Click for next page ( 13


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 12
CHAPTER 2 MEASURING EFFECTS ON HUMAN HEALTH FROM THE SUBTHERAPEUTIC USE OF ANTIMICROBIALS IN ANIMAL FEEDS A complex chain of events begins with the addition of sub- therapeutic levels of antimicrobials to animal feed, proceeds to the selection of bacteria bearing R plasmids in the animal gut and the transfer of these bacteria to humans (or their plasmids into the flora of humans), and ends with possibly adverse effects on human health. The entire chain can be thought of as a stochastic process. All steps of the chain are apparently possible, but have not been quantified. For example, the extent of the normal transfer of R+ organisms between animals or between animals and humans has not been adequately measured. The entire length of this chain is germane to this study, i.e., how does a change in the addition of antimicrobials to animal feed at the beginning of the chain alter the probability or severity of diseases or the ability to treat them at the end of the chain? A number of complications can arise when attempting to re- solve this question. This is illustrated by the following series of questions that should be addressed when treating a hospitalized patient suffering from septicemia due to Salmonella. 1. Is the Salmonella strain resistant to antimicrobials? 2. Is the resistance plasmid mediated? 3. Was the Salmonella resistant to drugs when the infec- tion occurred or did it acquire resistance through transfer of an R factor from the patientts resident microbial flora? 4. Did the patient acquire the Salmonella infection from contaminated meat or from another person? 5. If the infection was acquired from meat, had the animal received an antimicrobial? Was the antimicrobial given to the animal for growth promotion, for prophylaxis, or for treatment of an illness? 12

OCR for page 12
13 7. If the Salmonella acquired resistance by it-factor transfer from E. colt, was the resistant E. cold selected by an antimicrobial used in previous treat- ment or did the patient acquire a resistant E. cold strain from another person or from ingestion of con- taminated meat? If the resistant strain was acquired from another person, was that person infected via the food chain and did the food chain begin with an animal that had received subtherapeutic doses of antimicrobials? Determination of the answers to all of these questions for an individual patient is obviously very difficult and may well be impossible. Antimicrobial use varies greatly among producers of cattle, swine, and poultry, within each of these subsets of the industry, and among different parts of the country. Moreover, the distribu- tion channels are long and complex. Consequently, a consumer may purchase meat that was processed in a distant plant from animals raised in a still more distant place. The beef consumed by a household may have come from a feedlot in which antimicrobials were used only to treat sick animals, but its pork may have come from pigs fed tetracycline for growth promo- tion. Even if Salmonella may be judged more likely to have come from one source than from another, it may have acquired resistance by plasmid transfer from E. cold derived from either source. EPIDEMIOLOGICAL STUDIES The definitive epidemiological study of effects related to the subtherapeutic use of antimicrobials in animal feeds should encompass all aspects of meat production from animal breeding to consumption. Methods The committee examined the approaches that could be taken to relate the subtherapeutic use of antimicrobials in animal feed to the risk of increased morbidity and mortality in humans. It then grouped these approaches into six categoriese Cross-Sectional Studies (Population or Prevalence Sur- veys). Total communities, random samples of a total community, or selected subsets of a population may be identified and asked

