Antimicrobial Use and Resistance in Pigs and Chickens: A Review of the Science, Policy and Control Practices from Farm to Slaughter – Executive Summary

Over the past century, antimicrobials have evolved from the panacea for bacterial infections to an endangered tool because of rising resistance. Resistant bacteria in animals are one source of antimicrobial resistance (AMR) for humans – our strongest connection to animals is our daily food. The present report summarized and evaluated the scientific knowledge and pertinent national policies on AMR and antimicrobial use (AMU) in pigs and chickens. 
 
The three objectives of the present report were as follows: first, to serve as a reference for public heath, regulatory and agricultural policy-makers when debating this topic; second, to inform public health practitioners of notable AMR risks to the Canadian population related to food and agriculture; and third, to identify key research and policy gaps that compromise our ability to control this problem. 
 
The literature cited in the present comprehensive review was identified by systematically applying a priori search strings and medical subject headings, developed by the authors in conjunction with a professional librarian, to citation databases (Agricola, CAB, EMBASE, Medline and Scopus). Citations were restricted to English-language publication since 1990. After the removal of duplicates, more than 2000 potentially relevant citations were downloaded into commercially available software (RefWorks, USA). Titles were independently reviewed by two veterinary epidemiologists, and pertinent publications were reviewed in full. Additional ad hoc searches were conducted to identify grey literature or address specific questions. In total, 539 documents were cited. 
 
Humans can acquire antimicrobial-resistant bacteria from animals through food, but the risks of clinical disease, bacteria transmission to humans and horizontal transmission of resistance elements from bacteria in food to bacteria in the human intestinal tract have not been quantified. Likewise, the relative amounts that AMU in animals and humans each contribute to AMR in bacteria carried by humans have not been quantified. To address these questions, future research should expand on both ends of the current farm-to-fork research continuum. This could be achieved by studying the human health outcomes caused by resistant bacteria and investigating why producers, nutritionists and veterinarians use antimicrobials and what factors influence their decisions. Additional research on how various on-farm practices, only one of which is AMU, affect the rate and severity of foodborne diseases in general, and resistant bacterial infections in particular, would broaden on-farm food safety knowledge. The outcomes of such research would enable policy and interventions to expand from a specific focus on AMU and AMR to improving overall food safety. 
 
Antimicrobials are a necessary tool for appropriate veterinary care of food animals. Certainly, some AMU could be eliminated while still humanely raising animals; however data are lacking on the reasons why antimicrobials are used and the need for prophylactic and metaphylactic AMU to protect animal welfare. This has made distinguishing between necessary and inappropriate AMU contentious. The agriculture industry and regulators must continue to address this problem together so that AMU policy can preserve appropriate AMU, thus ensuring the production of safe food and humane rearing of livestock while eliminating inappropriate AMU practices and still ensuring economic viability of the industry. 
 
The causal relationship between AMU and AMR is complex. Use of some antimicrobials in some species in certain situations has clearly resulted in resistance in some bacteria. However, what has happened with one animal species, bacterium, antimicrobial or management system has not always happened in others. Evidence of increased resistance with increased AMU has been far more consistent than evidence of decreased resistance following cessation of use. Incomplete understanding of selective pressures for resistance in livestock, AMR transmission rates between animals and people, and management practices that help or hinder the emergence and persistence of resistance on farms has made evaluating the risk posed by AMU in livestock to public health contentious. Consequently, evidence-based interventions have been elusive and controversial. 
 
Links through travel and trade have made AMR in people and animals a global problem. International agencies have supported national solutions to this worldwide issue through prudent AMU guidelines, surveillance strategies and standardized risk-assessment techniques. Canada has established an integrated AMU and AMR surveillance program in humans and the major meat commodities. This program, along with an active research community, has provided government and industry strategists with a scientific foundation for decisions. Scrutiny of our veterinary drug regulations has resulted in preliminary policy and regulatory changes to eliminate inappropriate antimicrobial access and use, but this process has been slow and leaves much to be done. The Canadian government is responsible for ensuring the health of Canadians through policies and regulations that ensure safe food production. It must carry out this responsibility, while concurrently supporting a sustainable livestock industry that can produce food in a financially and environmentally sound manner. This task requires continued commitment to evidence-based policy and advocating for other nations to do the same. 
 
At the time of the present report, no prudent AMU education programs for producers or nutritionists existed in Canada. Education is desperately needed for the people who initiate much of the agricultural AMU in Canada. Yet concerns over inappropriate AMU have often overshadowed the good news stories in the industry. Agriculture and veterinary medicine have made advancements in animal health that decrease their reliance on antimicrobials. Livestock producers have embraced improvements in sanitation, nutrition and vaccine technology. Biosecurity has enabled flocks and herds to become negative for diseases that are endemic in the industry – diseases that were traditionally controlled with antimicrobials. Producer organizations have taken a proactive stance on food safety. Both chicken and pork industries have self-mandated on-farm food safety programs. These programs support best management practices and appropriate AMU. In conjunction with these farm-based improvements, the slaughter and meat processing industry has made substantial advancements in reducing the bacterial contamination of meat. In North America, the rate of foodborne diseases declined after slaughter and processing plants implemented hazard analysis and critical control point systems. These interventions have undoubtedly decreased the burden of illness in people from resistant bacteria in food, but the gains have not been quantified because our surveillance typically ends at ‘the fork,’ rather than with health outcomes. 
 
 
Although great progress has been made, there is still an enormous amount of work to be completed. We have advocated for continued research, but have also recognized that such research is ineffective without improvements in knowledge management. Novel techniques must be found to systematically assimilate the immense volume of discrepant research, ensure results are evaluated in context, and distribute the contextualized output to practitioners and policy-makers. Beyond this, the main recommendations from the present report were as follows: 
 
 
Seek and support research on the effectiveness of interventions including, but not limited to, AMU withdrawal to mitigate existing AMR. 
 
 
Seek and support research that expands on the current ‘farm-to-fork’ approach to account for diverse human health outcomes. 
 
 
Advocate for fair, transparent veterinary drug regulations, and AMR and AMU monitoring around the world based on scientific evidence, risk assessment, and appropriate precaution to ensure free and open trade of safe meat products. 
 
 
Commit to transparent policy evaluation to ensure Canada’s veterinary drug regulations result in prudent and safe AMU. 
 
 
Deliver AMU education to producers and nutritionists. 
 
 
Foster an innovative and collaborative relationship among regulators, public health officials and the agriculture industry.

Over the course of a century, antimicrobials have evolved from the panacea for bacterial infections to an endangered tool because of rising resistance. Resistant bacteria in animals are one source of antimicrobial resistance (AMR) for people and Canadians' strongest connection to animals is our daily food. This report summarizes and evaluates the scientific knowledge and pertinent policy around the world on AMR and antimicrobial use (AMU) in pigs and chickens. This document will serve as a reference document for public heath, regulatory, and agricultural policy makers and will hopefully initiate intense discussion among and between these groups. Public health practitioners may find it informative regarding notable AMR risks to the Canadian population related to food and agriculture. Key research and policy gaps have been identified that compromise our ability to control this problem.
It is irrefutable that people can acquire antimicrobial resistant bacteria from animals through food. What remains unclear is the frequency that pathogenic and commensal bacteria are transmitted to humans and either cause disease or transfer resistance elements to bacteria in people. A related knowledge gap is the relative amounts that AMU in animals and humans each contribute to AMR in humans. Future research should expand on both ends of the current farm-tofork research continuum. This could be achieved by studying health outcomes caused by resistant bacteria, and at the other end investigating why producers, nutritionists, and veterinarians use antimicrobials and the factors that influence their decisions. This research should be coupled with studies evaluating how various on farm practices, only one of which is AMU, affect the rate and severity of foodborne diseases in general and resistant bacterial infections in particular. By broadening on farm food safety knowledge, policy and interventions can expand from a specific focus on AMU and AMR to improving overall food safety.
Antimicrobials are a necessary tool for appropriate veterinary care of food animals. Certainly some AMU could be dispensed with while still humanely raising animals; but the data distinguishing why antimicrobials are used, and their range of effects, are limited so evaluating the appropriateness of agricultural AMU is contentious. For example, AMU to improve productivity is not necessary for good animal welfare yet prophylactic and metaphylactic use arguably could be. The agriculture industry and regulators must continue to address this problem together so that AMU policy can preserve appropriate AMU, thus ensuring the production of safe food and humane rearing of livestock, while eliminating inappropriate AMU practices but still ensuring economic viability of the industry.
Indisputably, AMU causes AMR. Yet, it is equally indisputable that the relationship is complex. Use of some antimicrobials in some species in certain situations clearly results in resistance in some bacteria. However, what happens with one animal species, bacterium, antimicrobial, or management system does not necessarily happen in others. Evidence showing increasing resistance with increasing AMU is far more consistent than evidence showing declining resistance with cessation of use. Our understanding is rudimentary of the complex selective pressures for resistance in livestock, the AMR transmission rates between animals and people, and the management practices that help or hinder the emergence and persistence of resistance on farms. Such knowledge gaps make evaluating the risk posed by AMU in livestock to public health a topic ripe for debate. Consequently, evidence-based interventions are elusive and controversial.
Links through travel and trade make AMR in people and animals a global problem. International agencies are supporting national solutions to this worldwide issue through prudent AMU guidelines, surveillance strategies and standardized risk assessment techniques. At home, Canada operates an integrated AMU and AMR surveillance program in humans and the major meat commodities. This program, along with an active research community, is providing government and industry strategists with a scientific foundation for decisions. Scrutiny of our veterinary drug regulations has resulted in preliminary policy and regulatory changes to eliminate inappropriate antimicrobial access and use, but this process has been slow and leaves much to be done. The Canadian government is responsible for ensuring the health of Canadians through policies and regulations that ensure safe food production. It must carry this responsibility out while concurrently supporting a sustainable livestock industry that can produce food in a financially and environmentally sound manner. This task requires continued commitment to evidence-based policy and advocating that other nations do the same.
Without doubt, unnecessary and inappropriate AMU occurs in agriculture. There is currently no prudent AMU education for producers or nutritionists. Education is desperately needed for the people who initiate much of the agricultural AMU in Canada. Yet concerns over inappropriate AMU often overshadow the goodnews stories in the industry. Agriculture and veterinary medicine have made advancements in animal health that decrease their reliance on antimicrobials. Livestock producers have embraced improvements in sanitation, nutrition, and vaccine technology. Biosecurity enables flocks and herds to remain negative for diseases that are endemic in the industry-diseases that were traditionally controlled with antimicrobials. Producerorganizations have taken a proactive stance on food safety. Both chicken and pork have self-mandated onfarm food safety programs. These programs support best management practices and appropriate AMU. In conjunction with these farm-based improvements, the slaughter and meat processing industry has made substantial advancements in reducing bacterial contamination of meat. In North America, the rate of foodborne diseases declined after slaughter and processing plants implementing hazard analysis and critical control point (HACCP) systems. These interventions undoubtedly decreased the burden of illness in people from resistant bacteria in food but have not been quantified as our surveillance ends at 'the fork,' rather with than health outcomes.
Although great progress has been made, there is still a great deal of work to be completed. We have advocated for continued research but also recognize that such research is ineffective without improvements in knowledge management. Novel techniques must be found to systematically assimilate the immense volume of discrepant research, ensure results are evaluated in context, and distribute the contextualized output to practitioners and policy makers. Beyond this, the main recommendations from this report are as follows: Seek and support research into the effectiveness • of interventions, including but not limited to AMU withdrawal, to mitigate existing AMR Seek and support research that expands on the • current 'farm-to-fork' approach to account for diverse human health outcomes Advocate for fair, transparent, veterinary drug • regulations, AMR and AMU monitoring around the world based on scientific evidence, risk assessment, and appropriate precaution to ensure free and open trade of safe meat products Change Canada's veterinary drug regulations to • ensure prudent and safe AMU while committing to transparent policy evaluation Deliver antimicrobial use education to producers • and nutritionists Foster an innovative and collaborative relationship • between regulators, public health officials and the agriculture industry Arguably, one of the most important discoveries in medicine was the development of antimicrobials. It was not until the start of the 'antibiotic era' that human and veterinary medicine became equipped to deal successfully with various bacterial diseases. The introduction of antimicrobials had profound effects on patient morbidity and mortality and ushered in an era of great optimism. The optimism was so great that in 1967, the U.S. Surgeon General somewhat infamously declared that it was "time to close the book on infectious diseases" (1). In hindsight, statements such as that seem ludicrous as the pendulum has now swung in the opposite direction. Some people have expressed concern that the end of the 'antibiotic era' may be near and questioned whether continued emergence of multidrug-resistant pathogens represents an 'unwinnable war' (2,3). Like most situations, the reality likely lies between those two extremes, but concerns over the emergence and dissemination of antimicrobial resistant pathogens in human and veterinary medicine are significant.
It was not long after antimicrobials became widely used in human medicine that antimicrobial resistance emerged in human pathogens. Biologically, there was no reason to believe that the response of bacteria in livestock to antimicrobial exposure would be any different and, indeed, the use of antimicrobials in animals has certainly led to the development and dissemination of AMR in various animal pathogens and commensal organisms. It is irrefutable that antimicrobial use (AMU) is a key component for development of antimicrobial resistance (AMR). However, it is equally clear that there is not a simple and consistent relationship between AMU and AMR. Use of some antimicrobials in some species in certain situations clearly results in resistance in some bacteria. However, something that happens with one animal species, bacterium, drug, or management system does not necessarily happen in all others, illustrating the complexity of the field. Responses to previous efforts to control AMR based solely on AMU have met with variable responses because of this complex and rather unpredictable biological variation, indicating the need for detailed interdisciplinary (and international) study to optimally address this pressing issue.
Antimicrobial use in animals and subsequent identification of AMR in pathogens and commensal bacteria has raised numerous concerns. The greatest involves the potential for AMU in animals to be reflected in AMR in human pathogens. The closest contact that most people have with livestock is through handling and consumption of food, and that is generally the main route of exposure of humans to bacteria of food-animal origin. As AMR increases in animals, and resistant bacteria contaminate food, the potential for human exposure increases. However, just as AMU can lead to AMR, and the presence of resistant bacteria in animals can lead to infections in people, defining what 'can' represents in terms of true incidence and risk is problematic. More objective information is needed to quantify the health risk and to develop interventions for the transmission of resistant bacteria to humans through food.
A broad, multidisciplinary approach is needed to address the relationships between AMU and AMR because this topic is constantly evolving with respect to management practices, AMU, pathogen distribution, monitoring methods, and research techniques. This document provides an overview of issues, concerns, research, policies, regulations, strategies, and legislations pertaining to AMU and AMR in swine and poultry, as well as an assessment of critical gaps in knowledge and efforts. Using a review and critical assessment of the literature for relevant research, policies and strategies that can, or have been used to address AMR in swine and poultry, along with consultation of relevant experts, this document takes a stepwise approach to understanding and assessing this topic. The first step is a basic description of AMR ecology, including mechanisms of resistance and means of acquisition or transmission of resistance, in order to provide the scientific background required for assessment of this field. An understanding of zoonotic pathogens of particular concern with respect to AMU and AMR in swine and in poultry is critical, but so is recognition of the role of the commensal microflora in AMR acquisition and transmission. Zoonotic bacteria of lesser concern are also discussed, as is the potential role of some emerging pathogens that may become a significant concern to the swine and poultry industries.
An understanding of current and historical antimicrobial use practices in livestock is provided, including why antimicrobials are used and how they are administered. The strength of evidence on the appropriateness of AMU in livestock production is evaluated. The logical extension of this is the merging of the two initial topics, and an assessment of links between AMU and AMR in poultry and swine. An important aspect of this topic is critical assessment of the current literature to address the contentious question of the extent that AMU in swine and poultry contributes to AMR in human pathogens.
Various policies, regulations, and legislations have been developed or enacted in an attempt to control AMU and AMR, with varying degrees of evidence and efficacy. The complexity of the field, differences in perspectives and absence of objective data in many key areas have led to much controversy regarding previous, current, and proposed control methods. The various organizations involved in AMU and AMR and their roles are described. This includes both international bodies and national systems. Additionally, the surveillance activities of these groups are compared. The varied approaches between groups underlie the differences in concern, perceptions, resources, and leadership in different countries and organizations. The structure and function of current systems are important for assessing current practices. There is particular focus on the Canadian perspective including surveillance, responsibility for AMU and AMR regulation, current regulatory practices and issues, and identified strengths and weaknesses within this system.
While objective data regarding the links between AMU and AMR are variable and information on effective interventions is sporadic and sometimes contradictory, both the livestock industry and regulatory bodies have developed various approaches to reduce the impact of AMR in chickens and pigs on human health. Key interventions are highlighted. Potential therapies such as prebiotics and probiotics are attractive but are largely unproven and under-investigated. Management measures to reduce the need for antimicrobials through improved animal health are an obvious area of interest. Management can have a tremendous effect on animal health, but research in many aspects is limited and a broad approach balancing animal health, animal welfare and production economics is needed, thereby complicating this area. Research on these topics is evaluated and presented with a description of the current production, management and marketing system differences between swine and poultry that can have an effect on feasibility, outcomes, and required measures. Additionally, current AMU guidelines that have been developed are discussed in terms of their content, practical application and potential consequences.
Zoonotic diseases will certainly remain as a pressing concern in human medicine, and animal diseases will be similarly important in veterinary medicine. Balancing human and animal health is a complex concept as measures taken in one group can have unpredictable or unknown effects in the other. Clearly, there are no straightforward answers for control of this issue. Cessation of all antimicrobial use is not practical for animal health, animal welfare, and production economics. Continuation of the status quo is similarly unacceptable because of evidence regarding the impact of agricultural AMU on some aspects of human disease. Identification of the appropriate middle ground is required to optimize human health while maintaining humane, safe, and economically viable swine and poultry production. Rhetoric, opinions and anecdotes are not acceptable foundations for widereaching policy, regulation, and legislation, but current knowledge gaps create an environment where such subjective information can have major impact. This document identifies important concepts and practices that should be considered for setting priorities, developing and implementing policy, regulation and legislation, and educating all relevant stakeholders.

Methods
This project reviews the literature on the ecology of antimicrobial resistance (AMR), AMR in zoonotic and commensal bacteria with the potential to cause foodborne disease, antimicrobial use (AMU) in pigs and chickens, and the surveillance of AMU and AMR in pigs and chickens. It evaluates the literature describing the relationships between AMU and AMR as well as the policies, strategies, and interventions to control AMR in bacteria carried by chickens and pigs. Throughout this document, the term 'antimicrobial' is defined as, "any substance of natural, semi-synthetic, or synthetic origin that kills or inhibits the growth of a microorganism but causes little or no damage to the host. " (4) and is used in the context of inhibiting or killing bacteria.
Publications utilized in this project were obtained either through publication databases subscribed to by the University of Saskatchewan library or through known resources (both written and personal) that the authors were aware of, or in contact with, as a result of their expertise in this area. The authors preferentially utilized peer-reviewed scientific literature. When publications were unavailable or insufficient, the alternatives of reports, websites and personal communications were employed. Relevant databases were identified and prioritized to ensure the literature search was comprehensive but not redundant.
The study inclusion and exclusion criteria were defined (Table 1) and used to establish a priori search terms, search strings, and medical subject heading (MeSH) terms (Appendix 1). Searches were restricted to English publications and the years 1990 to 2009. A professional librarian conducted each search, recorded the identified citations, and acquired necessary full texts. Three databases were searched systematically using consistent search terms: CAB, Embase, and Medline. Two databases were searched systematically with terms tailored to the database content: Agricola and Scopus. These searches returned a large number of citations (> 3,500) and a high rate of duplicates (approximately 40%), thus further databases were only used to address specific inclusion criteria. The citations identified by the searches were evaluated using a priori relevance screening criteria that elaborated on the inclusion and exclusion criteria. All relevant publications were briefly summarized as to which inclusion criteria they addressed.
Because of the volume of publications identified, a supplementary search strategy was applied to ensure critical publications were prioritized. A series of searches was completed using only the Medline database to identify articles indexed by the following: i) terms similar to foodborne, food, animal origin, antimicrobial resistance, or antibiotic resistance; and terms similar to Escherichia coli, Enterococcus, Salmonella, Campylobacter; MRSA or Clostridium difficile; and terms similar to swine, pig, pork, poultry, broiler, or chicken; ii) terms related to antimicrobial use in the broiler or swine industry, microbial ecology, or relationship between AMU and AMR and; iii) terms designed to look at the potential association between AMR in chickens and pigs and AMR and health in humans. This search was limited to publications since 1999 to emphasize the most current developments in the field. Again, the search strings and number of citations identified were recorded. Duplicate articles were removed and the results were saved in an electronic reference manager software (Refworks ©). The resulting data files were exported into a custom designed database, using Microsoft Access, for reviewers to describe and search the content of each paper for relevance to the project.
Two veterinary epidemiologists independently reviewed citation titles and abstracts and excluded citations that did not meet the inclusion criteria. The title and abstract of the remaining citations were used to describe the publication by a series of dichotomous variables that included the bacterial species, livestock species, stage of production or processing, AMU data, and type(s) of AMR data (prevalence, associations with human health, microbial ecology or relationship between AMU). The independent reviews were combined and discrepancies discussed and agreed upon. The full publications were acquired and reviewed by the author(s) of pertinent sections.
Following the conclusion of these searches, and while describing and evaluating the literature, authors occasionally conducted ad hoc searches of publication databases, the Internet, or known grey literature. These results of ad hoc searches were not recorded. All authors reviewed the entire document. An external review was graciously provided by veterinary epidemiologists specializing in swine, poultry, antimicrobial resistance, and surveillance techniques from the Public Health Agency of Canada.
In addition to the journal articles, original research was described in 21 conference proceedings, 19 reports, and 5 websites. Thus, 297 (55%) of the citations refer to original research. The remaining 242 citations (45%) were government reports (72, 33%), scientific reviews (136, 56%) and grey-literature (27, 11%). The proportion of citations that are original research is disproportionately low relative to the proportion of original research citations identified in the a priori literature searches because many of the grey literature citations and government reports were intentionally sought out to address the project objectives.

Introduction
People are increasingly affected by antimicrobial resistant bacteria; they can be exposed to bacteria through many environmental connections, one of which is the food they eat daily. Through food, people can be exposed to antimicrobial resistant zoonotic pathogens and commensal bacteria. Foodborne bacteria typically originate in the animal's gastrointestinal tract and reach people through fecal contamination of meat at slaughter. The gastrointestinal tract provides an ecological reservoir that supports a diverse bacterial population. Most bacteria in this population are beneficial or neutral to the animal host and are not the target of antimicrobial drug use. However, antimicrobial exposures intended to kill pathogens or improve growth also affect these bacteria. Resultantly, the normal flora becomes an unintended casualty or develops/acquires resistance. When people consume enteric bacteria through contaminated food, the bacteria are returned to a similar ecological niche and can either cause disease or share resistance elements with the diverse bacterial flora resident in people (5-7).

Scope and Objectives
The objective of this section is to provide sufficient knowledge of the hazards to public health, the source of bacteria, and the resistance types of greatest concern to facilitate the evaluation of the programs, policies, and strategies to control AMR that have been created by regulators and the agriculture industry. This chapter is limited to describing antimicrobial resistance (AMR) in bacteria that infect people through food. It is restricted to discussing resistance as it pertains to chicken and pigs/pork. This limitation in scope does not infer that these commodities pose a different risk to people than other meat commodities. After describing how people are exposed to resistant bacteria, the biology of AMR is explained. The main body of this chapter is devoted to describing concerning pathogen/ resistance combinations. Some are well known, like Campylobacter and Salmonella, while others are either of secondary importance to Canadians, such as Yersinia enterocolitica, or emerging food safety concerns such as Clostridium difficile. The relevance of resistance in these foodborne resistant pathogens for Canadians is the focus of this chapter.

The Link Between AMR Bacteria in Animals and Humans
Bacteria are ubiquitous in our environment and often infect multiple animal species including humans. The probability that people will be infected or colonized by bacteria from animals is affected by the route of exposure, frequency of exposure, the exposure dose, and the host-adaptation of the bacteria. Bacteria carried by animals that can cause disease in humans are termed 'zoonotic. ' The disease caused by these zoonotic bacteria can be exacerbated if the bacteria are resistant to antimicrobials (8-10). In addition, antimicrobial resistant bacteria can be a source of resistance elements for bacteria harboured by people (11)(12)(13). Direct physical contact, shared environments, and exposure through vectors and fomites are all routes for bacterial transmission between animal species. This review focuses on foodborne bacteria.
The dominant link between most Canadians and viable bacteria from animals is likely food consumption. Resistant bacteria undoubtedly spread into the environment through aerosol and animal wastes (14-16). However, modern livestock production has greatly reduced the frequency and extent of contact between humans and agricultural animals. Only 2.2% of Canadians live in the farm population; the remaining 97.8% have limited direct contact with animals and their environment is largely separate from the airspace and properly managed waste of livestock (17). People in direct contact with animals face different hazards and have different risk factors for acquiring AMR bacteria and, while further study is required to clarify the extent that resistance spreads to people by these routes, these links are beyond the scope of this project.
Antimicrobial resistance directly affects human health when infection with pathogenic bacteria leads to illness that requires antimicrobial therapy and the selected therapeutic is ineffective due to resistance. This results in prolonged illness with potentially more severe symptoms. The worst case scenario is a bacterial infection that is refractory to all available treatments leading to death (9,10,18). Although treatment failure is the most obvious effect from AMR, other health hazards exist due to AMR in bacteria. Numerous studies show the risk of hospitalization and the severity of disease are greater in people acquiring a resistant bacterial infection compared to those with a similar but susceptible strain (19)(20)(21)(22)(23). This increases the cost of treatment and the burden on the health care sector, and places people at risk for exposure to other nosocomial infections (10,24). The increased disease severity may partially be related to ineffective treatment early in the disease and subsequent disease progression. But increased hospitalization and disease severity is reported even after accounting for ineffective therapy. This could be due to co-selection of virulence traits, up-regulation of virulence traits, or improved fitness of resistant strains (25,26). Antimicrobial resistance also increases the incidence of foodborne disease as people taking antimicrobials for any reason have an increased likelihood of being colonized by resistant bacteria because the therapeutic drug alters the body's normal flora and concurrently selects for resistant bacteria (9,18).
Two types of resistant foodborne bacteria contribute to health burdens. Resistant pathogenic bacteria directly contribute to all four effects described above: treatment failure, increased hospitalization rates, increased disease severity, and increased disease incidence. Resistant commensal bacteria indirectly contribute to the problem by harbouring and spreading resistance genes to bacteria pathogenic to people (27)(28)(29)(30). Transmission can theoretically occur from commensal bacteria originating in livestock to pathogens in humans, or vice versa, from commensal bacteria in humans to pathogenic bacteria in livestock. Although this contribution is indirect, it may pose a risk to humans that is equal to or even greater than that posed by pathogenic bacteria. Commensal bacteria are ubiquitous in healthy animals and can contaminate carcasses at slaughter. The high prevalence of these bacteria drastically increases the probability of exposure compared to pathogenic bacteria.
introduction to AMR ecology

Resistance Development
Antimicrobials either kill (bacteriocidal) or inhibit (bacteriostatic) bacteria. Not all antimicrobials are effective against all bacteria: bacteria that are intrinsically resistant lack the structural or functional cellular mechanisms that are required for the antimicrobial to act (4,31). Intrinsic resistance is a genus or species-specific property of bacteria (32). While it is necessary to understand that some bacteria are inherently resistant to certain antimicrobials, this is not the focus of this review; therefore, the remainder of this report will pertain to acquired AMR. Acquired resistance occurs due to a change in the bacterial genome. The two major ways that susceptible bacteria acquire AMR are through mutation or horizontal acquisition of foreign genetic material (31,33).

