theoretical approaches describe the ecological and evolutionary dynamics of hostâpathogen interactions that generate disease patterns in space and time (Smith et al., 2002b). Mathematical models have been applied to the emergence and the spread of resistant bacteria, extending simple transmission models to reflect competition, such as simple infections
232 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS with colonization inhibition (Bonhoeffer et al., 1997), complex infections with resistance (Austin et al., 1997), or amplification of resistant bacteria owing to overgrowth following antibiotic use (Smith et al., 2002a). In general, these mod- els have focused on hospital settings (Bonten et al., 2001) and quantify the effects of different infection control measures (Bonhoeffer et al., 1997; Bootsma et al., 2006; Lipsitch et al., 2000; Massad et al., 1993). In hospital settings, health care workers are often modelled as vectors that spread resistant organisms among patients (Austin et al., 1999). Mathematical models can also offer important insights into the mechanisms and extent of the spread of AR in community settings, which are more difficult to study. Recent AR models have focused on movement of patients among hospitals (Austin et al., 1999), long-term care facilities (Smith et al., 2004), and the com- munity (Austin et al., 1997) and the role of antibiotic use in agriculture (Smith et al., 2002a). Emergence of AR can be modelled as an invasive pathogen (Smith et al., 2002b) into the human population (Smith et al., 2002a, 2005) using models that incorporate spatial and social processes (Singer et al., 2006). Evaluating the relative importance of individual medication use, environ- mental transmission, and rates of introduction of AR bacteria in explaining community AR patterns requires studying a region where there is variability in all of these factors. The construction of a new road in a previously roadless area of northern coastal Ecuador provides a valuable natural experiment to study how changes in the social and natural environment, mediated by road construction, affect the evolution and the spread of AR enterobacteria. This study area, comprising villages with varying degrees of remoteness relative to the main road (Figure A10-1), offers an ideal location for studying AR at a community scale. Since we postulate that the social and ecological changes that might affect the spread of AR bacteria will unfold over a large time scale, we use remote villages as a proxy for conditions prior to the construction of the road and close villages as a proxy for conditions after. We, therefore, use a cross-sectional design along with statistical models to examine AR as a function of remoteness, and we use mathematical models to explain the relative contribu- tions of: (i) antibiotic use; (ii) transmission of AR bacteria, generally mediated through standard water, sanitation, and hygiene environmental pathways; and (iii) rates of introduction of resistant bacteria, represented in our model as an ingestion factor, in explaining observed patterns of AR in 21 communities. The spread of resistant bacteria is framed here as a spatially inhomogeneous process that affects prevalence. This occurs through both environmental sources and hu- man movement patterns, whose effects are modified by conditions that increase the potential for human-to-human transmission, such as poor sanitation. Based on 5 years of data across 21 communities, we describe regional patterns of AR prevalence and use a transmission model to provide plausible explanations for these observed patterns.
APPENDIX A 233 FIGURE A10-1âMap of study region. The 21 villages are categorized by river basin (Santiago, Cayapas, Onzole, Bajo Borbón, and road), and by remoteness (close, medium, and far).
234 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Methods Study Site In the northern Ecuadorian province of Esmeraldas, approximately 150 vil- lages (ranging from 20â800 inhabitants) lie along the Cayapas, Santiago and Onzole rivers, which all flow towards Borbón, the main population centre of the region (with 5000 inhabitants). Villagers primarily consume untreated sur- face source water and sanitation facilities are inadequate. The region, populated primarily by Afro-Ecuadorians (Whitten, 1965), is undergoing intense environ- mental and social changes owing to the construction of a new highway along the coast, which connects previously remote villages to the outside world. Construc- tion of the road was completed from Borbón westward to the provincial capital of Esmeraldas in 1996 and from the coast eastward to the Andean mountains in 2003. Secondary and tertiary dirt roads off of this two-lane asphalt highway are continually being built, mostly for logging and the area has come to be known as one of the worldâs top 10 biodiversity hotspots (Myers et al., 2000). At the time these data were collected, 15 per cent of the 150 villages in the region were accessible by road. All villages in the region were categorized based on their geographical lo- cation relative to Borbón. A sample of 21 villages was selected by using block randomization to ensure that villages of varying remoteness and population sizes were represented; four of these were connected to the road when this study began. All households within each village were recruited, except in Borbón, where a random sample of 200 households (from approx. 1000) was selected for inclu- sion in the study. Consent was obtained at both the village and household level. Institutional review boards at the University of California Berkeley, University of Michigan, Trinity College, and Universidad San Francisco de Quito approved all protocols. Study Design Between August 2003 and February 2008, each enrolled village was visited seven times, with each visit lasting 15 days. Villages were visited on a rotating basis, during which time field staff identified all cases of diarrhoea through ac- tive surveillance. For each case of diarrhoea (defined using WHO standards as three or more loose stools in a 24 h period), two controls were randomly sampled from the same community, and one control was sampled from the case household. Controls were defined as someone with no signs of diarrhoea in the previous 6 days. Four 15-day case-control visits were conducted in Borbón. Antibiotic usage was measured through a sequential random sample of households where many of the households were measured more than once. A key informant was asked whether any household members had used antibiotics within the last week and, if so, they were asked to name the drug. Responses from the key informant were
APPENDIX A 235 converted to the individual level by recording usage for those identified by the survey and imputing a response of âNo usageâ for the remaining individuals who were known to live in the house from previous demographic surveys. Classifying Remoteness For each village, travel time and total cost of travel to Borbón were recorded by field staff members. Specifically, transport time was estimated assuming the use of a motorized canoe or bus, depending on location, and transport cost was determined through inquiries of key informants within each community. For each village, i, rank of remoteness, Ri, was calculated by summing normalized values of time, Ti and cost, Ci. Specifically, Ti Ci Ri = + â 21 j Tj â 21C j . j Since the metric is the result of two values standardized to a [0,1] scale, the possible range of Ri is from 0 (the town Borbón itself) to 2 (the theoretical far- thest community from Borbón). Villages were classified into three groups based on their remoteness metric: close, medium, and far from Borbón. The categorical breakpoints were selected by maximizing the differences in the mean remoteness values for each category. Microbiological Analysis Stool samples were collected by field staff from cases and controls, stored on ice and processed within 48 h and tested for the presence of Escherichia coli and AR. Lactose-positive isolates that were identified as E. coli were further analysed for antibiotic susceptibility (to ampicillin (amp), cefotaxime, chloram- phenicol, ciprofloxacin, gentamicin, sulphamethoxazoleâtrimethoprim (sxt) and tetracycline) using the disc-diffusion method following standard methods. To test for the presence of E. coli, stool was plated directly onto MacConkey agar; lactose-positive colonies were further cultured in Chromocult agar. The five most prominent lac+ isolates were initially selected and one confirmed E. coli isolate was randomly chosen for further AR analysis. All lactose-negative isolates were analysed for urease and oxidase, and with API 20 E (bioMérieux Corp) to speci- ate the bacterial isolates. Lactose-positive isolates that were identified as E. coli were further analysed for antibiotic susceptibility (to amp, cefotaxime, chloram- phenicol, ciprofloxacin, gentamicin, sxt and tetracycline) using the disc-diffusion method following standard methods (Bauer et al., 1966; Blake et al., 2003). As sulphamethoxazole and trimethoprim work synergistically, they are commonly
236 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS used together, often in the same pill. Therefore, one standard clinical approach is to screen for the combined resistance to both at the same time with discs impreg- nated with both antibiotics, and the resulting resistance to both antibiotics is then listed as sxt resistance. This was done as part of this study, with the limitation that we do not have information on E. coli isolates that were resistant to sulpha- methoxazole, but not trimethoprim, or vice versa. These seven antibiotics were chosen to be included in this study because they were reported to be the most commonly used antibiotics in the region both by physicians within our field staff and by other physicians who also work in the study region. Statistical Analysis Our statistical analysis consists of the following: (i) calculating prevalence of each AR profile correcting for the unequal sampling probabilities of cases and controls; (ii) estimating the variability of individual-level antibiotic use using random effects models to compare variability over time over space; (iii) estimat- ing the association between AR and remoteness using binary response general estimating equation (GEE) models; (iv) summarizing prevalence of antibiotic use in terms of drugs most frequently used, and in terms of prevalence of use; (v) exploring how antibiotic use rates vary as a function of remoteness to investigate their potential utility in explaining observed AR patterns; and (vi) examining the assumptions associated with aggregating our AR data over time. Calculating Prevalence The data used to estimate the distribution of AR was a non-standard case- control design consisting of cases, household controls, and community controls. Since cases are relatively rare, simple estimators of prevalence are potentially biased owing to over-representation of cases. To obtain community prevalence estimates, therefore, required different analytical techniques that use the follow- ing weighting procedure. Cases (those presenting with diarrhoea) were given a weight of 1, since all cases in each community were sampled, giving them a sampling probability of 1. Household controls (those sampled within a house with a case and not presenting with diarrhoea) are weighted by the inverse of the proportion of the susceptible population of household controls represented by the control sample. The equivalent weight is also calculated for the community controls (note, this weighting was done by community and collection cycle, and thus the weighted contribution of a community/cycle to the analysis is the same regardless of its total population size, i.e. the communities are the units). Using these weights, we calculate the standard HorvitzâThompson estimator (Lohr, 1999) of prevalence, which yields unbiased estimators of population means and proportions in unequal probability samples.
APPENDIX A 237 Variability of Antibiotic Use To compare the variability of antibiotic use over time and over space, two random effect models are fit with antibiotic use as the dependent variable. In the first model, the variance of the random offset corresponding to household is estimated; in the second, the variance of the random offset corresponding to time point. Comparison of the size of these variances is then used to give an indica- tion of whether there is more variability between households (spatial) or between time points (temporal). Further details on this analysis are given in the electronic supplementary material. Association Between Antibiotic Resistance Prevalence and Remoteness To explore the relationship between ampâsxt resistance prevalence and re- moteness, we estimate the odds ratio between the binary indicator of ampâsxt resistance and (i) the binary indicator of medium/close remoteness, using âfarâ as the reference category, as well as (ii) the binary indicator of residence in Borbón using the other communities as the reference category. To correct for unequal probability sampling, each observation is replicated a number of times equal to its sampling weight. Odds ratios are estimated by fitting a logistic regression model to this expanded dataset. To derive the statistical inference for the relevant measures of association, we relied on the clustered non-parametric bootstrap, specifically re-sampling 21 villages with replacement from the expanded dataset and estimate the odds ratio from this âbootstrap datasetâ (Efron and Tibshirani, 1993). This process is repeated 10,000 times to estimate the sampling distribution of the odds ratios, and we use the quantile method to derive the 95% CI. In the far versus medium/close comparisons, only bootstrap datasets that have at least two villages from each remoteness category are included, since the sampling of villages was done to create variability between villages in terms of remoteness. Therefore, datasets with 1 or 0 villages in one or more remoteness categories do not reflect the sampling distribution of interest. Similarly, in the Borbón versus community comparisons, bootstrap datasets that did not include Borbón at least once did not contribute to the reported confidence intervals (CI). Prevalence of Antibiotic Use To characterize the per-day prevalence of use, we calculate the proportion of individuals that report having used antibiotics within the last week and scale this quantity by 7, tacitly assuming that individuals only used drugs on 1 day within the last week and it was equally likely to have been any day. Since indi- viduals could have ingested drugs on more than 1 day, our use rate constitutes a lower bound. To look at what drugs are most commonly used, we summarize the relative frequency of drugs used among those that reported use (electronic supplementary material, table S1).
238 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Antibiotic Use and Remoteness Antibiotic use at a community level is estimated by the sample proportion of respondents who reported using antibiotics. We consider an individual to have used antibiotics if they indicate they have consumed any of: amp, amoxicillin, sulphamethoxazole, trimethiprim, or benzipenicillin. Ordinary least-squares re- gression was used to look at the relationship between the village-level propor- tion and remoteness. Although the proportion reporting use is clearly bounded between 0 and 1, the discrepancies from the regression line appeared symmetric (electronic supplementary material, figure S2), making ordinary least squares a tenable choice. Aggregation of Antibiotic Resistance Data Over Time To justify that the effect of remoteness on ampâsxt resistance is static, this relationship is assessed at each of the seven time points using a GEE model in the same way as was done in estimating the relationship between remoteness and AR prevalence on the full dataset. For each time point separately, we fit an independence GEE model with remoteness category as the lone predictor and ampâsxt resistance as the response variable. Confidence intervals for the odds ratios comparing âfarâ with the two other categories were produced using the same non-parametric bootstrap described for the full dataset, and intervals were examined for overlap (electronic supplementary material, figure S1). Greater detail is given in the electronic supplementary material. Modelling We use a village-level compartmental transmission model (Smith et al., 2002a) to examine the observed patterns of AR prevalence in our study com- munities. We chose a compartmental model, which assumes populations are well-mixed, because it provides better explanatory power than more complex model structures for understanding factors that drive transmission. A determin- istic model does not allow for the possibility of stochastic die-out, but at the phenotypic level, we do not observe this; i.e. all communities have non-zero prevalence. At the genotypic level, there could be stochastic die-out of specific strains, however, we do not have the genotypic information to illustrate this and therefore did not include this level of resolution in the model. The equation and parameters are shown in Figure A10-2. This model tracks four conditions among humans: (i) not colonized with resistant bacteria (W); (ii) transiently colonized with resistant bacteria, such that the bacteria have a high probability of dying out (X); (iii) colonized with resistant bacteria such that the population is more stable and less likely to die out compared with the exposed state (Y); and (iv) amplified or colonized with resistant bacteria such that bacterial species are present in high numbers and are actively reproducing (Z).
FIGURE A10-2â Deterministic antibiotic resistance model. W = 1âXâYâZ. See Smith et al. (2009a) for details. 239
240 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS The model assumes that human exposure to resistant bacteria comes from either: (i) the spread of these AR bacteria through standard water, sanitation, and hygiene pathways, or (ii) the ingestion of new antibiotic resistant strains that arise from either environmental sources (e.g., food or water) or introduction through movement of people to and from the region. AR spread is modelled as person- to-person transmission. Amplified resistant bacteria (Z) are assumed to transmit at a higher rate, β, than the unamplified or colonized bacteria, which transmit at a rate η. The rate of ingestion is described by the parameter λ. Antibiotic use, at a rate Ï, is assumed to alter the community ecology of the gut, eliminating competition with antibiotic-sensitive bacteria and allowing the population density of resistant bacteria to increase. The remaining five parameters that represent the rates of movements between states are described in Figure A10-2 as well as in Smith et al. (2009a). Although we observe that cases have higher prevalence of AR than do con- trols, both cases and controls have higher prevalence of AR in the less remote villages. Thus, in the simulation analysis, we do not make a distinction between cases and controls. An estimate of the transmission rate was established using E. coli prevalence data from our study region. Previous analysis of these data suggests an eightfold difference in E. coli prevalence comparing remote versus non-remote villages (Eisenberg et al., 2006). We use the prevalence values from this analysis for these two types of villages in conjunction with a susceptibleâinfectedâsusceptible (SIS) model (with disease duration of one week) to estimate β, the rate of transmission from amplified to susceptible individuals. β is estimated to be 0.154 new infec- tions per infectious individual per susceptible individual per day for the most remote village, and 0.325 for the least remote village, a transmission rate ratio of 2.11. To explore the sensitivity of transmission to AR prevalence, we vary this ratio in our simulation analysis from 0.9 to 9 keeping the baseline transmis- sion rate for remote villages at 0.154. The rate of transmission from colonized to susceptible individuals, η, is assumed to be one-tenth the value of β because colonized individuals have smaller populations of AR bacteria in their gut than amplified individuals. The antibiotic use rate, Ï, is based on survey data collected in each village, and does not vary by remoteness. The antibiotic use data, employed to estimate the antibiotic use rate parameter, Ï, are presented in §3. We specify the range for the antibiotic use rate by extending the 95% CI, resulting in the range Ï: 0.001 to 0.01 antibiotics per person per day. The per capita rate of human exposure to new strains (introduction rates), λ, is unknown for this region. We use the same per day baseline rate (0.001) reported in Smith et al. (2009a) to represent a remote village. To explore the sensitivity of λ to AR prevalence, we vary the rate of non-remote villages so that the ratio ranges from 1 to 10. The assumption that introduction rates are higher in non-remote villages is consistent with the observation that there is more human
APPENDIX A 241 movement to and from outside the region in these non-remote villages (Eisenberg et al., 2006), providing more opportunity to introduce AR bacteria. To examine the interaction between antibiotic use rates, transmission rate ratios comparing remote and non-remote villages, and introduction rate ratios comparing remote and non-remote villages, we simulate the model for a range of each of these three factors and use contour plots to present their relationship. The outcome measure is the risk ratio comparing a remote to a non-remote vil- lage. This risk ratio measure was compared with the empirical results presented in Table A10-2. Results Between 2003 and 2008, a total of 2,210 E. coli isolates were successfully analysed (518 were cases with diarrhoea and 1,692 were controls without diar- rhoea). We stratify our analysis by case/control status since the microbiota of those with diarrhoea is quite different from those without diarrhoea. Using results of screening isolates for sensitivities to seven antibiotics, we observed 39 unique profiles. The nine highest frequency profiles are listed in Table A10-1. The distri- bution of antibiotic profiles differs between cases and controls with cases having a tendency towards a higher frequency of resistance. Three of the most frequently observed profiles include resistance to amp and sxt. Sulphamethoxazole-resistant genes and trimethoprim-resistant genes are almost always present on the same integrons, while β-lactamase genes encoding resistance to amp can sometimes also be found in the same integron (Novais et al., 2006; Robin et al., 2010) or TABLE A10-1â Estimated Prevalence, Weighted by the Inverse Sampling Probability, of Antibiotic-Resistant E. coli Profiles Prevalence (per 100) Profile Total Cases Controls none 67.5 51.5 67.8 ampâsxtâtet 8.0 19.8 7.8 tet 6.9 4.0 7.0 other 3.5 4.7 3.5 amp 3.0 3.5 3.0 sxtâtet 2.9 1.8 2.9 ampâsxtâtetâclo 2.6 4.6 2.6 ampâtet 2.3 3.5 2.3 ampâsxt 2.1 6.3 2.1 sxt 1.0 0.3 1.1 Cases are defined as those with diarrhoea and controls are those without. All profiles with frequencies of less than 1 percent are placed in the âotherâ category. The antibiotics tested are: ampicillin (amp), tetracycline (tet), sulphamethoxazoleâtrimethoprim (sxt), chloramphenicol (clo), cefotaxime (ctx), gentamicin (gen) and ciprofloxacin (cip).
242 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS outside of the integron, but on the same plasmid (Miriagou et al., 2003; Woodford et al., 2009). In contrast, tetracycline resistance is never found as part of an in- tegron (Partridge et al., 2009). Thus, amp and sxt resistance are more likely to be horizontally and clonally transmitted together. For this reason, and because antibiotics that select for ampâsxt resistance are frequently used in the region (see below), we focus analysis on ampâsxt. We first report on the relationship between remoteness and ampâsxt re- sistance, showing that ampâsxt resistance decreases with remoteness. We next present our data on antibiotic use and show that there is no relationship between antibiotic use and remoteness, suggesting that the relationship between AR preva- lence and remoteness cannot be explained by differences in use rates alone. We present the results of an infection transmission model that examines the interac- tion between antibiotic use, transmission of resistant bacteria and introduction of resistant bacteria into villages in determining regional patterns of AR. The model analysis suggests that patterns of transmission as well as patterns of introductions of resistant bacteria into communities contribute to the regional-scale AR patterns we observed, and that antibiotic use rates determine which of these two factors predominate. AmpicillinâSulphamethoxazoleâTrimethoprim Resistance as a Function of Remoteness Ampicillinâsulphamethoxazoleâtrimethoprim (ampâsxt) resistance is signifi- cantly associated with lack of remoteness (Table A10-2). This trend is consistent TABLE A10-2â Prevalence and Odds Ratio of Simultaneous Antibiotic Resistance to amp and sxt Among Participants Living in 21 Villages in Ecuador Sulphamethoxazole and Ampicillin Resistance Control Overall Case prevalence prevalence prevalence (infections per (infections per (infections per Remoteness 100) 100) 100) OR (95% CI) Far 35.2 12.4 12.8 1.0 Medium 32.6 13.4 13.8 1.1 (0.6, 1.8) Close 43.0 20.1 20.5 1.8 (1.3, 2.3) Community 37.6 15.6 16.0 1.0 Borbón 46.4 19.4 20.0 1.3 (1.1. 1.6) Cases are defined as those with diarrhoea and controls are those without. Medium and close catego- ries are compared with the far category. Observations are weighted based on their inverse sampling probability to account for unequal probability sampling.
APPENDIX A 243 for both cases and controls. Estimating the community prevalence based on a weighted sum of the case and control observations, there was little difference in villages of far and medium remoteness (OR = 1.1 [0.6, 1.8]) whereas close vil- lages have higher prevalence relative to far villages (OR = 1.8 [1.3, 2.3]). Simi- larly, there are higher levels of resistance in Borbón, the main population centre of the region, compared with the communities collectively (OR = 1.3 [1.1, 1.6]). Although data were observed at seven different time points, we aggregate the data in this analysis, making an assumption about temporal stability of these relation- ships. This assumption is supported by data presented in electronic supplementary material, Figure S1, which show that the confidence intervals for the odds ratios stratified by time overlap. As with any symptom-based definition, there is the possibility of misclassifi- cation; however, if we assume that the disease misclassification is non-differential across our exposure (in this case remoteness of our study villages), then mis- classification will bias the results towards the null. We would, therefore, expect greater differences among our remoteness categories if we could adjust for this bias. Antibiotic Use During the study period, we surveyed 1,875 individuals about their antibiotic use in a population that averaged around 4,000 at any given time. On average, each sampled individual was surveyed 1.3 times over the study period, ranging from one to six times, also resulting in multiple measurements of each house- hold. A random effects analysis of these data supports our sampling strategy for added coverage across households rather than coverage over time (see electronic supplementary material). Among those individuals reporting use, the most fre- quently named antibiotics were amoxicillin (20% of antibiotics mentioned), amp (13%), sulphamethoxazole/trimethoprim (8%) and ciprofloxacin (8%) (electronic supplementary material, table S1). In the analysis presented in this manuscript, we restrict focus only to drugs that select for ampâsxt resistance. In addition to its constituent drugs, we also include amoxicillin and benzylpenicillin. These are in the family of beta lactams, and therefore their use potentially selects for amp resistance. Over the 5 years of collecting survey data across the region, the aver- age use rate was 0.05 per individual per week. Assuming use is evenly distributed throughout the week, this corresponds to a rate of 0.006 per individual per day, with an associated 95% CI of (0.003, 0.010). This rate per day is used in our subsequent simulation studies. There was no relationship between antibiotic use and remoteness at the community level (see electronic supplementary material, figure S2).
244 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS A Transmission Perspective on the Observed Antibiotic Resistance Patterns We use a transmission model to examine how the interaction among antibi- otic use, transmission rates of antibiotic resistant E. coli, and introduction rates of antibiotic resistant E. coli into villages affect the community-level AR patterns that we observed. As described in the transmission model, the transmission and introduction rates vary by remoteness, whereas antibiotic use does not. Our trans- mission model analysis suggests that the level of antibiotic use determines which factors explain the risk ratio of AR prevalence when comparing a close village with a far village: the ratio of transmission rates (close versus far) and/or the ratio of introduction rates (close versus far). This result is shown using contour plots of the risk ratio as a function of both the transmission rate and introduction rate ratios for both low and high antibiotic use rates (see electronic supplementary material, figure S3). To examine the marginal effects of transmission ratio and antibiotic use rate, we integrate out the introduction rate by calculating the geometric mean of the observed risk ratios across all introduction rate values (Figure A10-3). This FIGURE A10-3â The risk ratio of AR prevalence comparing a non-remote village (close) with a remote village (far) as a function of the ratio of transmission rates for close versus far villages. Each plot is for a different antibiotic use rate (Ï) ranging from 0.001 to 0.01 antibiotics per person per day. The transmission rate of the remote village is 0.154 (see text for justification). See Figure A10-2 for remaining parameter values. Circles with solid line, Ï = 0.001; squares with solid line, Ï = 0.002; triangles with solid line, Ï = 0.003; asterisks with solid line, Ï = 0.006; diamonds with solid line, Ï = 0.01.
APPENDIX A 245 is virtually identical to risk ratios corresponding to fixing the introduction rate ratio to its midpoint value of two. Figure A10-3, therefore, presents a plot of the effect of the ratio of transmission rates in close versus far communities on the risk ratio for sxtâamp resistance in close versus far communities for various antibiotic use rates to display the interaction between use rate and transmission ratio (Fig- ure A10-3). For extremely low antibiotic use rates (e.g. Ï = 0.001 per day), the transmission rate ratio has little effect on the risk ratio; i.e. given little selection pressure on AR in the village, transmission cannot amplify the prevalence levels. Under this scenario, the prevalence differences among villages can be attribut- able to differences in the introduction rates of resistant bacteria. The transition from no relationship to a very strong relationship between the transmission ratio and risk ratio can be seen as Ï increases. As this happens, the transmission rate ratio becomes the predominant determinant of the risk ratio; i.e. antibiotic use selects for AR and resistant bacteria spread throughout the villages via transmis- sion pathways. It appears that in our study region, AR prevalence is most sensi- tive to changes in the transmission rate ratio. This conclusion is based on our site-specific estimates of: (i) Ï (0.003 to 0.01); (ii) the ratio of the transmission rate, β, comparing close versus far villages (2.11); and (iii) the risk ratio of AR prevalence (1.8 [1.3, 2.3]). Discussion Roads have important impacts on social and ecological processes that in turn have impacts on health (Birley et al., 1998). The relationship between roads and disease has been examined for a variety of infectious diseases includ- ing HIV, malaria, dengue, and diarrhoeal disease (Carswell, 1987; Dutta et al., 1998; Eisenberg et al., 2006; Vittor et al., 2006). Here, we provide data from a 5-year regional-scale observational study showing that roads can also impact the spread of resistant bacteria. Focusing on E. coli resistance to ampâsxt, the most common pairing of antibiotics observed, we found a higher prevalence of antibiotic-resistant bacteria in villages along the road compared with more remote villages. These results are consistent with those of other researchers, who have found higher levels of AR organisms in sites with greater anthropogenic influence (Bartoloni et al., 2009; Pallecchi et al., 2008; Pei et al., 2006; Walson et al., 2001). However, we found no relationship between antibiotic use and remote- ness, which probably relates to the presence of both governmental and non- governmental organizations that deliver medical care, including antibiotics, throughout the region. Given its homogeneous distribution along the remoteness gradient, we employed a village-level transmission model to better understand how antibiotic use impacts prevalence patterns at a regional scale. Our model analysis suggests that at the regional-scale individual antibiotic use serves to modify the effect of two potentially important processes: the transmission of
246 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS E. coli from person to person mediated through environmental pathways, and the introduction of E. coli from outside the region owing to the movement patterns of people into and out of the region (Eisenberg et al., 2006). As antibiotic use rates decrease across the region, the differential rate of introduction becomes a more important determinant of our observed prevalence patterns. Transmission becomes an important determinant when antibiotic use increases; i.e. antibiotic use amplifies transmission. Thus, antibiotic use has a regional-scale impact that differs from those impacts that are derived from only considering the individual- level scale. At the individual scale, experimental evidence suggests that resistant bac- teria can be out-competed by their sensitive counterparts (Zeitouni and Kempf, 2011). The implication of this is that once the pressure of antibiotics is removed, the population of resistant bacteria may decrease relatively quickly, making an individualâs antibiotic use act primarily as a main effect on his/her probability of colonization with a resistant strain. However, at the community level, the effect of antibiotic use is more complex. Evidence suggests that the fitness costs of resistance can be very low (Andersson and Levin, 1999; Lenski, 1998; Schrag et al., 1997), and therefore the subsequent slow decline in the prevalence of re- sistant bacteria once the antibiotic use ceases provides continued opportunity for resistant organisms to spread from host to host, from host to the environment, and from the environment to the host. Therefore, interplay between antibiotic use, disease transmission rate, and rate of introduction from the environment must be considered when characterizing drivers of population-level prevalence of resistant bacteria. Our analysis suggests that the antibiotic use rate acts to modify the impact of the transmission rate and outside introduction rate, indicating that the effect of antibiotic use rate on community-level prevalence cannot be thought of in isolation. When antibiotic use is high (e.g. Ï = 0.01, antibiotics per person per day), the bacteria resistant to the antibiotic being used is selected for within the individual, thereby making it more likely for a transmission event to involve a resistant organism. Under these conditions, transmission becomes a major driver of AR prevalence, with outside introduction having a comparatively very small effect. When antibiotic use is low (e.g. Ï = 0.001, antibiotics per person per day), most transmission events involve sensitive bacteria, rendering the transmission rate impotent as a driver of AR prevalence (Figure A10-3). In this setting, oral exposure, which occurs through ingestion of bacteria into the gastrointestinal tract, is the primary driver of prevalence; this exposure comes from a variety of sources including introduction from outside the region. Many studies have demonstrated that AR can spread between individuals sharing the same home (Miller et al., 1996), day care centre (Fornasini et al., 1992; Reves et al., 1990), or even community (Mlander et al., 2000). For enteric organisms both transmis- sion and outside introduction occurs through water, sanitation, hygiene, and food
APPENDIX A 247 pathwaysâmodes of spread especially strong in agricultural settings (Marshal et al., 1990) and developing countries (Calva and Bojalil, 1996). The transmis- sion of bacteria can occur through these pathways in developed countries as well, albeit at lower rates. Typical models of AR are set in controlled environments such as hospitals, and focus on the competitive advantage given to resistant bacteria through an- tibiotic use. In such models, invasion of resistant bacteria from the outside is ignored, potentially because the focus of hospital settings is on the large amounts of antibiotic use and how they are optimally prescribed (e.g., Bonhoeffer et al., 1997). On the other hand, in a community setting, the invasion and the spread of resistant bacteria are an important determinant of prevalence. The inclusion of the rate of introduction of antibiotics and its interaction with transmission and antibiotic use, therefore, is a central piece of our analysis. The complete understanding of the dynamics of AR spread in the context of social and ecological changes can only be obtained through a systematic and ecological perspective as presented in this study. Our data and analysis support the proposal that understanding the mechanisms of the evolution and the spread of resistant bacteria require a consideration of the ecological dynamics that shape microbial population structure (Singer et al., 2006). These dynamics are mediated through factors that determine selection pressures, routes of transmission and the invasion of resistant bacteria (Singer et al., 2006), which may overwhelm the direct effects of individual antibiotic use in determining the emergence and dissemination of AR across communities or regions. In our study region, the major driver of selection pressure and routes of transmission appears to be a new network of roads, which have strong influence on the social and ecological environment and in turn on the health of communities (Airey, 1992; Coimbra, 1988; Dutta et al., 1998; Vittor et al., 2006). Roads may affect the evolution and the spread of resistant bacteria by influencing the use of antibiotics in the human population, changing hygiene and sanitation, and introducing resistant bacteria when people travel or migrate into a region. Acknowledgements The authors of this paper would like to thank the Ecologia, Desarrollo, Salud, y Sociedad (EcoDESS) field team for their invaluable contribution collecting the data, as well as Darlene Bhavnani for her helpful comments on the dataset and manuscript. This study was supported by grant number RO1-AI050038 from the National Institute of Allergy and Infectious Diseases (NIAID), and grant number 0811934 from the Ecology of Infectious Diseases programme, Fogarty Interna- tional Centre (FIC) of the National Institutes of Health (NIH) and the National Science Foundation (NSF).
248 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS References Airey T. 1992 The impact of road construction on the spatial characteristics of hospital utilization in the Meru district of Kenya. Soc. Sci. Med. 34, 1135â1146. Anderson R. M., May R. 1991 Infectious diseases of humans: dynamics and control. New York, NY: Oxford University Press. Andersson D. I., Levin B. R. 1999 The biological cost of antibiotic resistance. Curr. Opin. Microbiol. 2, 489â493. Austin D. J., Bonten M. J., Weinstein R. A., Slaughter S., Anderson R. M. 1999 Vancomycin-resistant enterococci in intensive-care hospital settings: transmission dynamics, persistence, and the impact of infection control programs. Proc. Natl Acad. Sci. USA 96, 6908â6913. Austin D. J., Kakehashi M., Anderson R. M. 1997 The transmission dynamics of antibiotic-resistant bacteria: the relationship between resistance in commensal organisms and antibiotic consump- tion. Proc. R. Soc. Lond. B 264, 1629-1638. Bartoloni A., et al. 2009 Antibiotic resistance in a very remote Amazonas community. Int. J. Antimi- crobiol. Agents 33, 125â129. Bauer A. A., Kirby W. M., Sherris J. C., Turck M. 1966 Antibiotic susceptibility testing by a standard- ized single disk method. Am. J. Clin. Path. 45, 493â496. Birley M. H., et al. 1998 Health and environmental impact assessment: an integrated approach, pp. 243. London, UK: Earthscan/British Medical Association. Blake D. P., Hillman K., Fenlon D. R., Low J. C. 2003 Transfer of antibiotic resistance between commensal and pathogenic members of the Enterobacteriaceae under ileal conditions. J. Appl. Microbiol. 95,428â436. Bonhoeffer S., Lipsitch M., Levin B. R. 1997 Evaluating treatment protocols to prevent antibiotic resistance. Proc. Natl Acad. Sci. USA 94,12 106â12 111. Bonten M. J., Austin D. J., Lipsitch M. 2001 Understanding the spread of antibiotic resistant patho- gens in hospitals: mathematical models as tools for control. Clin. Infect. Dis. 33, 1739â1746. Bonten M. J., Slaughter S., Ambergen A. W., Hayden M. K.,van Voorhis J., Nathan C., Weinstein R. A. 1998 The role of âcolonization pressureâ in the spread of vancomycin-resistant enterococci: an important infection control variable. Arch. Intern. Med. 158, 1127â1132. Bootsma M. C., Diekmann O., Bonten M. J.2006 Controlling methicillin-resistant Staphylococcus aureus: quantifying the effects of interventions and rapid diagnostic testing. Proc. Natl Acad. Sci. USA 103,5620â5625. Calva J., Bojalil R. 1996 Antibiotic use in a periurban community in Mexico: a household and drug- store survey. Soc. Sci. Med. 42,1121â1128. Carswell J. W. 1987 HIV infection in healthy persons in Uganda. AIDS 1, 223â227. Coimbra C. E. A. 1988 Human factors in the epidemiology of malaria in the Brazilian Amazon. Hu- man Organ. 47, 254â260. Collignon P., Powers J. H., Chiller T. M., Aidara-Kane A.,Aarestrup F. M. 2009 World Health Orga- nization ranking of antimicrobials according to their importance in human medicine: a critical step for developing risk management strategies for the use of antimicrobials in food production animals. Clin. Infect. Dis. 49, 132â141. Drusano G. L. 2003 Prevention of resistance: a goal for dose selection for antimicrobial agents. Clin. Infect. Dis. 36 (Suppl. 1), S42âS50. Dutta P., et al. 1998 Distribution of potential dengue vectors in major townships along the national highways and trunk roads of northeast India. Southeast Asian J. Trop. Med. Public Health 29, 173â176. Efron B., Tibshirani R. 1993 An introduction to the bootstrap. Boca Raton, FL: Chapman & Hall/ CRC. Eisenberg J. N., et al. 2006 Environmental change and infectious disease: how new roads affect the transmission of diarrheal pathogens in rural Ecuador. Proc. Natl Acad. Sci. USA 103, 19 460â19 465.
APPENDIX A 249 Fornasini M., Reves R. R., Murray B. E., Morrow A. L., Pickering L. K. 1992 Trimethoprim-resistant Escherichia coli in households of children attending day care centers. J. Infect. Dis. 166, 326â330. Foster S. D. 2010 The economic burden of antibiotic resistanceâevidence from three recent stud- ies. In Annual Conf. on Antimicrobial Resistance, Bethesda, MD. Washington, DC: National Academy Press. Hogberg L. D., Heddini A., Cars O. 2010 The global need for effective antibiotics: challenges and recent advances. Trends Pharmacol. Sci. 31, 509â515. Institute of Medicine. 2003 The resistance phenomenon in microbes and infectious disease vectors: implications for human health and strategies for containment. In Institute of medicine forum on emerging infections (eds S. L. Knobler, S. M. Lemon, M. Najafi & T. Burroughs). Washington, DC: National Academy Press. Lenski R. E. 1998 Bacterial evolution and the cost of antibiotic resistance. Int. Microbiol. 1, 265â270. Lipsitch M., Bergstrom C. T., Levin B. R. 2000 The epidemiology of antibiotic resistance in hospitals: paradoxes and prescriptions. Proc. Natl Acad. Sci. USA 97, 1938â1943. Lohr S. L. 1999 Sampling: design and analysis. Pacific Grove, CA: Duxbury Press. Love D. C., Davis M. F., Bassett A., Gunther A., Nachman K. E. 2011 Dose imprecision and resis- tance: free-choice medicated feeds in industrial food animal production in the United States. Environ. Health Perspect. 119, 279â283. MacLean R. C., Hall A. R., Perron G. G., Buckling A. 2010 The population genetics of antibiotic resistance: integrating molecular mechanisms and treatment contexts. Nat. Rev. Genet. 11, 405â414. Marshall B., Petrowski D., Levy S. B. 1990 Inter- and intraspecies spread of Escherichia coli in a farm environment in the absence of antibiotic usage. Proc. Natl Acad. Sci. USA 87, 6609â6613. Massad E., Lundberg S., Yang H. M.1993 Modeling and simulating the evolution of resistance against antibiotics. Intern. J. Bio-med. Comp. 33, 65â81. Melander E., Ekdahl K., Jonsson G., Molstad S. 2000 Frequency of penicillin-resistant pneumococci in children is correlated to community utilization of antibiotics. Pediatr. Infect. Dis. J. 19, 1172â1177. Miller Y. W., et al. 1996 Sequential antibiotic therapy for acne promotes the carriage of resistant staphylococci on the skin of contacts. J. Antimicrob. Chemother. 38, 829â837. Miriagou V., Tzelepi E., Gianneli D., Tzouvelekis L. S.2003 Escherichia coli with a self-transferable, multiresistant plasmid coding for metallo-beta-lactamase VIM-1. Antimicrob. Agents Che- mother. 47, 395â397. Myers N., Mittermeier R. A., Mittermeier C. G., da Fonseca G. A., Kent J. 2000 Biodiversity hotspots for conservation priorities. Nature 403,853â858. Novais A., Canton R., Valverde A., Machado E., Galan J.C., Peixe L.,Carattoli A., Baquero F., Coque T. M. 2006 Dissemination and persistence of blaCTX-M-9 are linked to class 1 integrons containing CR1 associated with defective transposon derivatives from Tn402 located in early antibiotic resistance plasmids of IncHI2, IncP1-alpha, and IncFI groups. Antimicrob. Agents Chemother. 50, 2741â2750. Pallecchi L., Bartoloni A., Paradisi F., Rossolini G. M. 2008 Antibiotic resistance in the absence of antimicrobial use: mechanisms and implications. Expert Rev. Anti Infect. Ther. 6, 725â732. Partridge S. R., Tsafnat G., Coiera E., Iredell J. R. 2009 Gene cassettes and cassette arrays in mobile resistance integrons. FEMS Microbiol. Rev. 33, 757â784. Pei R., Kim S. C., Carlson K. H., Pruden A. 2006 Effect of river landscape on the sediment con- centrations of antibiotics and corresponding antibiotic resistance genes (ARG). Water Res. 40, 2427â2435. Reves R. R., et al. 1990 Risk factors for fecal colonization with trimethoprim-resistant and multi- resistant Escherichia coli among children in day-care centers in Houston, Texas. Antimicrob. Agents Chemother. 34, 1429â1434.
