Under the Weather: Climate, Ecosystems, and Infectious Disease (2001)

Page 45

Page 46

Page 47

worldwide very year (WHO, 1998a). Dengue virus is carried by Aedes aegypti, a mosquito that preferentially feeds on humans and is particularly suited to urban environments. When a mosquito bites an infected person, it acquires the virus through the blood meal. In the insect's gut the virus replicates and spreads to the salivary gland, priming the mosquito for transmission of infection to another person.

Climate influences. The abundance of dengue vectors depends in part on the availability of breeding sites, primarily containers such as drums, discarded tires, and leaf axils that are filled with water either manually or by rainfall. Where containers are manually filled for water storage, vector abundance is largely independent of rainfall, as seen in some areas of Bangkok, Thailand (Sheppard et al., 1969). In contrast, in southwestern Puerto Rico a high proportion of available breeding sites are discarded containers such as tires, bottles, and tin cans that become filled with rainfall. In this case there is a clear correlation between rainfall and A. aegypti abundance (Moore et al., 1978). Heavy rainfall tends to overflow containers and can thus discourage vector breeding, while extended drought conditions have in some cases led to higher vector abundance due to an increased use of water storage containers.

Saturation deficit, a parameter taking into account both temperature and relative humidity, affects the survival of eggs and adults. Newly laid eggs are subject to desiccation, and adults can experience moisture-related reductions in survival throughout their lifetimes. However, saturation deficits high enough to reduce egg or adult survival, according to the few published studies, are rarely encountered in humid tropical locations (Southwood et al., 1972). Atmospheric moisture also influences the rate of water loss from containers, which in turn affects vector abundance.

Temperature affects the potential spread of the virus through each stage in the life cycle of the mosquito. Adult and immature A. aegypti survive in a broad range of temperatures, from about 5°C to 42°C, although temperatures below 20°C reduce or prevent eggs from hatching. Temperature also influences the time required for embryonic, larval, and pupal development and plays a major role in the frequency of biting. Temperature also affects the extrinsic incubation period (EIP), the period between when the mosquito imbibes virus-laden blood and actually becomes infectious. At lower temperatures the EIP is longer and the mosquito is less likely to survive long enough to transmit the virus. In the A. aegypti/dengue system, EIP is a non-linear function of temperature such that even small changes in temperature introduce a seasonality into transmission dynamics (Focks et al., 1995).

Recent studies of the distribution and epidemiology of dengue viruses suggest that projected climate warming is generally expected to increase the intensity of transmission. For instance, Patz et al. (1998) estimate that for regions where dengue is already present, a mean temperature increase of about 1°C

Page 48

increases the aggregate epidemic risk by an average of 31 to 47 percent. Higher infection rates translate into a greater number of individuals who have experienced multiple infections and thereby may have an elevated risk for developing serious dengue illness (Halstead, 1988; Jetten and Focks, 1997). Another possibility to consider, however, is that climate change may lead to lower average atmospheric moisture levels in some regions, which could actually reduce dengue transmission.

Page 49

desert and highland fringes. In northern Pakistan, higher temperatures associated with El Niño correlated with increased malaria incidence (Bouma et al., 1994). Malaria outbreaks in the former Punjab province of India were more likely in the year after an El Niño with a relative risk of 4.5, and in El Niño years in Sri Lanka with a relative risk of 3.6 (Bouma and van der Kaay, 1996). There was a highly consistent association between El Niño years and outbreaks of malaria in Venezuela (Bouma and Dye, 1997) and Colombia (Bouma et al., 1997). Linblade et al. (1999) found that in the highland region of southwestern Uganda during the El Niño event of 1997-1998, an increase in rainfall was positively correlated with an increase in vector density and incidence of malaria (almost three times the mean incidence of the preceding five years). These associations are strong and consistent and imply that ENSO forecasts may have value in predicting disease risk in particular regions, even though the mechanisms underlying these associations are not always apparent.