Mutation
Mutation is a spontaneous change in the genome resulting in a susceptible bacterium becoming resistant, usually during replication (33). Chromosomal mutations often result in structural changes to the bacterial cell wall which subsequently confers resistance (4,31). Mutation may lead to dramatic resistance development or to slower more gradual resistance development depending on the antimicrobial agent affected (4,31). Mutants may be disadvantaged compared to the parent and, therefore, less able to survive in the population in the absence of the selective pressure of an antimicrobial. Alternatively, mutants may be as or more viable than the original strain and may persist in the population with or without selective pressure from the antimicrobial (34). The emergence of resistance from mutational events happens at high frequencies for drugs such as streptomycin, nalidixic acid, and rifampin and has not been reported for others such as vancomycin and polymixin B (4).

Horizontal Transfer of Resistance
The horizontal transfer of resistance genes from donor to recipient bacteria is a second method through which bacteria can acquire resistance. Transformation, transduction, and conjugation are the three primary means for the horizontal transfer of resistance genes (35).
Transformation is the uptake of naked bacterial DNA from the environment by acceptor bacteria (4,35). It is an important method of gene transfer in vitro but less important in vivo (32). Transformation generally occurs between closely related genera and may result in gene recombination producing new forms of resistance genes. This method of resistance transfer is particularly important in bacteria species such as Streptococcus and Neisseria that have a high frequency of natural transformation (4).
Transduction is the transfer of resistant genes via a bacterial virus or phage (4,35). This is thought to be a relatively unimportant method of resistance transfer because bacteriophages are very specific to the bacterial host and can carry a limited amount of DNA; but occasionally, resistance plasmids can be accidentally packed into phage heads during phage assembly and subsequently be able to infect new cells by injecting plasmid DNA into a recipient cell (4,32). Neither transformation nor transduction requires a viable donor cell or a link between donor and recipient (31).
Conjugation is the transfer of resistance genes from a donor to a recipient bacteria through a temporary protein channel (4,35,36). Gene transfer in conjugation allows the spread of mobile genetic elements such as plasmids, transposons, or integrons/gene cassettes (35-37). These elements can possess multiple AMR genes and may be responsible for the rapid dissemination of resistance genes among different bacteria (28,38,39). Linked clusters of resistance genes on a single mobile element can aggregate in such a way that antimicrobials of a different class or even non-antimicrobial substances like heavy metals or disinfectants can select for AMR bacteria (40,41). Exchange of resistance genes between pathogens and non-pathogens or between gram-positive and gram-negative bacteria has been documented (40).

The Spread of Resistance: Mobile Genetic Elements
Of the three mechanisms for horizontal transfer of resistance genes: transformation, transduction, and conjugation, conjugation is undoubtedly the most important to understand. The acquisition of genetic elements such as plasmids, transposons, or integrons/gene cassettes is a critical part of horizontal transfer of AMR because these elements connect and re-assort resistance mechanism thus enabling the spread and establishment of resistance elements in bacteria populations. These elements vary considerably from each other in regard to their carriage of resistance, their replication, and transmission.
Plasmids are extra-chromosomal circular DNA that can replicate independently of chromosomal DNA. When resistance is transferred via plasmids, a copy of the plasmid is retained by the donor cell. Most plasmids carry the gene required for conjugation, but not all do. In these cases, plasmids can be mobilized by using the conjugal apparatus of other self-transmissible plasmids within the cell (4,32). Plasmids can harbour resistance genes for between one to ten different antimicrobials (multiple AMR) (4). Multi-resistant plasmids are often the result of interplasmidic recombination, integration of transposons, or insertion of gene cassettes (32). All resistance genes on a multi-resistant plasmid are transferred when the plasmid is transferred, whether there is selective pressure for all of the resistance genes on the plasmid or for just one of the resistance genes (32). Plasmids can act as vectors for transposons and integrons/gene cassettes (36).
Transposons (jumping genes) are short sequences of DNA that can move from plasmid to plasmid, or from plasmid to chromosome, and vice versa. Transposons do not possess replication systems and must be incorporated into chromosomal DNA or plasmids (32). Unlike plasmids, no copy of the transposon remains within the original cell as the transposon moves between the donor and recipient. All transposons can move and integrate into foreign DNA by non-homologous recombination, which permits the same transposon to be found in the genome or plasmids of highly unrelated organisms (4).
Integrons are a mobile element often found on plasmids and are distinct from transposons. They are a site-specific recombination system consisting of an integrase enzyme, a gene-capture site, and a captured gene cassette(s). Each gene cassette encodes a single resistance gene and a specific recombination site (4,41,42). An integron's site-specific integrases recognizes gene cassettes and catalyzes their insertion at a specific site. Repetition of this sequence results in integrons linking together multiple resistance gene cassettes (4).
Gene expression of an integron is dependent on various factors including promoter strength, gene copy number, the relative distance of the gene cassette from the promoter, and the presence of additional internal promoters. Expression is usually mediated via a common promoter situated upstream (5'-end) of the gene cassettes, rather than through individual promoter copies. Higher levels of gene expression can be achieved if a second promoter is included adjacent to the first, or if the gene in question is included as multiple copies. The relative distance between a gene cassette and the promoter plays a significant role regarding expression; proximal genes tend to be expressed more effectively than distal genes. As a result, distal genes may have very little effect on the susceptibility of the host bacterium to relevant antimicrobials (43,44). Integron carriage of resistance gene cassettes by host bacterium can be dependent on the environment; host bacteria can potentially lose integron-borne resistance genes in the absence of antimicrobial selective pressure (45).

Resistance Selection: Direct Selection, Cross-resistance and Co-selection
As mentioned earlier, the development of AMR is a complex process and the speed with which it develops depends on the bacteria involved, the selective pressure, and the availability and transferability of resistance genes (32). Recent studies have shown that the majority of multiresistant phenotypes are obtained by the acquisition of external genes that may provide resistance to an entire class of antimicrobials (46). Antimicrobial use can select for antimicrobial resistant bacteria in three ways: direct selection, cross-resistance, and co-selection.
Direct selection is the most simplistic form of selective pressure and occurs when a drug selects for bacteria resistant to it. For example, tetracycline is used and tetracycline resistant bacteria survive. Cross-resistance occurs when the expression of one antimicrobial resistance gene infers resistance to several related drugs that have similar targets or mechanisms of action. The blaCMY-2 gene provides an example of cross-resistance. This gene confers resistance to many potentiated ß-lactams (ampicillin and amoxicillinclavulanic acid) and cephalosporins (ceftiofur, cefoxitin, and ceftriaxone). Therefore, exposure to ceftiofur will both select for bacteria carrying this gene and indirectly increase the frequency of resistance to ampicillin. Cross-resistance is also common in the macrolide and fluoroquinolone classes (4).
Co-selection is the phenomenon of antimicrobial use selecting for resistance to completely unrelated drugs. Bacteria with multiple resistance genes can survive exposure to any drug affected by these genes. Therefore, the use of any of these drugs perpetuates resistance to all of the unrelated antimicrobials for which bacterium possesses resistance genes. (4).
Co-selection has important implications for policies designed to eliminate existing resistance and contributes to the complex relationship between antimicrobial use and resistance development (use of drug A can select for resistance to drug A; but because of co-selection, it may also be selecting for resistance to drugs B, C, and D).  10% (50,53). Although less significant than C. jejuni, C. coli can rank among the top four causes of enteric infection in people and should not be discounted as a food-safety concern (54).

Phenotype versus Genotype
Campylobacter primarily cause sporadic disease and to date, an effective typing system has not been developed. Together, these factors make source attribution difficult (55). Poultry is considered a leading source for foodborne infections while the role of pork is less clear (56-59). Chickens generally carry C. jejuni as part of their commensal flora and pigs typically harbour C. coli (56,(60)(61)(62)(63). Although both pigs and chickens commonly carry Campylobacter at slaughter, retail meat sampling consistently identifies much higher Campylobacter recovery rates from poultry than red meats (64-66), presumably because chicken intestines are more friable and prone to breaks that can cause carcass contamination. Case control studies routinely aggregate C. jejuni and C. coli cases. This may cause risk factors unique to C. coli to be missed because C. coli accounts for a smaller proportion of cases and could partially explain why pork is not consistently identified as a risk factor for Campylobacter infection (67).
It is believed that resistant cases of human campylobacteriosis generally arise from acquiring a resistant strain, versus human therapeutic drug use selecting for resistance (23,68). Initially, this seems counterintuitive given that resistance arises rapidly following therapy in people. However, AMU in uncomplicated cases is contraindicated so exposure in people should be limited. In contrast, antimicrobial consumption prior to an infection can increase the risk of a resistant infection in people (69), which suggests that human AMU provides a competitive advantage to already resistant strains. Campylobacter are ubiquitous in healthy pigs, so long durations of AMU can result in a prolonged and pervasive selective pressure on the Campylobacter population to acquire resistance and establish resistant strains. In reality, attributing resistance in Campylobacter to agricultural AMU is difficult as it requires data on AMR in isolates prior to therapy and should account for AMU prior to infection. The resistance rates in Campylobacter isolated from Canadians are available from regional research projects (70-72), but not from national surveillance. The most recent results from American surveillance found over half of the Campylobacter tested were resistant to one or more Clinical Laboratory Standards Institute (CLSI) antimicrobial classes and 13.6% were resistant to two or more subclasses (73).
Campylobacter's most concerning resistance types are macrolides and fluoroquinolones. Ciprofloxacin is a frontline drug for undifferentiated gastroenteritis, so fluoroquinolone resistance may cause treatment failure. High level fluoroquinolone resistance occurs with a single-step chromosomal mutation in the gryA (60,74). Resistance to fluoroquinolones in C. jejuni from poultry and humans is very common in certain parts of the world, including Mediterranean Europe and South East Asia (53,(75)(76)(77). In comparison, almost complete susceptibility to ciprofloxacin is reported in C. jejuni from Australia, Norway, and Sweden (78-81). In poultry, emerging resistance in C. jejuni correlates with the use of fluoroquinolones and this resistance does not appear to impose a fitness cost, and may even be advantageous (34,49,82). No fluoroquinolones are licensed for use in pigs or chickens in Canada although injectable enrofloxacin could be used in an extra-label manner. The rates of ciprofloxacin resistance in C. jejuni from retail chicken ranged from 0 to 2.3% from different parts of the country in 2006 (66). National surveillance for AMR in Campylobacter from pigs is not conducted in Canada, but Canadian research projects have reported 2.4 and 10% ciprofloxacin resistance in Campylobacter from pigs on farms (83,84).
Resistance to macrolides is also concerning in Campylobacter because this antimicrobial class is prescribed for severe cases of campylobacteriosis or immunocompromised cases. This has raised concern over widespread macrolide use in veterinary medicine (85,86). From American human cases, 1.6% of the C. jejuni and 3% of the C. coli were resistant to macrolides (87). The rate of resistance in human infections is highly variable world wide; rates as high as 51% in Singapore and 80% in Nigeria have been reported in Campylobacter isolates from children (88).
Macrolide resistant Campylobacter are relatively rare in chickens (generally <1%) but are prevalent in pigs (generally >50%) (66,79,89). This difference is partly due to the predominance of C. coli in pigs and C. jejuni in broilers; C. coli are genetically predisposed to macrolide resistance. This species characteristic is so strong that 40% of Campylobacter from antibiotic-free swine farms have been identified as macrolide resistant. This could reflect intrinsic resistance, persistence of acquired resistance from historical AMU, or transfer of resistance that arose from AMU in other groups of pigs on the farm (90,91). Campylobacter from pigs also demonstrate more macrolide resistance due to extensive macrolide use in swine (84). Comparing macrolide use between pigs and chickens is hampered by different drug-use metrics, but over half of Canadian swine producers use macrolides in feed while less than 5% of chickens in America are exposed to macrolides (66,85,92,93).
The food safety hazard posed by macrolide resistant Campylobacter could plausibly increase in the future. Over 20 erm genes have been identified, which are the genes responsible for many of the mutations in the rRNA genes that cause cross-resistance between macrolides, lincosamide and spectinomycin. These erm genes exist in gram-positive and gram-negative bacteria and many are located on transmissible genetic elements. Acquisition of these genes, along with Campylobacter's ability to pick up heterologous DNA through transformation, makes the likelihood of macrolide resistance establishing in C. jejuni a concern (88). Because C. jejuni predominates in human disease, this could have a far greater human health impact than the current high rates of macrolide resistance found in C. coli. The most direct threat posed by AMR in Salmonella is treatment failure following use of an antimicrobial to which the infecting strain is resistant. Accordingly, resistant infections are associated with higher case fatality ratios and increased hospitalization rates and durations (19,20,104,105). In addition to more severe disease, AMR may cause a higher disease incidence. A Danish model found that relative to their prevalence, quinolone-resistant Salmonella caused more disease than would be expected than quinolone-susceptible Salmonella. Also relative to their prevalence, multidrug-resistant Salmonella were associated with more illness than would be expected compared to pan-susceptible Salmonella. This trend was consistent across all serovars modelled. It was proposed that this trend could be due to enhanced ability to survive food processing or increased susceptibility in people consuming antimicrobials for other indications (98). Collectively, these studies demonstrate that antimicrobial resistance in Salmonella creates a health burden additional to baseline salmonellosis and that the increased burden is not limited to a particular serovar.

Salmonella
Public health officials recognize two resistance types in Salmonella pose an undue threat to human health. These include resistance to fluoroquinolones, the main treatment for invasive salmonellosis, and resistance to newer generation cephalosporins, the indicated treatment for salmonellosis in pregnant women and children (86,106). Historically, fluoroquinolone resistance was conferred by two stepwise chromosomal mutations, and therefore, resistance only developed when bacteria were exposed to a fluoroquinolone and subsequently only transmitted vertically to progeny. Over the last few years, plasmid-mediated fluoroquinolone resistance genes have been reported in Europe, Asia, and the United States (107-111).
The emergence of horizontally transmissible fluoroquinolone resistance genes raises two concerns. First, these can be transmitted between Salmonella and E. coli in vitro,and presumably,in vivo (107). This exponentially increases the bacterial reservoir that can harbour these genes and eliminates species or geographical barriers to the dissemination of resistance. So far reports of plasmid-mediated quinolone resistance genes in animals are rare, but it appears inevitable that these genes will become established in livestock and poultry (74,107). This will further exacerbate control efforts regarding fluoroquinolone resistance. The second concern is that the emerging qnr genes that moderate fluoroquinolone resistance can be associated with mobile genetic elements with novel combinations of resistance genes including extended spectrum ß-lactam resistance genes (ESBL) (107,112,113). This drastically increases the likelihood of treatment failure as these are the two drug classes most commonly used to treat Salmonella. Potentially of equal concern is that use of drugs outside the fluoroquinolone class can now supply the selective pressure necessary to establish and disseminate fluoroquinolone resistance.
Public health authorities are also concerned about ESBL resistance in Salmonella. In the 1980s, resistance was predominantly due to broad spectrum ß-lactamases that hydrolyzed penicillins and older generation cephalosporins (TEM-, SHV-, CTX-M-).
In the 1990s, resistance to newer cephalosporins emerged through two routes: one was via minor mutations of broad-spectrum ß-lactamases to become extended spectrum ß-lactamases and the other was through the emergence of the plasmid-mediated Ambler class C (AmpC) enzymes, predominantly mediated through CMY-genes. Each enzyme hydrolyzes a slightly different set of cephalosporins (114).
In humans, resistance types tend to cluster geographically, but extended spectrum ß-lactamases and AmpC resistance both occur in Europe and North America. In animals, extended spectrum ß-lactam resistance genes predominate in Salmonella in Europe with the few reported cases of AmpC-type resistance in European livestock linked to imported animals (111,114 and tetracycline (ACSSuT) (119,120). The ACSSuT resistance was carried on a chromosomal island, thus was vertically disseminated within this serovar (120). Yet, the predominance of blaCMY-2 in S. Heidelberg demonstrates that certain Salmonella serovars can have an affinity for particular plasmid-mediated resistances or plasmids. Salmonella serovars and clones can shift independent of AMU, but shifts can also be influenced by AMU (121). This makes it difficult to understand how AMU is affecting Salmonella epidemiology in general and resistance in particular. The relationship between serovar and AMR also makes it difficult to generalize about the prevalence or patterns of resistance in Salmonella. This challenge is addressed by limiting discussions to a specific serovar or, as will be discussed, circumvented by using generic E. coli as an indicator for Salmonella in particular and gram-negative bacteria in general.

E. coli
Salmonella and Campylobacter clearly demonstrate that bacteria can be transferred from animals to people through food. These zoonotic pathogens cause disease in people, and if the strain is resistant, it can increase the disease burden. However, these bacteria represent a small fraction of the total possible spread of resistance elements to people through food. The normal bacterial flora responds to antimicrobial exposures by establishing resistant populations. Fecal contamination of meat at slaughter is one mechanism that can result in human exposure to these bacteria, and once consumed, genetic exchange with pathogenic or commensal bacteria in the human gastrointestinal tract can occur. While transmission rates remain unknown, the sheer volume of these bacteria makes many experts believe that the global threat from AMR may be more greatly impacted by this commensal reservoir than emerging resistance in pathogens (12,122).
Fecal contamination of food is generally monitored through recovery of E. coli (123). Expanding on the role as an indicator of meat contamination, E. coli from healthy animals have also been adopted as an indicator of AMR for gram-negative bacteria. E. coli are used to understand resistance in bacteria with complex ecologies such as Salmonella. Although a relationship does exist between serotype and resistance types in E. coli, this relationship is largely ignored in E. coli isolated from healthy animals (aka generic E. coli). This is valid because, as opposed to pathogen overgrowth, a diverse population of E. coli can concurrently exist in the gut which differs from the clonal populations seen with pathogen overgrowth. The ubiquity, simple isolation, and diversity of E. coli makes it as a good model for understanding AMR in Salmonella because the genes operating in Salmonella also operate in E. coli and can be transmitted between these bacterial species (27,107).
Although E. coli is a useful model for AMR in Salmonella, differences certainly exist. The same resistance phenotypes, and often genes, operate in these bacteria, but their behavior and transmission can differ. For example, the genes encoding the ACSSuT phenotype in S. Typhimurium are located on the bacterial chromosome yet when this resistance phenotype is identified in E. coli, the genes are typically located on plasmids. Furthermore, the value of E. coli for understanding AMR in Salmonella does not extend to predicting resistance in Salmonella based on resistance in E. coli on a farm or regional level (124). This limitation is at least partly due to differences in the location and transmission of these genes in each bacterium as well as an inability to account for the relationship between resistance and Salmonella serovar.
Escherichia coli are used to study the selective pressure on the gram-negative bacterial population to develop and retain resistance. The intestinal normal flora is not the target of antimicrobial treatments, but nevertheless these bacteria are exposed and become resistant-much like civilian casualties in war. The degree of resistance in E. coli mirrors this selective pressure on gut bacteria (48,125,126). Thus the greatest utility of studying AMR in E. coli is improved understanding of the health risks posed by antimicrobial use.
As an indicator organism, the main resistance outcomes of concern in E. coli are those that are a concern in Salmonella. Resistance to fluoroquinolones and newer generation cephalosporins are monitored closely. In contrast to the relatively rare reports of plasmid-mediated fluoroquinolone resistance in Salmonella from the agri-food sector, several reports of plasmid-mediated resistance in E. coli exist (108,127-129 (25,132). VTEC infections are predominantly foodborne and associated with beef. Human illness from VTEC is largely independent of AMR. Antimicrobials are contraindicated in VTEC cases because this can induce release of the verotoxins, so treatment failure is a relatively minor concern (133). The attributable fraction (i.e. infection in people taking antimicrobials that select for a resistant foodborne pathogen) also seems small as studies that have compared AMR rates in VTEC and non-VTEC from healthy animals have found lower rates of resistance in the VTEC isolates (although higher rates of resistance may be reported in isolates from sick animals, these have likely been exposed to therapeutic AMU and are not representative of the background rate of bacteria that may contaminate food) (134,135). In pigs, the concern that antimicrobial use could select for virulent E. coli was raised because statistical relationships were identified between resistance and virulence genes (26,136). The virulent isolates were obtained from sick pigs and thus raised an animal health question. However, similar relationships between resistance and virulence genes in E. coli from healthy animals would be a food safety issue because this would suggest antimicrobial use could increase the prevalence of virulent isolates and increase foodborne disease risks.  (29). But the role of food as a source of human disease remains uncertain because illness is rarely caused by the same strains of E. faecium as are found in livestock, and communityacquired infections are rare (149,156,159,160). While ecological connections between VRE in pigs, chickens, community carriers, and ill people appear to exist, much remains to be learned about their relative importance. For example, E. faecium collected over a decade from hospitals, community infections, swine and poultry were compared, and some clonal groups contained isolates originating from more than one source, but overall the groups largely held isolates from one source (161). No risk assessments were identified describing the probability of avoparcin use in livestock and poultry resulting in treatment failure from vancomycin-resistant E. faecium. This is now an academic curiosity, as avoparcin has been removed from the world market (162).
In the late 1990s, a streptogramin antimicrobial, quinupristin/dalfopristin (QD) (Synercid®), was released to treat multidrug-resistant gram-positive infections. Primary indications included vancomycinresistant E. faecium (E. faecalis are innately resistant) and MRSA infections (162). Shortly after the release of QD, streptogramin-resistance was recognized in E. faecium. Once again, agricultural antimicrobial use was questioned because virginiamycin (a streptogramin) is used as a feed-grade antimicrobial. Virginiamycin use can select for the vat(D) or vat (E) genes which infer resistance to dalfopristin. These genes do not cause full resistance to virginiamycin or QD because they affect streptogramin A but not B. However, streptogramin B targets the same ribosomal subunit as macrolides and lincosamides and can be affected by resistance mechanisms in this antimicrobial class (163). Thus QD resistance may be selected for by virginiamycin and macrolide use in livestock. Prior to the release of QD, virginiamycin was not related to any antimicrobials used in humans.
European countries banned virginiamycin on the precaution that it could select for resistance to QD. The United States and Australia both conducted quantitative risk assessments evaluating the human health risk from virginiamycin use in food-producing animals (162,164). In 2004, Australia revoked the label for virginiamycin use for growth promotion and/or improved feed conversion and required modifications to the label for prophylactic and therapeutic claims. In contrast, neither the United States nor Canada changed the label indications or drug availability. In North America, virginiamycin continues to be extensively used in poultry and minimally used in pigs (66,92,93,165 Yersinia enterocolitica primarily causes sporadic infections so source attribution is difficult (97). Outbreak investigations and meat contamination rates suggest that pigs are the main livestock reservoir for foodborne infections (64,169,170). Yersinia enterocolitica is a commensal in pigs. Other species also carry Y. enterocolitica, but these belong to the non-pathogenic 1A biogroup. On a ranked basis, the reported prevalence in slaughter hogs tends to correlate with reported human incidence rates: Germany, 67%; Canada, 42%; United States, 13% (167,170,171). However, caution should be used when comparing these prevalence estimates because the sampling and laboratory methods differed between studies. Data were not available from many countries, making this comparison incomplete.
Extra-intestinal Yersinia infections can require antimicrobial therapy. Broad-spectrum cephalosporins with or without aminoglycosides are generally effective as are fluoroquinolones, trimethoprimsulfonamide combinations and doxycycline. Antimicrobial susceptibility to ß-lactams is serogroup specific with some strains possessing chromosomally mediated ß-lactamase genes (168). The resistance conferred by these genes to penicillins and cephalosporins is not entirely predictable but does not extend to newer generation cephalosporins such as ceftriaxone or cefoxitin.
German and Canadian surveys of Yersinia from pigs found that resistance to drugs other than ß-lactams or erythromycin is rare (167,170). Isolates from meat in Greece largely corroborated this, with no reported resistance to 3rd and 4th generation cephalosporins or fluoroquinolones. Low rates of resistance were noted to streptomycin, tetracycline, and co-trimoxazole, but isolates were obtained from a variety of sources and were not exclusively Y. enterocolitica. Statistical associations were identified between certain resistance phenotypes and virulence genes (172). Similar findings in E. coli have been interpreted as an increased potential for antimicrobial use to select for virulent strains (26,48,136). Therefore, interventions that minimize AMU in pigs could potentially decrease the morbidity attributed to Yersinia.

Listeria monocytogenes
The Listeria genus consists of six species, but only L. monocytogenes is an important foodborne pathogen. Mild infections typically present with flu-like symptoms. More severe cases can cause septicemia, meningitis, and miscarriage or stillbirth. The case fatality rate from L. monocytogenes can be as high as 30% (173). The incidence of L. monocytogenes infections is low relative to other foodborne bacterial pathogens. Europe and the United States report 0.3 cases per 100,000 people (97,166). People with impaired T-cell immunity are at increased risk for infection. This group includes young children, elderly people, pregnant women, and the immunocompromised (97,173).
Main reservoirs of L. monocytogenes include soil, forage, and water. Environmental contamination and cross-contamination of food in meat processing facilities are key factors in foodborne transmission of L. monocytogenes. The structure of modern food processing systems means that L. monocytogenes typically causes outbreaks rather than sporadic infections (173). Many domestic animals can harbour L. monocytogenes, which can lead to retail meat contamination. Retail meat testing in Ontario found 7% of pork chops and 28% of skin-on chicken breast were contaminated (96). Ready-to-eat foods are at increased risk of causing disease because this bacterium can multiply at refrigeration temperatures. Canada, the United States (US) and the European Union (EU) prohibit the sale of products containing ≥ 100 cfu/g of L. monocytogenes (106,174,175).
Effective treatment includes early diagnosis and antimicrobial therapy. Choice antimicrobials are typically a penicillin or ampicillin in combination with an aminoglycoside. Cases allergic to ß-lactams may be treated with a trimethoprim/sulfonamide combination, tetracycline, or chloramphenicol (176). Vancomycin and erythromycin are indicated for use in pregnant women (173). To date, AMR has not been a major concern with clinical L. monocytogenes infections. Studies that have examined L. monocytogenes from meat have found almost complete susceptibility to penicillin, ampicillin and aminoglycosides. Resistance to tetracycline and fluoroquinolones is more variable and high rates of resistance to sulfonamides, but not trimethoprimsulfonamide combinations, have been reported (176)(177)(178)(179)(180)(181). Listeria monocytogenes are intrinsically resistant to cephalosporins (106).
While resistance to clinically relevant drugs is currently very low, it is still of some concern given that not all patients can be treated with penicillins. Case reports of resistant infectious strains exist, and isolates with transposons and plasmids carrying antimicrobial resistant genes have been identified (176,182,183). From these, it has been determined that L. monocytogenes can acquire resistance elements from other gram-positive bacteria, particularly Enterococcus and Streptococcus while transiting the gastrointestinal tract (176,183). Enterococci carrying multi-resistance to ß-lactams and aminoglycosides exist which makes the potential for resistant Listeria monocytogenes that are refractory to treatment a concern. Monitoring of this pathogen will be important to ensure human therapy is predictable and to allow interventions should resistance become a problem, which would be devastating in such a virulent pathogen.

emerging issues
The following section presents two bacteria that do not directly fit this project's inclusion criteria of foodborne resistant bacteria. Despite this, it is imperative to discuss methicillin-resistant Staphylococcus aureus (MRSA) and Clostridium difficile because concerns are escalating over the public health risk from these bacteria in food animals. For both bacteria, concerns are being fuelled by a lack of information on the hazard. These bacteria remind us of the importance of 'thinking outsidethe-box. ' Neither pathogen has been traditionally considered zoonotic. Due to unique concerns and limited information on these novel problems, the organization of the following sections differs from the preceding bacteria.

