Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
AVIAN DISEASES 50:161–172, 2006 Invited Minireview— Potential Impacts of Antibiotic Use in Poultry Production Randall S. SingerAB and Charles L. HofacreC A Department of Veterinary and Biomedical Sciences, College of Veterinary Medicine, University of Minnesota, 300A VSB, 1971 Commonwealth Avenue, St. Paul, MN 55108 B Division of Environmental Health Sciences, School of Public Health, University of Minnesota, Minneapolis, MN 55455 C Department of Population Health, College of Veterinary Medicine, University of Georgia, Athens, GA 30602-4875 Received 31 March 2006; Accepted 7 April 2006 SUMMARY. The ways in which antibiotics are used in poultry production have changed considerably during the past decade, mainly because of concerns about potential negative human health consequences caused by these uses. Human health improvements directly attributable to these antibiotic-use changes are difficult to demonstrate. Given that some antibiotics will continue to be used in the poultry industry, methods are needed for estimating the causal relationship between these antibiotic uses and actual animal and human health impacts. This is a challenging task because of the numerous factors that are able to select for the emergence, dissemination, and persistence of antibiotic resistance. Managing the potential impacts of antibiotic use in poultry requires more than a simple estimation of the risks that can be attributed to the use of antibiotics in poultry. Risk models and empirical studies that evaluate interventions that are capable of minimizing the negative consequences associated with specific antibiotic uses are desperately needed. RESUMEN. Estudio Recapitulativo por Invitación—Impacto potencial del uso de antibióticos en la producción avı́cola. La forma como los antibióticos son utilizados en la producción avı́cola ha cambiado considerablemente durante la década pasada, principalmente debido a las preocupaciones sobre posibles consecuencias negativas en la salud humana causadas por la utilización de los mismos. Los progresos en la salud humana atribuibles directamente a estos cambios en la utilización de antibióticos son difı́ciles de demostrar. Sabiendo que algunos antibióticos continuarán siendo utilizados en la industria avı́cola, se requieren métodos para estimar la relación causal entre los cambios en la utilización de antibióticos y el impacto real en la salud tanto humana como animal. Esta es una tarea difı́cil debido al gran número de factores de selección que influyen en la aparición, diseminación y persistencia de la resistencia a los antibióticos. El manejo de los impactos potenciales de la utilización de antibióticos en la avicultura requiere más de un simple cálculo de los riesgos atribuibles al uso de los antibióticos en las aves domésticas. Existe una necesidad inmediata para desarrollar modelos de riesgo y estudios empı́ricos que evalúen estrategias capaces de minimizar las consecuencias negativas asociadas con la utilización de antibióticos especı́ficos. Key words: antibiotic resistance, antibiotic use, causal inference, public health, risk assessment, unintended consequences Abbreviations: APEC ¼ avian pathogenic Escherichia coli; FDA-CVM ¼ Food and Drug Administration, Center for Veterinary Medicine; GRE ¼ glycopeptide-resistant enterococci; QAC ¼ quaternary ammonium compound; QD ¼ quinupristin–dalfopristin; VRE ¼ vancomycin-resistant enterococci Over the past decade, the ways in which antibiotics are used in poultry production have changed. Antibiotic use in poultry is changing globally, although perhaps more dramatically in the European Union than in the United States (15). One reason for this shift is the removal of specific antibiotic uses, as mandated by regulatory agencies, such as the recent removal of fluoroquinolones from poultry production in the United States and the removal of growth-promotant use in many European countries (28,124). Another reason is the voluntary reduction in growth-promotant uses of antibiotics by some of the major poultry production companies and restaurant chains (30). The major driving force influencing these recent changes in antibiotic usage in poultry has been the perceived public health risks associated with this practice (88). Unfortunately, peer-reviewed documentation of these reductions, particularly in the United States, is difficult to find. Agricultural antibiotic usage data have been requested by varying groups but have not been easy to obtain (118,120), which has led some groups to generate estimates based on crude calculations (55). The notion that antibiotic uses in agriculture, and specifically in poultry, may pose a risk to public health and animal health is not new (121). Strains of Escherichia coli displaying resistance to multiple antimicrobial agents have been recovered from the fecal flora of broiler chickens since the earliest reports in the 1960s and 1970s (17,48,72,83,107,108,116). One of the first studies to demonstrate resistance among E. coli isolates from poultry was published in 1961 (111). The authors noticed that from 1957 to 1960, the practice of feeding broiler fowls on diets containing tetracyclines increased greatly and progressively in Great Britain. They further recorded the increasing incidence of tetracycline resistance among pathogenic avian E. coli isolates during this same time period from 3.5% in 1957, to 20.5% in 1958, 40.9% in 1959, and 63.2% in 1960 (111). A later study reported by Smith in 1966 examined antimicrobial resistance among E. coli isolates obtained from cases of human neonatal diarrhea, swine diarrhea, calf scours, and avian colisepticemia (107). Among the avian E. coli isolates tested, resistance was most often observed to sulfonamides (17.1%), tetracyclines (17.1%), and to a lesser extent, streptomycin (2.9%). Twenty-nine percent of avian isolates demonstrated resistance to one or more drugs with the most common multidrug-resistant patterns being streptomycin, sulfonamides, and tetracycline, or sulfonamides and tetracycline (107). The majority of publications from this time period echoed the same conclusions: that the observed antimicrobial-resistance phenotypes reflected the wide use of antibiotics in the livestock industry, and that these resistance phenotypes were primarily due to the presence of resistance plasmids (initially referred to as R factors) 161 R. S. Singer and C. L. Hofacre 162 Table 1. Therapeutic antibiotics used in poultry productionA (19). Egg layer chickens Antibiotic Feed Drinking water Bacitracin Bambermycin Chlortetracycline Erythromycin EnrofloxacinB Lincomycin Neomycin Novobiocin Oxytetracycline Penicillin Streptomycin Sulfonamides Tylosin Virginiamycin Yes No Yes No No No No No No No No No Yes No Yes No No No No No No No No No No No No No Broiler chickens Growth promotant No No No No No No No No No No No No Yes No Turkeys Feed Drinking water Growth promotant Yes Yes Yes Yes No Yes No Yes Yes Yes No Yes Yes No Yes Yes Yes Yes Yes Yes Yes No Yes No Yes Yes No No Yes Yes Yes No No Yes No No Yes Yes No No Yes Yes Feed Drinking water Growth promotant Yes Yes Yes Yes No No No Yes Yes Yes No Yes No No Yes Yes Yes Yes Yes Yes Yes No Yes Yes No Yes No No Yes Yes Yes No No Yes No No Yes Yes No No No Yes A B Gentamicin and ceftiofur are labeled for day-old chickens and turkeys by subcutaneous injection only. Enrofloxacin is no longer approved for use in poultry production in the United States. (48,79,83,107,116,122). These experiments increased both the scientific and public scrutiny of the administration of therapeutic and subtherapeutic doses of antimicrobials to animals and ultimately led to the release of the Swann report in the United Kingdom in 1969, which recommended that antibiotics used to treat infections in humans not be used as animal-feed additives (115). Despite implementing these recommendations, the United Kingdom and other countries have observed mixed results with regard to reducing antimicrobial-resistance phenotypes among pathogenic E. coli isolates. In fact, a survey conducted in 1980, 9 yr after the banning of the use of tetracyclines as feed additives in the United Kingdom, indicated that chickens and pigs were still a reservoir of tetracyclineresistant E. coli (109). When the potential public health impacts of antibiotic use in poultry are considered, antibiotic resistance is the major perceived risk. Various approaches have been used to document and estimate these potential risks and to relate resistance levels to actual human health impacts. The goal of this review is to highlight the ways in which antibiotics are used in poultry production, the ways in which these uses can impact public health, and the data gaps that exist in estimating this relationship. Hopefully, this brief review will identify key areas to which researchers, poultry producers, and the pharmaceutical industry can supply useful data. Finally, the review has the added goal of separating causality from association as we interpret the data collected from research projects, industry reports, and surveillance systems. ANTIBIOTIC USES IN POULTRY Antibiotic use in commercial poultry can be divided into two use categories: therapeutic antibiotics (Table 1) and growth-promotant antibiotics (20). Antibiotics used in feed for growth promotion and disease prevention purposes are administered at levels that are lower than antibiotics given for the treatment of disease, and therefore, these uses are termed subtherapeutic. These in-feed antibiotics are capable of exerting a selection pressure on bacterial populations and altering various microbial communities (74). Uses of antibiotics in either category have the potential to select for bacterial strains that are resistant to the antibiotic used while still having a positive impact on the health of the animal population. The antibiotics in the growth-promotant category are, in some cases, the same antibiotics used in the therapeutic category; the amount of the antibiotic administered to the birds is usually less for growth promotion. The growth-promotant antibiotics are administered only in the birds’ feed (55). The decision to use growthpromotant antibiotics in commercial