250 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Robin F., Aggoune-Khinache N., Delmas J., Naim M., Bonnet R. 2010 Novel VIM metallo-β- lactamase variant from clinical isolates of Enterobacteriaceae from Algeria. Antimicrob. Agents Chemother. 54,466â470. Schrag S. J., Perrot V., Levin B. R. 1997 Adaptation to the fitness costs of antibiotic resistance in Escherichia coli. Proc. R. Soc. Lond. B 264, 1287â1291. Singer R. S., Ward M. P., Maldonado G. 2006 Opinionâcan landscape ecology untangle the complex- ity of antibiotic resistance? Nat. Rev. Microbiol. 4, 943â952. Smith D. L., Dushoff J., Morris J. G. Jr. 2005 Agricultural antibiotics and human health: does antibi- otic use in agriculture have a greater impact than hospital use? PLoS Med. 2, 731â735. Smith D. L., Dushoff J., Perencevich E. N., Harris A. D., Levin S. A. 2004 Persistent colonization and the spread of antibiotic resistance in nosocomial pathogens: resistance is a regional problem. Proc. Natl Acad. Sci. USA 101, 3709â3714. Smith D. L., Harris A. D., Johnson J. A., Silbergeld E. K., Morris J. G. Jr. 2002a Animal antibiotic use has an early but important impact on the emergence of antibiotic resistance in human com- mensal bacteria. Proc. Natl Acad. Sci. USA 99, 6434â6439. Smith D. L., Lucey B., Waller L. A., Childs J. E., Real L. A. 2002b Predicting the spatial dynamics of rabies epidemics on heterogeneous landscapes. Proc. Natl Acad. Sci. USA 99, 3668â3672. Vittor A. Y., et al. 2006 The effect of deforestation on the human-biting rate of Anopheles darlingi, the primary vector of falciparum malaria in the Peruvian Amazon. Am. J. Trop. Med. Hyg. 74, 3â11. Walson J. L., Marshall B., Pokhrel B. M., Kafle K. K., Levy S. B. 2001 Carriage of antibiotic-resistant fecal bacteria in Nepal reflects proximity to Kathmandu. J. Infect. Dis. 184, 1163â1169. Whitten N. E. Jr. 1965 Class, kinship and power in an Ecuadorian town: the negroes of San Lorenzo. Stanford, CA: Stanford University Press. Woodford N., Carattoli A., Karisik E., Underwood A., Ellington M. J., Livermore D. M. 2009 Com- plete nucleotide sequences of plasmids pEK204, pEK499, and pEK516, encoding CTX-M enzymes in three major Escherichia coli lineages from the United Kingdom, all belonging to the international O25:H4-ST131 clone. Antimicrob. Agents Chemother. 53, 4472â4482. Zeitouni S., Kempf I. 2011 Fitness cost of fluoroquinolone resistance in Campylobacter coli and Campylobacter jejuni. Microb. Drug Resist. 17,171â179.
APPENDIX A 251 A11 SOCIAL CONNECTEDNESS CAN INHIBIT DISEASE TRANSMISSION: SOCIAL ORGANIZATION, COHESION, VILLAGE CONTEXT, AND INFECTION RISK IN RURAL ECUADOR40 Jonathan L. Zelner,41 James Trostle,42 Jason E. Goldstick,43 William Cevallos,44 James S. House,45 and Joseph N. S. Eisenberg46 Abstract Social networks are typically seen as conduits for the spread of disease and disease risk factors. However, social relationships also reduce the inci- dence of chronic disease and potentially infectious diseases. Seldom are these opposing effects considered simultaneously. We have shown how and why diarrheal disease spreads more slowly to and in rural Ecuadorian villages that are more remote from the areaâs population center. Reduced contact with outside individuals partially accounts for remote villagesâ relatively lower prevalence of diarrheal disease. But equally or more important is the greater density of social ties between individuals in remote communities, which facilitates the spread of individual and collective practices that reduce the transmission of diarrheal disease. Introduction Studies of the transmission of infectious diseases (Jolly et al., 2001; Klovdahl et al., 2001) often use social networks as maps of direct contact that facilitate person-to-person transmission of pathogens. From this perspective, relationships are increasingly associated with greater individual-level risk (Newman, 2002). The social cohesion and organization embodied in networks is, however, also critical to the functioning of communities (Hunt and Hunt, 1976; Pahl-Wostl 40ââReprinted with kind permission from the American Public Health Association. Originally printed as Zelner et al. 2012. Social connectedness can inhibit disease transmission: Social organization, cohesion, village context, and infection risk in rural Ecuador. American Journal of Public Health 102(12):2233-2239. 41ââDept. of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544; Research and Policy for Infectious Disease Dynamics (RAPIDD) Program, Fogarty International Center, Na- tional Institutes of Health, Bethesda, MD 20892. 42ââDept. of Anthropology, Trinity College, Hartford CT, 06016. 43ââDept. of Epidemiology School of Public Health, University of Michigan, Ann Arbor, MI 48109. 44ââDepartment of Microbiology. Universidad San Francisco de Quito, Quito Ecuador. 45ââInstitute for Social Research; Gerald R. Ford School of Public Policy; Dept. of Sociology, University of Michigan, Ann Arbor, MI 48109. 46ââDept. of Epidemiology School of Public Health, University of Michigan, Ann Arbor, MI 48109.
252 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS et al., 2007; Wallace, 1988), but researchers typically neglect the influence of these factors on community-level infectious disease risk. Social relationships have long been employed as contacts in transmission models (Aparicio and Pascual, 2007; Bansal et al., 2008; Klovdahl et al., 2001; Meyers et al., 2003) and as protective factors for chronic disease (Berkman and Glass, 2000; House et al., 1988). However, outside the literature on sexually transmitted diseases (Cohen et al., 2007; Holtgrave and Crosby, 2003) there are few examples of the protective role of social relationships in the epidemiology of infectious diseases (Cohen et al., 2003). Yet individuals in strongly connected, socially cohesive communities are more likely to perceive economic and social interests as shared. Consequently, they may be more motivated and better orga- nized to pursue collective goals such as building and maintaining effective water and sanitary infrastructure (Entwisle et al., 2007). This means that understanding infectious disease risk at the community level requires understanding not only how certain social networks may spread disease but also how other social networks may influence the infrastructure and behavior that can prevent population-level exposure. We examined 2 types of social networks from the same set of villages to test the hypothesis that increased social network connectedness predicts diminished risk of diarrheal illness, using a sample of 18 villages in rural, northern coastal Ecuador. Figure A11-1 illustrates our conceptual model. We sought to measure specific risk and protective effects of social relation- ships via survey and social network analysis methods. In the first part of the analysis, we examined the association of village social networks and different routes of exposure to self-reported illness. In the remainder of the analysis, we attempted to explain these associations in terms of factors that affect village so- cial networks (e.g., remoteness) and the mechanisms by which increased social cohesion is linked to diminished illness risk (e.g., improved water sanitation, education). FIGURE A11-1â Postulated conceptual model: Effects of social relationships on disease outcomes, Esmeraldas, Ecuador, 2007. NOTE: Solid arrows illustrate the hypothesized pathway by which remoteness impacts risk of infection. Arrows are marked by + or â to indicate the directionality of the relationship.
APPENDIX A 253 A road was recently built that connects some of these villages to the nearest large town, which has about 5000 inhabitants. Consequently, these villages now vary in their remoteness, measured by distance and time of travel to this trading center. Our previous analysis suggested that increasing remoteness is associated with increasing average degree in village social networks and that increasing average degree is associated with decreased prevalence of diarrheal disease (Trostle et al., 2008). Additionally, the connectivity of villages to communities in and outside the study region decreases with remoteness (Eisenberg et al., 2006). Consequently, less remote villages have more transient inhabitants and are more socially fragmented and therefore may be less able to build and maintain the water and sanitation infrastructure and promote hygiene practices than are more remote villages. We explicitly tested the relationships among these components, as described in Figure A11-1. We defined a contact network as a network comprising relationships that are likely to facilitate transmission of pathogens, that is, a structure of connec- tions through which an individual, denoted âego,â may infect or be infected by his or her network neighbors, denoted âalters.â This network contains all the pathways an infection may follow through the community via direct human contact. In contrast to contact networks, we defined links in sociality networks as connections between people that represent specific types of social engage- ment. Connections in sociality networks can correspond to casual acquaintance, close friendship and trust, or economic exchange. The presence or absence of these relationships affects infection risk because they often determine whether communities have effective sanitary infrastructure and health services. In this way, more network connections (e.g., friends) may indicate protective social support, instead of increasing exposure, as in a contact-only network (Christley et al., 2005). Community Social Structure and Risk Understanding how sociality networks influence infection risk in these vil- lages required us to answer the question of how social organization and action can inhibit or enhance pathogen transmission via the environment. Figure A11-1 illustrates the mechanism by which we posit that this occurs. Poor quality sanitary infrastructure is a leading cause of infection by enteric pathogens such as cholera (Checkley et al., 2004; Rego et al., 2005; Tumwine et al., 2002), and such infra- structure is usually a public good that requires ongoing funding and management by the community. Transmission of many enteric pathogens is often conceptual- ized as person-to-environment-to-person, with water acting as the environmental reservoir (Koelle and Pascual, 2004). Greater community cohesion may facilitate better overall water quality through the support of community education pro- grams that impart knowledge of household sanitary practices, such as water filtration, and social organization that produces infrastructural improvements,
254 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS such as sewage treatment. Alternatively, improved water quality or sanitation may result from the establishment of social norms and the reinforcement of those norms. If this is true, we would expect to find that the average number of social network connections in a village and risk of infection by enteric pathogens are inversely related. For example, if ego has many relationships (i.e., has high degree) in her or his village sociality network and belongs to a community organization focused on improving local water quality, it may help reduce the entire villageâs exposure to pathogens. Although these social relationships can also be transmission path- ways, the salutary effects of egoâs social engagement may preempt transmission via those connections by reducing village-wide exposure to enteric pathogens in the first place. Measuring Effects of Sociality and Contact Networks on Risk We analyzed our illness data with respect to 2 networks. The first network comprised individuals, excluding egoâs household members, with whom ego reported having spent time in the previous week. This is called the âpassing time network.â We used this inclusive definition of contact because a wide range of casual and close contacts can transmit gastrointestinal pathogens (Anderson and May, 1992). In addition to being conceptualized as a contact network, the passing time network may represent sociality in a village. This definition of a sociality relationship highlights many connections between people in the community, without capturing fine-grained social structures. If a widespread, but not neces- sarily strong, level of attachment to the community is sufficient to stimulate social organization and diffuse information that can reduce infection risk, we would expect that greater average degree in the passing time network would predict diminished risk. An alternate approach is to constrain membership in the sociality network to relationships corresponding to the question âOutside of members of your household, with whom can you talk about important matters?â This is the second network we used in our analysis, which we call the âimportant matters network.â This network typically contains fewer individuals than does the passing time net- work, but it may better expose the essential structure of the community. If attach- ment to the community stronger than that implied by the passing time network is necessary to reduce illness risk, relationships in the important matters network should be better predictors of risk than should those in the passing time network. By comparing results from both networks, we were better able to understand how the nature of relationships in the sociality network affected risk. Our analysis of sociality conceptualizes risk in terms of the networkâs village-level features and egoâs position in this village-wide network. By con- trast, the analysis of contact focuses on egoâs risk of infection by ill individuals in his or her household and contact network. This approach, therefore, allowed us
APPENDIX A 255 to examine the separate effects of the contact and sociological aspects of social relationships on disease outcomes. Methods We collected our data in 18 villages in the northern coastal Ecuadorian prov- ince of Esmeraldas. These villages are situated along 3 rivers: Cayapas, Santiago, and Ãnzole, all of which drain toward Borbón, which is the major population center of the region. In 1996, a new paved road was built westward from Borbón to the coast, and in 2001 a road connecting Borbón to the Andes was completed. A network of smaller roads linking villages to the main road is under continual construction. These villages vary by remoteness, a function of time and cost of travel to Borbón (for a map of the study region, see Eisenberg et al., 2006). Remoteness influences social relationships and network structure, migration into and out of the region, and other factors that affect both social network character- istics and exposure to infectious diseases. Outcome Measure: Recent Infectious Illness Our outcome measure is egoâs self-reported diarrheal disease or fever in the week before the survey. Diarrheal illness is defined as having 3 or more liquid stools in 1 day (WHO, 1988). Our initial analyses performed with each outcome in a separate model yielded broadly similar risk factors, so we combined these 2 categories of illness into a single binary response variable. The outcome variable was â1â if the individual had experienced either diarrhea or fever, indicating the individual had recently experienced illness that was likely of infectious origin. Measuring Community Cohesion and Household Attachment We took several approaches to measuring social cohesion and organization, utilizing data on the structure of community social networks, education, and participation in community organizations. We measured the average number of relationships in the sociality network for individuals aged 13 years and older. As the number of connections per person grows, the cohesion of the community is expected to grow as well (Bates et al., 2007; Trostle et al., 2008). Unless otherwise noted, we measured this quantity in 1-unit increments. Because the effects of social connectedness in villages affect household hygiene and water quality, we expected to see the salutary effects of cohesion at the household level. Because of this, we measured the effect of sociality (pass- ing time or important matters) network degree on risk using the sociality degree of the most connected individual in egoâs household, which we defined as egoâs household degree. We standardized each villageâs distribution of household
256 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS degree to have mean zero and unit variance, and we have presented household degree in SD units from village mean household degree. This allowed us to measure the effect of egoâs household social connectedness relative to the average household in her or his village. We performed data processing with Python 2.7 and social network analysis using igraph 0.5.4 for Python (http:// igraph.sf.net). Other Covariates Village remoteness is a composite of time and cost of travel to Borbón, the commercial center in the region. We normalized this scale so that the closest village had a remoteness value of zero and the most remote village had remote- ness equal to 1. For additional information on the construction of this scale, see Eisenberg et al., 2006. Contact network exposure is the number of alters in an egoâs contact (passing time) network that reported symptoms in the previous week. Household exposure is the number of individuals in an egoâs household reporting symptoms in the previous week. We also included several measures of village and household sanitation and hygiene: (1) observed hygiene is the average of 23 indicators of household clean- liness across all households in the village, (2) improved sanitation is the propor- tion of households in the village with a septic tank or a latrine, (3) improved water source measures the proportion of households using piped water or collected rainwater, and (4) water treatment measures the proportion of households in the community reporting that they used some kind of water treatment. For values of these measures by village, see Table C (available as a supplement to the online version of this article at http://www.ajph.org). In addition to these factors, we accounted for individual and village demo- graphics, contact with individuals outside the village, household wealth, and education. For information on the calculation of these covariates, see the supple- mentary materials (available as a supplement to the online version of this article at http://www.ajph.org). Modeling Risk for Individuals Nested in Communities Because we conceptualized individual outcomes as being influenced by potentially unobserved village-level factors, we expected that responses in a vil- lage would be correlated. We dealt with this correlation in all regression models by using mixed-effects logistic regression models with village-level random intercepts and estimates of individual-level fixed effects for all covariates over all villages (Laird and Ware, 1982; Stiratelli et al., 1984). We fitted all mixed- effects logistic regression models to data using the lme4 package in R 2.15 (http:// lme4.r-forge.r-project.org).
APPENDIX A 257 Indirect Effects of Village-Level Characteristics on Individual Risk Village remoteness and sociality networks do not directly affect disease but instead act through (or are mediated by) more immediate factors (e.g., sanita- tion), as illustrated by Figure A11-1. Because quantifying such indirect effects through the difference of regression coefficients (Baron and Kenny, 1986) is not readily extended to binary response variables, we used an alternate approach. We estimated the indirect effect as the difference between the total association of remoteness with illness and the residual direct association of remoteness and illness, adjusted for the proximal variable. These are quantified by the ratios of the expected probability of illness for individuals in far versus near villages, with and without the mediator in the model. We assessed statistical significance of this effect using a nonparametric boot- strap; we set the threshold for statistical significance at P < .05. Positive values of indirect effect indicate mediation and can be interpreted as the change in the risk ratio comparing far and near villages when the mediator is taken into account. We repeated this analysis to estimate the mediation of average village-level degree. For a detailed discussion of this analysis, see the supplementary materials (avail- able as a supplement to the online version of this article at http://www.ajph.org). Results Our data set consisted of 3413 cases obtained in a census with a greater than 95% response rate in the 18 villages in our analysis. To facilitate comparisons between different models using the Akaike information criterion, we included in our analysis only the 2912 (85%) individuals with complete observations for all social network, illness, and sanitation variables. Village-level descriptive statistics for remoteness, illness, water sanitation, water quality, and household hygiene appear in Table A11-1, with villages listed in order of increasing re- moteness. Descriptive characteristics of the important matters and passing time networks for each village include average degree and the global clustering coef- ficient (Table A, available as a supplement to the online version of this article at http://www.ajph.org). Additional village-level descriptive statistics on organization membership, education, and wealth are available in Table B (available as a supplement to the online version of this article at http://www.ajph.org). We used logistic regression models to examine the effects of exposures (contact outside villages, in households, and in social networks), household and village-level social network characteristics (degree), village-wide socioeconomic status (wealth, education), and social capital (membership in community organi- zations) on illness (whether a person had fever or diarrhea; Table A11-2). Model 1 (Akaike information criterion = 2110) shows risk associated with routes of expo- sure, adjusted for age and village size. This model shows that (1) a 10% increase in the proportion of households with visitors from outside the community in the
TABLE A11-1â Descriptive Characteristics of Villages: Effects of Social Relationships on Disease Outcomes, Esmeraldas, 258 Ecuador, 2007 Fever or Households Households Households Observed Sample diarrheal w/water w/improved w/improved household Remoteness size disease treatment sanitation water source hygiene index Village Continuous Category N Cases/100 % % % Mean 1 0.06 Close 158 15 25 43 43 0.64 2 0.07 Close 642 16 74 33 49 0.70 3 0.13 Close 407 13 18 55 59 0.69 4 0.20 Medium 110 11 14 61 7 0.69 5 0.20 Medium 41 14 0 64 15 0.63 6 0.20 Medium 30 23 93 11 2 0.53 7 0.25 Medium 49 8 33 100 0 0.79 8 0.25 Medium 37 30 72 55 100 0.51 9 0.31 Medium 101 12 0 15 0 0.45 10 0.40 Medium 64 15 0 26 100 0.68 11 0.57 Medium 89 18 23 50 77 0.71 12 0.62 Medium 119 19 19 7 19 0.31 13 0.71 Far 62 10 13 52 48 0.38 14 0.78 Far 185 8 33 55 55 0.71 15 0.80 Far 71 0 15 86 99 0.74 16 0.83 Far 285 8 0 41 82 0.73 17 0.96 Far 324 6 13 56 64 0.73 18 1.00 Far 138 14 5 50 28 0.68 Total â â 2912 12.3 30.0 45.4 52.4 0.66
TABLE A11-2â Multivariate Models for Risk of Disease in Previous Week: Effects of Social Relationships on Disease Outcomes, Esmeraldas, Ecuador, 2007 Model 1, None, OR Model 2 Passing Model 3, Important Sociality Network Type (95% CI) Time, OR (95% CI) Matters, OR (95% CI) Demographics â Age, decades 0.90*** (0.84, 0.96) 0.90*** (0.84, 0.96) 0.90*** (0.85, 0.96) â Village size 1.11*** (1.03, 1.19) 1.05* (0.99, 1.10) 1.04* (0.99, 1.10) Ownership of material goods by household 0.86 (0.35, 2.12) 0.90 (0.37, 2.20) 0.86 (0.36, 2.09) Outside contact, % â Households with outside visitor 1.12* (1.00, 1.25) 1.10 (0.99, 1.22) 1.08 (0.97, 1.21) â Households with outside trip 1.03 (0.91, 1.16) 1.03 (0.92, 1.15) 0.96 (0.86, 1.08) Food-sharing exposure 0.84 (0.45, 1.56) 0.84 (0.45, 1.55) 0.89 (0.48, 1.66) In-household exposure â No. infected in household 1.59*** (1.41, 1.79) 1.55*** (1.37, 1.74) 1.54*** (1.36, 1.73) â Mean-centered household size 0.86*** (0.81, 0.90) 0.86*** (0.82, 0.91) 0.87*** (0.82, 0.92) Contact network exposure â No. infected alters in passing time network 0.91 (0.74, 1.11) 0.97 (0.80, 1.19) 0.95 (0.78, 1.16) Sociality network â Household degree 0.64 (0.37, 1.10) 0.59** (0.40, 0.85) â Average degree 0.89** (0.81, 0.98) 0.83** (0.72, 0.95) â Average degree à household degree 1.06 (0.96, 1.17) 1.17** (1.04, 1.32) â Graph clustering 1.18 (0.94, 1.48) 1.12 (0.89, 1.42) Goodness of fit âLog-likelihood â1045 â1038 â1037 â Akaike information criterion 2110 2107 2103 NOTE: CI = confidence interval; OR = odds ratio. *P ⤠.05; **P ⤠.01; ***P ⤠.005. 259
260 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS week before the survey predicted an increased risk of illness (odds ratio [OR] = 1.11; 95% confidence interval [CI] = 1.00, 1.25), (2) a 1-person increase in the number of ill individuals in egoâs household predicted increased risk of illness (OR = 1.59; 95% CI = 1.40, 1.78), and (3) a 1-person increase in the size of egoâs household was associated with diminished risk (OR = 0.86; 95% CI = 0.81, 0.91). Model 1 also shows no significant change in risk associated with a 1-individual increase in the number of ill alters in egoâs community contact network (OR = 0.91; 95% CI = 0.74, 1.11). For both networks, a 1-unit increase in average village-level degree, adjusted for household and village-level network characteristics, was associated with di- minished risk when household degree was fixed at its village mean (passing time: OR = 0.89; 95% CI = 0.81, 0.98; important matters: OR = 0.83; 95% CI = 0.72, 0.95). This translated into an adjusted reduction in risk of 45% or 48% between the least connected and most connected villages for the important matters and passing time networks, respectively. This protective effect remained unchanged in the absence of controls for the number of ill contacts in the community. The statistically significant interaction in model 3 between village average and household important matters degree (OR = 1.17; 95% CI = 1.02, 1.34) sug- gests that the protective effect of village-level average degree applied to house- holds with degree less than 0.6 SD above the village mean. Above this level, the associations become nonsignificant, and our data cannot resolve the relationship. This indicates that in villages with high average degree, individuals were always protected regardless of the degree of their household. But in villages where av- erage degree was lower, household degree became protective. This relationship is analogous to herd immunity obtained through high vaccine coverage. (For further discussion of this interaction see the supplement to the online version of this article at http://www.ajph.org.) As with average degree, residence in the most versus the least remote village in our sample was associated with a large decrease in egoâs unadjusted risk of infectious illness (OR = 0.49; 95% CI = 0.29, 0.84). As shown in Table A11-3, this effect can be explained by 4 statistically significant village-level mediators (P ⤠.05): the percentage of households with an outside visitor in the previous week (indirect effect = 0.058; P = .013), improved sanitation (indirect effect = 0.040; P = .011), improved water treatment (indirect effect = 0.072; P = .035), and egoâs household size (indirect effect = 0.014; P = .007). We also included average degree in the passing time network (indirect effect = 0.045; P = .051) as a mediator, as it has a strong theoretical link with remoteness and was close to our cutoff for statistical significance. To assess whether these 5 variables could fully explain the association be- tween remoteness and illness, we fit a logistic regression model predicting egoâs illness as a function of remoteness, household size, village average passing time degree, and improved sanitation and water treatment. In this model, the relation- ship between remoteness and illness was no longer significant, and the point
TABLE A11-3â Indirect Effects of Remoteness and Village-Level Average Degree on Risk of Illness: Effects of Social Relationships on Disease Outcomes, Esmeraldas, Ecuador, 2007 Average Passing Time Average Important Remoteness, Indirect Degree, Indirect Effect Matters Degree, Indirect Pathogen Exposure Effect (95% CI) (95% CI) Effect (95% CI) Outside contact, % âHouseholds with outside visitor 0.058** (0.008, 0.099) âHouseholds with outside trip 0.002 (â0.081, 0.121) In-household exposure, mean-centered household size 0.014** (0.004, 0.041) Wealth, ownership of material goods by household 0.006 (â0.015, 0.019) 0.031*** (0.005, 0.067) 0.010** (0.001, 0.032) Sociality network âAverage degree, important matters 0.078 (â0.067, 0.352) âAverage degree, passing time 0.045* (â0.011, 0.184) Mean village years of education â0.019 (â0.087, 0.050) 0.017** (0.000, 0.042) 0.020 (â0.041, 0.115) Participation in community organizations âMean no. of organization memberships in village â0.041 (â0.136, 0.026) â0.039 (â0.039, 0.151) â0.028 (â0.116, 0.093) âMax no. of organization memberships in household 0.000 (â0.003, 0.011) 0.002 (0.006, 0.012) 0.005 (â0.016, 0.016) Water quality and sanitation âObserved hygiene index 0.006 (â0.010, 0.019) 0.072*** (0.020, 0.142) 0.095 (â0.079, 0.377) âCommunity improved sanitation 0.040* (0.006, 0.094) 0.146*** (0.058, 0.443) 0.133** (0.033, 0.310) âCommunity water treatment 0.072* (â0.008, 0.203) 0.168** (0.029, 0.362) 0.004 (â0.009, 0.037) âCommunity water source â0.019 (â0.076, 0.031) â0.007 (â0.025, 0.010) 0.007 (â0.043, 0.076) NOTE: CI = confidence interval. Positive values indicate mediation. Relative strengths of mediation may be interpreted in terms of differences between values of indirect effect for different mediators of the same distal variable, e.g., remoteness. *P ⤠.05; **P ⤠.01; ***P ⤠.005. P values reflect proportion of bootstrapped values of indirect effect. 261
262 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS estimate was closer to the null (OR = 0.75; 95% CI = 0.37, 1.53), suggesting that these variables explain much of the variability in risk associated with remoteness and are likely important mediators linking remoteness to illness. The analysis of indirect effects thus far suggests that remoteness influences risk through village networks and more proximal water and sanitation factors. Further analysis showed that village-level social networks may also exert influ- ence on risk through a number of mechanisms. Improved community sanitation was the strongest mediator of the effect of both average important matters and passing time degree (important matters: indirect effect = 0.133; P = .003; passing time: indirect effect = 0.146; P ⤠.001), whereas community water treatment (in- direct effect = 0.168; P = .006) is the strongest mediator of passing time degree. Observed hygiene (indirect effect = 0.072; P = .002) and average village educa- tion (indirect effect = 0.017; P = .027) also mediated passing time degree. Addi- tionally, household ownership mediated the relationship between both important matters and passing time degree and risk (important matters: indirect effect = 0.010; P = .015; passing time: indirect effect = 0.031; P = .009). After adjusting for these mediator variables, we found that the effect of living in the village with the highest versus lowest average passing time degree (OR = 0.83; 95% CI = 0.35, 1.98) was nonsignificant and slightly closer to the null, whereas the relationship between average important matters degree (OR = 0.47; 95% CI = 0.26, 0.87) and illness was essentially unchanged. This finding sug- gests that the relationship between degree in the passing time network and risk can be largely explained by community sanitation, community water, observed hygiene, and household ownership. The relationship between degree and illness in the important matters network was not explained by these variables: our mea- sures of nonnetwork protective factors may not be sensitive to all the pathways by which important matters network degree was associated with decreased risk. Discussion Highly connected social networks are usually represented as efficient trans- mission systems (Newman, 2002). By contrast, we have shown how greater connectivity at the village level may inhibit the prevalence of self-reported diar- rheal disease and fever. When controlling for sources of exposure to illness, our analysis shows that increasing village-wide average degree is associated with decreasing risk for all households in the passing time network and for households of average degree or above in all village important matters networks. Our analysis also connects social network, water sanitation, and hygiene fac- tors to the social and environmental context in which the village is situated, that is, its remoteness. The processes of environmental change reflected by a villageâs remoteness occur over a long time. As a result, analyzing a cross-sectional slice of a group of villages in the same region that are at different stages of social and
APPENDIX A 263 environmental transformation provides insight into the effects of these long-term processes. We postulated that remoteness would affect risk through contact networks and village cohesion (Eisenberg et al., 2006). To test this, we analyzed the pro- tective effects of local social networks as indirect effects of remoteness. Results from this analysis agree with that theory, showing that more remote villages experience decreased risk not only because of a lower rate of contact with indi- viduals from outside but also because the average individual in them has more relationships in the village passing time network and lives in a larger household than does a comparable person in a less remote village. Further mediation analy- sis suggests that villages with high average degree experience decreased risk of illness through improved water quality and sanitation. The finding that individuals in larger households experienced decreased risk may be explained by the fact that increasing household size explains some of the protective effects of remoteness. Larger households may indicate more traditional, cohesive communities. This would be consistent with our finding that the protective effect of remoteness manifested at least partly through increased social cohesion. The finding that household wealth explains some of the relationship between average degree and risk for both the important matters and passing time networks highlights the potential for social capital and household ownership to be mutually reinforcing. However, household ownership was not an independent predictor of risk of illness when we adjusted for village-level attributes associated with remoteness, and the size of this mediation effect relative to measures of water sanitation quality was small, indicating that these effects do not confound the relationship between social cohesion and risk. Our conceptual model (Figure A11-1) posited that village remoteness was re- lated to reduced risk through increasing village social organization and cohesion. We postulated that strong social organization supports infrastructure and behavior that decrease disease prevalence. Because we conceptualized sanitation and hy- giene as village-level constructs, the relatively small number of villages in our sample made it difficult to directly test the hypothesis that water sanitation and hygiene are outcomes of village-level social cohesion. However, ethnographic observations and interviews in these villages have shown how these effects might be produced. For example, we have observed that remote villages tend to have higher and more frequent participation in meetings designed to disseminate health information, whereas factionalism in villages along the road reduces the likeli- hood that all community members will participate in the same meeting. Although we have identified 7 factors mediating distal risk factors and dis- ease, additional mechanisms clearly relate remoteness to risk. However, we have demonstrated that relationships in social networks can protect against waterborne disease and that there are important mechanisms by which these relationships
264 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS may decrease risk; the scope of this analysis was not to rule out all alternative mechanisms linking remoteness to risk. In addition to the protective effects of social organization we have out- lined, we found that migration between villages, measured by the proportion of households with a visitor from outside the village in the previous week, predicts increased risk of infection. This confirms previous findings from these villages (Eisenberg et al., 2006). Networks of social relationships can reduce the individual-level risk of ill- ness from infectious diseases by mitigating population-level exposures, thereby preempting person-to-person transmission over these networks. These results expand on the theory that social connectedness and support are important predic- tors of chronic illness and mortality (House, 2002; Klinenberg, 2002) as well as risk of tuberculosis and HIV infection (Wallace and Wallace, 1998). Infectious disease epidemiologists and social scientists should incorporate these insights into mechanistic models that can explain outbreak and epidemic time series in terms of both the contact and sociality functions of networks. Such models can provide a more nuanced analysis of the relative contributions of social organiza- tion and contact to the risk of infectious diseases. Contributors J. L. Zelner and J. N. S. Eisenberg designed the study, analyzed the data, and interpreted the results. J. Trostle designed the study and interpreted the results. J. E. Goldstick contributed analytic tools and analyzed the data. W. Cevallos performed field research. All authors contributed to the writing of the article. Acknowledgments This study was supported by the National Institute of Allergy and Infectious Diseases (grant RO1-AI050038) and the Ecology of Infectious Diseases Program, Fogarty International Center of the National Institutes of Health and the National Science Foundation (grant 0811934). The authors would like to acknowledge the Ecologia, Desarrollo, Salud, y Sociedad field team, administered out of the Universidad San Francisco de Quito, for their invaluable contribution in collect- ing the data, as well as Darlene Bhavnani for her helpful comments on the data set and article. Human Participant Protection Institutional review boards at the University of California Berkeley, Uni- versity of Michigan, Trinity College, and Universidad San Francisco de Quito approved all protocols.
APPENDIX A 265 References Anderson RM, May RM. Infectious Diseases of Humans: Dynamics and Control. Oxford: Oxford Science Publications; 1992. Aparicio J, Pascual M. Building epidemiological models from R0: an implicit treatment of transmis- sion in networks. Proceedings of the Royal Society: B. 2007;274:505â512. Bansal S, Grenfell BT, Meyers LA. When individual behavior matters: homogeneous and network models in epidemiology. Journal of the Royal Society Interface. 2008;4:879â891. Baron R, Kenny D. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychol- ogy. 1986;51(6):1173â1182. Bates S, Trostle J, Cevallos W, et al. Relating diarrheal disease to social networks and the geo- graphic configuration of communities in rural Ecuador. American Journal of Epidemiology. 2007;166(9):1088â1095. Berkman LF, Glass T. Social integration, social networks, social support, and health. In: Berkman LF, Kawarchi I, editors. Social Epidemiology. New York: Oxford University Press; 2000. Checkley W, Gilman RH, Black RE, et al. Effect of water and sanitation on childhood health in a poor Peruvian peri-urban community. Lancet. 2004;363:112â118. Christley R, Pinchbeck G, Bowers R. Infection in social networks: using network analysis to identify high-risk individuals. American Journal of Epidemiology. 2005 Cohen JM, Wilson ML, Aiello A. Analysis of social epidemiology research on infectious diseases: historical patterns and future opportunities. Journal of Epidemiology and Community Health. 2007;61:1021â1027. Cohen S, Doyle WJ, Turner R, et al. Sociability and susceptibility to the common cold. Psychological Science. 2003;14(5):389â395. Eisenberg J, Cevallos W, Ponce K, et al. Environmental change and infectious disease: How new roads affect the transmission of diarrheal pathogens in rural Ecuador. Proceedings of the Na- tional Academy of Sciences. 2006;103(51):19460â19465. Entwisle B, Faust K, Rindfuss RR, et al. Networks and contexts: Variation in the structure of social ties. American Journal of Sociology. 2007;112(2):1495â1533. Holtgrave D, Crosby R. Social capital, poverty, and income inequality as predictors of gonorrhoea, syphilis, chlamydia and AIDS case rates in the United States. Sex Transm Infect. 2003;79(1):62. House J, Landis K, Umberson D. Social relationships and health. Science. 1988;241(4865):540â545. House JS. Understanding social factors and inequalities in health: 20th century progress and 21st century prospects. Journal of Health and Social Behavior. 2002;43(2):125â142. Hunt RC, Hunt E. Canal irrigation and local social organization. Current Anthropology. 1976;17(3):389â411. Jolly A, Muth S, Wylie J, et al. Sexual networks and sexually transmitted infections: a tale of two cities. Journal of Urban Health. 2001;78(3):433â445. Klinenberg E. Heat wave: A social autopsy of disaster in Chicago. Chicago: University of Chicago Press; 2002. Klovdahl AS, Graviss EA, Yaganehdoost A, et al. Networks and tuberculosis: an undetected com- munity outbreak involving public places. Social Science and Medicine. 2001;52:681â694. Koelle K, Pascual M. Disentangling extrinsic from intrinsic factors in disease dynamics: A nonlinear time series approach with an application to Cholera. American Naturalist. 2004;163(6):901â913. Laird N, Ware JH. Random-effects models for longitudinal data. Biometrics. 1982;38:963â974. Meyers L, Newman M, Martin M, et al. Applying network theory to epidemics: Control measures for Mycoplasma pneumonia outbreaks. Emerging Infectious Diseases. 2003;9(2) Newman MEJ. Spread of epidemic disease on networks. Physical Review E. 2002;66 (art. no.-016128). Pahl-Wostl C, Craps M, Dewulf A, et al. Social learning and water resources management. Ecology and Society. 2007;12(2):5.