Methicillin-Resistant Staphylococcus Aureus
Staphylococcus aureus are gram-positive bacteria that can colonize the skin or nose and mouth of people and animals. Staphylococcus aureus is a common commensal in humans and can be found in the nasal passages of approximately 30% of healthy individuals. Colonization rates are variable in other species. Colonized individuals are a source of infection to others and are at greater risk for infection themselves. ßeta-lactamase-mediated penicillin resistance is very common in S. aureus, so ß-lactamase-resistant penicillins are a main treatment for community-onset S. aureus infections. There may be various other therapeutic options as a range of antimicrobials are potentially effective against S. aureus. However, S. aureus has a tendency to develop or acquire antimicrobial resistance and multidrug-resistant staphylococci are a leading health concern in human medicine. Much of this is focused on methicillin resistance. This resistance, conferred by the mecA gene, provides resistance to all ß-lactam antimicrobials including methicillin, penicillins, cephalosporins, and carbapenems. This gene is carried within the staphylococcal cassette chromosome mec (SCCmec) and, once acquired, is stable within the bacterial chromosome. There are major lineages within S. aureus that can show host specificity for humans or animal species; this specificity extends to MRSA. Within a geographical area certain lineages or clones dominate, and these can differ between hospitals, the community, and agricultural settings. The clones associated with each setting tend to have somewhat predictable phenotypic characteristics and virulence traits (184). This discussion focuses on a specific clone relevant to food safety, hereafter referred to as livestockassociated MRSA (LA-MRSA). Other descriptions of MRSA are limited to the minimum necessary to provide context for understanding the role of LA-MRSA as a potential foodborne pathogen.
Staphylococcus aureus is an animal pathogen. Because of the clonal lineage, diseases in humans and animals were historically considered unrelated. This view has changed over the last decade with the recognition that dogs, cats, and horses can be colonized with and clinically affected by pathogenic MRSA strains that are indistinguishable from human strains (190)(191)(192). Epidemiological evidence suggests that infection in these species resulted from a spill-over from humans; such transmission has been dubbed a 'humanosis. ' These animal infections create a human health risk because people exposed to carrier animals can be infected by MRSA.
Concern over MRSA in animals expanded when a clonal strain, sequence type 398 (ST398), was identified in pigs (193). Further study identified a high prevalence of this clone in pigs, as well as some other livestock types, and an association between human infection or colonization with this strain and contact with food animals. Subsequently, MRSA associated with this clone has been referred to as LA-MRSA. In contrast to the situation in companion animals, the LA-MRSA clone is distinct from previously common human strains. Since its identification, LA-MRSA has been found in pigs in Austria, Belgium, Canada, Denmark, France, Germany, the Netherlands and Singapore. As summarized by the European Food Safety Authority (EFSA), the reported prevalence of LA-MRSA in pigs is high in these countries (184): Europe, 10% to 39%; United States, 70%; Canada, 25% (194)(195)(196)(197)(198)(199). Caution should be exercised in comparing these prevalence estimates because sampling and laboratory methods were not standardized. To date most investigations have focused on swine, thus pigs are often considered a primary reservoir of LA-MRSA. However, veal calves, dairy cows, and broiler chickens can be positive and may be important in the epidemiology of this clone (184). In general, LA-MRSA does not cause clinical disease in colonized agricultural animals.
For people working directly with food animals, LA-MRSA is an occupational hazard. Direct contact with colonized livestock can result in human colonization or infection with . Colonization risk increases with exposure: the highest rates occur in people that work directly with pigs (29%), followed by people entering barns but not working with pigs (12%), then people not working with pigs but living on positive farms (2%) (200). LA-MRSA has been associated with skin and soft tissue infection, pneumonia, and septicemia in people. Across the EU, LA-MRSA accounts for 0.7% of typed MRSA isolates, but the rate varies between countries; higher rates are reported in countries that have low overall rates of MRSA and high exposure to pig farming (184).
The zoonotic potential of LA-MRSA has raised concerns that LA-MRSA could be a food safety hazard. Food has served as a vector in a hospital outbreak of MRSA but food handlers, rather than animals, were considered the primary source (201). Theoretically, colonized animals could contaminate carcasses and meat during slaughter and create a source of exposure for people. MRSA has been detected in numerous foods of animal origin including chicken and pork (202-207). The majority of these studies found only human-related strains, which is suggestive of meat contamination from food handlers. The exceptions were a single pork sample by van Loo et al. and the study by de Boer et al. which found more than 90% of the MRSA from chicken, pork and a variety of other meats were LA-MRSA (204,205). Although a large proportion of tested meat samples were positive for LA-MRSA, the degree of contamination was below 10 cfu/g. Similar results have been found for retail meat in Canada, with MRSA being identified in retail pork, and beef, albeit at very low concentrations (208). Currently, beyond the 1995 report suggesting, but not proving, that food prepared by a colonized hospital worker was the source of an MRSA outbreak in a hospital (201), there are no reports of food as a source of human MRSA infection.
LA-MRSA is frequently multidrug-resistant, but the pattern differ from CA-MRSA and HA-MRSA. Resistance to tetracyclines is observed in almost 100% of LA-MRSA isolates, and resistance to trimethoprim, but not trimethoprim in combination with a sulfonamide, is common. Fluoroquinolone resistance has been reported from 0% to 35% (197,209). Based on this multi-resistance pattern, therapeutic treatment options are limited to valuable classes of drugs including the oxazolidinones and pleuromutilins. Resistance to these drugs can be conferred by the cfr gene. This gene affects the binding sites of 23S RNA and infers resistance to antimicrobials with this target site including oxazolidinone, pleuromutilins, phenicols, lincosamide, and streptogramin A. Two S. aureus isolates, one of which was a LA-MRSA clone carrying the cfr gene, were identified in pigs in Germany. Although not an indication of its prevalence, this report heightens concerns over limited treatment options should LA-MRSA become more established in agricultural animals and subsequently people (210).
The European Food Safety Authority has determined that current evidence indicates the food safety risk from LA-MRSA is small and is much lower than the risk from exposure to colonized animals or humans (184). Staphylococcus are killed by heat and will be eliminated with cooking or pasteurization. This limits the food exposure risk to unpasteurized dairy products and meat that is consumed raw or minimally cooked. The risk of exposure from handling meat also appears low as reports of colonization in meat handlers are infrequent.

Clostridium difficile
Clostridium difficile is the most commonly identified cause of antimicrobial-associated diarrhea and health care-associated (HA) diarrhea in most areas worldwide and is responsible for virtually all cases of pseudomembranous colitis (211,212). Clostridium difficile infection (CDI, previously called C. difficileassociated diarrhea or CDAD) is also emerging as an important cause of community-associated (CA) diarrhea with dramatic increases in the incidence rate reported over the last decade (213,214). Severe CDI is now being reported in populations previously considered to be at low risk with unique features that had not described including close contact transmission, high recurrence rate, young patient age, bloody diarrhea, and lack of antimicrobial exposure (215-217).
The source of infection of people in the community is unclear. Various sources are possible, including colonized individuals, people with CDI, the environment, pets, food, and water (218-227). This section of the report focuses on the potential role of food in CA-CDI. The main factor that has led to suspicion of foodborne transmission is the similarity of food animal and human CA-CDI strains. Currently in people, PCR ribotype 027 is the most important clone while ribotype 078 is increasingly attracting attention. Both of these clones are found in HA and CA-CDI cases (228-235). Although still preliminary, toxinotype V strains, predominantly ribotype 078, may be overrepresented among CA-CDI isolates (228,229,236).
The following section exclusively discusses C. difficile in pigs and chickens. Readers should be aware that parallel knowledge is emerging in other meat-animal commodities. Ribotype 078 has been reported in pigs from different countries (237,238). In an American study, 83% of C. difficile isolates from pigs were ribotype 078 (237). Although far less investigated, C. difficile has been identified in chickens (239,240). Reports of C. difficile in food animals led to concerns about the potential for retail meat contamination. An American study reported C. difficile contamination in 47% (21/45) of examined pork products (225). In that study, 67% of positive pork products harboured ribotype 078 while the remainder harboured ribotype 027. A study of retail pork using samples collected from four Canadian provinces only identified C. difficile in 1.4% (4/296) of ground pork and 3.1% (3/97) pork chops (241). The most common strain was ribotype 027 and ribotype 078 was not found. A second Canadian study of ground pork found C. difficile in 12% (14/115) of samples but most samples had low numbers of C. difficile and were only positive with enrichment culture (242). Where numbers could be determined, samples typically contained only 20-240 spores per gram. While the infectious dose of C. difficile is not known, it is assumed that low levels are of lesser concern than high levels. Further, one considers that C. difficile can be found in treated water, vegetables, and the household environment, so it is clear that simple exposure to low levels of C. difficile is not the sole factor in the pathophysiology of disease (219, 243,244).
Determining the relevance of C. difficile in food animals to CA-CDI will require much more investigation. Unanswered questions include the reasons for different ribotypes in American and Canadian livestock studies and the correlation between C. difficile types found in retail meat and those found in healthy food animals at slaughter. More work is also needed on the prevalence of contaminated meat and the degree of contamination. If meat is commonly contaminated with C. difficile at levels sufficient to cause disease, it could be important in the epidemiology and control of CA-CDI.
The concerns over antimicrobials and C. difficile differ from the other bacteria described in this report because therapeutic failure in humans is a minor concern. Instead, the concern with antimicrobial use is disruption of the normal intestinal flora and subsequent overgrowth of C. difficile. The recent recognition of a possible foodborne link and different concerns for AMR in C. difficile explain why studies to date have focused on elucidating potential transmission routes between food animals and people. Future studies may study AMR in C. difficile from food animals, with an application of understanding on-farm practices that affect the epidemiology in livestock populations.

Conclusion
This chapter has summarized the main foodborne hazards that humans face from antimicrobial resistant bacteria in food animals. It has focused on specific resistance types in individual bacterial species. This 'drug-bug' orientation is necessary to understand the issues that influenced many policies and regulations to control AMR. However, many scientists are evolving from a 'drug-bug' perspective to an ecological approach. This has been largely driven by the recognition that barriers to resistance exist only in our minds. Resistance elements move between bacterial populations and their animal hosts, between urban, rural and agricultural environments, and between bacterial hosts. Transmission and evolution can involve re-assortment of linked resistance genes, which consequently alters the selective pressures from co-selection. This has forced scientists and policy makers to seek solutions that extend beyond limiting the use of one antimicrobial in an attempt to mitigate a single resistance outcome.
This chapter also presented many foodborne links between AMR bacteria in pigs and chickens and humans. However, what is possible is not necessarily probable. Although experimental and observational studies have described many transmission routes, the probability of each occurring is often conditional on numerous sequential events. We felt it was important to describe these routes, but do not mean to leave the impression that every bacteria and resistance type described are in a crisis situation.
In conclusion, this chapter has presented numerous recent developments. These include the emergence of plasmid-mediated fluoroquinolone resistance, the description of the cfr gene, and an emerging connection to animals for the traditionally humanbased diseases of MRSA and C. difficile. These discoveries are shaping our approach to food safety and zoonotic disease. They remind us of the need to be constantly alert and that what we think we know today will likely be questioned tomorrow.

Chapter 2: Antimicrobial Use
Introduction Antimicrobial resistance (AMR) is an important issue facing both human and veterinary medicine. The increasing number of antimicrobial resistant pathogens in human medicine has raised both public and scientific interest. For human pathogens, most of the AMR development is considered attributable to the selection pressure from antimicrobial use (AMU) in people (245). However, the volume of antimicrobials used in food animal production has led to concerns in the public, regulatory, and scientific arenas that AMU in food animals is contributing to the AMR problem by creating a reservoir of resistant bacteria (134,246,247). For human health, the transfer of such resistance to zoonotic enteropathogens is of primary interest, but the development of antimicrobial resistance in animal pathogens, the associated subsequent loss of therapeutic options for veterinary medicine, and the potential need to use antimicrobials of greater importance in human medicine is also an important concern.
While AMR is an issue in veterinary medicine in Canada, its overall clinical impact is much less than in human medicine. However, the continued development of AMR in human medicine and the occurrence of AMR as a veterinary problem in other parts of the world indicate that AMR may become more of a concern in Canada with regard to continued therapeutic efficacy in veterinary medicine (248). The use of antimicrobials in animal agriculture is important for maintaining and improving animal health and welfare through disease prevention and treatment, and arguably for increasing carcass quality, as well as for enhancing the economic efficiency of growth and production. If the livestock industry loses efficacious antimicrobials because of resistance development, or experiences limited access because of tighter regulations, the consequences and costs to the industry are difficult to quantify but have the potential to be substantial.
The development and spread of AMR is a complex process involving interactions among antimicrobials, microorganisms, and the surrounding environment (249). There are multiple sources that may contribute to the creation and dissemination of AMR ( Figure 1).
While there is evidence indicating each pathway exists, our knowledge of the importance of each remains incomplete ( Figure 1) (12). There are many routes for AMR to spread within a local environment, including but not limited to, spread between humans, livestock, companion animals, wildlife, fish, water and soil, and vegetables (12), but the global aspect of AMR is also important to consider especially in the context of international travel, animal movement, and food trade.
The complex ecology of AMR development and dissemination can create considerable challenges in determining how much each potential source impacts AMR development in humans. As the issue of AMR is further explored and investigated, it becomes increasingly difficult to evaluate often conflicting results or statements with regard to the role of each of these potential sources. This point can be argued as being a fundamental question that needs to be addressed or a question of academic interest. It is the division between these two schools of thought that have lead countries to implementing either evidence-based AMR policy or a precautionary principle based-approach.

Scope and Objectives
In order to understand the potential impact that swine and poultry production may have on AMR development, the literature was examined and an overview was compiled. The objectives of this chapter are to: highlight the reasons for AMU in swine and broiler chickens, review the classes of antimicrobials used and the utility of continued availability of efficacious antimicrobials in Canadian swine and broiler chicken production, describe current known risk factors for AMR development in bacteria isolated from pigs and chickens and the potential relationship with AMR bacteria isolated from humans, and discuss the information gaps that still remain in our understanding of the effect of AMU in livestock on public health. This chapter provides examples rather than exhaustively listing all antimicrobial use practices, risk factors or proposed links between AMU in livestock and AMR development in humans. It is intended to provide a broad background for all readers regardless of their familiarity with agriculture and the issue of AMU in livestock and its impact on human health.
general Considerations, terminology, and Reasons for Antimicrobial Use Different AMU regimens can select for various resistance genes (251), and therefore, AMU patterns are expected to have some impact on the distribution of AMR phenotypes (252-254). Among the ramifications associated with resistance gene selection are the degree of resistance conferred and the carriage of linked resistance determinants (251). Sometimes only minimal antimicrobial exposure is necessary to select for continued persistence of resistance genes within enteric microflora (251). Figure 1. Epidemiology of the potential major pathways for antimicrobial resistance to transfer. After Linton (250), modified by R. Irwin, reproduced with permission.
Persistence of AMR in bacteria is generally related to the persistence of antimicrobial use/exposure, but AMR may also persist in the absence of known antimicrobial selective pressure. An example of this is provided by a study of the University of Kentucky swine farm. This farm ceased all antimicrobial use and then followed the animals housed on the farm over a period of 126 months. Resistance to tetracycline declined by less than 50% over this time, leading the authors to conclude that long-term withdrawal of antimicrobials failed to markedly reduce AMR and that long-term feeding of antimicrobials could lead to AMR that may not readily be reversed by withdrawal of antimicrobials (255). From this and other studies (256,257) it seems that some resistant bacteria can become stabilized in the intestinal tracts of animals and become the dominant intestinal flora potentially passing from one generation to the next (258). These findings therefore indicate the need to consider the long-term effects of AMU in any environment.

Antimicrobial Growth Promoter Terminology
Popular debate over AMU in agriculture often refers to drug use for growth promotion. While many deem 'growth-promotional AMU' inappropriate and potentially even unethical, what does this term mean? The phrase 'antimicrobial growth promotion/ promoter (AGP)' was a descriptive term coined early in the history of agricultural AMU to describe the overt effects of including drugs in the feed of apparently healthy animals. It became ingrained in AMU terminology in the 1970s when the European Union separated agricultural AMU legislation into two classes (See Chapter 3). One class contained non-prescription feed additives that producers could access without veterinary involvement and all other AMU (i.e. prophylactic, metaphylactic and therapeutic use of feed-grade, water-soluble, and injectable antimicrobials) required a veterinary prescription (259). In Europe, these uses were mutually exclusive meaning that an antimicrobial could not be licensed in both categories. Thus European references to AGP AMU pertain to a very specific and distinguishable type of AMU. It was these classes of drugs that were revoked in the 1999 and 2006 European Union drug bans (259,260).
The terminology surrounding AMU for growth promotion is less precise in North America.
Regulations have always permitted feed-antimicrobial labels to claim improved growth, disease prevention and therapy. In the 1970s and 1980s, the United States government investigated the potential implications to human health from feeding tetracyclines and penicillins to pigs and chickens (261,262). The evidence for potential human harm was deemed insufficient to justify action. As a result, neither Canada nor the United States changed the veterinary drug legislation to separate production and health indications and both countries have allowed continued access to these products without a veterinary prescription (263,264). So in North America, there are no drug use characteristics that reliably distinguish between AGP and other types of in-feed AMU. Describing in-feed AMU by the reported reason for inclusion is the only valid mechanism to proportion in-feed AMU into AGP versus other effects.
An outdated method of categorizing AMU as AGP is to consider all feed-grade antimicrobials included in diets at less than 200ppm for more than fourteen days to be AGP (261,265). Effectively, this means that in-feed AMU prevents sub-clinical disease. Throughout this document, the acronym AGP will exclusively refer to the regulated feed-grade antimicrobial additives within Europe prior to their prohibition. Outside of the European situation, all other in-feed AMU is either referred to by the reported purpose of use (including growth promotion) or, if that is not available, simply as in-feed or feed-grade AMU.

Reason for Antimicrobial Use in Livestock
Intensive livestock operations, such as swine or poultry operations, are typically confinement agricultural systems. These are defined as systems where "the movement of animals is confined and they are raised in high density, usually with stimulated feeding, and weight gain optimized so as to decrease time to mature weight. " (122). As demand for meat grows, so does the confined livestock production because it allows for economies of scale. Traditionally, raising large numbers of animals in a close proximity often required the use of the tools of prophylactic, metaphylactic, and growth promotional AMU in order to prevent morbidity and mortality, to ensure animal welfare, and for economic benefit.
It is not the need for AMU in livestock that is at question, but the extent to which antimicrobial use is necessary that is up for debate. Clearly animals can be raised without routine use of antimicrobials under some circumstances, but presumably there is always a need to access them if an animal or a group of animals requires treatment. How far the industry can go and needs to go in regularizing farming without antimicrobials is the question. The potential for increased mortality, days to market, feed costs, and overall increase in the cost of production may need to be passed on to the consumer resulting in higher food costs and precluding some individuals from accessing animal-based protein sources. Therefore, a balance between ensuring animal health and welfare, human health, and producing a reasonably priced safe and wholesome product needs to be achieved.
In swine, the majority of antimicrobial use is for treatment or prophylaxis of respiratory and enteric disease, while in poultry, antimicrobials are primarily used for intestinal infections, namely colibacillosis and necrotic enteritis (265,266). The method of administration and the volume of antimicrobial used will vary depending on the animal species, stage of production, and risk of disease. There are three primary reasons for AMU in food-producing animals: treatment of sick animals, prevention and control of disease, and improved productivity.
Prevention and control can be further divided into metaphylactic or prophylactic applications.
Metaphylaxis is a disease control measure involving the mass medication of a group of animals to prevent the spread of disease when only a few individuals have been identified as infected. Prophylaxis is a preventative treatment of an animal or group of animals at a time when it may be more susceptible to infection (267). Antimicrobials are given at critical points in production to help prevent the development of disease. Prophylactic treatment may involve the entire group of animals or may be targeted toward specific high risk individuals depending on the animal species, the production system, and the disease condition.
Growth promotion is another reason for AMU in livestock and it generally involves the use of antimicrobials licensed for this purpose. Normally, antimicrobials labelled for growth promotion are included at a dose lower than those approved for therapeutic purposes and are fed for a longer duration than antimicrobials used for prevention and control (263). Long-term mass exposure to antimicrobials can increase the selection pressure for AMR development and persistence (268-271).
Until recently in North America, the focus has been on minimizing the use of drugs of critical/very high importance in human medicine for all purposes including therapy and prophylaxis because these uses could increase selective pressure and the pool of bacteria resistant to these important antimicrobials (272,273). Lately, in the United States there has been increasing lobby to eliminate the use of other antimicrobials, particularly AMU for improved productivity rather than health purposes (274). This more closely mirrors Europe's premise that the human health risk from AGP does not justify its continued use. In Europe, AGP AMU was of particular concern because it was not strictly necessary for health, and as it was responsible for the majority of antimicrobial exposure by mass, it was presumed that this use caused most of the selection pressure leading to the indirect effect on the bacterial biomass and the role of reservoirs in AMR transmission (275).

Antimicrobial Use estimates in livestock Production
Many countries are attempting to gain a better understanding of the volumes of antimicrobials used in livestock production, but the problem is that these estimates can be misleading if the context is not understood. The main issue is that AMU data are difficult to collect and report. National, regional, or even farm level data are scarce. On a national level in numerous European countries, since antimicrobials are by prescription only, a central pharmacy database can provide information on the volume of AMU in animals (35,79,89). Other countries like Canada and the United States are still working on the best methodology for collecting these data since they do not have any legislation compelling pharmaceutical companies to provide this information nor do they have centralized pharmaceutical databases.
At the national level, AMU is often reported in kilograms or tonnes of active ingredient sold but use can also be reported by divided daily doses (DDD) (276), animal daily doses (ADD) (276), or as animal-units per treatment days (277). While each of the above methods tries to capture the true exposure of an animal to a drug, they all are limited and much debate still surrounds the best approach to reporting AMU information. Jensen et al. (276) provides a good overview and highlights the major potential methods for reporting drug use and the associated limitations.
In the case of reporting kilograms or tonnes of active ingredient, while these data provide the volume of drug used, they do not allow for assessment of how the drug was used and how this use may affect AMR. Data on the species drugs are delivered to, the number of animals exposed, the dose received, and the method of delivery are generally unavailable. Therefore, end-user data are often important to collect in order to gain a better appreciation of how and why antimicrobials are being used in livestock production.
While end user data can provide valuable information, it also has some serious limitations. Complete and accurate farm/hatchery-level AMU records are often difficult to obtain. In order to be able to generate useful national estimates, these data need to contain sufficient detail. Differences in record keeping methodologies and priorities result a lack of standardization and therefore summarizing information that is collected in multiple formats on individual farms is extremely challenging. Capture of use information can be expensive for the researcher and burdensome for the producer to accommodate especially during times of additional demands with limited resources. Under-reporting is potentially a problem since producers are busy with day-to-day operations on the farm and, therefore, record keeping may be relatively low on the priority list. Subsequently treatment records may be forgotten or incomplete. Dunlop et al. (277) reported a 35% under-reporting rate for AMU recorded by swine producers as compared to inventory and disappearance data collected by the research team. Reasons for underreporting include misunderstandings between researchers and producers and lack of time during periods of increased work load such as in disease outbreak situations (278). Current Canadian producers may be less prone to under-reporting as Dunlop et al's. study was conducted before on-farm food safety programs made AMU record keeping mandatory (See Chapter 4). Current producers have adjusted to maintaining AMU records but the accuracy of these records has not been assessed.
Although there are several challenges associated with AMU data collection in livestock, several countries are attempting to capture some AMU information. The next several paragraphs provide some estimates of AMU in animals from various countries. While reviewing these numbers, it is important to keep in mind the limitations of reporting tonnes of antimicrobials used so as to not over or misinterpret what these numbers are really telling us. Also, comparing one country to another is difficult because information on the denominator is lacking, i.e. the number of animals, or the type of animal and/or the dose given. Without this information, it may appear like one country is using considerably more antimicrobials in their livestock than another.
By not knowing how many animals were exposed and at what dose, etc., one cannot really compare one country or region to another, since this may equate to comparing apples to oranges. The main goal of presenting these data are to familiarize the reader with how these data are often presented and how they might be difficult to interpret and compare between regions and within species.
Estimates of antimicrobial use in Europe in 1997 were 10,494 metric tonnes (MT) of active ingredients (35). Human use accounted for 52%, animal treatments for 33%, and growth promotion for 15% (35). In-feed use accounted for the majority of the animal use with 90% of the antimicrobials administered this way (35). The breakdown of the total antimicrobial volume per species was as follows: pigs 60%, poultry and rabbits 20%, ruminants 18%, and fish and pets at 1%. For therapy, prevention, and control, the antimicrobial classes used were tetracyclines (66%), macrolides (12%), penicillin (9%), and others (12%) (35).
In the United States in 1989, the Institute of Medicine estimated that 50 million pounds of antimicrobials were produced in the States annually and 50% of this was used in animals (279). The Union of Concerned Scientists re-evaluated this and suggested that there are 35 million pounds of antimicrobials used annually in the United States with 87% of that being used in animals (280). Non-therapeutic was the primary reason for use (280). In 1999, the Federal Drug Administration (FDA) and Center for Disease Control and Prevention (CDC) began gathering AMU data from the Animal Health Institute. From these data, therapeutic and preventative use made up 83% of the total (281). The drug classes used included aminoglycosides, fluoroquinolones, ionophores, penicillins, sulfonamides, and tetracyclines. Current estimates of the volumes of antimicrobials used in the United States are unavailable because at this time the US does not routinely report AMU in livestock.
In the most recently available Canadian Integrated Program for Antimicrobial Resistance Surveillance (CIPARS) annual report, approximately 1.8 million kgs (~4 million lbs) of active antimicrobial ingredients were distributed for animal use in Canada (66). Tetracyclines were the most commonly used antimicrobial, followed by ionophores and macrolides (66). These estimates were provided by the Canadian Animal Health Institute (CAHI), which is the trade association representing the companies that manufacture and distribute drugs for administration to companion, sporting, and food animals in Canada (66). These data were aggregated to the class level and represent the distribution of these antimicrobials in 2006; they therefore do not represent what was actually used during this timeframe (66).
In order to gain a better understanding of AMU and to supplement this information, in 2006 CIPARS also began a pilot project to collect AMR and AMU information from Canadian swine herds. A preliminary report was released that includes the first animal agriculture AMU data collected by this national program (66). From these first few years of farm surveillance, AMU data collection challenges and limitations have been identified and data collection instruments have been refined to enable the team to gather the most useful AMU data available. While swine was the first commodity group that CIPARS began collecting data from, CIPARS has also been participating in large-scale research projects looking at AMR and AMU in feedlot cattle, dairy cattle, and sheep production, and is currently in discussions with the poultry industry with regard to developing a farm surveillance program in this industry. Canada is now facing challenges with how best to report their farm-based antimicrobial use data while acknowledging data limitations.
In order to substantially contribute to our understanding of AMU in animal production and its impact on AMR in people and/or animals, it is necessary to evaluate how antimicrobial use data are collected and reported so that it can be optimally utilized. While this may intuitively be easy to do, the reality is it is a challenge that still needs to be overcome. Even human antimicrobial use does not seem to be accurately estimated and reported (279,280), which leads one to surmise that there is additional work that needs to be done to not only understanding use in animals, but also in obtaining more consistent AMU estimates in human medicine.
OCT: available without a veterinary prescription; Pr: available with a veterinary prescription; -: no licensed products.

Antimicrobial Use: Routes, Antimicrobials and Reasons for Use
A background on the swine and poultry industries and common production practices is available in Chapter 4 and may assist those readers unfamiliar with these industries in considering the following information. As well, to further assist the reader, antimicrobials licensed for use in pigs and chickens in Canada are summarized by the labelled administration route(s) in Table 2.