poultry is primarily based on economic factors, i.e., whether the improvements in body weight, feed efficiency, and/or growth rates are worth the cost of the antibiotic (113). It should be noted that because there are limited therapeutic antibiotics approved for poultry, these antibiotics are rarely used by the commercial poultry producers as growth promotants (50). In the United States and many other countries, consumers of poultry products are increasingly requesting that poultry suppliers sell meat that has not been raised with growth-promotant antibiotics. This has resulted in the need for nonantibiotic alternatives, such as probiotics, intestinal acidifers, natural antibacterials, and enzymes to reduce the growth of unfavorable intestinal bacteria, such as Clostridium perfringens (32,51). These applications, in combination with management changes, must also improve animal health and promote size uniformity among animals in the herd or flock as the growth-promotant uses have done (15). Without these effects, the removal of the growth promotants could lead to the unintended consequence of increased levels of pathogen contamination in meat, as will be discussed later. Therapeutic antibiotics are used in those cases in which a disease has been introduced onto a farm. The poultry veterinarian must decide whether the birds can be treated with an antibiotic, and if so, which antibiotic and by what route of administration (in feed or in drinking water). The choice of antibiotic is often limited by the balance between the cost of the drug and the level of disease. In addition, antibiotic decisions are influenced by the regulated duration of withdrawal before slaughter to ensure that no drug residue is found in the meat. The vast majority of therapeutic treatments are given via the drinking water because sick birds may stop eating but often continue to consume water. In layer chickens, the antibiotics available to poultry veterinarians for the treatment of bacterial infections are limited (Table 1). Escherichia coli is the leading cause of economic loss due to disease for the poultry industry throughout the world (9). Therefore, it is the primary bacteria that therapeutic antibiotics are used to treat. In Antibiotic use in poultry most instances, these E. coli infections are secondary infections following a primary viral or environmental insult (39). This can mean that therapeutic antibiotics in commercial poultry are often used to relieve the suffering of the sick birds and to minimize the financial impact of the disease on bird performance until the primary insult can be identified and eliminated. Another reason for the use of therapeutic antibiotics is related to the potential increased public health risk associated with slaughtering birds from farms with sick birds. Poultry that are sick eat greater amounts of bedding material (litter), which can result in higher rates of Salmonella spp. and Campylobacter spp. in their intestinal tract (21). Also, one small study found that birds from flocks having higher airsacculitis condemnation rates had higher levels of E. coli and Campylobacter spp. in the processed meat (91). The therapeutic antibiotics available to treat the respiratory infections caused by E. coli are limited (40). The value of the individual animal is so low that it is cost prohibitive to individually dose each bird in a house, thus, eliminating the option of aminoglycosides and cephalosporins. Because sick birds continue to drink in a disease situation, therapeutic antibiotics labeled for use in the drinking water are most often used. Bacitracin, lincomycin, and neomycin are poorly absorbed from the intestine; therefore, tetracyclines, fluoroquinolones, and sulfonamides are the only drugs available to treat E. coli airsacculitis. It can be speculated that this limited choice of antibiotics to treat E. coli infections has resulted in many years of selection pressure on E. coli in the commercial poultry environment, resulting in the high levels of sulfonamide and tetracycline resistance in clinical E. coli isolates seen in many diagnostic laboratories (127). It is not the purpose of this review to comment on the amount of antibiotic used in the poultry industry, on the feasibility of collecting those data, or on the best ways to report them. Different organizations have arrived at different estimates of the amounts of antibiotics used, and differences between estimates are due to different equations for estimating usage as well as differences over what constitutes an antibiotic (55,82). These estimates are generally reported in gross amount of antibiotic used but do not provide information about indications for use, route of administration, or other qualitative information (104). A recent report by the Alliance for the Prudent Use of Antibiotics, Facts about Antibiotics in Animals and Their Impact on Resistance, addressed the ways in which these data can be collected and used. A general conclusion of this report is that we need new ways of describing antibiotic uses in animal agriculture, such as the defined daily doses equivalent in human medicine. A topic that is in need of additional research is the utility of antibiotic-use data in relation to the animal and public health impacts of antibiotic usage in animal agriculture. With respect to the poultry industry, data can be collected at the local, regional, national, or international level. The pressures that select for antibiotic resistance in bacteria may operate at many proximate levels (within an animal, within a group of animals, or on the farm). They may also operate on a broader scale, and therefore, it is plausible that some aspects of the bacterial response to exposure to antimicrobials may only be apparent when analyzed at the ecological scale. Consequently, to make a general statement that a national increase in the annualized amount of antibiotic used in poultry represents an increased risk to human health is overly simplistic and potentially fraught with biases (104). Better analytical tools as well as more detailed descriptions of biological mechanisms are needed to understand if and how these antibiotic usage estimates will be useful in predicting changes in human health risks. 163 ANTIBIOTIC USE, ANTIBIOTIC RESISTANCE, AND HEALTH With the changes being made in antibiotic use practices in poultry, it would appear that many natural experiments have been designed to determine the effects that antibiotic use has on antibiotic resistance. The types of studies that have been conducted are of two primary types. First, some investigators look for resistant bacteria in flocks that have used antibiotics, and most of the time, comparisons are made between poultry operations that use different levels of antibiotics. Second, some investigators perform temporal studies in which specific operations are followed before, during, and after they implement changes in antibiotic use practices. Both of these study designs have been used under experimental conditions as well as in observational settings, where normal production operations are compared. In several classic studies of antibiotic use in chickens, antibiotic administration resulted in an increased proportion of bacteria resistant to the antibiotic in chickens administered tetracycline or tylosin (49,68,69,71). In one of these studies, farm workers also experienced increased levels of tetracycline-resistant bacteria following the administration of in-feed tetracycline to the chickens (68,69). In another of these studies, a survey of resistance in normal broiler flocks was included (71). Resistance was often detected in birds that had received no antibiotics, a finding that, at the time, seemed to be in conflict with the assumption that resistance was a consequence of the selection pressure created by antibiotic use. When specific antibiotics are used, especially as growth promotants, resistance to that antibiotic is commonly detected in bacterial populations from that operation. In a small study of virginiamycin use in broilers in the United States, resistance to streptogramins was common in the Enterococcus faecium isolates (123). Unfortunately, that study had no comparison group, such as farms that were not using virginiamycin. A prospective, controlled trial was conducted in which virginiamycin was fed to chickens. No effect was observed on resistance in E. faecium or on weight gain in the birds (31). The study was very small, and the lack of an observed effect could be attributed to the small sample size (one trial with two groups of four chickens per group). In a larger study, multiple feeding trials with virginiamycin were conducted (80). The groups receiving virginiamycin in the feed had a higher prevalence of streptogramin-resistant E. faecium than the control groups, but the removal of virginiamycin from the feed resulted in a disappearance of the streptogramin-resistant isolates. The authors concluded that continuous feeding of virginiamycin was required to maintain the resistance. In one study of broilers in their normal production environment, there was no evidence that tetracycline usage for therapeutic purposes altered the tetracycline-resistance levels in E. coli, Enterococcus spp., or Campylobacter spp. (35). In that investigation of two flock houses per farm on three farms, E. coli and Enterococcus spp. isolates already had high tetracycline resistance in flocks with and without historical tetracycline usage, and Campylobacter spp. isolates remained largely susceptible, even after tetracycline administration. Perhaps one of the largest temporal investigations conducted under natural conditions was performed in Denmark, where a national surveillance system has assessed antibiotic-resistance levels in bacteria of food-animal origin for many years. Within Denmark, E. coli and Enterococcus spp. were monitored over time. Avoparcin was banned as a growth promotant in Denmark and Norway in 1995 and in the rest of the European Union by the end of 1997. Avoparcin is a glycopeptide that can induce cross-resistance to vancomycin (126), and therefore, there is concern that avoparcin use 164 R. S. Singer and C. L. Hofacre in animal agriculture can impact human