At present, there is insufficient evidence to clearly attribute an increase in malaria incidence or its geographic spread to long-term global warming patterns. While some studies have predicted that global climate change could potentially lead to widespread increases in malaria transmission by expanding mosquito habitat range and “vectorial capacity” (Martens et al., 1999), other models have projected only a negligible extension in potential malaria transmission due to climate change (Rogers and Randolph, 2000; these studies are discussed further in Chapter 5). Regardless, these models can only simulate the potential transmission risk; actual occurrence of the disease is determined largely by socioeconomic factors. For instance, historically malaria was common in temperate regions but is no longer found in such regions because of environmental sanitation, mosquito control (e.g., draining swamps that breed some species of Anopheles), and lifestyle changes that allow people to minimize exposure to mosquitos (e.g., by screening houses and using air conditioning). Similarly, Reiter (2000) emphasizes that an increase in the potential malaria burden due to future warming trends could be offset by other factors, especially human interventions, as documented by the fact that during the warming trend of the past three centuries, the geographic distribution of malaria diminished rather than increased.

Page 50

mer, during which the city's open sewage ditches did not flush and mosquito breeding was unusually abundant (Kinsella, 1935). There have been 4,478 reported human cases of St. Louis encephalitis in the United States since 1964, with an average of 128 cases reported annually (Shope and Tsai, 1998). One in 338 infected people becomes sick, and the case-fatality ratio is between 2 percent in young people and 23 percent in persons over 70 (Monath, 1980).

The urban mosquito vector of St. Louis encephalitis is Culex pipiens, which maintains the virus in a life cycle involving several species of wild birds. The mosquito is infected when it takes a blood meal from a viremic bird. The virus multiplies in the mosquito, and after an extrinsic incubation of about two weeks, the virus appears in the saliva and can be transmitted either to another bird or to a person. The virus also has a rural cycle in the western United States, where the Culex tarsalis mosquito is the major vector. This mosquito breeds in flood waters and is abundant in irrigated fields and riverine flood plains, and thus excess rainfall and abundant snowmelt can favor its breeding (Monath, 1980).

Climate influences. Temperature has a major influence on the development of the Culex mosquito and St. Louis encephalitis virus in the mosquito. Epidemics occur primarily in southern latitudes where the mean June temperature is 21°C or above. Monath and Tsai (1987) compared retrospectively 15 urban U.S. outbreaks of St. Louis encephalitis, and found statistically significant associations with increased precipitation and temperature in January, decreased temperature in April, and increased temperature in May. Seven of 15 epidemic years had all of these characteristics, as contrasted to only two of the 130 non-epidemic years.

Reeves et al. (1994) studied the ecology of St. Louis encephalitis in mosquitos of California by comparing the warm southern San Joachim Valley with the colder northern Sacramento Valley. Based on their findings about the dynamics of mosquito infection and life cycles, they predicted that if a 5°C warming were to occur, St. Louis encephalitis would become less prevalent in the warmer south and would be distributed farther north in California, Oregon, and Washington. This prediction, however, did not take into account socioeconomic factors that might modify exposure of persons to infected mosquitos.

Page 51

and cattle to abortion among pregnant ungulates and fatalities in young animals (Shimshony and Barzilai, 1983). Human disease may be mild but can involve abrupt onset of high fever, severe headache, myalgia, and incapacitation for several days (Peters and Meegan, 1981). Some patients develop fatal hemorrhagic fever or encephalitis. There is a 1 percent case fatality rate in humans. Low-level endemic transmission occurs regularly throughout much of the African continent, but most of this remains unrecognized due to inadequate surveillance and health care facilities (Meegan and Bailey, 1988).

