Embed Size (px)
Citation preview
REVIEW Open Access
Antimicrobial drug use in food-producinganimals and associated human health risks:what, and how strong, is the evidence?Karin Hoelzer* , Nora Wong, Joe Thomas, Kathy Talkington, Elizabeth Jungman and Allan Coukell
Abstract
Background: Antimicrobial resistance is a public health threat. Because antimicrobial consumption in food-producing animals contributes to the problem, policies restricting the inappropriate or unnecessary agricultural useof antimicrobial drugs are important. However, this link between agricultural antibiotic use and antibiotic resistancehas remained contested by some, with potentially disruptive effects on efforts to move towards the judicious orprudent use of these drugs.
Main text: The goal of this review is to systematically evaluate the types of evidence available for each step in thecausal pathway from antimicrobial use on farms to human public health risk, and to evaluate the strength ofevidence within a ‘Grades of Recommendations Assessment, Development and Evaluation‘(GRADE) framework. Thereview clearly demonstrates that there is compelling scientific evidence available to support each step in the causalpathway, from antimicrobial use on farms to a public health burden caused by infections with resistant pathogens.Importantly, the pathogen, antimicrobial drug and treatment regimen, and general setting (e.g., feed type) can havesignificant impacts on how quickly resistance emerges or spreads, for how long resistance may persist afterantimicrobial exposures cease, and what public health impacts may be associated with antimicrobial use on farms.Therefore an exact quantification of the public health burden attributable to antimicrobial drug use in animalagriculture compared to other sources remains challenging.
Conclusions: Even though more research is needed to close existing data gaps, obtain a better understanding ofhow antimicrobial drugs are actually used on farms or feedlots, and quantify the risk associated with antimicrobialuse in animal agriculture, these findings reinforce the need to act now and restrict antibiotic use in animalagriculture to those instances necessary to ensure the health and well-being of the animals.
Keywords: Antimicrobial drug use, Farm animals, Antimicrobial resistance, Public health risk
BackgroundAntimicrobial use and resistance in animal agricultureAntimicrobial resistance is a global public health threat,reflected in at least 2 million resistant infections and atleast 23,000 deaths in the United States each year [1].Antimicrobial drug use is considered to be the singlemost important factor leading to resistance [1]. However,effective policy interventions are hindered by the factthat the emergence and spread of antimicrobial resist-ance is complex and the underlying dynamics
incompletely understood [2, 3]. Antimicrobial drugs areused in a variety of settings including hospitals, out-patient clinics, and long-term care facilities as well asanimal-associated settings such as veterinary clinics,farms, and feedlots. There is general scientific consensusthat antimicrobial drug use in a variety of settings con-tributes to the burden of antimicrobial resistance [2, 3].However, the relative contribution of different anti-microbial uses to the overall burden has so far remainedunclear. Moreover, despite decades of research and aconsiderable body of scientific evidence, the link be-tween antimicrobial drug use on farms and antimicrobialresistant infections in humans remains contested bycritics, primarily in the U.S. [3]. This dispute is
* Correspondence: [email protected] Pew Charitable Trusts, 901 E Street NW, Washington, DC 20004, USA
© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Hoelzer et al. BMC Veterinary Research (2017) 13:211 DOI 10.1186/s12917-017-1131-3
disruptive to the debate around judicious use of anti-microbial drugs in animal agriculture, and has the po-tential to slow the implementation of policies aimed atassuring the responsible and prudent use of antibioticsin animal agriculture. Notably, various organizations in-cluding the World Health Organization for Animals(OIE), U.S. Food and Drug Administration, and theBritish and American Veterinary Medical Associations,have defined the concept of responsible, prudent or judi-cious antimicrobial use in animal agriculture. Optimiz-ing therapeutic efficacy and minimizing antimicrobialresistance are key objectives underpinning all of thesedefinitions, despite some definitional differences. For thepurposes of this study, judicious antimicrobial use shallbe equivalent to OIE’s ‘responsible and prudent use’ def-inition of ‘improv[ing]animal health and animal welfarewhile preventing or reducing the selection, emergenceand spread of antimicrobial-resistant bacteria in animalsand humans.’
Policy solutions to ensure judicious antimicrobial use inanimal agricultureAround the world, policy efforts to ensure judiciousantimicrobial use in animal agriculture have generally, asa first step, focused on restricting the use of those anti-microbial drugs important for human medicine for‘growth promotion’ purposes. For the purposes of thisarticle, the Codex Alimentarius definition of growth pro-motion shall be used, where ‘growth promotion refers tothe use of antimicrobial substances to increase the rateof weight gain and/or the efficiency of feed utilization inanimals by other than purely nutritional means’ and an-timicrobials are thus administered to healthy animalswith the primary goal of increasing growth rates and en-hancing feed conversion efficiency [4].As early as 1969, a report by the U.K. Joint Committee
on the Use of Antibiotics in Animal Husbandry, knownas the ‘Swann Report’, motivated by concerns about theemergence of antimicrobial resistance, called for the banof medically important (i.e., shared between humans andanimals) antimicrobials for growth promotion purposes[3]. However, it took nearly another two decades beforegovernments took concrete action and individual coun-tries differ drastically in their responsiveness on the issue[3]. Sweden outlawed the use of all antimicrobial drugsfor growth promotion in 1986 and Denmark banned theuse of the two medically important antimicrobials, avo-parcin and virginiamycin, as growth promoters in 1995[5]. Avoparcin was banned for growth promoting pur-poses across the European Union in 1997 and growthpromotion uses of the remaining four medically import-ant antibiotics (as defined by the World HealthOrganization [6]) bacitracin, spiramycin, tylosin and vir-giniamycin were banned in 1999 [5]. Following the
precautionary principle, the use of any antimicrobialdrug for growth promotion, including antimicrobial drugclasses not used in human medicine, has been banned inEurope since 2006 [7]. In the U.S., the use of medicallyimportant antimicrobial drugs (as defined by the U.S.Food and Drug Administration [8]) for growth promo-tion was phased out on January 1, 2017. In fact, accord-ing to data collected by the OIE for 2015, more than70% of member countries that responded to the surveydid not authorize antimicrobial drugs for growth promo-tion purposes [9]. However, outlawing the use of medic-ally important antimicrobial drugs for growth promotionis only the first step towards assuring their judicious usein animal agriculture. Several European countries havetaken concrete next steps towards curtailing the emer-gence and spread of antimicrobial resistance by assuringantimicrobial drugs are used judiciously and only whennecessary to ensure the health and well-being of the ani-mal, and the U.S. Food and Drug Administration has re-cently announced plans to examine potential actions tothat effect [10].
The dynamics of antimicrobial resistance and theirrelevance for this studyFor the purpose of this study, antimicrobial resistance isdefined as ‘microbiological resistance’ and refers to theincreased resistance of a bacterium in vitro compared toa population of wild-type bacteria which can beexpressed, for instance, as an increase in minimal inhibi-tory concentration (MIC). Infections with bacteria thatmeet this definition of resistance may in fact on occasionstill respond to treatments with the antimicrobial drug.In this respect, microbiological resistance is differentfrom the concept of ‘clinical resistance’, which is focusedon treatment failures and considers clinical factors suchas the therapeutic concentration that can be reached atthe site of infection. To evaluate clinical resistance, MICvalues can be compared to clinical breakpoints, whichshould be specific to the animal species and site of infec-tion [11]. However, for many situations encountered inveterinary medicine specific breakpoints have not beenestablished and extrapolations from existing breakpointscan be challenging.This study focuses on the emergence and spread of ac-
quired antimicrobial resistance. This resistance canemerge through point mutations, usually associated witha progression from low-level to high-level resistance assequential mutations occur [12], or through horizontalgene transfer (HGT), which usually shows immediatehigh-level resistance as resistance genes are sharedamong bacteria [12]. HGT can occur through severalmechanisms [12]; for instance, plasmids that carry resist-ance genes can be shared among bacterial strains, bacte-riophages can transfer resistance genes from one
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 2 of 38
bacterium to another, and bacteria can take up nakedDNA (e.g., genes originating from dead bacteria). Not-ably, the relative importance of each resistance pathwayappears to at least partially depend on the bacterial spe-cies [12]. Acquired resistance has to be distinguishedfrom inherent resistance. Some bacteria are inherentlyresistant to a drug because the bacterium is outside ofthe drug’s spectrum of action, for example because thebacterium lacks the drug target. In addition, some bac-teria can be transiently resistant to a drug without a cor-responding genetic change, probably because of atransient dormant state in which the bacterium’s metab-olism is lowered to a point where it virtually ceases tofunction. These resistance types are generally believednot to be affected by antimicrobial use and will not beconsidered here.The analysis of the link between antibiotic use on
farms and human health risk is complicated by the factthat the evolutionary dynamics of antimicrobial resist-ance do not followed a simple ‘necessary and sufficient’[13] model of epidemiological causation, where the pres-ence of antimicrobial drug use would be both necessaryand sufficient for the emergence of resistance. In manycases, the same acquired antimicrobial resistance type ortrait can be encoded by more than one genetic mechan-ism. Resistance traits often, although not always, carry afitness cost, at least in in vitro experimental systems,and fitness costs may differ for the same resistance traitbased on the determining genetic changes [12]. If thereis a fitness cost associated with a resistance trait, resist-ant strains are selected against in antimicrobial-free en-vironments and selected for in the presence ofantimicrobials. The fitness cost may be higher forchromosomally-mediated than plasmid-mediated muta-tions, may differ by resistance mechanism (e.g., whetherresistance is conferred by modification of a metabolicpathway, alteration of a target site, or upregulation ofmembrane channels), and may increase with the numberof point mutations required to express the trait [12]. Insome cases, however, the fitness cost is very low. More-over, compensatory mutations that correct the fitnessloss brought about by the resistance-conferring muta-tions are possible. It is unknown how closely fitnesscosts under in vitro conditions track those experiencedin vivo – for instance, multi-drug resistance may be as-sociated with a lower fitness cost in vivo than predictedbased on in vitro data [14].Antimicrobial resistance is not a new phenomenon.
There is strong scientific evidence for the pre-existenceof some antimicrobial resistance genes (e.g., beta-lactamase genes) in the absence of any antimicrobial use[14]. In fact, many antimicrobial drugs are naturally pro-duced by fungi or bacteria to stave off competition fromother bacteria; some resistance genes are necessary to
allow these antimicrobial-producing microbes to survive,and others have emerged as an adaptation to the naturalpresence of these antimicrobial compounds in the envir-onment, primarily in soil [15]. The term ‘resistome’ hasbeen proposed to describe the ecology of resistancegenes, broadly encompassing all resistance genes circu-lating in pathogenic and non-pathogenic bacteria, bethey from soil, animals, humans, or other sources [15].A comprehensive knowledge of the resistome and ad-equate phylogenetic analysis is necessary to determinewhether a newly-detected resistance gene truly emergedrecently, has been present for a while but recently be-came more prevalent, or simply has not been detectedpreviously.The evolutionary dynamics of antimicrobial resistance
are further complicated by the potential for cross-resistance (the simultaneous resistance to multiple re-lated drugs that share a drug target) and co-resistance(where several genes are transferred together, for in-stance on one plasmid, and selection for one of thegenes indirectly selects for the others as well). Moreover,antimicrobial resistance to a given drug can be mediatedby multiple genetic changes, with potentially differentevolutionary dynamics, and bacterial mutation ratesvary, potentially pre-disposing some bacteria (i.e., ‘hyper-mutators’) to the more rapid emergence of resistance-conferring mutations than others.For all of these reasons, the exact dynamics between
antimicrobial use and resistance development may differby bacterial species, drug or drug target, and resistance-conferring mutation, and may be influenced by externalfactors such as the selection pressures exerted by the useof related drugs. Even if antimicrobial use decreases orceases, this may not necessarily translate into a direct,measurable drop in antimicrobial resistance, and the im-pact of introducing, restraining or eliminating antimicro-bial use may vary by bacterial strain, bacterial target andenvironment (e.g., as a result of different resistancemechanisms with different evolutionary and ecologicalproperties). Perhaps most importantly, resistance maynot in all cases be reversible by discontinuing antimicro-bial use alone. As is also clear from this discussion, thereare many aspects about the emergence, ecology and evo-lution of antimicrobial resistance genes that are not yetfully understood. For these reasons it is difficult to ex-trapolate from individual research studies to other set-tings, and to adequately evaluate all direct and indirectimpacts of antimicrobial use on the emergence ofresistance.Because antimicrobial drug use and the emergence of
resistance may not follow a direct cause and effect rela-tionship, it can be very difficult to establish causalityoutside of tightly controlled experimental settings. How-ever, many chronic diseases also do not follow a direct
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 3 of 38
cause and effect relationship and research methods de-veloped to address these challenges may be applicable toantimicrobial resistance. One widely used epidemio-logical model of causation that fits these chronicdiseases and antimicrobial resistance is that of ‘suffi-cient-component causes,’ first articulated by Rothman[16]. According to this model, a sufficient cause can bemade up of multiple components, each of which needsto occur in order for the sufficient condition to occur.For example, in order for a certain bacterium in a cer-tain environment and with a given genetic make-up toacquire antimicrobial resistance both exposure to anantimicrobial drug and external factors such as contactwith bacteria carrying a plasmid with resistance againstthe antimicrobial drug have to occur. In addition, therecan be more than one sufficient condition for a cause.For example, another bacterium with a different geneticpredisposition or exposed to other environmental condi-tions may not require exposure to the antimicrobial drugto develop resistance, for instance because of co-selection for heavy-metal resistance. While antimicrobialexposure may be a necessary condition for resistance inone situation, it may not be necessary in all situations,depending on the genetic predisposition and externalfactors. This model of causation, which is widely ac-cepted by epidemiologists, will be used here. It providesa useful framework for the assessment of causality, in-cluding in instances with seemingly contradictory studyfindings. Moreover, it demonstrates the challenge associ-ated with quantifying the proportion of antimicrobial re-sistant bacteria associated with different causes: if eachcause in fact consists of multiple component causes eachoccurrence of antimicrobial resistant bacteria could beattributed to each of the component causes.
Study aims and objectivesThe goal of this review is to provide an objective meth-odical summary of the available scientific evidence for oragainst a relationship between antimicrobial drug use onfarms and antimicrobial resistant human infections. Be-cause this nexus remains contested by some groups, par-ticularly in the U.S., relevant antibiotic use policies andrestrictions in the U.S. are highlighted where relevant.To achieve this, we review the scientific evidence sup-porting or refuting each step in the causal pathwaysfrom on-farm antimicrobial use to human public healthrisk, placing particular emphasis on the strengths andlimitations of the available scientific evidence. Breakingdown the exceedingly complex pathway from farm topublic health risk into discrete intermediary steps con-siderably reduces complexity and allows for ahypothesis-driven approach. The goal of this study is tocharacterize the link between antimicrobial use on farmsand human infections with resistant bacteria, rather than
quantifying the relative importance of this link comparedto antimicrobial drug use in other settings even thoughthat quantification will ultimately be important to guidepublic health policy and data gaps that complicate orprevent quantification are highlighted. Similarly, quanti-fying the relative importance of different transmissionroutes from farm to humans is important but beyondthe scope of this study.
Pathways from antimicrobial drug use on farms to resistantinfections in humansThere are different pathways that can lead from anti-microbial use on farms to a public health risk, includingfoodborne and non-foodborne routes. (Drug residues area separate public health concern not considered in thisstudy.) Regardless of pathway, four distinct factors haveto be understood to characterize the public health riskassociated with antimicrobial drug use in animalagriculture.
Antimicrobial drug use on farms and feedlotsHow antimicrobial drugs are used on farms and feedlotsis central to understanding the emergence of antimicro-bial resistance. Unfortunately, the amount of informa-tion available on actual antimicrobial drug use on farmsand feedlots differs considerably across countries and isin many cases insufficient for understanding antimicro-bial exposures.
Risk of resistance emergence as a result of antimicrobialexposure on farms and feedlotsThis question concerns foodborne and zoonotic patho-gens as well animal pathogens and commensal bacteria(i.e., the very large number of naturally occurring micro-organisms that usually inhabit the body surfaces ofhumans and animals [17]). Resistance in commensal oranimal pathogens poses a human health risk only if theresistance genes can be transferred to human pathogens;this is regardless of whether the transfer occurs in thegut of farm animals, in the environment – be that onfarms or off - or within the human gut. Horizontal genetransfer between human commensals and human patho-gens has been widely demonstrated [18]. Therefore, evi-dence of resistance gene transfer from animal-associatedbacteria to human commensals can be regarded as indir-ect evidence for a potential transmission to human path-ogens and therefore a public health risk.