health by selecting for vancomycin resistance and by aiding in the dissemination and transfer of vancomycin-resistance genes (6). In some reports, the removal of avoparcin from broilers in Denmark was followed by a dramatic reduction in the percentage of E. faecium isolates resistant to vancomycin (vancomycin-resistant enterococci; VRE) (2), whereas, in other studies, the prevalence of VRE on broiler farms did not appear to change (46). A study in Denmark investigated the genetic diversity of VRE in broiler houses after the avoparcin ban (47). Based on high genetic similarity of isolates within a broiler house over time and high heterogeneity among houses, the authors concluded that the VRE may be persisting in the broiler house environment over time in the absence of selection pressure. A similar finding has been documented for E. coli in broiler houses (103). Additional studies have been conducted on the dynamics of antibiotic resistance following the removal of growth-promotant antibiotics in other parts of Europe. A report from Germany demonstrated a reduction in VRE prevalence in poultry meat and in the gut flora of humans in the community following the ban of avoparcin (61). The results following the ban of avoparcin in Norway have been more confusing. In a study conducted up to 8 yr after the ban of avoparcin, the prevalence of glycopeptide-resistant enterococci (GRE) had not declined significantly in either the fecal samples from broilers and turkeys or from poultry farmers (112). This finding has been partially attributed to the presence of a postsegregational killing system in the enterococci that maintains the antimicrobial resistance through a physical genetic linkage in the absence of selection pressure (112). The effects of antibiotic use on resistance are also evaluated by comparing farms that use antibiotics (conventional farms) to organic farms. The assumption is that differences in antibiotic-resistance levels between conventional and organic farms are due solely to the differences in antibiotic use, an assumption that will be further discussed later in this review. It is often assumed by the public that animals raised in an organic setting produce healthier and more wholesome meat products (63). One recent study compared VRE from conventional and organic broiler operations in the United Kingdom (37). Resistance was common in both types of operations, although the prevalence appeared to be higher on the conventional operations. Sampling was conducted in 2002 and 2003, years after the 1997 ban of avoparcin use in animal agriculture, and still resistance persisted. Finally, the potential impacts of antibiotic use in poultry have been assessed by analyzing bacterial populations on retail meat samples. The assumption in this type of study is that the bacteria on the meat reflect the organisms derived from the farm and that the resistance in these isolates is related to antibiotic use practices on the farm. These studies often include a comparison of conventional to organically reared chickens and, therefore, make the same assumption previously described. Some studies have shown that organic chicken can have higher rates of disease and higher rates of Campylobacter contamination at the time of slaughter than conventionally reared chicken (45), although resistance levels can be low in both types of operations. In one study conducted in Maryland retail stores, organic chicken had a higher prevalence of Salmonella and Campylobacter contamination than conventional chicken, but the overall resistance level in those isolates was higher in conventional product (26). Other studies have reported that the prevalence of Campylobacter contamination was not higher in the organic product but that the prevalence of fluoroquinolone-resistant Campylobacter was much higher in the conventional chicken (87). Importantly, these studies and most others that have investigated food-borne pathogen contamination on meats have not enumerated the microbial load on the product. These studies have focused on prevalence of resistance and of contamination; prevalence of samples on which specific bacteria can be found is not a measure of risk. Microbial loads, which relate directly to the dose that is capable of being ingested or disseminated, are more appropriate indicators of risk. When actual microbial loads are considered, organic products may have a food safety risk higher than that of the conventional product (34). With all of these different approaches to evaluating the relationship between antibiotic use and resistance, we are left asking ourselves whether we can now quantify, and subsequently reduce, the potential public health impacts of antibiotic use in poultry. How is antibiotic use in poultry actually impacting animal and human health? All antibiotic uses exert a selection pressure capable of selection for resistance. Increased levels of bacterial resistance on the farm will have obvious implications for the efficacy of therapeutic treatment of animal diseases on the farm. How, then, does this relate to human health? The most commonly assessed pathway linking antibiotic use in poultry to human health impacts is the farm-to-fork route. Pathogens that are resistant to antibiotics can be passed through the food chain. Following ingestion, these resistant pathogens might cause disease that is difficult to treat because of the antibiotic resistance. Therefore, human health outcomes that are used include treatment failures, increases in morbidity or mortality, and qualityof-life measurements, such as quality-adjusted life years (5,84). There are others ways in which antibiotic uses in poultry can impact human health. Instead of focusing on pathogens, some studies have begun to investigate resistance in commensal bacteria. It is now well recognized that commensals are capable of exchanging genes with other bacteria, including pathogens, in various environments (29,36,41,117). If commensal bacteria possessing antibiotic-resistance genes are ingested by humans through the same farm-to-fork pathway, the potential exists for these commensals to share these resistance genes with pathogens in the human gastrointestinal tract (92,93). The probability of this transfer might increase if the individual has recently taken antibiotics (73). Finally, bacteria can be disseminated into the environment from poultry farms. These bacteria can possess resistance genes and, therefore, present the same risk to human health as organisms passed through the farm-to-fork route. In that case, the organisms can be disseminated in the waterways, or they can be spread onto fields with poultry litter. Antibiotics and their metabolites are also capable of being disseminated in that fashion. There are many routes by which antibiotic use in poultry can influence animal and public health (Fig. 1). As we use the prevalence study to document changes in resistance levels or to compare resistance levels among farms with differing levels of antibiotic usage, how confident are we that we have assessed the causal impacts of antibiotic use in poultry? Have we succeeded in measuring the public health consequences caused by antibiotic use in poultry? How do we quantify the causal effects of antibiotic use on antibiotic resistance and public health risks? CAUSAL INFERENCE One of the primary goals of our research, surveillance, monitoring, and modeling is to determine the animal and public health impacts that are caused by antibiotic use. If the presence of resistant bacteria is the surrogate for a potential public health impact, then the question becomes how much of the observed bacterial resistance is caused by antibiotic use in poultry? Various studies have Antibiotic use in poultry 165 Fig. 1. Schematic representation of potential routes of transmission of antibiotic-susceptible or -resistant bacteria between animals, humans, and the environment. (Adapted from Phillips et al. [86]). conflicting results regarding the efficacy of growth-promotant antibiotics in the current poultry production system. Given that some antibiotic uses in poultry will remain, we must determine which uses are absolutely necessary and which uses pose the greatest risk to human and animal health. Unfortunately, a major factor in determining that potential risk relates to identifying whether the measured association between antibiotic use and antibiotic resistance (or some other indicator of risk) is actually a causal association. One goal of public health practitioners and veterinarians is to minimize risks to human health that are caused, either directly or indirectly, by the use of antibiotics in poultry. A healthy and safe food supply must be maintained. Identifying causal relationships is needed for making improvements to animal and human health, but unfortunately, that task is not as simple as it might sound. If we believe that antibiotic use in poultry causes antibiotic resistance, and that such usage causes human health risks, then the ideal study would be one in which all poultry farms are given the same set of antibiotics, and the human health effects are then characterized and quantified. At the exact same time, the exact same farms are then NOT given those same antibiotics; the only difference in the comparison trial, therefore, is the antibiotic usage, and consequently, the human health effects that are observed can be considered to be CAUSED by the antibiotic usage. That is an impossible study because we can not both expose and unexpose the same farms at the same time; therefore, the hypothetical alternative is termed counterfactual (52,78,90). The relationship between those two populations represents the real causal effect of antibiotic use on resistance. Because we can not observe the counterfactual situation, we use a substitute population, often called the control or comparison group. The relationship between the factual and substitute groups is often incorrectly interpreted as a causal relationship. As the differences between the counterfactual comparison group and the substitute comparison group grow, the degree of epidemiologic confounding and introduced bias also grows. Consequently, the observed effect of antibiotic use on resistance