Antimicrobial Use in Swine Production
Swine producers around the world use antimicrobials to treat and prevent disease, and depending on the country that they are being raised in swine, can also be exposed to antimicrobials to improve feed efficiency and daily gain. Antimicrobials can be administered to pigs through feed, water, and injection. Sows and suckling piglets are more often treated with injectable antimicrobials while nursery and grow-finish pigs more often are administered antimicrobials as groups through feed or water (93,277).
Overall, AMU is a common practice in Canadian swine production. Data from CIPARS indicated that 75% of the participating herds were exposed to at least one antimicrobial through feed, and 82% of the herds incorporated antimicrobials into either the feed or water (66). Assuming that reporting feed or water AMU means that all pigs in the batch were exposed, the majority of samples tested for antimicrobial susceptibility within the CIPARS Farm program were collected from pigs that had been exposed to an antimicrobial in the 20 weeks preceding sampling (66). These estimates relate only to grow-finish hogs but they do provide some estimates of conventional hog production antimicrobial use in this phase of production.
Injectable AMU is most important for treating clinically ill animals. The advantages of parenteral AMU are that the exposure to the appropriate dose can be ensured, treatment can be customized to each animal's condition, and only animals that require treatment need be exposed. The limitations of parenteral AMU are feasibility in treating large groups of animals and animal stress associated with handling and restraint. As pigs approach market weight, fewer herds use any injectable drugs and those continuing to use injectable antimicrobials report lower exposure rates (277). This is because of animal handling and disease factors. Market weight pigs are harder to handle and inject than small pigs. This increases the risk of injury to the staff or the pig and the risk of needle breakage. Some provinces allow pigs with a broken needle to be slaughtered if permanently marked while others require all pigs with a broken needle to be euthanized. The increased likelihood of a broken needle and greater economic loss from euthanizing older pigs both account for some of the decline in injectable AMU in near-to-market pigs. As well, there are decreasing infectious disease pressures as pigs age. However, when health problems occur in near-to-market pigs, producers may choose to inject individual animals to avoid having to hold groups of pigs until antimicrobial withdrawal times (period of time between the last administration of the drug and the collection of edible tissue or products from a treated animal that ensures the contents of the residues in food comply with the maximum residue limit for that drug) are observed. Additionally, producers may choose drugs with low or no withdrawal period; in particular, drugs like ceftiofur, a third-generation cephalosporin, are used close to market because of the zero day withdrawal.
In Chapter 1, the reasons for concern over resistance development to new-generation cephalosporins and fluoroquinolones were described.  (263) and extra-label use (see Chapter 3 for definition) in swine is legal (284-286).
In Canada, common injectable antimicrobials used include penicillin, trimethoprim-sulphadoxine, and oxytetracycline (92,93,93). Seventy-six percent of CIPARS sentinel swine herds reported injectable antimicrobial use in grow-finish pigs and the most commonly reported injectable antimicrobial used was penicillin (66).
Antimicrobials can also be administered through water. Water soluble antimicrobials can be administered to groups of animals without stress and are more practical for the treatment of infectious disease outbreaks than parenteral medications. Unlike injectable antimicrobials, exposure is not ensured because sick animals may stop drinking or may not drink sufficient water to achieve targeted blood antimicrobial levels. In this regard, water antimicrobials are often an intermediate between injectable and feed-grade antimicrobials because animals typically cease eating before they quit drinking. Water-soluble antimicrobials allow producers to rapidly respond to changing disease situations in groups of animals because treatment can be initiated and ceased virtually instantaneously with minimal labor. In contrast, medicated diets must be ordered or made in advance and then administered until the medicated feed is consumed. Despite these perceived advantages, fewer swine producers use water soluble antimicrobials than feed-grade drugs. This is likely because administration requires supervision in the barn and physical infrastructure to deliver the medication. Water AMU is most common in nursery pigs (92,93). When investigated, much shorter exposure times are reported for antimicrobials administered through water than feed (66,92,277,287).
In grow-finish pigs, 36% of Canadian producers reported use of water medication over a one year period (66). In general, water medication is used mostly for metaphylaxis and treatment (66,92). Penicillins, tetracyclines, and sulfonamides are the most common antimicrobial administered through water (66,92,93).
Feed-grade antimicrobials are most useful for managing static or predictable health situations. Medicated diets can be administered at high-risk times including weaning or transportation. The antimicrobials can be mixed in precise amounts and are stable, which makes it easier to achieve the desired concentration in feed than in water. Feed-grade medications are more cost effective than either watersoluble or injectable drugs. Feed-grade medications are the only medication route used to improve growth.
In Canada, most conventional swine producers use feed-grade antimicrobials and in many herds the exposure to antimicrobials is extensive. A survey of Alberta producers found 57% added feed medication to weaner, grower, and finisher rations more than 95% of the time (93). Of the remaining 43%, feed medication was more commonly administered to weaner and grower than finisher pigs (93). More extensive use of feed-grade antimicrobial use in younger pigs has been reported by many studies and is due to increased disease pressures and stressors in young pigs (92,93,282,283,287,288).
Tetracyclines, sulfonamides, and penicillin are commonly used in nursery pigs (93,287). Grow-finish pigs are most commonly exposed to macrolides for disease prevention or growth promotion followed by tetracyclines, which are predominantly used to prevent disease or treat respiratory disease (66). In-feed medication use is relatively static over time within herds. Dunlop et al. found few producers changed medication practices over eighteen months and Rajic et al. reported most producers used antimicrobials in diets more than 95% of the time in the previous twelve months (93,282). Un-medicated rations were more common during the latter part of the grower-finisher period (66). The use of un-medicated rations later in the feeding period most likely relates to decreased efficacy of growth promotion as animals age (289) and producers wanting to have flexibility in when they market these hogs and avoid any concerns over withdrawal times to prevent antimicrobial residues in the meat.

Antimicrobial Use in Broiler Chickens
Currently, no published data describe the types, amounts, or reasons for antimicrobial use in Canadian broilers and limited data are available from the United States. Injectable antimicrobials can be used in eggs or day-old chicks. Hatchery use of injectable antimicrobials is predominantly to control omphalitis caused by gram-negative bacteria (predominantly E. coli). Canadian hatchery drug use data are only available from Ontario; approximately 30% of the chicks hatched in Ontario are treated with an antimicrobial and the most common drug used was ceftiofur followed by gentamicin (290).
Beyond the hatchery, broiler chickens only receive antimicrobials through feed or water. As described in pigs, antimicrobials can be included in feed to improve growth and/or control disease (264). Industry representatives report extensive use of in-feed bacitracin and virginiamycin combined with ionophores (personal communication: various Canadian nutritionists). These drugs are primarily used for the control of coccidiosis and necrotic enteritis. Numerous water-soluble antimicrobials are available to producers without a veterinary prescription ( Table 2). The exception to this is amoxicillin (Paracillin® SP, Intervet) (263).
Agri-Stats is a private research company that collects data in the United States. Their analyses show a marked decrease in feed-grade antimicrobial use over the past decade (165). Medications in starter and grower diets (primarily bacitracin) fell from almost 100% of surveyed producers in 1995 to 65% in 2000. Antimicrobial use in withdrawal diets (primarily virginiamycin) fell from 75% to 48% over the same period. These declines may reflect tight profit margins. Producers will voluntarily remove inputs that do not increase production enough to offset their cost (165). Declining drug use may also reflect voluntarily restrictions on antimicrobial use to satisfy consumer concerns and target premium markets or satisfy production contracts (291,292). No data were found describing feed-grade antimicrobial use since 2001, nor were any publications identified on therapeutic antimicrobial exposures in American broilers.
European AMU data are also limited as few countries stratify the volume of antimicrobials sold by animal species. A good example of reporting by animal species comes from Denmark. Broiler chickens consumed 56 kg of the 121 tonnes (<0.001%) of the antimicrobials used in food-producing animals (89). The small proportion is because Denmark has a large export-based pig industry producing 1,957 million kg of pork and a domestic broiler industry producing 163 million kg of chicken (89)

Relationship between Antimicrobial Use in swine or Poultry and Antimicrobial Resistance
A challenge for any study is demonstrating causality. Cause is defined as an event, condition, or characteristic that preceded the disease or the disease event and without which the disease event either would not have occurred at all or would not have occurred until some later time (293). While the context of a single cause is a useful concept, the reality is there are often multiple causes working together to lead to an effect. In AMR, a more appropriate concept is sufficient cause. Sufficient cause is a set of minimal conditions and events that inevitably produce disease or in this case resistance. Minimal condition in this sense means that all conditions or events are necessary (293). It is in situations where a cause is multifaceted or has multiple contributing sources that proving causality becomes increasingly difficult.

Links between AMU in Swine and AMR in Bacteria Carried by Swine
There have been several articles describing associations between AMU and AMR in swine (84,92,125,283,(294)(295)(296)(297)(298)(299)(300). Depending on the organism and the resistance of interest, various risk factors for the development of AMR have been identified on swine farms. These include primarily AMU exposures, but they also reflect that stage of production may influence AMR development.
The expected relationship of exposure to an antimicrobial correlating to resistance to that antimicrobial and other antimicrobials within that class has been reported (125,295,298,299). But the use of an unrelated drug class being associated with increasing resistance to a particular antimicrobial has also been described. For example, tetracycline resistance is related to the use of sulfonamides or aminoglycosides (299) or the use of ceftiofur can select for tetracycline resistance (283).These associations demonstrate the likelihood of the presence of linked genetic elements and the coselection of AMR (295,298,299). Rosengren et al. (299) demonstrated possible coselection when they reported that macrolide use was associated with resistance to chloramphenicol, streptomycin, sulfamethoxazole, and trimethoprim sulfamethoxazole in E. coli isolates from finishing hogs. This was surprising because E. coli is intrinsically resistant to macrolides and yet macrolide use appeared to select for resistance to at least four other antimicrobials. Akwar et al. (295) found similar results. Further investigation to describe any genetic linkages between macrolide and other resistance genes is warranted to more fully understand potential repercussions of macrolide use on diverse bacterial pathogens (299).
In addition to AMU selecting for AMR, some negative associations, where AMU lead to decreased resistance, have also been reported (283,299). A possible explanation for these negative associations could include gene incompatibility between resistances to certain antimicrobials. For example, resistance genes catA1 and tetB are negatively associated with each other (136). Another explanation could be that AMU in this instance is a surrogate for some other risk factor that was not identified (283). A better understanding of why the use of a particular antimicrobial seems to lead to the reduction of AMR is needed and warrants additional research to attempt to identify if these negative associations are truly due to gene incompatibility or to try and identify 'unknown' factors that could be contributing to this phenomenon.
The impacts of exposure rates on AMR development have been investigated. Rosengren et al. (299) described the exposure rate of AMU and considered the relationship between use and resistance. Assuming a causal relationship, this study demonstrated that long-term use in pigs affects AMR more than targeted or short-term use. Similar results were also reported by Dunlop et al (277). Supporting this concept of increased use leading to increased resistance, the alternative has also been demonstrated where the removal of use has decreased resistance (296,301). Langlois et al. (255) demonstrated a reduction in AMR with the removal of AMU, but indicated that the reduction was less than what would have been anticipated over the study period, demonstrating that AMR can continue to persist in these herds without the selective pressure of AMU.
Antibiotic free (ABF) herds have been investigated and compared to conventional herds to elucidate the contribution of AMU to resistance over baseline. From these comparisons, varying results have been reported. Some indicated that resistance was more frequent on conventional farms that use antimicrobials than on ABF farms (302,303). These findings support AMU driving resistance. But, in a different study comparing ABF herds to conventional herds, not only did the authors report that at slaughter significantly more Salmonella were isolated from the carcasses of swine from ABF farms but that more multidrug-resistant Salmonella were isolated from ABF farms (300). Since the move of some farms to ABF production is relatively recent, there is paucity of data in this area. More research is needed to improve our understanding of the ecology of zoonotic bacteria and AMR on ABF farms and how these compare to conventional farms.

Links between AMU in Poultry and AMR in Bacteria Carried by Broilers
When looking at the impact of AMU on AMR in poultry production, there have been several studies supporting that AMU leads to AMR (302,304,305). Experimental trials have demonstrated that after therapy with a variety of antimicrobials there was a significant increase in resistance levels (302,306). While these experimental trials are useful, extrapolating these trial results to estimate in-barn selection pressure is challenging because the dynamics of the resident flora and historical AMU on the farm may impact the study results. Ultimately being able to use a wide variety of commercial farms for research purposes would be ideal since that would provide a range in management practices and exposures, but this is not always feasible from a management, economic or practical standpoint.
Due to the limitations of experimental research, an alternative approach is to evaluate the impact of widespread AMU changes on an industry. The most notable examples are the 1997 avoparcin and 1999 broad-AGP bans in Europe. The withdrawal of growth promoters resulted in a significant decline in resistance to avilamycin, erythromycin, vancomycin and virginiamycin in E. faecium isolated from broilers and broiler meat (304). But most recent reports acknowledge that these initial declines have been followed by stabilized persistence of resistance. A study from the United Kingdom indicated that vancomycinresistant enterococci have persisted on intensive broiler farms despite no avoparcin use for seven years (307). Similar results have also been shown in Norway, where three years after the ban of avoparcin, vancomycin-resistant enterococci were still being detected in poultry environments (308). These data demonstrate that in commercial barn settings resistant strains do persist on some farms for extended periods of time and support the arguments that husbandry practices and/or therapeutic AMU probably contribute to the persistence of AMR in these situations.
Studies looking at ABF poultry production and AMR are also available. One study looking at ampicillin-resistant E. coli in organically raised broilers found that older heavier birds were significantly more likely to carry ampicillin-resistant E. coli than younger birds, despite the fact that the total E. coli shedding was lower in older birds and that they had not been exposed to antimicrobials (309). This indicated that as birds age the proportion of ampicillin-resistant E. coli increased (309) even in the absence of any antimicrobial use. In another study, that examined barns that moved to ABF production that had previously used fluoroquinolones, it was determined that previous use increased the percentage of fluoroquinolone-resistant Campylobacter and that this resistance continued to persist in the poultry environment despite the ongoing absence of AMU selective pressure (310).
In a study comparing retail chicken from organicversus intensively reared poultry, all meat had Campylobacter isolates resistant to nalidixic acid and erythromycin, but the resistance levels for nalidixic acid were significantly higher in the conventional birds versus the organic (311). Also, all Campylobacter isolates from organically raised birds were susceptible to ciprofloxacin (311). The results from the above studies indicate that organic or ABF production does not guarantee that birds or retail meat will be free from AMR bacteria and that there are factors other than AMU that are impacting the presence of AMR.
Another example that demonstrates our limited understanding of the ecology of AMR comes from work examining the impact of enrofloxacin and tylosin treatment of broilers. Both enrofloxacin and tylosin are potential therapeutic agents for E. coli air sacculitis (8,263). A controlled experiment found erythromycin resistant Campylobacter did not emerge in broilers treated with tylosin (0.53g/ liter for three days, once or multiple times) (312).
In contrast, separate studies found fluoroquinolone treatment significantly increased the frequency of resistant Campylobacter; 10 mg of enrofloxacin or difloxacin per bird for five days in drinking water (313), 40 ppm of sarafloxacin or enrofloxacin for five days (314). In both fluoroquinolone studies, resistant Campylobacter dominated at slaughter. Therefore, although macrolides and fluoroquinoloneresistant Campylobacter each pose a public health hazard, the consequences to humans of therapeutic fluoroquinolone use in broilers appears substantially greater than macrolide use. Greater consideration for the effects of AMU, not only on AMR in the treated birds, but on meat contamination with resistant bacteria, could strengthen therapeutic guidelines developed to mitigate AMR.
The threat to public health is demonstrated by the development of resistance to third-generation cephalosporins by Salmonella. As mentioned in Chapter 1, Canada, through its national surveillance efforts, has identified a link between AMU and AMR in chickens (66). Salmonella Heidelberg is among the top five serovars isolated from human salmonellosis cases in Canada and the United States. Resistance to ceftiofur (a third-generation cephalosporin used only in animals) among S. Heidelberg isolates is highly correlated with reduced susceptibility to ceftriaxone, a third-generation cephalosporin used exclusively in humans. Ceftiofur is used as a prophylactic agent in broiler chicken hatcheries to reduce losses associated with omphalitis. Shortly after the inception of the CIPARS retail component, a sharp rise in resistance to ceftiofur was detected in both retail meat and human S. Heidelberg isolates (Figure 2)(66). When first detected, CIPARS presented these data to its stakeholders. To address this public health concern, hatcheries in Québec voluntarily stopped the use of ceftiofur in hatching eggs and day-old chicks. This voluntary withdrawal was followed by an observed decrease in ceftiofur resistance in retail chicken and human S. Heidelberg isolates ( Figure 2) (66). Subsequently, there has been anecdotal information that the use of ceftiofur in the hatcheries has been reinstituted at some level. According to CIPARS data, an increasing trend of ceftiofur resistance supports this return to ceftiofur use ( Figure 2). These data do show support for the correlation in trends over time between human S. Heidelberg resistance to ceftiofur and retail chicken S. Heidelberg resistance to ceftiofur. While this cannot show definitive causality, it does warrant further investigation to gain a better understanding how ceftiofur use in poultry is impacting human salmonellosis cases.
The above studies all support the fact that the ecology of AMR is not simplistic. There are biological factors, as well as exposure factors, that impact the prevalence and persistence of AMR. In some cases simply removing a drug exposure does not always lead to a reduction in AMR, and the lack of an exposure does not mean that AMR will not exist. There is still more research needed to try and identify what other factors may be impacting AMR development and/or persistence especially in the absence of AMU selective pressure.
information gaps and Challenges in Understanding the effects of Antimicrobial Use in livestock on Public Health Most classes of antimicrobials used in livestock are also used in human medicine. One of the problems with using the same class of antimicrobials in both humans and animals is that as a result of cross-resistance the development of resistance to a particular drug within a class of antimicrobials can result in resistance to the entire class. One class of antimicrobials used in animals and not in humans is the ionophores such as monensin, narasin, salinomycin and lasalocid, the quinoxalines, and avilamycin (316). While these drugs are not used in people, their use could still potentially impact AMR development through the process of co-selection. Although, a particular drug may or may not be used in humans, the reality is that through gene linkage and transfer any antimicrobial use could ultimately have a negative impact on human health.
One way to try and understand the role of AMU on AMR is through risk assessments. Risk assessments (RA) can be used to try to evaluate the consequences to human health from the use of antimicrobials in animals. Risk assessments involve the estimation of risk for situations that cannot be measured or observed and provide an interface between science and policy (293). The goal of a RA is to determine a quantitative or qualitative value of risk related to a concrete situation and a recognized hazard. These assessments involve the calculation of the magnitude of loss and the probability it will occur based on assumptions and data estimates that are not directly testable (293). As result, RA are often inconclusive.
Several RA have been performed to assess the impact of AMU in animals on the development of AMR in humans (317-321). One of the identified limiting factors of these assessments in providing a precise estimate is the need for a better quantitative data on the rate of transmission of AMR bacteria and resistance genes between animals and humans (317,319). Other limiting factors of these assessments include limitation to one bacteria, antimicrobial, outcome, animal species, and antimicrobial use (246). These factors, along with limiting the scope of the analysis to what has already happened and to the bacteria the resistance was identified in, ignores the effects of continuing the practice and the ability for AMR to spread to other organisms. The other potential biasing effect of RA is that they are generally commissioned for a reason and may be either subject to bias in the interpretation and reporting of results, or prone to criticisms of bias. For these reasons RA may be limited in what information they can provide. However, in the absence of other methodologies that can provide insight on the potential impact of a particular hazard, risk assessments can be an important tool especially if the results are considered in light of the limitations of RA.
The alternatives to risk assessments are; reviews of the literature, primary research, observational studies, and interpretation of longitudinal data. Again, these alternatives have potential limitations such as bias, misinterpretation, lack of power or study size, lack of ability to quantify certain critical data, and the inability to control for all potential confounders and risk factors. These methods are rarely able to put together the 'whole picture' as risk assessments attempt to do.
Regardless of the approach to summarizing data or to assessing the risk of AMU in animals on human health, there is still a great deal of controversy and conflicting evidence as to the actual risk. For example, in the case of invoking the precautionary principle there is a wide range of opinions as to the impact, not only on human health but to animal health and food safety. In the following paragraphs several examples have been provided in an attempt to demonstrate the divergence of opinion, with the ultimate goal of illustrating that this debate is far from being resolved.
A common position is that the bans were effective in achieving their goal and these proponents supported their position though evidence of decreased rates of AMR in bacteria from animals and food (13,304,322). But an opposing position is that policy that invoked the 'precautionary principle' to ban AGP lacked sufficient and conclusive scientific information.
Concerns exist that these bans may ultimately have potentially dangerous and inconsistent effects on animal and human health (323). There is concern that the AGP bans have increased rates of animal illness which may impact human health by increasing the microbial load of zoonotic pathogens as well as increasing the need for therapeutic antimicrobial use (266,324,325)(325). In addition to increasing the shedding of zoonotic pathogens, concerns exist that diseased flocks can have a higher rate of processing errors, cuts, fecal contamination, and microbial loads (325,326).
There may be some merit in speculating that the removal of in-feed antimicrobials may have adversely affected public health. In the years after the European AGP ban, there was a rise in antimicrobial resistance of hospitalized patients ( and seeing a rise in human health issues, the US was implementing hazard analysis and critical control points (HACCP), increasing public awareness through education and implementing prudent use guidelines (323), and as a result, the US subsequently saw a decrease in human illness. Whether these differences are like comparing apples and oranges or relate directly to the use of antimicrobials in livestock and poultry is impossible to determine, but they contribute to the scientific debate and rhetoric around AMU policy. Ideally if a similar phenomenon occurs in the future, national surveillance programs will be able to elucidate valid causes.
In addition to the impact on human health as a result of banning AGPs, there was also an impact on animal health. Some work has indicated that the ban of certain antimicrobials for animal use had a significant impact on animal health with an increase in necrotic enteritis in chickens and Lawsonia intracellularis in pigs (324,334). In the case of necrotic enteritis rates in broilers, it has been reported that morbidity rates went from approximately 0% to a transient high of 15% (323,335). In swine, Cox and Ricci indicated that it was difficult to find estimates for the impact of the ban on pig morbidity and mortality, but said that trade journals reported that from the time of the ban in Denmark until 2005 there was increase in mortality rates from 17% prior to the ban to 21% after the ban (323,336). Probably, much of this mortality in swine was related to increased diarrhea and weight loss due to E. coli or L. intracellularis early in the post weaning phase of production (324). Although producers experienced increased losses as a result of disease and death immediately subsequent to the bans, after the initial surge, as a result of increasing the use of therapeutic antimicrobials, animal illness rates were controlled within two to five years depending on the animal species (334,337,338).
In contrast to these findings, Wegner states, "in broilers in Denmark, necrotic enteritis was at most a minor broiler health problem following the termination of antibiotic growth promotants" (339). Similarly, Emborg et al. indicated that there were no changes in weight gains or mortality in broilers and that the effects of antimicrobial growth promotant termination in poultry production were small and limited to decreased feed efficiency (340). The cost from lost feed efficiency was estimated to be offset by producers' savings on not needing to purchase antimicrobials for growth promotion (341). Callesen indicated that in swine there was a significant increase in treatments for diarrhea in nursery piglets and that there was some loss of productivity in weaners but no real effect in finishers (342). These conclusions are in sharp contrast to the interpretations of the effect of the ban on animal health and welfare described earlier, and they demonstrate the inconsistencies of this debate and the varied interpretation of the impact of the ban. Some of these inconsistencies arise from a lack of quality animal health monitoring prior to the ban, while others may reflect different health experiences in different regions (342).
Conflicting conclusions about the impact of AGP bans continue to be at the forefront of debate. In some instances, there has been a paradoxical increase in AMR after the ban of antimicrobial use. For example, ciprofloxacin/naladixic acid-resistant Campylobacter coli isolated from swine increased significantly from 3% in to 16% in 2004 Final Thoughts on the Link Between AMU in Animals and AMR in Humans Cox and Ricci (323) suggest that the link between increased resistance in animals leading to increased resistance in humans is relatively weak, based on their work and the work of others (159,318,345,346). Others report that a link between AMU in livestock production and increasing AMR in people is virtually certain (275,339,(347)(348)(349). Turnidge (316) has a good summary of these opposing positions. He contends that the main reason that the debate continues is that there are virtually no studies that accurately quantify transmission rates of antimicrobial resistant organisms between humans and animals.
Although the degree of transfer is unclear, both sides agree that animals can be a reservoir for AMR and that they can act as a source of amplification for AMR as a result of AMU. At this point, we need to decide if we move forward from the debate and start controlling antimicrobial use without indisputable evidence of the effects on AMR of such use (316), or if we continue to study the complex ecology and epidemiology of AMR before intervening on AMU.

Conclusion
AMR is a complex subject. The issue of AMU in livestock and its subsequent impact on human health will probably continue to be debated because there is conflicting evidence, personal bias, and potentially significant political impacts on all sides of the question. Conflicting conclusions between researchers about the impact of AMU in livestock on AMR development in humans may be the result of strong regional biases in the data since large amounts of data can come from specific regions. Differences in methodology and agriculture production must also be considered when these studies are evaluated to ensure that results are taken into context. Effects of interventions may not be identical in regions with different management systems, climate, and pathogen prevalence. Therefore, results of studies in specific regions should certainly be considered, but development of policy or legislation based on data from potentially incomparable systems should be approached with caution.
In light of data indicating that the elimination of AMU in feed can lead to increased bacterial contamination of carcasses, a variety of effects of banning antimicrobials on public health must be considered.
Having food animal products that have minimal bacterial loads is critical to preventing foodborne illness. A part of reducing bacterial contamination on retail meat and poultry products may include ensuring good gut integrity and optimal animal health.
Since there is still a great deal to learn about the role of antimicrobials in potentially reducing carcass contamination with bacteria, it is essential to evaluate what impact, if any, the removal of antimicrobial use would have.
Due to the wide distribution and multi-faceted nature of AMR, it is going to be very difficult to quantify of the amount of AMR in humans because of AMU in livestock production. Additional research examining the causal effect of AMU in livestock on AMR in people, along with improved quantification of the transmission dynamics of AMR between people and animals, are needed but these will not be simple tasks. When it comes to understanding and quantifying AMU in livestock, the knowledge gap grows exponentially. In order to be able to appreciate the true impact of livestock AMU on AMR, a much better system of collecting and reporting AMU needs to be developed. For most countries, and in the majority of species, the unfortunate reality is that good AMU data are truly limited at this time. With ever changing and evolving understanding of both the molecular and global aspects of AMR ecology, there will need to be continually committed to modifying previous interventions and strategies. It is also very likely that because of the complexity of AMR development and dissemination, there will never be an easy resolution to contain the future development of AMR.
Chapter 3: overview of Antimicrobial Resistance surveillance or Monitoring Programs and subsequent Policy outcomes

Introduction
Links through travel and trade make antimicrobial resistance an international problem requiring a mix of national and international solutions. Canada cannot address agri-food antimicrobial resistance (AMR) without first considering the history of other nations and their responses to AMR. An evaluation of how international activities relate to Canada is also required. For those readers new to the story of AMR control in food and food animals, sorting out the history of the policies and regulations can feel like a convoluted journey. Many actions occurred close in time, but only some were a related sequence. Overall, the strategies, regulations, and policies to date have predominantly focused on two particular aspects of antimicrobial use (AMU) and on surveillance of AMR. Regulations have aimed to improve the appropriateness of AMU. In Europe, this largely meant banning antimicrobial growth promoters. In contrast, North America has focused on policy and regulations to improve prudent AMU. Surveillance of AMR, and to a lesser degree AMU, is occurring in many regions of the world. Most countries appear to recognize that they lack sufficient data to make optimal decisions and are investing in long-term programs that provide consistent and comparable data over time and space.