Risk of infection due to resistant bacteria that emerged onthe farmBacteria, including foodborne or zoonotic pathogens,can be transferred from food producing animals tohumans through direct contact, or indirectly throughfood or the environment. In addition to contact with the
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 4 of 38
environment in which the animals are raised, indirecttransmission may also include exposure to agriculturaloperations through manure run-offs, airborne particles,or other environmental exposures, even though thesetransmission routes are considerably less well under-stood and documented than the other possible transmis-sion routes [19–21]. Some evidence further suggests thathumans can transmit bacteria that originated on farmsto other humans through direct contact, food contamin-ation during processing, or contamination of shared en-vironments, but, again, the evidence is limited and oftencircumstantial, the underlying dynamics are not wellunderstood, and at least some animal-associated bacteriamay be poorly equipped to be transmitted from humanto human [22–24].In some cases (e.g., foodborne outbreaks) the direc-
tionality of infection is quite obvious, but this is not al-ways true. The question of directionality has to beconsidered in evaluations of the available scientific evi-dence. However, even if directionality may not always beclear, such studies imply a shared host range, whichmakes cross-species transmissions likely.
Excess morbidity and mortality caused by antimicrobialresistance traits that emerged on farmsSeveral specific mechanisms by which antimicrobial re-sistance can have adverse human health impacts havebeen identified in the literature, of which at least threeare directly relevant to resistance that emerged in animalagriculture [25]:
1. Linkage of virulence and resistance traits, leading todrug-resistant strains with increased virulence
2. Treatment delay because initial treatments areineffective
3. Necessity to choose less desirable treatment optionsbecause of resistance to more desirable antimicrobialdrugs
Observational studies on the topic can be complicatedby the presence of potential confounders such as differ-ences in age or underlying disease among patient groups[26, 27]; the choice of reference group can also have sig-nificant impacts on study findings [28]; in addition, bac-terial strains can differ in the severity of associatedhealth outcomes independent of antimicrobial resistance[29], thus potentially introducing another source ofconfounding.Given the vast amount of literature published on the
subject, this review does not strive to be comprehensivein reviewing the available literature associated with thetopic. Rather, for each step in the pathway a selectednumber of studies that exemplify each relevant studytype are discussed together with a general discussion of
the strength and limitations of the available evidence.Unanswered questions and areas requiring further studyare clearly highlighted. Therefore, this study documentsthe current scientific understanding about the link be-tween antimicrobial drug use on farms and humanhealth risks associated with antimicrobial resistant infec-tions, highlighting what is known and what remains tobe determined.
Main textA modified GRADE approach for evaluating the strengthof scientific evidenceThe approach for evaluating the scientific evidence for alink between antimicrobial use in food producing ani-mals, antimicrobial resistance, and a resulting humanhealth risk is based on the principles guiding theGRADE (Grades of Recommendation, Assessment,Development and Evaluation) approach to systematic re-views. The fundamental notion underpinning this ap-proach is that scientific studies vary in the strength ofevidence they provide, or, as stated by the Cochrane Col-laboration, the ‘extent to which one can be confidentthat an estimate of effect or association is close to thequantity of specific interest’. The GRADE approach con-siders the study type as well as factors relevant to studyquality that are specific to an individual study (e.g.,width of confidence intervals, inconsistency of results)to arrive at a quality rating for each study. The grade ap-proach considers three basic types of evidence [30]:
1. Controlled trials are intervention studies wheresubjects are allocated by the investigators intotreatment and control groups, ideally randomly butoccasionally based on another allocation scheme.The GRADE approach classifies randomizedcontrolled trials (RCTs) as ‘high quality of evidence’because further research is unlikely to change theconfidence in the effect estimate. However, asdetailed below, the presence of specific studylimitations can cause the quality of evidence to bedowngraded.
2. Observational studies include epidemiologicalstudy designs such as cohort, case-control andcross-sectional studies; contrary to RCTs the investi-gators do not allocate subjects to specific exposures,but compare outcomes among exposed and compar-able unexposed population subgroups. The GRADEapproach classifies observational studies as ‘low qual-ity of evidence’. Because allocation of individuals todifferent exposures cannot be controlled, biases aremore difficult to control than in RCTs and thereforefurther research is likely to have an important im-pact on the confidence in the effect estimate and islikely to change the estimate. The presence of
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 5 of 38
specific study characteristics can cause the quality ofevidence to be up- or downgraded.
3. Any other evidence is classified in the GRADEapproach as ‘very low quality of evidence’ becauseany effect estimate is uncertain; however, specificstudy characteristics can cause the quality ofevidence to be upgraded, and a variety of study typesin this category can provide useful information,including for example molecular subtyping studies,phylogenetic analyses, and case studies or outbreakinvestigations that lack a proper control group.
In the GRADE approach, the evidence ratings can beup- or downgraded based on a variety of study charac-teristics that impact study quality, including [31]:
1. Study limitations and risk of bias2. Directness of evidence (i.e., the extent to which
the subjects, interventions and outcome measuresare similar to those of interest)
3. Consistency of results (i.e., similarity of effectestimates across studies)
4. Precision of estimates (e.g., width of confidenceintervals around effect estimates)
5. Risk of residual confounding and publication bias
Finally, the strength of recommendations based on thisevidence is affected by [31]:
1. Quality of evidence2. Uncertainty about the balance between desirable
and undesirable effects3. Uncertainty or variability in values and preferences4. Uncertainty about whether the intervention
represents a wise use of resources
Importantly, practical as well as ethical considerationscan restrict the type of evidence that can be collected toaddress a specific research question in a meaningful way[32]. For instance, a randomized controlled trial toevaluate human health outcomes associated with expos-ure to drug resistant foodborne pathogens would inmost cases be unethical.In the following sections, each step in the causal path-
way from antimicrobial use in food producing animalsto public health risks caused by infections with anti-microbial drug resistant pathogens is described and thestrength of the available evidence is evaluated. A similarapproach has been used by the Australian governmentto review the scientific literature on antimicrobial resist-ance in animal bacteria and human disease in a studycompleted in 1999 [33].Notably, because a number of somewhat different defi-
nitions have been used in different scientific disciplines
it is important to clearly define the terms ‘antimicrobialdrug’ and ‘antibiotic’. For the purpose of his study, andfollowing similar definitions in other authoritative publi-cations [34], we use the term ‘antimicrobial drug’ torefer to ‘antibacterial agents classically used for therapy,prophylaxis, or growth promotion’ as adapted fromEFSA [34]. The effects of disinfectants, biocides, anti-viral, antifungal or antiparasitic drugs will not be consid-ered while synthetic antibacterial agents will beconsidered. For the purposes of this study and followingthe definitions herein, the terms ‘antimicrobial drug’ and‘antibiotic’ will be used synonymously.
Evaluating the link between antimicrobial use on farms tothe emergence of antimicrobial resistant humaninfectionsAntimicrobial drugs use on farms and feedlotsIntervention studies are not applicable for measuringantimicrobial use but a number of observational studieshave been conducted. Many follow a cross-sectional de-sign where antibiotic consumption is recorded at onepoint in time, even though longitudinal designs have alsobeen used and may be preferable to address certain re-search questions (see for instance [35]). As the examplesin Table 1 demonstrate, a number of different datasources are available that contain information regardingon-farm antimicrobial drug use. Data should ideally bebased directly on a review of treatment and/or prescrip-tion records (e.g., [35–39]). However, this data sourcemay not be readily available or the data collection maybe cost-prohibitive. In these cases, more indirect sourcesmay be used, for instance questionnaires administered toveterinarians [40], producers [41–43] or both [44]. Inaddition, other data sources such as antimicrobial salesdata reported by drug manufacturers (e.g., data reportedin the U.S. under section 105 of the Animal Drug UserFee Act) may be used albeit the information collected inthese cases is less direct and detailed, and consequentlyless informative. In some cases, it may be possible tocollect data from each relevant operation – for instanceevery producer in the target area (see for instance [36]).Often, however, this is impossible and a subset of opera-tions must be sampled. Ideally, the sampling designshould lead to nationally representative estimates (forexample [35, 40]). Realistically, it may be possible onlyto collect data that is representative of a subset of opera-tions, for instance operations representing a segment ofthe animal agriculture industry such as dairy herds [41]or broiler chicken and turkey flocks [35]. Similarly, stud-ies may focus on operations located in a specific geo-graphic region such as the densest pig production areain Belgium [37], Vietnam’s Tien Giang province [42] orGermany’s state of Northrhine-Westfalia [39]. Regardlessof study scope, a robust sampling frame and an
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 6 of 38
Table
1Exam
ples
ofselect
stud
iesavailableforevaluatin
gantim
icrobialconsum
ptionin
anim
alagriculture
Reference
Stud
yDatasource
Popu
latio
n;no
.of
farm
sor
prod
ucers
orveterin
arians
Nationally
represen
tative
Directne
ssof
eviden
ceInform
ationregarding
Stud
ytim
ing/
perio
dCou
ntry
Setting
Goalsor
focus
dosage
duratio
nreason
foruse
GRA
DElevel:ob
servationalstudy
[35]
France
governmen
tantib
iotic
consum
ption
metrics
treatm
entrecords(re
view
edmed
ication
purchases,daily
records,form
ssubm
itted
atslaugh
ter;cond
uctedfarm
erinterviews)
Turkey
andchicken
broilerflocks(n
=106;57
turkey
&49
chicken
flocks)
Restrictedto
western
France
actual
treatm
ent
records
yes
yes
limited
one
time
[36]
Den
mark
governmen
tantib
iotic
consum
ption
treatm
entrecords(datacollected
from
pharmacies,veterin
arians
andfeed
mills)
allfarm
anim
alop
erations
(all
prescriptio
ns)
Yes
actual
treatm
ent
records
yes
yes
yes
annu
al
[37]
Belgium
academ
icantib
iotic
consum
ption
treatm
entrecords
Pighe
rdsin
Belgium’s
densestpigprod
uctio
narea
Restrictedto
keypig
prod
uctio
narea
actual
treatm
ent
records
yes
yes
limited
one
time
[38]
Canada
academ
icantib
iotic
usepractices
treatm
entrecords&prod
ucer
questio
nnaires
Beef
herdsin
western
Canada(n
=203)
Restrictedto
western
Canada
actual
treatm
ent
records
yes
yes
yes
one
time
[39]
Germany
academ
icantib
iotic
usepractices
treatm
ent&prescriptio
nrecords
Cattle
farm
sin
Northrhine-Westfalia
(n=47)
Restrictedto
one
geog
raph
icarea
actual
treatm
ent
records
yes
yes
yes
one
time
[40]
Finland
academ
icantib
iotic
usepractices
veterin
arianqu
estio
nnaires
Veterin
arians
inFinland
Yes
respon
seto
survey
yes
yes
yes
one
time
[41]
U.S.
academ
icantib
iotic
useandfarm
managem
entpractices
onconven
tional&
organic
operations
prod
ucer
questio
nnaire
Dairy
farm
sin
4U.S.
states
(n=99
conven
tionaland
n=32
organic)
No
respon
seto
survey
nono
yes
one
time
[44]
Vietnam
academ
icantib
iotic
usepractices
and
farm
managem
ent
prod
ucer,veterinarianor
technicalstaff
questio
nnaires
Pigandpo
ultry
operations
in3locatio
nsin
Vietnam
(n=270)
Restrictedto
redriver
delta
respon
seto
survey
nono
limited
one
time
[42]
Vietnam
academ
icqu
antifyantib
iotic
use;
iden
tifyriskfactorsforuse
prod
ucer
questio
nnaire
Chicken
farm
sin
Vietnam’sTien
Giang
province
(n=208)
Restrictedto
oneprovince
respon
seto
survey
yes
yes
limited
one
time
[43]
U.S.
governmen
tfarm
managem
ent
prod
ucer
questio
nnaire
U.S.ope
ratio
nsa
Yes
respon
sesto
survey;
traine
dinterviewer
nolim
ited
limited
perio
dic
every5–
7years
[45]
U.S.
governmen
tfarm
financials
prod
ucer
questio
nnaire
U.S.ope
ratio
nsYes
respon
sesto
survey;
traine
dinterviewer
nono
noannu
al
GRA
DElevel:othe
r
[129]
U.S.
governmen
tantib
iotic
sales&distrib
ution
annu
alrepo
rtingof
antib
iotic
salesby
drug
spon
sors
AllU.S.d
rugspon
sors
Yes
Salesdata
nono
limited
annu
al
a Surveyspecificto
speciesan
dprod
uctio
nclass(e.g.,cow-calf,feed
lot);con
ducted
onrotatio
nalb
asis;app
roximatelyevery5–
7years
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 7 of 38
appropriate sample size are needed to ensure the studywill be representative of the target population, and ex-trapolations beyond that population may be challenging.Studies of antimicrobial consumption can be con-
ducted for surveillance or research purposes. Surveillancestudies are typically conducted by governmental organiza-tions (e.g., [36, 43]) and research studies more commonlydone by academic investigators (e.g., [37–42, 44]). In somecases, a study goal may for instance be to evaluate differentantibiotic use metrics (e.g., [35]), but more common re-search questions involve evaluating antibiotic use practices([38, 39]) or comparing antibiotic use and farm manage-ment practices across different types of operations, such asconventional and organic operations [41]. The study goalshave important implications for which study aspects toprioritize if resource limitations force trade-offs amongthem. For instance, for surveillance purposes it may bemore important to collect nationally representative dataand to repeat the data collection periodically to evaluatetemporal trends ([36, 43, 45]) than to collect highly granu-lar and detailed data on related issues such as correlation ofantibiotic use with farm management practices. In contrast,some research questions may require the collection of com-prehensive and highly detailed data on study operations butfunding may not be available to repeat the study periodic-ally (e.g., [38, 40, 41]). Regardless of purpose, though, infor-mation on antimicrobial dosage, duration and reason foruse are key determinants of antimicrobial use patterns andnecessary to evalaute antibiotic exposures. This informationshould be collected in surveillance as well as research stud-ies where possible. Unfortunately, this information is notconsistently collected in all studies (see Table 1) and for in-stance in the U.S. this data is currently very scarce.Antimicrobial sales data are the only nationally repre-
sentative, annually available data source in the U.S.However, these data do not directly capture antimicro-bial use on farms, and are currently not specific to indi-vidual species, even though drug companies are requiredto estimate sales for the major food-producing species(i.e., cattle, swine, chicken and turkey) separately startingwith the reporting of the 2016 data. Other nationallyrepresentative data sources in the U.S. are surveys con-ducted as part of the National Animal Health Monitor-ing System, which are focused on farm managementpractices, not quantitative with respect to dosage andduration, and currently only conducted every 5 to7 years; and surveys conducted as part of the Agricul-tural Resource Management Survey, which are focusedon farm finances. Despite these data limitations it isclear that antimicrobial drugs are used on U.S. farmsand feedlots. According to the most recent availablesales data, almost 10 million kilograms of medically im-portant antibiotics approved for food-producing specieswere sold in the U.S. in 2015, including 6.9 million
kilograms of tetracyclines, 900 thousand kilograms ofpenicillin and 600 thousand kilograms of macrolides.Also sold were drugs in a number of other classes in-cluding aminoglycosides, amphenicoles, cephalosporins,fluoroquinolones, lincosamides, and sulfonamides. Whileit is thus not currently possible to sufficientlycharacterize antibiotic exposures on U.S. farms and feed-lots they undoubtedly occur.
Risk of resistance emergence as a result of antimicrobialexposure on farms and feedlotsData based on studies in foodborne and zoonoticpathogensData on the emergence of resistance in foodborne orzoonotic pathogens is directly relevant for human healthbut studies have more commonly focused on commensalbacteria or animal pathogens. Yet, as shown in Table 2, avariety of study types are available that evaluate the linkbetween antimicrobial use on farms and the emergenceof resistance in foodborne or zoonotic pathogens. Theseinclude controlled trials, various observational studytypes, and other studies such as microbiological studiescomparing resistance levels across production types.Among the controlled trials allocation to treatment or
control groups should ideally be random and often is([46–48]), but this is unfortunately not in all studies suf-ficiently described [49, 50]. Treatment and control ani-mals are often experimentally challenged with foodbornepathogens such as Campylobacter [46, 49] or Salmonella[47, 48, 51]. In these cases bacterial strains used for theinoculum are well-characterized and the emergence ofresistance can easily be detected. However experimentalchallenges may differ from naturally occuring infectionsin important ways including for instance the types anddiversity of bacterial strains used for infection, the meta-bolic profiles of the bacteria at the time of infection, andthe bacterial quantities used for challenge. Naturallyoccuring bacterial populations of Campylobacter havealso been investigated [50], and in some cases naturallyoccuring populations of E. coli have been analyzed to-gether with experimentally inoculated human pathogens[48, 51]. In most cases, the outcome of interest isphenotypic resistance measured by dilution or, morerarely, disc diffusion test but phenotypic methodshave also been combined with genotyping, which pro-vides more detailed information about the emergingresistance traits [51].A number of controlled studies have demonstrated the
emergence of resistance in Campylobacter isolates afterexposure to antimicrobial drugs . For instance, one studyexperimentally infected groups of 15 broiler chickeneach on day 19 of life with a quinolone-sensitive Cam-pylobacter strain, and exposed the birds to enrofloxacinflumequine at low (15 ppm) or high (50 ppm) dosages in
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 8 of 38
Table
2Exam
ples
ofstud
iesevaluatin
gtheem
erge
nceof
resistance
infood
bornepathog
ensafterfood
prod
ucinganim
alswereexpo
sedto
antim
icrobialdrug
sReference
Stud
yKeyfinding
s
Type
aspecies
bacterialtarge
tantim
icrobialdrug
administration
route
stud
you
tcom
esrelatedto
antim
icrobial
resistance
stud
yde
sign
suscep
tibility
testing
controls
Grade
level:Con
trolledtrial
[46]
RCT
broiler
chicken
Campylobacter
expe
rimen
tal
infectionwith
5suscep
tible
strainson
day16
or24
oflife
fluoroq
uino
lone
sen
rofloxacin,
sarafloxacin,
ciprofloxacin;
treatedfor
5days,startingon
day30
oflife
drinking
water
emerge
nceof
resistance
2grou
ps;25or
50chicks
per
grou
p(half
treated)
Phen
otypic
agar
dilution
negativecontrols
resistance
emerge
drapidlyin
isolates
from
treatedbirdsand,
forbirds
treatedwith
enrofloxacin,
resistance
was
retained
throug
hout
the
samplingpe
riod;
resistance
remaine
dlow
inun
treatedcontrols
throug
hout
thestud
y
[47]
RCT
pigs
Salmon
ella
Typh
imurium
DT104;n
ativeE.coli
popu
latio
nmon
itored;
expe
rimen
talinfectio
nwith
5strainsisolated
from
apparentlyhe
althypigs
(pen
taresistantph
enotype)
aureom
ycin,stand
ardor
subthe
rape
uticdo
se,treated
for7days
startin
g48
hafter
inoculation
feed
emerge
nceof
resistance
3grou
ps;6
pigs
pergrou
pPh
enotypic
agar
dilution
negativecontrols
resistance
emerge
din
nativeE.coli
popu
latio
nsfro
mtreatedanim
als
andincreasedresistance
persisted
for2weeks
aftertreatm
ent;treated
pigs
shed
high
ernu
mbe
rsof
Salmon
ella
Typh
imurium
DT104
than
untreatedpigs.