RVF virus is transmitted by mosquitos to both domestic animals and people. Although more than 30 mosquito species are potential vectors, certain floodwater Aedes species that emerge from temporary ground pools following seasonal rains are recognized as the important enzootic vectors (McIntosh and Jupp, 1981). Epizootic transmission often occurs during periods of heavy rainfall in East Africa (Davies et al., 1985) when particular Culex species become infected. Transmission of RVF virus depends on complex interactions among mosquito vectors, non-human vertebrates, and human hosts, all of which are linked to various environmental factors (Wilson, 1994). Mosquitos perpetuate transmission either by direct infection of eggs or by acquiring infection during a blood meal and transmitting it during a subsequent feeding. Not only is species-specific mosquito abundance important, but densities of certain vertebrates, especially domestic ungulates, are crucial factors in zoonotic outbreaks. Such outbreaks induce protective immunity in all infected animals, thus reducing the number of susceptible animals after an epizootic. Human proximity to infected animals and mosquitos partly determines disease risk to people. Aedes mosquitos that infect animals rarely bite humans, so abundance of other mosquitos (notably Culex species) may be important for human disease.

Climate influences. Studies of climate variability and RVF activity have focused on precipitation and epizootics. Periods of excessive rainfall are believed to increase the egg hatching and larval survival of certain African Aedes floodwater mosquito species. Extensive transmission to humans, however, would seem to require the buildup of Culex populations. Recently, Linthicum et al. (1999) analyzed the historical pattern of eight recognized RVF outbreaks in East Africa since 1950 and claimed that each followed periods of abnormally high rainfall. Using Pacific and Indian ocean sea surface temperature anomalies and satellite-based vegetation indices, these authors suggest that outbreaks may be predicted up to five months in advance of their occurrence. Whether this association is strong and consistent enough to allow such forecasts remains to be seen.

Page 52

area of New Mexico, Arizona, Colorado, and Utah. The cluster of cases was caused by a newly recognized virus that was given the name Sin Nombre virus, a hantavirus (Nichol et al., 1993). The disease, hantavirus pulmonary syndrome (HPS), was soon recognized over a wide area of North and South America.

Patients with HPS typically exhibit fever and muscle aches lasting three to five days; shortness of breath and coughing symptoms then develop rapidly, requiring hospitalization and ventilation. Since first identified in May 1993, 260 cases of HPS have been reported in the United States, with the peak, 80 cases, occurring in 1993. The Sin Nombre virus is a rodent-borne pathogen belonging to the bunyavirus family of RNA viruses. Deer mice (Peromyscus maniculatus) are the most common reservoir in the southwestern United States, and they shed virus in their urine, droppings, and saliva (Childs et al., 1994). The virus is mainly transmitted to people when they breathe in contaminated air. Molecular tests of archived postmortem tissues from humans and rodents revealed that HPS was not a new disease, but that it was only newly recognized as a result of this unusual cluster of cases in 1993.

Climate influences. Data from an ongoing study of Peromyscus rodents revealed that in 1993 populations of this animal in some parts of New Mexico were as much as 10 times greater than average (Parmenter et al., 1993). The hypothesis was advanced that the unusually heavy rainfalls occurring during the 1992-1993 El Niño led to an abundant food supply for rodents, followed by a rodent population explosion. Deer mice readily entered homes and farm buildings, which resulted in greater exposure of humans to hantavirus-infected rodent excreta.

To test this hypothesis, investigators used satellite imagery and precipitation estimates to identify environmental conditions that were associated with the sites where HPS cases occurred (Glass et al., 2000). The clustering of cases was found to be associated with areas of heavy vegetation, but the data failed to fully support the connections between El Niño and HPS risk as proposed above. Also, at this point such associations may be confounded by the fact that more people have begun taking actions (such as rodent-proofing their houses) to mitigate risk of exposure to the virus.

Page 53

and is found at lower frequencies in the southeastern states and on parts of the West Coast, especially northwestern California. Lyme disease is also found in parts of Europe and eastward across Asia (Dennis, 1998). The annual number of cases in the United States per year has increased about 25-fold since national surveillance for the disease began in 1982. A mean of approximately 12,500 cases was reported annually to the Centers for Disease Control and Prevention from 1993 to 1997.