OCR for page 12
14 to provide answers to questions, undergo laboratory or clinical procedures, or both. Such surveys are conducted once, and the findings are customarily presented as prevalence ratios of a disease or condition between populations. Differences in the prevalence of the disease or condition in demographic subgroups may be interpreted as reflecting differences in risk caused by the exposure under study. Specifically-, populations that have had considerable exposure and others with minimal exposure to animals receiving antimicrobials via their feed could be surveyed to determine the prevalence of some+specific conditions, e.g., urinary tract infection caused by R organisms. The results of such surveys may be confounded by differences in prevalence of some associated causal or risk factors, such as socioeconomic sta- tus. For instance, elevated rates of diarrhea! disease in farm workers handling feed containing antimicrobials, when compared to rates in nonfarm groups, may be due as much to differences in sanitation in the work and home environments as to the acquisition of resistant pathogens from animals. Surveys are conducted frequently because they are relatively inexpensive and generally quite rapid. Survey data are generally unreliable in identifying the differences between statistical asso- ciation and cause and effect, partly because the effects of time are difficult to incorporate into a cross-sectional survey. This weakness presents a particular disadvantage in any situation that changes over time. Case-Comparison Studies. The characteristics of patients with a disease or other condition may be compared with those of a group of people who are free of the disease or condition, but whose other characteristics, such as age, sex, race, and history of exposure to therapeutic levels of ant~microbials, are similar. The comparison subjects are often selected from the same clinical setting in which the cases were discovered, but may be selected from groups living in the same neighborhood or perhaps by a random sampling of the general community. The characteristics of both groups and the exposure to which they have been subjected are sur- veyed retrospectively. For instance, patients with a disease be- lieved to be attributable to resistant microorganisms can be matched with comparison subjects. Exposure to animals, to animal products, or to feed containing antimicrobial agents is the deter- mined retrospectively. The efficiency of such a study may be im- proved by careful selection of comparison subjects to match the characteristics of the cases. The relative risk associated with certain characteristics, such as degrees of exposure to antimicro- bials in animal feed, may be estimated by calculating odds ratios. Such studies may be relatively inexpensive and can be conducted

OCR for page 12
15 rapidly, but they require careful predetermination of the char- acteristics to be matched and investigated retrospectively. If important variables are missed or mismatching occurs, results may be interpreted incorrectly. Cohort Studies. Variables related to risk of a population or a selected subset of a population may be identified and char- acterized, and the study group followed over time to observe new cases of illness. For instance, generally healthy individuals with high exposure to animal feeds containing antimicrobials, or to meat from animals that consumed such feed, and other members of the same community without such exposure may be followed for a year or longer. Populations with different exposures can be examined for differences in incidence or severity of diseases or for difficulties encountered during their treatment. The differences in these measurements will reflect the risk associated with the different exposures. Cohort studies are generally considered to be the most effec- tive approach to establishing differences in risk associated with various conditions, but they are expensive and time-consuming. Moreover, they require careful recordkeeping, meticulous followup of the cohort, and attentiveness to ensure that differences in medical care or inaccurate definitions of morbid events do not confound the study. Experimental Studies. After a defined study population is - identified and characterized, an appropriate subgroup is subjected to an experimental procedure. The test group is then followed to observe the effects of the procedure. The remainder of the initial group serves as a control. Experimental studies, particularly when the exposed group is randomly selected, are very effective in estab- lishing the cause-and-effect relationship between the experimental variable and the observed outcome. Thus, if it were possible to select a large number of subjects and randomly assign them to groups that are or are not exposed to animals fed antimicrobials or to the products of those animals, useful information would undoubtedly accrue. The performance of experimental studies with humans may be questioned on ethical grounds if there is any known or theorized hazard of the exposure being investigated. Moreover, such con- trolled trials are frequently very difficult to conduct because the subjects may not adhere to the regimen to which they have been randomly assigned. Experimental studies cannot yield definitive information on the likelihood that similar effects would result at similar frequencies under natural conditions.

OCR for page 12
16 Investigation of Epidemics. Outbreaks of a disease may be investigated retrospectively with epidemiological methods. Such investigations cannot quantitate the prevalence of diseases since only reported cases can be investigated. Nevertheless, they can be useful in establishing a probable chain of transmission. Case Reports/Case Series. Although not an epidemiological method, reports of individual cases or series of cases may also establish that a particular sequence or chain of events is possi- ble. They cannot be used to predict the prevalence of a disease or condition or the risk of any practice. WHAT IS THE IDEAL STUDY? The ideal study would start with the selection of animal- rearing facilities marked by sharply different practices in antimicrobial use--i.e., no use; subtherapeutic use in feeds for improving growth and feed conversion and for prophylaxis; intermittent therapeutic use; and both subtherapeutic and therapeutic use. A substantial number of facilities in each selected category should be studied to ensure that any observed differences in flora are more likely to be related to differences in antimicrobial use rather than to other confounding variables. Fecal specimens should be examined from a large enough s~m- ple of animals in each flock or herd to monitor changes in the proportion of animals carrying Salmonella spp. and E. cold with resistance and the percentage of these resistant organisms in each animal's flora. Ideally, full microbiological characteriza- tion of the flora, both aerobic and anaerobic, should be performed for both animals and humans. Special attention should be given to changes in those indices associated with intermittent therapeutic applications of antimicrobials. Similar studies should be conducted of employees, their fam- ilies, and a comparison group of neighbors employed in other occupations to determine whether their microflora follow patterns like those observed in animals. At the next several steps in the production chain--the slaught- erhouses and wholesale and retail butcher shops--bacteriological studies of carcasses, the work environments, employees, their fami- lies, and their neighbors should be pursued over time to determine whether changes in the indices of drug resistance occurring in the production facilities can be traced through the processing plants. The processing plants should, of course, receive meats from only one of the specified categories of animal producers.