Scope and Objectives
The objectives of this chapter are to familiarize readers with the history of policies and strategies to control AMR in other parts of the world to allow a sufficient evaluation of Canada's current approach and future options. Once again, the scope is limited to AMR as it pertains to bacteria in pigs and chickens from farm to abattoir and processing. This chapter does not investigate the effects of these policies because examples of these have been discussed in Chapter 2. This chapter begins by describing the actions and policies of four international organizations with mandates in human health, food safety, and animal health. Although these groups largely acted in the 21st century, which was after the major regulatory changes in Europe in the 1990s, they are presented first because they provide insight into the prevailing global attitudes about AMR and the recommendations for a cohesive policy approach. They are presented separately from the national regulations and legislations as these organizations can only issue recommendations which member countries may choose to implement.
The second segment describes regulations and policies pertaining to veterinary drug use in Europe, the United States, and finally Canada. Europe is presented first as it has the longest history with this issue, and many decisions there have shaped the international and North American responses. Canada is presented last to allow a comparison of our veterinary drug regulations with others. The final section describes international surveillance activities in AMR and AMU. Canada is presented first with a detailed description of our program followed by other noteworthy programs, which are compared to Canada.

international organizations' Response to AMR in Agriculture
International organizations recognize that AMR transcends national boundaries. Several international organizations have, in accordance with their mandates, provided recommendations to foster a globally-cohesive response to this threat. Most of the international activity addressing AMR has occurred in the 21st century. The following discusses the mandates, positions, and recommendations of the United Nations' World Health Organization (WHO) and Food and Agriculture Organization (FAO), the World Organization for Animal Health (OIE), and Codex Alimentarius.
Under the United Nations' system, the WHO is responsible for providing leadership on global health matters, shaping the health research agenda, setting standards, articulating evidence-based policy options, providing technical support to countries, and monitoring and assessing health trends (350). The WHO has acknowledged that the inappropriate use of antimicrobials has exacerbated the selection for antimicrobial resistant organisms and in response developed the first global strategy for the containment of antimicrobial resistance published in 2001 (270). This summarized six key recommendations regarding the use of antimicrobials in food-producing animals. These are as follows: i) obligatory prescriptions for all antimicrobials used for disease control in animals; ii) in the absence of a public health safety evaluation, terminate or rapidly phase out the use of antimicrobials for growth promotion if they are also used for treatment of humans; iii) create national systems to monitor antimicrobial usage in food animals; iv) introduce pre-licensing safety evaluation of antimicrobials with consideration of potential resistance to human drugs; v) monitor resistance to identify emerging health problems and take timely corrective actions to protect human health; and vi) develop guidelines for veterinarians to reduce overuse and misuse of antimicrobials in food animals. The WHO regularly works in conjunction with FAO and the OIE, which have food and animal oriented mandates, to address this issue.
The agricultural arm of the United Nations, the FAO, leads international efforts to achieve food security and aims to raise nutrition, improve agricultural productivity, improve the lives of rural populations, and contribute to the growth of the world economy.
The FAO provides a neutral forum for developing and developed countries to negotiate agreements and debate policy (351). Jointly established by the WHO and OIE, the Codex Alimentarius is an international organization responsible for generating standards to protect the health of consumers while ensuring fair trade practices of food (352). In the case of antimicrobial use and resistance, Codex largely bases its recommendations on those issued by the WHO (353). The Codex Alimentarius Commission has released a code of practice to minimize and contain antimicrobial resistance in food. The aim of the code is to provide recommendations to prevent or reduce selections of antimicrobial resistant microorganisms in animals and humans through the responsible use of antimicrobial drugs in food-producing animals.
The code provides guidance for the prudent use of antimicrobials in food-animals, addressing guidelines on the prescription, application, distribution and control of drugs used for treating animals, preserving animal health, and improving animal production (353).
While the mandate of Codex is limited to food, the OIE is an international organization that is responsible for improving animal health worldwide (354). The OIE has outlined in its terrestrial animal health code: i) criteria for development and harmonization of national AMR surveillance and monitoring programs; ii) recommendations for AMU monitoring in animals; iii) guidelines for responsible and prudent use of antimicrobials in veterinary medicine; and iv) risk assessment methodologies for assessing AMR arising from AMU in animals (355). The terrestrial code is published annually and is aimed to assure the sanitary safety of international trade in terrestrial animals and their products. The guidelines that are currently being developed for risk assessment aim to provide a transparent, objective, and scientifically defensible way for member countries to assess and manage the human and animal health risks of antimicrobial resistance from the use of antimicrobials in animals. These international organizations collectively hold a global mandate to address AMR from farm to fork. Each organization is responsible for a distinct segment of the supply chain. Thus inter-organizational co-operation has become the mainstay for most of the policies and recommendations released. All of these organizations have identified the need for national surveillance or monitoring systems for antimicrobial resistance (270,358-361). The global diversity of animal production systems, and veterinary and public health infrastructures, necessitates that these guidelines be general and overarching.

Ranking of Critically Important Antimicrobials
In 2003, the FAO, OIE, and WHO initiated a joint meeting to discuss issues related to antimicrobial use in agriculture and veterinary medicine, while understanding the essential role antimicrobials play in human and animal medicine (359). The outcome was a ranking system for critically important antimicrobials (CIA) (362). These were concluded as, "antimicrobial classes that provide specific treatment or one of a limited number of treatments for serious human diseases or pathogens that cause foodborne diseases. " WHO developed a list of critically important antimicrobials based on the following two criteria: i) sole therapy or one of few alternatives to treat serious human disease and; ii) antibacterial used to treat diseases caused by organisms that may be transmitted via non-human sources or diseases caused by organisms that may acquire resistance genes from non-human sources. Antimicrobial classes meeting criteria 1 and 2 were ranked as 'critically important;' antimicrobials meeting 1 or 2 were 'highly important;' and if neither criteria were met, then the antimicrobial class was deemed 'important' (86,362).
Following recommendations from this meeting, the OIE created a similar ranking of antimicrobials for their importance in treating animal diseases (362). The overlaps of both critical lists for human and veterinary medicine collectively provide insight to allow for a balance of what is needed in veterinary medicine while taking into account public health considerations. Australia, Canada, Japan, the United States, and possibly others have developed their own national ranking of antimicrobials for their importance in human medicine. Canada has outlined specific criteria for how it determines critically important antimicrobials (Table 3) (363). While the international list provides a global perspective and facilitates a consistent approach between countries, the national lists more precisely reflect the antimicrobial needs in each country. This makes them more applicable to regulatory decisions within those countries. For example, Canada and the United States use their respective lists for risk prioritization and assessment of which antimicrobials used in agriculture require risk management to avoid undue risk to human health, and then apply this information to reviews of new veterinary drug applications (272,364).

Policy, strategy, Regulation, and legislation to Address AMR in Agri-food europe
In 1963, the United Kingdom experienced an epidemic of multidrug-resistant Salmonella Typhimurium that was linked to therapeutic antimicrobial use in veal calves. In response to the possibility that antimicrobial use in food-producing animals could be causing disease in people, the Joint Committee on the Use of Antimicrobials in Animal Husbandry and Veterinary Medicine was established and published a seminal report in 1969 (365). This report is widely referred to as the 'Swann report' after the lead author. In it, the following three recommendations stand out: i) antimicrobial growth promoters (AGPs) should be restricted to antimicrobials that are not used as therapeutics in animals or people and restricted to drugs that will not impair the efficacy of therapeutic drugs through the development of resistant bacteria; ii) antimicrobials used in animals should be divided into feed (i.e. growth promoting) and therapeutic classes; and iii) therapeutic antimicrobials should be available only by veterinary prescription (365).
Although the Swann report acknowledged gaps and uncertainties in the scientific evidence, European countries responded to its recommendations and separated the legislation for in-feed antimicrobial growth promoters from all other veterinary drugs. This resulted in prohibition of using tetracyclines and penicillins as AGP (259,260). Concurrently, European countries made all veterinary drugs available by prescription only. In 1986, Swedish producers opted to ban all AGP to assuage consumer concerns over AMR, while acknowledging that the scientific evidence was incomplete (337,366). Thus, it has been over forty years since Europe and North America initially diverged in their regulation of veterinary drugs, and over this time these differences have continued to grow.
The catalyst for the current global concern over AMR in livestock was the emergence of communityacquired vancomycin-resistant enterococci (VRE) in Europe. This raised concerns over the use of a related glycopeptide, avoparcin, as a growth promoter in animals (367-369). In 1995, countries began to independently ban avoparcin, and in 1997 an EU-wide ban was enacted (259). As described more fully in Chapter 1, the pharmaceutical industry responded to the need to treat VRE by releasing a new antimicrobial under the trade name Synercid® (162). The active pharmaceutical ingredient in Synercid® is quinupristin/dalfopristin which is related to a feed-grade antimicrobial, virginiamycin. Prior to the release of Synercid, the use of virginiamycin in animals was of minimal concern because there was no product from this class of antimicrobials that was being used in human medicine (162,164).
The premise of avoiding AGPs that are related to antimicrobials used in people is almost impossible to comply with in any system that permits this type of AMU. The lack of new antimicrobial discoveries has meant that the human pharmaceutical industry has remodelled old classes of antimicrobials to address emerging problems. This situation is not limited to virginiamycin. Bambermycins were historically limited to use in animals, but have recently been touted as the "holy grail for antibiotic drug discoveries" under the name moenomycin, and ionophores are being developed for topical therapy of bacterial and viral diseases under the name gramicidin (370-372). The limited discovery of truly novel antimicrobials means that drugs previously not used in human medicine are now being used to help combat resistance issues that have continued to emerge.
Following the ban on avoparcin in 1997, the EU banned four additional AGPs in 1999. These included spiramycin, virginiamycin, bacitracin, and tylosin (259). This ban did not target a specific bacterial pathogen or resistance outcome but was designed to decrease the ecological selective pressure from long-term, low-dose AMU. The EU's policy on AMU in agriculture was formally completed through the 2006 legislation that revoked the remaining four AGP feed additives: monensin sodium, salinomycin sodium, avilamycin, and flavophospholipol. The poultry industry was granted continued access to ionophores without prescription until 2013 (259,260,324

The United States
The United Sates and Canada face many similar issues regarding veterinary drug regulations. Specifically relating to antimicrobials in food-producing animals, American concerns exist around the following: i) extra-label drug use (ELDU); ii) evidencebased risk assessment of human health risks from AMU in livestock; iii) non-prescription access to antimicrobials (commonly known as over-the-counter or OTC); and iv) use of in-feed antimicrobials for growth promotion or improved productivity. The first two points are being addressed through FDA regulations, no evidence of action was found on the third point, and the final point has fuelled numerous bills supporting more restrictive antimicrobial drug legislation.

Extra-label Drug Use
Extra-label drug use is the use of any drug in a way not indicated on the label. It is described in more detail under Canada, with only American specific legislation presented here. In 1994, the Animal Medicinal Drug Use Clarification Act (AMDUCA) gave American veterinarians the flexibility to use animal pharmaceuticals in situations not specifically listed on the label providing the use occurred in the context of a valid veterinary-client-patient relationship (VCPR). The AMDUCA specifically excluded certain ELDU practices. Feed-grade antimicrobials could not be used extra-label under any circumstance. As well, the FDA can impose a restriction of ELDU on a certain product or group of antimicrobials in food animals. This has been applied to chloramphenicol, fluoroquinolones, and glycopeptides and, as will be discussed, was recently considered for cephalosporins in food-producing animals (281).

Risk Assessment of Human Health from AMU in Food Animals
The FDA is addressing concerns of AMR in bacteria from food animals through a multifaceted regulatory framework that includes pre-approval safety evaluations, post-marketing surveillance, and risk management strategies. Evaluating new antimicrobial drug submissions for both safety and microbial effects on bacteria of human health concern was embarked on in 1998 and rolled out to industry in 2002 (272,373). These regulations focus on enteric bacteria and require a risk assessment approach, which can be based on the OIE Ad Hoc Group on Antimicrobial Resistance recommendations (272,374,375).
The post-marketing requirements depend on the perceived risk to human and animal health and may include surveillance through the National Antimicrobial Resistance Monitoring System (NARMS). Risk management occurs via controlling drug access. Antimicrobials that require veterinary supervision to ensure appropriate use are licensed as prescription drugs. Other antimicrobials may be licensed as nonprescription drugs. The continued option to register new drugs as non-prescription is interesting, given the new regulations provided an opportunity to decrease the number of antimicrobials available without a veterinary prescription. The FDA may also prohibit the extra-label use of a drug or limit the extent of use. For example, drugs may be limited in exposure duration or to use in individual animals (272).
Through these regulations, the FDA has revoked, evaluated, and considered relabelling drugs to address risks to public health from AMU in food animals. The only antimicrobial that has been removed from the market is water-soluble enrofloxacin licensed for use in poultry. Enrofloxacin and another fluoroquinolone, sarafloxacin, were licensed in 1995 and 1996 respectively (376). As described in Chapter 2, rising rates of ciprofloxacin resistant Campylobacter were identified in poultry and human clinical isolates (68,262,376). Together, surveillance and quantitative risk assessment results were used to scientifically justify the removal of water soluble enrofloxacin from the American market in September 2005 (8,343,376). This legislation did not apply to sarafloxacin because its manufacturer Abbott Laboratories voluntarily removed it from the American market prior to 2000 (377).
Virginiamycin was the second antimicrobial evaluated by the FDA for AMR risks in humans. The human health risk from virginiamycin use in animals and the FDA-CVM risk assessment has been described in Chapter 1 (164). Following release of this risk assessment, no regulatory action has been taken (164); possibly because other antimicrobials have since replaced QDA as the most important treatment for vancomycin-resistant enterococci.
Most recently, the FDA proposed prohibiting extralabel cephalosporin use in food-producing animals (378). Ceftiofur is a third-generation cephalosporin available as an injectable product for use in chickens, pigs, and other food-animals (379). NARMS surveillance identified rising resistance to thirdgeneration cephalosporins in Salmonella from cattle, pigs, chickens, and turkeys and these findings were corroborated by the Canadian Integrated Program for Antimicrobial Resistance Surveillance (CIPARS) surveillance data from retail chicken (66,115). The proposed prohibition cited concerns that hatcheries were using ceftiofur extra-label, a practice also noted by CIPARS (378,380). Following an extended public comment period, with response from many agricultural sectors that restricting ELDU of cephalosporins would cause an animal welfare and disease crisis, the FDA revoked the extra-label cephalosporin prohibition in November 2008. Evaluation of this concern continues, but if the FDA wishes to prohibit extra-label cephalosporin use, it must re-initiate the entire regulatory processes (381).

Antimicrobial Use in Food Animals to Improve Growth and Productivity
In These bills propose that new antimicrobial drug applications be denied unless there is a reasonable certainty of no harm to humans due to AMR from non-therapeutic use of the drug. They also propose that approvals for non-therapeutic drug use in foodproducing animals be withdrawn two years after enactment of the Act unless safety requirements are met. To date, such bills have not been passed, but the persistent submissions demonstrate a strong support base and the potential for AMU to become managed by legislation rather than regulations. If passed, these changes would move the United States closer to the precautionary principle approach of the EU.

Canada
In Canada, the role of food safety policy and regulation is mandated to Health Canada and enforcement to the Canadian Food Inspection Agency (CFIA). Within Health Canada, the Veterinary Drugs Directorate (VDD) has the directive to protect human and animal health and the safety of Canada's food supply with a focus on veterinary drugs administered to food-producing animals (383). Hence, licensing of veterinary pharmaceuticals is a federal responsibility that relates to the provincial responsibility for regulating the sale and distribution of veterinary drugs (248). Although there are is substantial differences between the Canadian provinces in their regulation of veterinary drugs, this document is limited to discussing federal regulations.
In 1997, Health Canada hosted a national consensus conference entitled 'Agriculture's Role in Managing Antimicrobial Resistance. ' Delegates attending this conference called for Health Canada to address AMR and AMU in the agri-food and aquaculture sectors (384). This conference supported initiating AMR surveillance and influenced changes to Canada's veterinary drug use regulations.
Three years after this conference, the European Commission (EC) audited Canada's control of chemical residues in animals and meat. This audit was conducted to ensure Canada continued to meet the consumer needs of the EU and was sweepingly critical of Canada's veterinary drug regulations (385).
Two concerns that were raised in this report, and Health Canada's subsequent response, are pertinent to the current discussion. The first concern was that Canada address ELDU. The second concern was over Canada's lack of action to ban the sale of drugs that, based on scientific evidence, pose a risk to human health. This concern pertained specifically to a non-antimicrobial drug, diethylstilbesteral, but was also applicable to in-feed non-therapeutic AMU which had been banned in Europe the previous year (385). The changes in Canada's veterinary drug regulations subsequent to the 1997 consensus conference and the 2000 EC audit are the focus of the following section.
In 2002, an Expert Advisory Committee that was convened by the Veterinary Drug Directorate (VDD), Health Canada, submitted an extensive report on "Animal Uses of Antimicrobials and the Impact on Resistance and Human Health" (248). This Committee was established in 1999 as a result of the 1997 consensus conference. In total, the Committee made thirty-eight recommendations, twelve of which specifically pertained to veterinary drug regulations and five of which directed specific actions. These five recommendations are as follows: i) develop methods and criteria for human health safety assessment of veterinary drugs with respect to antimicrobial resistance; ii) define threshold levels of resistance for post-approval surveillance and provide for appropriate remedial action if thresholds are surpassed, up to and including modification of approval or suspension of marketing; iii) develop an extra-label use policy that ensures this practice does not endanger human health. Such a policy should include the ability to prohibit the extra-label use of specific drugs of critical importance to human health; iv) evaluate, register, and assign a drug identification number (DIN) to all antimicrobials used in food animals in Canada, whether manufactured domestically or imported. This includes bulk active pharmaceutical ingredient (API) products with intent to stop the direct use of APIs in food animals and; v) make all antimicrobials used for disease treatment and control available by prescription only.
In addition to these five suggestions, a recommendation was made to conduct risk assessments of currently approved antimicrobials, including but not limited to those with a growth promotion claim, for their potential effects on human health. This recommendation echoed the concerns of the European Commission in 2000. In response to the recommended strategies from this expert Committee, Health Canada has made efforts to address the first four of the above recommendations of the Advisory Committee. We did not identify any activities pertaining to changing veterinary drug access to prescription only or to conducting risk assessments for currently approved antimicrobials.

Criteria for Human Health Safety Assessment of Veterinary Drugs with Respect to AMR
The VDD has established a multi-disciplinary Expert Advisory Committee on Antimicrobial Resistance Risk Assessment with a mandate to provide advice on scientific methods to evaluate the human health risk from antimicrobial resistance (386). In particular, this Committee is focused on the validity of various risk assessment methodologies for establishing thresholds and application to drug monitoring. Over a three-year history, a major contribution from this Committee has been guidance on a classification scheme for antimicrobials according to their importance in human medicine (as described above) (363,387,388). This classification is used by the VDD in new antimicrobial drug applications (364).
The VDD requires new antimicrobial drug applications to include information pertaining to AMR, including applicant submitted risk assessments. These include data on the phenotypic and genotypic resistance mechanisms in the target bacteria and the resistance patterns expected in non-target bacteria. Applicants must consider AMR development and transmission including cross-resistance and co-selection in target pathogens, relevant foodborne pathogens, and commensal bacteria. There is specific focus on the possible effects of the antimicrobial on the flora of the intestinal and colonic microbiota with emphasis on Salmonella,Campylobacter,commensal E. coli and enterococci. (364).
The VDD has post-approval marketing requirements. Guidance documents for industry emphasize surveillance for resistance to antimicrobial classes considered critical to human medicine (364,388). This surveillance may be conducted through a combination of federal programs, the most applicable of which is CIPARS (364,383). However, given that critical thresholds triggering action have not been established due to their complexity, postapproval surveillance is in truth actually monitoring (389). A role of the Expert Advisory Committee on Antimicrobial Risk Assessment is to consider how risk assessments may contribute to the development of thresholds for action.

Extra-label Drug Use
Extra-label drug use (ELDU) is the actual or intended use of a drug in animals in a manner that is not in accordance with the approved label (285,390). For example, any use that is in a species, for indications, at dosages, frequencies, or administration routes that are not on specifically stated on a product's label are ELDU. Deviation from label use creates concerns about drug residues as the appropriate withdrawal time has not been determined for the circumstances of use. Antimicrobial resistance may be a concern if the use alters the selective pressure on bacteria.
Extra-label drug use is an important tool for veterinary medicine. Minor or exotic species have few drugs licensed for use in them, which limits labeled treatment options (286). Likewise, less common diseases may have limited or no label treatment options. Pharmaceutical companies typically seek label claims for major commodities and prevalent pathogens; demonstrating safety and efficacy of drugs is costly and time consuming so labels for minor uses often lack sufficient return on investment.
In Canada, extra-label use occurs in broilers and pigs (personal communication; Dr. Patricia Dowling, Director of Canadian Global Food Animal Residue Avoidance Databank (CgFARAD)). The CgFARAD has extensive data on possible ELDU in poultry because the CFIA requires that veterinarians obtain a CgFARAD issued drug withdrawal time recommendation for all poultry exposed to ELDU drugs. Producers report the ELDU and withdrawal time when they ship birds to slaughter. In chickens, most inquiries pertain to in-feed unlicensed drug combinations. Inquiries regarding AMU in-feed at doses that are not on the label or in age categories of birds that are not on the label is less common. Swine data are less complete because abattoirs do not require a CgFARAD withdrawal recommendation. Thus the inquiries are incomplete and more varied. To date, they have largely pertained to therapeutic AMU to control an endemic pathogen, Streptococcus suis.
Currently, extra-label use of non-prescription drugs is not limited to veterinarians but can be practiced by pharmacists, trainers, producers, animal owners, and other non-health care professionals. Untrained users may not appreciate the risks of ELDU to public health and food safety (285,286,391). In 2002, the VDD established a committee to provide expert advice and guidance on a commissioned study entitled 'Drug Use in Animals Study. ' This committee became the ELDU Advisory Committee and released an issues identification paper in 2004. Since then, Health Canada has released a policy on ELDU, recognizing its importance but also recommending against ELDU outside of a VCPR and for antimicrobials considered of very high importance to human health (388). The federal government's jurisdiction of ELDU is limited because of the provincial regulation of veterinary drug sales. This seems to have limited the VDD to making a recommendation, which is less enforceable than the American prohibition of ELDU for certain critically important antimicrobials, but does match the approach and tone of the United States. Overall, this policy aims to promote the prudent use of drugs in food producing animals in order to minimize the risks to public, animal, and environmental health and addresses the concerns raised by the Expert Committee and Advisory Committee (285).

Active Pharmaceutical Ingredients
The Health Products and Food Branch (HPFB) have been addressing concerns over the use of active pharmaceutical ingredients (API) in animals. Licensed establishments (i.e. pharmaceutical companies or research institutions), veterinarians, and pharmacists can import API. Some API are in dosage form, have demonstrated manufacturing standards that meet Canadian requirements, and are issued a Canadian DIN and permitted for sale. Others are imported as bulk products for further compounding and do not receive a DIN until they have been modified to dosage form using processes regulated by the government. Health Canada's greatest concern is with bulk API that are being sold as finished products for veterinary use thus bypassing manufacturing standards and controls. If importers lack good manufacturing practices, these products could include impurities leading to human or animal toxicity or inaccurate active ingredient concentrations. This could potentially result in residues in excess of maximum limits or inadequate levels resulting in inappropriate dosing and the potential for further resistance-pressure. Inappropriate bulk API use puts public health and export markets at risk. Within the HPFB, stakeholder consultation has occurred and a policy statement was released in 2007. The policy specifically pertains to concerns over the sale and distribution of bulk API to farmers, pharmacists, feed mill operators, and veterinarians for use as dosage drugs (392). The HPFB policy outlines how API should be handled and the appropriate regulation that governs importation and use of API.

Own use Importation
Importation of antimicrobials without a Canadian DIN by producers is another concern in Canada's veterinary drug regulations. The 'own-use loophole,' as it has become known, is a regulatory omission that permits ninety days of worth of a pharmaceutical drug that is not licensed in Canada to be imported for personal use. The intent of this regulation was to allow people receiving medical care internationally to return to Canada without interruption in therapy (393). However, it is used by producers to import pharmaceuticals that are either not licensed in Canada or are available more economically elsewhere. The ability to import drugs for 'own-use' purposes raises concerns about the impact that these drugs may have on antimicrobial resistance development and meat residues and toxic contaminants.
Health Canada has faced opposition to changing this regulation. Stakeholders argue that out-of-country access to pharmaceuticals is the only mechanism that ensures pharmaceutical prices remain competitive in Canada (393,394). Others have raised concerns about the VDD backlog in approving new-drug submissions. They argue that closing the own-use loophole puts industry at a competitive disadvantage due to delayed product access. An estimate from the Canadian Animal Health Institute suggests that about one-third of all veterinary drugs (not only antimicrobials) used in Canada are drugs imported by producers for their own use (380), but to date there are limited data on the effects that these importations may have on food safety or the economic effects on producers and the pharmaceutical industry. This issue remains at the consultation level. The task force that was put in place to examine this issue has called for further study before drawing any conclusions (394).

Limiting Antimicrobial Access through Prescription-Only Licenses
Continued access to non-prescription antimicrobials is counterproductive to prudent AMU guidelines, which are based on the premise that AMU should occur within the confines of a VCPR to ensure evidencebased decisions (395). It also could have a negative impact on animal health based on the lack of a veterinarian in the decision-making process and the potential for inadequate or contraindicated therapies. Over-the-counter (OTC) access to antimicrobials also complicates AMU data collection (281,396). The sale of veterinary pharmaceuticals is under provincial jurisdiction and Quebec has a prescription only system (248). Other provinces have struggled to address this issue because it requires changes to the pharmaceutical distribution system. Ethical debate over veterinarians concurrently prescribing and selling pharmaceuticals exists, and restricting antimicrobial access to prescription could be perceived as supporting a veterinary monopoly on drug sales (397,398). In some jurisdictions, this has been addressed by limiting veterinary profit on antimicrobial sales to a pre-defined handling fee. Another challenge for limiting OTC pharmaceuticals has been ensuring alternative access. In some rural areas competitor veterinarians may be distant and licensed pharmacists may not be trained in veterinary science. Our understanding is the only avenues open to the VDD are to re-license all antimicrobials as prescription drugs, effectively passing the distribution problem on to the provinces, or to continue supporting its policy statements that ELDU should be limited to VCPR and limit approvals of new applications for non-prescription antimicrobials.

Comparison of International Veterinary Drug Regulations and Policy
In summary, Canada has some distinct differences in veterinary drug regulations from Europe and the United States. First, in Europe all veterinary AMU must be accompanied by a prescription. In contrast, both Canada (excluding Quebec) and the United States permit non-prescription sales and use of antimicrobials. Second, European authorities do not allow antimicrobials to be included in animal feeds without a veterinary prescription and then use is only to address a disease indication rather than improve growth or productivity. In contrast, there are currently no feed-grade antimicrobials licensed for use in pigs or chickens in Canada that require a veterinary prescription when used according to the label (263,264). Label indications can include a list of specific diseases as well as a growth promotion claim. In Canada, feed-grade antimicrobials may be included in diets in an extra-label manner if accompanied by a veterinary prescription. Although non-prescription in-feed AMU is permitted in the United States, extra-label use is prohibited (281).
Superficially, the differences between Europe and North America may appear minor; but they belie a fundamentally different approach to food safety policy. Canada and the United States have committed to evidence-based policy while Europe has employed the precautionary principle in its legal statutes. The European Environmental Agency has defined and clarified the precautionary principle as, "the Precautionary Principle provides justification for public policy actions in situations of scientific uncertainty and ignorance, where there may be a need to act in order to avoid, or reduce, potentially serious or irreversible threats to health or the environment, using an appropriate level of scientific evidence, and taking into account the likely pros and cons of action and inaction" (399). The disparate approaches to regulating AMU in livestock mean that North America and Europe tend to interpret risk prioritization, assessment, and management differently. Ideally, risk assessment, evidence-based science, cost-benefit and precautionary approaches should be combined to provide a balanced decision-making process (400). As scientific research and observational studies help clarify the links between people, food, and animals, our approaches to managing AMR and veterinary AMU may become increasingly similar.

surveillance and Monitoring of Antimicrobial Resistance and Use
There is international consensus that countries should conduct AMR surveillance in animals and/or food to monitor trends, provide a basis for the development of national policies, and assess interventions. Antimicrobial use data are essential to understand the causes of AMR. The WHO, FAO, and OIE recommend that countries collect veterinary AMU data (359). At a minimum, these data should include the national use of antimicrobial agents in kilograms of active ingredient on an annual basis and be reported using the Anatomical Therapeutic Classification system (ATC). When possible, data should be stratified by animal species (359,401). Many countries have made great progress in conducting AMR surveillance and most have faced challenges in responding to recommendations for AMU data collection.