[48]
RCT
pigs
Salmon
ella
Typh
imurium;
nativeE.colipo
pulatio
n;expe
rimen
talchalleng
eof
piglets2days
afterweaning
aparam
ycin
indifferent
dosing
sche
mes,insome
casesfollowed
bysulfamethazine
andcarbadox
orge
ntam
icin
andne
omycin;
treatm
entinitiated
7days
afterinfection,
for14
days
feed
ordrinking
water
emerge
nceof
resistance
2trials;6
grou
psof
8pigs
each
Phen
otypic
dilutiontest
negativecontrols
resistance
ofSalmon
ella
isolates
was
notim
pacted
bytreatm
entregimen
;resistance
emerge
din
nativeE.coli
popu
latio
nsandwas
effected
bytreatm
entregimen
;use
ofsimilar
antib
ioticsin
rotatio
nresultedin
greaterresistance
than
ifdissim
ilar
antib
ioticswereused
[49]
CT
broiler
chicken
Campylobacter;
expe
rimen
talinfectio
nwith
suscep
tiblestrain
onday19
oflife
fluoroq
uino
lone
s:en
rofloxacinor
flumeq
uine
indifferent
concen
trations;
treatedfor4days,b
eginning
7days
afterinfection(oron
day1of
theexpe
rimen
tfor
onegrou
p)
drinking
water
emerge
nceof
resistance
8grou
ps;15
chicks
pergrou
pPh
enotypic
disc
diffusio
nne
gativecontrols
resistance
emerge
din
Campylobacter
isolates
from
enrofloxacintreated
birds(isolates
from
allo
ther
grou
psremaine
dsuscep
tible)
[50]
CT
beef
cattle
Campylobacter;n
aturally
occurringbacterial
popu
latio
ns
chlortetracyclinewith
orwith
out
sulfamethazine
;virg
iniamycin;
mon
ensin;
ortylosin;
treated
for56
days,starting18
days
afterarrivalat
feed
lot;
antim
icrobialsremoved
for
91days,reintrodu
cedfor
42days
feed
emerge
nceof
resistance
30grou
ps;10
steerspe
rgrou
pPh
enotypic
agar
dilution
negativecontrols
administrationof
antim
icrobialdrug
sincreasedthecarriage
rate
for
resistantisolates,b
utresults
differed
byantim
icrobialdrug
and
Campylobacter
species
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 9 of 38
Table
2Exam
ples
ofstud
iesevaluatin
gtheem
erge
nceof
resistance
infood
bornepathog
ensafterfood
prod
ucinganim
alswereexpo
sedto
antim
icrobialdrug
s(Con
tinued)
[51]
RCT
pigs
Salmon
ella
Typh
imurium
with
nalidixicacid
resistance;n
ativeE.coli
popu
latio
n;expe
rimen
tal
challeng
eof
piglets(from
treatedor
untreatedsows)
oxytetracycline,aparam
ycin;
sowstreatedpriorto
farrow
ing
feed
emerge
nceof
resistance
(inpigletsafter
treatm
entof
sows)
3grou
psof
pigs
Phen
otypic
dilutiontest
Gen
otypic
PCR-based
negativecontrols
Expo
sure
ofsowsto
oxytetracycline
associated
with
increasedresistance
inbacterialisolatesfro
mpigs
[52]
RCT
Pigs
Campylobacter;n
aturally
occurringpo
pulatio
nsen
rofloxacin;
treatedwith
standard
therapeutic
dose
for
5days
peros
emerge
nceof
resistance
2grou
psof
pigs;
6pigs
pergrou
pPh
enotypic
agar
dilution
negativecontrols
expo
sure
toen
rofloxacinresultedin
Campylobacter
isolates
resistantto
nalidixicacid
andciprofloxacin(no
quinolon
eresistance
detected
prior
toantib
iotic
expo
sure);resistant
isolates
detected
forup
to35
days
[53]
RCT
Pigs
Salmon
ella
Typh
imurium;
expe
rimen
talchalleng
ewith
markedstrain
containing
nalidixicacid
resistance
marker
ceftiofur,aparam
ycin,o
rcarbadox,followed
byoxytetracycline;Treated
startin
g2days
post
inoculation
feed
(i.m.for
ceftiofur)
emerge
nceof
resistance
4grou
psof
pigs;
12pigs
per
grou
p
Phen
otypic
Disc
diffusio
nne
gativecontrols
frequ
ency
ofantim
icrobialresistance
variedover
time–resistance
was
loweston
day4po
stinoculationbu
tincreasedsteadilyafterw
ards
and
was
high
eston
thelastdayof
sampling;
continuedincrease
ofselect
antim
icrobialresistance
after
expo
sure
was
discon
tinued
Grade
level:ob
servationalstudy
[54]
Coh
ort
perio
d:2000–2001
dairy
cattle
Salmon
ella
vario
usused
onconven
tional
farm
sn/a
antim
icrobial
resistance
levels
95farm
s:26
organicand69
conven
tional
Phen
otypic
broth
dilution
organicvs.
conven
tional
results
differedby
antib
iotic;for
most
antim
icrobialdrug
s,resistance
levels
wereno
tsign
ificantlydifferent;
resistance
tostreptom
ycin
and
sulfametho
xazole
was
greateron
conven
tionalfarms.
[55]
CS
dairy
cattle
Campylobacter
vario
usused
onconven
tional
farm
sn/a
antim
icrobial
resistance
levels
60farm
s:30
organicand30
neighb
oring
conven
tional
Phen
otypic
disk
diffusio
norganicvs.
conven
tional
nostatisticallysign
ificant
difference
inresistance
ratesfororganicvs.
neighb
oringconven
tionald
airy
herds
[56]
CS
broiler
chicken
and
turkey
Campylobacter
vario
usused
onconven
tional
farm
s(gen
tamicin,
lincomycin,b
acitracin,
virginiamycin,etc.)
feed
and
water
antim
icrobial
resistance
levels
30farm
s:10
organic,20
conven
tional
Phen
otypic
agar
dilution
organicvs.
conven
tional
antim
icrobialresistance
rateswere
sign
ificantlyhigh
erin
isolates
from
conven
tionalthanorganicfarm
s,and
conven
tionalisolateswere
sign
ificantlymorelikelyto
bemulti-
drug
resistant;resistance
levelsvaried
acrossantim
icrobialdrug
s
Grade
level:othe
r
[57]
Long
itudinal
stud
ybroiler
chicken
Campylobacter
fluoroq
uino
lone
s;difloxacin
oren
roflaxcin;treatmen
tfor
5days
totreatclinically
relevant
infection;
bacterial
samples
takenbe
fore,d
uring
andaftertreatm
entof
flocks
drinking
water
prevalen
ceof
antim
icrobial
resistance
5commercial
flocks
Phen
otypic
agar
dilution
Gen
otypic
PCR-based
resistance
before,
durin
gandafter
treatm
ent
prevalen
ceof
resistantstrains
increaseddu
ringtreatm
entand
remaine
delevated
aftertreatm
ent
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 10 of 38
Table
2Exam
ples
ofstud
iesevaluatin
gtheem
erge
nceof
resistance
infood
bornepathog
ensafterfood
prod
ucinganim
alswereexpo
sedto
antim
icrobialdrug
s(Con
tinued)
[58]
bacterial
samples
atslaugh
ter
broiler
chicken
Campylobacter
vario
usused
onconven
tional
farm
svario
usprevalen
ceof
antim
icrobial
resistance
atslaugh
ter
n=600ceacal
samples
from
chicken
Phen
otypic
dilution
metho
d
resistance
levels
atslaugh
teras
afunctio
nof
prod
uctio
ntype
andantim
icrobial
usehistory
antim
icrobialresistance
levels
differedsign
ificantlyby
prod
uctio
ntype
andantim
icrobial
administrationhistory
[59]
bacterial
samples
ofretailmeats
Chicken
meat
Campylobacter
fluoroq
uino
lone
presum
ably
water
Prevalen
ceof
resistance
5prod
ucers,
meatssampled
atretailfor20
consecutive
weeks
in2004
and15
consecutive
weeks
in2006
Phen
otypic
NAEvs.
conven
tional
Nostatisticallysign
ificant
difference
influoroq
uino
lone
resistance
levels
afterusewas
repo
rted
lydiscon
tinued(nosamplecollection
orvalidationat
thefarm
orprod
ucer
level)
[60]
bacterial
samples
chicken
and
turkey
meat;
healthy
human
volunteers
Enterococci;VanA
-type
vancom
ycin
resistant(VRE)
avop
arcinuseforgrow
thprom
otionph
ased
out
durin
gstud
ype
riod
vario
uslevelsof
VRE
prevalen
ceconven
ience
samples
N/A
VREcarrierrates
before
andafter
theban
VREde
tectionin
poultrymeatand
intestinalbacteriaof
healthy
volunteersde
creasedaftertheban
[61]
bacterial
samples
humans
and
poultry
prod
ucts
Campylobacter
fluroqu
inolon
esn/a
prevalen
ceof
resistance
conven
ience
sample,collected
from
1982
to1989
Phen
otypic
disk
diffusio
nprevalen
ceof
resistance
over
time
prevalen
ceof
quinolon
eresistance
increasedfro
m0to
14%
inpo
ultry
prod
uctsandfro
m0to
11%
inhu
man
samples
[62]
bacterial
samples
atslaugh
ter
broiler
chicken
Campylobacter
vario
usused
onconven
tional
farm
s;1992–1996tim
epe
riod
falls
before
banforantib
iotic
grow
thprom
oters,2001–
2002
perio
ddo
esno
t
vario
usprevalen
ceof
antim
icrobial
resistance
atslaugh
ter
22flocks;10
chickens
per
flock
Phen
otypic
agar
dilution
resistance
levels
atslaugh
teras
afunctio
nof
prod
uctio
ntype
(free-range
orstandard)and
stud
ype
riod
antim
icrobialresistancewas
morelikely
inisolatesfrom
standard
than
free-rang
echicken,differedby
Campylobacterstrain,
andform
anyantim
icrobialdrugsand
strainsresistanceincreasedovertim
e
[63]
correlation
stud
yisolates
from
chicken
meatand
humans
Salmon
ella
Heide
lberg,
ceftiofurresistant
commen
salE.coli
2003–2008
n/a
prevalen
ceof
ceftiofur-
resistance
inbacterialiso-
latesfro
mchickenmeat
andhu
man
cases
Correlatio
nacross
provinces
andtim
epe
riods
Phen
otypic
broth
dilution
Resistance
levels
before,d
uring,
andaftera
voluntary
discon
tinuatio
nof
ceftiofurusein
hatche
riesin
Quebe
c
Statisticallysign
ificant
correlation
betw
eenprevalen
ceof
ceftiofur-
resistantSalmon
ella
Heide
lbergon
retailchickenandhu
man
infections;
inQuebe
c,levelsof
ceftiofurusein
hatche
riesdu
ringstud
ype
riod
seem
edcorrelated
with
ceftiofur
resistan
cein
chickenSalm
onella
andE.
coliisolates.
[64]
Long
itudinal
stud
ypigs
Enterococcus
faecium
tylosin;
resistance
toerythrom
ycin,spiramycin,
clindamycin,p
enicillin,
tetracycline,chloramph
enicol,
streptom
ycin,g
entamycin
feed
prevalen
ceof
antim
icrobial
resistance
16farm
s;13
farm
swere
exam
ined
again
183to
287days
later
Phen
otypic
disk
diffusio
nresistance
before
andaftertheban
oftylosinas
agrow
thprom
oter
Resistance
tomacrolides,
lincosamides
andtetracyclines
decreasedaftertheuseof
tylosinfor
grow
thprom
otionwas
bann
ed.
a CTcontrolledtrial(allocatio
nsche
meno
tclearly
describ
ed),RC
Trand
omized
controlledtrial,CS
cross-sectiona
lstudy
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 11 of 38
drinking water for 4 days, beginning on day 26 of life,7 days after infection [49]. On days 29, 33 and at slaugh-ter on day 43, Campylobacter isolates from both exposedgroups harbored resistance to enrofloxacin as well asnalidixic acid and flumequine, while those from unex-posed control groups remained sensitive to these drugsand non-infected control groups did not yield Campylo-bacter isolates, making introduction of other Campylo-bacter strains from outside sources unlikely. In anothertrial, broiler chicken were experimentally infected withsusceptible Campylobacter strains on days 16 or 24 oflife and exposed to enrofloxacin or sarafloxacin in drink-ing water for 5 days, starting on day 30 of life [46]. Flur-oquinolone resistance emerged rapidly in Campylobacterisolates from exposed birds, and, for birds treated withenrofloxacin, resistance was retained throughout thesampling period. Another controlled trial evaluated theimpact of a single enrofloxacin exposure on naturallyoccuring Campylobacter populations isolated from pigs.Before exposure, all 867 collected isolates were suscep-tible to nalidixic acid and ciprofloxacin. After treatment,isolates with high resistance to nalidixic acid and cipro-floxacin were detected in the exposed groups, up to35 days after treatment [52]. Similar observations havealso been made for naturally occuring Campylobacterpopulations in beef cattle. In this study, steers were ex-posed to a variety of antimicrobial treatments (i.e., chlor-tetracycline with or without sulfamethazione,virginiamycin, monensin and tylosin) in feed for 56 days,and then again for 42 days after 91 days without anyantimicrobial exposures [50]. Exposure to antimicrobialsresulted in an increased carriage rate for resistant Cam-pylobacter isolates but the authors noted marked differ-ences among antimicrobial drugs and Campylobacterspecies. For Campylobacter jejuni, administration of anti-microbials significiantly increased resistance rates toampicillin and tetracycline but not erythromycin. In con-trast, for Campylobacter hyointestinalis, results were notstatistically signficiant for ampicillin and erythromycinresistance was statistically signficiant only in cattle fedtetracycline, highlighting the potentially important dif-ferences among antimicrobial drugs and bacterialspecies.In fact, studies for Salmonella have yielded somewhat
different results. A randomized controlled trial investi-gating the impact of different treatment regimen admin-istered to pigs through feed or drinking water failed tofind a statistically significant impact of antimicrobial ex-posure on resistance in the Salmonella Typhimuriumisolates used in the experimental challenge, but did findan impact on resistance in the native E. coli populationwhich was effected by treatment regimen [48]. In thatstudy, piglets were experimentally challenged with Sal-monella two days after weaning and treatement through
feed or drinking water was initiated 7 days after infec-tion, for 14 days. Notably, data for Salmonella Typhi-murium were not actually presented in the study andisolation rates were not reported. Still, the reasons forthis apparent difference in results for Salmonella com-pared to native E. coli are not clear. In another con-trolled trial, sows were treated with oxytetracycline priorto farrowing (or not), their piglets challenged with Sal-monella Typhimurium, and then piglets were treatedwith oxytetracycline or apramycin (or not at all) [51].Resistance levels in the Salmonella isolates were not sig-nificantly effected by the treatment regimen, but resist-ance to both aparamycin and oxytetracycline wasreportedly highest in E. coli isolates from pigs thatnursed sows treated with oxytetracycline prior to farrow-ing. In yet another randomized controlled trial, pigletswere experimentally infected with Salmonella Typhimur-ium and exposed to aureomycin fourty-eight hours afterchallenge. Exposed pigs consistently shed statisticallysignificantly higher numbers of Salmonella than unex-posed controls but resistance was not reported for Sal-monella isolates. On the contrary, in anotherrandomized controlled trial pigs were experimentallychallenged with a Salmonella Typhimurium strain carry-ing a nalidixic acid resistance marker and then exposedto ceftiofur and oxytetracycline, aparmycin and oxytetra-cycline, or carbadox and oxytetracycline [53]. Resistancelevels varied over the study period and differed by anti-biotic treatment but for all treatments were highest onthe last day of the study. Notably, for aparamycin, carba-dox and ceftiofur, resistance continued to increase afterthe antimicrobial exposure was discontinued.Evidence from observational studies supports a risk of
resistance emergence in foodborne pathogens exposedto antimicrobial drugs on farms but also reinforces thecomplexity of the issue. For instance, one cohort studyduring the years 2000 and 2001 compared phenotypicresistance to various antimicrobial drugs in Salmonellaisolates from 69 conventional and 26 organic daries. Re-sults differed by antibiotic. Resistance to streptomycinand sulfamethoxazole was greater on conventional farmswhile for most antimicrobial drugs, resistance levelswere not significantly different [54]. In contrast, a cross-sectional study of Campylobacter resistance in insolatesfrom 30 organic dairies and 30 neighboring conventionaloperations failed to find a statistically significant differ-ence in resistance rates [55]. Yet another cross-sectionalstudy of Campylobacter resistance in conventional andorganic broiler chicken and turkey flocks found signfi-ciantly higher antimicrobial resistance rates in isolatesfrom conventional compared to organic farms [56]. Inaddition, isolates from conventional farms were signifi-cantly more likely to be multi-drug resistance, and resist-ance levels varied across drugs.