266 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Rego RF, Moraes LR, Dourado I. Diarrhoea and garbage disposal in Salvador, Brazil. Trans R Soc Trop Medi Hyg. 2005;99:48â54. Stiratelli R, Laird N, Ware JH. Random-effects models for serial observations with binary response. Biometrics. 1984;40:961â971. Trostle JA, Hubbard A, Scott J, et al. Raising the level of analysis of food-borne outbreaks. Epide- miology. 2008;19(3):384â390. Tumwine JK, Thompson J, Katua-Katua M, et al. Diarrhoea and effects of different water sources, sanitation and hygiene behaviour in East Africa. Trop Med Int Health. 2002;7:750â756. Wallace D, Wallace R. A plague on your houses: How New York was burned down and national public health crumbled. New York: Verso; 1998. Wallace R. A Synergism of plagues: âPlanned shrinkage,â contagious housing destruction, and AIDS in the Bronx. Environmental Research. 1988;47(1):1â33. WHO. Persistent diarrhoea in children in developing countries: Memorandum from a WHO meeting. Bulletin of the World Health Organization. 1988;66:709â717. A12 CLIMATE, WIND STORMS, AND THE RISK OF VALLEY FEVER (COCCIDIOIDOMYCOSIS) Heidi E. Brown,47 Andrew C. Comrie,48 James Tamerius,49 Mohammed Khan,50 Joseph A. Tabor,51 and John N. Galgiani52 Introduction Valley fever (coccidioidomycosis) is a Western Hemisphere disease, having been found in several South American, Central American, and North American countries (Laniado-Laborin, 2007). For the most part, the areas endemic for Coccidioides spp. are rural areas of low population densities. However in the United States there are exceptions such as Bakersfield in Kern County, California 47ââHeidi E. Brown, Ph.D., M.P.H.; Assistant Professor, University of Arizona; Mel and Enid Zuckerman College of Public Health; Division of Epidemiology and Biostatistics; 1295 N. Martin Ave.; Tucson, AZ 85724. 48ââAndrew C. Comrie, Ph.D.; Professor, University of Arizona; School of Geography & Develop- ment; 409 Harvill Building; Tucson, AZ 85721. 49ââJames Tamerius, Ph.D.; Assistant Professor, University of Iowa; Department of Geographical and Sustainability Sciences; 316 Jessup Hall; Iowa City, IA 52245. 50ââMohammed Khan, M.S.P.H.; Epidemiologist; Infectious Disease Epidemiology and Surveillance; Office of Infectious Disease Services; Arizona Department of Health Services; 150 N. 18th Ave. Suite 140; Phoenix, AZ 85007. 51ââJoseph A Tabor, Ph.D., M.P.H.; Assistant Professor, University of Arizona; Mel and Enid Zuckerman College of Public Health; Division of Community, Environment and Policy; 1295 N. Martin Ave.; Tucson, AZ 85724. 52ââCorresponding author: John N Galgiani, M.D.; Professor, University of Arizona College of Medi- cine; Director, Valley Fever Center for Excellence; PO Box 245215; Tucson, AZ 85724.
APPENDIX A 267 (population nearly one million), and populations surrounding Phoenix in Mari- copa County and Tucson in Pima County of Arizona whose combined populations are approximately five million. The total number of infections reported from endemic states (Arizona, California, Nevada, New Mexico, and Utah) in 2011 were 10-fold greater than in 1998 (CDC, 2013). That the case rates for these populations have also increased eight-fold indicates that the rise is not simply due to population growth. In this report, we review some of the factors that are responsible for these changes with particular attention to how weather patterns may influence infection rates. The Problem of Valley Fever Coccidioidomycosis is a systemic fungal infection caused by Coccidioides spp. Spores (arthroconidia) of the fungus that develop in the soil of endemic regions are aerosolized by wind or mechanical disturbance of endemic soil. Inhalation of an arthroconidium into the lungs of a human or another mammal can initiate a respiratory infection. It is estimated that 150,000 such infections annually occur in U.S. residents. The consequences of infection range from no apparent illness in 60 percent of infections, to a self-limited community-acquired pneumonia in another 35 percent. The remaining 5 percent result in a variety of progressive, even life-threatening, complications, either in the lungs or outside of the chest if the fungus travels through the bloodstream to other organs such as the brain, bones, and skin (Figure A12-1). Since two-thirds of all U.S. infec- tions occur in Arizona, the Arizona Department of Health Services (ADHS) has been investigating the overall impact of this problem to the state. A questionnaire survey of newly diagnosed patients with valley fever in 2007 (Tsang et al., 2010) demonstrated the severe consequences associated with infection: · Illness lasted an average of 6 months. · 75 percent of employed persons stopped working, half missing 2 or more weeks. · 40 percent were hospitalized. In a more recent report, hospital costs alone in 2012 amounted to over $100 million.53 This, taken with outpatient care costs and lost productivity, suggests that the economic impact of valley fever on Arizona is easily several hundred million dollars annually. In California for 2000 through 2011, hospital costs were greater than $2 billion (Sondermeyer et al., 2013). 53ââSee http://azdhs.gov/phs/oids/epi/disease/valley-fever/documents/reports/valley-fever-2012.pdf (accessed December 2, 2013).
268 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A12-1â Coccidioidomycosis. Estimated numbers of total annual U.S. infections with Coccidioides spp. and resulting clinical consequences. SOURCE: Courtesy of Galgiani, 2013. How Passive Surveillance May Affect Reported Numbers of Valley Fever Infection There is no question that the nationally reported annual number of patients with coccidioidomycosis has progressively increased since the 1990s. This has been noted in several reports in recent years (CDC, 1996, 2003; Chen et al., 2011; Hector et al., 2011; Leake et al., 2000; Park et al., 2005; Sunenshine et al., 2007; Tsang et al., 2010), but the most recent Centers for Disease Control and Prevention report (CDC, 2013) gives this trend much needed visibility in both the medical community and the general media, which reported on this story na- tionally. Headlines such as âValley Fever Cases Skyrocketâ and âValley Fever Cases Are at Their Highest Numbers in Nearly Two Decadesâ dotted news stories for more than a week following the CDC report. There have been 111,717 cases of coccidioidomycosis reported to the CDC from 1998 through 2011. Arizona
APPENDIX A 269 was responsible for 66 percent of these infections and California for 31 percent. Nearly all of the remaining infections were reported from New Mexico, Utah, and Nevada. The annual number of valley fever cases in Arizona and Califor- nia are shown in Figure A12-2. Case rates reported by the CDC are prevalence estimates based on the general population and not the susceptible population as are disease incidence rates. Care needs to be used when comparing reported case rates between different parts of the country since they do not take into account the proportion of previously infected (now immune) individuals (Tabor and OâRourke, 2010). Notably absent from these numbers is representation from Texas, well known to be endemic along its western border (Edwards and Palmer, 1957). There are no reports of coccidioidomycosis from Texas because it is not a reportable disease in that state, underscoring one of the limitations of the National Notifiable Disease Surveillance System (NNDSS): state reporting to NNDSS is voluntary. In addi- tion to missing information from Texas, underreporting by some or many of the nonendemic states is likely as well. Beyond state reporting decisions, several other surveillance factors need to be met for a patientâs infection to be incorporated into the overall case tally. First, only persons sick enough to seek medical care will be included. Second, a clini- cian needs to consider the diagnosis and order the necessary tests. Third, the tests need to have sufficient sensitivity and specificity to enable the correct diagnosis. FIGURE A12-2âAnnual coccidioidomycosis. Numbers of cases of coccidioidomycosis reported to the CDC by Arizona and California from 1990 through 2013. SOURCE: Courtesy of Galgiani, 2013.
270 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Fourth, once diagnosed, the infection must be reported to public health authori- ties. How completely the second, third, and fourth of these steps is conducted has a direct effect on the resulting estimates of disease activity. Physician awareness of valley fever is variable even within the endemic regions. For example, in a survey by the ADHS, Arizona clinicians were asked about their knowledge, attitudes, and practice with respect to valley fever (Chen et al., 2011). Only 12 percent of respondents indicated they had learned medi- cine in Arizona schools and 47 percent had no clinical training in Arizona prior to starting practice here. Moreover, 40 percent lacked confidence in diagnosing a coccidioidal infection. In another study of two physician group practices, only 2â13 percent of patients with community-acquired pneumonia were evaluated for Coccidioides infection (Chang et al., 2008). These relatively recent studies strongly suggest that the actual number of patients seeking care for valley fever infections is greatly underreported. Moreover, if clinicians improve their capac- ity for detecting new coccidioidal infections, their changed practices have the potential of significantly increasing the number of reported cases. Although available serologic tests are effective in diagnosing patients with widespread and long-standing infection, false negatives are common among newly infected patients with disease limited to the lungs. In one study, standard serologic testing missed such infections in about half of first sera tested (Wieden et al., 1996). Improving the sensitivity of clinical testing is under active investiga- tion, and progress in this area could significantly change the number of reported cases. For example, in 2009 a major clinical laboratory in Arizona began using a more sensitive test as indicative of infection. As a result the number of reported cases to the state nearly doubled (Hector et al., 2011). A puzzling observation in the 2013 CDC report is the disproportionate in- crease in reported coccidioidomycosis among females in Arizona. The percent- age of females with coccidioidomycosis before 2009 was 44 percent, but since 2009 this has risen to 55 percent. One possible explanation for this shift comes from a pair of studies conducted by the University of Arizona Campus Health (Lundergan et al., 1985; Stern and Galgiani, 2010). In their 1985 report, women comprised 44 percent of valley fever cases, but in 2010, females comprised 56 percent of infected scholarship athletes. Between these two studies, screening serologic tests for coccidioidomycosis at Campus Health changed from the less sensitive standard coccidioidal serology (immunodiffusion tests) to the more sensitive enzyme-linked immunoassays (EIAs) (Wieden et al., 1996). This was the same change that was made in 2009 by the major Arizona clinical labora- tory mentioned above. Women at the University of Arizona, on average, were found to have lower complement fixation titers (Lundergan et al., 1985), raising the plausible possibility that the increased statewide percentage of females with valley fever could be due to the increased sensitivity of EIAs to lower levels of anticoccidioidal antibodies in women. Certainly further studies are needed to clarify these findings.
APPENDIX A 271 Reporting newly diagnosed patients with coccidioidomycosis may not al- ways be complete. For example, although clinicians are required to report new coccidioidomycosis, it may be difficult because of busy schedules. In contrast, if reporting is asked of the clinical laboratory that identifies the positive test, the likelihood of reporting is much greater. In Arizona, the reporting responsibilities were shifted in 1997 to include laboratory reporting, and it is possible that some of the increase in the ensuing years was due to that change. Relationship Between Weather Patterns and Valley Fever Infections in Arizona Despite several surveillance considerations just described, none adequately account for the episodic increases seen in California in 1993 and 1994 or in Ari- zona in 2011 (Figure A12-2). Weather factors such as wind, precipitation, and heat may provide an explanation. The 37 percent increase in Arizona cases in 2011 highlights the interaction of valley fever and weather with the co-occurrence of several spectacular dust storms, known as haboobs. These haboobs were so severe and so directly affecting the urban Phoenix area that time-lapse video foot- age featured prominently on national evening news programs.54 As early as 1940, Smith described the relationship between seasonal weather patterns and valley fever incidence: the lowest incidence occurs during the wet seasons; incidence increases with the onset of the dry weather of spring and early summer; the peak season follows the hot summer and increased winds of fall. Smith (1940) also de- scribed human activity (harvesting) as an exposure risk. It bears noting, however, that the strength of the association varies across populations and time periods. Wind is an important factor to generate aerosols of Coccidioides spores. For example, in December 1977, a major Santa Anna wind swept across the Central Valley of California, resulting in cases of valley fever in distant, nonendemic areas such as the San Francisco Bay Area (Flynn et al., 1979). In Kern County there were 120 excess cases in the following 3 months (Pappagianis and Einstein, 1978). That was with a Kern County population of approximately 400,000 of which three quarters were likely immune because of prior infection and therefore not susceptible to new infection. In contrast, the Phoenix area population down- wind to the July 2011 haboobs was 10 times larger, and three-quarters were likely to be susceptible because of the large in-migration population. Simple extrapola- tion using these figures results in a prediction of an excess 3,600 infections in the months following July. However, this prediction was not borne out. As shown in Figure A12-3, the week-to-week numbers of reported valley fever cases for the three major urban areas within the endemic region (Maricopa, Pima, and Pinal Counties) were strikingly stable with no apparent increase in cases in the months following the first major July storm. 54ââSee http://www.cbsnews.com/2100-201_162-20094755.html (accessed December 2, 2013).
272 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A12-3â Dust storms have little effect on Arizona case rates. Weekly reported cases of coccidioidomycosis in selected Arizona counties for 2011. SOURCE: Courtesy of Galgiani, 2013. There are at least three possible explanations for how these 2011 obser- vations in the Phoenix area could be so different from what was observed in California in the 1977 storm. First, the California storm occurred in the winter, a time when infections are minimal in Kern County. Therefore, with very low background numbers of cases, the excess cases were very apparent. In contrast, the Phoenix storms occurred when many cases normally are reported. Thus, it is possible that there in fact were excess cases that could not be detected by the passive surveillance that is in place. Second, summer haboobs, although excep- tionally spectacular, usually last only a matter of hours. It very well may be that these very short-term peak concentrations and exposures of coccidioidal spores in the air cannot be detected from case reporting due to the attenuating effects of variable disease onset and reporting time. The spores, being 3â5 microns in size, may require only slight turbulence to be lifted from the soil surface into the air. Figure A12-4 is a hypothetical representation of such a situation to illustrate that if spores are being picked up on a regular basis the overall area under the curve for spore density might only slightly be increased from exceptional but brief dust storms. Third, haboobs in the Phoenix area are generally associated with sum- mer thunderstorms, but in different locales across central and southern Arizona they may possess different amounts and types of dust. Much of the dust in and around Phoenix may come from disturbed agricultural and urban landâneither
APPENDIX A 273 FIGURE A12-4â How dust storm contribution on spore density could be minimal. Hypo- thetical contribution of episodic wind storms to ambient atmospheric spore density if only small breezes are sufficient to produce an aerosol. SOURCE: Courtesy of Galgiani, 2013. of which represents the natural desert soils that Coccidioides spp. are thought to favor. Thus the path that these storms cover may largely determine the exposure risk they pose to downwind populations. Although it is difficult to show a causal relationship between wind storms and the 2011 increase in valley fever cases, a recently developed model shows a strong relationship at seasonal time scales between precipitation patterns and the year-to-year changes in valley fever cases in Arizona. In particular, winter pre- cipitation promotes fungal growth in the soil (Hugenholtz, 1957), and increased precipitation during this period seems to be related to increased incidence of valley fever the following summer and fall (Comrie, 2005; Kolivras and Comrie, 2003; Smith et al., 1946; Tamerius and Comrie, 2011). High summer precipita- tion, conversely, has a negative effect on incidence, perhaps by reducing the chances of aerosolized spores (Kolivras and Comrie, 2003). Tamerius and Comrie (2011) used data from 1995 through 2006 for rainfall and cases in the Arizona counties of Maricopa and Pima to develop a precipitation-driven model of val- ley fever cases. They reasoned that the overall trend evident in Figure A12-2 of increasing cases over that period was not climate related and removed the trend to reduce biasing any association between weather and reported cases. Using
274 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS an autocorrelation statistical procedure, they were then able to define a primary coccidioidal exposure season of August through March. By examining rainfall patterns before and during these exposure seasons, these authors identified two countervailing relationships. First, the magnitude of rainfall during the winter was positively correlated with the number of reported infections the following season. Second, higher rain amounts during an exposure season resulted in fewer reported infections. Combining these two relationships in a single model during the train- ing period resulted in a correlation coefficient of 0.83 for Maricopa data and 0.73 for Pima County. In preparation for this workshop, this model was updated to the present period for Maricopa County. The model was also used to estimate disease activity backward to 1950 (hindcasting) using historical precipitation data. With the hindcasting, there are no surveillance data available for the majority of this period to validate the accuracy of the results. Updating the model for Maricopa Countyâ The model was updated to investi- gate the predictive power of winter precipitation and valley fever exposure using current data (1995â2013). The greatest challenge with these data was the afore- mentioned changes in case definition, reporting, and testing around 2009 that led to a doubling of cases: from under 5,000 to over 10,000 (Nguyen et al., 2013). As in the original publication, the case data were adjusted to estimate exposure dates rather than date of diagnosis or report. Monthly reported incidence (number of cases per 100,000) for each was calculated by dividing the number of cases in Maricopa County by the U.S. Census Bureau estimated annual population for the county. As previously, the data were detrended for the period January 1995âFebruary 2009 by removing the best-fit linear regression (i.e., modeling the residuals). To adjust for the change in the case definition and reporting, the median incidence was subtracted out for the period from March 2009âMarch 2013. While this standardizes the two distinct periods, the variance for the latter period (2009â2013) is greater and creates difficulty when comparing incidence across study periods. Finally, seasonal incidence was calculated by summing across months during the exposure season (AugustâMarch). Monthly precipitation totals for the grid points in which the Maricopa Na- tional Weather Service Station locations fall were acquired from the Oregon State PRISM (Parameter-elevation Regressions on Independent Slopes Model) climate mapping system (http://prism.oregonstate.edu/index.phtml, last accessed Decem- ber 2, 2013) and averaged for use in the model. The previously identified regres- sion coefficients were applied on the data to estimate the number of exposures per month for the primary exposure season (AugustâMarch). There is concordance between the predictions from the model and the observed reported cases for the exposure seasons (Figure A12-5). The model fits relatively well, though it errs in 1998, 2008, 2009, and 2010 (i.e., predicts high when the exposures are low or vice versa). Note that the means of the detrended (observed) time series and the modeled time series have been standardized (forced to equal) for comparison.
APPENDIX A 275 FIGURE A12-5â Model results (1996â2013). Comparison of modeled versus the predicted exposure rates based on OctoberâDecember precipitation prior to the exposure season and concurrent exposure season (August to March) precipitation. The solid line indicates the model exposure rate and the dashed line the observed rate. â * Tamerius and Comrie, 2010. SOURCE: Courtesy of James Tamerius. Hindcasting (1950 through 2013)â To estimate the number of cases back in time, we also applied the model to a precipitation time series beginning in 1950 using the same PRISM-based precipitation data for one location, Maricopa County (downloaded from http://www.cefa.dri.edu/Westmap). The model was applied in order to estimate the number of cases that might have been expected in the past. The outcome of this model is the estimated exposure rate for the primary exposure period (AugustâMarch) per 100,000. Average case numbers predicted by the model for the periods 1950â1979 and 1908â2009 are shown in Figure A12-6. The dashed line indicates the time period for which data were available and upon which the original model was built (Tamerius and Comrie, 2011), and the solid line indicates the predictions. Unlike observed temperature increases, the Southwest has not experienced shift- ing trends in precipitation for the past 110 years (Hoerling et al., 2013). Our precipitation-based model similarly does not appear to indicate any significant change in the predicted seasonal exposure. Finally, in addition to wind and precipitation, heat and human behavior also play a role the incidence of valley fever. Evidence suggests temperature has a
276 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A12-6â Hindcasting estimates (1950â2013). Modeled coccidioidal exposure in Maricopa County during AugustâMarch for the years 1950 through 2012. The model training period, 1995â2006, is shown as a dashed line while the prediction is a solid line. SOURCE: Courtesy of Andrew Comrie. role in regulating competition between fungal species. Laboratory experiments showed Coccidioides spp. are well adapted to the arid environment, tolerating extreme heat and a wide range of humidity levels (Friedman et al., 1956). High heat and sun is proposed to have a sterilizing effect on soils that may provide a competitive advantage for Coccidioides spp. over other soil fungi (Duran et al., 1973; Hugenholtz, 1957; Maddy, 1965). Heat is also thought to facilitate the aero- solization of spores by drying soils. All of these external forces acting upon the fungus are further affected by host behavior. Human or other animal activity such as digging, excavating, and other soil disrupting activities may result in cases outside the climate-driven seasonality (Converse and Reed, 1966; Maddy, 1965). This is further supported by work in Kern County, California, where weather has only a weak association with incidenceâthe authors attribute human behavior to the observed disease trends in their area (Talamantes et al., 2007; Zender and Talamantes, 2006). Future Approaches to Better Understand the Effects of Environmental Change on Risk of Coccidioidomycosis Current knowledge of Coccidioides ecology is based mostly on pre-1970 studies and, more recently, inferred from reported human cases that are aggregated
APPENDIX A 277 at the county level (Baptista-Rosas et al., 2007; Comrie, 2005; Tamerius and Comrie, 2011). Spatial and temporal precision is low when relying on reported case data, whereas the alternative of analyzing environmental samples collected precisely is expensive and problematic. Indeed, our accumulated understanding about the environmental biology and ecology of Coccidioides spp. is largely incomplete, due to specific challenges in identifying the fungus in its natural state. Improvements in PCR detection and direct plating isolation methods for environmental samples are needed in order to replace the standard method of induced infections in laboratory mice (Barker et al., 2012). Here we briefly sum- marize the literature and identify a pathway to completing this ecological puzzle. Geographic distributionâ Much groundwork for identifying environmental cor- relates of Coccidioides spp. habitat was performed in the 1940s and 1950s. Researchers like Maddy (1958) showed an association between valley fever in- cidence and the lower Sonoran life zone, though subsequent research expanded the endemic zone more broadly. Epidemiological evidence led researchers to be- lieve exposure was inhalational, often windborne (Emmons, 1942a; Smith et al., 1946), despite difficulty recovering Coccidioides spp. from the air (Ajello et al., 1965; Anderson, 1958; Converse and Reed, 1966). Weather was also found to play a role in identifying endemic regions that shared a characteristic wet period followed by a dry period with blowing winds (Hugenholtz, 1957; Maddy, 1958; Smith et al., 1946). The association with precipitation patterns has been borne out in modern times as well (Kolivras and Comrie, 2003; Park et al., 2005; Stacy et al., 2012; Tamerius and Comrie, 2011; Zender and Talamantes, 2006). Rodents and microhabitatâ Ideas regarding the role of animals in the evolution of Coccidioides spp. has been inconsistent. Some researchers have concluded that infected animals, like humans, are accidental hosts to the fungus (Cummins et al., 1929; Pulford and Larson, 1929), while others have proposed a role for the animal carcass as a medium for fungal growth within the soil (Emmons, 1942b; Emmons and Ashburn, 1942). While the mechanism is unknown, there is significant evi- dence that rodents play a prominent role in the saprophytic phase of this fungus. For example, although the fungus is notoriously difficult to recover from soil (Greene et al., 2000), soil samples collected near animal burrows are often posi- tive for the fungus (Barker et al., 2012; Eulalio et al., 2001; Maddy, 1959, 1965). Challenges and Future Work While considerable strides were made in those early years of research on valley fever, Ajello (1967) concluded that âthe ecology of these fungi, i.e., the study of their relationship to their environment, is rather superficial and scanty.â This sentiment holds today: âecology of the pathogen, Coccidioides, remains obscureâ (Barker et al., 2012). Recent outbreaks of coccidioidomycosis in eastern
278 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Washington and northern Utah (Mardo et al., 2002; Marsden-Haug et al., 2013), far outside of known endemic areas, illustrate our poor understanding of these fungiâs ecologies. The outbreaks could be due to soil disturbances in isolated patches of Coccidioides spp. or due to range expansion. Genetic analyses have shown that valley fever is caused by C. immitis in central/southern California and a genetically related, but distinct C. posadasii in southern Arizona, Texas, and Mexico (Barker et al., 2007; Fisher et al., 2002). Consideration of differences among these two fungal species will likely help further elucidate distinctions and nuances in the range and habitat of these fungi. However, the difficulty in locating areas of infected soilâwhich, as noted, may be related to burrowing animals, soil salinity (Ajello, 1967; Friedman et al., 1956), soil type (Maddy, 1958), and veg- etation (Egeberg, 1953; Swatek, 1970)âcontinues to stymie accurate exposure tracking. From a public health standpoint, it is this latter aspectâmore narrowly defining exposure riskâthat will be critical in reducing the impact of this disease. Currently, soil samples must be collected in the field, transported to the labo- ratory, and mouse models exposed to the potentially contaminated soil (Barker et al., 2012; Levine and Winn, 1964; Maddy, 1965) in order to identify Coc- cidioides spp. This is a labor- and cost-intensive method for identifying possible sources and does not lend itself well to large-scale risk mapping. Sorely needed is a means to quickly and accurately identify the microhabitat wherein this fungus thrives. Conclusions In this paper, we discuss trends in occurrence of valley fever in the United States. Of note are the challenges this disease presents to prevention efforts: re- porting issues, changes in diagnostics and detection, changes in surveillance, and limited tools for assessing risk. We provide an updated version of a precipitation- driven model that fits well for Maricopa County, Arizona. We end with a discus- sion of the ecology of this disease of the American Southwest and the challenges facing improvements to disease control. The increase in the number of reported cases and interest following these re- ports have generated a justifiable spotlight on this disease. As physiciansâ aware- ness of the disease increases, there will likely be an increase in testing for valley fever. Accordingly, reported incidence of this disease will likely increase. While improving our capacity to describe the burden of this disease, these increases will be a result of enhanced case identification rather than changes in susceptibility or exposure. In all likelihood, individual patient outcomes will be improved by accurate and earlier diagnosis; however, these changes will not translate into valley fever prevention. Currently, a serious impediment to prevention is a lack of knowledge about the ecological processes that modulate the presence of Coccidioides spp. in the environment. Ecological research is needed that examines the relationships
APPENDIX A 279 between the occurrence of Coccidioides spp. and soil moisture, soil temperature, and rodent populations. Defining the habitat of this pathogen (identifying specific soils, areas, regions) would enable the identification of risk related to specific human activities or periods of time; possibilities for developing treatment of land- scapes to reduce fungal proliferation; and early warning for protection against aerosols (e.g., knowing that sufficient strength winds are blowing across known endemic soils). Until a means to detect the pathogen in the soil is discovered, we are left with broad and sometimes conflicting studies on the risk of acquiring valley fever from the environment. Also needed is the exploration into the etiological relationships for spore exposure. Not all dust contains viable spores. Many validated models exist for predicting dispersal and exposure of airborne particulates, but not for predicting when and where viable Coccidioides spp. spores are entrained into the air from the soil surface. Likely predictors for airborne dispersal and exposure are soil surface temperature, soil surface moisture, vegetative cover, and UV exposure from the sun. Once these ecological and etiological questions can be answered then actionable occupational and land use practices can be identified for disease prevention. References Ajello, L. 1967. Comparative ecology of respiratory mycotic disease agents. Bacteriological Reviews 31(1):6-24. Ajello, L., K. Maddy, G. Crecelius, P. G. Hugenholtz, and L. B. Hall. 1965. Recovery of Coccidioides immitis from the air. Sabouraudia 4(June):92-95. Anderson, A. A. 1958. New sampler for the collection, sizing, and enumeration of viable airborne particles. Journal of Bacteriology 76(5):13. Baptista-Rosas, R. C., A. Hinojosa, and M. Riquelme. 2007. Ecological niche modeling of Coccidi- oides spp. in western North American deserts. Annals of the New York Academy of Sciences 1111:35-46. Barker, B. M., K. A. Jewell, S. Kroken, and M. J. Orbach. 2007. The population biology of coccidi- oides: Epidemiologic implications for disease outbreaks. Annals of the New York Academy of Sciences 1111:147-163. Barker, B. M., J. A. Tabor, L. F. Shubitz, R. Perill, and M. J. Orbach. 2012. Detection and phyloge- netic analysis of Coccidioides posadasii in Arizona soil samples. Fungal Ecology 5:13. CDC (Centers for Disease Control and Prevention). 1996. CoccidioidomycosisâArizona, 1990-1995. Morbidity and Mortality Weekly Report 45:1069-1073. CDC. 2003. Increase in coccidioidomycosisâArizona, 1998-2001. Morbidity and Mortality Weekly Report 52(6):109-112. CDC. 2013. Increase in reported coccidioidomycosisâUnited States, 1998-2011. Morbidity and Mortality Weekly Report 62:217-221. Chang, D. C., S. Anderson, K. Wannemuehler, D. M. Engelthaler, L. Erhart, R. H. Sunenshine, L. A. Burwell, and B. J. Park. 2008. Testing for coccidioidomycosis among patients with community- acquired pneumonia. Emerging Infectious Diseases 14(7):1053-1059. Chen, S., L. M. Erhart, S. Anderson, K. Komatsu, B. Park, T. Chiller, and R. Sunenshine. 2011. Coc- cidioidomycosis: Knowledge, attitudes, and practices among healthcare providersâArizona, 2007. Medical Mycology 49(6):649-656.
280 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Comrie, A. C. 2005. Climate factors influencing coccidioidomycosis seasonality and outbreaks. En- vironmental Health Perspectives 113(6):688-692. Converse, J. L., and R. E. Reed. 1966. Experimental epidemiology of coccidioidomycosis. Bacterio- logical Reviews 30(3):678-695. Cummins, W. T., J. K. Smith, and C. H. Halliday. 1929. Coccidioidal granuloma, an epidemiologic survey, with a report of 24 additional cases. Journal of the American Medical Association 93:1046-1049. Duran, F., Jr, G. W. Robertstad, and E. Donowho. 1973. The distribution of Coccidioides immitis in the soil in El Paso, Texas. Sabouraudia 11(2):143-148. Edwards, P. Q., and C. E. Palmer. 1957. Prevalence of sensitivity to coccidioidin, with special refer- ence to specific and nonspecific reactions to coccidioidin and to histoplasmin. Diseases of the Chest 31:35-60. Egeberg, R. O. 1953. Coccidioidomycosis: Its clinical and climatological aspects with remarks on treatment. Transactions of the American Clinical and Climatological Association 65:116-126. Emmons, C. W. 1942a. Coccidioidomycosis. Mycologia 34:452-463. Emmons, C. W. 1942b. Isolation of coccidioides from soil and rodents. Public Health Reports 57:109-111. Emmons, C. W., and L. L. Ashburn. 1942. The isolation of Haplosporangium parvum n. sp. and Coc- cidioides immitis from wild rodents. Their relationship to coccidioidomycosis. Public Health Reports 57:1715-1727. Eulalio, K. D., R. L. de Macedo, M. A. Cavalcanti, L. M. Martins, M. S. Lazera, and B. Wanke. 2001. Coccidioides immitis isolated from armadillos (Dasypus novemcinctus) in the state of Piaui, northeast Brazil. Mycopathologia 149(2):57-61. Fisher, M. C., G. L. Koenig, T. J. White, and J. W. Taylor. 2002. Molecular and phenotypic descrip- tion of Coccidioides posadasii sp nov., previously recognized as the non-California population of Coccidioides immitis. Mycologia 94(1):73-84. Flynn, N. M., P. D. Hoeprich, M. M. Kawachi, K. K. Lee, R. M. Lawrence, E. Goldstein, G. W. Jordan, R. S. Kundargi, and G. A. Wong. 1979. An unusual outbreak of windborne coccidioi- domycosis. New England Journal of Medicine 301(7):358-361. Friedman, L., C. E. Smith, D. Pappagianis, and R. J. Berman. 1956. Survival of Coccidioides immitis under controlled conditions of temperature and humidity. American Journal of Public Health 46:1317-1324. Greene, D. R., G. Koenig, M. C. Fisher, and J. W. Taylor. 2000. Soil isolation and molecular identi- fication of Coccidioides immitis. Mycologia 92(3):406-410. Hector, R. F., G. W. Rutherford, C. A. Tsang, L. M. Erhart, O. McCotter, K. Komatsu, S. M. Anderson, F. Tabnak, D. J. Vugia, Y. Yang, and J. N. Galgiani. 2011. Public health impact of coccidioi- domycosis in California and Arizona. International Journal of Environmental Research and Public Health 8(4):1150-1173. Hoerling, M. P. D., K. Wolter, J. Lukas, J. Eischeid, R. Nemani, B. Leibmann, and K. E. Kunkel. 2013. Present weather and climate: Evolving conditions. In Assessment of climate change in the Southwest United States, edited by G. J. Garfin, R. Merideth, M. Black, and S. LeRoy. Washington, DC: Island Press. Pp. 74-100. Hugenholtz, P. G. 1957. Climate and coccidioidomycosis. In Proceedings of Symposium on Coc- cidioidomycosis, Phoenix, AZ. Atlanta: Public Health Service Publication No. 575. Pp. 136-143. Kolivras, K. N., and A. C. Comrie. 2003. Modeling valley fever (coccidioidomycosis) incidence on the basis of climate conditions. International Journal of Biometeorology 47(2):87-101. Laniado-Laborin, R. 2007. Expanding understanding of epidemiology of coccidioidomycosis in the Western Hemisphere. Annals of the New York Academy of Sciences 1111:19-34. Leake, J. A., D. G. Mosley, B. England, J. V. Graham, B. D. Plikaytis, N. M. Ampel, B. A. Perkins, and R. A. Hajjeh. 2000. Risk factors for acute symptomatic coccidioidomycosis among elderly persons in Arizona, 1996-1997. Journal of Infectious Diseases 181(4):1435-1440.
APPENDIX A 281 Levine, H. B., and W. A. Winn. 1964. Isolation of Coccidioides immitis from soil. Health Laboratory Science 1:29-32. Lundergan, L. L., S. S. Kerrick, and J. N. Galgiani. 1985. Coccidioidomycosis at a university outpatient clinic: A clinical description. In Coccidioidomycosis. Proceedings of the Fourth In- ternational Conference, edited by H. E. Einstein and A. Catanzaro. Washington, DC: National Foundation for Infectious Diseases. Pp. 47-54. Maddy, K. T. 1958. The geographic distribution of Coccidioides immitis and possible ecologic impli- cations. Arizona Medicine 15:178-188. Maddy, K. T. 1959. A study of a site in Arizona where a dog apparently acquired a Coccidioides im- mitis infection. American Journal of Veterinary Research 20:642-646. Maddy, K. T. 1965. Observations on Coccidioides immitis found growing naturally in soil. Arizona Medicine 22:281-288. Mardo, D., R. A. Christensen, N. Nielson, S. Hutt, R. Hyun, J. Shaffer, C. Barton, G. Dowdle, M. Mottice, C. Brokopp, R. Rolfs, and D. Panebaker. 2002. Coccidioidomycosis in workers at an archeologic siteâDinosaur National Monument, Utah, June-July 2001. Archives of Dermatol- ogy 138(3):424-425. Marsden-Haug, N., M. Goldoft, C. Ralston, A. P. Limaye, J. Chua, H. Hill, L. Jecha, G. R. Thompson, 3rd, and T. Chiller. 2013. Coccidioidomycosis acquired in Washington State. Clinical Infectious Diseases 56(6):847-850. Nguyen, C., B. M. Barker, S. Hoover, D. E. Nix, N. M. Ampel, J. A. Frelinger, M. J. Orbach, and J. N. Galgiani. 2013. Recent advances in our understanding of the environmental, epidemiological, immunological, and clinical dimensions of coccidioidomycosis. Clinical Microbiology Reviews 26(3):505-525. Pappagianis, D., and H. Einstein. 1978. Tempest from Tehachapi takes toll or coccidioides conveyed aloft and afar. Western Journal of Medicine 129:527-530. Park, B. J., K. Sigel, V. Vaz, K. Komatsu, C. McRill, M. Phelan, T. Colman, A. C. Comrie, D. W. Warnock, J. N. Galgiani, and R. A. Hajjeh. 2005. An epidemic of coccidioidomycosis in Arizona associated with climatic changes, 1998-2001. Journal of Infectious Diseases 191(11):1981-1987. Pulford, D. S., and E. E. Larson. 1929. Coccidioidal granuloma; report of a case treated by intra- venous dye, colloidal lead, and colloidal copper, with autopsy observations. Journal of the American Medical Association 93:1049-1056. Smith, C. E. 1940. Epidemiology of acute coccidioidomycosis wth erythema nodosum. American Journal of Health 30:600-611. Smith, C. E., R. R. Beard, H. G. Rosenberger, and E. G. Whiting. 1946. Effect of season and dust control on coccidioidomycosis. Journal of the American Medical Association 132(14):833-838. Sondermeyer, G. L., D. Gilliss, F. Tabnak, and D. Vugia. 2013. Coccidioidomycosis-associated hos- pitalizations, California, USA, 2000-2011. Emerging Infectious Diseases 19(10):8. Stacy, P. K., A. C. Comrie, and S. R. Yool. 2012. Modeling valley fever incidence in Arizona using a satellite-derived soil moisture proxy. GIScience and Remote Sensing 49(2). Stern, N. G., and J. N. Galgiani. 2010. Coccidioidomycosis among scholarship athletes and other college students, Arizona, USA. Emerging Infectious Diseases 16(2):321-323. Sunenshine, R. H., S. Anderson, L. Erhart, A. Vossbrink, P. C. Kelly, D. Engelthaler, and K. Komatsu. 2007. Public health surveillance for coccidioidomycosis in Arizona. Annals of the New York Academy of Sciences 1111:96-102. Swatek, F. E. 1970. Ecology of Coccidioides immitis. Mycopathologia et Mycologia Applicata 40(1-2):3-12. Tabor, J. A., and M. K. OâRourke. 2010. A risk factor study of coccidioidomycosis by controlling dif- ferential misclassifications of exposure and susceptibility using a landscape ecology approach. Science of the Total Environment 408(10):2199-2207. Talamantes, J., S. Behseta, and C. S. Zender. 2007. Statistical modeling of valley fever data in Kern County, California. International Journal of Biometeorology 51(4):307-313.