In all areas where Lyme disease occurs, ticks of the genus Ixodes appear to be the vector responsible for most transmission. The predominant vector in the eastern United States is the deer tick (Ixodes scapularis) and in the western United States the western black-legged tick (Ixodes pacificus). The tick has three motile stages (larvae, nymphs, adults), and each of these stages must find, attach to, and feed on the blood of a vertebrate to eventually molt to the next stage or, in the case of adult females, to reproduce. Mated and blood-fed female ticks lay perhaps 2,000 to 3,000 eggs, which hatch into larvae and begin another generation. Adult ticks in North America feed predominantly on large vertebrates, particularly white-tailed deer (Odocoileus virginianus). Abundance of deer has been linked to abundance of ticks in infested areas. The principal reservoir of the spirochete, however, is the white-footed mouse (Peromyscus leucopus). Larval or nymphal ticks become infected while feeding on an infected mouse and are then capable of transmitting infection to another vertebrate during a subsequent blood meal. Occasionally, a tick will bite and feed on a human, sometimes resulting in Lyme disease (Dennis, 1998; Wilson, 1998).

Climate Influences. Many factors appear to influence the spatial and temporal patterns of Lyme disease risk. Transmission among natural hosts and humans is determined by local abundance and survival of ticks, the percentage of ticks that are infected, abundance of hosts, human activity in tick habitats, and people's knowledge and awareness of Lyme disease prevention. A few studies have demonstrated how micro-climate influences the survival and activity of deer ticks (e.g., Bertrand and Wilson, 1996), suggesting that climate patterns may be important in determining Lyme disease risk; but research thus far has not yet fully elucidated the contributions of climate and other factors to transmission dynamics.

Climate change scenarios often forecast that some regions of the United States may become warmer and moister, leading to speculation that the range of deer ticks carrying Lyme disease might expand. However, the current distribution of the deer tick and of Lyme disease in the Unites States spans a wide range of climatic conditions, and deer ticks are already abundant in parts of the country where cold extremes are common. Extremely cold regions of North America and Northern Europe could perhaps support survival of local Ixodes ticks if climate warming were to occur, but whether this would translate to new cases of Lyme disease is highly speculative. If the range of ticks were to expand, this

Page 54

could set the stage for expansion of disease, but many other factors that contribute to human risk would also have to develop (e.g., competent reservoirs, human behavior, seasonal activity patterns).

Although climate is important, various other factors appear to be primarily responsible for risk of this vector-borne disease. The range of factors that presently limit distribution of the vector remain poorly understood, but research suggests that microclimate, abundant hosts, and suitable vegetation and soil habitat are important. The tick has not yet become established or widespread in apparently appropriate environments in many areas of the United States, thus indicating that even where suitable microclimate, host, pathogen, and human contact conditions appear to exist, Lyme disease may not be present.

Page 55

large numbers of people, and sometimes an entire population, are vulnerable to infection because they have no antibodies that recognize the virus. Influenza B viruses occur almost exclusively in humans, whereas all human influenza A viruses infect avian species, and a few subtypes infect other animals, particularly pigs and horses. There is evidence that the three pandemic influenza strains of the twentieth century arose from the incorporation of genetic material from animal influenza viruses (Webster et al., 1992).

Climate influences. Influenza has a clear seasonal cycle, occurring in North America mainly in late fall, winter, and early spring. The peak number of reported cases averaged over five seasons from 1994 to 1999 occurred in mid-December through February. It is reasonable to assume that the disease transmission cycle is influenced by climate, but the actual driving mechanisms are not well understood and have been the subject of few quantitative studies. Annual influenza outbreaks do not appear to correlate with mean winter or monthly temperature (Langford and Bentham, 1995). The interannual variability in the virulence of influenza strains makes interpretation of the relevant data difficult.

One common explanation for influenza's seasonal cycle is that there is more indoor crowding in the winter, which leads to greater disease transmission. Evidence for the effect of crowding includes the fact that flu epidemics tend to correlate with the start of school and peak during winter holidays, and that outbreaks occur frequently on cruise ships and other “contained” environments. However, laboratory studies of influenza transmission in mice show that, even under identical crowding conditions, flu transmission can still show a seasonal component. This may be due to the effects of humidity on the survival rate of the virus contained in the aerosolized droplets spread by coughing and sneezing (Schulman and Kilbourne, 1963).