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 12 of 38
A variety of other study designs have also been used toinvestigate the issue and generally support a risk of re-sistance emergence after exposure, even though thestrength of evidene is significantly lower compared tothat from controlled trials or observational studies. Forinstance, one study investigated the prevalence of anti-microbial resistance in Campylobacter isolates from 5commercial broiler flocks before, during and after theywere exposed to difloxacin or enrofloxacin in the drink-ing water [57]. Even though this study lacks proper con-trol groups it was conducted in commercial flocks andtherefore provides a valuable complement to the con-trolled trials discussed above. The prevalence of resistantstrains increased significantly during treatment andremained elevated after treatment. Other studies havecompared rates of resistance across production systems[58, 59], over time periods that coincided with the intro-duction or ban of certain antimicrobials [60, 61], or both[62]. These studies generally support a risk of antimicro-bial resistance emergence after on-farm exposure. Adrop in vanA –type vancomycin resistance in Entero-cocci from poultry meat and health volunteers, for in-stance, coincided with a ban of avoparcin as growthpromoter [62], ceftiofur resistance levels in SalmonellaHeidelberg from retail chicken and human cases de-creased while ceftiofur use in hatcheries was tempor-arily discontinued and increased when it wasreinstated [63], and a rise in fluoroquinolone resistantCampylobacter coincided with the introductin ofenrofloxacin [61]. Notably, another study failed tofind a statistically significant drop in fluoroquinoloneresistance among Campylobacter isolates after theiruse was supposedly discontinued [59].Taken together, these studies provide evidence that
antimicrobial use on farms can and does lead to an in-crease in resistance among foodborne pathogens isolatedfrom the exposed animals. How quickly resistanceemerges seems to depend on many factors, including thepathogen species, strain and drug treatment of interest.In some instances, discontinuation of use may lead to arapid decrease in resistance; for example, a decrease invancomycin-resistant Enterococci (VRE) coincided with aban of avoparcin as growth promoter [60], and a de-crease in vancomycin resistance was demonstrated onindividual farms after the ban of antimicrobial growthpromoters [64]. In other cases, however, resistance levelshave remained increased for extended periods of timeafter antibiotic uses were discontinued. For example, ina longitudinal study of farms that discontinued fluoro-quinolone use as well as control farms, resistance ratesin Campylobacter isolates remained high for years afteruse was discontinued [59]. Similarly, vancomycin resist-ance in Enterococci isolates from pigs did not decreasein response to a discontinuation of use until the use of
tylosin, a macrolide, was also discontinued, because bothresistance traits were genetically linked [2]. Also notably,antimicrobial exposure does not always lead to a statisti-cally significant increase in resistance rates during thestudied time periods [54]. Similarly, resistance levels inbacteria associated with operations that do not use anti-microbial drugs are not in all cases significantly lowerthan those associated with operations that do [55].Therefore, while exposure of foodborne pathogens toantimicrobial drugs on farms or feedlots clearly poses arisk of resistance emergence that risk is currently diffi-cult to quantify.
Data based on studies in animal pathogens or commensalbacteriaBacteria can readily share their genetic material. Theemergence of antimicrobial resistance in commensalbacteria or animal pathogens contributes to the abun-dance of resistance genes that may subsequently betransferred to human pathogens. Because commensalbacteria are ubiquitous in the environment and theirstudy tends to be easier than that of foodborne or zoo-notic pathogens, considerably more studies have focusedon the impact of antimicrobial exposure on commensalbacteria than foodborne or zoonotic pathogens. Only afew representative studies are highlighted in Table 3, asillustrative examples. Additional examples are providedin other reviews, including that performed by the JointExpert Technical Advisory Committee on Antibiotic Re-sistance (JETACAR) [32].As shown in Table 3, a variety of study types are avail-
able that investigate the emergence of resistance in com-mensal bacteria in response to antimicrobial exposure.These include controlled trials, observational studies,and correlation studies that compare bacterial popula-tions across settings. Similar to the situation describedin the preceding section, most studies rely on pheno-typic methods to evaluate resistance levels, and studiesvary vastly in the amount of control they exert on treat-ment allocation and antimicrobial exposures. Becausethe primary focus is on commensal bacteria, studies typ-ically follow naturally-occurring bacterial populations,even though, as described in the preceding section, insome cases animals are experimentally challenged withpathogens and both pathogens and commensals arefollowed over time.A number of randomized controlled trials have dem-
onstrated an increase in the prevalence of antimicrobialresistant bacteria following antimicrobial exposure. Forinstance, in one study feedlot steers were fed chlortetra-cycline with or without sulfamethazine, monensin, tylo-sin or virginiamycin for 61 days in a silage-based diet,followed by 86 days without antimicrobials, and then an-other 42 days of exposure while being on a grain-based
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 13 of 38
Table
3Exam
ples
ofstud
ytype
sevaluatin
gtheem
erge
nceof
antim
icrobialresistance
innaturalb
acterialb
ackgroun
dpo
pulatio
nsafterfood
prod
ucinganim
alswereexpo
sed
toantim
icrobialdrug
s
Reference
Stud
yKeyfinding
s
Type
aspecies
bacterialtarge
tantim
icrobialdrug
administration
route
stud
you
tcom
esrelatedto
antim
icrobial
resistance
stud
yde
sign
suscep
tibility
testing
controls
Grade
level:Con
trolledtrial
[65]
RCT
cattle
E.coli
chlortetracycline(with
orwith
out
sulfamethazine
),mon
ensin,tylosin,
virginiamycin;allas
grow
thprom
oters
feed
emerge
nceof
antim
icrobial
resistance
6treatm
ents,5
pens
pertreatm
ent,
10steerpe
rpe
n
Phen
otypic
agar
dilutio
nne
gative
control
expo
sure
tochlortetracyclineand
sulfamethazine
increasedthe
prevalen
ceof
resistance;d
iethadan
impo
rtantim
pact
onshed
ding
ofresistantstrains
[66]
RCT
cattle
E.coli
florfe
nicol
s.c.injectio
nem
erge
nceof
antim
icrobial
resistance
42pe
ns(8–10cows
perpe
n),2
cows
perpe
nen
rolledin
stud
y
Phen
otypic
disk
diffusio
nne
gative
control
afterantim
icrobialtreatm
ent,increased
antim
icrobialresistance;p
rior
treatm
entpred
ictiveof
resistance;
anim
alsource
andprevious
managem
entim
pact
results
[67]
RCT
Cattle
fecalb
acteria
ceftiofur,tetracycline
s.c.injectio
nantim
icrobial
resistance
176steers;2
replicates
of88
steerseach,8
pens
of11
steerspe
rreplicate;totalo
f4treatm
entgrou
ps
Gen
otypic
PCRbased
positiveand
negative
controls
Initialandsubseq
uent
antim
icrobial
treatm
entsledto
increasedfre
quen
cyof
resistance
gene
s
[68]
RCT
pigs
Enterococci
Staphylococcus
hyicus
tylosin(asgrow
thprom
oter)
feed
erythrom
ycin
resistance
levels
2trialsa10
pigs,
n=5pe
rgrou
pPh
enotypic
negative
control
Prevalen
ceof
resistance
inen
terococci
immed
iatelyincreased2.4×
inrespon
seto
tylosinexpo
sure;resistance
inStaphylococcus
hyicus
occurred
more
gradually,atarate
of8%
perdayand
5×over
20days
[130]
RCT
pigs
aerobicgram
negative
bacteria;
Salmon
ella
chlortetracycline
feed
antim
icrobial
resistance
3farm
s(con
venien
cesample);12,2and8
barnspe
rfarm
Phen
otypic
broth
dilution
negative
control
antim
icrobialexpo
sure
was
associated
with
increasedprevalen
ceof
resistance
inaerobicgram
-neg
ativebacteria
GRA
DElevel:ob
servationalstudy
[69]
coho
rtpigs
coliform
bacteria
olaquind
oxfeed
emerge
nceof
resistance
12farm
sPh
enotypic
neighb
oring
farm
sthat
dono
tuse
olaquind
ox
prevalen
ceandlevelo
fresistance
increasedon
farm
sthat
used
it,andto
lesser
extent
onadjacent
farm
s
[70]
CS
pigs
E.coli
vario
usfeed
emerge
nceof
antim
icrobial
resistance
34farm
sPh
enotypic
negative
control
antim
icrobialuseincreasesthe
emerge
nceof
resistance
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 14 of 38
Table
3Exam
ples
ofstud
ytype
sevaluatin
gtheem
erge
nceof
antim
icrobialresistance
innaturalb
acterialb
ackgroun
dpo
pulatio
nsafterfood
prod
ucinganim
alswereexpo
sed
toantim
icrobialdrug
s(Con
tinued)
[71]
CS
pigs
E.coli
vario
usvario
usantim
icrobial
resistance
47farm
sPh
enotypic
negative
control
antim
icrobialuse(in
weane
rrather
than
finishe
rpigs)w
asassociated
with
resistance;som
eeviden
ceforcross-
resistance
andco-selectio
n
GRA
DElevel:othe
r
[72]
correlation
stud
ypigs,
cattle,
poultry
E.coli
vario
usvario
uscorrelation
betw
een
antim
icrobial
useand
resistance
7coun
tries
phen
otypic
comparison
across
coun
tries
useof
antim
icrobialdrug
sisstrong
lycorrelated
with
resistance
inE.coli
[73]
correlation
stud
yPo
ultry
Enterococci,
Vancom
ycin-
resistant
avop
arcin
vario
usCom
parison
ofresistance
prevalen
cebe
fore
and
after
avop
arcinban
Italy
phen
otypic
comparison
over
time
prevalen
ceof
vancom
ycin-resistant
Enterococcide
creasedin
poultrymeat
samples
afterthe
avop
arcinban
a CTcontrolledtrial(allocatio
nsche
meno
tclearly
describ
ed),RC
Trand
omized
controlledtrial,CS
cross-sectiona
lstudy
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 15 of 38
diet [65]. Expsoure to antimicrobial drugs significantlyincrased the prevalence of steers shedding resistantbaceria and the data suggested a potential genetic linkbetween tetracycline and ampicillin resistance. Notably,the type of feed significantly impacted shedding rateswith higher rates associated with the grain-based diet,emphasizing the importance of external factors. In an-other randomized controlled trial, feedlot steers were ex-posed to a single dose of florfenicol by subcutaneousinjection and resistance in fecal E. coli populations weremeasured before and immediately after exposure, and upto a month later [66]. Notably, treated and control ani-mals were housed in the same pen together with othersteers. Immediately after treatment, all isolates fromtreated animals were resistant to multiple antimicrobialdrugs including chloramphenicol, and resistance per-sisted in later samplings. The source of the animals andtheir previous management were found to have a statis-tically significant impact on the resistance dynamics. Yetanother study used a metagenomic approach [67]. Feed-lot steers were exposed to chlortetracycline – adminis-tering the drug either to one or all steers in the pen –followed by administration of chlortetracycline (or noantimicrobial drugs). Fecal bacterial communities wereanalyzed for common resistance genes (i.e., blaCMY-2,blaCTX-M, tet(A) and tet(B)) by real-time PCR. A deter-mination of the bacterial species carrying the resistancegenes was not performed but 16S rRNA gene quantitywas determined to standardize the number of resistancegene copies. The frequency of ceftiofur resistance genesincreased significantly when all steers received ceftiofurwhile tetracycline resistance decreased at the same time,likely indicating a shift in the composition of the micro-bial community. In response to subsequent chlortetra-cycline exposure both ceftiofur and tetracyclineresistance genes increased.Similar results have also been observed in pigs. In one
trial, for instance, pigs on 3 farms were exposed tochlortetracycline in the feed and resistance was evalu-ated in their aerobic gram negative bacteria as well asSalmonella. Antimicrobial exposure was associated withincreased prevalence of resistance in aerobic gram-negative bacteria. In another study, pigs were fed tylosinand weekly samples were taken to evaluate macrolide re-sistance in fecal enterococci and skin staphylococci [68].The prevalence of resistance in enterococci immediatelyincreased 2.4 times in response to tylosin exposure. Bycomparison, resistance in Staphylococcus hyicus oc-curred more gradually, at a rate of 8% per day and 5times over 20 days.As shown in Table 3, there is also strong evidence
from observational studies as well as the analysis of con-venience samples that antimicrobial use on farms leadsto an increase in resistance among commensals and
animal pathogens isolated from the exposed animals. Inone cohort study, for instance, the impact of olaquindoxexposure on coliform bacteria from pigs was analyzedand compared to resistance in coliform bacteria frompigs on neighboring farms not exposed to olaquindox[69]. Both the prevalence and level of resistance increasedon farms that used olaquindox, and to lesser extent on ad-jacent farms. Two cross-sectional studies of 34 and 47 pigfarms provide further evidence for a risk of resistance inresponse to antimicrobial exposure [70, 71]. Notably, oneof the studies provided some evidence for co-resistanceand cross-selection, and noted that antimicrobial use inweaners was associated with incrased resistance [71]. Inaddition, a study of E. coli convenience samples across 7countries revealed a correlation between antimicrobial useand resistance [72] and an analysis of Enterococci frompoultry samples collected before and after the avopar-cin ban found the prevalence of vancomycin-resistantEnterococci decreased in poultry meat samples afterthe avoparcin ban [73].Taken together there is ample evidence supporting a
risk of resistance emergence in commensal bacteria afterexposure to antimicrobial drugs. How quickly resistanceemerges is affected by a variety of factors, including thebacterial species and the antimicrobial treatment choiceas well as potential preceding antimicrobial exposures[66, 68]. A variety of external factors, such as the type offeed chosen, can also influence the emergence of resist-ance [65]. Resistance may be reversible [73], but can re-main at elevated levels after use is discontinued [47],and a single exposure can lead to increased resistance inthe commensal bacterial population [66]. Therefore,while exposure to antimicrobial drugs on farms or feed-lots clearly poses a risk of resistance emergence in com-mensal bacteria that risk is currently difficult toquantify.