is an uncertain distance from the true causal effect. There is a general perception that increased antibiotic use implies an increased risk. However, not all antibiotic uses are equivalent in terms of their selection pressure and potential animal and human health risks. In addition, many of the studies that have assessed this relationship would suffer when scrutinized under the counterfactual model of causation. For example, in the case of antibiotic use in poultry and its effects on public health, the counterfactual model could sound like this: if poultry operations were simultaneously exposed to high and low levels of antibiotics, would the public health impacts be reduced when lower levels were used? As described previously, the standard study designs are those that compare poultry operations that use high levels to those that use low levels of antibiotics or designs in which antibiotic usage is changed over time in selected poultry operations. Public health impacts are then estimated from these comparisons. Neither of these approaches is a perfect substitute for the unobtainable counterfactual model, and consequently, we must critically assess the interpretations of these studies in light of their imperfections (78,81). Regardless, studies will often make causal conclusions from imperfect comparisons, often in the absence of any estimated epidemiological measure of effect (14). Finally, most studies compare the prevalence of resistant organisms in samples that have been affected by the antibiotic use (such as farms treated with an antibiotic) to those that are supposedly free of R. S. Singer and C. L. Hofacre 166 the selection pressure (farms that do not use the antibiotic). As was described in a recent review of epidemiological interpretations of antibiotic resistance data, prevalence data can lead to spurious conclusions about the relationship between antibiotic use and resistance (97). For example, an effective antibiotic might wipe out all susceptible bacterial targets while the resistant population will remain. There may be no effect on the resistant population, as evidenced by no increase in the microbial load of the resistant population. Based on prevalence data, a standard strength of association criterion of causation (89) might lead to the erroneous conclusion that antibiotic use in poultry was ineffective and caused an increase in resistance levels. If the dose–response criterion of causation is used (89), we might again reach the erroneous conclusion that higher levels of antibiotics result in increased resistance, when in fact they are just more effective at eliminating the susceptible bacteria. COMPETING CAUSES The aim of quantifying the causal association between antibiotic use and resistance on poultry farms is far from simple. Without precise and accurate estimates of that relationship, strategies aimed at reducing animal and human health risks may be ineffective. A major reason for the difficulty in assessing the relationship between antibiotic use and potential public health risks is the diversity of factors that can affect antibiotic-resistance levels other than antibiotic usage. In other words, the substitute or control group rarely resembles the causal counterfactual comparison group. Antimicrobial-resistance genes can often be found in physically linked assemblages within the bacterial cell (114). Consequently, resistance to a specific antibiotic can be under positive selection in the absence of a selection pressure from that antibiotic. Many examples of these effects have been described. For example, as some countries have reduced the amount of antibiotic used in animal agriculture, specific resistances have persisted in the apparent absence of primary selection pressure. In Norway, an isolate of Enterococcus hirae was found in poultry that possessed the vanA and erm(B) genes, conferring glycopeptide and macrolide resistance to this isolate (13). In that case, the potential existed for macrolide use in poultry to aid in the persistence of VRE; tylosin was also banned from use in poultry, thus, limiting that scenario. In Denmark, VRE has persisted after the ban of avoparcin in poultry (46). Tylosin use was continued for a short time after the avoparcin ban, and in swine, glycopeptideresistant E. faecium (GRE) prevalence did not decline rapidly. The same genetic linkage between vanA and erm(B) was detected (1), and following the decrease in tylosin usage in pigs, the GRE prevalence declined (2). Commonly used disinfectants can also help explain certain patterns of antimicrobial resistance. For example, quaternary ammonium compounds (QAC) are commonly used in hatcheries, as well as in processing plants and in retail establishments, and resistance can develop to these disinfectants (94,100). There are many examples of QAC-resistance genes having a genetic linkage with antimicrobial-resistance genes (98,99). The location of the QAC-resistance gene is often on a plasmid, enabling an increased potential for horizontal gene transfer among bacteria. In addition, a QAC-resistance gene cassette has been found within the class I integron in various bacteria, including Salmonella spp. (42). In this example, QAC usage may select for resistance to many other antibiotics. Many bacteria possess mechanisms of resistance to specific metals to prevent potential toxicity from metal exposure. Copper is commonly used in food-animal production, for example, in poultry feed in the form of copper sulfate (65). Copper has been shown to select for copper-resistant bacterial isolates (12), and this process is also capable of selecting for increased antibiotic resistance (12). A recent study of E. faecium from broiler and turkey farms demonstrated the presence of copper resistance conferred by the tcrB gene. Of concern was the finding that the tcrB gene was linked to the vatE gene on the same plasmid, thus enabling a copperselection pressure to maintain virginiamycin resistance, even in the absence of virginiamycin use (101). Continued use of copper sulfate in swine production in Denmark did not appear to maintain the combined resistance to copper, glycopeptides, and macrolides in E. faecium, suggesting that the resistance genes might not be directly linked to metal resistances in all cases (44). Another metal resistance that has been commonly linked to antibiotic-resistance genes is mercury, which is commonly found in the environment. Mercuryresistance genes can be found in association with mobile genetic elements (8), and these mercury-resistance genes are often linked to other antibiotic-resistance genes (125). For example, the transposon Tn21 possesses the mercury-resistance mer operon and a class 1 integron that includes antibiotic-resistance genes (11,42,70). Overall, the selection of antibiotic resistance in poultry production, which includes the poultry environment, may have more to do with the presence of additional genes that confer resistance to chemicals and metals or that provide an ecological fitness advantage to the cell than to the presence of a primary antibiotic-resistance gene (3). Environmental-selection pressures also exist in the form of antibiotics and antibiotic metabolites. Given that certain microorganisms, especially those that live predominantly in the soil, are capable of producing antibiotics, increased levels of resistance would be expected in areas where these antibiotic-producing organisms are in abundance (27). However, antibiotics can also be found in the waterways (62) after release from hospitals, sewage treatment plants, factories, and agricultural runoff (64). When poultry litter is used as fertilizer, the potential exists for spreading resistant organisms and active antibiotic compounds into the environment. This route of potential risk to public health has been understudied. Perhaps the most difficult aspect of the design of antibioticresistance studies is the inclusion of the routes and probabilities of transmission of antibiotic resistant organisms. When considering the effects of antibiotic use, the indirect effects (which include transmission) might represent the most important contributors to antibiotic-resistance emergence and dissemination at the ecological level (73). Bacteria are capable of being disseminated among sites, and therefore, resistant organisms can be transmitted through a variety of routes. Resistance can be found in areas irrespective of any selection pressures. For example, effluents from hospitals and sewage treatment plants represent just one source of resistant organisms. When considering the dissemination of organisms, agricultural runoff will, of course, be an important contributor. Airborne dissemination of resistant bacteria from farms has been observed (16,38). People returning to the community after hospital stays can carry resistant organisms (73). Foreign travel is always a major risk factor for the acquisition of resistant food-borne infections (33,56,85). Finally, organisms can be moved from site to site, and potentially at large geographic distances, via mechanical transmission or via biological vectors. Antibiotic use in poultry can have public health impacts through the selection of resistant organisms and subsequent dissemination by one of these understudied routes. It must be emphasized that resistant organisms are also capable of being introduced into the poultry farm via these same processes (60), further complicating our ability to distinguish and quantify the causal impact of antibiotic usage in poultry on antibiotic resistance or on public health. Antibiotic use in poultry RISK ASSESSMENT AND MANAGEMENT Quantifying the causal relationship between antibiotic use in poultry and public health impacts is challenging and difficult to do in empirical studies. Related to the ways in which antibiotic use can impact public health, we know that antibiotic use in poultry can increase the pool of resistant bacteria and resistance genes. This can potentially lead to increased carriage of resistant bacteria and resistance genes in the human community. These antibiotic uses can also lead to the release of resistant organisms and selection pressures into the environment. There should be no need to find