Basic Premises of AMR Surveillance
Surveillance is a continuous and systematic process of collecting, analyzing, interpreting, and disseminating descriptive information for the purposes of understanding and describing health issues (402). Over time, surveillance can be used to identify trends and emerging situations. Surveillance can detect and report on the incidence or prevalence of health outcomes and identify at-risk groups. Surveillance data may be useful for assessing the effects of interventions applied to the issue of concern. There is often a balance between the desire for the best information and feasible data collection (402). The ideal national system should include monitoring both animal and human use as well as resistance levels in bacterial species from animals, animal derived foods, and humans. Several factors are important to consider when designing an integrated monitoring system, including: the main purpose of the system, the reservoir of interest, the bacterial species, and the antimicrobial agents to include.
As a result of needing long-term cooperation from participants, and due to the breadth of surveillance in time and geography, surveillance is generally less detailed and precise than research projects. However, surveillance and research can complement each other. If through surveillance a particular area of interest or concern is detected, then research can be used to investigate the issue in more depth and detail. Conversely, research studies can identify new issues that may benefit from formal ongoing surveillance.
Surveillance can be active or passive (403). Active surveillance employs statistically valid sample collection to obtain unbiased data from a targeted area or group. It must be implemented consistently across groups and over time. There is a need for ongoing contact with the participants in order to be able to collect specific data from the identified population in a consistent manner. It may involve the use of sentinel sites that act on behalf of the surveillance system to provide the required information. Active surveillance requires substantial financial and human resources. Passive surveillance is less demanding on human and financial resources. In passive surveillance, the data or samples often arise from submissions to a lab, clinic, hospital, etc., and are subsequently provided to the surveillance system using a standardized protocol. In passive surveillance, the surveillance program has no influence on sample collection.
Both active and passive surveillance can provide useful information. Passive surveillance is more prone to bias than active surveillance (402,403). People seeking treatment could have differences in their infecting pathogens compared to people with mild symptoms that do not seek treatment. In food animal AMR, passive data typically represent clinical isolates while active data often represent healthy individuals. These clinical isolates may have different rates and patterns of resistance than isolates of the same bacterial species obtained from healthy animals. This may be due to linkages between virulence and resistance gene, differences between commensal and pathogenic strains of a bacterial species, and therapeutic drug exposure (26,48). This bias is amplified in passive surveillance of clinical isolates from animals because producers incur laboratory expenses. Diagnostic submissions are often only made for severe or non-responsive disease outbreaks, thereby creating a tendency to receive more highly resistant bacteria (404). Conversely, utilizing samples from unrelated mandatory monitoring programs can artificially increase the number of submission from a certain species or region.
Passive surveillance of AMR in bacteria from clinically ill animals increases the likelihood of detecting emerging resistance, as this is a form of targeted surveillance. It does not reflect the probability of human exposure to resistant bacteria because only healthy animals enter the food chain. While both healthy and sick animals harbour resistant bacteria, the patterns, frequencies and genetic basis for resistance differ (26,136). For these reasons, passive surveillance of clinical isolates is used as only one component of an AMR agri-food surveillance system. Surveillance programs must take all of the relative strengths and weaknesses of passive and active data collection from animals, food, and people into consideration, along with the objectives and resources available, to determine the most appropriate program design. Each piece can provide information on different populations, and therefore, studying both presents the most complete picture.
Surveillance from farm to fork is useful for assessing the impact of AMU in livestock and poultry production on AMR throughout the food production chain. Farm level AMU and AMR surveillance allows considerations of what drug are used and how they impact the AMR patterns on the farm. Collecting samples at abattoir can provide information about changes in AMR patterns between what is detected on farm and what is being detected in the processing plant. Depending on the stage of processing that these samples are taken at and how they are taken, this may allow for assessment of the effectiveness of HACCP programs, cross contamination, or changes in intestinal microbial flora as a result of transport and/ or stress in the animals. Abattoir surveillance is the easiest point to ensure samples are representative of the livestock population and unbiased. Between slaughter and retail, the microbial load, resistance rates, and resistance patterns can again change. So, retail surveillance provides the best estimate of human exposure to resistant organisms. A challenge of interpreting retail surveillance is incomplete information identifying a product's source. Knowing if a product is domestic and where it came from within the country or if it is imported would obviously assist with the interpretation of data collected at all points of the surveillance program.
In summary, the 'farm-to-fork' approach is useful for understanding the transmission of resistance through the food chain and identifying potential areas for interventions and areas requiring further investigation.
In future research and surveillance, the farm-to-fork concept should be expanded to foodborne disease in humans. This is currently being attempted outside of AMR surveillance programs via source attribution projects (96,405) but to our knowledge, no country is attempting to monitor the effect of AMR on clinical outcomes of foodborne disease or the proportion of resistant foodborne infections that occur in individuals taking antimicrobials prior to infection.

National AMR and AMU Monitoring and Surveillance Programs
There are many national AMR monitoring and surveillance programs. These programs are challenging to catalogue and many have not published their methodologies in peer reviewed literature. Websites and annual reports were reviewed for methodological information, but were often incomplete. Summarizing all of the identified programs would have been fraught with errors by omission and somewhat redundant (406). The following section highlights surveillance programs as they relate to pigs and chicken and their meat and evaluates their relevance to Canada. We have selected the programs from the United States because Canada was designed to harmonize with this program, Europe because they have been instrumental in establishing methodologies and ensuring ongoing prioritization of surveillance, and Japan because it was the only program identified outside of Canada that uses on-farm sampling, and the only well described program outside North America and Europe.

Canada
The Canadian Integrated Program for Antimicrobial Resistance Surveillance (CIPARS) was established in 2002 by Health Canada (66). This is a national program that collects, integrates, analyzes, and communicates trends in antimicrobial use and resistance in select species of enteric bacteria from humans, animals, and meat across Canada. The goals of CIPARS are to: i) use a unified approach to monitor trends in AMR and AMU in humans and animals; ii) generate timely reports; iii) generate data to facilitate the assessment of the public health impact of AMU in people and livestock; and iv) allow accurate international comparisons with other countries that use similar surveillance systems.
The several components of CIPARS (Figure 3) are harmonized to monitor temporal and regional trends in the prevalence and patterns of AMR across regions, bacterial species, and hosts. Results of CIPARS surveillance are published in annual reports, and results of detailed studies are published in peer reviewed journals.
The human component of CIPARS involves passive surveillance of Salmonella isolates from clinical cases. Salmonella isolates are sent from provincial public health laboratories and reference laboratories across the country to the National Microbiology Laboratory for phage typing and antimicrobial susceptibility testing. As Salmonella is a reportable disease in humans, these isolates should represent the vast majority of clinically diagnosed cases. Clinical Salmonella isolates from animals are also tested. These originate from veterinarian or producer submissions to veterinary diagnostic laboratories. In animals, Salmonella is not reportable and coverage between provinces is variable.
In agri-food, surveillance is conducted in beef, chicken, and pork, but only chicken and pork are described. The retail surveillance component was initiated in 2002 and provides data on antimicrobial resistance in Salmonella, Campylobacter, and Enterococcus from raw chicken (skin on wings and legs), and generic E. coli from pork (chops) (Figure 2, Table 4). Weekly samples are submitted from retail stores. Stores are selected by a stratified sampling scheme which randomly selects census divisions, weighted by the population of each of the provinces. The abattoir component isolates bacteria from swine and poultry cecal contents. The unit of concern is the bacterial isolate. Salmonella and generic E. coli are isolated. The sampling design is a two-stage sampling. Each commodity is handled separately. The first stage randomly selects slaughterhouses by a probability that is proportional to their annual slaughter volume. Antimicrobial use data are currently collected in people and pigs. Human drug use data are obtained through Intercontinental Medical Statistics (IMS) data and describe oral human antimicrobial consumption at the community pharmacy level. Data are computed using DDD from dispensed prescription data. As more fully described in Chapter 2, animal drug use data are not readily available from a single source in Canada or the United States. Use data for pigs are obtained through farm surveillance. Insight into AMU in other commodities currently occurs through collaborative research projects. Recently, antimicrobial distribution data have been provided by the Canadian Animal Health Institute. These data will provide a context for interpreting livestock AMU data generated through research and farm surveillance (66).
CIPARS has consistently, and continues, to expand to include more commodities, bacterial species, and geographical regions within Canada ( Figure 4). As CIPARS expands, it will continue to act as a research platform to investigate AMR and as a critical source for gaining insight into emerging trends in AMR and AMU over time.   (Table 4). NARMS differs from CIPARS by rolling its sampling focus through the livestock commodities so evaluating results for trends is more difficult.
The United States does not currently collect any AMR data from animals on farms. A pilot program called Collaboration in Animal Health and Food Safety Epidemiology (CAHFSE) was run in swine from 2003 to 2005 (73). From the beginning, this program was designed to only run a few years followed by an evaluation period. American studies have recognized that an on-farm surveillance component is one of the few feasible ways to collect AMU data (281,396,408). No information was found pertaining to the potential re-initiation in swine or expansion of farm surveillance or pilot projects in other commodities. The United States describes AMU through surveys conducted by the National Animal Health Monitoring System (NAHMS) program. These surveys rotate between commodities and target appropriate regions of the country to achieve representative data for that industry. Commercial swine production was last targeted in 2006 and commercial poultry 2004. A small enterprise study that covered both poultry and swine was conducted in 2007 (409).

Europe
Antimicrobial resistance is monitored in most countries within Europe. A survey performed in 1998 outlined the methodologies of 12 countries in the European Union (EU). At that time, the most frequent bacterial group tested was Salmonella (410). The European Antimicrobial Resistance Surveillance System (EARSS) monitors resistance in human isolates for a variety of bacterium including E. coli and Enterococcus (151). Results from these bacteria in particular are of importance due to their potential foodborne connection. Participating EARSS laboratories within each country collect data obtained from routine susceptibility testing from invasive isolates of these bacterial species along with patient information such as age, sex, birth date, and level of care. These data are then provided to EARSS from the participatory nations. There are currently thirty-one European nations who participate in EARSS. Data provided to EARSS are reviewed by experts before publication in annual reports. This allows for a more thorough representation of the current status of antimicrobial resistance within Europe given resistance does not obey national boundaries.
The new Zoonoses Monitoring Directive was adopted by the European Food Safety Authority (EFSA) in 2003 to improve the comparability of data of member countries and to consistently monitor AMR (97). Data from 2005 onward are reported based on the new directive. These data include AMR in Salmonella, Campylobacter, Enterococcus, and E. coli. Bacterial isolates are obtained from humans, animals, and foods for susceptibility testing. Considering that extensive harmonization in surveillance has occurred across European countries, and that a single report summarizes both the results and methodologies of individual country's programs for AMR data from food and animals, we felt a detailed description of each European country's surveillance program was redundant (97,360,361). Instead, select national programs are described to highlight similarities and differences from Canada. Denmark was selected as the longest-running and most comprehensive AMR and AMU surveillance program in the world. Norway was selected as an example of a comprehensive program that is representative of Northern Europe. Both of these programs have been in place for over a decade. Many of the more recently developed European programs are less comprehensive but may still be in the development stages.

Denmark
Denmark established the world's first systematic and continuous monitoring program in 1995 (89,322,411). The Danish Antimicrobial Resistance Monitoring and Research Program (DANMAP) examines representative bacterial isolates from animals, foods, and humans in Denmark. Isolates include zoonotic and indicator bacteria (Table 4). Sampling occurs in food animals at abattoirs and through diagnostic submissions. Cecal contents of pigs and cloacal swabs of broilers are obtained and susceptibility testing is carried out for Salmonella, Campylobacter, Escherichia coli, and Enterococcus faecium/faecalis. Salmonella isolates collected from systematic random sampling from diagnostic submissions to the Danish Veterinary Laboratory and the laboratory run by the Federation of Danish Pig Producers and Slaughterhouses are also tested. Isolates are also included from submissions to the National Food Institute. Sampling of predetermined foodstuffs from retail outlets and wholesalers also occurs. The foods sampled include both Danish and imported foods, and Salmonella, Campylobacter, E. coli, and Enterococcus are tested.
Antimicrobial use data in people are available electronically from pharmacy sales (89). Data on animal use are also available electronically through pharmacy sales data. Veterinarians can sell medicines to farmers but profits are limited to 5%. Therefore, most antimicrobials are sold on prescriptions. The Danish Plant Directorate monitors the consumption of in-feed antimicrobials which can only be sold at feed mills in the form of premixes. Producers of premixes are required to report quantities of active ingredients used. This provides consumption data but no information on the recipient animal species (89). These limitations were addressed by the VETSTAT program, which was established in 2001 to continuously monitor the use of all prescription medicines in animals at the individual herd level (412). Veterinarians are required by law to report the sale of medicines to a central database. Information submitted includes the identity of the farm, receiving medicine, species of animal, age group, and reason for prescription.

Norway
The Norwegian AMR surveillance program (NORM) was established in 1999 to monitor resistance in human pathogens and the veterinary arm (NORM-VET) was established in 2000. An action plan developed by the Norwegian Ministry of Health and Social Affairs described the need to address antimicrobial resistance at a national level and emphasized the importance of ongoing surveillance and monitoring of both human and animal sectors to aid in controlling antimicrobial resistance. The NORM-VET program rotates through the food-animal commodities with a particular emphasis on one species per year. In the target commodity, commensal E. coli and Enterococcus spp. are described. Chickens were last targeted in 2006 and pigs in 2004. Four enteropathogenic bacteria are monitored in animals and people annually: Salmonella, Campylobacter,Yersinia,and Shigella (79). Salmonella from all food-animal species are obtained during clinical examinations or necropsies at the National Veterinary Institute. Campylobacter is cultured from samples collected at slaughter plants. One isolate per farm is included. The NORM program obtains human isolates from clinical specimens.
In Norway, all antimicrobials intended for human use are prescription and sold through pharmacies. Therefore, the human use data presented include total sales of antimicrobials for humans in Norway. The unit of measurement for human antimicrobial use is defined daily doses (DDD), reflecting the usage of the Anatomical Therapeutic Chemical (ATC)/ DDD system. In Norway veterinary antimicrobials intended for domestic animals and farmed fish for therapeutic use are available through prescription only and only dispensed through pharmacies. Medicated feeds have to be prescribed by veterinarians and produced by feed mills and are only available for farmed fish. Given the small size of herds/flocks in Norway, livestock are treated with antimicrobial agents, prescribed by veterinarians, through drinking water or injection. The reporting of sales of veterinary drugs is mandatory and, therefore, the number of items sold per year, drug formulations, strengths, and package sizes are able to be obtained. Drugs are classified using the Anatomical Therapeutic Chemical for Veterinary medicinal products (ATCvet) classification system and presented as the kg of active substance sold in the country. Data are not stratified by species (79).

Japan
The Japanese Veterinary Antimicrobial Resistance Monitoring Program (JVARM) was established in 1999. The objectives of JVARM are to monitor antimicrobial use and resistance amongst zoonotic and indicator bacteria from healthy animals and pathogens amongst diseased animals. Zoonotic (Salmonella and Campylobacter) and indicator (Escherichia coli and Enterococcus) bacterial isolates are obtained from healthy animal (cattle, pigs, broilers, and layers) fecal samples. The JVARM was the only program identified that collected samples from farms. Six samples are collected per prefecture (region) annually with a limit of one sample per farm. Antimicrobial use data are acquired from pharmaceutical companies which produce and import antimicrobials for animals. The annual weight in kilograms of the active ingredient of approved antimicrobials used in animals is collected. This only includes therapeutic AMU in animals (413-418).

Studies and Pilot Projects
Many countries have not met WHO and OIE recommendations to implement national AMR surveillance or monitoring programs. In some, studies or pilot projects have evaluated AMR. A surveillance system pilot was implemented for Salmonella spp. in Mexico in 2002 through to August 2005. Active surveillance began in 2002 integrating samples from ill and asymptomatic persons and retail pork, chicken, and beef. In 2003, intestines of chicken, swine, and cattle were collected from slaughter (419). A study from South Africa was conducted to aid in the establishment of an integrated AMR monitoring program. The study was performed to make recommendations for a practical and sustainable monitoring program that could provide information on a national scale (420). Australia currently does not have a national AMR monitoring or surveillance program. However, the Joint Expert Technical Advisory Committee on Antibiotic Resistance (JETACAR) has made a recommendation to establish an AMR surveillance program which will follow OIE recommendations (421). In other countries, research studies have been conducted. The following references provide evidence that research in AMR in food animals is occurring on every continent except Antarctica (66,77,97,115,419,(422)(423)(424)(425)(426)(427).

(C) Campylobacter (S) Salmonella (Ec) Escherichia coli (En) Enterococcus
* Surveillance or Monitoring Program rotates between species annually for data collection.
** Information unavailable. Does not imply the program does not perform type of monitoring but rather information has not been found within the scope of this paper. † Agricultural AMU data are obtained through several different methods, such as studies and national distribution data.
-Monitor AMR in animals and humans but methodologies are unclear. No differentiation made between active monitoring from a retail or abattoir program or diagnostic submissions.
Note: Information lacking within this chart does not mean the program does not perform such monitoring but rather information has not been found within the scope of the literature reviewed for this paper.

International Comparisons
International surveillance programs have become increasingly harmonized over the last decade. Many of the differences in sample collection and susceptibility testing are being overcome, which makes data comparisons more valid. The described programs differ in the regularity of their sampling schemes (Table 4). The CIPARS collects a predetermined number of isolates, for each studied bacterium, from each commodity every year (66).
Most other programs rotate their focus to a particular commodity each year with or without a base level of sampling across all commodities (97,407). The CIPARS approach helps to identify changing trends in resistance over time. It also makes random variation due to chance appear less important. For example, CIPARS presents data trends and the 'noise' around resistance estimates is obvious. However, if data were collected sporadically, this noise could be considered representative of reality. Although the CIPARS approach of regular sampling is more expensive from a laboratory perspective, it may allow efficiencies in labour and program management. Effective relationships are established with sample collectors and once trained, minimal interaction is required.
Most integrated programs collect human clinical isolates along with food-animal isolates from abattoirs. Retail sampling is relatively common while only Canada and Japan collect data and samples from farms (415). The merits of on-farm AMR data have been discussed and largely relate to avoiding ecological bias between farm-level AMU data and abattoir-level AMR data, as well as the ability to look at management factors beyond AMU for their influence on AMR. There are currently no other viable sources of valid AMU data in Canada, although the possibilities of using surveys or data extrapolation continue to be evaluated. The main benefit of such alternative AMU methods would be cost and labour savings.
Other countries have also struggled to collect valid and useful AMU data. Many European countries report sales volumes which are not stratified by species. These data are of minimal use and are not available in Canada because of privacy regulations for pharmaceutical corporations (281,396,408). The Danish VETSTAT program is a census style farm-level data collection system that inventories the national AMU (412). This program is detailed and world-renowned. However, it cannot be applied to Canada for at least three reasons. First, in Denmark, all AMU is prescription only and largely occurs through pharmacies. This facilitates drug tracking and allows validation of the farm-level data. Second, the VETSTAT system is mandatory while Canada would require a voluntary program. Denmark appears to have different privacy regulations than Canada which facilitate regulated data collection. But even beyond these, operating the VETSTAT program is simplified by the structure of the livestock industry. The beef, dairy, and poultry sectors are relatively small and although the pork sector is large, it is integrated with producer co-operatives as the owners (428). This makes implementing and enforcing farm-to-fork food safety systems in Demark much easier because producers have a unified and influential voice in negotiations. The final reason the Danish VETSTAT system is not applicable to Canada is that Canada does not possess the pharmaceutical industry infrastructure that enables electronic data collection.
Canada's AMU data collection system most closely mirrors the pilot program run by CAHFSE in the United States, but the CAHFSE pilot program ceased operations in 2005 (73). As evidenced by the discussed veterinary drug regulations, substantial differences exist between North America and Europe. We expect that a North America AMU surveillance approach will emerge with minor modifications to account for Canadian and American differences (408). This will benefit both countries as harmonized AMR surveillance already occurs. In contrast, harmonization with existing European AMU surveillance is not realistic. The greatest issue that we foresee for AMU and AMR surveillance in Canada is financial sustainability. Continued commitment from the federal government is needed to ensure the long-term viability of this world-class program.
A weakness of this report is an inability to describe or compare the financial costs of national surveillance programs. An economic estimation of the costs to the Danish livestock industry from the AGP ban has been published, but did not extend to considering the cost of DANMAP or VETSTAT (341). The lack of information of cost efficiency of surveillance systems is an important information gap for decision-makers.
The surveillance programs described would be more appropriately referred to as monitoring programs. A monitoring system is an ongoing effort to describe disease in a population. Surveillance is distinguished from monitoring by having predetermined actions that occur when data indicate disease prevalence or incidence has risen above a threshold (403). The importance of antimicrobials for human medicine is only one component required to establish thresholds beyond which the risk to public health from AMR in food animals is deemed unacceptable. The second piece is the likelihood of humans being affected by the resistance elements in animals. In 1999, the FDA proposed to categorize antimicrobials according to the likelihood of human exposure to resistant bacteria from food animals that are either human pathogens or may transfer resistance elements to human pathogens. This would involve considering four likelihoods: i) that use in food animals will induce resistance in bacteria; ii) that food-producing animals will promote such resistance; iii) that resistant bacteria will be transmitted to people; and iv) that transfer will result in the loss of available human antimicrobial therapies (429). To our knowledge this categorization has not been released by the FDA or any other regulatory body.
In this same document, the FDA proposed that these two categorization schemes could be combined to establish and monitor thresholds for AMR. Such thresholds would be based on two premises. First, that a regulatory agency could determine a threshold of resistance that would adequately protect public health, and second, that the regulatory agency has the ability to detect when that threshold is reached.
Numerous concerns were raised about the validity of these premises (429,430). These included: What determines a safe threshold? Should thresholds identify resistance frequencies where public health is initially affected, based on the premise that interventions will reduce resistance, or should the threshold be lowered to ensure that limiting further increases is sufficient to protect public health? Should resistance be monitored in food animals, food, or human cases? And should it account for resistance in pathogenic bacteria or sentinel bacteria? Who is responsible for the surveillance and its costs? And what actions should be taken when a threshold is exceeded: increased monitoring, restricted use, or complete cessation of use?
These are a few of the questions that have hindered regulators' abilities to establish action thresholds for AMR and that require consensus from scientists and stakeholders for progress. These concerns and questions are by no means limited to the United States but pertain to all countries attempting to understand and address AMR and AMU in food animals.

Conclusion
Even though North America has based, and continues to base, regulatory changes on scientific evidence, determining the appropriate time, sufficient evidence and appropriate response to scientific evidence is influenced by political pressures from public health authorities, the agriculture industry, the pharmaceutical industry, trading partners, and the public. What seems insufficient to some often seems too drastic to others. Establishing action thresholds for regulations is extremely difficult and potentially fraught with significant resistance from either side of the debate. There is a continued need for surveillance activities that can monitor trends over time in both AMU and AMR for humans and animals to help provide the critical data that are needed to assist policy makers in their decision-making process.

Chapter 4: Agricultural and Agri-food interventions to Reduce the impact of AMR bacteria in Pigs and Chickens on Human Health
Introduction Pressures to decrease antimicrobial use (AMU) in meat animal production are being prompted by public health concerns, consumer demands, and narrowing profit margins (11,291,431). Canada's regulatory agencies are addressing international recommendations and standards (269,359). The livestock industry is also facing recommendations from national and international groups to eliminate unnecessary AMU and to decrease their reliance on antimicrobials by improving management and increasing the use of antimicrobial alternatives. Some recommendations are specific, such as the requirements imposed by fast food chains on their suppliers, while others give generic guidelines such as emphasizing prevention of infectious diseases to avoid AMU (122,291,395,432,433).

Scope and Objectives
This chapter is written for professionals in human medicine and public health who have minimal to no background in agriculture. It aims to inform these professionals about industry-led initiatives in pigs and chickens that mitigate foodborne AMR risks. It also describes alternative inputs or management practices that are available to producers to improve animal health including vaccines, competitive exclusion strategies, and biosecurity practices. It provides examples rather than an exhaustive catalogue of all alternatives and does not quantify the extent that each are used. Practices that control pathogens on-farm or bacterial contamination of meat during slaughter and processing with no plausible influence on the prevalence of resistant bacteria are excluded.

Basics of Chicken and Pig Production
The term 'broiler chicken' refers to birds raised for meat. Broiler producers receive day-old chicks from the hatchery. These birds are raised as a flock using all-in-all-out management which means that all chicks arrive together, are raised together and are marketed together. Depending on the target market, birds are finished between 1.75 and 2.2 kg and generally range from 34 and 42 days of age. Although the birds are managed as a single group and remain in the same barn for their lifespan, they are managed in phases. Each phase lasts approximately 14 days and progresses from the starter phase, through the grower and finisher phase, to the withdrawal phase. Disease pressures vary between phases and are greatest early in life. At each phase change, the birds' diets are altered and feed medications or coccidiostats may also be changed (personal communication: Dr. Tom Ingles, Poultry Health Services, Airdrie, Alberta).
In North America, pigs are also managed in phases, but progression between phases typically entails physical movement of the pigs. The suckling phase extends from birth until weaning. Pigs are weaned between 15 and 30 days, with most herds weaning pigs slightly less than three weeks (126,409). Weaning marks the beginning of the nursery phase. Piglets may be moved to a separate pen in the same room as sows but are more commonly moved to separate rooms or even sites. The nursery phase is associated with increased disease risk because stressors, including mixing and diet changes, occur concurrently with declining maternal immunity. Pigs generally remain in the nursery phase until they are 6 to 10 weeks of age (average, 61.8 days; S, 0.6) (409). Movement to the grow-finish phase completes the production cycle. Pigs remain in this phase until marketed or selected for the breeding herd. In Canada, pigs are typically marketed at 110 to 115 kg live weight and are generally between 24 and 30 weeks old (66).