282 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Tamerius, J. D., and A. C. Comrie. 2011. Coccidioidomycosis incidence in Arizona predicted by seasonal precipitation. PLoS One. 6(6):e21009. Tsang, C. A., S. M. Anderson, S. B. Imholte, L. M. Erhart, S. Chen, B. J. Park, C. Christ, K. K. Komatsu, T. Chiller, and R. H. Sunenshine. 2010. Enhanced surveillance of coccidioidomycosis, Arizona, USA, 2007-2008. Emerging Infectious Diseases 16(11):1738-1744. Wieden, M. A., L. L. Lundergan, J. Blum, K. L. Delgado, R. Coolbaugh, R. Howard, T. Peng, E. Pugh, N. Reis, J. Theis, and J. N. Galgiani. 1996. Detection of coccidioidal antibodies by 33-kDa spherule antigen, Coccidioides EIA, and standard serologic tests in sera from patients evaluated for coccidioidomycosis. Journal of Infectious Diseases 173:1273-1277. Zender, C. S., and J. Talamantes. 2006. Climate controls on valley fever incidence in Kern County, California. International Journal of Biometeorology 50(3):174-182. A13 ZOONOTIC DISEASE RISKS ASSOCIATED WITH TRADE AND MOVEMENT OF ANIMALS Nina Marano,55 Adam J. Langer,55 G. Gale Galland,55 Nicole J. Cohen,55 Emily Lankau,56 Ashley Marrone,55 David McAdam,55 Casey Barton Behravesh,57 and Nicki Pesik55 Globalization of the market for live animals and animal productsâcombined with human behaviors and preferences for the exoticâare ever-growing risk factors for the translocation of zoonotic diseases to the United States from parts of the world where they are endemic or exist in a reservoir state. Why is there global trade in animals? Animals and animal products are transported across borders for many reasons. They are used for exhibitions at zoos; scientific edu- cation, research, and conservation programs; food and nonedible products; for the pet trade; and in the case of companion animals, tourism and immigration. The United States is one of the worldâs largest consumers of imported wildlife and wildlife products. A recent analysis showed that during 1999â2010, over 80 million vertebrate species were imported to the United States, including 2 million mammalian species; of these, 46 percent were rodent species (Romagosa, 2010; CDC, unpublished data). In the United States, there is a network of federal, state, and local agency regulations in place to prevent the transmission of diseases carried by animals that could be harmful to humans, other animals, or the environment. There are five U.S. federal agencies whose authorities pertain to movement of live animals: the 55ââDivisionof Global Migration and Quarantine, U.S. Centers for Disease Control and Prevention. 56ââLandCow Consulting. 57ââDivision of Foodborne, Waterborne, and Environmental Diseases, U.S. Centers for Disease Control and Prevention.
APPENDIX A 283 U.S. Department of Health and Human Services Centers for Disease Control and Prevention (HHS/CDC), U.S. Department of Agriculture Animal and Plant Health Inspection Service (USDA-APHIS), U.S. Fish and Wildlife Service (USFWS), the U.S. Department of Homeland Security (DHS), and the U.S. Food and Drug Administration (FDA). Section 361 of the Public Health Service Act58 (42 U.S.C. 264) gives the Secretary of Health and Human Services the authority to make and enforce regulations to prevent the introduction into, and the spread of com- municable diseases within, the United States. The secretary has delegated the responsibility to oversee the importation of animals, animal products, and other potentially infectious items to CDC. Because of the known human health risks associated with certain animal species, HHS/CDC has promulgated regulations that specifically restrict the importation of animals and animal products such as nonhuman primates (NHPs), dogs and cats, small turtles (those with a shell length of less than 4 inches), African rodents, and goat skin drums from Haiti. Additionally, the importation of animals and animal products of civets (family Viverridae) as well as infectious biological agents, infectious substances, and animal vectors of human disease (such as bats) is restricted under HHS/CDC regulations (HHS, 2001). CDCâs Regulations for Dog and Cat Importation Importation of dogs and cats is restricted by CDC because these domestic animals carry zoonotic diseases (HHS/CDC, 2003a). They are the most common animals kept as pets in the United States; thus, they have very close contact with humans. Dogs and cats are both capable of transmitting rabies, which is a major reason why CDCâs dog and cat regulations are in place. In the United States, widespread mandatory vaccination of dogs has eliminated the canine variant of rabies and dramatically reduced the number of human cases (Velasco-Villa et al., 2008). However, the risk of reintroduction of canine rabies virus variants exists via importation of unvaccinated dogs from areas where rabies is enzootic, such as Asia, Africa, the Middle East, and parts of Latin America. Globally, canine vari- ants are responsible for most of the 55,000 human deaths from rabies estimated worldwide each year (HHS, 2001; WHO, 2009). Since canine variants of rabies remain a very serious health threat in many other countries, preventing the entry of potentially infected dogs into the United States is a critical public health priority. Each year CDC publishes a list of coun- tries that reported no indigenous cases of terrestrial rabies in the previous year; these countries were formerly known as ârabies-freeâ countries (CDC, 2014a). At 58ââThePublic Health Service Act is a U.S. federal law enacted in 1946. The full act is captured under Title 42 of the United States Code âThe Public Health and Welfare,â Chapter 6A âPublic Health Serviceâ under section 361 of the Public Health Service Act (42 U.S.C. 264). The U.S. Secretary of Health and Human Services is authorized to take measures to prevent the entry and spread of com- municable diseases from foreign countries into the United States and between states.
284 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS present, CDC allows the entry of dogs from rabies-free countries. Unvaccinated dogs may be imported without a requirement for proof of rabies vaccination if they have been located for a minimum of 6 months or more in a rabies-free country. Unvaccinated dogs from rabies-endemic countries pose a risk to rabies- free countries for reintroduction of terrestrial rabies. This was evidenced by the reintroduction of terrestrial rabies in Greece and Taiwan (Huang et al., 2013; Tsiodras et al., 2013) Unfortunately, since May 2004 there have been at least four documented instances of dogs being imported to the United States from rabies-endemic areas that subsequently were diagnosed with rabies, necessitating extensive public health investigations to identify persons at risk of exposure and in need of postexposure prophylaxis (PEP) (Table A13-1). HHS/CDC regulations require that dogs entering the United States be vac- cinated for rabies or, if they are not vaccinated, that the importer agree to have the dog vaccinated and confined for 30 days after rabies vaccination to allow for acquisition of vaccine-induced immunity (HHS/CDC, 2003a). While HHS/CDC regulations do not currently require that cats be vaccinated against rabies, HHS/ CDC does require that they appear to be healthy upon entry into the United States and recommends that all cats receive rabies vaccinations. HHS/CDC recognizes the importance of updating its importation regulations to reflect the current global epidemiology of rabies, particularly the importance of preventing the reintroduction of the canine rabies variant into the United States. HHS/CDC published an Advance Notice of Proposed Rulemaking in 2007 seek- ing input from the public on revising its importation regulations, among other issues, and has used the feedback to draft a new proposed importation regulation. HHS/CDC hopes to publish additional revisions in a Notice of Proposed Rule- making in the coming year that would strengthen protection of public health by addressing some gaps found in the importation of dogs, cats, and other animals. TABLE A13-1â Importations of Rabid Dogs to the Continental United States, 2004â2008 No. of dogs with No. of persons rabies/No. of animals Territory or country receiving PEP/No. of Month/Year in shipment of origin persons assessed May 20041 1/6 Puerto Rico 6/11 June 20042 1/1 Thailand 12/40 March 20073 1/2 India 8/20 June 20084 1/24 Iraq 13/38 Based on data from: 1Personal communication, Fredric Cantor, Massachusetts Department of Public Health. 2Personal communication, Ben Sun, California Department of Public Health. 3Castrodale et al., 2008. 4CDC, 2008b.
APPENDIX A 285 CDCâs Regulations for Turtles: Reemerging Concerns About Salmonella Due to Increased Human Contact with Small Turtles Although Salmonella is prevalent in all reptile populations, to help mini- mize the risk to the U.S. human population, CDC limits imports of small turtles. Turtles with a shell length of less than 4 inches and viable turtle eggs may not be imported for any commercial purpose (HHS/CDC, 2003c); however, noncom- mercial imports of up to six turtles with shells less than 4 inches long or viable eggs are permitted. Seven or more small turtles may be imported for science, edu- cational, or exhibition purposes, but the importer is required to obtain a letter of permission from CDC. This rule was implemented in 1975 after it was discovered that small turtles frequently transmitted Salmonella to humans, particularly young children for whom the small turtles were very popular as pets. Young children are more susceptible to Salmonella infection due to the nature of their interactions with the turtles and lack of hand hygiene. Also in 1975, the FDA prohibited the sale and interstate distribution of viable turtle eggs and live turtles with a shell less than 4 inches long. FDA restricted the legal sale and distribution of small turtles grown in the United States to the export market only, and required the outside of turtle shipping packages to be labeled conspicuously âfor export only.â In 1980, the CDC and FDA bans were estimated to prevent 100,000 Salmonella infections annually in children younger than 10 years old in the United States (Harris et al., 2010). The global trade in pet turtles has changed according to consumer prefer- ences, and as a result zoonotic health risks to humans in the United States as- sociated with turtle contact are resurfacing. The United States produced nearly 9 million baby red-eared slider turtles each year from 1997â2003, and 32 million live turtles were farmed and exported from the United States from 2003 to 2005 (National Geographic News, 2009). However, the Chinese turtle industry has been steadily rising, while the supply of U.S.-grown turtles is increasing with no legal outlet. Thus some U.S. suppliers are skirting the FDA regulations that ban the domestic sale of live turtles with a shell less than 4 inches long. They accomplish this by selling turtles in flea markets or from roadside vendors for âeducationalâ purposes, or by selling the terrarium and giving the turtle for âfree.â These factors, combined with the lack of regulations prohibiting Internet turtle sales, may be contributing to the dramatic increase in turtle ownership over the last 5 years (American Veterinary Medical Association, 2012). From May 2011 to September 2013, CDC reported 473 human Salmonella infections from 41 different U.S. states; 70 percent of ill persons were children 10 years of age or younger; 88 percent of ill persons specifically reported exposure to small turtles (shell length less than 4 inches) (CDC, 2013). In response to this emerging issue, in 2013 CDC warned public health officials globally about the public health risks associated with exportation of animals known to be infected with Salmonella.
286 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS CDCâs Regulations for Nonhuman Primates NHPs, particularly those recently captured in the wild, may harbor agents in their blood or other body tissues that are infectious to humans. NHPs and NHP products are a potential source of pathogens that can cause severe or fatal dis- ease in humans, including filoviruses, hepatitis A and B viruses, herpes B virus, rabies, tuberculosis, and parasitic infections (NRC, 2003). Cynomolgus, African green, and rhesus monkeys have been associated with outbreaks of hemorrhagic fever viruses such as Ebola and Marburg. Cynomolgus monkeys imported into the United States were previously demonstrated to be infected with Ebola Reston virus (CDC, 1990). Persons working in temporary and long-term animal holding facilities, and individuals involved in transporting animals (e.g., cargo handlers and inspectors) are especially at risk for infection. An epidemiologic link between hepatitis A infections in NHPs, particularly chimpanzees, and their caretakers has been demonstrated (Robertson, 2001). Herpes B virus is a zoonotic agent that naturally infects only macaque monkeys, causing mild illness or no illness, but can cause a fatal encephalomyelitis in humans. Previously reported fatal cases of herpes B virus disease in humans have been caused by animal bites, scratches, or mucous membrane contact with infected materials (Cohen et al., 2002). NHPs, especially macaques, are highly susceptible to tuberculosis; most are imported from areas of the world with a high prevalence of these diseases in humans and animals (CDC, 1993). NHPs might also be a source of flaviviruses (e.g., yel- low fever virus), which can be transmitted to humans by mosquitoes that have previously fed on an infected NHP (Mansfield and King, 1998). Transmission of yellow fever from NHPs to humans through vectors has occurred (Richardson, 1987). Imported NHPs have also been known to contract melioidosis, caused by Burkholderia pseudomallei (Johnson et al., 2013). U.S. quarantine requirements for imported NHPs are designed to reduce these infectious disease risks. While NHPs have been regulated by public health authorities since the 1950s, CDC has prohibited the importation of NHPs except for scientific, educational, or exhibition purposes since 1975. Current regulations maintain that imported NHPs and the offspring of NHPs imported after 1975 may not be maintained as pets or as an avocation with occasional display to the general public. Also under current regulations, NHP importers are required to register with CDC, and this registration must be renewed every 2 years. NHPs are required to be held in quarantine for a minimum of 31 days following entry into the United States. CDCâs regulations also require registered importers to maintain records on imported NHPs and to immediately report NHP illnesses or deaths that occur during the 31-day quarantine period to CDC. During the quarantine period, all imported NHPs must complete three negative tuberculin skin tests. Additional requirements for importers of NHPs were developed and imple- mented in response to specific public health threats. On January 19, 1990, CDC published interim guidelines for handling NHPs during transit and quarantine in response to identification of Ebola virus (Reston strain) in NHPs imported to the
APPENDIX A 287 United States from the Philippines (CDC, 1990). In April 1990, there was confir- mation of Ebola virus infection but no apparent disease in four NHP caretakers. As a result of these findings, as well as serologic evidence that African green, cynomolgus, and rhesus monkeys potentially pose a threat of human exposure to filoviruses, CDC placed additional restrictions and permit requirements on im- porters wishing to import these species. In 2013, CDC codified these additional restrictions into comprehensive nonhuman primate regulations found at 42 Code of Federal Regulations (CFR) 71.53 (HHS/CDC, 2013). CDCâs Regulations for Bats Under 42 CFR 71.54, CDC restricts the importation of infectious biological agents, infectious substances, and animal vectors of human disease. Importers must satisfy several conditions and obtain a CDC import permit before import- ing any infectious biological agents, infectious substances, or animal vectors of human disease. Because they are viewed as âanimal vectors of human disease,â all imported bats require an import permit from CDC (HHS/CDC, 2003b). Bat imports additionally require a permit from the U.S. Department of Interiorâs Fish and Wildlife Service for reasons related to wildlife conservation and prevention of the introduction of invasive wildlife species into the United States. Bats can serve as reservoirs for many zoonotic infectious agents and are prohibited from import for personal use as a pet. Marburg virus is clearly associated with a species of bat called Rousettus aegyptiacus, at least in Uganda, and one or more other species are almost surely associated with Ebola virus (Calisher et al., 2006). In addition, bats are known to be the keystone reservoirs for rabies virus, other lyssaviruses related to rabies, and henipaviruses, and were identified as the reservoir for severe acute respiratory syndrome (SARS) coronavirus (Cui et al., 2007). CDCâs Regulations for Rodents: Monkeypox Case Study The emergence of human monkeypox in the Western Hemisphere in Mayâ June 2003 was a vivid reminder of why importation of wild animals into the United States is a concern from a public health perspective. Monkeypox is a zoo- notic disease endemic to Central and West Africa. African rodents are considered to be the natural host of the virus which, in humans, causes a systemic febrile illness and rashes similar to smallpox (CDC, 2004b; Khodakevich et al., 1988). Human infections during the 2003 outbreak were traced back to contact with pet prairie dogs that had contracted monkeypox from diseased African rodents im- ported for the commercial pet trade (CDC, 2003; Hutson et al., 2007; Reed et al., 2004). The shipment of mammals imported from Ghana contained more than six species and a total of 762 African rodents, some of which were confirmed to be infected with monkeypox virus. The U.S. monkeypox outbreak resulted in 72 hu- man cases, 37 of which were laboratory confirmed (CDC, 2003). Most patients
288 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS had direct or close contact with the infected prairie dogs, including 28 children at a day care center and veterinary clinic staff (Reynolds et al., 2007). On June 11, 2003, CDC and the FDA, pursuant to 42 CFR 70.32(b) and 21 CFR 1240.30, respectively, issued a joint order prohibiting, until further notice, the transportation or offering of transportation in interstate commerce, or the sale, offering for sale, or offering for any other type of commercial or public distribution, including release into the environment of prairie dogs and the six implicated species of African rodents (FDA, 2003; Gerberding and McClellan, 2003). In addition, pursuant to 42 CFR 71.32(b), CDC implemented an immedi- ate embargo on the importation of all rodents (order Rodentia) from Africa. This emergency order was superseded on November 4, 2003, when the two agencies issued an interim final rule creating two complementary regulations restricting both domestic trade and importation, intended to prevent the further introduction, establishment, and spread of the monkeypox virus in the United States. In 2008 the FDA portion restricting interstate movement was lifted because there was no evidence that monkeypox virus was continuing to circulate in the United States. It was agreed that CDCâs ban on importation of African rodents was sufficient to protect public health and therefore should remain in place (HHS/CDC, 2003d). Rodents that originate outside of Africa, from other parts of the world in- cluding Asia, Europe, and South America, also constitute a significant public health risk. In addition to harboring poxviruses, rodents are also known to carry hemorrhagic fever viruses, arenaviruses, hantaviruses, rickettsioses, and parasites (Azad, 1990; Eremeeva and Dasch, 2008; Heymann, 2008; Hugh-Jones et al., 1995). CDC conducted an analysis of the numbers and origins of rodents imported to the United States since CDCâs African rodent ban was instituted in 2003. CDC analyzed data from the United States Fish and Wildlife Serviceâs Law Enforce- ment Management Information System (LEMIS) database, which records the entry of wildlife species to the United States. Since 2003, the CDC ban has ef- fectively limited legal importation of African rodents; the number of rodents from Africa entering the United States decreased by 90 percent. However, the commer- cial pet market has found a new niche in rodents from other parts of the world, as the number of rodents from Europe, Canada, and South America imported to the United States increased by 300 to 29,800 percent (CDC, unpublished data). These data illustrate the need to reevaluate whether the rodent ban should be expanded to restrict the importation of all rodents to the United States. CDCâs Regulations for Civets: SARS Case Study In 2003, CDC issued an emergency embargo restricting the importation of civets under 42 CFR 71.32(b). This action was taken because of civetsâ potential to serve as an amplifying host for transmission of SARS coronavirus to humans.
APPENDIX A 289 CDC has interpreted this ban broadly to include all members of the family Vi- verridae. Although bats were ultimately discovered to be the vector of SARS, concerns about the unique susceptibility of members of the family Viverridae to SARS coronavirus, and the extremely high viral load that they experience as a result of infection, keeps the embargo in place, with exceptions allowing importa- tion by permit for science, exhibition, or education (CDC, 2004a). CDCâs Regulations on Products from Restricted Animals: Bushmeat In other parts of the world, especially parts of West Africa, mammalian wild- life species (including rodents, bats, antelope, and NHPs) serve as an important human food source called bushmeat. Bushmeat is highly desired among many African expatriates, and when shipped it is typically not treated (e.g., cooked) to render it noninfectious. Because of concerns about bushmeatâs potential to transmit zoonotic diseases, CDC prohibits the importation of mammalian species listed in 42 CFR part 71. The Bushmeat Crisis Taskforce, an organization that works to prevent the illegal harvesting of meat from wild animals, estimates that approximately 15,000 pounds of meat harvested from African wildlife are illegally imported into the United States each month (Goldman, 2007). Inspection capabilities at U.S. ports of entry are limited, given the scope of trade and travel. For example, in 2012 there were 373,441 nonresident passenger arrivals in the United States from Af- rica, representing an 86 percent increase in travel since 2004 (International Trade Association, 2012). From September 2005 to December 2010, there were 543 seizures of CDC-prohibited bushmeat at U.S. ports of entry. Nearly 80 percent of bushmeat confiscated by CDC was from West Africa. Half of these confiscations were rodents; included among the confiscations were specimens of NHPs, birds, porcupines, and bats (Bair-Brake et al., 2013). Many of the animals eaten as bushmeat harbor pathogens that are dangerous to humans, and there are no regulations to ensure that bushmeat is properly de- contaminated or otherwise safe for human consumption. Viral pathogens carried by NHPs pose the greatest public health risk. Human immunodeficiency virus infections probably originated from chimpanzee-to-human transmission through hunting and butchering the animals eaten as bushmeat; the virus then adapted to human hosts and was subsequently transmitted from human to human (Hahn et al., 2000). Ebola virus has been detected in chimpanzees in Côte dâIvoire, and human infections with the virus were found to be associated with contact between chimpanzee hunters and dead chimpanzees (Formenty et al., 1999). Simian im- munodeficiency virus was detected in 131 out of 788 (16.6 percent) serum sam- ples from monkeys in Cameroon (Peters et al., 2002), and simian T-lymphotropic virus has been identified in monkeys from Cameroon (Courgnaud et al., 2004). During 2008â2010, a pilot project conducted by CDC and partner organizations
290 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS identified simian foamy virus, cytomegalovirus, and lymphocryptovirus in NHP specimens confiscated at U.S. ports of entry. These viruses have been shown to infect humans (Smith et al., 2012a). Even with regulations in place, inconsistent surveillance and enforcement of penalties at U.S. ports of entry makes bushmeat importation difficult to con- trol; bushmeat continues to be found in mail, checked or hand-carried luggage, and cargo on international flights. CDC, FDA, and other federal agencies are forming a multiagency working group to coordinate efforts to prevent bushmeat importation. Challenges to Preventing Zoonotic Diseases Associated with Animal Importation and Exportation As stated earlier, five federal agencies are responsible for regulating live ani- mal movement. A 2010 Government Accountability Office report on live animal importation concluded that agencies needed to work more closely together to reduce the risk of importing zoonotic animal-related diseases (U.S. Government Accountability Office, 2010). Areas of major concern cited by the reportâs authors included the need to identify and resolve differing program priorities, identify and leverage resources, examine ways to improve data sharing on live animal imports, and assess whether any additional legislative authority was needed to ensure that live animal imports posing a risk of zoonotic and animal diseases do not enter the United States. There are loopholes in current regulations, and instances where CDC lacks statutory authority to enact certain regulations. For example, CDCâs existing regulations allow the importation of dogs too young to be vaccinated for rabies. Furthermore, CDC lacks the statutory authority under the Public Health Service Act to regulate exportation of animals that carry pathogens known to infect hu- mans. Reports of human Salmonella infections were traced to frozen mice sold by a U.S. company to a distributor in the United Kingdom, who then sold the mice over the Internet to U.S. and Canadian customers. Possible routes of trans- mission to humans included handling frozen or thawed mice, handling reptiles infected by consumption of the mice, handling or cleaning the reptileâs habitat, or cross-contamination in the kitchen where mice were thawed. The sale of these infected mice resulted in hundreds of human Salmonella infections, including at least 34 cases in 17 U.S. states and over 500 cases in the United Kingdom (CDC, 2010; Harker et al., 2011). In addition, there have been recent reports of human Salmonella infections abroad associated with contact with small turtles that had been exported from the United States (Angulo et al., 2010). Fiscal and human resource constraints limit regular inspection and surveil- lance of shipments, cargo, and luggage and consistent enforcement of CDC regulations at U.S. ports of entry. State and local public health agencies are often asked to help enforce CDC regulations such as dog confinement agreements.
APPENDIX A 291 With increasing strain on their budgets, state and local public health officials are less able to comply. Recommendations There are multiple regulatory and operational challenges to preventing zoo- notic diseases associated with importation of animals, animal products, and other potentially infectious items into the United States. In these instances, updating and strengthening regulations and import restrictions applied to a wider range of species than currently regulated could be the only effective means of preventing the introduction of exotic infections into the United States. CDC is working to- wards amending its regulations to institute further requirements and restrictions for entry of dogs and other animals of concern into the United States, and it is working on a proposal to limit the list of the rabies-free countries only to those that have no lyssaviruses. The capacity of U.S. federal agencies that oversee animal importation to identify and track imported animal species and quantify shipments should be enhanced. For example, under the Security and Accountability for Every Port Act of 2006 (SAFE Port Act, Public Law 109-347), federal agencies with regu- latory responsibilities related to importation and exportation of goods are re- quired to participate in the International Trade Data System (ITDS). ITDS, working through U.S. Customs and Border Protectionâs Automated Commercial Environment system, is an electronic information exchange capability, or âsingle window,â through which businesses will transmit data required by participating agencies for the importation or exportation of cargo. ITDS is intended to signifi- cantly enhance compliance with federal regulations related to trade, including CDCâs animal importation regulations. Once fully implemented in FY2017, ITDS will enable CDC, as well as other agencies with a role in regulating the importation of animals, to receive cargo manifest and other data in advance of arrival at a U.S. port of entry, and benefit from automated targeting algorithms that will target anomalies in inbound animal shipments. ITDS should not only improve compliance with CDC animal importation regulations, but also enable more effective coordination with other federal agencies to address risks posed by imported animals. The U.S. federal government, as well as state and local agencies, have a ma- jor role to play in the promulgation, enforcement, and coordination of policy and regulations. However, prevention of diseases arising from animal importation is also a shared responsibility with animal importers, the pet industry, veterinarians, and other health care providers. When establishing an animal as a new patient, veterinarians should inquire about the animalâs origin and its recent history of travel. Veterinarians should also know how to recognize and report foreign animal and zoonotic diseases. Human health care providers should obtain animal contact
292 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS history from their patients and be familiar with zoonotic diseases, including those that are not endemic to the United States. Educational strategies have already been implemented, but they need to be expanded to inform the public about the risks of zoonotic diseases. In 2007, CDC and some of its partners conducted focus group discussions among African expatriates in the United States, many of whom reported consuming bushmeat. A theme that emerged from the discussions included evidence of long-standing cultural practices of hunting and eating bushmeat in their countries of origin, making it difficult for consumers to understand the potential health risks. Focus group participants mentioned that âsince U.S. merchants sell bushmeat, it must be legalâ; thus there is a lack of understanding among consumers about the il- legality of importation (Bair-Brake et al., 2013). This scenario points to the need for further education of consumers and sellers of bushmeat in the United States. Another example is seen through zoonotic transmissions of infectious diseases from pets, such as tularemia, salmonellosis, and lymphocytic choriomeningitis virus infection (CDC, 2004c, 2005, 2008a) and how these incidents have served as opportunities to educate the public about safe handling of animals. Educational materials on the prevention of reptile- and amphibian-associated salmonellosis are available in four languages. Pet retailers have been and can continue to be valuable partners in this effort, and CDC works with this industry in collabora- tion with the Pet Industry Joint Advisory Council (2014). Multiple guidances published by the American Academy of Pediatrics, CDC, and the National As- sociation of State Public Health Veterinarians (American Academy of Pediatrics, 2012; CDC, 2014b; National Association of State Public Health Veterinarians, 2013; Pickering et al., 2008) also remind the public of the dangers of contact with any wildlife, whether imported or domestic. This paper has described selected aspects of CDCâs animal regulations that exist to prevent the introduction of zoonotic diseases to the United States. Each section illustrated examples where (1) regulations exist, but the fluidity of the pet trade has managed to circumvent them, or there are loopholes in the regula- tions, or (2) no specific regulation or policy exists, or (3) human behavior and attitudes towards interaction with animals, animal products, and pet ownership preferences pose public health risks. Multiple challenges exist, but educating the public; assessing whether additional authority is needed; strengthening the regu- lations; instituting stronger surveillance and tracking systems; and recognizing that prevention is a shared responsibility between government, industry, health practitioners, and pet owners are all strategies that can reduce the introduction and spread of zoonotic diseases associated with trade and movement of animals.
APPENDIX A 293 References American Academy of Pediatrics. 2012. Section 2: Recommendations for care of children in special circumstances diseases transmitted by animals (zoonoses): Household pets, including nontradi- tional pets, and exposure to animals in public settings. Red Book: 2012 Report of the Commit- tee on Infectious Disease, edited by L. K. Pickering, 29th ed. Elk Grove Village, IL: American Academy of Pediatrics. American Veterinary Medical Association. 2012. US Pet Ownership Statistics 2012. https://www. avma.org/KB/Resources/Statistics/Pages/Market-research-statistics-US-pet-ownership.aspx (ac- cessed January 9, 2014). Angulo, F. J., J. R. Harris, K. P. Neil, C. B. Behravesh, and M. J. Sotir. 2010. Recent multistate outbreaks of human Salmonella infections acquired from turtles: A continuing public health challenge. Clinical Infectious Diseases 50(4):554-559. Azad, A. F. 1990. Epidemiology of murine typhus. Annual Review of Entomology 35:553-569. Bair-Brake, H., T. Bell, A. Higgins, N. Bailey, M. Duda, S. Shapiro, H. Eves, N. Marano, and G. Galland. 2013. Is that a rodent in your luggage? A mixed method approach to describe bushmeat importation into the United States. Zoonoses and Public Health 61(2):97-104. Calisher, C. H., J. E. Childs, H. E. Field, K. V. Holmes, and T. Schountz. 2006. Bats: Important res- ervoir hosts of emerging viruses. Clinical Microbiology Reviews 19(3):531-545. Castrodale, L., V. Walker, J. Baldwin, J. Hofmann, and C. Hanlon. 2008. Rabies in a puppy imported from India to the USA, March 2007. Zoonoses and Public Health 55(8-10):427-430. CDC (Centers for Disease Control and Prevention). 1990. Update: Ebola-related filovirus infection in nonhuman primates and interim guidelines for handling nonhuman primates during transit and quarantine. Morbidity and Mortality Weekly Report 39(2):22-24, 29-30. CDC. 1993. Tuberculosis in imported nonhuman primatesâUnited States, June 1990-May 1993. Morbidity and Mortality Weekly Report 42(29):572-576. CDC. 2003. Update: Multistate outbreak of monkeypoxâIllinois, Indiana, Kansas, Missouri, Ohio, and Wisconsin, 2003. Morbidity and Mortality Weekly Report 52(27):642-646. CDC. 2004a. Notice of embargo of civets. http://www.cdc.gov/sars/media/civet-ban.html (accessed March 14, 2014). CDC. 2004b. Updated interim case definition for human monkeypox, January 2004, http://www.cdc. gov/ncidod/monkeypox/casedefinition.htm (accessed March 14, 2014). CDC, 2004c. Brief report: Tularemia associated with a hamster biteâColorado, 2004. http://www. cdc.gov/mmwr/preview/mmwrhtml/mm5351a3.htm (accessed March 14, 2014). CDC. 2005. Outbreak of multidrug-resistant Salmonella typhimurium associated with rodents pur- chased at retail pet storesâUnited States, December 2003-October 2004. Morbidity and Mortal- ity Weekly Report 54(17):429-433. CDC. 2008a. Brief report: Lymphocytic choriomeningitis virus transmitted through solid organ transplantationâMassachusetts, 2008. Morbidity and Mortality Weekly Report 57(29):799-801. CDC. 2008b. Rabies in a dog imported from IraqâNew Jersey, June 2008. Morbidity and Mortality Weekly Report 57(40):1076-1078. CDC. 2010. Multistate outbreak of human Salmonella Infections associated with frozen rodents (final update). http://www.cdc.gov/salmonella/frozenrodents (accessed February 8 2014). CDC. 2013. Eight multistate outbreaks of human Salmonella infections linked to small turtles (fi- nal update). http://www.cdc.gov/salmonella/small-turtles-03-12/index.html (accessed January 9 2014). CDC. 2014a. Rabies-free countries and political units. http://www.cdc.gov/animalimportation/rabies- free-countries.html (accessed Jan 4, 2014). CDC. 2014b. Healthy pets healthy people. http://www.cdc.gov/healthypets/ (accessed February 8, 2014).
294 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Cohen, J. I., D. S. Davenport, J. A. Stewart, S. Deitchman, J. K. Hilliard, and L. E. Chapman. 2002. Recommendations for prevention of and therapy for exposure to B virus (Cercopithecine her- pesvirus 1). Clinical Infectious Diseases 35(10):1191-1203. Courgnaud, V., S. Van Dooren, F. Liegeois, X. Pourrut, B. Abela, S. Loul, E. Mpoudi-Ngole, A. Van- damme, E. Delaporte, and M. Peeters. 2004. Simian T-cell leukemia virus (STLV) infection in wild primate populations in Cameroon: Evidence for dual STLV type 1 and type 3 infection in agile mangabeys (Cercocebus agilis). Journal of Virology 78(9):4700-4709. Cui, J., N. Han, D. Streicker, G. Li, X. Tang, Z. Shi, Z. Hu, G. Zhao, A. Fontanet, and Y. Guan. 2007. Evolutionary relationships between bat coronaviruses and their hosts. Emerging Infectious Diseases 13(10). Eremeeva, M. E., and G. A. Dasch. 2008. Rickettsial infections. In CDC health information for inter- national travel, edited by P. M. Arguin, P. E. Kozarsky, and C. Reed. Philadelphia, PA: Elsevier. FDA (Food and Drug Administration). 2003. Control of communicable diseases. Federal Register 68(117) http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?CFRPart= 1240 (accessed March 14, 2014). Formenty, P., C. Boesch, M. Wyers, C. Steiner, F. Donati, F. Dind, F. Walker, and B. Le Guenno. 1999. Ebola virus outbreak among wild chimpanzees living in a rain forest of Cote dâIvoire. Journal of Infectious Diseases 179(Supplement 1):S120-S126. Gerberding, J. L., and M. B. McClellan. 2003. Joint order of the Centers for Disease Control and Prevention and the Food and Drug Administration, Department of Health and Human Services, http://www.gpo.gov/fdsys/pkg/FR-2003-11-04/html/03-27557.htm (accessed March 14, 2014). Goldman, R. 2007. Bushmeat: Curse of the monkeyâs paw. ABC News. http://abcnews.go.com/Health/ story?id=2952077 (accessed March 14, 2014). Hahn, B. H., G. M. Shaw, K. M. De, and P. M. Sharp. 2000. AIDS as a zoonosis: Scientific and public health implications. Science 287(5453):607-614. Harker, K., C. Lane, E. De Pinna, and G. Adak. 2011. An outbreak of Salmonella typhimurium DT191a associated with reptile feeder mice. Epidemiology and Infection 139(08):1254-1261. Harris, J. R., K. P. Neil, C. B. Behravesh, M. J. Sotir, and F. J. Angulo. 2010. Recent multistate outbreaks of human Salmonella infections acquired from turtles: A continuing public health challenge. Clinical Infectious Diseases 50(4):554-559. Heymann, D. L. 2008. Control of communicable diseases manual. Washington, DC: American Public Health Association. HHS (Health and Human Services). 2001. Part 71: Foreign Quarantine. Federal Code of Regulations: Title 42âPublic Health). http://www.gpo.gov/fdsys/pkg/CFR-2003-title42-vol1/content-detail. html (accessed March 14, 2014). HHS/CDC. 2003a. Regulations on the importation of dogs and cats, http://www.gpo.gov/fdsys/pkg/ CFR-2003-title42-vol1/pdf/CFR-2003-title42-vol1-sec71-51.pdf (accessed March 14, 2014). HHS/CDC. 2003b. Regulations on etiologic agents and vectors, http://www.gpo.gov/fdsys/pkg/ CFR-2003-title42-vol1/pdf/CFR-2003-title42-vol1-sec71-54.pdf (accessed March 14, 2014). HHS/CDC. 2003c. Regulations on turtles, http://www.gpo.gov/fdsys/pkg/CFR-2003-title42-vol1/pdf/ CFR-2003-title42-vol1-sec71-52.pdf (accessed March 14, 2014). HHS/CDC 2003d. Regulations on rodents, http://www.gpo.gov/fdsys/pkg/FR-2003-11-04/html/03- 27557.htmn (accessed January 9, 2014). HHS/CDC. 2013. Regulations on control of communicable disease; Foreign requirements for importers of nonhuman primates (NHP). https://www.federalregister.gov/articles/2013/02/15/2013-03064/ control-of-communicable-disease-foreign-requirements-for-importers-of-nonhuman-primates- nhp (accessed January 9, 2014). Huang, J. J., K. H. You, C. F. Lin, C. Y. Wang, C. L. Liu, and J. J. Yen. 2013. Taiwanâs strategies in response to re-emergence of animal rabies. Taiwan Epidemiology Bulletin 29 (Supplement): 11-26-2013. http://www.cdc.gov.tw/english/info.aspx?treeid=412DF4F760DD5617&nowtreei d=16112473A81AA6A7&tid=F0C7C38B01171DBB (accessed Jan 4, 2014).