OCR for page 12
17 Finally, these same investigative techniques should be used in comparable communities where the meats and meat products are sold, prepared, and consumed, still separated by the type of anti- microbial use in animal feeds at their sources. Selecting c~mpar- able communities may be difficult since many factors need to be matched, e.g., the degree of hygiene in butcher shops. Samples of households should be selected, and the enteric bacteria of the family members should be characterized. The sampling would permit community-wide estimates of the prevalence of both resistant and susceptible Salmonella infection, as well as the prevalence of colonization with resistant E. colt. A system of defining, identi- fying, and recording illnesses should be established in order to compute the rates of disease caused by Salmonella and to determine the special problems that result from illnesses attributable to resistant strains. Attempts should be made to trace the spread of Salmonella infection in these communities, identifying wherever possible cases attributable to contact with or consumption of contaminated meat and those due to secondary spread from person to person. The nature of antimicrobial resistance in other pathogens causing illness in these communities should also be investigated to ascertain the extent to which R factors arising from therapeu- tic and subtherapeutic antimicrobial use in animals are transferred between microbial species and constitute a health problem. A characterization of the R plasmids, which can be achieved by physical, genetic, or enzyme techniques (Jacoby and Low, Appendix C; O'Brien, Appendix I), could provide corroborative evidence of the direction of transmission and information on qualitative changes in resistance. If the inquiries conducted on the breeding farms and the feedlots reveal no differences in the prevalence of resistant organisms in animals related to the differing use of antimicrobials, the study could be terminated. If such differences are apparent, further work would be needed. The second phase of the study would determine the occupationally associated risk of acquisition of resistant organisms and provide some information on the likelihood of spread from these foci. The carrier rates in samples from the neighborhoods of the abattoir workers should provide data on the prevalence of resistant enteric bacteria in general communities, as would the information derived from the household samples in the third phase. The community-based studies would relate the carrier rates in a community to the use of antimicrobials in animals and measure the associated burden of disease (see Table 2-1~.

OCR for page 12
18 TABLE 2-1 The Ideal Comprehensive Study Design PHASE 1: BREEDING FARMS AND FEEDLOTS Study Groups Herds and flocks in four cate- gories of antimicrobial use: no antimicrobial use subtherapeutic use only therapeutic use only both subtherapeutic and therapeutic use Animal handlers Family members Neighbors Study Procedures Bacteriological indices: Prevalence of carriage of anti- microbial-resistant Salmonella spp./relative frequency of resistant organisms in each specimen Prevalence of carriage of antimi- crobial-resistant E. colt/ relative frequency of resistant . . . organisms In eac ~ specimen Characterization of fecal flora and plasmids PHASE 2: SLAUGHTERHOUSES, PROCESSING PLANTS, AND RETAIL BUTCHER SHOPS ~ _ . Study Groups Carcasses Meat handlers Family members Neighbors PHASE 3: COMMUNITIES Study Groups - Households selected on the basis of probability sample Cases of salmonellosis Other cases of infectious diseases caused by bacteria Study Procedures Bacteriological indices (as above) Human illness caused by bacteria Study Procedures Bacteriological indices (as above) Tracing of source of infection