Industry Background
Canada produces over a million metric tonnes of chicken on 2,000 farms annually. Most of this chicken is retained for domestic consumption while 15% is exported. This makes Canada the 13th largest global chicken producer and the 7th largest exporter (although our exports amount to less than 5% of the global leaders; Brazil and the United States) (434-436). In contrast, Canada produces a considerable proportion of the world's pork; ranking among the five largest producers in the world and accounting for 17% of the world's exports. In 2007, 31 million pigs were produced with one-third exported live. The remaining 21 million pigs resulted in slightly less than 2 million tonnes of pork, of which 55% was exported (435,437,438). Despite Canadians consuming slightly more chicken than pork (2008 average per capita consumption: Chicken, 11.2 kg; Pork, 9.7 kg (439), our pork industry is far larger than our chicken industry because it is export based.
Canada's poultry industry is supply managed. This is a marketing system that regulates domestic chicken production and imports. It matches demand to supply thereby ensuring that there is a consistent national market with stable commodity prices. Production is allocated to farms based on the amount of quota they own. Only producers that own quota can raise more than 300 birds annually and sell them to processors at the industry set price (440-442). This has protected the chicken industry from much of the market volatility experienced in the pork sector. Vertical integration is the term describing a single corporation owning all, or most, of the steps in a production chain. This business model is pervasive in much of the poultry industry worldwide (443). In Canada, the poultry industry has been described as vertically coordinated. While few corporations control the majority of slaughter, processing and distribution of chicken, and these corporations typically own or have strategic alliances with hatcheries, independent farms raise the broilers on a production contract with the processor. Feed mills are generally independent of both the producers and processors (441) (personal communications: Dr. Stewart Ritchie, Canadian Poultry Consultants and Dr. Agnes Agunos, Public Health Agency of Canada).
In Canada, the swine industry is not supply managed. Although the last three years have seen a contraction in the industry, this contraction follows twenty years of substantial growth. The expansion of the Canadian swine industry was accompanied by industry consolidation to capture economies of scale. The number of herds decreased, while herd size increased and large corporate producers became established. In some instances, there is vertical integration with single corporations owning multiple or all aspects of production including feed mills, pigs and slaughtering facilities. But much of the industry remains segmented, particularly in comparison with the American industry (444).
Vertical integration, or the lack of it, has implications for AMU decisions. Integrated companies can consider decisions from numerous perspectives, and the costs incurred in one division may be sufficiently offset in another to justify their incurrence (445,446). These operations may have better production data, ability to track the effect of decisions, and specialized staff, which could all influence the scientific and economic rigour applied to AMU decisions. The lack of vertical integration in the Canadian industry, particularly when animals change production phases and ownership, also affects AMU decisions. Hatcheries may treat eggs differently from breeder flocks with a known versus unknown health status, and may make different AMU decisions for domestic versus imported eggs. In the swine industry, some producers purchase weanling pigs and raise them to market weight. Again, AMU may differ in herds purchasing from a single supplier with a known-health status versus multiple suppliers.
Large producers and corporate farms that own multiple barns make AMU decisions that affect many animals. For example, in 2006, 40% of American farms used pre-weaning antimicrobials in feed but only 10% of pigs produced were exposed (409). Thus, 60% did not use pre-weaning antimicrobials and 90% of pigs were not exposed at weaning. This example shows that industry consolidation can positively affect AMU when large producers decide to limit or discontinue AMU. Supply chains can also affect AMU decisions. For example, numerous international fast-food chains impose the National Chicken Council Animal Welfare Guidelines on their poultry suppliers across North America (The Chicken Farmers of Canada's recently released Animal Care Guidelines may supersede these in Canada in the future). Under this and the related Canadian programs, diseases indicative of poor welfare, including mortality and lameness, require immediate euthanasia, veterinary advice, and appropriate therapy (447-449).
Supply management may have implications for AMU decisions. Supply management provides financial stability to chicken producers, relative to producers of export-based commodities (namely pork and beef) (441). This may contribute to an industry stance on antimicrobial use that has financial consequences but consumer and/or political appeal. Examples include an industry-mandated recommendation against active pharmaceutical ingredient (API) use and the own-use importation of drugs. In contrast, the Canadian pork industry is export-based and drug-use decisions may be influenced by importing nations' and customers' standards for antimicrobial use or resistance (450,451). As such, the Canadian Pork Council has stated in its Canadian Quality Assurance Policy on Drug use that own-use importation of drugs is not permitted, but that API use may occur providing that producers do not use bulk API and that the final dosage form is accompanied by a veterinary prescription and has been identity tested (452).

HACCP: on the farm and At slaughter
Hazard Analysis Critical Control Point (HACCP) is a food safety management system that enables the food industry to pre-empt problems rather than rely on end product testing. HACCP consists of six generic steps. These systematically apply science to control and document the safety of a food product. Since inception in the 1960s, this generic system has been effectively applied to a wide-range of food processing facilities (453-455). On-farm food safety programs (OFFS) can be HACCP-based (284,458). They are not true HACCP systems because biological processes are inherently more variable and less controllable than industrial processes. The Canadian programs described next provide a general HACCP plan that is applied to all producers. In contrast, a true HACCP program is tailored to each process. Rather than using CCP, OFFS programs identify good production practices (GPP) that might minimize the risk of the hazard at any point in production, not just at a specific point. Finally, OFFS programs are not HACCP programs because while the OFFS requires a plan, with compliance ensured through external audits and verification, they lack monitored thresholds at CCP (284,458). This prevents processors from knowing the entry contamination levels from which they must act to achieve their CCP levels.
Canadian OFFS programs cannot yet feasibly address AMR because intervention points and thresholds must be science-based. More research is needed on how and where resistance elements enter flocks/herds, the transmission dynamics once they are present, and the factors that allow persistence. These same questions also require investigation at the industry level. However, it is foreseeable that OFFS programs could identify batches or flocks with AMR frequencies exceeding a threshold, which could subsequently be moved into low-risk product streams, such as further processed and pre-cooked meats. Such an approach has been taken with Salmonella and Campylobacter control in Europe (64,459,460).

on-farm food safety Programs
The Canadian Pork Council's Quality Assurance (CQA)

and the Chicken Farmers of Canada's Safe, Safer, Safest Programs
A lack of data describing AMU in agriculture has been a major stumbling block for understanding AMR. In Canada, producer-led organizations in swine and poultry administer on-farm food safety programs that collect data on antimicrobial drug use (284,458). The 'Safe, Safer, Safest' program was initiated by the Chicken Farmers of Canada in 1998, and the Canadian Pork Council implemented the Canadian Quality Assurance (CQA) program in 2001. These programs are designed to ensure best management practices in biosecurity, disease control, sanitation, drug handling, and residue avoidance.
Producers are required to record AMU data. The AMU data are not compiled or analyzed across the industry. The primary objective of recording AMU data is to mitigate risk from antimicrobial drug residues, and so these programs focus their description of AMU near to market. In pigs, AMU through feed is described through the entire production cycle and through injection and water after weaning. In poultry the AMU data are limited to the two weeks closest to slaughter. The AMU data do not accurately capture exposure dose, duration or drug combinations. In our opinion, the most valuable data from these programs for addressing AMR is the information on producer decisions and attitudes towards AMU. If compiled, these answers could facilitate the development of effective programs by describing producer knowledge uptake and compliance with OFFS programs. Although Canada's OFFS programs were not designed specifically to control AMR, they do provide an on-farm link to inform producers about risks from AMU. The existence and survival of these programs demonstrates an industry-wide willingness to respond to regulatory and consumer concerns over food safety.

Prudent Antimicrobial Use guidelines
Prudent use, judicious use, or antimicrobial stewardship involves reducing inappropriate AMU and selecting the optimal drug, dose, and duration when use is necessary to ultimately reduce the emergence of AMR (461). A necessary tool to ensure prudent use is education, thus the standardized AMR education for Canadian veterinary students that was jointly developed by the Canadian Veterinary Medical Association (CVMA) and the Canadian Committee on Antibiotic Resistance (CCAR) in 2007 should improve AMU by future veterinarians (462). In veterinary medicine, several organizations have designed prudent use guidelines which have been summarized by Weese (395). These consistently recommend avoiding unnecessary AMU. They also emphasize the importance of basing necessary use decisions on susceptibility testing, and they suggest that when an antimicrobial is required that the narrowest spectrum drug is administered for the shortest time possible. These guidelines all support the notion that AMU should involve veterinary advice (268). For example, the CVMA has stated that all antimicrobials, including non-prescription antimicrobials, should be used within the confines of a veterinarian-client-patient relationship (VCPR) (390).
Published AMU recommendations target veterinarians because of the premise that AMU should occur within a VCPR. Again, using the CVMA as an example, the recently released Prudent Use Guidelines 2008 for beef cattle, dairy cattle, poultry, and swine recommend that veterinarians: i) take efforts to design a health management program to prevent and reduce disease; ii) look for antimicrobial alternatives; iii) prescribe all antimicrobials within a valid VCPR; and iv) select and use antimicrobials appropriately (390). Detailed recommendations are accompanied by treatment protocols. These guidelines should be a valuable resource for veterinarians, but their influence is unknown because their uptake and use is not tracked. More specifically, the uptake of treatment protocols is unmeasured and may change over time if these guidelines are not flexible and kept current.
Critics have raised concerns that, although these protocols are written for veterinarians, they may serve as a 'therapeutic recipe book' for producers, thereby undermining efforts to bolster VCPR. This, too, would be interesting to monitor.
Focusing guidelines on veterinarians may be valuable for initiating discussions about prudent AMU amongst veterinarians and between veterinarians and clients, but does not address the fact that most antimicrobials in Canada are available without a veterinary prescription. Administering antimicrobials within a VCPR is ideal because the veterinarian understands animal health as well as the herd's disease status and so can design treatment regimens to maximize efficacy while minimizing bacterial resistance. Veterinarians are trained to account for complex interactions between microbe pharmacodynamics and antimicrobial pharmacokinetics through their knowledge of physiology, anatomy, and disease pathology (463,464). However, non-prescription antimicrobials are available in Canada and the challenges involved in limiting antimicrobials to prescription only were discussed in Chapter 3. So, failing to target non-veterinarians in prudent AMU guidelines is an important oversight in Canada.
Only one education program for producers pertaining to appropriate antimicrobial drug use was identified. Ten years ago, the Ontario Ministry of Agriculture, Food and Rural Affairs (OMAFRA) developed a swine medications course as a proactive response to address the possibility that certification would be required to purchase non-prescription antimicrobials. Veterinarians delivered this program to producers. Such certification was never required and the program was discontinued after Ontario Pork took over its operation. In the initial five years, producer uptake of the program was good, but it is now available only on an ad hoc basis through OMAFRA swine experts (personal communication: Dr. Janet Alsop, OMAFRA). No other programs for producers, and no programs for nutritionists, were identified in Canada.
Considering that no feed-grade antimicrobials, all but one water antimicrobial, and over one-third of the injectable antimicrobials licensed for use in pigs and chickens in Canada are non-prescription drugs (263), the number of prudent use guidelines targeted at veterinarians versus producers seems unbalanced. Producers and animal nutritionists should have access to training on which practices elevate the risk of antimicrobial residues at slaughter, antimicrobial resistance in bacteria carried by food animals, and approaches to prudent antimicrobial use. Such training should emphasize the value of seeking veterinary advice for antimicrobial use decisions rather than attempt to train the producer or nutritionist in the optimal antimicrobial decision.
Therapeutic recommendations within AMU guidelines often overshadow the crucial point that prudent AMU demands that animal health be optimized through husbandry (much like the reduce component of reduce-reuse-recycle is often overlooked). Veterinarians can play an important role in designing health programs. In order to do this effectively, veterinarians must continually update their knowledge of disease prevention, therapeutics, and AMR trends to ensure the most appropriate use of antimicrobials (390). Effective health programs that include vaccination, biosecurity, good hygiene, and improved management should prevent disease and thereby reduce the need for antimicrobials. Veterinarians must also seek current continuing education to ensure their ability to design and implement evidence-based health programs (390). Unless Canada licenses all antimicrobials as prescription drugs, building and establishing trust in VCPR is the only insurance that producers will seek veterinary advice for all AMU, regardless of availability. Continuing to build such relationships may ultimately foster efforts to capture people's imagination that through AMU stewardship they can refuse (do not use), reduce, and potentially affect AMR development and dissemination (268).
The following sections describe examples of interventions that improve health and resultantly minimize AMU.

Alternatives to Antimicrobials
Vaccines Antimicrobial resistance is most likely to emerge and persist when AMU involves low doses, mass medication, and long exposure durations (268-271). Some antimicrobial classes that are critically important in human therapy are also used as feed additives to control disease in pigs and chickens.
The following examples relate to macrolide use to control ileitis (causative agent, Lawsonia intracellularis) in pigs and virginiamycin use to control bacterial enteritis (causative agent, Clostridium perfringens) in chickens (86,100,465,466). If expert opinion is correct and AMR can in part be mitigated by addressing longterm, low-dose antimicrobial exposures, reducing the use of these drugs should alleviate selective pressures for AMR in bacteria carried by pigs and chickens. Three recently developed vaccines allow nonantimicrobial control of these endemic conditions. These vaccines were selected as examples of advancements in animal health that positively affect public health.
Ileitis causes chronic diarrhea in pigs between 6 and 20 weeks of age. It is endemic in pigs worldwide, cannot be eradicated from herds, and until recently was extremely difficult to diagnosis. The combination of diagnostic limitations and financial losses has led to widespread AMU in feed for disease control. The most common prophylactic regimens include tiamulin, tylosin, chlortetracycline, or lincomycin fed continuously (100,465). In 2002, an avirulent live vaccine (Enterosol®, Boehringer Ingelheim Vetmedica, Inc.) was released on the Canadian market (467). This vaccine is administered through drinking water and significantly reduces disease prevalence and severity (465,468). For financial reasons, this product is used more extensively in grow-finish pigs in Europe and more commonly in breeding animals in Canada (468). The CIPARS sentinel herds in 2007 reported 76% of sow herds supplying sentinel grow-finish sites were positive for ileitis and 26% vaccinated. In the grow-finish herds, 66% reported a positive status and 17% reported vaccinating (469).
In broilers, bacterial enteritis is among the most prevalent and financially costly diseases. An overgrowth of C. perfringens causes disease symptoms ranging from wet litter to death. In feed, bacitracin or virginiamycin, combined with a coccidiostat, is used to control this syndrome and these products account for much of the AMU in broilers (165). The coccidiostat is necessary because coccidia damage the gut and predispose birds to more severe disease (466,466,470

Competitive Exclusion Strategies
Competitive exclusion strategies (CES) encompass a broad range of products that manipulate gut bacteria. As categories, these include competitive exclusion products, probiotics, and prebiotics. Most are oral products. Competitive exclusion (CE) and probiotics are live bacterial or yeast preparations that moderate the normal flora, while prebiotics are inert nondigestible additives that support beneficial bacteria. Competitive exclusion strategies are so named because CE or probiotics can be used synergistically with prebiotics (268,475,476). The objective of these products is to shift the gut ecology away from pathogens by displacing them with beneficial commensal bacteria (475).
Neonatal animals are born with an immature intestinal flora. Competitive exclusion products are live obligate and facultative anaerobic bacteria, typically originating from the intestinal contents of healthy adults. They are administered in the first days of life. This expedites establishment of the climax population; a stable intestinal microflora that is more resistant to invasion from pathogens (335,475). Competitive exclusion technology has particular application in chickens. While pigs are exposed to maternal flora, chickens are hatched in clean environments and raised separate from breeder flocks. Establishing the climax population takes six weeks, which is the majority of a modern broiler's life, and is predominantly influenced by the environment (477,478).
Probiotics are live bacterial cultures that differ from CE products because they are fed continuously (478). They can be monostrain, multistrain (of one species), or multispecies combinations and are primarily composed of gram-positive bacteria (479). Unlike competitive exclusion products, probiotics are not always derived from the animal species they are administered to (475). The mechanisms of probiotics are not completely understood and likely vary with different organisms. Factors such as organic acid production, bacteriocin production, and stimulation of the immune system have been demonstrated, with an impact on colonization or infection by pathogenic bacteria or parasites (480).
Undifferentiated CE and probiotic products are derived from intestinal content. These are essentially crude preparations of bacteria from healthy animals. The bacterial content is not identified or quantified. Early studies suggested these may be more efficacious than defined cultures, but this may simply reflect the infancy of this field's ability to identify optimal mixtures of beneficial bacteria (481) Undifferentiated but quality-controlled and pathogen-free products are available in Europe but prohibited in Canada and the United States (364,479,482). In North America, cultures must be composed of identified bacteria demonstrated to be beneficial (478). This limits the range of products available in North America.
There are at least two reasons for the regulatory restrictions on undifferentiated CE and probiotic products. Health Canada's Veterinary Drugs Directorate (VDD) uses the same registration process for pharmaceuticals and natural health products. Although this is being reviewed, it is difficult to register natural health products because many requirements are difficult to achieve with biological processes (483). Delays in licensing veterinary pharmaceuticals are a serious concern for producers and the pharmaceutical industry, and natural health products are only one area requiring immediate attention to maintain the Canadian livestock industry's competitiveness internationally (484). Yet, amendments to VDD regulation may not result in market approval for undifferentiated bacterial products. Undifferentiated cultures may pose an AMR risk if they possess resistance elements. These could be rapidly disseminated from a single source to many commercial flocks/herds (482). An experimental study with a porcine-derived undifferentiated CE culture found higher levels of tetracycline and streptomycin resistance in E. coli from exposed pigs compared to controls (485). Undifferentiated CE products are derived from mature animals, which generally carry resistant bacteria, so these results were not unexpected. Ideally, future preparations will distinguish between bacteria that are non-therapeutic and commonly carry resistance elements, such as E. coli, and therapeutic bacteria that rarely carry resistance genes, such as lactic acid bacteria. If achieved with minimal impact on efficacy, the risk of AMR from these products could be diminished.
The European Union requires antimicrobial resistance susceptibility testing of undifferentiated products (479). This approach is not foolproof. Traditional culture-based techniques cannot demonstrate freedom of resistance because resistance genes can be silent (carried but not expressed) and culture-based techniques test a small, and potentially unrepresentative, sample of the bacteria in the product. Molecular testing for resistance genes addresses these problems but will not detect novel resistance mechanisms. But together phenotype and genotype testing can provide some assurance that concerning resistance types are not being widely disseminated through undifferentiated cultures.
Studies of CE and probiotics products have had wide-ranging results, both beneficial and detrimental.
Although this promising research field may some day help address the pharmaceutical void created by removing in-feed antimicrobials, advancements in basic knowledge along with improved study designs and more precise outcomes are needed before inconsistent findings are overcome. Serious concerns exist over the use of enterococci in CE products due to their propensity to carry resistance genes. As discussed above, other bacterial species may be more promising. A cautious regulatory approach should help to ensure that these products ameliorate rather than contribute to agri-food AMR.
Prebiotics are a dietary supplement that are available to microbes and provide limiting nutrients to the intestinal mucosa but provide no direct nutritional benefit to the animal host (335,475,486). Prebiotics are commonly used in people, while uptake in agriculture has been limited due to expense (475).
Research from humans and animals suggest various prebiotics can modulate the immune system, neutralize pathogenic enterotoxins, and bind E. coli and Salmonella (471,475,487). Prebiotics are not restricted to antimicrobial-free production because the effects complement in-feed antimicrobials (488,489). Therefore, these can be considered an antimicrobial alternative or simply an advancement in human and animal intestinal management.
Studies of CES have shown variable results on animal productivity and health (489,490), which along with expense, has limited commercial uptake. Early studies sought to improve growth and did not evaluate the microbial ecosystem. But Europe's experience following in-feed antimicrobial bans made it clear that in-feed antimicrobials concurrently suppressed disease while improving productivity (491-494). Thus, recent efforts have focused on more specific objectives such as manipulating the microbial population to optimize niche dominance of beneficial bacteria and exclude pathogens. Neonatal animals with naïve flora are now commonly studied to minimize uncontrolled confounding and decrease variability. Beyond study design, disappointing results with live bacterial products have been attributed to challenges in maintaining an exogenous population of bacteria without continuous administration. This may be difficult to overcome as the balance of intestinal microflora appears to not only be species and gut location specific but to have important variability between individuals (454,475,478,481,495).
Food safety may be improved by CES if these prevent the colonization of animals with foodborne pathogens. A variety of studies in pigs and chickens have found these products can affect colonization or shedding of Salmonella and Campylobacter, but results are preliminary, variable, and are preparation and pathogen specific. (496,497). Commercial application will require continued research. However, this research field holds great potential because these products can concurrently address animal and human health challenges.

Bacteriophages
Another area that may have promise in the future is the use of bacteriophages, viruses that infect bacterial cells. There are two forms of bacteriophage, temperate and lytic. Temperate bacteriophages infect, but do not damage, bacterial cells and are a potential mechanism for transmission of virulence and antimicrobial resistance genes. Accordingly, they are not useful therapeutically. Lytic bacteriophages, in contrast, kill infected cells. Some bacteriophages have broad-spectrum activity while others only infect a narrow range of bacterial species, or even strains within a species. In addition to the potential for efficacy, lytic bacteriophages are appealing because of their safety. Bacteriophages are unable to infect mammalian cells and therefore are non-pathogenic to animals. This field is relatively young and ongoing safety evaluations are needed. However, temperate bacteriophages are classified as 'generally regarded as safe' (GRAS) by the FDA, are classified as Biosafety Level 1 microorganisms (as with Lactobacillus spp.) and have fewer regulatory hurdles to clear, and have minimal to no evidence of ability to convert to temperate forms. The potential efficacy of lytic bacteriophages is of great interest and has been discussed for decades, however objective data has been lacking until recently.
Bacteriophages could be used for the elimination of bacteria in or on animals, or in the environment.
Intensive research on bacteriophages in swine and poultry has been fairly recent, but preliminary data are promising. Bacteriophages with in vitro efficacy against swine pathogens such as ETEC, Salmonella, and Streptococcus suis have been identified (498-502). In vivo efficacy has been reported as well, with certain bacteriophages being effective at moderating the course of experimentally-induced ETEC (E. coli F4) diarrhea (503). Bacteriophages with in vitro efficacy against important pathogens in poultry such as Salmonella and E. coli have been identified (504,505). Bacteriophage therapy has been shown to reduce Salmonella and Campylobacter shedding in broiler chickens (506-509). Research is still needed on the rate and extent that bacteria will develop resistance to lytic phages.
While not acting directly against AMR, bacteriophages could have an indirect impact by reducing disease and concurrent need for antimicrobials. Further, bacteriophages could be used against multidrugresistant pathogens such as MRSA. This is a field that is receiving increasing attention and might constitute a useful adjunctive therapeutic option in future years.

best Management Practices
There is no 'magic bullet' that eliminates the need for AMU in food-animal production. European Union producers identified a return to 'good management' as the most effective intervention following antimicrobial growth promoter bans. Likewise, the North American experience in antimicrobial-free production emphasizes barn hygiene, stocking density, ventilation, and for chickens lighting, and litter management. Particular attention must be paid when animals are immunologically compromised.
In broilers, this occurs during brooding, and in pigs, post-weaning (100,471,478,510

Nutrition and Feed
Animal nutrition is constantly advancing to optimize genetic potential and protect health. In both chickens and pigs, advancements in high quality starter diets have eased the transition to solid food. Anorexic animals can experience gastric stasis which places them at increased risk for pathogen colonization (100,101,476). Ensuring chicks and piglets get onto feed as soon as possible after brooding/weaning supports the gastrointestinal flora through this challenging period for the immune system.
In the late 1990s, the North American swine industry initiated a management system known as segregated early weaning. Pigs were weaned as young as fourteen days and moved to ultra-clean off-site nurseries. Separating piglets and sows was effective in containing many respiratory pathogens to the breeding herd, but it created nutritional challenges for feeding immature pigs. The transition from milk to a plant-based diet in young animals is facilitated by high quality animal protein, which in North America is typically derived from milk or blood (spray-dried plasma) products. Using quality ingredients, palatable diet forms (i.e. crumble), managing feeders and waterers, and occasionally offering creep feed before weaning have all worked to get weaned pigs onto feed quickly. By eating and minimizing gastric stasis, a healthy gut flora is supported and enteric diseases are prevented. The standard industry weaning age has increased since the 1990s, but these advances continue to influence nutrition in the nursery (100). This is significant since nursery production uses more antimicrobials than the other production phases. (92,93,282,409).
Poultry nutritionists have modified diets to minimize the risk of bacterial enteritis (337,366,511). This was stimulated by European experience following the antimicrobial growth promoter bans when wheatbased diets fuelled necrotic enteritis outbreaks (493). Along with adjustments to the carbohydrate source, feed grind is now balanced for health (larger) and productivity (smaller), and whole grains are occasionally included to stimulate gizzard development. Other advancements have included increased use of synthetic amino acids, a focus on fat quality, and inclusion of enzymes to facilitate digestion of non-starch polysaccharides. Canadian nutritionists employ these techniques resulting in an industry with decreased risk of enteric disease.
Nutritional advances are relevant to AMR because they support health and decrease reliance on AMU. These advances also support the healthy intestinal flora, and like CES, help preclude colonization with enteric pathogens. Finally, this section has relevance for addressing endemic resistance problems by emphasizing the benefits that can arise when new technologies are applied to old husbandry questions.