APPENDIX A 295 Hugh-Jones, M. E., W. T. Hubbert, and H. V. Hagstad. 1995. Zoonoses: Recognition, control, and prevention. Ames, IA: Iowa State University Press. Hutson, C. L., K. N. Lee, J. Abel, D. S. Carroll, J. M. Montgomery, V. A. Olson, Y. Li, W. Davidson, C. Hughes, M. Dillon, P. Spurlock, J. J. Kazmierczak, C. Austin, L. Miser, F. E. Sorhage, J. Howell, J. P. Davis, M. G. Reynolds, Z. Braden, K. L. Karem, I. K. Damon, and R. L. Regnery. 2007. Monkeypox zoonotic associations: Insights from laboratory evaluation of animals asso- ciated with the multi-state US outbreak. American Journal of Tropical Medicine and Hygiene 76(4):757-767. International Trade Association. 2012. Office of Travel and Tourism Industries 2012 nonresident arrivals from Africa. http://travel.trade.gov/view/m-2012-I-001/table2.html (accessed January 9, 2014). Johnson, C. H., B. L. Skinner, S. M. Dietz, D. Blaney, R. M. Engel, G. W. Lathrop, A. R. Hoffmaster, J. E. Gee, M. G. Elrod, and N. Powell. 2013. Natural infection of Burkholderia pseudomallei in an imported pigtail macaque (Macaca nemestrina) and management of the exposed colony. Comparative Medicine 63(6):528-535. Khodakevich, L., Z. Jezek, and D. Messinger. 1988. Monkeypox virus: Ecology and public health significance. Bulletin of the World Health Organization 66(6):747-752. Mansfield, K., and N. King. 1998 Nonhuman primates in biomedical research. In Viral Disease, edited by B. T. Bennett, C. R. Abee, and R. Henrickson. San Diego, CA: Academic Press. National Association of State Public Health Veterinarians. 2013. Compendium of measures to prevent disease associated with animals in public settings, 2013. Journal of the American Veterinary Medical Association 243 (9):1270-1288. National Geographic News. 2009. http://news.nationalgeographic.com/news/2009/07/090724-turtles- china_2.html (accessed January 9, 2014). Pet Industry Joint Advisory Council. 2014. http://www.pijac.org/ (accessed February 6, 2014). Peters, M., V. Courgnaud, B. Abela, P. Auzel, X. Pourrut, F. Bibollet-Ruche, S. Loul, F. Liegeois, C. Butel, and D. Koulagna. 2002. Risk to human health from a plethora of simian immunodefi- ciency viruses in primate bushmeat. Emerging Infectious Diseases 8(5):451-457. Pickering, L. K., N. Marano, J. A. Bocchini, and F. J. Angulo. 2008. Exposure to nontraditional pets at home and to animals in public settings: Risks to children. Pediatrics 122(4):876-86. Reed, K. D., J. W. Melski, M. B. Graham, R. L. Regnery, M. J. Sotir, M. V. Wegner, J. J. Kazmierczak, E. J. Stratman, Y. Li, J. A. Fairley, G. R. Swain, V. A. Olson, E. K. Sargent, S. C. Kehl, M. A. Frace, R. Kline, S. L. Foldy, J. P. Davis, and I. K. Damon. 2004. The detection of monkeypox in humans in the Western Hemisphere. New England Journal of Medicine 350(4):342-350. Reynolds, M. G., W. B. Davidson, A. T. Curns, C. S. Conover, G. Huhn, J. P. Davis, M. Wegner, D. R. Croft, A. Newman, N. N. Obiesie, G. R. Hansen, P. L. Hays, P. Pontones, B. Beard, R. Teclaw, J. F. Howell, Z. Braden, R. C. Holman, K. L. Karem, and I. K. Damon. 2007. Spectrum of infection and risk factors for human monkeypox, United States, 2003. Emerging Infectious Diseases 13(9):1332-1339. Richardson, J. H. 1987. Basic considerations in assessing and preventing occupational infections in personnel working with nonhuman primates. Journal of Medical Primatology 16(2):83-89. Robertson, B. H. 2001. Viral hepatitis and primates: Historical and molecular analysis of human and nonhuman primate hepatitis A, B, and the GB-related viruses. Journal of Viral Hepatitis 8(4):233-242. Romagosa, C. M. 2010. Center for Forest Sustainability, School of Forestry and Wildlife Sciences, Auburn University; A summary of live animal importation by the United States. http://www. evergladescisma.org/SummaryofUSliveanimalimports.pdf (accessed December 23, 2013). Smith, K. M., S. J. Anthony, W. M. Switzer, J. H. Epstein, T. Seimon, H. Jia, M. D. Sanchez, T. T. Huynh, G. G. Galland, and S. E. Shapiro. 2012. Zoonotic viruses associated with illegally im- ported wildlife products. PloS One 7(1):e29505.
296 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Tsiodras, S., G. Dougas, A. Baka, C. Billinis, S. Doudounakis, A. Balaska, T. Georgakopoulou, G. Rigakos, V. Kontos, K. E. Tasioudi, M. Tzani, P. Tsarouxa, P. Iliadou, O. Mangana-Vougiouka, D. Iliopoulos, S. Sapounas, P. Efstathiou, A. Tsakris, C. Hadjichristodoulou, and J. Kremasti- nou. 2013. Re-emergence of animal rabies in northern Greece and subsequent human exposure, October 2012 - March 2013. Eurosurveillance 18(18):20474. U.S. Government Accountability Office. 2010. Live animal imports: Agencies need better collabo- ration to reduce the risk of animal-related diseases. http://www.gao.gov/products/GAO-11-9 (accessed January 9, 2014). Velasco-Villa, A., S. A. Reeder, L. A. Orciari, P. A. Yager, R. Franka, J. D. Blanton, L. Zuckero, P. Hunt, E. H. Oertli, L. E. Robinson, and C. E. Rupprecht. 2008. Enzootic rabies elimination from dogs and reemergence in wild terrestrial carnivores, United States. Emerging Infectious Diseases 14(12):1849-1854. WHO (World Health Organization). 2009. RabiesâBulletinâEurope. Rabies Information System of the WHO Collaborating Centre for Rabies Surveillance and Research. http://www.who-rabies- bulletin.org/Queries/Surveillan ce.aspx (accessed March 14 2014).
APPENDIX A 297 A14 THE GLOBAL DISTRIBUTION AND BURDEN OF DENGUE59 Samir Bhatt,60 Peter W. Gething,60 Oliver J. Brady,60,61 Jane P. Messina,60 Andrew W. Farlow,60 Catherine L. Moyes,60 John M. Drake,60,62 John S. Brownstein,63 Anne G. Hoen,64 Osman Sankoh,65,66,67 Monica F. Myers,60 Dylan B. George,68 Thomas Jaenisch,69 G. R. William Wint,60,70 Cameron P. Simmons,71,72 Thomas W. Scott,60,73 Jeremy J. Farrar,63,64,74 and Simon I. Hay60,68 Dengue is a systemic viral infection transmitted between humans by Aedes mosquitoes (Simmons et al., 2012). For some patients, dengue is a life- threatening illness (WHO, 2009). There are currently no licensed vaccines or specific therapeutics, and substantial vector control efforts have not stopped its rapid emergence and global spread (Tatem et al., 2006). The contempo- rary worldwide distribution of the risk of dengue virus infection (Brady et al., 2012) and its public health burden are poorly known (Halstead, 1988; WHO, 2009). Here we undertake an exhaustive assembly of known records of dengue occurrence worldwide, and use a formal modelling framework to map the global distribution of dengue risk. We then pair the resulting risk map with detailed longitudinal information from dengue cohort studies and 59ââBhatt et al. 2013. The global distribution and burden of dengue. Nature 496:504-507. Reprinted with permission from Nature Publishing Group. 60ââSpatial Ecology and Epidemiology Group, Tinbergen Building, Department of Zoology, Univer- sity of Oxford, South Parks Road, Oxford OX1 3PS, UK. 61ââOxitec Limited, Milton Park, Abingdon OX14 4RX, UK. 62ââOdum School of Ecology, University of Georgia, Athens, GA 30602, USA. 63ââDepartment of Pediatrics, Harvard Medical School and Childrenâs Hospital Informatics Program, Boston Childrenâs Hospital, Boston, Massachusetts 02115, USA. 64ââDepartment of Community and Family Medicine, Geisel School of Medicine, Dartmouth College, Hanover, NH 03755, USA. 65ââINDEPTH Network Secretariat, East Legon, PO Box KD 213, Accra, Ghana. 66ââSchool of Public Health, University of the Witwatersrand, Braamfontein 2000, Johannesburg, South Africa. 67ââInstitute of Public Health, University of Heidelberg, 69120 Heidelberg, Germany. 68ââFogarty International Center, National Institutes of Health, Bethesda, MD 20892, USA. 69ââSection Clinical Tropical Medicine, Department of Infectious Diseases, Heidelberg University Hospital, INF 324, D 69120 Heidelberg, Germany. 70ââEnvironmental Research Group Oxford (ERGO), Tinbergen Building, Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK. 71ââOxford University Clinical Research Unit, Hospital for Tropical Diseases, Ho Chi Minh City, Vietnam. 72ââCentre for Tropical Medicine, University of Oxford, Churchill Hospital, Oxford OX3 7LJ, UK. 73ââDepartment of Entomology, University of California Davis, Davis, CA 95616, USA. 74ââDepartment of Medicine, National University of Singapore, 119228 Singapore.
298 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS population surfaces to infer the public health burden of dengue in 2010. We predict dengue to be ubiquitous throughout the tropics, with local spatial variations in risk influenced strongly by rainfall, temperature and the degree of urbanization. Using cartographic approaches, we estimate there to be 390 million (95% credible interval 284â528) dengue infections per year, of which 96 million (67â136) manifest apparently (any level of clinical or subclinical severity). This infection total is more than three times the dengue burden estimate of the World Health Organization (2009). Stratification of our esti- mates by country allows comparison with national dengue reporting, after taking into account the probability of an apparent infection being formally reported. The most notable differences are discussed. These new risk maps and infection estimates provide novel insights into the global, regional and national public health burden imposed by dengue. We anticipate that they will provide a starting point for a wider discussion about the global impact of this disease and will help to guide improvements in disease control strate- gies using vaccine, drug and vector control methods, and in their economic evaluation. Dengue is an acute systemic viral disease that has established itself glob- ally in both endemic and epidemic transmission cycles. Dengue virus infection in humans is often inapparent (Endy et al., 2011; Simmons et al., 2012) but can lead to a wide range of clinical manifestations, from mild fever to potentially fatal dengue shock syndrome (WHO, 2009). The lifelong immunity developed after infection with one of the four virus types is type-specific (Simmons et al., 2012), and progression to more serious disease is frequently, but not exclusively, associated with secondary infection by heterologous types (Halstead, 1988; WHO, 2009). No effective antiviral agents yet exist to treat dengue infection and treatment therefore remains supportive (WHO, 2009). Furthermore, no licensed vaccine against dengue infection is available, and the most advanced dengue vac- cine candidate did not meet expectations in a recent large trial (Halstead, 2012; Sabchareon et al., 2012). Current efforts to curb dengue transmission focus on the vector, using combinations of chemical and biological targeting of Aedes mosqui- toes and management of breeding sites (WHO, 2009). These control efforts have failed to stem the increasing incidence of dengue fever epidemics and expansion of the geographical range of endemic transmission (Gubler, 1998). Although the historical expansion of this disease is well documented, the potentially large bur- den of ill-health attributable to dengue across much of the tropical and subtropical world remains poorly enumerated. Knowledge of the geographical distribution and burden of dengue is essential for understanding its contribution to global morbidity and mortality burdens, in determining how to allocate optimally the limited resources available for dengue control, and in evaluating the impact of such activities internationally. Addition- ally, estimates of both apparent and inapparent infection distributions form a key
APPENDIX A 299 requirement for assessing clinical surveillance and for scoping reliably future vaccine demand and delivery strategies. Previous maps of dengue risk have used various approaches combining historical occurrence records and expert opinion to demarcate areas at endemic risk (Beatty et al., 2009; Van Kleef et al., 2009; WHO, 2012). More sophisticated risk-mapping techniques have also been imple- mented (Hales et al., 2002; Rogers et al., 2006), but the empirical evidence base has since been improved, alongside advances in disease modelling approaches. Furthermore, no studies have used a continuous global risk map as the foundation for dengue burden estimation. The first global estimates of total dengue virus infections were based on an assumed constant annual infection rate among a crude approximation of the population at risk (10% in 1 billion [Halstead, 1988] or 4% in 2 billion [Monath, 1994]), yielding figures of 80â100 million infections per year worldwide in 1988 (Halstead, 1988; Monath, 1994). As more information was collated on the ratio of dengue haemorrhagic fever to dengue fever cases, and the ratio of deaths to dengue haemorrhagic fever cases, the global figure was revised to 50â100 million infections (Rigau-Pérez et al., 1998; Rodhain, 1996), although larger estimates of 100â200 million have also been made (Van Kleef et al., 2009) (Figure A14-1). These estimates were intended solely as approximations but, in the absence of better evidence, the resulting figure of 50â100 million infections per year is widely cited and currently used by the World Health Organization (WHO). As the methods used were informal, these estimates were presented without confidence intervals, and no attempt was made to assess geographical or temporal variation in incidence or the inapparent infection reservoir. Here we present the outcome of a new project to derive an evidence-based map of dengue risk and estimates of apparent and inapparent infections world- wide on the basis of the global population in 2010. We compiled a database of 8,309 geo-located records of dengue occurrence from a systematic search, resulting from 2,838 published literature sources as well as newer online re- sources (Freifeld et al., 2008) (see Supplementary Information, section A; the full bibliography (Brady et al., 2012) and occurrence data are available from authors on request). Using these occurrence records we: chose a set of gridded en- vironmental and socioeconomic covariates known, or proposed, to affect dengue transmission (see Supplementary Information, section B); incorporated recent work assessing the strength of evidence on national and subnational-level dengue present/absent status (Brady et al., 2012) (Figure A14-2a); and built a boosted regression tree (BRT) statistical model of dengue risk that addressed the limita- tions of previous risk maps (see Supplementary Information, section C) to define the probability of occurrence of dengue infection (dengue risk) within each 5 km à 5 km pixel globally (Figure A14-2b). The model was run 336 times to reflect parameter uncertainty and an ensemble mean map was created (see Supplemen- tary Information, section C). We then combined this ensemble map with detailed longitudinal information on dengue infection incidence from cohort studies and
300 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A14-1âGlobal estimates of total dengue infections. Comparison of previous estimates of total global dengue infections in individuals of all ages, 1985â2010. Black triangle (Halstead, 1988); dark blue triangle (Monath, 1998); green triangle (Rodhain, 1996); orange triangle (Rigau-Pérez, 1998); light blue triangle (TDR/WHO, 2006); pink triangle (Beatty et al., 2009); red triangle, apparent infections from this study. Estimates are aligned to the year of estimate and, if not stated, aligned to the publication date. Red shading marks the credible interval of our current estimate, for comparison. Error bars from Beatty et al. (2010) and Rigau-Pérez (1998) replicated the confidence intervals pro- vided in these publications. built a non-parametric Bayesian hierarchical model to describe the relationship between dengue risk and incidence (see Supplementary Information, section D). Finally, we used the estimated relationship to predict the number of apparent and inapparent dengue infections in 2010 (see Supplementary Information, section E). Our definition of an apparent infection is consistent with that used by the cohort studies: an infection with sufficient severity to modify a personâs regular sched- ule, such as attending school. This definition encompasses any level of severity of the disease, including both clinical and subclinical manifestations. We predict that dengue transmission is ubiquitous throughout the tropics, with the highest risk zones in the Americas and Asia (Figure A14-2b). Validation
APPENDIX A 301 FIGURE A14-2â Global evidence consensus, risk and burden of dengue in 2010. a, Na- tional and subnational evidence consensus on complete absence (green) through to com- plete presence (red) of dengue (Brady et al., 2012). b, Probability of dengue occurrence at 5 km à 5 km spatial resolution of the mean predicted map (area under the receiver operator curve of 0.81 (±0.02 s.d., n = 336)) from 336 boosted regression tree models. Areas with a high probability of dengue occurrence are shown in red and areas with a low probability in green. c, Cartogram of the annual number of infections for all ages as a proportion of national or subnational (China) geographical area.
302 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS statistics indicated high predictive performance of the BRT ensemble mean map with area under the receiver operating characteristic (AUC) of 0.81 (±0.02 s.d., n = 336) (see Supplementary Information, section C). Predicted risk in Africa, although more unevenly distributed than in other tropical endemic regions, is much more widespread than suggested previously. Africa has the poorest record of occurrence data and, as such, increased information from this continent would help to define better the spatial distribution of dengue within it and to improve its derivative burden estimates. We found high levels of precipitation and tempera- ture suitability for dengue transmission to be most strongly associated among the variables considered with elevated dengue risk, although low precipitation was not found to limit transmission strongly (see Supplementary Information, section C). Proximity to low-income urban and peri-urban centres was also linked to greater risk, particularly in highly connected areas, indicating that human move- ment between population centres is an important facilitator of dengue spread. These associations have previously been cited (Gubler, 1998), but have not been demonstrated at the global scale and highlight the importance of including socio- economic covariates when assessing dengue risk. We estimate that there were 96 million apparent dengue infections globally in 2010 (Table A14-1). Asia bore 70% (67 (47â94) million infections) of this bur- den, and is characterized by large swathes of densely populated regions coincid- ing with very high suitability for disease transmission. India (Chakravarti et al., 2012; Kakkar, 2012) alone contributed 34% (33 (24â44) million infections) of the global total. The disproportionate infection burden borne by Asian countries is emphasized in the cartogram shown in Figure A14-2c. The Americas contrib- uted 14% (13 (9â18) million infections) of apparent infections worldwide, of which over half occurred in Brazil and Mexico. Our results indicate that Africaâs dengue burden is nearly equivalent to that of the Americas (16 (11â22) million infections, or 16% of the global total), representing a significantly larger burden than previously estimated. This disparity supports the notion of a largely hid- den African dengue burden, being masked by symptomatically similar illnesses, under-reporting and highly variable treatment-seeking behaviour (Endy et al., TABLE A14-1â Estimated Burden of Dengue in 2010, by Continent Apparent Inapparent Millions (credible interval) Millions (credible interval) Africa 15.7 (10.5â22.5) 48.4 (34.3â65.2) Asia 66.8 (47.0â94.4) 204.4 (151.8â273.0) Americas 13.3 (9.5â18.5) 40.5 (30.5â53.3) Oceania 0.18 (0.11â0.28) 0.55 (0.35â0.82) Global 96 (67.1â135.6) 293.9 (217.0â392.3)
APPENDIX A 303 2011; Gubler, 1998; Kakkar, 2012). The countries of Oceania contributed less than 0.2% of global apparent infections. We estimate that an additional 294 (217â392) million inapparent infections occurred worldwide in 2010. These mild ambulatory or asymptomatic infections are not detected by the public health surveillance system and have no immediate implications for clinical management. However, the presence of this huge poten- tial reservoir of infection has profound implications for: (1) correctly enumerating economic impact (for example, how many vaccinations are needed to avert an apparent infection) and triangulating with independent assessments of disability adjusted life years (DALYs) (Murray et al., 2012); (2) elucidating the population dynamics of dengue viruses (Cummings et al., 2009); and (3) making hypotheses about population effects of future vaccine programmes (Johansson et al., 2011) (volume, targeting efficacy, impacts in combination with vector control), which will need to be administered to maximize cross-protection and minimize post- vaccination susceptibility. The absolute uncertainties in the national burden estimates are inevitably a function of population size, with the greatest uncertainties in India, Indonesia, Brazil and China (see full rankings in Supplementary Table 4). In addition, com- paring the ratio of the mean to the width of the confidence interval (Hay et al., 2010) revealed the greatest contributors to relative uncertainty (see full rankings in Supplementary Table 4). These were countries with sparse occurrence points and low evidence consensus on dengue presence, such as Afghanistan or Rwanda (see Figure A14-2a), or those with ubiquitous high risk, such as Singapore or Djibouti, for which our burden prediction confidence interval is at its widest (see Supplementary Information, section D, Figure 2). Therefore, increasing evidence consensus and occurrence data availability in low consensus countries and assembling new cohort studies, particularly in areas of high transmission, will reduce uncertainty in future burden estimates. Our approach, uniquely, pro- vides new evidence to help maximize the value and cost-effectiveness of surveil- lance efforts, by indicating where limited resources can be targeted to have their maximum possible impact in improving our knowledge of the global burden and distribution of dengue. Our estimates of total infection burden (apparent and inapparent) are more than three times higher than the WHO predicted figure (Supplementary Informa- tion, section E). Our definition of an apparent infection is broad, encompassing any disruption to the daily routine of the infected individual, and consequently is an inclusive measurement of the total population affected adversely by the disease. Within this broad class, the severity of symptoms will affect treatment- seeking behaviours and the probability of a correct diagnosis in response to a given infection. Our definition is therefore more comprehensive than those of traditional surveillance systems which, even in the most efficient system, re- port a much narrower range of dengue infections. By reviewing our database of longitudinal cohort studies, in which total infections in the community were
304 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS documented exhaustively, we find that the biggest source of disparity between actual and reported infection numbers is the low proportion of individuals with apparent infections seeking care from formal health facilities (see Supplementary Information, section E, Fig. 5 for full analysis). Additional biases are introduced by misdiagnosis and the systematic failure of health management information systems to capture and report presenting dengue cases. By extracting the aver- age magnitude of each of these sequential disparities from published cohort and clinical studies, we can recreate a hypothetical reporting chain with idealized reporting and arrive at estimates that are broadly comparable to those countries reported to the WHO. This is most clear in more reliable reporting regions such as the Americas. Systemic under-reporting and low hospitalization rates have important implications, for example, in the evaluation of vaccine efficacy based on reduced hospitalized caseloads. Inferences about these biases may be made from the comparison of estimated versus reported infection burdens in 2010, highlighting areas where particularly poor reporting might be strengthened (see Supplementary Information, section E). We have strived to be exhaustive in the assembly of contemporary data on dengue occurrence and clinical incidence and have applied new modelling ap- proaches to maximize the predictive power of these data. It remains the case, however, that the empirical evidence base for global dengue risk is more limited than that available, for example, for Plasmodium falciparum (Gething et al., 2011) and Plasmodium vivax (Gething et al., 2012) malaria. Records of disease occurrence carry less information than those of prevalence and, as databases of the latter become more widespread, future approaches should focus on assessing relationships between seroprevalence and clinical incidence as a means of assess- ing risk (Anders and Hay, 2012). Additional cartographic refinements are also required to help differentiate endemic from epidemic-prone areas, to determine the geographic diversity of dengue virus types and to predict the distributions of future risk under scenarios of socioeconomic and environmental change. The global burden of dengue is formidable and represents a growing chal- lenge to public health officials and policymakers. Success in tackling this grow- ing global threat is, in part, contingent on strengthening the evidence base on which control planning decisions and their impact are evaluated. It is hoped that this evaluation of contemporary dengue risk distribution and burden will help to advance that goal. Methods Assembly of the Occurrence Database and Its Quality Control Occurrence data comprised of point or polygon locations of confirmed den- gue infection presence derived from both peer-reviewed literature and Health- Map alerts (Brownstein et al., 2008; Freifeld et al., 2008) (see Supplementary
APPENDIX A 305 Information, section A). An occurrence was defined as one or more laboratory or clinically confirmed infection(s) of dengue occurring at a unique location (a 5 km à 5 km pixel) within one calendar year. All occurrence data underwent manual review and automatic quality control to ensure information fidelity and precise geo-positioning. In total, 9,648 and 1,622 occurrence locations were obtained from literature searches and HealthMap, respectively. After the quality control procedures, our final data set contained 8,309 occurrence locations (5,216 point locations and 3,093 small polygon centroids) spanning a period from 1960 to 2012. We assume any record of dengue occurrence, regardless of its age, repre- sented an environment permissible for the disease, as dengue has expanded from a focal disease in Asia to a cosmopolitan disease of the tropics. Explanatory Covariates We assembled gridded global data for a suite of eight explanatory covari- ates. The covariates were chosen based on factors known or hypothesized to contribute to suitability for dengue transmission (see Supplementary Information, section B). These covariates included: (1) annual maximum and minimum pre- cipitation variables from a Fourier processed (Scharlemann et al., 2008) synoptic annual series interpolated from global meteorological stations (Hijmans et al., 2005); (2) a biological model combining the effects of temperature on the extrin- sic incubation period of dengue virus and lifespan of the Aedes aegypti vector to quantify the dengue-specific temperature suitability for transmission (Focks et al., 1993a,b; Gething et al., 2012); (3) Fourier-processed annual average normalized difference vegetation index (Hay et al., 2009); (4) categorical demarcations of urban and peri-urban areas (Hay et al., 2009); (5) an urban accessibility metric defining the travel time to nearest city of 50,000 people or more by land- or water- based travel (Nelson, 2008); and (6) an indicator of relative poverty derived from the finest geographic scale data available for economic productivity and adjusted for purchasing power parity (Nordhaus, 2006). No covariate grids were shown to be adversely affected by multicollinearity (see Supplementary Information, sec- tion B) and were standardized to ensure identical spatial resolution, extent and boundaries. For point records, covariate values corresponded to the pixel value containing the location of the point. For polygon occurrence records, covariate values were averaged across the whole polygon. Predicting the Probability of Occurrence (Risk) of Dengue Transmission We used a boosted regression tree (BRT) approach to establish a multivariate empirical relationship between the probability of occurrence of a dengue virus infection and the environmental conditions sampled at each site from the covari- ate suite. The BRT method has been shown to fit complicated response functions efficiently, while guarding against overfitting, and is therefore widely used for
306 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS vector and disease distribution mapping (Elith et al., 2006; Stevens and Pfeiffer, 2011). The BRT approach combines regression trees (Breiman, 1984) with gra- dient boosting (Friedman, 2001), whereby an initial regression tree is fitted and iteratively improved upon in a forward stage-wise manner (boosting) by minimiz- ing the variation in the response not explained by the model at each iteration (see Supplementary Information, section C). Like other niche mapping approaches, the BRT models require not only presence data but also absence data defining areas of disease absence and poten- tially unsuitable environmental conditions at unsampled locations. Because data on absence of disease are not definitive, pseudo-absence data estimate areas of disease absence instead. No consensus approach has been developed to optimize the generation of pseudo-absence data and we therefore created an evidence- based probabilistic framework for generating pseudo-absences, incorporating the main biasing factors in pseudo-absence generation, namely: (1) geographical extent; (2) number; (3) contamination bias; and (4) sampling bias. To represent areas of absence, na pseudo-absence points (Chefaoui and Lobo, 2008; Lobo and Tognelli, 2011; Stokland et al., 2011) were randomly generated based on dengue presence or absence certainty measures at a national or subnational level (Brady et al., 2012). Pseudo-absence locations were restricted to a maximum distance μ from any recorded presence site (Barbet-Massin et al., 2012; VanDerWal et al., 2009). Additionally, to compensate for âcontaminationâ of true but unobserved presences within the generated pseudo-absences (Ward et al., 2009), np pseudo- presence points were generated using the same procedure used to generate the pseudo-absences. Variation in the parameter set Ï = {μ, na, np} resulted in inde- pendent samples of the possible states of the real distribution, with all parameter combinations representing a null distribution of possible states. Therefore, rather than using an individual parameter combination from Ï, we created an ensemble (Araújo and New, 2007) of 336 BRT models spanning reasonable ranges in Ï and evaluated the central tendency as the mean across all 336 BRT models (see Supplementary Information, section C). The final ensemble BRT model was used to predict a global map of the probability of occurrence of dengue virus infection at a 5 km à 5 km resolution. Estimation of Dengue Burden and Populations at Risk Formal literature searches were conducted for serological dengue virus in- cidence surveys. Inclusion criteria were restricted to longitudinal surveys of seroconversion to dengue-virus-specific antibodies carried out in parallel with active symptom surveillance in a defined cohort. The surveys were abstracted, standardized and geopositioned (see Supplementary Information, section D). In total, 54 dengue incidence surveys were collected. Of these, 39 contained infor- mation about the ratio of inapparent to apparent infections.
APPENDIX A 307 The empirical relationship between incidence and the probability of occur- rence was represented using a Bayesian hierarchical model. We defined a negative binomial likelihood function (Hilbe, 2011) with constant dispersion and a rate characterized by a highly flexible data-driven Gaussian process prior (Banerjee et al., 2004). The Gaussian process prior was parameterized with a quadratic mean function and a squared exponential covariance function (Banerjee et al., 2004). Uninformative hyperpriors were assigned hierarchically to the prior pa- rameters and the full posterior distribution determined by Markov Chain Monte Carlo (MCMC) sampling (Patil et al., 2010). The entire model was fitted sepa- rately for apparent and inapparent infection incidences, with missing inapparent to apparent ratio values imputed in the MCMC. Using human population gridded data for the year 2010 (Balk et al., 2006), estimates of apparent and inapparent dengue infections were calculated nationally, regionally and globally. These estimates were then compared to national clinical cases reported to the WHO and differences between our cartographic estimates of infections and the WHO surveillance estimates were reconciled in a comparative analysis addressing key factors in traditional surveillance under-reporting (see Supplementary Informa- tion, section E). Acknowledgements S.I.H. is funded by a Senior Research Fellowship from the Wellcome Trust (095066), which also supports S.B. and P.W.G. C.P.S. is also funded by a Senior Research Fellowship from the Wellcome Trust (084368). O.J.B. is funded by a BBSRC Industrial CASE studentship. J.P.M., A.W.F., T.J., G.R.W.W., C.P.S., T.W.S. and S.I.H. received funding from, and with S.B., P.W.G., O.J.B. and J.J.F. acknowledge the contribution of, the International Research Consortium on Dengue Risk Assessment Management and Surveillance (IDAMS, 21803, http://www.idams.eu). This work was funded in part by EU grant 2011-261504 EDENEXT, and the paper is catalogued by the EDENEXT Steering Commit- tee as EDENEXT. S.I.H. and T.W.S. also acknowledge funding support from the RAPIDD program of the Science & Technology Directorate, Department of Homeland Security, and the Fogarty International Center, National Institutes of Health. Contributions S.I.H. and J.J.F. conceived the research. S.B. and S.I.H. drafted the man- uscript. S.B. drafted the Supplementary Information with significant support on sections A (O.J.B., C.L.M.), B (J.P.M., G.R.W.W.), C (P.W.G.), D (O.J.B., T.W.S.), and O.J.B. wrote section E. J.S.B. and A.G.H. provided HealthMap oc- currence data and advice on its provenance. O.J.B. reviewed all the occurrence data. S.B. did the modelling and analysis with advice from J.M.D., P.W.G., and
308 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS S.I.H. J.P.M. created all maps. All authors discussed the results and contributed to the revision of the final manuscript. Competing Financial Interests The authors declare no competing financial interests. References Anders, K. L. & Hay, S. I. Lessons from malaria control to help meet the rising challenge of dengue. Lancet Infect. Dis. 12, 977â984 (2012). Araújo, M. B. & New, M. Ensemble forecasting of species distributions. Trends Ecol. Evol. 22, 42â47 (2007). Balk, D. L. et al. Determining global population distribution: methods, applications and data. Adv. Parasitol. 62, 119â156 (2006). Banerjee, S., Carlin, B. P. & Gelfand, A. E. Hierarchical modeling and analysis for spatial data. Monographs on Statistics and Applied Probability 101 (Chapman & Hall/CRC, 2004). Barbet-Massin, M., Jiguet, F., Albert, C. H. & Thuiller, W. Selecting pseudo-absences for species distribution models: how, where and how many? Methods Ecol. Evol. 3, 327â338 (2012). Beatty, M. E., Letson, G. W. & Margolis, H. S. Estimating the global burden of dengue. Am. J. Trop. Med. Hyg. 81 (Suppl. 1), 231 (2009). Brady, O. J. et al. Refining the global spatial limits of dengue virus transmission by evidence-based consensus. PLoS Negl. Trop. Dis. 6, e1760 (2012). Breiman, L. Classification and Regression Trees (Chapman & Hall/CRC, 1984). Brownstein, J. S., Freifeld, C. C., Reis, B. Y. & Mandl, K. D. Surveillance sans frontières: internet- based emerging infectious disease intelligence and the HealthMap project. PLoS Med. 5, e151 (2008). Chakravarti, A., Arora, R. & Luxemburger, C. Fifty years of dengue in India. Trans. R. Soc. Trop. Med. Hyg. 106, 273â282 (2012). Chefaoui, R. M. & Lobo, J. M. Assessing the effects of pseudo-absences on predictive distribution model performance. Ecol. Modell. 210, 478â486 (2008). Cummings, D. A. et al. The impact of the demographic transition on dengue in Thailand: insights from a statistical analysis and mathematical modeling. PLoS Med. 6, e1000139 (2009). Elith, J. et al. Novel methods improve prediction of speciesâ distributions from occurrence data. Ecography 29, 129â151 (2006). Endy, T. P. et al. Determinants of inapparent and symptomatic dengue infection in a prospective study of primary school children in Kamphaeng Phet, Thailand. PLoS Negl. Trop. Dis. 5, e975 (2011). Focks, D. A., Haile, D. G., Daniels, E. & Mount, G. A. Dynamic life table model for Aedes aegypti (Diptera: Culcidae): analysis of the literature and model development. J. Med. Entomol. 30, 1003â1017 (1993a). Focks, D. A., Haile, D. G., Daniels, E. & Mount, G. A. Dynamic life table model for Aedes aegypti (Diptera: Culicidae): simulation and validation. J. Med. Entomol. 30, 1018â1028 (1993b). Freifeld, C. C., Mandl, K. D., Reis, B. Y. & Brownstein, J. S. HealthMap: global infectious disease monitoring through automated classification and visualization of Internet media reports. J. Am. Med. Inform. Assoc. 15, 150â157 (2008). Friedman, J. H. Greedy function approximation: a gradient boosting machine. Ann. Stat. 29, 1189â 1232 (2001). Gething, P. W. et al. A long neglected world malaria map: Plasmodium vivax endemicity in 2010. PLoS Negl. Trop. Dis. 6, e1814 (2012).
APPENDIX A 309 Gething, P. W. et al. A new world malaria map: Plasmodium falciparum endemicity in 2010. Malar. J. 10, 378 (2011a). Gething, P. W. et al. Modelling the global constraints of temperature on transmission of Plasmodium falciparum and P. vivax. Parasites Vectors 4, 92 (2011b). Gubler, D. J. Dengue and dengue hemorrhagic fever. Clin. Microbiol. Rev. 11, 480â496 (1998). Hales, S., de Wet, N., Maindonald, J. & Woodward, A. Potential effect of population and climate changes on global distribution of dengue fever: an empirical model. Lancet 360, 830â834 (2002). Halstead, S. B. Dengue vaccine development: a 75% solution? Lancet 380, 1535â1536 (2012). Halstead, S. B. Pathogenesis of dengue: challenges to molecular biology. Science 239, 476â481 (1988). Hay, S. I. et al. A world malaria map: Plasmodium falciparum endemicity in 2007. PLoS Med. 6, e48 (2009). Hay, S. I. et al. Estimating the global clinical burden of Plasmodium falciparum malaria in 2007. PLoS Med. 7, e1000290 (2010). Hay, S. I., Tatem, A. J., Graham, A. J., Goetz, S. J. & Rogers, D. J. Global environmental data for mapping infectious disease distribution. Adv. Parasitol. 62, 37â77 (2006). Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G. & Jarvis, A. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965â1978 (2005). Hilbe, J. M. Negative Binomial Regression 2nd edn, 251 (Cambridge Univ. Press, 2011). Johansson, M. A., Hombach, J. & Cummings, D. A. Models of the impact of dengue vaccines: a review of current research and potential approaches. Vaccine 29, 5860â5868 (2011). Kakkar, M. Dengue fever is massively under-reported in India, hampering our response. Br. Med. J. 345, e8574 (2012). Lobo, J. M. & Tognelli, M. F. Exploring the effects of quantity and location of pseudo-absences and sampling biases on the performance of distribution models with limited point occurrence data. J. Nat. Conserv. 19, 1â7 (2011). Monath, T. P. Yellow fever and dengue-the interactions of virus, vector and host in the re-emergence of epidemic disease. Semin. Virol. 5, 133â145 (1994). Murray, C. J. L. et al. Disability-adjusted life years (DALYs) for 291 diseases and injuries in 21 re- gions, 1990â2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380, 2197â2223 (2012). Nelson, A. Estimated travel time to the nearest city of 50,000 or more people in year 2000. http:// bioval.jrc.ec.europa.eu/products/gam (accessed 1 January 2012) (Global Environment Monitor- ing Unit â Joint Research Centre of the European Commission, 2008). Nordhaus, W. D. Geography and macroeconomics: new data and new findings. Proc. Natl Acad. Sci. USA 103, 3510â3517 (2006). Patil, A., Huard, D. & Fonnesbeck, C. J. PyMC: Bayesian stochastic modelling in Python. J. Stat. Softw. 35, e1000301 (2010). Rigau-Pérez, J. G. et al. Dengue and dengue haemorrhagic fever. Lancet 352, 971â977 (1998). Rodhain, F. La situation de la dengue dans le monde. Bull. Soc. Pathol. Exot. 89, 87â90 (1996). Rogers, D. J., Wilson, A. J., Hay, S. I. & Graham, A. J. The global distribution of yellow fever and dengue. Adv. Parasitol. 62, 181â220 (2006). Sabchareon, A. et al. Protective efficacy of the recombinant, live-attenuated, CYD tetravalent dengue vaccine in Thai schoolchildren: a randomised, controlled phase 2b trial. Lancet 380, 1559â1567 (2012). Scharlemann, J. P. W. et al. Global data for ecology and epidemiology: a novel algorithm for temporal Fourier processing MODIS data. PLoS ONE 3, e1408 (2008). Simmons, C. P., Farrar, J. J., van Vinh Chau, N. & Wills, B. Dengue. N. Engl. J. Med. 366, 1423â1432 (2012). Stevens, K. B. & Pfeiffer, D. U. Spatial modelling of disease using data- and knowledge-driven ap- proaches. Spat. Spatiotemporal Epidemiol. 2, 125â133 (2011).
310 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Stokland, J. N., Halvorsen, R. & Stoa, B. Species distribution modelling. Effect of design and sample size of pseudo-absence observations. Ecol. Modell. 222, 1800â1809 (2011). Tatem, A. J., Hay, S. I. & Rogers, D. J. Global traffic and disease vector dispersal. Proc. Natl Acad. Sci. USA 103, 6242â6247 (2006). TDR/World Health Organization. Report of the Scientific Working Group on Dengue, 2006. TDR/ SWG/08 (TDR/World Health Organization, 2006). Van Kleef, E., Bambrick, H. & Hales, S. The geographic distribution of dengue fever and the potential influence of global climate change. TropIKA. net http://journal.tropika.net/scielo. php?script5sci_arttext&pid5S2078-86062010005000001&lng5en&nrm5iso (2009). VanDerWal, J., Shoo, L. P., Graham, C. & William, S. E. Selecting pseudo-absence data for presence- only distribution modeling: how far should you stray from what you know? Ecol. Modell. 220, 589â594 (2009). Ward, G., Hastie, T., Barry, S., Elith, J. & Leathwick, J. R. Presence-only data and the EM algorithm. Biometrics 65, 554â563 (2009). World Health Organization. Dengue: Guidelines for Diagnosis, Treatment, Prevention and Control. WHO/HTM/NTD/DEN/2009.1 (World Health Organization, 2009). World Health Organization. International Travel and Health: Situation as on 1 January 2012 (World Health Organization, 2012). A15 CIRCUMPOLAR POPULATIONS, CLIMATE AND ENVIRONMENTAL CHANGE, AND THE IMPACT ON INFECTIOUS DISEASE PATTERNS Alan J. Parkinson75 The Arctic Environment and Populations The Arctic is home to 4 million people of whom almost half reside in the northern part of the Russian Federation. People in the Arctic live in social and physical environments that differ from their more southern dwelling counterparts. Approximately 400,000 (10 percent) of persons are of indigenous ancestry, half of whom live in the northern part of the Russian Federation (Parkinson, 2009; Stefansson Arctic Institute, 2004). The indigenous populations of northern Canada, Alaska, Greenland, and the northern Russian Federation generally reside in remote isolated communities consisting of 150 to several thousand inhabitants. In some regions, the only ac- cess to communities is by small aircraft or boat in summer and by small aircraft and snow machine in winter. Arctic communities, once isolated, are now very much a part of the global village we all live in and are as vulnerable to health 75ââDeputyDirector, Arctic Investigations Program, Division of Preparedness and Emerging Infec- tious Diseases. National Center for Emerging and Zoonotic Infectious Diseases, Centers for Disease Control and Prevention, Anchorage, Alaska.