Data based on studies in humans, laboratory animalmodels or in vitro systemsData on the correlation between antimicrobial use andthe emergence of antimicrobial resistance have also beencollected in human healthcare, laboratory animal modelsand in vitro systems. Results collected in these settingsare generally consistent with the observations made inagricultural settings. As shown in Table 4, a variety ofstudy types are available, including RCTs, observationalstudies, and other types of evidence. Many of the scien-tific studies have analyzed the emergence of resistance inhumans during treatment, monitoring the resistance ofbacteria isolated from the patients over time. In somecases, patients were randomly allocated to treatmentarms, whereas other studies compared isolates from pa-tients that received treatments with different antimicro-bial drugs or treatment regimen. Yet other studies
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 16 of 38
Table
4Exam
ples
ofstud
iesevaluatin
gtheem
erge
nceof
antim
icrobialresistance
inrespon
seto
antim
icrobialexpo
sure
Reference
Stud
yKeyfinding
s
Type
astud
ypo
pulatio
ndrug
ofconcern
bacterialtarge
ten
dpoint
orou
tcom
erelevant
toantim
icrobial
resistance
controls
perio
dof concernb
Grade
level:Con
trolledtrial
[74]
SRof
RCTs;
n=46
stud
ies
pneumon
iapatients
imipen
emPseudomon
asaerugino
sainitialresistance;
resistance
emerge
nceon
treatm
ent
othe
rantim
icrobials:
beta-lactam
s,am
ino-
glycosides,vanco-
mycin,
fluoroq
uino
lone
1993–
2008
resistance
emerge
d:im
ipen
em:38.7%
(rang
e:5.6–77.8%)of
isolates
comparatorgrou
ps:21.9%
(rang
e:4.8–
56.5%)
[75]
MAof
RCTs;
n=8stud
ies
patientswith
vario
usinfections
beta-lactam
vario
usem
erge
nceof
antim
icrobial
resistance
othe
rantim
icrobials:
beta-lactam
andam
i-no
glycoside
combinatio
n
1980–
2004
resistance
emerge
din
vario
usbacterial
species(upto
20%
ofisolates);no
sign
ificant
differencebe
tweenthetw
otreatm
entarmsin
resistance
emerge
nce
[85]
RCT;n=313
patients
patientswith
nosocomial
pneumon
ia,sep
sis,or
severe
diffu
sepe
riton
itis
imipen
emPseudomon
asaerugino
saem
erge
nceof
antim
icrobial
resistance
imipen
emplus
netilmicin
1988–
1992
resistance
toim
ipen
emoccurred
inmon
othe
rapy
(n=8)
andcombinatio
ntherapy(n
=13)patients
Grade
level:Observatio
nalstudy
[76]
SR&MAof
RCTs
andOS;
n=24
stud
ies
patientswith
vario
usinfections
andhe
althyvolunteers
vario
usvario
usresistance
emerge
nce
unde
rtreatm
ent
othe
rantim
icrobials
1955–
2009
prescriptio
nof
antib
ioticsin
prim
ary
care
isassociated
with
resistance
inbacteria
[77]
coho
rtstud
y;n=271
patients
hospitalized
patientswith
lab-
confirm
edPseudomon
asinfec-
tionwhe
reinitialisolatewas
suscep
tible
cefta
zidime
ciprofloxacin
imipen
empipe
racillin
Pseudomon
asaerugino
saem
erge
nceof
resistance
durin
gtreatm
ent
compare
patients
with
resistance
emerge
nceto
those
with
out
1994–
1996
resistance
emerge
din
10%
ofpatients
(n=28);riskdifferedby
antim
icrobial
drug
[78]
coho
rtstud
y;n=2641
patients
coronary
artery
bypass
graft
surgerypatients
Vario
us;sho
rtvs.
prolon
ged
antib
iotic
prop
hylaxis
Enterobacteriaceae
(cep
halosporin
resistant);Enterococci
(vancomycin
resistant)
emerge
nceof
resistance
durin
gprop
hylaxis
compare
patients
with
acqu
ired
antib
iotic
resistance
tothosewith
out
1993–
1997
Prolon
gedantib
iotic
prop
hylaxiswas
associated
with
anincreasedriskof
acqu
iredantib
iotic
resistance
[79]
CCstud
ypatientswith
lab-confirm
edbacterem
ia,resistant
Pseudomon
as
Pipe
racillin,
cefta
zidime,
imipen
em,
ciprofloxacin,
gentam
icin,
amikacin
Pseudomon
asaerugino
saod
dsof
previous
antib
iotic
treatm
ent
patientswith
lab-
confirm
edbacterem
iacaused
bysuscep
tible
Pseudomon
asstrains
1989–
1998
previous
expo
sure
toantim
icrobial
mon
othe
rapy
isassociated
with
increasedriskof
subseq
uent
resistance
tothat
antim
icrobial
Grade
level:Other
[80]
invivo
trial
miceexpe
rimen
tally
challeng
edwith
bacteria
amikacin,
ceftriaxone
,pe
rfloxacin
ortheircombinatio
n
Pseudomon
asaerugino
sa,Klebsiella
pneumon
iae,
resistance
emerge
nceafter
treatm
ent
othe
rantim
icrobials
(singlevs.
combinatio
ntherapy)
1986
resistance
emerge
din
respon
seto
treatm
ent;thefre
quen
cyof
resistance
emerge
ncevariedby
pathog
enand
antib
iotic
/combinatio
n
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 17 of 38
Table
4Exam
ples
ofstud
iesevaluatin
gtheem
erge
nceof
antim
icrobialresistance
inrespon
seto
antim
icrobialexpo
sure
(Con
tinued)
Enterobactercloacae,
Serratiamarcescens
[81]
invitrotrial
(labevolution
system
)
bacteriaexpe
rimen
tally
expo
sedto
antim
icrobial
colistin
(differen
ttreatm
ent
regimen
)
Acinetobacter
baum
annii
resistance
emerge
nceafter
expo
sure
negativecontrol
2007
extensiveresistance
emerge
ddu
ring
re-growth
(asearly
as6hafter
treatment)
[82]
Cross-national
database
stud
y(26
coun
triesin
Europe
)
antib
iotic
usein
outpatient
settings
(calculatedas
defined
daily
dose
per1000
inhabitants)andantib
iotic
resistance
rates
vario
usvario
uscorrelation
betw
een
antib
iotic
use
andresistance
across
coun
tries
n/a
1997–
2002
ratesof
antib
iotic
resistance
arehigh
erin
high
consum
ingcoun
tries
a SRsystem
aticreview
,MAmeta-an
alysis,R
CTrand
omized
controlledtrial,OSob
servationa
lstudy
,CCcase-con
trol
stud
ybFo
rexpe
rimen
tallab
oratorystud
iesthat
dono
tspecify
theda
teof
theexpe
rimen
tsthepu
blicationda
teissubstituted
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 18 of 38
followed bacterial populations in laboratory animals orin vitro systems, or evaluated correlations between anti-biotic use practices and the prevalence of resistance.A systematic review of randomized controlled trials in
pneumonia patients with initially sensitive Pseudomonasaeruginosa infection, for instance, found that resistanceemerged readily de novo in response to imipenemexposure, with an average of 38.7% of isolates (range:5.6–77.8%) acquiring resistance [74]. Notably, de novoresistance emergence was less likely in the comparatorgroup treated with other antimicrobials (i.e., beta-lactams, aminoglycosides, vancomycin, or fluoroquino-lone), where resistance emerged in an average of 21.9%of isolates (range 4.8–56.5%). Another meta-analysis ofhuman patients with various types of infections treatedwith beta-lactam antibiotics or a combination of beta-lactams and aminoglycosides found that resistancerapidly emerged de novo, with up to 20% of isolates be-coming resistant, and there was no statistically signifi-cant difference between the two treatments [75]. Resultsfrom a variety of observational studies as well as meta-analyses of individual observational studies also supporta risk of resistance emergence in response to antimicro-bial exposure [76–79]. Studies in laboratory animals andin vitro models have also demonstrated for a number ofdifferent bacteria and drugs that antimicrobial resistancereadily emerges in response to exposure, sometimes veryquickly, even though the dynamics of that resistance de-pend on the specific bacterium and drug [80, 81]. Simi-lar to the situation in animal agriculture [72], a cross-national study has demonstrated a correlation betweenantibiotic use patterns and resistance levels, showingthat rates of antimicrobial resistance are higher in high-consuming countries [82].Together, these data further support that antimicrobial
drug use can and does lead to the emergence of resistancein exposed bacteria. The data also suggest that in general,longer exposure periods and preceding antimicrobial ex-posure tend to be associated with an increased risk of re-sistance compared to short exposure times [78, 79, 83].However, how quickly resistance emerges depends on thepathogen, drug, and resistance mechanism, among otherfactors [80, 83]; exposure to a drug can also select for re-sistance to a related drug due to cross-selection, and someresistance may be maintained in the absence of antimicro-bial selection, for instance due to co-selection [84]. Thedata further suggest that certain treatment regimens maybe associated with an increased risk of resistance com-pared to others (see for instance [77, 78, 85]), but moredata are needed to allow the optimization of treatmentregimen to minimize the risk [83, 86] and data specific toanimal agriculture are needed to determine how to trans-late the information to optimize treatments in this regardin veterinary settings.
Risk of infection due to resistant bacteria that emergedon the farmTransmission of foodborne or zoonotic pathogens tohumansBecause of ethical considerations, RCTs that evaluatethe risk of pathogen transmission from animals tohumans are generally not feasible. However, there isconsiderable evidence from observational studies, as wellas other study types, such as outbreak investigations,case reports, and bacterial studies of correlation amongbacteria from different sources that foodborne or zoo-notic pathogens carrying antimicrobial resistance genesare shared between animals and humans (Table 5). Manyof the studies have focused on Salmonella ormethicillin-resistant Staphylococcus aureus (MRSA), andmany are based on case-control studies that comparerisk factors for infections with resistant or susceptibleisolates retrospectively. In at least some studies, the dir-ectionality of transfer remains unknown; and for at leastsome pathogens and settings, cross-species transfersmay be relatively infrequent. Moreover, in some instancethe study analyzed colonization with a potential humanpathogen such as MRSA, rather than actual infectionthat led to clinical disease. While colonization with thesepotential pathogens clearly demonstrates transmission,the associated public health risk is less clear.A number of observational studies have evaluated the
human health risk posed by infections with antimicrobialresistant foodborne or zoonotic pathogens. A retrospect-ive case-control study in New England (U.S.), for in-stance, evaluated risk factors for infections with amultidrug-resistant Salmonella Newport strain (MDRAmpC), comparing 34 case patients to 37 controls in-fected with susceptible Salmonella Newport strains and94 healthy community controls [87]. Infections with themultidrug-resistant Salmonella Newport strain were sig-nificantly associated with dairy farm exposure.(Adjusted odd ratio 12.2; 95% Confidence Interval:
1.2–640). Investigation of two regional dairy farmswhere the Salmonella Newport strain had recently beenidentified detected the strain in animals and ill farmworkers. In fact, bacterial isolates from humans and cat-tle were indistinguishable or closely related based onantibiotic resistance profiles and PFGE patterns. Anothercase-control study, also focusing on Salmonella NewportMDR AmpC in the U.S. but national in scope and com-paring 54 case patients to 146 controls infected withpan-susceptible Salmonella Newport strains and 1154healthy community controls, identified consumption ofuncooked ground beef or home prepared runny scram-bled eggs or omelets during the 5 days prior to illnessonset as a significant risk factor [88]. A third case-control study of Salmonella Newport MDR AmpC, com-paring 268 cases from Wisconsin to 402 controls from
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 19 of 38
Table
5Exam
ples
ofstud
ytype
sevaluatin
gthehu
man
health
riskpo
sedby
infections
with
antim
icrobialresistantfood
borneor
zoon
oticpathog
ens
Reference
Stud
yKeyfinding
srelevant
toantim
icrobialresistance
Type
asubjects
coun
try
time
perio
dbacterialtarge
trelevant
stud
you
tcom
es
samplesize
controls
commen
ts
GRA
DElevel:Con
trolledtrial
Not
applicablebe
causeof
ethicalcon
cerns
GRA
DElevel:ob
servationalstudy
[87]
CC
human
patients
U.S.
1999–2001
Salmon
ella
New
port;
MDRA
mpC
resistant
riskfactors
forinfection
34cases;37
infected
controlsand94
commun
itycontrols
Salmon
ella
patientswith
suscep
tible
New
port
infection;
healthy
commun
itycontrols
followed
upby
investigation
ofcattleisolates
andrisk
factorsforshed
ding
ondairy
farm
s;region
alin
scop
e(New
England)
Associatio
nof
infectionwith
Salmon
ellaNew
port
MDRA
mpC
with
dairy
farm
expo
sure;b
acterialisolates
from
humansandcattlewere
indistingu
ishableor
closely
related(based
onantib
iotic
resistance
profile
andPFGE
pattern)
[88]
CC
human
patients
U.S.
2002–2003
Salmon
ella
New
port;
MDRA
mpC
resistant
riskfactors
forinfection
215case
patients;54
with
MDRA
mpC
resistantstrain
and146
pansusceptible;
1154
commun
itycontrols
healthy
commun
itycontrols;
volunteers
with
out
diarrhea
intheprevious
28days
Food
Net
stud
y;natio
nalin
scop
e(8
of9Food
Net
sites)
Associatio
nof
infectionwith
Salmon
ellaNew
port
MDRA
mpC
with
consum
ption
ofun
cooked
grou
ndbe
efor
homeprep
ared
runn
yscrambled
eggs
orom
elets
durin
gthe5days
before
onset
ofillne
ss
[89]
CC
human
patientsin
Wisconsin
U.S.
2003–2005
Salmon
ella
New
portMDR
AmpC
associations
betw
een
antim
icrobial
resistance
andrepo
rted
expo
sures
268isolates
from
Wisconsin
and402fro
mtheremaining
U.S.
Cases
inWisconsin
vs.
restof
the
U.S.
Con
trolswerecollected
2003–
2004
Infections
with
multid
rug-
resistantSalmon
ellaNew
port
strainswereassociated
with
expo
suresto
cattle,farmsand
unpasteurized
milk
[90]
CC
human
patientsin
California
U.S.
1985
Salmon
ella
New
port
(chloram
phen
icol
resistant)
riskfactors
forinfection
45patientsand
89matched
controls
healthy
volunteers
Epidem
icSalmon
ellastrain
was
isolated
from
hambu
rger
prod
uctseatenby
cases,as
wellastheabattoirs
whe
rethe
anim
alswereslaugh
tered,
dairies
that
sent
theanim
alsto
slaugh
ter,andilldairy
cows.
Associatio
nof
infectionwith
resistantSalmon
ellaNew
port
isolates
with
eatin
ggrou
ndbe
efdu
ringtheweekbe
fore
onsetandpe
nicillinand
tetracyclineusedu
ringthe
mon
thbe
fore
onsetof
symptom
s;chloramph
enicol
resistantSalmon
ellafro
mdairy
farm
swereassociated
with
chloramph
enicol
useon
those
dairies
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 20 of 38
Table
5Exam
ples
ofstud
ytype
sevaluatin
gthehu
man
health
riskpo
sedby
infections
with
antim
icrobialresistantfood
borneor
zoon
oticpathog
ens(Con
tinued)
[91]
CC
human
patients
from
natio
nal
database
Nethe
rland
s2003–2005
MRSA(non
-typable)
riskfactors
forinfection
35cases,76
controls
MRSA
infections
with
othe
rstrains
Analysisaimed
toexclud
esecond
arycases
Living
inruralareas
and
contactwith
pigs
andcattle
wereassociated
with
case
patients
[92]
CS
vealcalf
farm
ers
Nethe
rland
s2007–2008
MRSAST
398
riskfactors
forcarriage
102farm
s,390
questio
nnaires,
2151
nasal
swabsfro
mcalves
farm
ers
negativefor
MRSAST
398
nasalswabstakenfro
mfarm
ersandfarm
calves;farms
rand
omlyselected
Hum
anMRSAST
398carriage
was
associated
with
intensity
ofanim
alcontactandnu
mbe
rof
positiveanim
alson
farm
;calves
weremorelikelyto
becarrierswhe
ntreatedwith
antib
iotics,andgo
odfarm
hygien
ehadaprotective
effect.
[93]
CS
swine
farm
ers
Belgium
2007
MRSAST
398
riskfactors
forcarriage
49sw
inefarm
s,127pe
rson
son
thefarm
sparticipated
and
1500
pigs
were
sampled
farm
ers
negativefor
MRSAST
398
nasalswabsandwou
ndsamples;farmsrand
omly
collected
MRSAST
398carriage
byfarm
erswas
associated
with
prevalen
ceam
ongpigs
onthe
farm
,havingregu
larcontact
with
pigs,d
ogs,andho
rses,
anduseof
protective
equipm
ent.
GRA
DElevel:othe
r
[22]
phylog
enetic
stud
ybacterial
isolates
vario
us1997–2011
MRSACC398
evolution
and
transm
ission
dynamics
includ
ing
transm
ission
betw
een
species
n/a
none
Phylog
eneticandpo
pulatio
nge
netic
analysis
human
andanim
alpo
pulatio
nsof
CC398are
separate,but
with
multip
letransfersback
andforth;same
data
sugg
estin
gon
ward
transm
issionof
animal-
associated
strainsincommun
itysettings
(e.g.,ho
spitals);different
tetracyclineresistancegenes
show
eddifferent
evolutionary
patterns
[94]
outbreak
human
patients
with
rare
strain
Den
mark
1998
Salmon
ella
Typh
imurium
DT104
source
ofmultid
rug
resistant
infection
outbreak
27cases
none
rare
strain
isolated
from
patientsandpo
rksamples;
pork
samples
weretraced
back
tohe
rd;resistant
infections
inhu
mansweremoredifficultto
treat
Asw
inehe
rdwas
thelikely
source
ofahu
man
outbreak
with
MDRSalmon
ella
Typh
imurium
DT104
[95]
outbreak
human
patients
infected
with
rare
strains
U.S.
1983
Salmon
ella
New
port
source
ofmultid
rug
resistant
infection
outbreak
18cases
none
rare
isolatestrain
foun
din
human
patientsandcattle
hambu
rger
from
beef
cattle
fedsubtherapeutic
chlortetracyclineforgrow
thprom
otionlikelyou
tbreak
source
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 21 of 38
Table
5Exam
ples
ofstud
ytype
sevaluatin
gthehu
man
health
riskpo
sedby
infections
with
antim
icrobialresistantfood
borneor
zoon
oticpathog
ens(Con
tinued)
[96]
human
case
repo
rthu
man
patient
infected
with
rare
strain
U.S.