Biosecurity
Biosecurity is the collective measures taken to ensure security from exposure to harmful biological agents. Thus biosecurity is both the plan and the actions to implement that plan. Biosecurity is a risk management tool, much like insurance. In the case of on-farm biosecurity, producers incur costs up-front to minimize future risks. Many biosecurity protocols also apply to bio-containment, which in this context is the control, eradication, or prevention, of new incurrence of infectious agents.
Biosecurity can be applied to farms, regions, or nations. Protocols are implemented to control pathogens with sufficient consequences or probability of incursion (i.e. risk) to justify the expense and effort. These protocols address animal, human, fomite, and vector borne risks. Sanitation is not truly biosecurity, but is often included because it is a barrier to disease between groups of animals in a flock/herd. In Canada, on-farm biosecurity protocols vary from non existent to extremely stringent (476,512-514).
Control of animal movement is essential for disease control. Many swine barns and all broiler operations are closed to live animals after they fill. Many swine and broiler operations prefer to receive animals from a single source. Within barns, an animal movement pattern known as 'all-in-all-out' is commonly used. This management approach batches animals by age, moves them as a group, and prevents contact between batches. When combined with sanitation, this helps to prevent aerosol and fecal-oral disease transmission between groups of animals (100). All-inall-out is inherently more effective in poultry than pigs because hatcheries supply chicks that have never had physical contact with the breeder herd. Additionally, the entire poultry barn is emptied between groups, while swine barns often employ all-in-all-out by room within a barn. All-in-all-out is less effective without sanitation. Although it prevents the spread of disease between animals through direct contact, barns should be emptied of litter/manure, washed with detergent, disinfected, and allowed a dry down time between groups for optimal efficacy (515). Roughly threequarters of American swine producers use all-in-allout management and the uptake of this practice has increased from approximately half of producers in 2000 (409).
In theory, preventing bird/pig contact between groups could help to control AMR. A study investigating AMR in Campylobacter found pig herds with allin-all-out management had lower odds of multiple drug resistance than herds with partial all-in-all-out or continuous flow management (516). Although western Canadian field studies examining E. coli and Campylobacter from grow-finish pigs failed to find a relationship between pig flow and AMR, this could be attributed to insufficient power. Conversely, a lack of relationship might exist in the grow-finish production phase because pigs enter the phase with high resistance levels (84,299).
In broilers, fluoroquinolone resistance in E. coli has been traced to vertical transmission from the breeder flock, most likely through hatchery contamination and re-infection (517). (Note: vertical disease transmission is the spread of disease from one generation to its progeny, typically from mother to offspring. It is used to differentiate from disease transmission within a group of contemporary animals which is termed horizontal transmission.) Hence, transmission dynamics are important to the question of AMR control through animal segregation. Demonstrating common transmission from breeder flocks to chicks (via the hatchery) or from sows to suckling pigs before weaning would mean that animal flow does not break resistance and an industry-wide control effort is needed. Alternatively, rare vertical transmission would provide an opportunity for individual operations to address resistance levels. These dynamics are likely to be bacterial speciesspecific and are likely to differ between pigs and chickens. In pigs, we would expect that industry-wide programs will be needed because piglets are exposed to the sow's feces.
Pigs and chickens can acquire disease from humans. To address this risk, visitors may be required to have no chicken/pig contact for a certain duration before entering the barn (i.e. swine barns often require 48-72 pig-free hours prior to entry). The entry requirements for people are more stringent in breeding flocks/herds than grow-out operations because of the continuous presence of valuable animals. Upon entry, visitors may be required to shower and/or change clothing. This minimizes the risk of people acting as a fomite. An experimental study demonstrated that enterotoxigenic E. coli can be transmitted between groups of pigs by handlers despite hand washing and outer-clothing change between groups. Transmission did not occur when handlers showered between groups (518). Downtime and showering between farms is important, as a British study found flocks using 'thinning' (split marketing heavy birds and leaving lighter birds to continue growing) are at increased risk of Campylobacter infection. The risk is attributed to the catchers who move from flock to flock (519).
People can act as a source of infectious disease to birds/pigs, as presumably occurred in the recent H1N1 (pandemic strain) infected swine farm in Alberta (520). No studies were identified that investigated the rate of transmission of resistant bacteria from healthy people to chickens/pigs under normal contact conditions. This is another area of transmission that needs study before AMR control programs can be applied on-farm. Drugs critically important for human medicine are used most commonly in people and thus novel resistance is most likely to emerge in people. If people subsequently transmit these resistance elements into otherwise isolated livestock or poultry operations, and amplification, re-assortment, and re-transmission to humans occurs, it exponentially increases the reservoir of resistance. This concern mirrors those over the role of pigs and poultry in influenza and has been demonstrated to occur with methicillin-resistant Staphylococcus aureus (MRSA) in horses and companion animals (184). 'Humanosis' control is an understudied area of AMR control. This area of study could be expanded to evaluate the risks of resistant bacteria derived from humans entering barns through alternative routes, such as water contaminated by urban effluent (161).
Biosecurity can be an effective tool to keep herds/ flocks negative for diseases that are enzootic for the industry. A study of broiler farms found the odds of being Campylobacter-positive at slaughter were nine times higher in control flocks than in flocks implementing a biosecurity program. Control flocks were also infected earlier than the 'biosecure' flocks that broke with disease (521). More research quantifying the effect of biosecurity practices or systems on animal and zoonotic pathogens is desperately needed. Such work could be expanded to evaluating associations between biosecurity practices and AMR, as has been done to a degree in studies that evaluated management practices as risk factors for AMR in bacteria from healthy pigs (299,516).
The 'art' of biosecurity has traditionally been empiric with minimal scientific rigor applied to recommendations (514). Many practices have not been proven to improve productivity or health, and despite increased uptake of biosecurity practices in the United States, pig health worsened from 2000 to 2006 (409). While this may reflect changes in endemic disease or management factors other than biosecurity, it reminds us that biosecurity and husbandry practices require rigorous scientific scrutiny to ensure they actually are an improvement.
Biosecurity decisions are made on an individual flock/ herd basis, and producers implement systems based on the perceived costs and benefits (522,523).
A variable system places the Canadian livestock industry at risk for foreign animal disease. A review of biosecurity practices in broilers was conducted after the avian influenza outbreaks in British Columbia in 2004. Lapses in biosecurity were deemed to contribute to disease spread between farms. Data do not exist to determine if the practices in British Columbia are representative of those across Canada (524). Addressing the reasons for breaches in biosecurity, along with the reasons for differences in biosecurity between flocks, could be relevant for understanding what changes are needed in biosecurity before programs could consider attempting to prevent novel resistance elements from becoming established in swine and poultry populations.
The CIPARS farm swine surveillance collects data describing biosecurity practices on sentinel farms (66). To date, these have not been published and are, to our knowledge, the only national data describing biosecurity practices in the Canadian swine industry. Biosecurity in American swine herds has been described but should not be extrapolated to Canada as half of the herds reported outdoor exposure of sows (409). Biosecurity requirements for indoor closed herds differ from free ranging pigs. In Canada, maintaining sows outdoors would be a relatively uncommon practice. Outdoor rearing of sows would impact sow exposures and disease, thus making any comparisons virtually impossible.
The swine and broiler industries are embarking on a new chapter of on-farm programs by developing on national biosecurity standards. These efforts are at the conceptual stage but will provide an outlet for scientific, Canadian-based research. This is a valuable opportunity to investigate potential controls for AMR dynamics and transmission. The resultant findings could be incorporated into these national standards. Without a concerted effort from researchers and policy makers, there is a risk that antimicrobial resistance issues being delegated to one of the many public-good issues will provide insufficient incentive for producers to act voluntarily. If this is the case, spin-off benefits will continue to be the main source of AMR control in Canadian livestock operations.

slaughter and Processing
It is widely held that the best way to mitigate foodborne bacterial risk is to reduce the prevalence of pathogens in animals pre-harvest (7,455). Yet, post-harvest interventions have been responsible for most of the advancements in food-safety from bacterial disease (321). A growing burden could be placed on the slaughter and processing industries if producers change AMU practices. Decreased in-feed AMU could result in a subsequent decline in carcass uniformity, an increase in sub-clinical disease (and therefore more friable intestines), and an increase in adhesions from past clinical disease. An increase in these carcass and intestinal characteristics would lead to an increased risk of intestinal perforations and fecal contamination (325,326,(525)(526)(527), and therefore, a reduction in food safety.  456,528,529). A summary of the advancements in poultry processing is provided by Bolder et al. (Table 5) (455). Since these interventions control the bacterial host contamination, they indirectly affect resistance elements on meat and control AMR dissemination. Further innovation to directly address the risk of contamination of carcasses or meat with AMR bacteria currently lacks obvious or simple controls. If the reduction of foodborne AMR bacteria becomes a goal, basic scientific research will be needed to address this challenge. Only three aspects of slaughter and meat processing were found that specifically pertained to food safety risks from AMR. Interestingly, all were concerns raised in Europe, and all evaluated the potential for introducing risk rather than risk mitigation.
Microbial control agents (also referred to as antimicrobial treatment substances in Europe) are licensed in Canada, the United States, and Australia. These include hydrogen peroxide, peroxyacetic acid (POAA), octanoic acid, peroxyoctanoic acid (POOA), and 1-hyroxyethylidene-1, 1-diphosphonic (HEDP) acid (528,530). In these countries, microbial control agents are registered for use on fruits, vegetables, meat, and poultry. Most microbial activity is attributed to POAA with reports of 2-to 9-fold reductions in total microbial loads and activity against Listeria monocytogenes, E. coli O157 and Salmonella (530). The safety of these compounds was reviewed by the FAO/WHO Expert Joint Committee of Food Additives, which ruled that the residual quantities of these acids on foods at the time of consumption posed no food safety risk (531). In a 2005 report, the European Food Safety Commission echoed this finding (532), but further assessed the issue in 2007, when potential effects on antimicrobial resistance were considered (533). Resistance was defined as both a decreased effectiveness for reducing microbial load and resistance to therapeutic antimicrobials. The panel concluded that there was no evidence to indicate that use will lead to either type of resistance but did recommend research on the likelihood of emergence of resistance (533).
The second and third concerns regarding processing interventions leading to AMR were summarized by the European Food Safety Authority (106). One concern was that laboratory studies showed that industrial non-thermal processing and preservation techniques can damage cell membranes, enzymes, and DNA. This could "promote the generation or transfer of resistance" (106). The other concern was that bacteria intentionally added to the food chain to assist in preservation, or fermentation could serve as a source of resistance elements. Much like the concern over commensal bacteria from animals, these bacteria could theoretically pass resistance elements on to commensal or pathogenic bacteria in humans. Although these concerns are theoretical, they deserve investigation to determine the likelihood and repercussions of such events.

Conclusion
This chapter summarized the industry-wide and producer-level activities that reduce the need for antimicrobials in chickens and pigs. From this summary, it is evident that the effect of individual interventions on AMU is unmeasured and a cohesive approach is lacking. The two notable gaps in activity were the lack of education programs on appropriate AMU for producers and animal nutritionists and a lack of publications related to AMR control at slaughter and processing. In contrast, the advancements in animal husbandry, management, and health that have occurred concurrently with the intensification of our livestock production have helped to mitigate the reliance on antimicrobials that is inherent with large confinement operations.
There are two impediments that we believe would speed industry-led action to control AMR if they were removed. The first is a lack of scientific evidence regarding the efficacy of interventions. Most veterinarians and producers recognize that decreasing AMU is the goal, but the best approach remains elusive because rigorously-tested, evidence-based research has largely not occurred. A cohesive industryled approach requires a strategy based on science.
Appropriately designed observational studies must be able to attribute AMR changes to an intervention, measure confounding effects, evaluate challenges and costs that may lead to non-compliance, and ensure that the AMU alternatives or therapeutic drug use shift in response to the intervention are beneficial. This is a massive undertaking especially since AMR ecology is so complex and to date not completely understood. However, it is absolutely necessary to begin to take these steps in order to scientifically address this problem while maintaining a viable agriculture industry.
Economics are the second impediment to industryled AMR interventions. Producers are financially rewarded for healthy, uniform animals. Antimicrobials currently help to achieve that. Currently, beyond the niche market for antibiotic-free pork/chicken, there are no tangible incentives to offset production risk and losses from interventions to control AMR. At present there are also many industry challenges, such as market access, foreign animal diseases, and extensive financial losses, especially in the pork industry, that are directly jeopardizing producers' viability. It is very difficult to inspire producers to address AMR issues when their livelihood is threatened. Meat production is how these individuals make their living. Hence, they must see a demonstrable benefit from making changes and implementing new practices, especially to address an issue that is not negatively impacting their production. This would involve producer and industry education, but it would also need to include a component that would help the livestock industry to mitigate risks to animal health, welfare, product quality, and return on investment for making changes to address AMR. Such action is required as incentive to increase the priority of this issue. Widespread industry change will also need innovative ways to ensure cost-sharing between society and industry.

AMR Bacteria in Agri-food
The emergence, dissemination, and maintenance of antimicrobial resistance (AMR) are complex processes that cannot be explained or addressed using simplistic methods. Factors such as co-selection of resistance genes, interactions between bacterial populations, and variable changes in bacterial fitness after resistance acquisition are but a few complex issues that require equally complex investigations and interventions. Research regarding AMR must be adequately broad to be able to identify true causal relationships. This knowledge can expedite implementing effective interventions and minimize trial and error in policy changes.
Innumerable connections link antimicrobial resistant bacteria in humans, food-producing animals, and the environment, and believing that we understand all of the possible links would be presumptuous. The transmission rates for these links are unknown, but food is a universal and direct connection between people and animals. It is irrefutable that people can acquire antimicrobial resistant bacteria from animals through food. What remains disputed is the frequency that pathogenic and commensal bacteria are transmitted to humans and either cause disease or transfer resistance elements to bacteria in people. A related knowledge gap is the relative amounts that AMU in animals and humans each contribute to AMR in humans. Answers to these questions would allow policy makers to evaluate the magnitude of risk imposed on the population by contaminated food and would allow appropriately aggressive intervention strategies.
Resistant pathogenic bacteria are the most obvious food safety concern. Within these bacteria, new threats are constantly emerging, such as the novel plasmid-mediated fluoroquinolone resistance genes. It is likely, and perhaps even inevitable, that fluoroquinolone resistance genes will become established in commensal and pathogenic bacteria in swine and poultry (74,107). Equally concerning is the potential for fluoroquinolone resistance to be encoded on mobile genetic elements together with other critically important resistance determinants such as extended-spectrum ß-lactamase genes (112,113). If fluoroquinolone resistance is linked to other resistance genes, then the use of any of the linked antimicrobials could increase fluoroquinolone resistance. This emerging issue emphasizes the need for research into transmission dynamics and possible interventions that could prevent human AMR problems from becoming established in food-animal populations.
While the potential role of the commensal gut microflora is frequently discussed, investigation of this diverse and poorly understood bacterial population is quite superficial. A large percentage, if not the majority of bacteria present in the intestinal tracts of livestock, have likely never been cultured and identified. So, current assessments of the commensal microflora are presumably being based on a small, and not necessarily representative, component of the bacterial population. Considering that this bacterial population may be of greater relevance to AMR than pathogenic bacteria, greater study, in quantity and depth, of the commensal microflora is required. This should include broader study of the true nature of the microflora, the effects of antimicrobial use (AMU) on population dynamics and on AMR, and the interaction between the commensal microflora and pathogens. Furthermore, more research is needed on environmental bacteria to investigate connections between AMR in soil bacteria and human commensal and pathogenic bacteria (534). Our limited understanding of commensal bacterial ecology is illustrated by the much studied relationship between E. faecium in livestock and people. This commensal bacterium is ubiquitous in humans, swine, and poultry. But determining the extent to which these bacterial populations are host adapted, interact, and share resistance elements has challenged the limits of science (156,(159)(160)(161). Molecular epidemiology is constantly responding to this and similar challenges through new tools. Ultimately, we hope this will enable scientists to quantify these relationships, which would allow interventions to focus on the greatest risks to human health.
The public health community has seen resistant nosocomial pathogens expand their niche into communities. Concurrently, health and veterinary authorities are acknowledging the porousity of the human/animal boundary to many bacteria. Controlling the 'infectious traffic' between nosocomial, community and animal bacterial populations will be an increasingly important task for AMR control. Future challenges are highly unpredictable, as already demonstrated by the potential for methicillin-resistant S. aureus (MRSA) and C. difficile to be zoonotic. The identification of MRSA in food animals and food has raised concerns about the risk of foodborne human MRSA infection (202)(203)(204)(205)(206)(207). Basic food handling and cooking practices should greatly reduce the risk of food acting as a mode of transmission, but the risks cannot (and should not) be assumed to be negligible in the absence of evidence. More intensive research in this field is required, particularly characterization of strains found in pigs in Canada, mechanisms of foodborne contamination, and epidemiological investigation of food contact as a risk factor for community-associated MRSA infection in people. Clostridium difficile is also a high profile pathogen with recent concerns about foodborne transmission (225,241,242). The role of food in human C. difficile infection is unclear and is coupled with a rather superficial understanding of the role of C. difficile in community-onset diarrhea in humans. Study of the role of C. difficile in community-onset diarrhea and food as a risk factor for community-associated C. difficile infection is needed.
Despite the abundance of data regarding AMR and livestock, some significant deficiencies exist. While large volumes of data are present, there is tremendous repetition of the same methodologies in the same regions or between regions. This can provide information about inter or intra-region differences or changes, but it can also result in generation of data that contribute little to the overall understanding of AMR and AMU. While continuation of such studies is relatively easy, expansion of research to encompass a broader understanding of AMU and AMR is required. If this field is to truly develop, new methodologies and populations need to be used, and better collaboration between the fields of epidemiology, microbiology, molecular biology, ecology, food safety, and animal management are required. This type of integrated study can be technically challenging and expensive, but has great potential to generate true and accurate understanding of causal determinants of AMR.
Funding agencies need to be open to supporting broader and novel methods as, in the long run, these may be a more efficient use of resources.
Active research is exponentially increasing the information available about AMR in food animals. Each study adds a piece to the puzzle. But the sheer volume of scientific publications is becoming unwieldy and is presenting challenges for concise and current knowledge translation to researchers and practitioners who need to apply these findings. We face the risk of data overload, where policy and practice is unable to assimilate new advancements and contextualize novel discoveries. The rapid expansion of information, often with conflicting results, means literature reviews quickly become obsolete and are prone to biased interpretations. In the health sciences, meta-analyses and systematic reviews are used to compile multiple research findings. These techniques can evaluate specific and extensively-studied questions, but are insufficient for understanding and communicating knowledge on many issues because the topic and concerns with AMR are broad, and understanding of diverse but inter-related aspects is necessary. Yet, we believe that analogous approaches must be developed through multidisciplinary teams involving information technology specialists, librarians, and scientists to assimilate the vast scope of AMR knowledge and reliably evaluate conflicting findings. Without such tools, information paralysis may undermine the value of current and future research. In a related vein, decision analyses have been used to model health decisions at individual and population levels. We suggest it would be useful to investigate whether similar techniques could assist in incorporating new information into intervention or policy decisions.

Antimicrobial use in Food Animals
Antimicrobials are used in food-animal production around the world. The food safety hazards arising from this vary based on the types and amounts of antimicrobials used. But despite the concerns about AMU on farms and the interventions that are recommended or implemented, our understanding of AMU practices remains limited. This is problematic because assumptions and weak data are being used to determine policy, strategy, and even legislation. Antimicrobial volumes collected through sales data may be reliable, but are of limited value by themselves for understanding AMR. Many products are licensed in numerous species for multiple indications and may also be used in an extra-label manner. Antimicrobial volumes cannot reflect selective pressure without accounting for the recipient population (535). In contrast, end-user data hold great potential for identifying risk factors for AMR but are more prone to bias and are more costly to collect. We have seen global progress in standardizing antimicrobial susceptibility testing and surveillance techniques (281,408). Ideally, advancements in AMU data collection will soon follow. This will allow for valid analysis of the selective pressures for resistance and will bolster the validity of ecological level studies.
In the meantime, it is widely stated that antimicrobials are necessary for animal health and welfare. We agree that AMU is vital for humane animal production but AMU can also compensate for sub-optimal management. The true role of antimicrobials in animal health and welfare needs to be objectively quantified. The AMU that is unnecessary or unreasonable must be eliminated while preserving use that is required to produce safe food, humanely rear livestock, and ensure economic viability of the industry. Research should broadly consider animal health, economics, and AMR. Answering these questions could be facilitated by evaluating differences between conventional, antibiotic-free, and organic production systems.
It was traditionally believed that ceasing AMU would inevitably and promptly lead to reductions in AMR. But data supporting an increase in AMR with increasing AMU are stronger and more consistent than studies showing a decrease in AMR following withdrawal of AMU. Declining resistance after drug cessation has been demonstrated for some pathogen/drug combinations, but it is certainly not universal and has often been less substantial and sustained than was expected. This is important because it suggests that established AMR is not consistently resolved by removing the inciting cause. Our ability to contain or decrease specific AMR trends may be limited. Many factors, including co-selection, bacterial fitness, and linkages with virulence and other important genes, means that a direct relationship between AMU and AMR is not always present. Therefore, efforts focused simply on one organism or one antimicrobial could be inadequate. Research demonstrating the effectiveness of a variety of interventions to mitigate existing AMR, only one of which should be drug withdrawal, will be necessary for future policy. This should include interventions through sanitation, animal movement, and husbandry.
The apparent relationship between in ovo ceftiofur use and resistance in human Salmonella isolates illustrates how agricultural AMU can affect humans. The correlation between decreasing AMR in Salmonella and E. coli from chicken following reports of voluntary cessation of the practice and return of resistance after use was reportedly reinstituted are one of the clearest demonstrations of the link between AMU and AMR (117). However, these surveillance findings need additional research to clarify the total effect of AMR/AMU interventions on human health. For example, research is needed on human health repercussions from the rise in ESBL resistant E. coli in chicken that paralleled the rise in Salmonella Heidelberg (Chapter 2; Figure 2). Food safety research is also needed, such as comparing pathogen loads on the carcasses derived from treated and untreated eggs, and comparing the health and subsequent AMU in chicks that did and did not receive ceftiofur in ovo. Surveillance should be used as a time sensitive tool to instigate research activities that expand beyond the farm-to-fork continuum to measure a wide range of effects on human health.
Quantitative risk assessments have examined the magnitude of risk imposed on people from foodborne resistant bacteria. Several American publications estimated substantially lower risks than were intuitive based on a qualitative evaluation of the literature (8,85,526). These models can provide a tool to investigate the relative importance of transmission pathways, but are limited in scope to known and studied connections. Therefore, our confidence in the input parameters determines their validity. To date, quantitative risk assessments have been confined to single pathogen/antimicrobial combinations. Perhaps in the future it will be possible to expand the scope of risk assessments to account for a wide variety of repercussions including environmental contamination from AMU, environmental pollution from lack of AMU, animal health and carcass characteristics, pathogen loads on meat, and nutritional repercussion from changing meat prices. This would be a colossal undertaking but would help ensure policy results in the greatest benefits for society.

Legislation, Regulation, and Policy to Address Antimicrobial Resistance in Agri-Food
The WHO, OIE, and FAO have led international co-operation to address foodborne resistant bacteria. These agencies have released prudent use guidelines and collaborated on risk assessment techniques. The effects of these actions on AMU and AMR are difficult to judge because countries that have implemented regulations and policies may have done so in the absence of these mandates, while others may be preparing to act but appear unresponsive. Some of the countries that have not developed formalized systems to address AMR have either piloted surveillance programs or conducted research into AMR in foodborne bacteria. This indicates that concern is widespread.
National differences in veterinary drug regulations and AMU/AMR monitoring raise concerns about the international movement of animals and food.
Importing nations are concerned about the safety of imported products and exporting nations are concerned about open and transparent trade.
International standards, such as the draft methods released by Codex to conduct risk assessment, can help to smooth trade but are not enforceable regulations (536-538). Thus, member countries can still set their own requirements. For example, risk prioritization relies on the ranking of antimicrobials for their importance to human medicine. But countries may rank antimicrobials using a list developed through international consensus or one reflecting the concerns of the member country (86,388). Trade barriers to control animal or foodborne resistant bacteria have not yet occurred but are a realistic concern given the increasing divergence in regulations and controls between countries. Until such a barrier is invoked, challenged, and ruled on by the World Trade Organization, the ability of countries to block importation of products based on antimicrobial use or resistance remains speculative.
Canada has played an active international role in AMR policy. We applaud the efforts to support evidencebased regulations. It is important that Canada continue to push other nations to address high-risk drug use practices because the problem of AMR is truly global. Movement of animals, people, and food means that Canadians are affected by AMR in distant countries. The Canadian government also has a responsibility to the agriculture sector to hold our competitors to similar standards as are imposed at home. Much remains to be achieved on this front as many competitors have permissive veterinary drug regulations and scant to non-existent reporting. The reverse of this statement is also true. Canada has more permissive veterinary drug regulations than Europe or the United States. If we expect to hold other countries accountable to our standards, we will be expected to respond to the concerns raised over our inability to ensure prudent antimicrobial use (248,385).
Canada is one of several countries that has established an agri-food AMR surveillance system. Amongst these, Canada stands out as one of the most comprehensive. Unlike many other programs, Canada monitors three steps of the production chain-the farm, abattoir, and retail-and annually collects samples for a range of bacteria in every major food commodity (66). Advancements in collecting AMU data from pigs, with parallel initiatives in other commodities, have made this a robust and comprehensive program. From the swine component, positive news has emerged including the absence of fluoroquinolone use and close to 10% of sentinel farms voluntarily using no antimicrobials. Concerning practices have also been identified including ceftiofur use and long-term macrolide use. The swine and broiler industries have recognized the value of third-party reporting to demonstrate transparency. Ideally, this system should enhance the effectiveness of policy addressing AMU in Canada.
Collecting multi-source AMR data allow 'farm-to-fork' risks to be studied. We recommend continued strengthening of the CIPARS program design to draw causal links from AMU in animals to AMR in animals to AMR in humans. Surveillance programs are most effective when the data collected are consistent over time. Canada has developed a comprehensive program, but due to budgetary limitations, there are some concerns about long-term sustainability of the farm component in its current form. Continued research into alternative AMU data collection methods, such as surveys and data extrapolation, may identify valid and cost-effective alternatives. Preparing for the worst is an unfortunate reality of shifting budget priorities. However, we advocate continued support for Canada's CIPARS program and its expansion to other provinces and commodities. The Canadian regulatory system is committed to evidence-based policy. This has led to risk assessments being applied to new drug applications and potentially to future post-approval evaluations. Canada has been criticized for its veterinary drug regulations by national and international experts. Clearly, the VDD acknowledges these criticisms and is striving to address them as shown by the extensive stakeholder consultations that have occurred since 2002 (285,364,391,394). From an outside perspective, it appears that stakeholder paralysis may be an issue. This will be challenging for the VDD to address, as science will not be able to provide definitive solutions to many of the current concerns before regulatory changes are needed. We recommend the VDD commit to evaluating policy for efficacy with established targets to ensure it is working. In the case of ineffectiveness or (even worse) harm, the VDD must be committed to revoking or changing such policy. We urge the VDD to work in conjunction with the appropriate provincial regulators to address the thoroughly investigated recommendations of the 2002 Advisory Committee on Animal Uses of Antimicrobials and Impact on Resistance in Human Health before they are obsolete (248).

Agri-food Industry Actions to Address AMR
The broiler, swine, veterinary, and meat industries continue to make advancements that improve animal health and decrease their reliance on AMU. Excluding the prudent AMU guidelines for veterinarians, these actions are not undertaken specifically to address AMU and so their effects on AMR are presumed but unconfirmed. Viable on-farm food safety (OFFS) programs in the chicken and pork sectors are consistently implemented across Canada (284,458). When sufficient knowledge pertaining to on-farm AMR is available, these could provide a foundation for creating and delivering an AMR control program. Beyond these, we have limited knowledge of how extensively the other activities are used across the industry. Prudent AMU guidelines were the only identified activity that specifically addresses AMR. These guidelines tend to be broadbased and provide little clear information about how to incorporate prudent use into current practices (395). Of the numerous guidelines that exist, we found none that were monitoring their influence on AMU. This information void is unfortunate because, without evidence of their effect, it is difficult to sustain the necessary funding to promote, update, and modify the standards in order to keep them viable. Also it is impossible to identify the ineffective areas for improvement. The Canadian Veterinary Medical Association (CVMA) generated the most recent guidelines for Canadian swine and poultry veterinarians (390). We urge the CVMA to survey its membership on AMU practices and attitudes immediately to provide a baseline for the recently released guidelines, with follow-up surveys on the use of these guidelines. Compared to the efforts that have gone into the creation of this document, this seems like a small and invaluable undertaking. In conjunction, we support a collaborative relationship between the CVMA, the Chicken Farmers of Canada (CFC), the Canadian Pork Council (CPC), and the Animal Nutrition Association of Canada (ANAC) to develop and deliver prudent use education for producers and animal nutritionists. These groups are the greatest users of non-prescription antimicrobials, thus this may counterbalance Canada's controversial regulations. Data describing how and why producers make AMU decisions exist but have not been utilized. This could strengthen the creation of prudent use programs for producers and shape effective policy. Thus, our final recommendation to the CFC and CPC is to analyze and disseminate the valuable information they possess describing AMU attitudes, OFFS program sustainability, and the implementation of best management practices.
Vaccines, competitive exclusion strategies, nutritional advancements, and biosecurity were presented as examples of industry-led advancements to decrease reliance on antimicrobials. Research to develop new products and to identify new biosecurity techniques will continue to improve animal health. Indirect support for prudent AMU must not be undervalued. Funding agencies should continue to invest in healthbuilding research. Data on the extent that these alternative management practices are implemented nationally, and their alleviation of AMU dependence, could be studied.
Empirically, it appears that the implementation of interventions by industry correlates with financial reward and/or risk mitigation. Perhaps Canada's public health and agriculture sectors can learn from two recent American experiences. A positive example was set when the United States' Waxman-Markey clean energy bill was strongly supported by corporate America because it fostered innovation in the corporate sector without stifling financial gain from innovation and early adoption (539). In contrast, a negative example has been set by the repeated introduction of bills that threaten to revoke antimicrobials, which has fostered distrust, fear, and non-cooperation between corporate agriculture and health activists. Canada should strive for an innovative and collaborative relationship between the medical, veterinary, and agriculture communities. If public health authorities explored the risks of inaction in conjunction with the agriculture industry, and funding agencies concurrently provided resources for applied innovations, while regulators minimized impediments to novel alternatives, we could collectively enable the agriculture industry to drive, rather than drag, solutions for AMU and AMR.

Main Recommendations and Knowledge needs
Initiate investigations into techniques for knowledge • assimilation, evaluation, and utilization Seek and support research into the effectiveness • of interventions, including but not limited to AMU withdrawal, to mitigate existing AMR and apply such findings to future policies and actions Seek and support innovative research that expands • on the current 'farm-to-fork' approach to truly account for diverse human health outcomes Continue to advocate for fair, transparent, veterinary • drug regulations, AMR and AMU monitoring around the world based on scientific evidence, risk assessment, and appropriate precaution to ensure free and open trade of safe meat products Change Canada's veterinary drug regulations • to ensure prudent and safe antimicrobial use in animals and commit to transparent policy evaluation and action in the absence of efficacy or demonstrable harm Deliver antimicrobial use education to producers • and nutritionists Foster an innovative and collaborative relationship • between regulators, public health officials and the agriculture industry