Flu is often regarded as a “high-latitude” disease, yet it does occur every year in the tropics. Recent outbreaks occurring on cruise ships have shown that flu can be introduced from the Southern Hemisphere (Australia) and lead to outbreaks even in the middle of the summer (Miller et al., 2000b). More research is needed to gain a better understanding of the basis of the climatic influences on influenza. Particularly useful would be additional animal model studies in which dose and environmental conditions can be controlled.

There are many ways that global warming could conceivably impact influenza transmission. For instance, warming may change bird migration patterns and thus patterns of interaction between humans and infected animals; if warmer weather reduces indoor crowding, this could reduce virus transmission; higher relative humidity and ultraviolet flux could impair virus survival and slow the spread of disease. Other factors that may have greater influence on future transmission patterns include changes in population density, urbanization, and increased air travel.

Page 56

Page 57

suggested that waterborne transmission of Cryptosporidium was related to the seasonality of the cases associated with rainfall (Wuhib et al., 1994).

Waterborne disease due to any fecal-oral agent such as Cryptosporidium is not only influenced by climate. The incidence of infection in the animal or human population and the type of animal waste handling and sewage treatment will influence the likelihood of oocysts ending up in the environment. The size and hydrology of the watershed and the type and reliability of drinking water treatment will influence the impact on the drinking water. Thus, human, infrastructure, and engineering factors all play important roles in the possibility of waterborne disease.

Page 58

between sea surface temperature in the Bay of Bengal and cholera cases reported in Bangladesh (Colwell, 1996). Sea surface height was also found to be associated with cholera outbreaks and may be an indicator of incursion of plankton-laden water inland (Lobitz et al., 2000). More recently, Pascual et al. (2000) carried out a time-series analysis of an 18-year cholera record from Bangladesh and found a significant association between ENSO and the interannual variability of cholera, likely mediated by regional climate variables such as temperature. The confirmation that V. cholerae occurs in aquatic environments in association with zooplankton and phytoplankton, and the associations found between cholera cases and sea surface temperature and sea surface height, combine to provide evidence that some cholera epidemics are indeed influenced by climate.

Other vibrios that cause human disease also have growth characteristics dependent on climate. V. vulnificus, acquired by eating uncooked or undercooked shellfish, causes primary septicemia and gastroenteritis. V. vulnificus proliferates at warmer temperatures and thus could be influenced by climate-induced increases in water temperature. O'Neill et al. (1992) showed that in a New England estuary subject to extreme seasonal temperature fluctuations, there was a strong correlation between levels of V. vulnificus recovered from oysters and water temperature and salinity. Motes et al. (1998) investigated the temperature and salinity parameters of waters associated with oysters linked to V. vulnificus infections and found that abundance of this pathogen is directly related to water temperature.

Vibrio parahaemolyticus, the second most common vibrio disease in humans after V. cholerae, is a frequent food-borne pathogen in Japan. It typically causes acute gastroenteritis when associated with eating uncooked or undercooked shellfish. V. parahaemolyticus is the first known example of a human pathogenic bacterium whose growth fluctuates with environmental temperatures. In 1973, Kaneko and Colwell (1973) reported that V. parahaemolyticus overwinters in Chesapeake Bay sediment and enters the water column only when temperatures exceed 14°C. They also observed that V. parahaemolyticus abundance was proportional to the concentration of zooplankton in the water column, especially copepods. Watkins and Cabelli (1985) observed a similar relationship in Narragansett Bay, Rhode Island. They determined that nutrients associated with waste water stimulated phytoplankton. The phytoplankton supported larger numbers of grazing zooplankton with a resultant increase in the abundance of the associated V. parahaemolyticus. From an ecological perspective, it is probable that V. parahaemolyticus behaves in a manner similar to that described for V. cholerae because both are estuarine bacteria that associate with copepods, both have reasonably strict temperature and salinity requirements, and both can accidentally enter human hosts and cause disease. Hence, it is likely that climate fluctuations conducive to their growth (warmer temperatures and intermediate salinities) lead to increased incidence of disease.