the ‘‘smoking gun’’ example of an actual instance in which antibiotic use in poultry causes an antibiotic resistant organism to cause harm in humans; the risks associated with any antibiotic use are well known, and thus, the potential risk to human health exists from antibiotic use in poultry. The question is, how do we begin to estimate the strength of this causal relationship and then manage the potential risks? For this purpose, a variety of mathematical models and quantitative risk assessments have been developed to predict the effects of antibiotic usage on resistance and human health. These models have been used to estimate the likely impacts of continued antibiotic use but rarely to evaluate management changes and efficacious interventions for reducing public health risks. One of the first mathematical models relating antibiotic use in animals and humans to human health risks predicted that animal use of antibiotics would decrease the efficacy duration of antibiotics in human medicine but that human uses of antibiotics would not only decrease the efficacy but also would increase the overall prevalence of resistance (105). In a subsequent model, virginiamycin use in poultry was assessed as a risk for the emergence of streptogramin resistance (106). Although this model did not explicitly consider the risks of infection with a streptograminresistant isolate and subsequent treatment failure, the model evaluated the combined medical and agricultural use of streptogramins and predicted that virginiamycin use in poultry would reduce the efficacy of quinupristin–dalfopristin (QD). In other models that have included actual health outcomes, the risks of increased morbidity and mortality due to QD-resistant E. faecium infection in humans associated with virginiamycin use in poultry were predicted to be very low (24) and were highly sensitive to QD prescription rates in humans (23). The model assumed that transfer of resistant E. faecium actually occurs and that it is caused by virginiamycin use in chickens. The model suggested that the benefit of removing virginiamycin use from poultry rapidly decreased as human QD use increased and high-level QD resistance spreads. The impact of virginiamycin use appears to depend on the rate of transmission of the QD resistance in humans (more precisely modeled as R0, the basic reproductive rate of transmission) (57). The lower the potential spread of resistance among humans, the more important the virginiamycin use in poultry becomes. Risk analysis methodologies have also been used to determine the likelihood of specific outcomes, including negative public health impacts, resulting from the use of antibiotics in animal agriculture (18,119). A recent publication has reviewed many of the microbial risk assessments that have been conducted in the area of antimicrobial resistance (110). In one high-profile example, a risk assessment was conducted by the U.S. Food and Drug Administration, Center for Veterinary Medicine (FDA-CVM) to estimate the public health burden associated with fluoroquinolone-resistant Campylobacter infections in people that were caused by the use of fluoroquinolones in chickens (10). The model was a major contributor to the recent decision to remove fluoroquinolones from 167 use in poultry production. In a farm-to-fork type of risk assessment model, the impacts of macrolide use in animal agriculture were assessed (53). In this model, resistance that was generated in animal production as a result of macrolide use was followed through the food chain to determine how many people would be infected with these resistant strains, would subsequently seek medical care, and would be treated with a macrolide. Risk was defined as the probability of this chain of events, combined with the consequence of macrolide-treatment failure due to resistant Campylobacter spp. or E. faecium. In general the estimated risks were very small, with the largest risk being an annual probability of less than 1 in 10 million for macrolide-resistant Campylobacter derived from poultry. Many mathematical and risk assessment models have now been developed, and each seems to take a different approach to quantifying and predicting the human health risks associated with antibiotic use in animal agriculture. Some of these have been used for regulating the uses of animal antibiotics, even though some have argued that the models do not actually predict causal relationships (22). Many of these models have been subjected to a rigorous sensitivity analysis, in which all model parameters are assessed for their relative importance in influencing the outcome of the model. The sensitivity analysis can reveal not only those parameters that appear to be biologically important in affecting the outcome but also those parameters about which we need to know more information. For example, a parameter to which the model is highly sensitive might indicate the need to generate more precise estimates for the value of this parameter. In a recent review, the argument was made that some of these mathematical and risk assessment models are missing key parameters, and others have very broad and potentially biased assumptions, and thus, the estimated risks may be over- or underestimated (7). One of the primary reasons for conducting a rigorous modeling exercise is to enable management to make informed decisions. Ultimately, though, there are two ways in which antibiotic use in poultry can be managed. One approach is to employ the precautionary principle. In this argument, the consequences of an action are often unknown, such as the actual public health risks associated with animal antibiotic use. There is a perceived potential for serious negative consequences of this action, and therefore, it is better to avoid the action rather than to suffer the potential consequences. Under this principle, which on the surface makes good public health sense, certain antibiotic uses have been withdrawn from animal agriculture. One reason why this approach is often needed, especially in the case of antibiotic use and resistance, is the difficulty in designing, implementing, and analyzing the decisive studies that will prove a negative health effect. By the time such a study is complete, the negative effects associated with continued antibiotic use might be irreversible. Consequently, the precautionary principle approach to managing antibiotic use in poultry has only one real option: withdraw the antibiotic use that might result in a negative human health consequence. Another way to evaluate the potential consequences of antibiotic use in poultry is to develop scientifically based predictions, and through these models, to evaluate interventions that reduce the human and animal health risks associated with the continued antibiotic use in poultry. Unfortunately, most risk assessments conducted to date in antibiotic resistance that have been used for regulatory purposes have not been based on interventions. Instead, the assessments have attempted to qualify the risks associated with the antibiotic use or else to quantify the risks for the sole purpose of making the dichotomous decision of whether or not to withdraw the antibiotic from use. R. S. Singer and C. L. Hofacre 168 FUTURE DIRECTIONS Estimating the potential impacts of antibiotic use in poultry production is complex. To make more accurate assessments of these impacts and to manage these antibiotic uses in a rational and efficacious manner, research must continue in specific directions. An area in which more work is needed is in relation to the physical linkage of genes in the bacterial chromosome and in mobile genetic elements, such as plasmids (29,36,41,117). For example, a plasmid from an avian pathogenic E. coli (APEC) was recently sequenced in its entirety (54). This plasmid, labeled pAPEC-O2-R, contains many different types of genes, such as heavy metal– resistance genes, a QAC-resistance gene, and up to eight different antibiotic-resistance genes. All of these genes can be under selection pressure in the poultry environment, resulting in the coselection and potential persistence of many different antibiotic resistances in the absence of antibiotic pressure. In some of the early studies of antibiotic use and resistance in chickens, these R plasmids were found to be moveable to humans and were considered a risk in the form of a potential reservoir of antibiotic-resistance genes (69). As described earlier, there are few therapeutic options available for treating infections such as airsacculitis, and these limited choices may have helped contribute to the presence and persistence of tetracycline- and sulfonamide-resistance genes residing on this APEC plasmid. Until we have a better understanding of the diversity of factors that can select for the persistence of antibiotic-resistance genes, it will be difficult to find management strategies that can actually reduce the levels of antibiotic resistance currently circulating in the poultry production system. In addition to elucidating the diversity and interdependence of genes located within the bacterial cell, we also need to evaluate how the entire bacterial community is affected by antibiotic uses. Bacteria are known to transfer resistance genes, even among distantly related bacteria (36,41,117). The selection pressures exerted by antibiotic uses affect the entire population of bacteria, not just those that are the intended targets. The emerging field of metagenomics, which assesses the genomic composition of the bacterial community in complex samples, is being used more frequently to fill this data gap (43). Some studies have begun to characterize this bacterial community, for instance, in the intestinal tract of the maturing broiler chicken (75), in the broiler litter (76), and in the cecal bacterial community, following exposure to oxytetracycline (35). There is still the common belief that resistance will decrease when an antibiotic is removed. The assumption is based primarily on the ecological theory that the bacterium incurs a fitness cost for maintaining the resistance gene but that this cost is less than the benefit when an antibiotic is present. In the absence of the selection pressure, the