1998
Salmon
ella
Typh
imurium
source
ofmultid
rug
resistant
infection
1case
none
Infectiontraced
back
tofarm
;Thefarm
was
iden
tifiedas
the
source
ofamultid
rugresistant
Salmon
ellaisolatefro
ma
human
case
patient
[97]
human
case
repo
rtMRSA
cluster
arou
ndpigfarm
er
Nethe
rland
s2004
MRSA
source
ofmultid
rug
resistant
infection
pigfarm
er,
family
andco-
workers,p
igs
none
anim
alson
farm
sampled
Infectionwith
MRSAtraced
back
tofarm
,onw
ard
transm
ission
infarm
family
[98]
correlation
stud
yIsolates
from
swineand
swine
workers
U.S.
Unspe
cified
MRSA
riskfactors
forMRSA
colonizatio
n
20workersand
299sw
ine
none
anim
alsandhu
manssampled
Colon
izationwith
MRSAST
398was
very
common
insw
inehe
rds,similarisolates
inhu
mansandanim
als
[99]
correlation
stud
yisolates
from
farm
person
nel
andmilk
samples
Hun
gary
2002–2004
MRSA
riskfactors
forMRSA
colonizatio
n
1farm
,12
workers
sampled
,595
milk
samples
none
MRSAstrainsweresubtyped
MRSAisolates
from
workersin
closecontactwith
cowswere
indistingu
ishablefro
mcow
isolates
a CCcase-con
trol
stud
y(orcase-casestud
ywhe
recontrolsarecasesof
infectionwith
othe
rstrains),C
Scross-sectiona
lstudy
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 22 of 38
the remaining U.S. found infections with SalmonellaNewport MDR AmpC significantly associated with ex-posure to cattle, farms and unpasteurized milk [89]. Yetanother case-control study of 45 case patients infectedwith Salmonella Newport strains carrying chlorampheni-col resistance and 89 matched healthy controls identifiedeating ground beef during the week before illness onsetand penicillin and tetracycline use during the month be-fore onset of symptoms as significant risk factors [90].Notably, the same Salmonella strain was isolated fromhamburger products eaten by case patients, as well asthe abattoirs where the animals were slaughtered, dairiesthat sent the animals to slaughter, and ill dairy cows.Chloramphenicol resistant Salmonella from dairy farmswere associated with chloramphenicol use on those dair-ies. In addition, case-control [91] and cross-sectional[92, 93] studies in the Netherlands and Belgium haveidentified animal contacts as risk factors for carriage orinfections with MRSA. In one study of veal calf farmersin the Netherlands, human MRSA ST 398 carriage wasassociated with intensity of animal contact and numberof positive animals on farm; calves were more likely tobe carriers when treated with antibiotics, and good farmhygiene had a protective effect [92]. In another study ofswine farmers in Belgium, MRSA ST 398 carriage byfarmers was associated with prevalence among pigs onthe farm, having regular contact with pigs, dogs, andhorses, and use of protective equipment [93]. A case-control study of human infections with non-typableMRSA in the Netherlands, comparing 35 cases to 76controls based on a national database of human patients,identified living in rural areas and contact with pigs andcattle as significantly associated with case patients [91].In addition, Salmonella outbreaks have been traced
back to farm animals. For instance, an outbreak ofmultidrug-resistant Salmonella Typhimurium DT104that occurred in Denmark in 1998 was traced back to aswine herd [94]. The outbreak strain was a rare strain,simplifying the trace-back. In fact, the same rare strainwas isolated from patients and pork samples and porksamples were traced back to the swine herd of origin.Notable, the resistant infections in humans were moredifficult to treat due to the resistance profile. In anotheroutbreak involving a rare strain of Salmonella Newportthat occurred in the U.S., hamburgers from beef cattlefed chlortetracycline were identified as the likely out-break source [95]. Trace-back investigations revealed thesame rare Salmonella strain found in human patientswas shed by cattle. Human case reports have identifiedfarm exposures as the likely source of infections withmultidrug resistant Salmonella Typhimurium in the U.S.[96] and MRSA in the Netherlands [97]. In fact, in thelatter case infection with MRSA was traced back a pigfarm, where it was isolated from the pigs and co-
workers, and onward transmission to the farm familywas documented [97]. Correlation studies and phylogen-etic analyses also provide evidence supporting a transmi-sison of MRSA between animals and human contacts[22, 98, 99]. In Canada, the rate of ceftiofur-resistantSalmonella Heidelberg infections was statistically signifi-cantly associated with contamination rates on retailchicken [63].Taken together, the evidence for a link between food-
borne or zoonotic bacteria with antimicrobial resistancegenes on farms or feedlots and human health risks is un-disputable. Exposure to farms, consumption of raw orundercooked foods of animal origin, and living in ruralareas have been identified as risk factors for infectionswith those foodborne pathogens that carry antimicrobialresistance genes [87, 88, 90, 91]. Outbreaks of foodborneillness associated with resistant bacteria have been tracedback to the farm of origin [94–96]; identical bacterialstrains have been identified in animals and humans incontact with them or the food they produce [97–99].And in at least some cases, increased intensity of animalcontact led to a measurably increased risk of human in-fection with resistant strains [92, 93]. While the magni-tude of the risk may vary across pathogens andcircumstance, it is clear that foodborne or zoonotic bac-teria on farms or feedlots that carry resistance genespose a human health risk.
Transmission of commensal bacteria from food-producinganimals to humansMost of the research on the emergence of resistance as aresult of on-farm use of antimicrobial drugs has beenconducted on commensal bacteria, and observationsfrom studies which simultaneously analyzed resistanceemergence in foodborne as well as commensal bacteriasuggest that resistance may emerge more quickly in thelatter. The question of whether the emergence of resist-ance among commensal or animal pathogens on farmsposes a human public health risk is therefore an import-ant one.As shown in Table 6, a variety of studies have evalu-
ated the link between antimicrobial resistance in animalcommensals and human health risks. The available evi-dence types include observational studies and otherstudy designs such as correspondence of bacterial strainsfrom different sources. Notably, because of the complex-ity of the events evaluated, study designs are often intri-cate and may include, for instance, a randomizedcontrolled trial of chicken exposed to antimicrobials,paired with an observational study evaluating thecolonization of their human caretakers [100]. Whilesome studies choose prevalence of resistance as studyoutcome, others also include degree of resistance,expressed for instance as MIC. Farm workers as well as
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 23 of 38
Table
6Exam
ples
ofstud
ytype
sevaluatin
gthehu
man
health
riskpo
sedby
antim
icrobialresistance
gene
sthat
emerge
din
anim
alpathog
ensor
commen
salb
acteria
Reference
Stud
yKeyfinding
srelevant
toantim
icrobial
resistance
Type
asubjects
coun
try
time
perio
dbacterialtarge
trelevant
stud
you
tcom
essamplesize
controls
commen
ts
GRA
DElevel:Con
trolledtrial
Not
available
GRA
DElevel:ob
servationalstudy
[100]
CC
poultryworkers
Nigeria
publishe
din
1989
E.coli,nalidixic
acid
resistant
colonizatio
nof
poultry
workerswith
challeng
eE.
colistrain
Com
plex
stud
yde
sign
:-Birdswereexpe
rimen
tally
inoculated
with
challeng
estrain
ofE.
coli(n
=36
birdson
university
farm
,and
n=16
oncommercial
farm
)-Po
ultryworks
indirect
contactwith
challeng
edbirdswere
sampled
forcolonizatio
nwith
challeng
estrain
-Thiswas
comparedto
controlw
orkerswith
orwith
outdirect
contactto
controlb
irds,andto
samples
collected
from
theexpo
sed
workerspriorto
expo
sure
-birdsweresampled
forcolonizatio
n
afterbirdswere
challeng
ed,the
poultryworkersin
direct
contactwith
challeng
edbirdswere
colonized;
results
weresimilaron
the
university
and
commercialfarm
[101]
CC
poultryfarm
ers
U.S.
publishe
din
1978
aerobic
bacterial
cultu
res,
tetracycline
resistance
resistance
inbacterial
microflora
ofchickenand
contact
farm
ers
Com
plex
stud
yde
sign
:-RC
Tof
chicken,n=50
chickenpe
rgrou
p,testchickenwere
expo
sedto
tetracyclinein
feed
,con
trolswereno
t;em
erge
nceof
resistance
was
tested
incommen
salb
acteria
-farm
family
incontactwith
chickenwas
tested
forem
erge
nceof
resistance
inbackgrou
ndmicroflora
-farm
family
was
comparedto
othe
rfarm
families
inproxim
ity,and
tomed
icalstud
ents
antim
icrobial
resistance
emerge
din
expo
sedchickenand
farm
contacts
[102]
coho
rtpo
ultryfarm
ers
U.S.
publishe
din
1976
intestinal
bacterialflora,
tetracycline
resistance
resistance
inbacterial
microflora
ofchickenand
contact
farm
ers
n=11
farm
mem
bers,24
neighb
ors
neighb
ors
birdswereexpo
sed
totetracyclinein
feed
,emerge
nceof
resistance
was
traced
inbirdsandcontact
humans,and
comparedto
neighb
ors
antim
icrobial
resistance
emerge
dun
derexpo
sure
onthefarm
;the
prevalen
ceof
antim
icrobial
resistance
was
high
erin
bacteriafro
mexpo
sedfarm
families
than
inne
ighb
ors;
[103]
CS
pigfarm
ersand
abattoirworkers
Nethe
rland
spu
blishe
din
1994
E.coli
Prevalen
ceof
resistance
n=290pigfarm
ers;
n=316abattoir
workers;n
=160
urban/subu
rban
reside
nts
urban/subu
rban
reside
nts
fecalE.colifrom
the
threehu
man
grou
psweretested
for
resistance
tomultip
leantim
icrobialdrug
s
Pigfarm
ershadthe
high
estpe
rcen
tage
ofresistantE.coli,and
urban/subu
rban
reside
ntshadthe
lowest
[104]
CC
turkey,b
roiler
andlayerfarm
ers
andslaugh
terers
Nethe
rland
spu
blishe
din
2001
E.coli
Prevalen
ceandde
gree
ofresistance
n=47
turkey
farm
ersandtheir
flocks;n=51
broiler
farm
ersandn=50
broilerflocks;25
layerfarm
ersand
comparison
across
human
(and
correspo
nding
anim
al)po
pulatio
ns;
ciprofloxacin-resistant
isolates
werefurthe
rsubtyped
;meat
humansandbirds
sampled
;antibiotic
useon
farm
srecorded
Prevalen
ceof
resistance
sign
ificantly
high
erin
turkey
and
broilersamples
than
laying
hens
(which
correlated
with
antib
iotic
use);
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 24 of 38
Table
6Exam
ples
ofstud
ytype
sevaluatin
gthehu
man
health
riskpo
sedby
antim
icrobialresistance
gene
sthat
emerge
din
anim
alpathog
ensor
commen
salb
acteria
(Con
tinued)
theirflocks;n=46
poultryslaugh
terers
samples
werecol-
lected
immed
iately
afterslaugh
ter
prevalen
ceof
resistance
was
high
erin
turkey
andbroiler
farm
ersand
slaugh
terersthan
inlaying
henfarm
ers;
isolates
from
farm
ers/
slaugh
terersandbirds
/meatseem
edto
match,d
espite
variabilityacross
farm
s
GRA
DElevel:othe
r
[63]
correlation
stud
yisolates
from
chickenmeatand
humans
Canada
2003–
2008
Salmon
ella
Heide
lberg,
E.coli
Prevalen
ceof
ceftiofur-
resistance
inchickensam-
ples
andhu
-man
cases
Correlatio
nacross
provincesandtim
epe
riods
n/a
n/a
Statisticallysign
ificant
correlationbe
tween
ceftiofur-resistan
tSalm
onella
Heidelberg
onretailchickenand
human
infections;in
Quebec,changing
levelsofceftiofur
usein
hatcheriesdu
ringstudy
perio
dseem
edto
impa
ctdynamics
from
before
voluntary
with
draw
alto
reintrod
uctio
n.
[105]
uncontrolled
transm
ission
stud
y
farm
ers,livestock
anden
vironm
ent
U.S.
publishe
din
1990
E.coli
Spread
ofresistant
bacteriaon
farm
;em
erge
nceof
resistance
2trialswith
n=2
cattleeach;n
=4
pigs,n
=5micepe
rcage
;bovine
expe
rimen
tsrepe
ated
inindo
or/
outdoo
rsettings
onechalleng
edpig
exchange
dwith
one
unchalleng
edpig
from
unchalleng
edpe
n
N/A;study
mon
itored
contactanim
als,mice
inpe
nsof
challeng
edandno
n-challeng
edpigor
cow,enviro
n-men
tinclud
ingflies,
andhu
man
caretakers
stud
yalso
evaluated
impact
ofchlortetra-
cyclineuseon
emer-
genceof
resistance
onepigandon
ecow
wereinoculated
with
markedstrains
ofE.coliwith
resistance;con
tact
anim
als(pigsor
cows
andmice)
were
sampled
fortheE.
colistrain,aswere
human
caretakers
andtheen
vironm
ent
Con
tact
anim
als,
mice,flies
and
caretakersexcreted
challeng
estrain
ofE.
coli;leng
thof
colonizatio
nvaried,
butin
severalcases
exceed
ed4weeks;E.
colistrain
was
foun
din
environm
entand
housingsystem
impacted
spread;
chlortetracyclineuse
ledto
increased
resistance;transferof
resistance
plasmid
toothe
rbacteriano
tde
tected
[106]
US
E.coli
1trial
N/A
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 25 of 38
Table
6Exam
ples
ofstud
ytype
sevaluatin
gthehu
man
health
riskpo
sedby
antim
icrobialresistance
gene
sthat
emerge
din
anim
alpathog
ensor
commen
salb
acteria
(Con
tinued) un
controlled
transm
ission
stud
y
Con
sumer
hand
lingchicken
carcasses
publishe
din
1977
spread
ofresistant
bacteriafro
mcontam
inated
carcassto
volunteer
hand
lingit
Bacterialstrainwith
antib
iotic
resistance
transferredfro
mchickencarcassto
human
volunteer
hand
lingit
[107]
Con
jugatio
nstud
yTransfer
ofresistance
gene
sam
ongbacteria
ofdifferent
origin
Norway
Publishe
din
1994
E.colias
well
asVibrio,
Aeromon
as
Con
jugatio
nfre
quen
cies
across
bacteria
and
environm
ental
settings
Con
jugatio
nwas
stud
iedin
multip
leen
vironm
ents(e.g.,
seaw
ater,hand
towel,m
eat)
N/A
Plasmidswerereadily
transferredfro
mon
ebacterium
toanothe
r,across
bacterial
speciesandin
anu
mbe
rof
different
environm
ents
[108]
Con
jugatio
nstud
yTransfer
ofresistance
gene
sbe
tween
Enterobacteriaceae
UK
Publishe
din
2003
CommensalE.
coli,E.coil
O157,
Salmon
ella
Con
jugatio
nacross
bacterial
species
Con
jugatio
nacross
anu
mbe
rof
bacterialp
airswas
evaluatedun
der
cond
ition
sin
the
gut(i.e.,ileum
)
N/A
Antibiotic
resistance
gene
swerereadily
transferredfro
mon
ebacterialstrainto
anothe
r
[109]
Con
jugatio
nstud
yTransfer
ofresistance
gene
sam
ong
Enterococciinthe
mou
segu
t
France
Publishe
din
2003
Enterococcus
faecium
isolates
from
human
san
dpigs
Con
jugatio
nacross
Enterococcio
fdifferent
origin
Inthegu
tof
mice
N/A
micewere
gnotob
iotic
(i.e.
reared
unde
rcond
ition
sso
that
colonizatio
nwith
bacteriaisfully
know
n)
Vancom
ycin
resistance
was
readily
transferredfro
mpo
rcineto
human
isolates
inthegu
tof
mice;tylosinexpo
sure
(throu
ghdrinking
water)favored
colonizatio
ndu
eto
conjug
ation.
[110]
Con
jugatio
nstud
yTransfer
ofresistance
gene
sbe
tween
Enterococcistrains
Swed
enPu
blishe
din
2006
Enterococci
faecium
ofhu
man
and
animalorigin
Con
jugatio
nacross
Enterococcio
fdifferent
origin
Invitroandin
the
gutof
mice
N/A
Micewerege
rm-free
Vancom
ycin
resistance
was
readily
transferredam
ong
Enterococcio
fdifferent
origin;
conjug
ation
frequ
ency
was
high
erin
themou
segu
tthan
inthe
environm
ent;in
most
casesresistance
disapp
earedwith
in3days
buton
eof
the
bacterialstrains
persistedformore
than
20days
with
out
antib
iotic
selection
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 26 of 38
Table
6Exam
ples
ofstud
ytype
sevaluatin
gthehu
man
health
riskpo
sedby
antim
icrobialresistance
gene
sthat
emerge
din
anim
alpathog
ensor
commen
salb
acteria
(Con
tinued)
[111]
Con
jugatio
nstud
yTransfer
ofresistance
gene
sin
thegu
tof
human
volunteer
Den
mark
Publishe
din
2006
Enterococcus
faecium
Con
jugatio
nbe
tween
Enterococcus
faecium
ofchickenand
human
origin
n=6volunteers
n/a
Inthegu
tof
human
volunteers
Resistance
gene
swerereadily
transferredfro
mchickento
human
isolates
inthegu
tof
human
volunteers;in
onevolunteer,
additio
nalresistance
gene
swere
transferred
[112]
Con
jugatio
nstud
yTransfer
ofresistance
from
Klebsiella
toE.coli
inthemou
seintestine
Den
mark
Publishe
din
2008
Klebsiella
andE.
coli
Con
jugatio
nbe
tween
Klebsiella
and
E.coli
Invitroandin
the
gutof
mice
n/a
Somemicewere
expo
sedto
antim
icrobial
treatm
ent
Resistance
gene
swerereadily
transferredin
vitro
andin
vivo.