APPENDIX A 311 threats as any other community on the globe. Through their unique relationship with nature, many of these peoples are more vulnerable to health threats gener- ated by climate change. These communities often have little economic infrastructure and are still largely dependent on subsistence harvesting of wildlife resources from terrestrial, fresh water, and marine ecosystems for a significant proportion of their diet. Food security is often dependent on subsistence wildlife migration patterns, predict- able weather, and some method of food storage. In these remote regions, access to public health and acute care systems is often marginal and poorly supported. The health of indigenous peoples of the circumpolar region has improved over the last 50 years or so. Much of this improvement can be attributed not only to the implementation of prevention and treatment activities that have resulted in reductions in morbidity and mortality from infectious diseases, such as tubercu- losis, and the vaccine-preventable diseases of childhood, but also to the provision of safe water supplies and sewage disposal in many communities. However, life expectancy of the indigenous populations of Alaska, northern Canada, Greenland, and the northern Russian Federation is lower than that of the respective national populations. Infant mortality remains higher than respective populations of the United States, northern Canada, Greenland, and northern Rus- sian Federation (Young and Bjerregaard, 2008). Mortality rates for heart disease and cancer were once lower among the indigenous populations of the United States, Canada, and northern European countries, but are now similar to their national rates. According to the U.S. CDC the cancer incidence rate is 1.5 times higher for Alaska Natives than the general U.S. population. However, it is evident that the indigenous populations of Alaska, Canada, and Greenland have much higher rates of unintentional injury and suicide. Uninten- tional injuries have always been a fact of life for those who live close to the land. These have often been related to hunting (animal attacks, shootings, and boating accidents) and hypothermia. With modernization, motor vehicle accidents and house fires have recently assumed more importance. Other health concerns of indigenous populations of the north include the high prevalence of certain infectious diseases such as hepatitis B, tuberculosis, Helicobacter pylori, respiratory syncytial virus in infants, influenza, and sexu- ally transmitted infections, as well as the potential health impacts associated with exposure to environmental pollutants through the traditional food supply, the health impacts of rapid economic change and modernization, and now the health impacts of climate change. Climate and Environmental Change The Arctic has warmed substantially more than the rest of the world over the last century, principally in recent decades. Figure A15-1 shows the global
312 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A15-1â Global temperature anomalies for 2000â2009 compared to 1951â1980. Global temperatures were on average about 0.6°C higher than they were 1951â1980. The Arctic, however, was about 2°C warmer. SOURCE: NASA image by Robert Simmon based on GISS surface temperature analysis data, including ship and buoy data from the Hadley Centre. Caption by Adam Voiland. The Climate of 2012 report: http://www.ncdc.noaa.gov/bams-state-of-the-climate/2012. php The State of the Climate in 2012 is a supplement to the August 2013 issue of the Bulletin of the American Meteorological Society (BAMS Vol. 94, No. 8). temperature anomalies for 2000â2009 compared to 1951â1980. Global tempera- tures were on average about 0.6°C higher than they were 1951â1980. The Arctic, however, was about 2°C warmer. Arctic climate models project continued warm- ing with a 3â7°C mean increase by 2100 (Symon et al., 2005). The greatest warming will occur in the winter months. The mean annual precipitation will also increase, and continued melting of land and sea ice is ex- pected to increase river discharge and contribute to a 1 m sea level rise by 2100, which will greatly impact many low-lying coastal communities not only in the Arctic but worldwide. Climate change is already impacting indigenous communities of the Arctic. Indigenous people have reported that the weather is less predictable than in recent memory (Furberg et al., 2011; Huntington and Weller, 2005). There is already a considerable impact on the sea and coastline in many regions of the Arctic. The average extent of sea ice in summer has declined by 15â20 percent over the last 30 years. The sea ice is getting thinner. There has been a significant loss of multiyear ice over the last decade (www.nsida.org). Ice is important for hunting and fishing. It provides easy access to open water, a platform to hunt fromânot only for indigenous people but also for the prey they hunt. Thin ice means fewer
APPENDIX A 313 seals, walrus, and whales to hunt, impacting diet nutrition and cultural well-being (Brubaker et al., 2011). Thin ice also may increase unintentional injury and death by drowning to those traveling the surface to hunting grounds (Fleischer et al., 2013). Reduction in sea ice will have widespread effects on the marine ecosystem, coastal climate, human settlements, and subsistence activities. The reduction in sea ice has, for the first time, created ice-free shipping lanes to the northwest from Labrador to the Bering sea, and to the northeast from the Bering sea to Norway, representing fuel saving short cuts for transportation by sea, and access to oil gas and mineral reserves once inaccessible (Arctic Council, 2009). This traffic is projected to increase rapidly. Benefits to isolated communities will include construction of military bases and the development of other industrial and commercial ventures such as tourism, which will result in infrastructure support and employment. Public and private services will increase to support emerging economies. However, these ventures will affect population distribution, dynam- ics, culture, and local environments and will challenge the traditional subsistence way of life for many communities and lead to accelerated and long-term cultural change, which will create additional stress on an already vulnerable population. Delayed freeze-up, lack of the sea ice barrier, and increased storm intensity have accelerated coastal erosion and damage to water and sanitation systems, forcing some communities in northwestern Alaska to evacuate during storm events and to consider eventual permanent relocation (Brubaker et al., 2011). Such storm events also place residents at higher risk for unintentional injury and for chronic stress. Fear for safety and security will have long-term effects on men- tal and behavioral health in these communities. The movement of rural residents to urban centers is occurring in some regions of the circumpolar north (Driscoll et al., 2010; Stefansson Arctic Institute, 2004). This is currently being driven by economic, educational, and health care opportunities. However, this trend may accelerate due to the impact of climate change. For many communities, the river is a byway that connects a community to inland-based subsistence resources such as caribou, salmon trout, waterfowl, and wild berries. With warming has come widespread thawing of shallow permafrost resulting in the collapse of tundra into the rivers. This increased erosion changes river flow, increases turbidity, and restricts access to upriver hunting grounds. Rivers are becoming wider, shallower, warmer, and dirtierâaffecting navigation, critical fish habitat, and water quality and quantity when used as a community water source (Brubaker et al., 2011; EvengÃ¥rd et al., 2011). Lakes are also changing. Some are expanding with water from thawing per- mafrost; others are shrinking as an underlying ice lens thaws and drains the lake (Smith et al., 2005). Warmer water allows new vegetation to grow; algae, aquatic plants, and mosses flourish, creating problems if the lake is used as a community water source. As warming temperatures move northward, associated plants and wildlife will follow. Biologic responses to a warming Arctic are expected to outpace those
314 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS at lower latitudes. Spring will occur earlier and the growing season will be longer. The tree line is projected to reach the Arctic Ocean in most of Asia and western North American by the end of this century. This is likely to lead to a near loss of tundra vegetation in these areas with important consequences for many types of wildlife (Weller, 2005). Thus, climate change is likely to have a significant impact on key terrestrial species used as subsistence food by shifting range and abundance of key species such as caribou, moose, water fowl, and sea birds. The health impacts of a decline in the proportion of traditional food consumed by the indigenous population may be significant. A shift away from a traditional diet to a more western diet, higher in carbohydrates and sugars, has been associated with increased levels of cardiovascular disease, diabetes, vitamin-deficiency disorders, dental cavities, anemia, obesity, and lower resistance to infection. Climate and Environmental Change and Infectious Diseases It is well known that climate and weather affect the distribution and risk of many vector-borne diseases such as malaria, Rift Valley fever, plague, and den- gue fever in tropical regions of the globe. Weather also affects the distribution of food- and water-borne diseases and emerging infectious diseases such as West Nile virus, Hantavirus, and Ebola hemorrhagic fever (Haines et al., 2006). Less is known about the impact of climate change and the risk and distribution of infec- tious diseases in Arctic regions. It is known that Arctic populations have a long history of both endemic and epidemic infectious diseases (IOM, 2008; Parkinson and Butler, 2005). However, with the introduction of antimicrobial drugs, vac- cines, and public health systems, morbidity and mortality due to infectious dis- eases have been greatly reduced. The impact of climate on the incidence of these existing infectious disease challenges is unknown. In many Arctic regions, however, inadequate housing and sanitation are already important determinants of infectious disease transmission. The cold northern climate keeps people indoors, amplifying the effects of household crowding, smoking, and inadequate ventila- tion. Crowded living conditions increase person-to-person spread of infectious diseases and favor the transmission of respiratory and gastrointestinal diseases and skin infections. Impact on the Water and Sanitation In many communities in the north, the built infrastructure is supported on permafrost. Loss of this support will result in damage to water intake systems and pipes and may result in contamination of community water supplies and damage to water and sanitation infrastructures and distribution systems, forcing commu- nities to rely more on untreated (or traditional) water sources (Brubaker et al., 2011; EvengÃ¥rd et al., 2011). This may result in an increase in clinic visits and hospitalizations for various âwater washedâ infectious diseases, those commonly
APPENDIX A 315 prevented by hand washing such as gastroenteritis, respiratory infections caused by respiratory syncytial virus (RSV), influenza, skin infections, impetigo, and boils caused by MRSA (Brubaker et al., 2011; Hennessy et al., 2008; Wenger et al., 2010). A study in western Alaska demonstrated two to four times higher hospitalization rates among children less than 3 years of age for pneumonia, in- fluenza, and childhood RSV infections in villages where the majority of homes had no in-house piped water, compared with villages where the majority of homes had in-house piped water service. Likewise, outpatient multiple-resistant Staphy- lococcus aureus infections and hospitalizations for skin infections among persons of all ages were higher in villages with no in-house piped water service compared to villages with water service (Hennessy et al., 2008; Wenger et al., 2010). Lack of water for hygiene may also contribute to the transmission of zoonotic pathogens such as Giardia, Cryptosporidium, and Echinococcus from the envi- ronment to people. The climate-related northern expansion of the boreal forest in Alaska and northern Canada has favored the steady northward advance of the bea- ver, potentially extending the range of Giardia, a parasitic infection of beaver that can infect other mammals, including humans, who consume untreated surface water. Giardia is currently well established in northern climates where cooler, wetter conditions favor survival and transmission of cysts. This parasite is the most significant enteric protozoan in the entire North American Arctic (Jenkins et al., 2013). In Alaska, rates of giardiasis have been consistently two times higher than is found overall in the United States (Yoder et al., 2012a). It is possible that warming temperatures will decrease environmental survival of Giardia cysts. However, this will likely be offset by increased transmission through changes in animal reservoir dynamics, regional hydrology, and flooding events caused by heavy rain, snowfall, and melting, leading to outbreaks of waterborne infections. Elevated runoff from snow melt and increased precipitation could also exac- erbate contamination of water supplies with Cryptosporidium cysts and oocysts (Davidson et al., 2011). The association between infection and increased precipi- tation is well recognized. In a recent outbreak in two towns in northern Sweden, more than 50,000 residents developed Cryptosporidium-related gastroenteritis after drinking contaminated municipal water following heavy rainfall that over- whelmed water purification systems (Evengård et al., 2011). However, compared to giardiasis, cryptosporidiosis in Arctic populations appears relatively uncom- mon. In Alaska, for example, the mean annual incidence (1.0 and 0.8/100,000) is almost three times lower than the U.S. average (Yoder et al., 2012b). The lack of livestock, and apparent low level of infection in both marine and terrestrial wildlife in these regions, suggests that zoonotic transmission of cryptosporidiosis may be uncommon in Arctic regions. Alveola echinococcosis caused by Echinococcus multilocularis was common in two regions of northwestern Alaska prior to 1986 (State of Alaska, 2003). Only one case has been reported in northern Canada (Jenkins et al., 2013). Echinococ- cus multilocularus maintains a cycle in foxes and voles. Disease in humans was
316 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS mainly associated with contact with sled dogs. However, improvements in hous- ing, water and sanitation, sled dog lot management, and the transition from dog sled teams to snow machines have largely eliminated dog-to-human transmission in Alaska. Climate change will influence the transmission of E. multilocularis through the effects on the distribution and abundance of rodent intermediate hosts and the sylvatic definitive hostâthe Arctic fox. Increased precipitation might lead to increased stability and density of rodent populations facilitating transmission of E. multilocularis. However, increased frequency of severe weather events may decrease overall transmission. These events are occurring in parallel with other drivers of disease emergence, landscape change, and translocation of hosts. E. multilocularis was recently detected in Svalbard in the Norwegian Arctic follow- ing introduction of a suitable intermediate host likely from shipping (Henttonen et al., 2001). Such events will increase in frequency with the opening of ice-free shipping lanes across the Arctic bringing increasing cargo, tourist traffic, and other flora and fauna to regions once inaccessible to invasion from the sea, requir- ing a continued vigilance for this disease in these regions (Jenkins et al., 2011). The Northern-strain cystic hydatid disease is caused by Echinococcus granu- losus, which maintains a cycle that includes an adult cestode stage in the de- finitive host such as wolf, coyote, fox, or dog, and a larval cestode stage in an intermediate host in cervids such as moose, deer, caribou, and reindeer. Humans usually acquire the infection via exposure to eggs that are shed in canid feces and are an accidental host. In Alaska, the first human case was recorded in 1941. Human hydatid disease is reportable in Alaska; peak numbers of cases were de- tected from 1953 to 1973, nearly all of which were in Alaska Natives (State of Alaska, 2003). Similarly in Canada, 99 percent of 141 cases in the 1950s occurred in indigenous peoples. Today, cases of cystic hydatid disease occur infrequently in North America, in part, as a result of improving housing, water and sanitation infrastructure, and the gradual phasing out of sled dogs as a method of transporta- tion in these regions (Jenkins et al., 2013). Populations of cervid hosts are already being affected by climate change. Caribou populations in the Arctic are declin- ing due to secondary effects of climate and landscape changes (Kerby and Post, 2013). Increased precipitation and extreme weather events will likely contribute to further declines in population and parasite transmission in Arctic regions. Impact on the Food Supply Some infectious diseases are unique to the Arctic and lifestyles of the indig- enous populations and may increase in a warming Arctic. For example, many Arc- tic residents depend on subsistence hunting, fishing, and gathering for foodâand on a predictable climate for food storage. Traditional food storage methods often include aboveground air-drying of fish and meat at ambient temperature, below- ground cold storage on or near the permafrost, and fermentation. Changes in cli- mate may prevent the drying of fish or meat, resulting in spoilage. Similarly, loss
APPENDIX A 317 of the permafrost may result in spoilage of food stored below ground (Brubaker et al., 2011). Climate change could exacerbate the potential for the food and/or waterborne transmission of toxoplasmosis in the Arctic (Jenkins et al., 2013). Toxoplasmosis is caused by infection with Toxoplasma gondii, a widespread protozoan parasite of mammals and birds. Members of the cat family are the only known definitive host for the sexual infectious stages (oocysts) of T. gondii; however, the asexual encysted stage is found in muscle tissues of animals and can serve as the main reservoirs of infection in cat-free areas. Humans become infected by ingesting raw or insufficiently cooked meat and foods that have come into contact with in- fected meat; by, indirectly or directly, ingesting cysts from soil, such as items that have come into contact with cat feces (unwashed vegetables); or transplacentally in humans from a mother to her fetus. A serosurvey conducted among the Inuit of Nunavik showed a seropreva- lence of 60 percent (Messier et al., 2009). Because of the absence of a felid host in Nunavik, it is unclear how Toxoplasma infection would be maintained in this region of the Arctic. It would appear to require a nonfelid definitive host, pos- sibly rodents or various migratory species (e.g., barren ground caribou, birds, marine mammals) with terrestrial runoff feeding into a marine cycle. It is also possible that it could be maintained by carnivores and/or vertical transmission (marine mammals, herbivores). The recent discovery of Toxoplasma in polar bears and Arctic foxes in Svalbard underscores the widespread nature of this infection (Elmore et al., 2012). It has been hypothesized that this infection was introduced to this region by migratory birds. The prevalence in polar bears in Svalbard, the Barents Sea Region, and Eastern Greenland areas has doubled in the last decade (now 46 percent), and detection in ring seals for the first time highlights the predatorâprey cycles in this region, and increasing risk to popula- tions that rely on marine mammals for food. In the Arctic, the consumption of undercooked meat from marine mammals seems a much more important risk factor for human infection than drinking water. A recent serosurvey among Inuit in northern Quebec showed that 80 percent of Inuit with a dietary preference for dried meat from sea mammals were seropositive compared to 10 percent among ethnic Cree in the same community, who preferred cooked terrestrial mammals (Messier et al., 2009). Another important meat-borne parasite in the Arctic is Trichinella, com- monly responsible for outbreaks related to the consumption of undercooked bear or walrus meat (Davidson et al., 2011). The most common species is T. nativa, which, unlike other Trichinella species (such as Trichinella T6), survives freez- ing. In addition, smoking, drying, fermenting, or salting are not reliable methods for killing the parasite, thus placing the consumer at risk of infection. The geo- graphical distribution of cold-tolerant versus freeze-tolerant Trichinella sp. follow the January isothermal lines (-5°C for T. nativa). Thus, shifts in host diversity and environmental temperature could lead to altered distribution. There is no current
318 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS evidence that climate change has contributed to an increase in Trichinella preva- lence in Alaska or northern Canada. Loss of sea ice could interfere with resting, feeding, and breeding of marine mammals. The decline in sea ice has already resulted in large haulouts of walruses on beaches in northwestern Alaska and may contribute to increased transmission of Trichinella to other land or marine scavengers and carnivores. In the Arctic, human infections caused by Brucella suis biovar IV have been linked to the consumption or processing of raw caribou meat, and the infection has been shown to be endemic in many caribou and reindeer herds across the Arctic (Hueffer et al., 2013). This is in contrast to infection in other parts of the United States where the most common route of transmission is through the consumption of raw dairy products or meat. However, there is little data on the prevalence of brucellosis in humans or wildlife not only in Alaska, but also in other Arctic regions. In Alaska, there have been 17 human cases reported since 1973. Reporting in wildlife is complicated by the absence of a standardized diag- nostic test for different animal species. Other Brucella spp. can also infect other land and marine mammals. While human infections caused by a marine Brucella found in seals (B. pinnipedialis) has been documented, evidence for the direct transmission from seals to humans has yet to be established, as does the risk of infection to those who depend on marine mammals as a subsistence food source. Past outbreaks of anthrax among cattle and reindeer have resulted in more than 13,000 burial grounds in Russia containing the carcasses of infected animals. More than half of these are located on permafrost in Siberia. There is concern that with a warming of the Arctic, melting permafrost in these regions will ex- pose many of these burial sites together with anthrax spores that will result in an epizootic among grazing animals and increase the risk of infection in humans who come into contact with infected animal products (undercooked meat, hides, bone) (Revich et al., 2012). Impact on Vector-Borne Diseases Vector-borne diseases are those transmitted to humans, or between humans, via an arthropod vector. In Sweden, the incidence of tick-borne encephalitis (TBE) has substantially increased since the mid-1980s (Lindgren and Gustafson, 2001). This increase corresponds to a trend of milder winters and an earlier onset of spring, resulting in an increase in the tick population (Ixodes ricinus) that car- ries the virus responsible for TBE and other potential pathogens (Skarphédinsson et al., 2005). Similar movement of TBE has been documented in northern north- western Russia where Ixodes persulcatus is the predominant vector. This move- ment corresponds to the estimated climate-induced changes in the I. persulcatus habitat (Revich et al., 2012; Tokarevich et al., 2011). In northeastern Canada, climate change is projected to result in a northward shift in the range of Ixo- des scapularis, a tick that carries Borrelia burgdorferi, the etiologic agent of
APPENDIX A 319 Lyme disease. The current northern limit of Ix. scapularis is southern Ontario, including the shoreline of Lake Erie and the southern coast of Nova Scotia. Some temperature-based models show the potential for a northward expansion of Ix. scapularis above 60°N latitude and into the Northwest Territories by 2080 (Ogden et al., 2006). Alaska, once thought to be tick free (Zarnke et al., 1990), is now reporting the presence of the moose winter tick (Dermacentor albipictus), which transmits anaplasmosis in moose and elk herds in southwestern Alaska, as well as dog ticks (Dermacentor variablis and Rhipicephalus sanguineus)âboth vectors of Rocky Mountain Spotted Fever in south central and interior Alaska (Beckman, 2013). While it may be predicted that the warming of Alaska may be contributing to this recent invasion, tick distribution is influenced by additional factors such as habitat suitability and dispersal patterns, which can affect the accuracy of these predictions. The contribution of climate change-induced altera- tions in vector range to human disease, thus, depends on many other factors such as land use practices, human behavior (suburban development in wooded areas, outdoor recreational activities, transport of pets, use of insect repellents, etc.), and human population density, as well as adequacy of the public health infrastructure. Linkage Between Climate and Infectious Diseases For most diseases that we consider to be climate sensitive, we have very little data on the relationship between weather climate and infectious disease emergence in the Arctic. Much can be learned by promptly investigating out- breaks that may be climate related. For example, outbreaks of gastroenteritis caused by Vibrio parahaemolyticus have been related to the consumption of raw or inadequately cooked shellfish collected from seawater at temperatures higher than 15°C (Figure A15-2). Prior to 2004, the most northerly outbreak occurred in northern British Columbia in 1997. However, in July 2004, an outbreak of gastroenteritis caused by V. parahaemolyticus was documented among cruise ship passengers consuming raw oysters while visiting an oyster farm in Prince William Sound, Alaska (McLaughlin et al., 2005). The outbreak investigation documented an increase of 0.21°C per year in the July-August water temperature since 1997, and reported that 2004 was the first year that the oyster farm water temperature exceeded 15°C in July. This event provides direct evidence of an association be- tween rising seawater temperature and the onset of illness. Warmer temperatures may allow infected host animal species to survive winters in larger numbers, increase in population, and expand their range of habitation, thus increasing the opportunity to pass infections to humans. For example, milder weather and less snow cover may have contributed to a large outbreak of Puumala virus infection in northern Sweden in 2007 (Figure A15-3). Puumala virus is endemic in bank voles, and causes hemorrhagic fever with renal syndrome in humans (Pettersson et al., 2008). Similar outbreaks have been noted in Finland and in the Russian Federation (Makary et al., 2010; Revich, 2008).
320 FIGURE A15-2â Climate-related outbreak of Vibrio parahaemolyticus gastroenteritis, Alaska 2004. Graph shows the mean daily water tempera- ture at an oyster farm in Prince William Sound Alaska, together case patients by date of consumed farmed oysters. The sea water temperature had increased by 0.21°C per year since 1997 (r² = 0.14, P < 0.001) reaching an optimal temperature for bacterial growth, above 15°C, in oysters in June 2004. SOURCE: McLaughlin et al., 2005.
APPENDIX A 321 FIGURE A15-3â Climate-related outbreak of Puumala virus infection in Sweden 2007. (A) Monthly incidence of nephropathia epidemica (NE) in Sweden and Västerbotten County, Sweden, January 1997âSeptember 2007. Also shown are climate conditions, De- cember 1998â2006, in the NE outbreak area of Västerbotten County, Sweden. (B) Number of days with a snow cover. (C) Average temperature. SOURCE: Pettersson et al., 2008.
322 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Recommended Actions It is apparent that for many of these diseases the risk to human and animal populations is not known, largely because surveillance systems are inadequate, lacking the sensitivity and specificity to be able to determine, with any accuracy, the prevalence of disease in these populations. It is not clear whether this is be- cause of underreporting or underdiagnosis due to the lack of diagnostics, staff issues in remote locations, or logistical difficulties associated with remote speci- men collection handling and shipping. Baseline levels of infection in humans and animals are also unknown. More research is needed on establishing current preva- lence of infection, determining the disease ecology of these and other emerging pathogens, and the potential impact of climate on disease occurrence in both human and animal populations in Arctic regions. To begin to assess the potential emergence and health impact of climate-sensitive infectious diseases in northern human and animal populations, the following actions need to be undertaken: 1. Enhance the surveillance capacity to monitor potentially climate-sensitive infectious diseases that are likely to have the most impact on human and animal populations. An initial step would include conducting surveys to assess reportable climate-sensitive infectious diseases by Arctic country or northern region. This catalog can then be used to conduct surveillance evaluations for those climate-sensitive infectious diseases with the great- est potential for impacting public or animal health. All facets of each surveillance system should be examined to determine the number of cases identified, case definitions used, data collection, analysis, and reporting and distribution systems used, including feedback to those providing the data. 2. Determine baseline levels of infection by conducting seroprevalence sur- veys in both human and animal populations. Conduct a survey of avail- able human and animal specimen banks in the circumpolar north. Results could be used to target communities or regions for specific prospective serosurveys and risk factor analysis, and could lead to the implementa- tion of prevention and control outreach, education, and communication activities. 3. Conduct research into the relationship between weather, climate, and infectious disease emergence to guide early detection and intervention by promptly investigating outbreaks that may be climate related. 4. Adopt a âOne Healthâ strategy. Throughout the world the close link between ecosystem health and health of food species and humans has been recognized and is the foundation of the One Health concept. This concept is nowhere more apparent than in the Arctic, making the circum- polar north the place where One Health can be the organizing concept to understand the disease ecology and the potential impact of climate on disease occurrence in both human and animal populations in Arctic
APPENDIX A 323 regions (Dudley et al., 2013). The key to capitalizing on the One Health approach is to use and expand interdisciplinary networks to establish and integrate disease surveillance using human, animal, and environmental data to detect emergence of climate-sensitive infectious diseases in human and animal populations. 5. Networks must be expanded. One network that focuses on human health in Arctic populations is the International Union for Circumpolar Health (www.iuch.net), which includes working groups on infectious diseases, and climate change and infectious diseases (www.arcticinfdis.com). A network that could facilitate greater intersectorial cooperation efforts between human, animal, and environmental professionals in the Arctic is the Arctic Council. The Arctic Council (www.arctic-council.org) is a ministerial intergovernmental forum promoting cooperation, coordination, and interaction between the eight Arctic States (the United States, Canada, Denmark/Greenland, Iceland, Norway, Sweden, Finland, and the Russian Federation), including Arctic indigenous populations, on common Arctic concerns such as sustainable development and environmental protection in the Arctic and more recently on climate change (Arctic Council, 2005). The scientific work of the Arctic Council is carried out in six work- ing groups, which include the Arctic Contaminants Action Program, the Monitoring and Assessment Program, Conservation of Arctic Flora and Fauna, Protection of the Marine Environment, Emergency Prevention Pre- paredness and Response, and Sustainable Development Working Group. Using contacts within the International Union for Circumpolar Health and Arctic Council working group structures allowed the formation of the International Circumpolar Surveillance of Emerging Infectious Diseases (ICS) in 1999 (Figure A15-4). ICS links public health laboratories, insti- tutes, and academic centers across the circumpolar north for the purpose of monitoring and sharing information on infectious diseases of concern, collaborating on research, and prevention and control activities (IOM, 2008; Zulz et al., 2009). 6. Develop communication strategies for data sharing with communities, circumpolar countries, and other organizations and agencies with wildlife, human, and environmental health responsibilities. In the Arctic, systems for monitoring and communicating changes in environment, wildlife, and human health are very limited. Therefore, establishing communication networks, locally, regionally, and internationally, is critical. For example, the Alaska Native Tribal Health Consortium has developed the Local En- vironmental Observer Network (LEO), a system for sharing information on environmental impacts and community health effects (Brubaker et al., 2013). LEO uses trained community members to document time and loca- tion of specific events and encourages communication between communi- ties, academic centers, and resource agencies to increase understanding
324 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A15-4â International Circumpolar Surveillance of Emerging Infectious Diseases. Established in 1998 ICS links public health laboratories, institutes, and academic centers across the circumpolar north for the purpose of monitoring and sharing information on infectious diseases of concern, collaborating on research and prevention and control ac- tivities. Participating regions (shown in dark grey), reference laboratories (large red dots), and laboratories (small red dots). SOURCE: Parkinson, 2009; Zulz et al., 2009. about climate and other drivers of climate change and to develop adap- tation strategies. Such networks can be linked to circumpolar networks such as the Atlas of Community-Based Monitoring (www.arcticcbm.org), which will allow communities and networks to link and expand informa- tion sharing and establish collaboration on a circumpolar scale. Conclusion Climate change is already affecting indigenous communities of the Arctic. It is well known that in more temperate parts of the world, climate can affect the distribution of many food-, vector- and waterborne infectious diseases. Less is known about the impact of climate change and the risk and distribution of infec- tious diseases in Arctic regions. Melting permafrost can destabilize water and sanitation infrastructure, resulting in damage to water intake systems and pipes, forcing communities to rely on untreated water sources. This may result in an increase in clinic visits and hospitalizations for various âwater-washedâ infectious
APPENDIX A 325 diseases, those commonly prevented by hand washing such as respiratory infec- tions, skin infections, echinococcosis, and gastroenteritis caused by Giardia or Cryptosporidium. Many Arctic residents depend on subsistence hunting, fishing, and gathering for food; consequently, changes in climate may increase the po- tential for the food-borne transmission of toxoplasmosis, trichinosis, and brucel- losis in the Arctic. Milder winters and an earlier onset of spring may result in an increase and northward shift in tick populations increasing the incidence of tick-borne diseases in Arctic regions. It is apparent that for many of these diseases, the risk to human and animal populations is not known, and for most diseases that we consider to be potentially climate sensitive we have very little data on the relationship between weather, climate, and infectious disease emergence. More needs to be done to determine baseline levels of infection and to enhance surveillance capacities in Arctic coun- tries for those infectious diseases that are likely to be potentially climate sensitive, and could have the most impact on human and animal populations. More research needs to be done on establishing the relationship between weather, climate, and infectious disease emergence to guide early detection and intervention. The Arctic provides a unique opportunity to use a One Health approach as an organizing concept to understand the disease ecology and the potential impact of climate on disease occurrence in both human and animal populations in Arctic regions and to expand local, regional, and international net- works to increase interdisciplinary collaboration and understanding about climate change and infectious disease emergence, prevention, and control. References Arctic Council. 2005. Arctic climate impact assessment scientific report. Cambridge, UK: Cambridge University Press. http://www.acia.uaf.edu (accessed January 30, 2014). Arctic Council. 2009. Arctic marine shipping assessment report. http://www.pame.is/amsa-2009- report (accessed January 30, 2014). Beckman, K. B. 2013. Dog ticks introduced and established in Alaska: Increased risks for tick-borne zoonoses. Poster presentation November 7, 2013, Alaska Department of Fish and Game Divi- sion of Wild-Life Conservation, Fairbanks, Alaska. Brubaker, M., J. Berner, R. Chavan, and J. Warren. 2011. Climate change and health effects in North- west Alaska. Global Health Action 4. Brubaker, M., J. Berner, and M. Tcheripanoff. 2013. LEO, the Local Environmental Observer Net- work: A community-based system for surveillance of climate, environment, and health events. International Journal of Circumpolar Health 72:513-514. Davidson, R., M. Simard, S. J. Kutz, C. M. Kapel, I. S. Hamnes, and L. J. Robertson. 2011. Arctic parasitology: Why should we care? Trends in Parasitology 27(6):239-245. Driscoll, D., B. Dotterrer, J. Miller, and H. Voorhees. 2010. Assessing the influence of health on rural outmigration in Alaska. International Journal of Circumpolar Health 69(5). Dudley, J. P., E. P. Hoberg, E. J. Jenkins, and A. J. Parkinson. 2013. Climate change in the North American Arctic: A one health perspective. Ecohealth. Submitted. Elmore, S. A., E. J. Jenkins, K. P. Huyvaert, L. Polley, J. J. Root, and C. G. Moore. 2012. Toxoplasma gondii in circumpolar people and wildlife. Vector Borne and Zoonotic Diseases 12(1):1-9.
326 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS EvengÃ¥rd, B., J. Berner, M. Brubaker, G. Mulvad, and B. Revich. 2011. Climate change and water security with a focus on the Arctic. Global Health Action 4. Fleischer, N., P. Melstrom, E. Yard, M. Brubaker, and T. Thomas. 2013. The epidemiology of falling- through-the-ice in Alaska, 1990â2010. Journal of Public Health doi: 10.1093/pubmed/fdt081. Furberg, M., B. EvengÃ¥rd, and M. Nilsson. 2011. Facing the limit of resilience: Perceptions of climate change among reindeer herding Sami in Sweden. Global Health Action 4. Haines, A., R. S. Kovats, D. Campbell-Lendrum, and C. Corvalan. 2006. Climate change and human health: Impacts, vulnerability, and mitigation. Lancet 367(9528):2101-2109. Hennessy, T. W., T. Ritter, R. C. Holman, D. L. Bruden, K. L. Yorita, L. Bulkow, J. E. Cheek, R. J. Singleton, and J. Smith. 2008. The relationship between in-home water service and the risk of respiratory tract, skin, and gastrointestinal tract infections among rural Alaska natives. American Journal of Public Health 98(11):2072-2078. Henttonen, H., E. Fuglei, C. N. Gower, V. Haukisalmi, R. A. Ims, J. Niemimaa, and N. G. Yoccoz. 2001. Echinococcus multilocularis on Svalbard: Introduction of an intermediate host has en- abled the local life-cycle. Parasitology 123(Pt 6):547-552. Hueffer, K., A. J. Parkinson, R. Gerlach, and J. Berner. 2013. Zoonotic infections in Alaska: Disease prevalence, potential impact of climate change and recommended actions for earlier disease detection, research, prevention and control. International Journal of Circumpolar Health 72. Huntington, H., and G. Weller. 2005. Arctic Climate Impact Assessment. Cambridge, UK: Cambridge University Press. http://www.acia.uaf.edu (accessed January 30, 2014). IOM (Institute of Medicine). 2008. Global climate change and extreme weather events: Understand- ing the contributions to infectious disease emergence. Washington, DC: The National Academies Press. Jenkins, E. J., J. M. Schurer, and K. M. Gesy. 2011. Old problems on a new playing field: Helminth zoonoses transmitted among dogs, wildlife, and people in a changing northern climate. Veteri- nary Parasitology 182(1):54-69. Jenkins, E. J., L. J. Castrodale, S. J. de Rosemond, B. R. Dixon, S. A. Elmore, K. M. Gesy, E. P. Hoberg, L. Polley, J. M. Schurer, M. Simard, and R. C. Thompson. 2013. Tradition and transi- tion: Parasitic zoonoses of people and animals in Alaska, northern Canada, and Greenland. Advances in Parasitology 82:33-204. Kerby, J. T., and E. Post. 2013. Advancing plant phenology and reduced herbivore production in a terrestrial system associated with sea ice decline. Nature Communications 4. Lindgren, E., and R. Gustafson. 2001. Tick-borne encephalitis in Sweden and climate change. Lancet 358(9275):16-18. Makary, P., M. Kanerva, J. Ollgren, M. J. Virtanen, O. Vapalahti, and O. Lyytikainen. 2010. Disease burden of Puumala virus infections, 1995-2008. Epidemiology and Infection 138(10):1484-1492. McLaughlin, J. B., A. DePaola, C. A. Bopp, K. A. Martinek, N. P. Napolilli, C. G. Allison, S. L. Murray, E. C. Thompson, M. M. Bird, and J. P. Middaugh. 2005. Outbreak of Vibrio parahae- molyticus gastroenteritis associated with Alaskan oysters. New England Journal of Medicine 353(14):1463-1470. Messier, V., B. Levesque, J. F. Proulx, L. Rochette, M. D. Libman, B. J. Ward, B. Serhir, M. Couillard, N. H. Ogden, E. Dewailly, B. Hubert, S. Dery, C. Barthe, D. Murphy, and B. Dixon. 2009. Seroprevalence of Toxoplasma gondii among Nunavik Inuit (Canada). Zoonoses and Public Health 56(4):188-197. Ogden, N. H., A. Maarouf, I. K. Barker, M. Bigras-Poulin, L. R. Lindsay, M. G. Morshed, J. OâCallaghan C, F. Ramay, D. Waltner-Toews, and D. F. Charron. 2006. Climate change and the potential for range expansion of the Lyme disease vector Ixodes scapularis in Canada. Interna- tional Journal for Parasitology 36(1):63-70. Parkinson, A. J. 2009. The International Polar Year, 2007â2008, an opportunity to focus on infectious diseases in Arctic regions. Emerging Infectious Diseases 14(1):1. Parkinson, A. J., and J. C. Butler. 2005. Potential impacts of climate change on infectious diseases in the Arctic. International Journal of Circumpolar Health 64(5).