theory would state that the susceptible strain will either out-compete the resistant strain or the resistant strain will lose the resistance gene. Unfortunately, recent studies indicate that the maintenance of resistance may not impose a significant fitness cost or that this cost can be overcome (4,58,59,66,67,77,95,96). To better manage our current antibiotic uses, we need to better comprehend the ways in which bacteria can harbor resistance and how resistant strains are likely to persist and spread, even in the absence of the antibiotic-selection pressure. The models that we build to assess the potential risks of antibiotic use in poultry must begin to take a more holistic view of health. Specifically, these models need to include the potential risks and the potential benefits associated with antibiotic use. Phrased another way, are there potential unintended consequences of removing antibiotics from use in poultry? Recent models have predicted that there might be some significant negative human health consequences associated with the removal of certain antibiotics from poultry production (25,102). In these models, it is assumed that the antibiotics are helping maintain a healthier population of birds as well as promoting size uniformity of the birds in the flock. These factors can potentially lead to reduced microbial loads on the carcass, which is a direct measure of the potential risk of foodborne illness associated with consumption of that product. Consequently, antibiotic use has potential benefits and risks, and all of these potential consequences should be a part of our future assessments of the ways in which antibiotics are used in poultry production. Finally, and probably most important, the assessments that are used to determine the potential impacts of antibiotic use in animal agriculture must be improved: these assessments must include evaluations of potential interventions for reducing the risks to human and animal health. This is especially true given that the United States and some other countries have explored a sciencebased approach to assessing and managing risk. For example, in the United States FDA-CVM risk assessment of fluoroquinolone use in chickens, the aim was never to evaluate potential interventions for reducing the risks to human health associated with fluoroquinoloneresistant Campylobacter infections that were caused by fluoroquinolone use in poultry; the model only estimated the potential human health impact. By using a different modeling format, however, potential interventions could have been evaluated, and many different solutions could have been assessed. For example, the processing of flocks that have received fluoroquinolones in a separated fashion from flocks that have not been treated could have been evaluated as a risk-reduction strategy. In this case, treated flocks could be processed as the final group of birds before a full processing plant cleaning and disinfection. This would reduce the chance of cross-contamination of chicken meat from nontreated houses with the bacteria from treated houses. In another potential intervention, houses or farms that have received fluoroquinolones could be cleaned in a more intensive manner than the normal cleaning, and all litter from these flocks could be sterilized. Perhaps, flocks from these farms during successive grow-outs would have to be processed in a similar manner as the treated flocks. Finally, an intervention in which flocks that have been treated with antibiotics would have to wait for a longer period of time before processing could have been assessed. This type of approach would resemble the mandatory withdrawal times associated with antibiotic residues. Feeding studies of other antibiotics have shown that resistance increases after exposure to the antibiotic but that the feeding of the antibiotic might be required to maintain the resistance (80). Consequently, management strategies could also include guidelines of when antibiotic uses should be ceased in flocks before they go to processing to reduce the load of resistant organisms in the birds. These types of interventions might sound labor-intensive and costly. They are, and that is the point. Under certain circumstances, it might be cost-effective and ethical for a producer to use a powerful antibiotic to control a severe disease in the flock, but this use would then have major repercussions on how the flock and the farm are subsequently managed. Producers might not opt for this intensive measure, but at least, they would have a choice that is accepted as scientifically sound for reducing both the human and animal risks associated with the antibiotic use in their flocks. As we begin to gain a better understanding of the ecology of resistance and its relation to animal and human health, we will need these scientifically based strategies for minimizing the impacts of antibiotic use on animal, human, and environmental health. Antibiotic use in poultry REFERENCES 1. Aarestrup, F. M. Characterization of glycopeptide-resistant Enterococcus faecium (GRE) from broilers and pigs in Denmark: genetic evidence that persistence of GRE in pig herds is associated with coselection by resistance to macrolides. J. Clin. Microbiol. 38:2774–2777. 2000. 2. Aarestrup, F. M., A. M. Seyfarth, H. D. Emborg, K. Pedersen, R. S. Hendriksen, and F. Bager. Effect of abolishment of the use of antimicrobial agents for growth promotion on occurrence of antimicrobial resistance in fecal enterococci from food animals in Denmark. Antimicrob. Agents Chemother. 45:2054–2059. 2001. 3. Alonso, A., P. Sanchez, and J. L. Martinez. Environmental selection of antibiotic resistance genes. Environ. Microbiol. 3:1–9. 2001. 4. Andersson, D. I., and B. R. Levin. The biological cost of antibiotic resistance. Curr. Opin. Microbiol. 2:489–493. 1999. 5. Angulo, F. J., V. N. Nargund, and T. C. Chiller. Evidence of an association between use of anti-microbial agents in food animals and antimicrobial resistance among bacteria isolated from humans and the human health consequences of such resistance. J. Vet. Med. B Infect. Dis. Vet. Public Health 51:374–379. 2004. 6. Bager, F., M. Madsen, J. Christensen, and F. M. Aarestrup. Avoparcin used as a growth promoter is associated with the occurrence of vancomycin-resistant Enterococcus faecium on Danish poultry and pig farms. Prev. Vet. Med. 31:95–112. 1997. 7. Barber, D. A., G. Y. Miller, and P. E. McNamara. Models of antimicrobial resistance and foodborne illness: examining assumptions and practical applications. J. Food Prot. 66:700–709. 2003. 8. Barkay, T., S. M. Miller, and A. O. Summers. Bacterial mercury resistance from atoms to ecosystems. FEMS Microbiol. Rev. 27:355–384. 2003. 9. Barnes, H. J., J. P. Valliancourt, and W. B. Gross. Colibacillosis. In: Diseases of poultry, 11th ed. Y. M. Saif, ed. Iowa State Press, Ames, IA. pp. 631–652. 2003. 10. Bartholomew, M. J., D. J. Vose, L. R. Tollefson, and C. C. Travis. A linear model for managing the risk of antimicrobial resistance originating in food animals. Risk Anal. 25:99–108. 2005. 11. Bass, L., C. A. Liebert, M. D. Lee, A. O. Summers, D. G. White, S. G. Thayer, and J. J. Maurer. Incidence and characterization of integrons, genetic elements mediating multiple-drug resistance, in avian Escherichia coli. Antimicrob. Agents Chemother. 43:2925–2929. 1999. 12. Berg, J., A. Tom-Petersen, and O. Nybroe. Copper amendment of agricultural soil selects for bacterial antibiotic resistance in the field. Lett. Appl. Microbiol. 40:146–151. 2005. 13. Borgen, K., M. Sorum, Y. Wasteson, H. Kruse, and H. Oppegaard. Genetic linkage between erm(B) and vanA in Enterococcus hirae of poultry origin. Microb. Drug Resist. 8:363–368. 2002. 14. Carraminana, J. J., C. Rota, I. Agustin, and A. Herrera. High prevalence of multiple resistance to antibiotics in Salmonella serovars isolated from a poultry slaughterhouse in Spain. Vet. Microbiol. 104:133–139. 2004. 15. Casewell, M., C. Friis, E. Marco, P. McMullin, and I. Phillips. The European ban on growth-promoting antibiotics and emerging consequences for human and animal health. J. Antimicrob. Chemother. 52:159–161. 2003. 16. Chapin, A., A. Rule, K. Gibson, T. Buckley, and K. Schwab. Airborne multidrug-resistant bacteria isolated from a concentrated swine feeding operation. Environ. Health Perspect. 113:137–142. 2005. 17. Chaslus-Dancla, E., G. Gerbaud, M. Lagorce, J. P. Lafont, and P. Courvalin. Persistence of an antibiotic resistance plasmid in intestinal Escherichia coli of chickens in the absence of selective pressure. Antimicrob. Agents Chemother. 31:784–788. 1987. 18. Claycamp, H. G., and B. H. Hooberman. Antimicrobial resistance risk assessment in food safety. J. Food Prot. 67:2063–2071. 2004. 19. Code of Federal Regulations. 21 CFR. Title 21: Food and drugs. Office of the Federal Register, Washington, DC. 2005. 20. Compendium of Veterinary Products. 2003. Compendium of veterinary products. North American Compendiums, Inc., Port Huron, MI. 2003. 169 21. Corrier, D. E., J. A. Byrd, B. M. Hargis, M. E. Hume, R. H. Bailey, and L. H. Stanker. Presence of Salmonella in the crop and ceca of broiler chickens before and after preslaughter feed withdrawal. Poult. Sci. 78:45–49. 1999. 22. Cox, L. A., Jr. Some limitations of a proposed linear model for antimicrobial risk management. Risk Anal. 25:1327–1332. 2005. 23. Cox, L. A., Jr., and D. A. Popken. Bayesian Monte Carlo uncertainty analysis of human health risks from animal antimicrobial use in a dynamic model of emerging resistance. Risk Anal. 24:1153–1164. 2004. 24. Cox, L. A., Jr., and D. A. Popken. Quantifying human health risks from virginiamycin used in chickens. Risk Anal. 24:271–288. 2004. 25. Cox, L. A., Jr., and D. A. Popken. Quantifying potential human health impacts of animal antibiotic use: enrofloxacin and macrolides in chickens. Risk Anal. 26:135–146. 2006. 26. Cui, S., B. Ge, J. Zheng, and J. Meng. Prevalence and antimicrobial resistance of Campylobacter spp. and Salmonella serovars in organic chickens from Maryland retail stores. Appl. Environ. Microbiol. 71:4108–4111. 2005. 27. D’Costa, V. M., K. M. McGrann, D. W. Hughes, and G. D. Wright. Sampling the antibiotic resistome. Science 311:374–377. 2006. 