Antim
icrobial
treatm
entselected
for
resistantstrains,
which
rapidly
disapp
earedin
the
gutsof
miceno
texpo
sedto
antib
iotics.
[113]
uncontrolled
transm
ission
stud
y
Calvesand
humansin
contactwith
them
US
publishe
din
1978
E.coli
Spread
ofresistant
bacteriafro
mcalves
tohu
man
contacts,and
impact
oftetracycline
expo
sure
onrisk
1trialrep
eated3
times
N/A
calves
were
inoculated
with
rare
E.colistrain
carrying
resistance
gene
s
Bacteriawere
transferredfro
mcalves
tohu
man
contacts;no
statisticallysign
ificant
impact
oftetracycline
expo
sure
ontransm
ission
risk;
expo
sure
ofcalves
totetracyclinedidno
tresultin
statistically
sign
ificant
differences
inresistance
levelsof
bacteriain
human
volunteers
[114]
correlation
stud
yCattle
andhu
man
contacts
Norway
1996
E.coli
Presen
ceof
drug
-resistant
E.coliin
ani-
malsandhu
-man
contacts
n=13
cattle;n
=3
family
mem
bers;
n=1veterin
arian
(sam
plingof
humansrepe
ated
after1year,and
samples
from
4othe
rveterin
arians
adde
d)
N/A
stud
yof
concurrence
ofE.colistrains
Multi-drug
resistantE.
coliwerefoun
din
the
anim
alsandhu
man
contacts.
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 27 of 38
Table
6Exam
ples
ofstud
ytype
sevaluatin
gthehu
man
health
riskpo
sedby
antim
icrobialresistance
gene
sthat
emerge
din
anim
alpathog
ensor
commen
salb
acteria
(Con
tinued)
[115]
correlation
stud
yPo
ultryfarm
ers
andtheirbirds
Norway
1998
Enterococcus
faecium
(VRE)
Gen
etic
relatedn
essof
human
and
anim
alisolates
n=5farm
ersand
n=7broiler
chicken
N/A
stud
yof
concurrence
ofE.colistrains
Onon
eof
thefarm
s,hu
man
andanim
albacteriawere
gene
ticallyclosely
related;
gene
tically
unrelatedstrains
shared
related
vancom
ycin
resistance
gene
s,sugg
estin
gHGT
[116]
correlation
stud
yfarm
families
and
theiranim
als(i.e.,
cattleor
swine)
US
Publishe
din
1975
E.coli
Gen
etic
relatedn
essof
human
and
anim
alisolates
n=14
farm
families
N/A
Farm
families
inMissouri;im
pact
offactorssuch
asanim
alcontact
Freq
uencyof
anim
alcontactwas
not
sign
ificantly
correlated
with
gene
ticrelatedn
essof
human
andanim
alisolates;consum
ptionof
home-raisedbe
efap-
pearsto
beassociated
with
concordance
betw
eenhu
man
and
anim
alisolates
[118]
Con
jugatio
nstud
yTransfer
ofresistance
gene
sfro
mE.colifro
mpigto
thosefro
mhu
man
gut
Den
mark
2007–
2008
E.coli
Con
jugatio
nbe
tween
resistance
gene
sam
ong
E.coliof
pig
andhu
man
origin
n=9hu
man
volunteers;
n/a
dono
rstrain
ofE.coli
from
Danishpig;
recipien
tstrain
E.coli
ofhu
man
origin
Transfer
ofresistance
gene
sbe
tweenE.coli
bacteriain
human
intestine
[119]
Con
jugatio
nstud
yTransfer
ofmob
ilege
netic
elem
ents
associated
with
MRSA
UK
Publishe
din
2014
Staphylococcus
aureus
bacterial
popu
latio
ns
Freq
uencyof
horizon
tal
gene
transfer
ofpigand
human
origins
Gno
tobioticpiglets
n/a
Co-colonizatio
nwith
human
andpig
associated
variantsof
MRSA
Highfre
quen
cyof
horizon
talg
ene
transfer;transferof
mob
ilege
netic
elem
entsfro
mpigto
human
with
in4hof
co-colon
ization
[120]
Phylog
enetic
stud
yPresen
ceof
tetQ
resistance
gene
indifferent
bacteria
US
Publishe
din
1994
Bacteroides
Sequ
ence
comparison
for
tetQ
resistance
gene
n/a
n/a
Presen
ceof
iden
tical
orcloselyrelated
resistance
gene
sacross
distantly
relatedbacteria
Virtually
iden
tical
resistance
gene
sfoun
din
different
bacterialspe
cies
and
isolates
ofdifferent
geog
raph
icorigin
a CCcase-con
trol
stud
y(orcase-casestud
ywhe
recontrolsarecasesof
infectionwith
othe
rstrains),C
Scross-sectiona
lstudy
,ETexpe
rimen
taltria
l(i.e.,no
rand
omizationan
dno
controls)
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 28 of 38
workers in slaughterhouses tend to be the primary popula-tion subgroups of concern in these studies, even thoughconsumers handling potentially contaminated carcasses,as well as animals in contact with infected animals havealso been considered in some studies [100–106]. Studiesalso vary in the bacterial population of concern and rangefrom narrowly defined, for instance E. coli carrying nali-dixic acid resistance [100] to broad, for example anaerobicbacterial populations in general [107]. Conjugation studiesare a special study type relevant in this context, whichevaluates the transfer rates of bacterial resistance genesbetween bacterial populations in vivo, in the gut of ani-mals or human volunteers [108–112], or in vitro [107].A number of observational studies have investigated
the effect of antimicrobial exposures on farms on resist-ance in commensal bacteria isolated from farm animalsand human contacts. In one cohort study in the U.S., forinstance, when broiler chicken were exposed to tetracyc-line in the feed, the emergence of tetracycline resistancewas traced in the intestinal bacteria isolated from thebirds, and in commensal bacteria isolated from 11 mem-bers of the farms where the exposed birds were housed[102]. Resistance to tetracycline in the intestinal bacteriafrom farm members in contact with the exposed birdswas compared to that of 24 neighbors. Antimicrobial re-sistance emerged under exposure on the farm and theprevalence of antimicrobial resistance was higher in bac-teria from exposed farm families than in neighbors. Inanother study of poultry farmers in the U.S. a random-ized controlled trial design was combined with a case-control study design [101]. In the randomized,controlled part of the study, groups of 50 chicken wereexposed to tetracycline in feed while controls did not re-ceive any antimicrobial drugs. The emergence of resist-ance in the commensal bacteria on the chicken wastracked. At the same time, the commensal bacteria offarm families in contact with the chicken was tested forthe emergence of resistance, and this was compared tothe commensal bacteria from other farm families in theproximity, and to commensal bacteria from medical stu-dents. The study clearly demonstrated that antimicrobialresistance emerged in exposed chicken and the farmfamily contacts. Similarly, in a study of 47 turkey,51 broiler and 25 layer farmers and their flocks as wellas 46 poultry slaughterers in the Netherlands, the preva-lence and degree of resistance in E. coli populations wasanalyzed and antibiotic use on the farms was recorded[104]. The prevalence of resistance was significantlyhigher in E. coli samples from turkey and broiler flocksthan in those from laying hens, which was correlatedwith antibiotic use practices. The prevalence of resist-ance was also higher in bacteria from turkey and broilerfarmers and slaughterers than in laying hen farmers. In-dividual isolates from farmers or slaughterers and birds
or their meat seemed to match, despite some variabilityacross farms. In another cross-sectional study, the preva-lence of resistance in E. coli isolates from 290 pigfarmers and 316 abattoir workers in the Netherlandswas compared to that of 160 urban or suburban resi-dents [103]. Fecal E. coli from the three human groupswere tested for resistance to multiple antimicrobialdrugs. The highest percentage of resistant E. coli wasfound in pig farmers, while urban/suburban residentshad the lowest prevalence. In another very complexstudy design, the colonization of poultry workers with aspecifically marked E. coli strain was evaluated [100].Birds were experimentally inoculated with a challengestrain of E. coli (36 birds on a university farm, and 16 ona commercial farm). Poultry workers in direct contactwith the challenged birds were sampled for colonizationwith the challenge strain. The results were compared tocontrol workers with or without direct contact to con-trol birds, and to samples collected from the exposedworkers prior to exposure. In addition, the birds weresampled for colonization. The study clearly showed thatafter birds were challenged, the poultry workers in directcontact with the challenged birds were colonized; resultswere similar on the university and commercial farm.Evidence for a transmission risk from farm animals to
humans is also provided by other study types. For in-stance, a number of studies lack proper negative controlsand apply complex study designs but, despite these limi-tations, provide evidence in support of a transmissionrisk. For instance, one study performed in the U.S. eval-uated the spread of resistant E. coli to livestock, farmersand the environment [105]. In separate experiments, onepig and one cow were experimentally inoculated with aspecific E. coli strain that carried a particular resistanceto make it easier to detect. Animals in contact with theinoculated animals (pigs or cows and mice) were sam-pled for the E. coli strain, as were human caretakers andthe environment. Contact animals, mice, flies and care-takers were found to excrete the challenge strain of E.coli. The length of colonization varied, but in severalcases exceeded 4 weeks. The E. coli strain was also foundin the environment and housing system and this findingimpacted the dynamics of spread. Exposure to chlor-tetracycline use in the experiments led to increased re-sistance, even though transfer of resistance plasmid toother bacteria was not detected. The number of animalsin this experiment was low (Table 6) and only the cattleexperiment was repeated once, with one ‘replicate’ in in-door and one in outdoor settings. Nonetheless, the ex-periments provide evidence that transmission ofresistant bacteria from livestock to human contacts canand does occur. In another uncontrolled transmissionstudy, also conducted in the U.S., the spread of resistantE. coli from calves to human contacts was evaluated, and
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 29 of 38
the impact of tetracycline exposure on the risk wasassessed [113]. In this case, the experiment was repeatedthree times. Calves were inoculated with a rare E. colistrain carrying resistance genes and human volunteers incontact with the calves were sampled for carriage of thechallenge strain. The study provided evidence that chal-lenge bacteria were transferred from calves to humancontacts. However, no statistically significant impact oftetracycline exposure on the transmission risk was de-tected. The exposure of calves to tetracycline did not re-sult in a statistically significant difference in resistancelevels of bacteria in the human volunteers. In yet an-other uncontrolled study, the transmission risk associ-ated with consumer handling of contaminated carcasseswas assessed [106]. An E. coli strain with an antimicro-bial resistance marker was used to inoculate a chickencarcass, and the volunteer handling the carcass was eval-uated for carriage of the marked strain. The study clearlyshowed that the E. coli strain was transferred from thechicken carcass to the volunteer handling it. Eventhough this study also lacked controls and replicates itdoes provide evidence that bacteria can be transferred toconsumers through the handling of contaminated meator poultry products.Results from correlation studies, which compare bacter-
ial populations in food producing animals and humancontacts, also support a transmission risk even though theevidence is less-well controlled and less rigorous. In onestudy in Norway, for instance, 13 cattle, and 3 familymembers in contact with the cattle as well as the veterin-arian, were evaluated for the presence of antimicrobial-resistant E. coli, and the sampling of human volunteerswas repeated after one year, at which point samples from4 other veterinarians were added [114]. The study foundconcurrent strains of multi-drug resistant E. coli in the an-imals and human contacts. In another Norwegian study,poultry farmers and their birds were sampled forvancomycin-resistant Enterococcus faecium and the gen-etic relatedness among the bacterial isolates was assessed[115]. On one of the farms, human and animal bacteriawere genetically closely related. Notably, even geneticallyunrelated strains shared related vancomycin resistancegenes, suggesting HGT. In another study in the U.S., 14farm families in Missouri and their livestock (i.e., cattle orswine) were sampled for E.coli and the genetic relatednessamong the human and animal isolates was assessed [116].The frequency of animal contact was recorded and its po-tential impact on the genetic relatedness among the iso-lates was assessed. Notably, the frequency of animalcontact was not significantly associated with the geneticrelatedness among human and animal isolates. However,consumption of home-raised beef appeared to be associ-ated with concordance between human and animalisolates.
Taken together, the data clearly support the notionthat antimicrobial-resistant commensal bacteria thatemerged in food-producing animals can be and aretransferred to humans. This may occur through directcontact with the animals, or indirectly, for instancethrough handling of contaminated carcasses. Notably, incases of transmission through direct contact directional-ity may not always be clear and in some instances bac-teria may also be transmitted from humans to theanimals they are in contact with. Nonetheless, the factthat commensal bacteria can be transferred to humansfrom food-producing animals is undoubtable.
Risk of resistance gene transfer from food-animal associatedcommensal bacteria to human pathogensResistance that emerged in commensal bacteria on farmsor feedlots poses a human health risk if it can be trans-ferred to human pathogens. This transfer may occur inthe human gut, in the environment, or in the food-producing animals. The ability of human commensals totransfer their resistance genes to human pathogens hasbeen extensively studied and is clearly possible, eventhough certain transfers are clearly more likely thanothers (see [117] for a review). Therefore this study fo-cuses on evidence for a transfer of resistance genes be-tween bacteria of human and animal origin regardless ofthe human-pathogenic potential of the human-associated recipient strain.A number of in-vivo and in-vitro studies have investi-
gated the ability of commensal bacteria of animal andhuman origin to transfer their resistance genes. In onestudy of six human volunteers, for instance, the transferof resistance genes between Enterococcus faecium strainsof chicken and human origin was investigated [108]. Re-sistance genes were readily transferred from chicken tohuman isolates in the gut of the human volunteers. Not-ably, in one volunteer, additional resistance genes werealso transferred. In another study, transfer of vanco-mycin resistance between Enterococcus faecium isolatesof human and swine origin was investigated in the gut ofgnotobiotic mice (i.e., mice reared under conditions sothat colonization with bacteria is fully known) [106].Vancomycin resistance was readily transferred from por-cine to human isolates in the gut of the mice. Notably,tylosin exposure through drinking water favoredcolonization with strains that had undergone conjuga-tion. Similar results were observed in another study,which evaluated the transfer of vancomycin resistancebetween Enterococci faecium strains of human and ani-mal origin in the environment and in the gut of germ-free mice [107]. Vancomycin resistance was readilytransferred among Enterococci of different origin. Not-ably, the frequency of conjugation was higher in themouse gut than in the environment. In most cases,
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 30 of 38
resistance disappeared within 3 days but one of the bac-terial strains persisted for more than 20 days in the ab-sence of antimicrobial exposure.Exchange of resistance genes between isolates of ani-
mal and human origin has also been demonstrated forother bacteria [118]. In one study, for instance, thetransfer of resistance genes from a multi-drug resistantKlebsiella strain isolated from a human pneumonia pa-tient to a mouse-adapted E. coli strain was investigatedin the mouse intestine and under in-vitro conditions[112]. Some of the mice were exposed to antimicrobialdrugs during the study. Resistance genes were readilytransferred in vitro and in vivo. Antimicrobial exposureselected for resistant strains, which rapidly disappearedin the guts of mice not exposed to antibiotics. Eventhough transfer in this case occurred from human toanimal –associated bacteria it clearly demonstrates thatthe exchange of genetic material is possible. In anotherstudy, the transfer of resistance genes between com-mensal E. coli and two pathogens (E. coli O157 and Sal-monella), all of porcine origin, was investigated in an in-vitro model of the porcine gut [108]. Antimicrobialresistance genes were readily transferred between thebacteria and in one example persisted throughout theduration of the study. In yet another study, the transferof resistance genes among E.coli, Vibrio and Aeromonaswas investigated in a number of environments (e.g.,meat, seawater, hand towel) [107]. Plasmids were readilytransferred from one bacterium to another, across bac-terial species and in a number of different environments.Similar findings have also been reported for MRSA. Inone study, for instance, gnotobiotic piglets were co-colonized with human- and pig-associated variants ofMRSA ST 398 and the bacterial populations werefollowed for 16 days and analyzed by whole-genome se-quencing [119]. Transfer of mobile genetic elementsfrom the pig to the human isolates was detected as earlyas 4 h after co-colonization. Plasmids and bacterio-phages were extensively and repeatedly transferredamong the isolates, indicating a very high frequency ofhorizontal gene transfer. Interestingly, bacterial popula-tions differed considerably across pigs, implying limitedtransfer among the pigs in this study. In another study,nine human volunteers ingested an E. coli strain of hu-man origin that carried resistance to rifampicin but wassusceptible to sulfonamides. This was followed threehours later by ingestion of another E. coli strain, of pigorigin and carrying sulfonamide resistance. Stool sam-ples from the volunteers were collected 24 h prior to thechallenge, and on days 1 to 7 as well as 14 and 35 after.Conjugation was detected on day 2 in one of the volun-teers, mediated through the exchange of a plasmid con-taining the sulfonamide resistance gene sul2, and co-transfer of ampicillin resistance was also demonstrated.