APPENDIX A 327 Pettersson, L., J. Boman, P. Juto, M. Evander, and C. Ahlm. 2008. Outbreak of Puumala virus infec- tion, Sweden. Emerging Infectious Diseases 14(5):808-810. Revich, B. 2008. Climate change alters human health in Russia. Studies on Russian Economic De- velopment 19(3):311-317. Revich, B., N. Tokarevich, and A. J. Parkinson. 2012. Climate change and zoonotic infections in the Russian Arctic. International Journal of Circumpolar Health 71. Skarphédinsson, S., P. M. Jensen, and K. Kristiansen. 2005. Survey of tickborne infections in Den- mark. Emerging Infectious Diseases 11(7):1055-1061. Smith, L., Y. Sheng, G. MacDonald, and L. Hinzman. 2005. Disappearing Arctic lakes. Science 308(5727):1429-1429. State of Alaska Epidemiology Bulletin. 2003. http://www.epi.alaska.gov/bulletins/docs/b2003_02. pdf (accessed May 20, 2014). Stefansson Arctic Institute. 2004. Arctic human development report. Stefansson Arctic Institute in Akureyri, Iceland. http://www.svs.is/AHDR/AHDR%20chapters/English%20version/AHDR_ chp%201.pdf (accessed May 2012). Symon, C., L. Arris, and B. Heal. 2005. Arctic climate impact assessment. Cambridge, UK: Cam- bridge University Press. Tokarevich, N., A. A. Tronin, O. V. Blinova, R. V. Buzinov, V. P. Boltenkov, E. D. Yurasova, and J. Nurse. 2011. The impact of climate change on the expansion of Ixodes persulcatus habitat and the incidence of tickborne encephalitis in the north of European Russia. Global Health Action. Weller, G. 2005. Arctic climate impact assessment. Cambridge, UK: Cambridge University Press. Pp. 991. http://www.acia.uaf.edu (accessed January 30, 2014). Wenger, J. D., T. Zulz, D. Bruden, R. Singleton, M. G. Bruce, L. Bulkow, D. Parks, K. Rudolph, D. Hurlburt, T. Ritter, J. Klejka, and T. Hennessy. 2010. Invasive pneumococcal disease in Alaskan children: Impact of the seven-valent pneumococcal conjugate vaccine and the role of water sup- ply. Pediatric Infectious Disease Journal 29(3):251-256. Yoder, J. S., J. W. Gargano, R. M. Wallace, and M. J. Beach. 2012a. Giardiasis surveillanceâUnited States, 2009â2010. MMWR: Surveillance Summaries 61(5):13-23. Yoder, J. S., R. M. Wallace, S. A. Collier, M. J. Beach, and M. C. Hlavsa. 2012b. Cryptosporidiosis surveillanceâUnited States, 2009â2010. MMWR: Surveillance Summaries 5:1-12. Young, T. K., and P. Bjerregaard. 2008. Health transitions in Arctic populations: Toronto, Canada: University of Toronto Press. Zarnke, R. L., W. Samuel, A. W. Franzmann, and R. Barrett. 1990. Factors influencing the potential establishment of the winter tick (Dermacentor albipictus) in Alaska. Journal of Wildlife Dis- eases 26(3):412-415. Zulz, T., M. Bruce, and A. Parkinson. 2009. International circumpolar surveillance: Prevention and control of infectious diseases: 1999â2008. Circumpolar Health Supplement 4:20-23.
328 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS A16 CLIMATE CHANGE AND HUMAN HEALTH: A ONE HEALTH APPROACH76 Jonathan A. Patz77 and Micah B. Hahn77 Abstract The electronic version of Climate change and human health: A One Health approach is unavailable due to copyright restrictions. 76ââReprinted with kind permission from Springer Science+Business Media. Springer-Verlag Berlin Heidelberg. Current Topics in Microbiology and Immunology. Climate change and human health: A One Health approach, 366(2013):141-171, J. A. Patz and M. B. Hahn. 77ââNelson Institute, Center for Sustainability and the Global Environment (SAGE), University of Wisconsin, 1710 University Avenue, Madison, WI 53726, USA.
APPENDIX A 329
330 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 331
332 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 333
334
335
336 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 337
338 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 339
340
341
342 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 343
344 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 345
346 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 347
348 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 349
350 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 351
352 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 353
354 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 355
356 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 357
358 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS
APPENDIX A 359 A17 IMPACTS OF CLIMATE CHANGE ON PLANT DISEASES: NEW SCENARIOS FOR THE FUTURE Marco Pautasso78 and Michael J. Jeger79 Abstract In this overview, we selectively discuss recent literature on the develop- ment of new scenarios of the impacts of climate change on plant health. The literature on human health is much larger than the one on plant health, but plants are essential to human health and well-being. Impacts of climate change on plant health have been mostly studied for North America and Europe, although research has also taken place in South America, Asia, and Australasia. New scenarios will need to take into account not just the expected temperature and precipitation trends, but also the increased likeli- hood of introduction of exotic plant pathogens due to long-distance plant trade networks. The example of Phytophthora ramorum shows that historical factors and the distribution of the hosts are important and need to be consid- ered in new scenarios too. At the same time, there is scope for integration in such scenarios of human responses to climate change (e.g., large-scale plan- tations of biofuel crops), effects of extreme weather events, the uncertainty in precipitation shifts, and climate change modifications of plant host defence systems. More precise and useable prediction will come from scenarios in- cluding the diversity of strategies available to prevent and manage the emer- gence of exotic plant pathogens under changing environmental conditions. 78ââForest Pathology and Dendrology, Institute of Integrative Biology, ETHZ, Zurich, Switzerland. 79ââDivision of Ecology and Evolution & Centre for Environmental Policy, Imperial College Lon- don, UK. ââKeywords: global change, migration pathways, network epidemiology, Phytophthora ramorum, plant health
360 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Introduction Plants are essential for human health and well-being, not only because they provide ecosystem services such as photosynthesis, food production, evapo- transpiration, and carbon storage, but also due to their cultural, psychological, and aesthetic benefits (Pearson-Mims and Lohr, 2000; Bringslimark et al., 2009; Eyles et al., 2010; Russell et al., 2013). Plants are vital to human beings, but plant health is an understudied issue. About 90,000 items are retrieved with Google Scholar using the keyword plant health, whereas the same search engine retrieves about 1,640,000 publications when searching for human health, or about 18 times more material (as of October 2013, no time limits). The difference is even more pronounced when searching the same keywords in Web of Science (~1,000 (plant health) vs. ~28,000 (human health) findings, or ~28 times more material in human medicine). Interestingly, when searching for plant disease and for human disease, there are still more findings for human beings compared to plants, but the ratio is only about four, both for Google Scholar (~356,000 vs. ~1,510,000 findings, respectively) and for Web of Science (~3,700 vs. 16,000 findings, respectively). Whereas there appear to be more publications using the keyword human health than human disease, the opposite is the case for plant health and plant disease (Figure A17-1). This is likely to be due to the traditional focus of plant patholo- gists on specific plant diseases. The recent development of the One Health concept calls for more recogni- tion of the interconnections between the health of human beings, animals, and ecosystems and for common approaches among researchers and practitioners in the three domains (Pautasso et al., 2012b). A study of whether epidemic models can be used to represent the diffusion of topics among disciplines suggests a potential for interdisciplinary collaboration between researchers in human and plant epidemiology (Kiss et al. 2010), but there is still little overlap between the two areas, for example, in terms of curricula (Gómez et al., 2013). Nonetheless, there is increasing recognition that integrating ecological considerations into human epidemiology and public health is likely to bring many benefits (Preston et al., 2013), just as this has been the case for plant disease and pest management (Kogan, 1998). Patz et al. (this volume) provide an excellent overview of the importance of the environment for human health. Three recent examples specific to plant health are: · The report of increased human mortality following widespread ash tree mortality in the Midwest of the United States due to the introduction from Asia of the Emerald Ash Borer, which led to more severe heat waves (Donovan et al., 2013); · A study showing the benefits for child health and nutrition of retaining patches of forest in Malawi (Johnson et al., 2013); and
APPENDIX A 361 · A call for multidisciplinary research to tackle the increasing challenge posed by human pathogens harboured by plants (Fletcher et al., 2013). Given the number of plant species on the planet (estimated at about 420,000; Crane, 2004), plant health is a rather more complex issue than human health, where there is only a single host species to be kept healthy, except for zoonotic diseases. Plants range from annual weeds (that germinate and develop to maturity and death in a few months) to long-lived trees, able to cope with diseases for decades, at the same time sustaining an intricate ecosystem in their canopy and roots. The diversity of plant life forms is matched by a variety of plant pathogen strategies, which translates into a corresponding diversity of plant pathosystems (Ingram, 2002). Moreover, there are fundamental differences between human and FIGURE A17-1â Number of publications retrieved in (a) Web of Science and (b) Google Scholar using the keyword human disease relative to the number of publications retrieved using the keyword human health, and for plant disease in comparison with plant health (1991â2012, as of October 2013).
362 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS plant health because of emotive and ethical considerations. The death of every child is a tragedy, but we do not hesitate to harvest vegetables, long before these plants have completed their life cycle. While there has been a long-standing philosophical debate about the definition of human health, plant health is also understudied from a conceptual perspective (Döring et al., 2012). Plants are traditionally seen as sedentary organisms, living their whole life at the location where they germinated. This makes it unlikely that a tree in, say, Washington, D.C., might move temporarily to a tropical location and then come back with a tropical disease, as many people now sometimes do (Richaud et al., 2013). However, human beings have developed a trade network of plants for planting, seed, and other plant material that has grown rapidly over the last de- cades and is now global in scope (Dehnen-Schmutz et al., 2010). Some plants for planting might travel more in their life span than many humans do. This pathway has become the major route of invasion for forest pathogens and pests, both in North America (Liebhold et al., 2012) and in Europe (Santini et al., 2013). Most published research on global change in general (Pasgaard and Strange, 2013) and on the effects of climate change on plant health in particular (Andrew et al., 2013) is indeed from North America and Europe, although research on the topic has also taken place in South America, Australasia, and Asia (Chakraborty et al., 1998; Ganley et al., 2011; Ghini et al., 2011; Savary et al., 2011a). There is thus the need for more empirical studies and scenarios of climate change impacts on plant health from other continents. The diversity of plant pathosystems makes it difficult to achieve success- ful generalizations about the effects of climate change on plant health, because specific scenarios are needed (Savary et al., 2011b). Nonetheless, comparative epidemiology can deliver useful insights also on the likely effects of climate change (West et al., 2012). Moreover, many plant pathogens are nowadays dis- persed artificially due to long-distance trade in plants, so scenarios of the effects of climate change on such pathogens need to take into account that their intro- duction into areas newly suitable from a climatic point of view will be facilitated by plant trade. Without increasing global trade and the associated movements of pathogens, climate change would be less of a problem for health, and this is likely to apply to both human beings and plants (Engering et al., 2013). However, for plants, increased trade in seed of food crops would be necessary anyway fol- lowing climate shifts to support new crops being grown in regions previously unsuitable. As an example, take Phytophthora ramorum, the oomycete responsible for Sudden Oak Death in California (Rizzo et al., 2005) and for Sudden Larch Death in the British Isles (Brasier and Webber, 2010). This quarantine pathogen, which was unknown to science until 2001 (Werres et al., 2001), has been intercepted a number of times in shipments of ornamental plants among European countries (EFSA PLH, 2011). Border interceptions of plant pathogens are probably the tip of the iceberg, given that only a fraction of the massive amounts of traded plants
APPENDIX A 363 can be inspected (Brasier, 2008). In Britain, P. ramorum findings are concentrated in the southwestern part of the country. This could reflect the distribution of · The hostsâThe pathogen affects plants that are typically found in Corn- wall, from holm oak (Quercus ilex) to Rhododendron ponticum (an exotic shrub), but also more widespread species such as heath (Calluna vulgaris) and bilberry (Vaccinium myrtillus) (Purse et al., 2013); · The initial foci of introductionâWhether these were nurseries or historic gardens in Cornwall is likely, but remains to be demonstrated (Xu et al., 2009); · Climatic suitabilityâThe western part of Britain is generally milder and wetter (EFSA PLH, 2001; Chadfield and Pautasso, 2012); or · A combination of these three issues. This example confirms that climate plays an important role also in determin- ing the distribution of exotic plant pathogens (Ireland et al., 2013), but also that historical factors and the host distribution are important and need to be considered in new scenarios too (Croucher et al., 2013; Liebhold et al., 2013). The 2013 P. ramorum outbreaks in thousands of hectares of Japanese larch (Larix kaempferi) plantations in southwestern Scotland underline the importance of establishing the climatic suitability to exotic plant pathogens of entire countries. In this case, the Scottish region with the majority of outbreaks (now an infection âred zoneâ) was the only part of Scotland considered at high risk from P. ramorum. Assessing regional plant disease risk helps prioritizing monitoring also under changing environmental conditions. One interesting insight achieved by mod- ellers of P. ramorum in California is that predictive modelling is more reliable if based not just on data on where the pathogen has been detected, but also on where the pathogen has not been detected (absences) (VáclavÃk and Meentemeyer, 2009; Pautasso, 2013b). Scenarios of the future distribution of plant pathogens are thus likely to be fraught with considerable uncertainty, because there are few data available on the absences of plant pathogens under changed climatic conditions. The evidence base (and lack thereof, i.e., knowledge gaps) on plant health and climate change has been summarized many times before, so there is no lack of reviews of the literature on the topic (e.g., Coakley et al., 1999; Garrett et al., 2006, 2012; Jeger et al., 2011; Newton et al., 2011; Sturrock et al., 2011; Boonekamp, 2012; Dixon, 2012; Ghini et al., 2012; Chakraborty, 2013; Pangga et al., 2013; Juroszek and von Tiedemann, 2013). In this overview, we make one more attempt at reviewing such literature, but also take into account the related body of knowledge on exotic plant pathogens. Our main point is that prediction of the effects of climate change on plant health is likely to fail without considering at the same time the effects of globalization on the likelihood of introduction of plant pathogens (Figure A17-2). We also provide a summary of strategies with which to respond to emerging plant diseases. We argue that these strategies are
364 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A17-2â New scenarios of climate change impacts on plant health will need to take into account the likely introductions of exotic plant pathogens due to increased plant trade, as well as human responses to both climate change (e.g., large-scale cultivation of biofuels) and exotic plant pathogens. likely to gain even more importance in the presence of changing environmental and climatic conditions. New Scenarios to Predict Plant Health Under Climate Change New plant health scenarios under climate change need to integrate likely de- velopments of global trade taking place at the same time as the predicted changes in climate. Scenarios linking climate change and global trade are a subset of those predicting plant health under global change, because global change is not just driven by climate change and long-distance trade, but also by urbanization, pollution, and land use change (Gregory and Ingram, 2000). These drivers are fuelling each other. For example, the more land is converted from tropical forest to agriculture, the more there will be carbon emissions, and the likelier it will become that the most severe climate change scenarios will take place. Similarly, further urbanization spurs increased long-distance trade, which leads to increased pollution, including greenhouse gas emissions. Plant health is also the outcome of various interacting factors, including hostâpathogen interactions, host and pathogen migration, the action of vectors, and environmental and human influences. However, establishing how the various plant health components are influencing each other is less straightforward than with global change, not just because of the variety of pathosystems (Sutherst et al., 2011). For example, landscape connectivity can make it more likely that
APPENDIX A 365 exotic pathogens will disperse, but it will also tend to maintain host genetic di- versity, which can result in enhanced resilience of plant hosts in the face of a new disease. Variation in host genetic diversity also influences plant phenology, which can affect hostâpathogen interactions, such as by shifting the main period of host susceptibility away from (or closer to) the main period of pathogen virulence (Dodd et al., 2008; Grulke, 2011; Caffarra et al., 2012). Similar mismatches be- tween plants and their pollinators can affect plant and ecosystem health (Ogawa- Onishi and Berry, 2013). An additional complication is that each of these components, whose interac- tions result in the continuum between plant health and plant disease epidemics, is affected by each of the global change drivers (Anderson et al., 2004; Pautasso et al., 2012a). Researchers have started investigating the outcome of such inter- actions, but we are still far away from a comprehensive understanding of how various pathosystems will behave under changing environmental conditions. We know though that climate change effects deleterious to plant health are likely to result in enhanced climate change, because of the resulting increased carbon emissions due to such things as large-scale tree mortality (Kurz et al., 2008; Hicke et al., 2013), although in the long term these emissions may be compen- sated by forest regrowth. Not only plant health scenarios, but also climate change models, need to include this kind of feedback. To make realistic predictions about the feedbacks between climate change and plant health, scenarios also have to take into account that climate change will result not just in changed average conditions, but also in the increased fre- quency of extreme events (Garrett, 2008; Reichstein et al., 2013). Heat waves, wind storms, drought, and floods can all affect plant health (and food safety) both indirectly, by moving the environmental conditions outside of the optimal physiological window for many plant species, and directly, by affecting the likeli- hood and severity of plant disease epidemics (Marvin et al., 2013). Floods and heavy rains dramatically increase moisture and can thus increase the prevalence of leaf and soil pathogens, but they can also contribute in dispersing plant patho- gens in air currents, transported soil, and water droplets (Munkvold and Yang, 1995). Drought and heat waves, conversely, favour secondary parasites because they weaken plants, and make viral epidemics more likely as insect vectors are favoured by dry and warm conditions (Rosenzweig et al., 2001). Some extreme weather events are reported to be on the rise (e.g., extreme rain events in India between the 1950s and 2000; Goswami et al., 2006), but (due to the many confounding factors) it is difficult to establish whether plant disease epidemics due to extreme weather events are also becoming more frequent. What is apparent, despite the differing survey effort in various countries, is a shift in first reports of plant pathogens towards the poles, which would suggest an al- ready perceptible influence of climate warming on plant pathosystems worldwide (Bebber et al., 2013). Climate warming is only part of the observed and foreseen climate changes: precipitation and humidity are two other essential variables to
366 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS predict the development of future plant pathosystems (Thompson et al., 2013). Unfortunately, the uncertainty of the predicted precipitation regimes under cli- mate change is rather high for many regions, thus making prediction of outcomes for plant pathosystems also insecure (Shaw and Osborne, 2011). Further work is thus needed to model likely shifts in precipitation for regions such as China, Indonesia, the Rocky Mountains, and western tropical Africa. Uncertainty in future precipitation is an important knowledge gap because scenarios of plant diseases in the presence of increased temperature and precipita- tion or in the presence of increased temperature and reduced precipitation differ markedly (Kliejunas, 2011). For example, a foliar pathogen such as P. ramorum is not likely to become more virulent in the West Coast of the United States in case of increased temperature but decreased precipitation, whereas increased temperature and precipitation are likely to lead to more sporulation, increased inoculum, and thus more likely expansion of Sudden Oak Death into new re- gions. However, if sexual reproduction does take place in P. ramorum, as with some other exotic and homothallic Phytophthora species in California, then a wider range of environmental conditions and the potential for greater diversity in virulence may occur (EFSA PLH, 2011; Grünwald et al., 2012). On the contrary, an abiotic tree health issue such as the dieback of yellow cedar (Chamaecyparis nootkatensis) in Alaska (which appears to be due to yellow cedar vulnerability to fine-root freezing following reduced snow depth; Hennon et al., 2012) would be exacerbated by increased temperature and reduced precipitation, whereas the tree species might recover in case of increased temperature and precipitation (Hennon and Shaw, 1994; Kliejunas, 2011). Several reports of increased tree mortality due to more frequent drought have appeared, for example for quaking aspen (Populus tremuloides) in North America (Worrall et al., 2013). These reports are worrying, because of the associated car- bon emissions and the loss of ecosystem services we normally take for granted from healthy forests. Widespread tree mortality due to more frequent and severe drought is a problem which is likely to take place at the same time as the pre- dicted lack of drinking, agricultural, and industrial water for many regions of the world. Whether precipitation will decrease or not, climate warming is expected to reduce the presence and severity of frost events, and thus to extend the grow- ing season at high altitudes and latitudes (Roos et al., 2011). This phenomenon is likely to make it possible to introduce new crops in such regions, but it is also likely that the cultivation of those crops will be followed by the migration of their associated pathogens (Chakraborty, 2013). What is needed to develop successful scenarios of such shifts in crops and plant pathogens is also an improved under- standing of how plant host defence systems will behave in response to increased temperatures (Cheng et al., 2013). Shifts in the range and severity of plant diseases will also be a consequence of climate change mitigation measures such as large-scale biofuel cultivation. In- creased use of plant biomass could provide new opportunities for plant pathogens,
APPENDIX A 367 particularly in the case of large-scale deployment of monocultures (Paterson et al., 2013). Even without the threat posed by climate change, there is the need to develop more diverse timber and biomass plantations, so as to diminish the losses due to outbreaks of plant pathogens and pests. In many cases, emerging pathogens affecting crops used to produce biomass are exotic. This is a further reminder that scenarios to predict plant health under climate change need to take into account patterns in the long-distance trade of plants, because such trade networks amplify the likelihood of introducing new plant pathogens. Integrating Our Responses to Exotic Plant Pathogens in Scenarios to Predict Plant Health Under Climate Change Expanded cultivation of biofuels, whether this makes sense as a climate change mitigation measure or not, is likely to provide new opportunities for plant pathogens. This makes it clear that, to predict how plant health will develop under changing environmental conditions, we need to take into account human responses to climate change and new plant diseases. A diversity of strategies is available to respond to emerging plant diseases. In general, prevention is better than cure, but prevention is at best difficult for pathogens that are still unknown to science or are understudied. Interdisciplinary approaches between plant pathologists, economists, modellers, and social scientists to develop pre- ventive strategies are important and should be encouraged, as shown by the Rural Economy and Land Use (RELU) project on the growing risks to the U.K. rural economy posed by crop diseases (Mills et al., 2011; Pautasso et al., 2012b; Ilbery et al., 2013). One way to respond to new plant diseases is to model the spread of exotic plant pathogens in space and time. This was achieved for P. ramorum in Britain, taking into account the distribution of the main hosts (deciduous trees such as Quercus spp. and heathland) and a realistic reconstruction of the trade in orna- mental plants susceptible to the pathogen (Harwood et al., 2009). Such models can help predict the further development of plant disease epidemics under various inspection policies, thus improving the focus of monitoring surveys. There is the need to include changing climatic conditions in such modelling exercises. The importance of including trade network information in spatial simula- tion models is exemplified by the 2012 reports from Britain of ash dieback. This disease, caused by the exotic ascomycete Hymenoscyphus pseudoalbidus (Gross et al., 2013), is now present throughout much of the distributional range of the host, Fraxinus excelsior, and is likely to have arrived to the eastern part of Britain from the European continent through wind-blown spores, due to the reports in East Anglia and Kent of affected mature woodlands. However, the trade in ash saplings contributed to establishing the disease more widely, because the patho- gen was detected across the United Kiingdom at recently planted sites (Pautasso et al., 2013b).
368 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Despite the increasing plant health issues associated with movements of plants by the nursery industry, there is still little information about the structure of such plant trade networks. This is of concern, given that network structure, even in the case of small-size networks, has been shown to affect the likelihood of epidemics taking place, particularly in the presence of super-connected indi- viduals (Moslonka-Lefebvre et al., 2011); in the case considered here, nursery sites which act as major hubs in the nursery trade. The amounts of plants shipped within and among continents are massive (for example, millions of ash saplings were imported by the United Kingdom from other European countries between 2002 and 2011), so that only a small fraction of such consignments can be in- spected. It is thus sobering that most models of climate change impacts on plant health do not take into account plant trade networks. The sudden appearance of ash dieback in Britain was not just a wake-up call to improve plant biosecurity (Woodward and Boa, 2013), but also an example of how stakeholder involvement can make a difference in terms of research funds made available to plant health researchers. Following the first reports of the pathogen in the country, a huge public and media outcry made it possible for the U.K. government to find the resources to commit to funding research on ash dieback, despite the climate of austerity and the previous years of cuts to plant pathology departments and institutes. Scenarios to predict plant health in the face of climate change and the introduction of new plant pathogens will thus need to be aware of such potential rapid developments in research capacity. Nevertheless, long-term monitoring and research would be better served by a more secure funding environment. It would be beneficial if botanic gardens and arboreta, traditionally operating with a long-term perspective, could become a sentinel network warning about potential plant health risks before these material- ize (Britton et al., 2010). Because of limited resources, economic considerations are indeed at the heart of any disease management strategy. Although there is increased use of bio-economic models in plant disease management, with promising work on efficient control of plant disease epidemics spreading in networks (Ole Å et al., 2012, 2013), there is still the need to apply such approaches in the modelling of climate change impacts on plant pathosystems. This is an important knowledge gap because plant diseases already cause substantial losses to global food produc- tion (Savary et al., 2012). In the coming decades, climate change will not happen in a vacuum, but together with a further rise in human population, which will make intensification of agriculture necessary in many regions. New scenarios of climate change impacts on plant health need to take into account such shifts in cultivation intensity. Further intensification of agriculture is just one of the land use changes po- tentially affecting how climate change will influence plant pathosystems. Particu- larly important and underexplored in scenarios is the assisted migration of plant species to enable them to cope with the rapid climate warming. There has been
APPENDIX A 369 little consideration in the assisted migration debate among conservation biolo- gists that such plant translocations could result in the inadvertent long-distance movement of plant pathogens (Garbelotto and Pautasso, 2012). One practical way to respond to both climate change and to the likely effects of climate change on plant health would be to decrease emissions of greenhouse gases (Pautasso et al., 2010), but there is little evidence for the moment that such a decrease may hap- pen soon, so it is likely that scenarios of climate change impacts on plant health based on current emission trends may be more realistic than those expecting a drop in emissions. Long-term monitoring of (i) carbon emissions, (ii) climate change effects on plant health, and (iii) the literature on this important topic is essential to inform scenarios with the most updated information from experiments and surveys (Jeger and Pautasso, 2008). For example, there is evidence that fungi are increasingly mentioned in the climate change literature of the last two decades. A similar increasing trend in the proportion of papers mentioning fungi was reported also for the literature on disease, health, infection, and immunity (Pautasso, 2013a). It is possible that such a pattern in the literature reflects an increased importance of fungi as human pathogens. Much of the literature on climate change impacts on crop health has focused on a handful of widespread crops (e.g., grape, potato, oilseed rape, rice, soybean, and wheat; Luck et al., 2011; Bregaglio et al., 2013; Siebold and Tiedemann, 2013; van der Waals et al., 2013). Similarly, most of the available literature on exotic tree diseases is about a few pathogens (e.g., P. ramorum, P. cinnamomi, and P. cactorum) (Pautasso et al., 2013a). There is the need to increase our epi- demiological knowledge of less studied pathosystems that might become prob- lematic under changing environmental conditions, including increased CO2 levels (Kaczynski and Cooper, 2013). This knowledge includes how such pathogens will respond to changing climates and plant trade patterns, but also how we are likely to respond to such changes in pathogen behaviour (in agro- vs. wild ecosystems), and the associated uncertainties (Gouache et al., 2013). Acknowledgements Many thanks to C. Allen, J. Fletcher, O. Holdenrieder, D. Rizzo, and M.W. Shaw for insights and discussions, and to O. Holdenrieder for helpful feedback on a previous draft. References Anderson, P. K., A. A. Cunningham, N. G. Patel, F. J. Morales, P. R. Epstein, and P. Daszak. 2004. Emerging infectious diseases of plants: Pathogen pollution, climate change and agrotechnology drivers. Trends in Ecology & Evolution 19(10):535-544.
370 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Andrew, N. R., S. J. Hill, M. Binns, M. H. Bahar, E. V. Ridley, M. P. Jung, C. Fyfe, M. Yates, and M. Khusro. 2013. Assessing insect responses to climate change: What are we testing for? Where should we be heading? PeerJ 1:e11. Bebber, D. P., M. A. T. Ramotowski, and S. J. Gurr. 2013. Crop pests and pathogens move polewards in a warming world. Nature Climate Change 3(11):985-988. Boonekamp, P. 2012. Are plant diseases too much ignored in the climate change debate? European Journal of Plant Pathology 133(1):291-294. Brasier, C., and J. Webber. 2010. Plant pathology: Sudden larch death. Nature 466(7308):824-825. Brasier, C. M. 2008. The biosecurity threat to the UK and global environment from international trade in plants. Plant Pathology 57(5):792-808. Bregaglio, S., M. Donatelli, and R. Confalonieri. 2013. Fungal infections of rice, wheat, and grape in Europe in 2030â2050. Agronomy for Sustainable Development 33(4):767-776. Bringslimark, T., T. Hartig, and G. G. Patil. 2009. The psychological benefits of indoor plants: A criti- cal review of the experimental literature. Journal of Environmental Psychology 29(4):422-433. Britton, K. O., P. White, A. Kramer, and G. Hudler. 2010. A new approach to stopping the spread of invasive insects and pathogens: Early detection and rapid response via a global network of sentinel plantings. New Zealand Journal of Forestry Science 40:109-114. Caffarra, A., M. Rinaldi, E. Eccel, V. Rossi, and I. Pertot. 2012. Modelling the impact of climate change on the interaction between grapevine and its pests and pathogens: European grapevine moth and powdery mildew. Agriculture, Ecosystems & Environment 148(0):89-101. Chadfield, V., and M. Pautasso. 2012. Phytophthora ramorum in England and Wales: Which environ- mental variables predict county disease incidence? Forest Pathology 42(2):150-159. Chakraborty, S. 2013. Migrate or evolve: Options for plant pathogens under climate change. Global Change Biology 19(7):1985-2000. Chakraborty, S., G. M. Murray, P. A. Magarey, T. Yonow, R. G. OâBrien, B. J. Croft, M. J. Barbetti, K. Sivasithamparam, K. M. Old, M. J. Dudzinski, R. W. Sutherst, L. J. Penrose, C. Archer, and R. W. Emmett. 1998. Potential impact of climate change on plant diseases of economic signifi- cance to Australia. Australasian Plant Pathology 27(1):15-35. Cheng, C., X. Gao, B. Feng, J. Sheen, L. Shan, and P. He. 2013. Plant immune response to pathogens differs with changing temperatures. Nature Communications 4. Coakley, S. M., H. Scherm, and S. Chakraborty. 1999. Climate change and plant disease management. Annual Review of Phytopathology 37:399-426. Crane, P. R. 2004. Documenting plant diversity: Unfinished business. Philosophical Transactions: Biological Sciences 359(1444):735-737. Croucher, P. P., S. Mascheretti, and M. Garbelotto. 2013. Combining field epidemiological informa- tion and genetic data to comprehensively reconstruct the invasion history and the microevolution of the sudden oak death agent Phytophthora ramorum (Stramenopila: Oomycetes) in California. Biological Invasions 15(10):2281-2297. Dehnen-Schmutz, K., O. Holdenrieder, M. J. Jeger, and M. Pautasso. 2010. Structural change in the international horticultural industry: Some implications for plant health. Scientia Horticulturae 125(1):1-15. Dixon, G. R. 2012. Climate changeâImpact on crop growth and food production, and plant patho- gens. Canadian Journal of Plant Pathology 34(3):362-379. Dodd, R. S., D. Huberli, W. Mayer, T. Y. Harnik, Z. Afzal-Rafli, and M. Garbelotto. 2008. Evidence for the role of synchronicity between host phenology and pathogen activity in the distribution of sudden oak death canker disease. New Phytologist 179(2):505-514. Donovan, G. H., D. T. Butry, Y. L. Michael, J. P. Prestemon, A. M. Liebhold, D. Gatziolis, and M. Y. Mao. 2013. The relationship between trees and human health: Evidence from the spread of the emerald ash borer. American Journal of Preventive Medicine 44(2):139-145. Döring, T. F., M. Pautasso, M. R. Finckh, and M. S. Wolfe. 2012. Concepts of plant healthâReview- ing and challenging the foundations of plant protection. Plant Pathology 61(1):1-15.
APPENDIX A 371 EFSA PLH. 2011. Scientific opinion on the pest risk analysis on Phytophthora ramorum. EFSA Journal 9(6):2186. Engering, A., L. Hogerwerf, and J. Slingenbergh. 2013. Pathogen-host-environment interplay and disease emergence. Emerging Microbes & Infections 2:e5. Eyles, A., P. Bonello, R. Ganley, and C. Mohammed. 2010. Induced resistance to pests and pathogens in trees. New Phytologist 185(4):893-908. Fletcher, J., J. E. Leach, K. Eversole, and R. Tauxe. 2013. Human pathogens on plants: Designing a multidisciplinary strategy for research. Phytopathology 103(4):306-315. Ganley, R., M. Watt, D. Kriticos, A. M. Hopkins, and L. Manning. 2011. Increased risk of pitch canker to Australasia under climate change. Australasian Plant Pathology 40(3):228-237. Garbelotto, M., and M. Pautasso. 2012. Impacts of exotic forest pathogens on Mediterranean ecosys- tems: Four case studies. European Journal of Plant Pathology 133(1):101-116. Garrett, K. A. 2008. Climate change and plant disease risk. In Global climate change and extreme weather events: Understanding the contributions to infectious disease emergence, edited by D. A. Relman, M. A. Hamburg, E. R. Choffnes, and A. Mack. Washington, DC: The National Academies Press. Garrett, K. A., S. P. Dendy, E. E. Frank, M. N. Rouse, and S. E. Travers. 2006. Climate change effects on plant disease: Genomes to ecosystems. Annual Review of Phytopathology 44(1):489-509. Garrett, K. A., A. Jumpponen, C. Toomajian, and L. Gomez-Montano. 2012. Climate change and plant health: Designing research spillover from plant genomics for understanding the role of microbial communities. Canadian Journal of Plant Pathology 34(3):349-361. Ghini, R., W. Bettiol, and E. Hamada. 2011. Diseases in tropical and plantation crops as affected by climate changes: Current knowledge and perspectives. Plant Pathology 60(1):122-132. Ghini, R., E. Hamada, F. Angelotti, L. B. Costa, and W. Bettiol. 2012. Research approaches, adapta- tion strategies, and knowledge gaps concerning the impacts of climate change on plant diseases. Tropical Plant Pathology 37:5-24. Gomez, A., S. Balsari, J. Nusbaum, A. Heerboth, and J. Lemery. 2013. Perspective: Environment, bio- diversity, and the education of the physician of the future. Academic Medicine 88(2):168-172. Goswami, B. N., V. Venugopal, D. Sengupta, M. S. Madhusoodanan, and P. K. Xavier. 2006. Increasing trend of extreme rain events over India in a warming environment. Science 314(5804):1442-1445. Gouache, D., A. Bensadoun, F. Brun, C. Pagé, D. Makowski, and D. Wallach. 2013. Modelling cli- mate change impact on Septoria tritici blotch (STB) in France: Accounting for climate model and disease model uncertainty. Agricultural and Forest Meteorology 170:242-252. Gregory, P. J., and J. S. I. Ingram. 2000. Global change and food and forest production: Future sci- entific challenges. Agriculture, Ecosystems & Environment 82(1â3):3-14. Gross, A., O. Holdenrieder, M. Pautasso, V. Queloz, and T. N. Sieber. 2014. Hymenoscyphus pseu- doalbidus, the causal agent of European ash dieback. Molecular Plant Pathology 15(1):5-21. Grulke, N. E. 2011. The nexus of host and pathogen phenology: Understanding the disease triangle with climate change. New Phytologist 189(1):8-11. Grünwald, N. J., M. Garbelotto, E. M. Goss, K. Heungens, and S. Prospero. 2012. Emergence of the sudden oak death pathogen Phytophthora ramorum. Trends in Microbiology 20(3):131-138. Harwood, T. D., X. Xu, M. Pautasso, M. J. Jeger, and M. W. Shaw. 2009. Epidemiological risk assess- ment using linked network and grid based modelling: Phytophthora ramorum and Phytophthora kernoviae in the UK. Ecological Modelling 220(23):3353-3361. Hennon, P. E., and C. G. Shaw. 1994. Did climatic warming trigger the onset and development of yel- low-cedar decline in southeast Alaska? European Journal of Forest Pathology 24(6-7):399-418. Hennon, P. E., D. V. DâAmore, P. G. Schaberg, D. T. Wittwer, and C. S. Shanley. 2012. Shifting cli- mate, altered niche, and a dynamic conservation strategy for yellow-cedar in the north Pacific coastal rainforest. Bioscience 62(2):147-158. Hicke, J. A., A. J. Meddens, C. D. Allen, and C. A. Kolden. 2013. Carbon stocks of trees killed by bark beetles and wildfire in the western United States. Environmental Research Letters 8(3):035032.