28. Danish Integrated Antimicrobial Resistance Monitoring and Research Programme. DANMAP 2001—use of antimicrobial agents and occurrence of antimicrobial resistance in bacteria form food animals, foods and humans in Denmark. Danish Veterinary Laboratory, Copenhagen, Denmark. 2002. 29. Davison, J. Genetic exchange between bacteria in the environment. Plasmid 42:73–91. 1999. 30. Dibner, J. J., and J. D. Richards. Antibiotic growth promoters in agriculture: history and mode of action. Poult. Sci. 84:634–643. 2005. 31. Donabedian, S., L. A. Thal, P. Bozigar, T. Zervos, E. Hershberger, and M. Zervos. Antimicrobial resistance in swine and chickens fed virginiamycin for growth promotion. J. Microbiol. Methods 55:739–743. 2003. 32. Elwinger, K., C. Schneitz, E. Berndtson, O. Fossum, B. Teglof, and B. Engstom. Factors affecting the incidence of necrotic enteritis, caecal carriage of Clostridium perfringens and bird performance in broiler chicks. Acta Vet. Scand. 33:369–378. 1992. 33. Engberg, J., J. Neimann, E. M. Nielsen, F. M. Aerestrup, and V. Fussing. Quinolone-resistant Campylobacter infections: risk factors and clinical consequences. Emerg. Infect. Dis. 10:1056–1063. 2004. 34. Engvall, A. May organically farmed animals pose a risk for Campylobacter infections in humans? Acta Vet. Scand. Suppl 95:85–87. 2001. 35. Fairchild, A. S., J. L. Smith, U. Idris, J. Lu, S. Sanchez, L. B. Purvis, C. Hofacre, and M. D. Lee. Effects of orally administered tetracycline on the intestinal community structure of chickens and on tet determinant carriage by commensal bacteria and Campylobacter jejuni. Appl. Environ. Microbiol. 71:5865–5872. 2005. 36. Frost, L. S., R. Leplae, A. O. Summers, and A. Toussaint. Mobile genetic elements: the agents of open source evolution. Nat. Rev. Microbiol. 3:722–732. 2005. 37. Garcia-Migura, L., E. Pleydell, S. Barnes, R. H. Davies, and E. Liebana. Characterization of vancomycin-resistant Enterococcus faecium isolates from broiler poultry and pig farms in England and Wales. J. Clin. Microbiol. 43:3283–3289. 2005. 38. Gibbs, S. G., C. F. Green, P. M. Tarwater, and P. V. Scarpino. Airborne antibiotic resistant and nonresistant bacteria and fungi recovered from two swine herd confined animal feeding operations. J. Occup. Environ. Hyg. 1:699–706. 2004. 39. Glisson, J. R. Bacterial respiratory disease of poultry. Poult. Sci. 77: 1139–1142. 1998. 40. Glisson, J. R., C. L. Hofacre, and G. F. Mathis. Comparative efficacy of enrofloxacin, oxytetracycline, and sulfadimethoxine for the control of morbidity and mortality caused by Escherichia coli in broiler chickens. Avian Dis. 48:658–662. 2004. 41. Gogarten, J. P., and J. P. Townsend. Horizontal gene transfer, genome innovation, and evolution. Nat. Rev. Microbiol. 3:679–687. 2005. 42. Guerra, B., S. Soto, R. Helmuth, and M. C. Mendoza. Characterization of a self-transferable plasmid from Salmonella enterica serotype 170 R. S. Singer and C. L. Hofacre typhimurium clinical isolates carrying two integron-borne gene cassettes together with virulence and drug resistance genes. Antimicrob. Agents Chemother. 46:2977–2981. 2002. 43. Handelsman, J. Metagenomics: application of genomics to uncultured microorganisms. Microbiol. Mol. Biol. Rev. 68:669–685. 2004. 44. Hasman, H., and F. M. Aarestrup. Relationship between copper, glycopeptide, and macrolide resistance among Enterococcus faecium strains isolated from pigs in Denmark between 1997 and 2003. Antimicrob. Agents Chemother. 49:454–456. 2005. 45. Heuer, O. E., K. Pedersen, J. S. Andersen, and M. Madsen. Prevalence and antimicrobial susceptibility of thermophilic Campylobacter in organic and conventional broiler flocks. Lett. Appl. Microbiol. 33:269–274. 2001. 46. Heuer, O. E., K. Pedersen, J. S. Andersen, and M. Madsen. Vancomycin-resistant enterococci (VRE) in broiler flocks 5 years after the avoparcin ban. Microb. Drug Resist. 8:133–138. 2002. 47. Heuer, O. E., K. Pedersen, L. B. Jensen, M. Madsen, and J. E. Olsen. Persistence of vancomycin-resistant enterococci (VRE) in broiler houses after the avoparcin ban. Microb. Drug Resist. 8:355–361. 2002. 48. Hinton, M. The ecology of Escherichia coli in animals including man with particular reference to drug resistance. Vet. Rec. 119:420–426. 1986. 49. Hinton, M., A. Kaukas, S. K. Lim, and A. H. Linton. Preliminary observations on the influence of antibiotics on the ecology of Escherichia coli and the enterococci in the faecal flora of healthy young chickens. J. Antimicrob. Chemother. 18(Suppl C):165–173. 1986. 50. Hofacre, C. L. The health and management of poultry production. Int. J. Infect. Dis. 6(Suppl 3):3–7. 2002. 51. Hofacre, C. L., R. Froyman, B. Gautrias, B. George, M. A. Goodwin, and J. Brown. Use of Aviguard and other intestinal bioproducts in experimental Clostridium perfringens–associated necrotizing enteritis in broiler chickens. Avian Dis. 42:579–584. 1998. 52. Hofler, M. Causal inference based on counterfactuals. BMC Med. Res. Methodol. 5:28. 2005. 53. Hurd, H. S., S. Doores, D. Hayes, A. Mathew, J. Maurer, P. Silley, R. S. Singer, and R. N. Jones. Public health consequences of macrolide use in food animals: a deterministic risk assessment. J. Food Prot. 67:980–992. 2004. 54. Johnson, T. J., K. E. Siek, S. J. Johnson, and L. K. Nolan. DNA sequence and comparative genomics of pAPEC-O2-R, an avian pathogenic Escherichia coli transmissible R plasmid. Antimicrob. Agents Chemother. 49: 4681–4688. 2005. 55. Jones, F. T., and S. C. Ricke. Observations on the history of the development of antimicrobials and their use in poultry feeds. Poult. Sci. 82: 613–617. 2003. 56. Kassenborg, H. D., K. E. Smith, D. J. Vugia, T. Rabatsky-Ehr, M. R. Bates, M. A. Carter, N. B. Dumas, M. P. Cassidy, N. Marano, R. V. Tauxe, and F. J. Angulo. Fluoroquinolone-resistant Campylobacter infections: eating poultry outside of the home and foreign travel are risk factors. Clin. Infect. Dis. 38(Suppl 3):S279–S284. 2004. 57. Kelly, L., D. L. Smith, E. L. Snary, J. A. Johnson, A. D. Harris, M. Wooldridge, and J. G. Morris, Jr. Animal growth promoters: to ban or not to ban? a risk assessment approach. Int. J. Antimicrob. Agents 24:205–212. 2004. 58. Khachatryan, A. R., D. D. Hancock, T. E. Besser, and D. R. Call. Role of calf-adapted Escherichia coli in maintenance of antimicrobial drug resistance in dairy calves. Appl. Environ. Microbiol. 70:752–757. 2004. 59. Khachatryan, A. R., D. D. Hancock, T. E. Besser, and D. R. Call. Antimicrobial drug resistance genes do not convey a secondary fitness advantage to calf-adapted Escherichia coli. Appl. Environ. Microbiol. 72: 443–448. 2006. 60. Kinde, H., D. H. Read, A. Ardans, R. E. Breitmeyer, D. Willoughby, H. E. Little, D. Kerr, R. Gireesh, and K. V. Nagaraja. Sewage effluent: likely source of Salmonella enteritidis, phage type 4 infection in a commercial chicken layer flock in southern California. Avian Dis. 40: 672–676. 1996. 61. Klare, I., D. Badstubner, C. Konstabel, G. Bohme, H. Claus, and W. Witte. Decreased incidence of VanA-type vancomycin-resistant enterococci isolated from poultry meat and from fecal samples of humans in the community after discontinuation of avoparcin usage in animal husbandry. Microb. Drug Resist. 5:45–52. 1999. 62. Kolpin, D. W., E. T. Furlong, M. T. Meyer, E. M. Thurman, S. D. Zaugg, L. B. Barber, and H. T. Buxton. Pharmaceuticals, hormones, and other organic wastewater contaminants in U.S. streams, 1999–2000: a national reconnaissance. Environ. Sci. Technol. 36:1202– 1211. 2002. 63. Kouba, M. Quality of organic animal products. Livest. Prod. Sci. 80: 33–40. 2003. 64. Kummerer, K., and A. Henninger. Promoting resistance by the emission of antibiotics from hospitals and households into effluent. Clin. Microbiol. Infect. 9:1203–1214. 2003. 65. Leeson, S., and J. D. Summers. Minerals. In: Nutrition of the chicken, 4th ed. University Books, Guelph, Ontario, Canada. pp. 394–397. 2001. 66. Lenski, R. E. Bacterial evolution and the cost of antibiotic resistance. Int. Microbiol. 1:265–270. 1998. 67. Levin, B. R., V. Perrot, and N. Walker. Compensatory mutations, antibiotic resistance and the population genetics of adaptive evolution in bacteria. Genetics 154:985–997. 2000. 68. Levy, S. B., G. B. FitzGerald, and A. B. Macone. Changes in intestinal flora of farm personnel after introduction of a tetracyclinesupplemented feed on a farm. N. Engl. J. Med. 295:583–588. 1976. 69. Levy, S. B., G. B. FitzGerald, and A. B. Macone. Spread of antibioticresistant plasmids from chicken to chicken and from chicken to man. Nature 260:40–42. 1976. 70. Liebert, C. A., R. M. Hall, and A. O. Summers. Transposon Tn21, flagship of the floating genome. Microbiol. Mol. Biol. Rev. 63:507–522. 1999. 71. Linton, A. H., M. H. Hinton, and Z. A. Al Chalaby. Monitoring for antibiotic resistance in enterococci consequent upon feeding growth promoters active against gram-positive bacteria. J. Vet. Pharmacol. Ther. 8:62–70. 1985. 72. Linton, A. H., K. Howe, C. L. Hartley, H. M. Clements, M. H. Richmond, and A. D. Osborne. Antibiotic resistance among Escherichia coli O-serotypes from the gut and carcasses of commercially slaughtered broiler chickens: a potential public health hazard. J. Appl. Bacteriol. 42:365–378. 1977. 73. Lipsitch, M., and M. H. Samore. Antimicrobial use and antimicrobial resistance: a population perspective. Emerg. Infect. Dis. 8:347–354. 2002. 74. Lu, J., C. L. Hofacre, and M. D. Lee. Emerging technologies in microbial ecology aid in understanding the effect of monensin on necrotic enteritis. J. Appl. Poult. Res. 15:145–153. 2006. 75. Lu, J., U. Idris, B. Harmon, C. Hofacre, J. J. Maurer, and M. D. Lee. Diversity and succession of the intestinal bacterial community of the maturing broiler chicken. Appl. Environ Microbiol 69:6816–6824. 2003. 76. Lu, J., S. Sanchez, C. Hofacre, J. J. Maurer, B. G. Harmon, and M. D. Lee. Evaluation of broiler litter with reference to the microbial composition as assessed by using 16S rRNA and functional gene markers. Appl. Environ. Microbiol. 69:901–908. 2003. 77. Luo, N., O. Sahin, J. Lin, L. O. Michel, and Q. Zhang. In vivo selection of Campylobacter isolates with high levels of fluoroquinolone resistance associated with gyrA mutations and the function of the CmeABC efflux pump. Antimicrob. Agents Chemother. 