OCR for page 12
19 FEASIBILITY OF A COMPREHENSIVE STUDY The committee reviewed methods for gathering information on the effects of various exposures on subsequent morbidity and mortality and for determining the specific needs for information on the subtherapeutic use of antimicrobials in animal feeds. It concluded that the comprehensive study described above could not be realized or even approximated. This decision reflects a number of facts, which are discussed below: The use of antimicrobials differs markedly for the three major meat animals--cattle, swine, and poultry. Moreover, for each type of animal the use of antimicrobials varies in various parts of the country and at different times, e.g., with seasons or weather con- ditions. Moreover, it is often not possible to differentiate whether the antimicrobials had been given for growth promotion, prophylaxis, or treatment of manifest illness. In much of the industry, only insignif icant numbers of animals have never received any antimicro- bials, and in normal rearing and processing operations, it is not practical to identify these animals. During shipment from breeding farms to feedlots, groups of animals with different exposures to antimicrobials are often combined. The likely exchange of bacteria between animals under these circumstances further hinders the identi- fication of the origin of any observed R+ organisms. The processing of meats and meat products also contributes to the mixing of meats from animals with different antimicrobial his- tories. Cross-contamination can also occur in these processes, e.g., via cutting boards and instruments. Currently, there is no hope of identifying communities in which residents can purchase only the meat of animals that had been exposed to only one regimen of antimicrobial usage. The diseases and conditions likely to result from resistant microorganisms in the general population exposed to meat are rela- tively rare. Overt diarrhea caused by Salmonella, acute urinary tract infections in young women, and other illnesses related to infections with enteric pathogens all have extremely low incidence rates. Thus, any study attempting to relate an increase (po+sibly a small increase) in morbidity or mortality to exposure to R orga- nisms on meat (whether selected by subtherapeutic or therapeutic antimicrobial use) would be so massive that it would probably be unmanageable. Additionally, the chain of events linking antimi- crobial-resistant bacteria in the animal gut to overt human infec- tions is extraordinarily difficult to trace in any given case. This difficulty would increase the uncertainty in the study and jeopardize the validity of any risk estimates that might be developed.

OCR for page 12
20 ALTERNATIVES TO A COMPREHENSIVE STUDY AND THEIR DEFICIENCIES . Less comprehensive approaches, although more realistic, can- not provide direct evidence of a consistent chain of causation from subtherapeutic antimicrobial use in animal feeds to human illness. Moreover, the fragmentary data that are available (discussed below) not only suffer from deficiencies of method and design but also leave gaps that can be bridged only by conjecture or speculation. Better research may repair the former, but little can be done about the latter. Some deficiencies of the narrower studies are listed below: ~ Studies of the prevalence of antimicrobial-resistant bacteria in animals cannot be used to determine the extent of the transmission of R+ enteric organisms from animals to human popula- tions. ~ In studies of the prevalence of R+ organisms in farm workers, the R+ bacteria acquired directly from animals cannot easily be distinguished from those R+ organisms resulting from the selection pressure exerted by ingested or inhaled antimicrobials from the feeds. Studies of the prevalence of resistant organisms on car- casses or in abattoir workers do not provide direct measurements of the extent to which these organisms are transmitted to the general population, nor can they distinguish the effects of subtherapeutic use from the therapeutic use of antimicrobials. Studies comparing the prevalence of R+ organisms in meat- eaters and vegetarians cannot differentiate whether observed differ- ences in the meat-eaters can be attributed to the selective pressure for resistance exerted by the subtherapeutic use or by the therapeutic use of antimicrobials in the animals consumed. Such studies would also need to take into account the use of antimicrobials in the sub- jects themselves or in persons in their immediate environment. Studies on the prevalence rates of R+ organisms in different populations can only imply the causes for differences in rates. They do not relate R+ prevalence rates to increases in morbidity, mortality, or complications in therapy caused by resistance in the pathogen. The foregoing discussion makes it clear that isolated studies on parts of the transmission chain cannot be used to quantitate the overall effects on human health resulting from the subtherapeutic use of antimicrobials in animal feeds.

OCR for page 12
21 This committee therefore concludes that it is not possible to conduct a feasible, comprehensive direct study of the effects on human health arising from the subtherapeutic use of antimicrobials in animal feeds. However, after examining some of the issues and the research conducted to date, the committee outlined several studies that might quantitate some of the stages in the chain of causation on which speculation of hazard is based. The studies are presented as an indication of what the com- mittee believes to be the most fruitful approaches. They will not provide a direct assessment of the effects on human health resulting from subtherapeutic levels of antimicrobials. Chapter 4 contains descriptions of these studies and some caveats on the interpretation of their results.