372 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Ilbery, B., D. Maye, J. Ingram, and R. Little. 2013. Risk perception, crop protection and plant disease in the UK wheat sector. Geoforum 50(0):129-137. Ingram, D. S. 2002. The diversity of plant pathogens and conservation: Bacteria and fungi sensu lato. In Microorganisms in plant conservation and biodiversity, edited by K. Sivasithamparama, K. W. Dixon, and R. L. Barrett. The Netherlands: Springer. Pp. 241-267. Ireland, K. B., G. E. S. J. Hardy, and D. J. Kriticos. 2013. Combining inferential and deductive ap- proaches to estimate the potential geographical range of the invasive plant pathogen, Phytoph- thora ramorum. PloS One 8(5):1-13. Jeger, M. J., and M. Pautasso. 2008. Plant disease and global change â the importance of long-term data sets. New Phytologist 177(1):8-11. Jeger, M. J., M. Pautasso, and J. Stack. 2011. Climate, globalization and trade: Impacts on dispersal and invasion of fungal plant pathogens. In Fungal diseases: An emerging challenge to human, animal and plant health, edited by L. A. Olsen, D. A. Relman, E. R. Choffnes, and L. Pray. Washington, DC: The National Academies Press. Johnson, K. B., A. Jacob, and M. E. Brown. 2013. Forest cover associated with improved child health and nutrition: Evidence from the Malawi Demographic and Health Survey and satellite data. Global Health: Science and Practice 1(2):237-248. Juroszek, P., and A. Tiedemann. 2013. Climate change and potential future risks through wheat dis- eases: A review. European Journal of Plant Pathology 136(1):21-33. Kaczynski, K. M., and D. J. Cooper. 2013. Susceptibility of Salix monticola to Cytospora canker under increased temperatures and decreased water levels. Forest Ecology and Management 305:223-228. Kiss, I. Z., M. Broom, P. G. Craze, and I. Rafols. 2010. Can epidemic models describe the diffusion of topics across disciplines? Journal of Informetrics 4(1):74-82. Kliejunas, J. T. 2011. A risk assessment of climate change and the impact of forest diseases on forest ecosystems in the Western United States and Canada: U.S. Department of Agriculture, Forest Service, Pacific Southwest Research Station. Kogan, M. 1998. Integrated pest management: Historical perspectives and contemporary develop- ments. Annual Review of Entomology 43:243-270. Kurz, W. A., C. C. Dymond, G. Stinson, G. J. Rampley, E. T. Neilson, A. L. Carroll, T. Ebata, and L. Safranyik. 2008. Mountain pine beetle and forest carbon feedback to climate change. Nature 452(7190):987-990. Liebhold, A. M., E. G. Brockerhoff, L. J. Garrett, J. L. Parke, and K. O. Britton. 2012. Live plant imports: The major pathway for forest insect and pathogen invasions of the US. Frontiers in Ecology and the Environment 10(3):135-143. Liebhold, A. M., D. G. McCullough, L. M. Blackburn, S. J. Frankel, B. Von Holle, and J. E. Aukema. 2013. A highly aggregated geographical distribution of forest pest invasions in the USA. Diver- sity and Distributions 19(9):1208-1216. Luck, J., M. Spackman, A. Freeman, P. Trebicki, W. Griffiths, K. Finlay, and S. Chakraborty. 2011. Climate change and diseases of food crops. Plant Pathology 60(1):113-121. Marvin, H. J. P., G. A. Kleter, H. J. Van der Fels-Klerx, M. Y. Noordam, E. Franz, D. J. M. Willems, and A. Boxall. 2013. Proactive systems for early warning of potential impacts of natural di- sasters on food safety: Climate-change-induced extreme events as case in point. Food Control 34(2):444-456. Mills, P., K. Dehnen-Schmutz, B. Ilbery, M. Jeger, G. Jones, R. Little, A. MacLeod, S. Parker, M. Pautasso, S. Pietravalle, and D. Maye. 2011. Integrating natural and social science perspectives on plant disease risk, management and policy formulation. Philosophical Transactions of the Royal Society of London B Biological Sciences 366(1573):2035-2044. Moslonka-Lefebvre, M., A. Finley, I. Dorigatti, K. Dehnen-Schmutz, T. Harwood, M. J. Jeger, X. Xu, O. Holdenrieder, and M. Pautasso. 2011. Networks in plant epidemiology: From genes to landscapes, countries, and continents. Phytopathology 101(4):392-403.
APPENDIX A 373 Munkvold, G. P., and X. Yang. 1995. Crop damage and epidemics associated with 1993 floods in Iowa. Plant Disease 79(1):95-101. Newton, A., S. Johnson, and P. Gregory. 2011. Implications of climate change for diseases, crop yields, and food security. Euphytica 179(1):3-18. Ogawa-Onishi, Y., and P. M. Berry. 2013. Ecological impacts of climate change in Japan: The importance of integrating local and international publications. Biological Conservation 157(0):361-371. OleÅ, K., E. Gudowska-Nowak, and A. Kleczkowski. 2012. Understanding disease control: Influence of epidemiological and economic factors. PloS One 7(5):e36026. OleÅ, K., E. Gudowska-Nowak, and A. Kleczkowski. 2013. Efficient control of epidemics spreading on networks: Balance between treatment and recovery. PloS One 8(6):e63813. Pangga, I. B., J. Hanan, and S. Chakraborty. 2013. Climate change impacts on plant canopy archi- tecture: Implications for pest and pathogen management. European Journal of Plant Pathology 135(3):595-610. Pasgaard, M., and N. Strange. 2013. A quantitative analysis of the causes of the global climate change research distribution. Global Environmental Change 23(6):1684-1693. Paterson, R., M. Sariah, and N. Lima. 2013. How will climate change affect oil palm fungal diseases? Crop Protection 46:113-120. Pautasso, M. 2013a. Fungal under-representation is (slowly) diminishing in the life sciences. Fungal Ecology 6(1):129-135. Pautasso, M. 2013b. Phytophthora ramorumâa pathogen linking network epidemiology, landscape pathology and conservation biogeography. CAB Reviews 8:1-14. Pautasso, M., K. Dehnen-Schmutz, O. Holdenrieder, S. Pietravalle, N. Salama, M. J. Jeger, E. Lange, and S. Hehl-Lange. 2010. Plant health and global changeâsome implications for landscape management. Biological Reviews of the Cambridge Philosophical Society 85(4):729-755. Pautasso, M., T. Döring, M. Garbelotto, L. Pellis, and M. Jeger. 2012a. Impacts of climate change on plant diseasesâopinions and trends. European Journal of Plant Pathology 133(1):295-313. Pautasso, M., K. Dehnen-Schmutz, B. Ilbery, M. J. Jeger, G. Jones, R. Little, A. MacLeod, D. Maye, S. Parker, and S. Pietravalle. 2012b. Plant health challenges for a sustainable land use and rural economy. CAB Reviews 7:63. Pautasso, M., P. Gonthier, and G. Nicolotti. 2013a. Responding to diseases caused by exotic tree pathogens. Infectious Forest Diseases: 592-612. Pautasso, M., G. Aas, V. Queloz, and O. Holdenrieder. 2013b. European ash (Fraxinus excelsior) diebackâA conservation biology challenge. Biological Conservation 158:37-49. Pearson-Mims, C. H., and V. I. Lohr. 2000. Reported impacts of interior plantscaping in office envi- ronments in the United States. HortTechnology 10(1):82-86. Preston, N. D., P. Daszak, and R. R. Colwell. 2013. The human environment interface: Applying ecosystem concepts to health. In One Health: The Human-Animal-Environment Interfaces in Emerging Infectious Diseases. Berlin: Springer. Pp. 83-100. Purse, B. V., P. Graeser, K. Searle, C. Edwards, and C. Harris. 2013. Challenges in predicting inva- sive reservoir hosts of emerging pathogens: Mapping Rhododendron ponticum as a foliar host for Phytophthora ramorum and Phytophthora kernoviae in the UK. Biological Invasions:1-17. Reichstein, M., M. Bahn, P. Ciais, D. Frank, M. D. Mahecha, S. I. Seneviratne, J. Zscheischler, C. Beer, N. Buchmann, and D. C. Frank. 2013. Climate extremes and the carbon cycle. Nature 500(7462):287-295. Richaud, C., D. Lebeaux, and O. Lortholary. 2013. Fungal infections associated with travel. Current Fungal Infection Reports 7(4):311-319. Rizzo, D. M., M. Garbelotto, and E. M. Hansen. 2005. Phytophthora ramorum: Integrative research and management of an emerging pathogen in California and Oregon forests. Annual Review of Phytopathology 43:309-335. Roos, J., R. Hopkins, A. Kvarnheden, and C. Dixelius. 2011. The impact of global warming on plant diseases and insect vectors in Sweden. European Journal of Plant Pathology 129(1):9-19.
374 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Rosenzweig, C., A. Iglesias, X. Yang, P. R. Epstein, and E. Chivian. 2001. Climate change and extreme weather events: Implications for food production, plant diseases, and pests. Global Change & Human Health 2(2):90-104. Russell, R., A. D. Guerry, P. Balvanera, R. K. Gould, X. Basurto, K. M. A. Chan, S. Klain, J. Levine, and J. Tam. 2013. Humans and nature: How knowing and experiencing nature affect well-being. Annual Review of Environment and Resources 38(1):473-502. Santini, A., L. Ghelardini, C. d. Pace, M.-L. Desprez-Loustau, P. Capretti, A. Chandelier, T. Cech, D. Chira, S. Diamandis, and T. Gaitniekis. 2013. Biogeographical patterns and determinants of invasion by forest pathogens in Europe. New Phytologist 197(1):238-250. Savary, S., A. Mila, L. Willocquet, P. Esker, O. Carisse, and N. McRoberts. 2011a. Risk factors for crop health under global change and agricultural shifts: A framework of analyses using rice in tropical and subtropical Asia as a model. Phytopathology 101(6):696-709. Savary, S., A. Nelson, A. H. Sparks, L. Willocquet, E. Duveiller, G. Mahuku, G. Forbes, K. A. Garrett, D. Hodson, and J. Padgham. 2011b. International agricultural research tackling the effects of global and climate changes on plant diseases in the developing world. Plant Disease 95(10):1204-1216. Savary, S., A. Ficke, J.-N. Aubertot, and C. Hollier. 2012. Crop losses due to diseases and their implications for global food production losses and food security. Food Security 4(4):519-537. Shaw, M. W., and T. M. Osborne. 2011. Geographic distribution of plant pathogens in response to climate change. Plant Pathology 60(1):31-43. Siebold, M., and A. Tiedemann. 2013. Effects of experimental warming on fungal disease progress in oilseed rape. Global Change Biology 19(6):1736-1747. Sturrock, R., S. Frankel, A. Brown, P. Hennon, J. Kliejunas, K. Lewis, J. Worrall, and A. Woods. 2011. Climate change and forest diseases. Plant Pathology 60(1):133-149. Sutherst, R. W., F. Constable, K. J. Finlay, R. Harrington, J. Luck, and M. P. Zalucki. 2011. Adapt- ing to crop pest and pathogen risks under a changing climate. Wiley Interdisciplinary Reviews: Climate Change 2(2):220-237. Thompson, S., S. Levin, and I. Rodriguez-Iturbe. 2013. Linking plant disease risk and precipitation drivers: A dynamical systems framework. American Naturalist 181(1):E1-E16. VáclavÃk, T., and R. K. Meentemeyer. 2009. Invasive species distribution modeling (iSDM): Are absence data and dispersal constraints needed to predict actual distributions? Ecological Model- ling 220(23):3248-3258. van der Waals, J. E., K. Krüger, A. C. Franke, A. J. Haverkort, and J. M. Steyn. 2013. Climate change and potato production in contrasting South African agro-ecosystems 3. Effects on relative devel- opment rates of selected pathogens and pests. Potato Research 56(1):67-84. Werres, S., R. Marwitz, W. A. Man Inât veld, A. W. A. M. De Cock, P. J. M. Bonants, M. De Weerdt, K. Themann, E. Ilieva, and R. P. Baayen. 2001. Phytophthora ramorum sp. nov., a new pathogen on Rhododendron and Viburnum. Mycological Research 105(10):1155-1165. West, J., J. Townsend, M. Stevens, and B. L. Fitt. 2012. Comparative biology of different plant pathogens to estimate effects of climate change on crop diseases in Europe. European Journal of Plant Pathology 133(1):315-331. Woodward, S., and E. Boa. 2013. Ash dieback in the UK: A wake-up call. Molecular Plant Pathol- ogy 14(9):856-860. Worrall, J. J., G. E. Rehfeldt, A. Hamann, E. H. Hogg, S. B. Marchetti, M. Micaelian, and L. K. Gray. 2013. Recent declines of Populus tremuloides in North America linked to climate. Resilience in quaking aspen: Restoring ecosystem processes through applied science. Forest Ecology and Management 299:35â51. Xu, X., T. D. Harwood, M. Pautasso, and M. J. Jeger. 2009. Spatio-temporal analysis of an invasive plant pathogen (Phytophthora ramorum) in England and Wales. Ecography 32(3):504-516.
APPENDIX A 375 A18 WATER QUALITY AND HEALTH FOR A SUSTAINABLE SOCIETY Joan B. Rose,80 Georgia Mavrommati,80 and Erin A. Dreelin80 Abstract Sustaining high-quality aquatic ecosystems is critical for ensuring global biohealth. However, critical knowledge gaps limit our ability for efficient and effective management of human and natural water systems to ensure human health. We recommend a major program in water and health, promoting wa- ter quality diagnostic tools and pathogen discovery focusing on key exposure pathways. This program should: 1. Fill key knowledge gaps necessary for using the quantitative micro- bial risk assessment framework: a. Critical water quality diagnostic tools should be used to address exposure assessment for recalcitrant and emerging pathogens, which includes exploration of the water microbiome along the human exposure pathway. b. Health risks should be calculated and known for all waters in the United States. 2. Develop a framework for transdisciplinary research with respect to biohealth linked to water systems. System dynamics methodology should be used to: a. Create and explore the theoretical couplings between the vari- ous components of water systems (including the socioeconomic [health] and biophysical [water quality]). b. Develop and operationalize indicators for assessing and managing key linkages among various components of human and natural systems. Background Water is one of the most critical of the worldâs life support systems. Water quantity and quality are inextricably linked with global biohealth. Biohealth is the health of all animals, humans, and plants (including the microbiome) on Earth linked to a healthy biosphere. High-quality, functioning aquatic ecosystems sup- port sustainable plant, animal, and human communities and provide critical ser- vices to humans, from food and drinking water to industry and recreation. In the last 60 years we have seen a great acceleration of population growth in people and 80ââMichigan State University.
376 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS animals, land use change, use of fertilizers, and water use as well as the global transport of humans and animals. This has led us into the Anthropocene, in which continued water quality degradation as demonstrated by increased eutrophica- tion and fecal contamination associated with microbial hazards and antibiotic resistance is a global phenomenon. This degradation in global water quality and quantity is exacerbated by climate change and extreme events. Despite our invest- ment in infrastructure and better environmental protection policies, water pollu- tion shows a continual and dramatic impact on health in the developed world and devastates communities in the developing world. Waterborne diseases in humans are characterized by pathogens that are persistent, potent, excreted at high num- bers, and often zoonotic. Waterborne disease outbreaks of Clostridium difficile and Guillian-Barré syndrome in Europe and the United States are emerging. Rare amoeba associated with high mortality in children are showing up in association with tap water. Waterborne poliovirus and cholera have not been controlled, and zoonotic diseases including bacteria such as E. coli 0157H7, Campylobacter, and Salmonella; parasites like Giardia and Cryptosporidium; and new emerging viruses like Cyclovirus remain global threats to animal and human health. Animal and human health can be addressed through targeted monitoring and management strategies using the quantitative microbial risk assessment (QMRA) framework and molecular tools. Point and diffuse pollution sources and specific hazards are now identifiable using technology such as microbial source tracking. In addition, system dynamics models can improve decision making regarding complex systems by using integrated socioeconomic, water quality, and health data to couple human and water systems, thereby explicitly linking human health and well-being with ecosystems services. In the future, it will be more important than ever to operationalize QMRA and develop system dynamics models in order to effectively and efficiently mitigate the impacts of climate, an aging infrastructure (or lack thereof), and the global changes that are now occurring to protect and restore the biohealth of the planet. Key scientific questions around water quality and health remain: · How is water quality changing and affecting human health? · What are the sources of emerging, recalcitrant, and problematic pathogens? · How does ecosystem health relate to human health? · How do we restore and protect water systems to protect the biohealth of the planet? We recommend a major program in water and health, promoting water quality diagnostic tools and pathogen discovery focusing on key exposure path- ways. These data then must be used within community-based and health risk
APPENDIX A 377 frameworks to improve decision making for enriching the future of our complex human-coupled water systems and health. Waterborne Disease: Emerging and Zoonotic Pathogens Waterborne disease problems seem to have been solved in the United States, yet large or dramatic outbreaks continue to occur in both drinking water systems and recreational waters. Understanding the exposure pathways (Figure A18-1) that move pathogens from humans and animals into water and back to susceptible populations via drinking water, recreational water, and food is critical due to emerging, recalcitrant, and problematic infectious agents associated with human or animal wastes and the water environment. Waterborne outbreaks continue to occur in the United States, and these are the âplane crashesâ for the water industry and for communities. From 1971 to 2008 in the United States, more than a half million people (576,853 persons) were ill during 747 documented waterborne outbreaks caused by bacteria (15 percent), viruses (9 percent), protozoa (19 percent), and chemicals (11 percent) (Haas et al., 2014). The etiological agents causing acute gastrointestinal illness in a large percentage of the outbreaks were not identified (45 percent), and those with multiple types of pathogens were less than 1 percent. In the last decade, outbreaks in community drinking water have decreased and averaged about five FIGURE A18-1 Exposure pathways in the human-water coupled system.
378 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A18-2â Outbreaks in drinking water in the United States. per year (Figure A18-2) with more outbreaks associated with multiple pathogens and animal manure or human sewage as the source of the pathogens. In ambient recreational waters where outbreaks are much more difficult to identify, no such decrease in disease is evident (Figure A18-3), and in the last few years between 5 and 15 outbreaks have occurred (Haas et al., 2014). The outbreak data do not capture endemic waterborne disease risks, referred to as the iceberg because most of these cases go unreported (the tip being the reportable outbreaks). There are several estimates of endemic waterborne disease risks: · 12 million cases/year (Eisenberg et al., 2006), · 16 million cases/year (Messner et al., 1996), and · ~19 million waterborne illnesses/year for community water systems in the United States (5.4 million illnesses from groundwater and 13 million illnesses from surface water systems) (Reynolds et al., 2008). New disease concerns emerge every year as a result of water contami- nation. A waterborne cluster of 24 cases of Guillain-Barré syndrome (GBS) along the U.S.-Mexico border near Yuma, Arizona, was identified in 2011 due to Campylobacter (AZ Department of Health, http://www.azdhs.gov/diro/pio/ news/2011/110718_Yuma_GBS.pdf). The first documentation of waterborne dis- ease caused by Clostridium difficile was due to sewage contamination of tap water (Kotila et al., 2013). Nagleria fowleri infections associated with high mortality
APPENDIX A 379 FIGURE A18-3â Outbreaks in ambient recreational water per year in the United States. in children are now showing up in association with tap water.81 Waterborne po- liovirus and cholera have not been controlled, and zoonotic diseases including E. coli 0157H7, Campylobacter, and Salmonella, and parasites like Giardia and Cryptosporidium, remain global threats to animal and human health. Finally, concerns about antibiotic resistance are coming to the forefront, and studies on polluted water show that these markers are prolific (Ashbolt et al., 2013). Emerging viral diseases in humans are now a new threat to water systems. Enteric viruses were once thought to be host specific, yet the human Cyclovirus is related to the porcine circovirus (PCV1 and PCV2) that infects pigs. New genetic variants and the genetic diversity of the virus are not well understood. Recent global studies have identified the following: · A fecal-oral Cyclovirus causing acute central nervous system infections in humans was sequenced, and 60 percent of the pigs tested positive for this novel virus in Vietnam (Tan et al., 2013). · Studies of human feces and food products in Minnesota, Nigeria, and Pakistan reported that the viruses in the stools of U.S. adults were porcine circoviruses. They were also found in most U.S. pork products, yet were more diverse compared to Asian and African samples (Li et al., 2010). The Cyclovirus is a single-stranded DNA virus, which is smaller than all 81ââSee http://www.cdc.gov/parasites/naegleria; http://www.cdc.gov/parasites/naegleria/public-water- systems.html (accessed August 6, 2014).
380 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS other known enteric viruses and little is known about its resistance to wa- ter treatment. These viruses remain a significant risk as they are excreted in high concentrations, survive in the environment, and can be spread through contaminated manure to people, other animals, and water. In summary, waterborne disease continues to occur in the United States and throughout the world. Climate variability and increasing intensity of storms could increase outbreaks (Curriero et al., 2001), and aging infrastructure is leading us toward a future of a greater probability of pathogen contamination of water. Emerging pathogens, those associated with chronic diseases, zoonotic transmis- sion, and antibiotic resistance in water, are a threat to human health. Develop- ment of molecular-based water diagnostic tools are needed to characterize these risks along recreational, drinking water, and irrigation water exposure pathways. Yet ultimately this characterization of the water pollution biome needs to be put into the context of public health decision frameworks. Both dynamic models of water quality and health linked to management and quantitative microbial risk assessment provide the approaches for understanding and controlling waterborne diseases. System Dynamics Modeling for Water and Health Fostering pathways of biohealth associated with water systems requires (i) definition of the goals of a sustainable society in operational terms with respect to human health, water quality, and quantity, and (ii) an understanding of the process by which the various subsystems associated with water interact and affect public health (Mavrommati et al., 2013a). Defining biohealth pathways at an operational level involves not only scientific knowledge but also ethical considerations about how future generations should be treated by the current generation. Protecting earthâs key life support systems is a prerequisite for a sustainable society, a goal reiterated during the discussions of the United Nations Rio+20 Summit in 2012 (Griggs et al., 2013). Recent approaches for sustainability at the global level suggest that life support systems related to human health and water quality are at stake of exceeding the planetary boundaries within which humanity is safe (Barnosky et al., 2012; Rockstrom et al., 2009). Thus, there is an urgent need to identify and understand the boundaries (planetary, regional, or local) within which water systems can function and secure biohealth. Studying the couplings among the elements of water systems transcends the boundaries of one discipline and requires the use of methodologies employing transdisciplinary approaches to include components of socioeconomic, public health, and biophysical systems. Mainstream approaches in which each disci- pline develops separated models are not sufficient for water system analysis (Simonovic and Davies, 2006). The methodology of system dynamics (SD) has been extensively used for coupling various disciplines into a common framework
APPENDIX A 381 of analysis that can be understandable by decision makers and other stakeholders. The methodology of SD is based on systems thinking and recognizes that the in- ability to capture the structure and dynamics of complex systems stems from the lack of holistic approaches (Sterman, 2000). SD models for water systems can determine the roots of unsustainable water resources management by identifying the source of the problem both in qualitative and quantitative terms (Mirchi et al., 2012; Sterman, 2012). In this way, decision makers and other stakeholders can better evaluate the effects of their actions in relation to the problematic behavior of water systems; hence, they may adopt more informed and efficient policies for sustaining biohealth in water systems (Hopkins et al., 2012; Mavrommati et al., 2013b). A system dynamics model displays positive and negative feedbacks (rein- forcing and counteractive, respectively) between different variables and drivers, and it helps us understand how a system changes over time. Failure to identify the feedback loops related to the problems of the studied system results in mis- leading models for decision making. SD models can be both qualitative and quantitative. Qualitative models are represented through causal loop diagrams (CLDs), and quantitative models through stock-flow diagrams where parameter values and functions are estimated through various types of data (e.g., mental, numerical, written). The recreational waterborne disease exposure pathway is shown as an ex- ample CLD for the Lake St. Clair (LSC) region, which is part of the Lauren- tian Great Lakes System, as a case study (Figure A18-4) (Mavrommati et al., 2013a). This CLD includes the analysis of two counteractive and reinforcing loops with respect to water quality, lake visitors, and human health/well-being (Figure A18-5a,b). Wastewater discharge and beach use create pollutants/exposure pathways, and the impact from these pollutants entering lake waters can be measured by chlorophyll a, a proxy for primary production and toxic algal exposure, and con- centrations of the fecal indicator bacteria Escherichia coli (E. coli) that is used to measure pathogen risks. High concentrations of E. coli directly affect public health. In addition, when E. coli is above a certain level defined through current water quality regulations, then local beaches are closed to protect public health. Beach closures result in the decrease of the number of lake visitors and impact public health policies. Drinking and recreational water pollution associated with sewage and nonpoint sources of pathogens can be examined along with climate factors (precipitation). This example presents specific aspects of biohealth in water systems and the ability of system dynamics modeling to capture their complexity. In summary, recreational pathways for waterborne disease may be becom- ing increasingly important, yet lack of data and appropriate models to capture the interactions between society, climate, water quality, ecosystem services, and health make it difficult to address the complexities of what makes a sustainable
382 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS FIGURE A18-4â A causal loop diagram that represents coupled socioeconomic and bio- physical systems in Lake St. Clair. Reproduced from Mavrommati et al. (2013a). NOTE: HWB: Human well-being. and safe water system. Models allow one to explore changes in pathogen sources associated with various infrastructure; drivers of deteriorating water quality such as human dynamics, climate change, eutrophication, and algal blooms; and the role of policies in protecting health. This will be particularly important for ex- amining future scenarios. Quantitative Microbial Risk Assessment (QMRA) One of the missions of the National Institute of Environmental Health Sci- ences is the investigation of environmental exposures and human disease. How- ever, research has focused on chemical pollutants, whereas exposure to microbial infectious agents through the environment was not considered. This gap has begun to be filled by the field of risk assessment for microbial pathogens us- ing innovative water genomic technology, quantitative data on persistence in the environment, dose-response, and understanding of acute and chronic health outcomes. QMRA was developed through the integration of environmental mi- crobiology, environmental engineering, and infectious disease epidemiology. Clinical experimental data were first examined by Haas (1983) demonstrating
FIGURE A18-5â A causal loop diagram that represents a reinforcing feedback loop (symbolized with R) and a counteractive feedback loop (symbolized with C). The positive (+) arrows represent a cause-and-effect relationship in which the two parameters change in the same direc- tion, while the negative (-) rows represent two parameters that change in the opposite direction (Ford, 1999). Reproduced from Mavrommati et al. (2013a). NOTE: HWB: Human well-being. 383
384 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS that dose-response modeling could be used to examine probability of infection. The Center for Advancing Microbial Risk Assessment (CAMRA) funded by the Environmental Protection Agency (EPA) and Department of Homeland Security has now produced a QMRA wiki that houses over 70 dose-response data sets and models for addressing pathogens spread through air, fomites, soil, and water (http://qmrawiki.msu.edu/index.php?title=Dose_Response). QMRA includes four steps: 1. Hazard identification: This step includes gathering information regard- ing the identification of the pathogen(s) and description of the pathogenic- ity and illness(es). Community-based information on endemic baseline disease levels and attack rates during outbreaks of environmentally trans- mitted pathogens are also needed. Pathogen discovery is critical, and new genomic tools will be extremely useful. Recent studies on detection and characterization of viruses in community sewage using microarray technology with a total of 780 unique probes targeting 27 different groups revealed a high variety of RNA (astroviruses and enteroviruses) and DNA viruses (adenoviruses, particularly type 41 and BK polyomavirus) (Wong et al., 2013). 2. Dose-response: It is now possible to describe the quantitative relationship between doses of the pathogen due to specific exposure pathways and health outcomes as a probability of infection, disease, or in some cases mortality depending on the data sets. Comparative assessments of doses using the ID50 or LD50 is an approach used to compare the potency of pathogens (Figure A18-6) based on the QMRAwiki (http://qmrawiki.msu. edu/index.php?title=Dose_Response) and illustrates the variability across pathogens. 3. Exposure assessment: This is the most challenging part of the QMRA because exposure pathways are complex and data on prevalence, con- centrations, distributions (in time and space), and the volume of water consumed are necessary. There are many exposure pathways that require specific parameterization for water. Pathogen persistence under a range of environmental conditions is particularly important. Reductions of the microorganisms during various treatment processes (e.g., disinfection and filtration), and an understanding of the natural history of the pathogen in the built and natural water environment are key to undertaking exposure assessment. Molecular methods are used to improve the pathogen resolu- tion and temporal and spatial distributions for the various polluted water environs. Databases for the water microbiome are needed. 4. Risk characterization: The quantitative likelihood of a potential adverse health outcome is based on the above three steps. Risk characterization estimates the magnitude of the public health problem, while also deter- mining the variability and uncertainty of the hazard (Haas et al., 2014).
APPENDIX A 385 FIGURE A18-6â Comparative assessments of ID50 or LD50 (Y axis represents the dose in colonies, virons, or cells). The smaller the bar the more potent the organism. NOTE: EHEC = enterohemorrhagic E. coli.
386 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS This step is generally pathogen specific and has been used for static esti- mates focusing on describing the most reasonably exposed individual or describing the distributions of risk with techniques such as Monte Carlo (probabilistic) analysis (Medema et al., 2006; Smeets et al., 2008). 5. Risk management: QMRA is used to inform policy and management decisions. Following the new framework suggested by the recent report Science and Decisions: Advancing Risk Assessment from the National Academy of Sciences (NRC, 2009), QMRA should integrate the problem identification and risk management strategies into the four-step process. As an example, the effectiveness of household water treatment at reducing diarrheal disease caused by viruses, parasites, and bacteria was evaluated based on treatment efficacy and compliance (Enger et al., 2013) using a QMRA model. The quantitative relationship between compliance and ef- fectiveness is poorly understood; however, this study showed that benefits associated with water treatment were strongly dependent on how many of the households had adequate treatment within the community. Thus policy debates around community water systems and access to safe water and treatment need to include data that protect public health, particularly from emerging contaminants. QMRA methodology has proven invaluable for assisting local, national, and international decision making for protection of health and water quality. In summary, hundreds of human pathogens can be found in water, and while dose-response models exist for many, there is very little information on exposure including the concentrations, fate, and transport of microbial hazards in the en- vironment. Exposure pathways for emerging and recalcitrant pathogens are not well described. QMRA frameworks now allow data to be systematically evaluated so disease can be better understood. These approaches are just now being further expanded to animal and plant diseases. Through such analyses, critical informa- tion that is needed for decision making to address water security and safety can now be identified, gathered, synthesized, and communicated. Research Recommendations for Addressing Biohealth in Urban Water Systems There are a tremendous number of scientific, technological, and societal questions that need to be answered as we attempt to address health and water quality in the Anthropocene. Several key questions are: · What is the risk of waterborne disease? · Will it change with changes in animal populations? · Will it change with population growth? · Will it change with new emerging pathogens?
APPENDIX A 387 · What is the risk of an outbreak occurring? · Will it change with precipitation? · Will it change with the failure of our infrastructure and treatment? Global water quality is deteriorating and threatens public health. Human wastewater and animal wastes remain some of the most important sources of pathogens. Preparing for a future where climate, water infrastructure, and emerg- ing pathogens are colliding and impacting the components of human well-being (e.g., human health, subjective well-being derived from recreation) and ecosys- tem services (e.g., drinking water and recreation) necessitates research on the essential elements for sustainable water and biohealth of the planet. The research recommendations that should be considered include the following: · Fill key knowledge gaps necessary for using the QMRA framework: Critical water quality diagnostic tools should be used to address ex- posure assessment for recalcitrant and emerging pathogens, which in- cludes exploration of the water microbiome along the human exposure pathway. Health risks should be calculated and known for all waters in the United States. · Develop a framework for transdisciplinary research with respect to bio- health linked to water systems. The systems dynamics methodology should be used to: Create and explore the theoretical couplings between the various com- ponents of water systems (including the socioeconomic [health] and biophysical [water quality]). Develop and operationalize indicators for assessing and managing key linkages among various components of human and natural systems. References Ashbolt, N. J., A. Amezquita, T. Backhaus, P. Borriello, K. K. Brandt, P. Collignon, A. Coors, R. Finley, W. H. Gaze, T. Heberer, J. R. Lawrence, D. G. Larsson, S. A. McEwen, J. J. Ryan, J. Schonfeld, P. Silley, J. R. Snape, C. Van den Eede, and E. Topp. 2013. Human Health Risk Assessment (HHRA) for environmental development and transfer of antibiotic resistance. En- vironmental Health Perspectives 121(9):993-1001. Barnosky, A. D., E. A. Hadly, J. Bascompte, E. L. Berlow, J. H. Brown, M. Fortelius, W. M. Getz, J. Harte, A. Hastings, P. A. Marquet, N. D. Martinez, A. Mooers, P. Roopnarine, G. Vermeij, J. W. Williams, R. Gillespie, J. Kitzes, C. Marshall, N. Matzke, D. P. Mindell, E. Revilla, and A. B. Smith. 2012. Approaching a state shift in Earthâs biosphere. Nature 486(7401):52-58. Curriero, F. C., J. A. Patz, J. B. Rose, and S. Lele. 2001. The association between extreme precipita- tion and waterborne disease outbreaks in the United States, 1948-1994. American Journal of Public Health 91(8):1194-1199.
388 GLOBAL CHANGE AND INFECTIOUS DISEASE DYNAMICS Eisenberg, J. N., A. Hubbard, T. J. Wade, M. D. Sylvester, M. W. LeChevallier, D. A. Levy, and J. M. Colford, Jr. 2006. Inferences drawn from a risk assessment compared directly with a randomized trial of a home drinking water intervention. Environmental Health Perspectives 114(8):1199-1204. Enger, K. S., K. L. Nelson, J. B. Rose, and J. N. Eisenberg. 2013. The joint effects of efficacy and compliance: A study of household water treatment effectiveness against childhood diarrhea. Water Research 47(3):1181-1190. Ford, F. A. 1999. Modeling the environment: An introduction to system dynamics modeling of envi- ronmental systems. Washington, DC: Island Press. Griggs, D., M. Stafford-Smith, O. Gaffney, J. Rockstrom, M. C. Ohman, P. Shyamsundar, W. Steffen, G. Glaser, N. Kanie, and I. Noble. 2013. Policy: Sustainable development goals for people and planet. Nature 495(7441):305-307. Haas, C. N. 1983. Estimation of risk due to low doses of microorganisms: A comparison of alternative methodologies. American Journal of Epidemiology 118(4):573-582. Haas, C. H., J. B. Rose, and C. P. Gerba, (eds). 2014. Quantitative microbial risk assessment, 2nd edition. New York: John Wiley and Sons. Hopkins, T. S., D. Bailly, R. Elmgren, G. Glegg, A. Sandberg, and J. G. Støttrup. 2012. A systems approach framework for the transition to sustainable development: Potential value based on coastal experiments. Ecology and Society 17(3). Kotila, S. M., T. Pitkanen, J. Brazier, E. Eerola, J. Jalava, M. Kuusi, E. Kononen, J. Laine, I. T. Mi- ettinen, R. Vuento, and A. Virolainen. 2013. Clostridium difficile contamination of public tap water distribution system during a waterborne outbreak in Finland. Scandinavian Journal of Public Health 41(5):541-545. Li, L., A. Kapoor, B. Slikas, O. S. Bamidele, C. Wang, S. Shaukat, M. A. Masroor, M. L. Wilson, J. B. Ndjango, M. Peeters, N. D. Gross-Camp, M. N. Muller, B. H. Hahn, N. D. Wolfe, H. Triki, J. Bartkus, S. Z. Zaidi, and E. Delwart. 2010. Multiple diverse circoviruses infect farm animals and are commonly found in human and chimpanzee feces. Journal of Virology 84(4):1674-1682. Mavrommati, G., K. Bithas, and P. Panayiotidis. 2013a. Operationalizing sustainability in urban coastal systems: A system dynamics analysis. Water Research 47(20):7235-7250. Mavrommati, G., M. M. Baustian, and E. A. Dreelin. 2013b. Coupling socioeconomic and lake sys- tems for sustainability: A conceptual analysis using Lake St. Clair region as a case study. Ambio doi:10.â1007/âs13280-013-0432-4. Medema, G., J.-F. Loret, T. A. Stenström, and N. Ashbolt. 2006. Quantitative microbial risk assess- ment in the water safety plan. Final Report on the EU MicroRisk Project. Brussels: European Commission. Messner, M., S. Shaw, S. Regli, K. Rotert, V. Blank, and J. Soller. 2006. An approach for developing a national estimate of waterborne disease due to drinking water and a national estimate model application. Journal of Water and Health 4(Suppl 2):201-240. Mirchi, A., K. Madani, D. Watkins, Jr., and S. Ahmad. 2012. Synthesis of system dynamics tools for holistic conceptualization of water resources problems. Water Resources Management 26(9):2421-2442. NRC (National Research Council). 2009. Science and decisions: Advancing risk assessment. Wash- ington, DC: The National Academies Press. Reynolds, K. A., K. D. Mena, and C. P. Gerba. 2008. Risk of waterborne illness via drinking water in the United States. Reviews of Environmental Contamination and Toxicology 192:117-158. Rockstrom, J., W. Steffen, K. Noone, A. Persson, F. S. Chapin, 3rd, E. F. Lambin, T. M. Lenton, M. Scheffer, C. Folke, H. J. Schellnhuber, B. Nykvist, C. A. de Wit, T. Hughes, S. van der Leeuw, H. Rodhe, S. Sorlin, P. K. Snyder, R. Costanza, U. Svedin, M. Falkenmark, L. Karlberg, R. W. Corell, V. J. Fabry, J. Hansen, B. Walker, D. Liverman, K. Richardson, P. Crutzen, and J. A. Foley. 2009. A safe operating space for humanity. Nature 461(7263):472-475. Simonovic, S. P., and E. G. Davies. 2006. Are we modelling impacts of climatic change properly? Hydrological Processes 20(2):431-433.
APPENDIX A 389 Smeets, P., G. Medema, Y. Dullemont, P. Van Gelder, and J. Van Dijk. 2008. Improved methods for modelling drinking water treatment in quantitative microbial risk assessment; A case study of Campylobacter reduction by filtration and ozonation. Journal of Water and Health 6(3):301-314. Sterman, J. 2000. Business dynamics. New York: Irwin-McGraw-Hill. Sterman, J. D. 2012. Sustaining sustainability: Creating a systems science in a fragmented academy and polarized world. In Sustainability Science. Springer. Pp. 21-58. Tan, le V., H. R. van Doorn, H. D. Nghia, T. T. Chau, T. P. Tu le, M. de Vries, M. Canuti, M. Deijs, M. F. Jebbink, S. Baker, J. E. Bryant, N. T. Tham, B. K. NT, M. F. Boni, T. Q. Loi, T. Phuong le, J. T. Verhoeven, M. Crusat, R. E. Jeeninga, C. Schultsz, N. V. Chau, T. T. Hien, L. van der Hoek, J. Farrar, and M. D. de Jong. 2013. Identification of a new cyclovirus in cerebrospinal fluid of patients with acute central nervous system infections. mBio 4(3):e00231-00213. Wong, M. V., S. A. Hashsham, E. Gulari, J. M. Rouillard, T. G. Aw, and J. B. Rose. 2013. Detec- tion and characterization of human pathogenic viruses circulating in community wastewater using multitarget microarrays and polymerase chain reaction. Journal of Water and Health 11(4):659-670.