47:390–394. 2003. 78. Maldonado, G., and S. Greenland. Estimating causal effects. Int. J. Epidemiol. 31:422–429. 2002. 79. Margard, W. L., A. C. Peters, R. N. Pesut, and J. H. Litchfield. Chlortetracycline resistance in enteric microorganisms in chickens and swine. In: Developments in industrial microbiology. American Institute of Biological Sciences, Washington, DC. pp. 376–392. 1971. 80. McDermott, P. F., P. Cullen, S. K. Hubert, S. D. McDermott, M. Bartholomew, S. Simjee, and D. D. Wagner. Changes in antimicrobial susceptibility of native Enterococcus faecium in chickens fed virginiamycin. Appl. Environ. Microbiol. 71:4986–4991. 2005. 81. McGowan, J. E., Jr. Antimicrobial resistance in hospital organisms and its relation to antibiotic use. Rev. Infect. Dis. 5:1033–1048. 1983. 82. Mellon, M., and C. B. K. L. Benbrook. Hogging it: estimates of antimicrobial abuse in livestock. UCS Publications, Cambridge, MA. 2001. Antibiotic use in poultry 83. Mitsuhashi, S., H. Hashimoto, and K. Suzuki. Drug resistance of enteric bacteria, 13: distribution of R factors in Escherichia coli strains isolated from livestock. J. Bacteriol. 94:1166–1169. 1967. 84. Molbak, K. Spread of resistant bacteria and resistance genes from animals to humans—the public health consequences. J. Vet. Med. B Infect. Dis. Vet. Public Health 51:364–369. 2004. 85. Nelson, J. M., K. E. Smith, D. J. Vugia, T. Rabatsky-Ehr, S. D. Segler, H. D. Kassenborg, S. M. Zansky, K. Joyce, N. Marano, R. M. Hoekstra, and F. J. Angulo. Prolonged diarrhea due to ciprofloxacinresistant Campylobacter infection. J. Infect. Dis. 190:1150–1157. 2004. 86. Phillips, I., M. Casewell, T. Cox, B. De Groot, C. Friis, R. Jones, C. Nightingale, R. Preston, and J. Waddell. Does the use of antibiotics in food animals pose a risk to human health? a critical review of published data. J. Antimicrob. Chemother. 53:28–52. 2004. 87. Price, L. B., E. Johnson, R. Vailes, and E. Silbergeld. Fluoroquinolone-resistant Campylobacter isolates from conventional and antibiotic-free chicken products. Environ. Health Perspect. 113:557–560. 2005. 88. Robb-Nicholson, C. By the way, doctor, some brands of fresh chicken claim to be healthier because they’re antibiotic-free; are the antibiotics in chicken unsafe to eat? what are they for? Harv. Womens Health Watch. 13:8. 2005. 89. Rothman, K. J., and S. Greenland. Modern epidemiology. 2nd ed. Lippincott-Raven, Philadelphia. 1998. 90. Rubin, D. B. Estimating causal effects of treatments in randomized and nonrandomized studies. J. Educ. Psychol. 66:688–701. 1974. 91. Russell, S. M. The effect of airsacculitis on bird weights, uniformity, fecal contamination, processing errors, and populations of Campylobacter spp. and Escherichia coli. Poult. Sci. 82:1326–1331. 2003. 92. Salyers, A. A. Antibiotic resistance transfer in the mammalian intestinal tract: implications for human health, food safety and biotechnology, Springer-Verlag, New York. 1995. 93. Salyers, A. A., A. Gupta, and Y. Wang. Human intestinal bacteria as reservoirs for antibiotic resistance genes. Trends Microbiol. 12:412–416. 2004. 94. Sander, J. E., C. L. Hofacre, I. H. Cheng, and R. D. Wyatt. Investigation of resistance of bacteria from commercial poultry sources to commercial disinfectants. Avian Dis. 46:997–1000. 2002. 95. Sander, P., B. Springer, T. Prammananan, A. Sturmfels, M. Kappler, M. Pletschette, and E. C. Bottger. Fitness cost of chromosomal drug resistance-conferring mutations. Antimicrob. Agents Chemother. 46:1204– 1211. 2002. 96. Schrag, S. J., V. Perrot, and B. R. Levin. Adaptation to the fitness costs of antibiotic resistance in Escherichia coli. Proc. Biol. Sci. 264:1287– 1291. 1997. 97. Schwaber, M. J., T. De Medina, and Y. Carmeli. Epidemiological interpretation of antibiotic resistance studies—what are we missing? Nat. Rev. Microbiol. 2:979–983. 2004. 98. Sidhu, M. S., E. Heir, T. Leegaard, K. Wiger, and A. Holck. Frequency of disinfectant resistance genes and genetic linkage with blactamase transposon Tn552 among clinical staphylococci. Antimicrob. Agents Chemother. 46:2797–2803. 2002. 99. Sidhu, M. S., E. Heir, H. Sorum, and A. Holck. Genetic linkage between resistance to quaternary ammonium compounds and beta-lactam antibiotics in food-related Staphylococcus spp. Microb. Drug Resist. 7:363– 371. 2001. 100. Sidhu, M. S., H. Sorum, and A. Holck. Resistance to quaternary ammonium compounds in food-related bacteria. Microb. Drug Resist. 8: 393–399. 2002. 101. Simjee, S. In chickens and poultry farms, antibiotic resistance genes are present in the absence of antibiotic use as they are found on the same piece of DNA as heavy metal resistance genes. In: Proc. 44th Interscience Conference on Antimicrobial Agents and Chemotherapy, Washington, DC. 2004. 102. Singer, R. S., L. A. Cox, J. S. Dickson, H. S. Hurd, I. Phillips, and G. Y. Miller. Potential risks and benefits of tylosin use in poultry. In: Proc. 44th Interscience Conference on Antimicrobial Agents and Chemotherapy, Washington, DC. 2004. 103. Singer, R. S., J. S. Jeffrey, T. E. Carpenter, C. L. Cooke, E. R. Atwill, W. O. Johnson, and D. C. Hirsh. Persistence of cellulitis-associated 171 E. coli DNA fingerprints in successive broiler chicken flocks. Vet. Microbiol. 75:59–71. 2000. 104. Singer, R. S., R. Reid-Smith, and W. M. Sischo. Stakeholder position paper: epidemiological perspectives on antibiotic use in animals. Prev. Vet. Med. 73:153–161. 2006. 105. Smith, D. L., A. D. Harris, J. A. Johnson, E. K. Silbergeld, and J. G. Morris, Jr. Animal antibiotic use has an early but important impact on the emergence of antibiotic resistance in human commensal bacteria. Proc. Natl. Acad. Sci. U S A. 99:6434–6439. 2002. 106. Smith, D. L., J. A. Johnson, A. D. Harris, J. P. Furuno, E. N. Perencevich, and J. G. Morris, Jr. Assessing risks for a pre-emergent pathogen: virginiamycin use and the emergence of streptogramin resistance in Enterococcus faecium. Lancet Infect. Dis. 3:241–249. 2003. 107. Smith, H. W. The incidence of infective drug resistance in strains of Escherichia coli isolated from diseased human beings and domestic animals. J. Hyg. (Lond) 64:465–474. 1966. 108. Smith, H. W. The effect of the use of antibacterial drugs on the emergence of drug-resistant bacteria in animals. Adv. Vet. Sci. Comp Med. 15:67–100. 1971. 109. Smith, H. W., and M. A. Lovell. Escherichia coli resistant to tetracyclines and to other antibiotics in the faeces of U.K. chickens and pigs in 1980. J. Hyg. (Lond) 87:477–483. 1981. 110. Snary, E. L., L. A. Kelly, H. C. Davison, C. J. Teale, and M. Wooldridge. Antimicrobial resistance: a microbial risk assessment perspective. J. Antimicrob. Chemother. 53:906–917. 2004. 111. Sojka, W. J., and R. B. A. Carnaghan. Escherichia coli infection in poultry. Res. Vet. Sci. 2:340–352. 1961. 112. Sorum, M., P. J. Johnsen, B. Aasnes, T. Rosvoll, H. Kruse, A. Sundsfjord, and G. S. Simonsen. Prevalence, persistence, and molecular characterization of glycopeptide-resistant enterococci in Norwegian poultry and poultry farmers 3 to 8 years after the ban on avoparcin. Appl. Environ. Microbiol. 72:516–521. 2006. 113. Stutz, M. W., and G. C. Lawton. Effects of diet and antimicrobials on growth, feed efficiency, intestinal Clostridium perfringens, and ileal weight of broiler chicks. Poult. Sci. 63:2036–2042. 1984. 114. Summers, A. O. Generally overlooked fundamentals of bacterial genetics and ecology. Clin. Infect. Dis. 34(Suppl 3):S85–S92. 2002. 115. Swann, M. M. The Joint Committee on the Use of Antibiotics in Animal Husbandry and Veterinary Medicine. Her Majesty’s Stationery Office, London. 1969. 116. Thiele, E. H., E. L. Dulaney, M. J. Carey, and D. Hendlin. Detection of R factor-bearing microorganisms in laboratory animals. J. Bacteriol. 95:1184. 1968. 117. Thomas, C. M., and K. M. Nielsen. Mechanisms of, and barriers to, horizontal gene transfer between bacteria. Nat. Rev. Microbiol. 3:711–721. 2005. 118. Viola, C., and S. J. Devincent. Overview of issues pertaining to the manufacture, distribution, and use of antimicrobials in animals and other information relevant to animal antimicrobial use data collection in the United States. Prev. Vet. Med. 73:111–131. 2006. 119. Vose, D., J. Acar, F. Anthony, A. Franklin, R. Gupta, T. Nicholls, Y. Tamura, S. Thompson, E. J. Threlfall, M. van Vuuren, D. G. White, H. C. Wegener, and M. L. Costarrica. Antimicrobial resistance: risk analysis methodology for the potential impact on public health of antimicrobial resistant bacteria of animal origin. Rev. Sci. Tech. 20:811– 827. 2001. 120. Wallinga, D. Public health advocate. Prev. Vet. Med. 73:221–228. 2006. 121. Walton, J. R. The public health implications of drug-resistant bacteria in farm animals. Ann. N Y Acad. Sci. 182:358–361. 1971. 122. Watanabe, T. Infective heredity of multiple drug resistance in bacteria. Bacteriol. Rev. 27:87–115. 1963. 123. Welton, L. A., L. A. Thal, M. B. Perri, S. Donabedian, J. McMahon, J. W. Chow, and M. J. Zervos. Antimicrobial resistance in enterococci isolated from turkey flocks fed virginiamycin. Antimicrob. Agents Chemother. 42:705–708. 1998. 124. Wierup, M. The Swedish experience of the 1986 year ban of antimicrobial growth promoters, with special reference to animal health, 172 R. S. Singer and C. L. Hofacre disease prevention, productivity, and usage of antimicrobials. Microb. Drug Resist. 7:183–190. 2001. 125. Wireman, J., C. A. Liebert, T. Smith, and A. O. Summers. Association of mercury resistance with antibiotic resistance in the gramnegative fecal bacteria of primates. Appl. Environ. Microbiol. 63:4494– 4503. 1997. 126. Witte, W. Selective pressure by antibiotic use in livestock. Int. J. Antimicrob. Agents 16(Suppl 1):19–24. 2000. 127. Zhao, S., J. J. Maurer, S. Hubert, J. F. De Villena, P. F. McDermott, J. Meng, S. Ayers, L. English, and D. G. White. Antimicrobial susceptibility and molecular characterization of avian pathogenic Escherichia coli isolates. Vet. Microbiol. 107:215–224. 2005. ACKNOWLEDGMENTS R. Singer was partially supported by USDA National Research Initiative Competitive Grant 00-35212-9398. C. Hofacre was partially supported by USDA National Research Initiative Competitive Grant 01-35212-10877.