Notably, however, the sul2 resistance gene was identicalto one carried by commensal bacteria present in the gutof the volunteer before the experiment but differed fromthat carried by the pig strain, demonstrating the com-plexity of these types of investigations, in particular ifthe resident bacterial populations are not fully known orcontrolled. A number of other studies have also inves-tigated the transfer of various resistance genes amonga diverse group of bacteria in the guts of various ani-mals and other environments and found ample evi-dence that resistance genes are readily transferred(see [117] for a review). In addition, phylogeneticstudies provide further support for horizontal genetransfer of resistance by identifying virtually identicalresistance genes in distantly related bacterial speciesfrom varying geographic regions [120].Taken together, there is considerable evidence that re-
sistant animal-associated commensals can and do, atleast transiently, colonize humans that are in contactwith the animals or, in some cases, their food and thatthese commensals can transfer resistance genes tohuman-associated commensals and to human pathogens.Notably, antimicrobial exposure seems to favor the se-lection and maintenance of conjugates with newly-acquired antimicrobial resistance genes, even thoughthis selection pressure does not seem to be necessary inall cases. There is also strong evidence that the same re-sistance genes are present in bacteria from animals andhumans. What has so far remained less clear is how fre-quently this transfer occurs under real-world conditions,what other factors such as management practices orantibiotic exposures favor or limit this transfer, exactlywhen, where and how the transfers occur, and what frac-tion of resistance genes present in populations of humanpathogens originated in commensals on farms.
Excess morbidity and mortality caused by antimicrobialresistance traits that emerged on farmsTo fully evaluate the impact of antimicrobial resistanceas a public health burden, it is important to quantify theexcess morbidity and mortality caused by the fact that acertain pathogen is resistant to antimicrobial drugs. Thissection focuses on antimicrobial resistant zoonotic andfoodborne pathogens. However, the origin of resistancegenes is complex and it is not clear that the resistancegenes carried by these foodborne or zoonotic pathogensnecessarily emerged in animal agriculture rather thanother settings such as human healthcare. Regardless ofthe origin of the resistance genes, the studies are rele-vant to the question of whether infections withantimicrobial-resistant bacteria pose an increased publichealth risk.For ethical reasons, RCTs comparing health outcomes
associated with resistant as compared to susceptible
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 31 of 38
human pathogens are not feasible. However, observa-tional studies as well as other study types such as out-break analyses are available. Observational studiestypically compare outcomes such as hospitalizations ordeaths among patients infected with resistant or suscep-tible strains.A number of observational studies have compared
health outcomes for patients infected with susceptibleversus resistant strains of foodborne or zoonotic patho-gens such as Salmonella, Campylobacter or Staphylo-coccus aureus (Table 7). One study in the U.S., forinstance, compared the frequency of hospitalizationand bloodstream infections in patients with resistantand susceptible Salmonella strains [121]. Resistant iso-lates were significantly more likely to be associated withbloodstream infections than susceptible isolates. For ex-ample, for serotype Typhimurium, 3% of patients in-fected with susceptible isolates had bloodstreaminfections, compared to 6% of patients infected with re-sistant strains. Patients infected with resistant isolatesalso were more likely to be hospitalized because ofbloodstream infection than those infected with suscep-tible isolates. Among the hospitalized patients, thoseinfected with resistant isolates had longer stays thanthose infected with susceptible isolates. Similar resultswere obtained in another U.S. based study of 875 pa-tients with gastro-intestinal and bloodstream infectionsdue to Salmonella [122]. Bloodstream infections andhospitalizations were significantly more likely amongpatients with resistant compared to susceptible Salmon-ella isolates. Results from yet another U.S. based studyof patients infected with Salmonella further supportthis finding [123]. Patients with resistant infectionswere significantly more likely to be hospitalized thanthose with susceptible infections. Notably, recent anti-microbial treatment as well as age and race were sig-nificant predictors of resistance.Another observational study, conducted in Canada
to compare the risk of hospitalization in response toinfection with resistant Salmonella Typhimuriumstrains found similar results [124]. Hospitalization wassignificantly more likely in patients infected with re-sistant compared to susceptible strains. In fact, an es-timated 57–72% of hospitalizations were attributableto the resistance patterns. A Danish study of patientsinfected with Salmonella Typhimurium infections thatoccurred between 1995 and 1999 evaluated deathrates in patients infected with Salmonella Typhimur-ium strains carrying various antimicrobial resistancegenes [125]. For every patient enrolled in the study,10 random controls were found, matched by age, sexand county of residence. Subjects were followed forup to 2 years. Patients with susceptible SalmonellaTyphimurium infection were 2.3 times (95% CI 1.5 to
3.5) more likely to die during the study period thanthe general population. By comparison, patients in-fected with the most resistant strains were 4.8 times(95% CI: 2.2–10.2) more likely, and patients infectedwith quinolone-resistant isolates were 10.3 times (95%CI 2.8-37.8) more likely to die than the generalpopulation.Similar results have also been obtained for other bac-
teria. One Danish study of Campylobacter infections,for instance, compared the risk of invasive illness anddeath within up to 90 days of sample collection in 3471patients infected with quinolone and erythromycin re-sistant or susceptible strains [126]. Patients infectedwith quinolone-resistant Campylobacter strains had a6-fold increased risk of invasive illness or death within30 days of sample collection compared with patients in-fected with susceptible isolates. Similarly, infection witherythromycin-resistant Campylobacter strains was asso-ciated with more than a 5-fold higher risk within90 days compared to infections with susceptible strains.A meta-analysis of cohort studies focused on patientswith bloodstream infections due to Staphylococcus aur-eus in multiple countries found similar results [127].Data from a total of 3963 patients from 31 individualstudies were analyzed, comparing mortality rates due toinfections with methicillin susceptible and resistantstrains. A significantly increased mortality rate was as-sociated with MRSA infections compared to methicillin–susceptible strains (OR: 1.93; 95% CI – 1.52 – 2.42).Data from outbreaks further support worse health out-comes for infections with resistant compared to suscep-tible strains (see for instance [128]), even thoughpotential confounding effects such as virulence differ-ences among strains or differences in exposed popula-tions have to be considered.Taken together, there is unequivocal evidence, from epi-
demiological studies, meta-analyses, and the analysis ofoutbreaks, that infections with antimicrobial resistant bac-teria tend to be associated with worse public health out-comes than infections with susceptible strains. Notably,the impact appears to differ by pathogen as well as resist-ance involved [121, 126]. The underlying reasons may notalways be clear, but can include delayed treatment onset,the need to choose less-desirable treatment options, and,at least in some cases, increased virulence of antimicrobialdrug resistant strains because of co-selection for resistanceand virulence genes. Regardless of what drives the in-creased public health cost, infections with antimicrobialresistant bacteria are a public health threat that needs tobe addressed.
ConclusionsThe review clearly demonstrates that there is ample scien-tific support, based on observational studies, a variety of
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 32 of 38
Table
7Exam
ples
ofstud
ytype
sevaluatin
gthehe
alth
impactsassociated
with
human
infections
dueto
resistantas
comparedto
suscep
tiblebacterialstrains
Reference
Stud
yKeyfinding
srelevant
toantim
icrobialresistance
Type
asubjects
coun
try
time
perio
dbacterialtarge
trelevant
stud
you
tcom
essample
size
controls
commen
ts
GRA
DElevel:Con
trolledtrial
n/a
GRA
DElevel:ob
servationaltrial
[121]
Coh
ort
Patientswith
Salmon
ella
infections
US
1996–
2001
Salmon
ella
Freq
uencyof
hospitalization
and
bloo
dstream
infection
n=1415
patients
(Foo
dNet)
and
n=7370
isolates
(NARM
S)
Patientsinfected
with
suscep
tible
Salmon
ella
strains
Twoanalyses
werepe
rform
ed,
oneof
Salmon
ellaisolates
collected
throug
hNARM
S,and
oneof
patientswith
NARM
SSalmon
ellaisolates
and
correspo
ndingFood
Net
interviews
Resistantisolates
weremorelikely
associated
with
bloo
dstream
infections
than
suscep
tible
isolates
(forexam
ple,forserotype
Typh
imurium,3%
ofpatients
infected
with
suscep
tibleisolates
hadbloo
dstream
infections,
comparedto
6%of
patients
infected
with
resistantstrains);
patientsinfected
with
resistant
isolates
weremorelikelyto
beho
spitalized
becauseof
bloo
dstream
infectionthan
those
infected
with
suscep
tibleisolates;
amon
gho
spitalized
patients,
thoseinfected
with
resistant
isolates
hadlong
erstaysthan
thoseinfected
with
suscep
tible
isolates;
[122]
Coh
ort
Patientswith
Salmon
ella
infections
(blood
stream
andGI
infection)
US
2006–
2008
Salmon
ella
Clinical
outcom
esn=875
patients
Patientsinfected
with
suscep
tible
Salmon
ella
strains
Bloo
dstream
infections
and
hospitalizations
morelikely
amon
gpatientswith
resistant
than
suscep
tibleisolates
[123]
Coh
ort
Patients
infected
with
Salmon
ella
US
1989–
1990
Salmon
ella
Infectionand
hospitalization
Patientsinfected
with
suscep
tible
Salmon
ella
strains
Results
werecomparedto
similarstud
ycond
uctedin
1979–1980
Patientswith
resistantinfections
weremorelikelyto
beho
spitalized
than
thosewith
suscep
tibleinfections;recen
tantim
icrobialtreatm
entas
wellas
ageandrace
weresign
ificant
pred
ictorsof
resistance
[124]
Coh
ort
Patientswith
Salmon
ella
infections
Canada
1999–
2000
Salmon
ella
Typh
imurium
Risk
ofho
spitalization
n=440
patients
Patientsinfected
with
suscep
tible
Salmon
ella
strains
Hospitalizationwas
morelikelyin
patientsinfected
with
resistant
than
suscep
tiblestrains;an
estim
ated
57–72%
ofho
spitalizations
wereattributable
toresistance
patterns
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 33 of 38
Table
7Exam
ples
ofstud
ytype
sevaluatin
gthehe
alth
impactsassociated
with
human
infections
dueto
resistantas
comparedto
suscep
tiblebacterialstrains
(Con
tinued)
[125]
Coh
ort
Patients
infected
with
Salmon
ella
Typh
imurium
Den
mark
1995–
1999
Salmon
ella
Typh
imurium
Death
rates
N=2047
10rand
omcontrols
(matched
byage,sex
andcoun
tyof
reside
nce);
comparison
betw
een
suscep
tibleand
resistantinfections
Subjectswerefollowed
forup
to2yearsafteren
terin
gthe
stud
y
Patientswith
suscep
tible
Salmon
ellaTyph
imurium
infection
were2.3tim
esmorelikelyto
die
durin
gthestud
ype
riodthan
the
gene
ralp
opulation;patientswith
mostresistantstrainswere4.8
times
(95%
CI:2.2–10.2)more
likely;patientsinfected
with
quinolon
e-resistantisolates
were
10.3tim
esmorelikelyto
diethan
thege
neralp
opulation.
[126]
Coh
ort
Patientswith
Campylobacter
infections
Den
mark
Campylobacter
(quino
lone
and
erythrom
ycin-
resistant)
Risk
ofinvasive
illne
ssand
deathwith
inup
to90
days
ofsamplereceipt
n=3471
patients
Patientsinfected
with
suscep
tible
Campylobacter
isolates
Patientsinfected
with
quinolon
e-resistantCa
mpylobacterstrains
hada6-fold
increasedriskof
in-
vasive
illne
ssor
deathwith
in30
days
ofsample
collection
comparedwith
patientsinfected
with
susceptibleisolates;infectionwith
erythrom
ycin-resistant
Campylobacter
strainswasassociated
with
>5-fold
high
erriskwithin90
days
compared
toinfections
with
susceptiblestrains
[127]
Meta-
analysis
ofcoho
rtstud
ies
Stud
iesof
patientswith
bloo
dstream
infectiondu
eto
Staph.
aureus
multip
lecoun
tries
1980–
2000
Staphylococcus
aureus,M
RSA
Mortalityrates
dueto
infections
with
methicillin
suscep
tibleand
resistant
Staphylococcus
aureus
n=3963
patientsin
n=31
stud
ies
Patientsinfected
with
methicillinsuscep
tible
strain
Results
pooled
with
rand
omeffectsmod
elIncreasedmortalityassociated
with
MRSAinfections
compared
tomethicillin–susceptiblestrains
(OR:1.93;95%
CI–
1.52
–2.42)
GRA
DElevel:othe
r
[128]
Outbreak
analysis
Outbreaks
ofpatients
infected
with
Salmon
ella
US
1984–
2002
Salmon
ella
Hospitalization
dueto
infection
with
suscep
tible
orresistant
Salmon
ella
strains
n=23,206
patientsin
n=32
outbreaks
Outbreaks
with
suscep
tible
Salmon
ella
strains
Themed
ianprop
ortio
nho
spitalized
foreach
type
ofou
tbreak
caused
byresistant
strainswas
26%,2.5tim
eshigh
erthan
themed
ianprop
ortio
nho
spitalized
forou
tbreakscaused
bypan-suscep
tiblestrains
a CCcase-con
trol
stud
y(orcase-casestud
ywhe
recontrolsarecasesof
infectionwith
othe
rstrains),C
Scross-sectiona
lstudy
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 34 of 38
other relevant scientific evidence, and, where ethicallyfeasible, also RCTs to support each step in the causalpathway from antimicrobial use on farms to the publichealth burden caused by antimicrobial resistant infectionsin humans. These studies reach across the major food-producing species and a variety of bacterial species includ-ing Salmonella, Campylobacter, E. coli and Enterococcieven though there seem to be some important differencesamong the bacterial species and data are not readily avail-able for all bacteria, species and antimicrobial drugs ofinterest. Yet, taken together, the data support what the sci-entific community, national governments, and inter-national organizations such as the World HealthOrganization, the Food and Agricultural Organization ofthe United Nations, and the World Organization for theHealth of Animals (OIE) have long recognized: antimicro-bial use on farms clearly contributes to the emergence ofresistance and poses a human public health risk.Because the molecular and evolutionary determinants
of antimicrobial resistance are complex, resistance doesnot strictly follow a simple epidemiological model ofcausation; rather, a more sophisticated model typicallyused for chronic diseases applies. For example, anti-microbial use may not in all cases lead to the immediateemergence of resistance, and its discontinuation may notlead to an immediate drop in resistance. The antimicro-bial drug and treatment regimen as well as external fac-tors such as the feed type used on a given operation canhave significant impacts on resistance, and rates of re-sistance emergence may be markedly different acrossbacterial species or even strains.Bacterial resistance genes can be readily shared across
bacteria, and some may be pre-existing or persist in en-vironmental reservoirs such as soils. Tracing their originback across bacterial backgrounds is challenging, eventhough technological advances such as whole-genomesequencing and more advanced phylogenetic and popu-lation genetic models make their trace-back increasinglyfeasible. Currently, it may be impossible to exactly quan-tify the contribution of animal agriculture to the overallburden of resistance, but this goal is becoming increas-ingly feasible.Importantly, even if some data gaps remain to be
filled, there is no doubt that antimicrobial use on farmsor feedlots contributes to the problem of antimicrobialresistance. It is therefore important to take action nowto ensure antimicrobial drugs are used judiciously, andonly when needed to ensure animal health and well-being. Almost 50 years have passed since the Swann re-port called for the discontinuation of growth promotionuses for medically important antibiotics, and the practicehas just now been phased out in the U.S. Many of thestudies reviewed here are not new – in fact, many well-designed, compelling studies were published more than
20 or 30 years ago. Time is running out on curtailingantimicrobial resistance. Antimicrobial use on farmscontributes to the burden of antimicrobial resistance.Science has shown it, many times over, and stakeholdersaround the world have come to accept it. It is time tomove the public debate away from the problem to itspotential solutions.
AbbreviationsGRADE: Grades of recommendations assessment, development andevaluation; HGT: Horizontal gene transfer; JETACAR: Joint experttechnical advisory committee on antibiotic resistance; MIC: Minimuminhibitory concentration; MRSA: Methicillin resistant Staphylococcus aureus;OIE: World Organization for Animal Health; RCT: Randomized controlled trial;VRE: Vancomycin resistant Enterococci
AcknowledgementsNot applicable.
Availability of data and materialsData sharing not applicable to this article as no datasets were generated oranalyzed during the current study.
Authors’ contributionsKH, KT, EJ and AC conceived the study. KH, NW and JT provided analysis forthe manuscript. KH drafted the manuscript. KT, EJ and AC critically andsubstantially revised the manuscript. All authors approved the manuscript.
Competing interestsThe authors declare that they have no competing interests.
Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.
Received: 1 August 2016 Accepted: 23 June 2017
References1. U.S. Centers for Disease Control and Prevention. Antibiotic Resistance
Threats in the United States, 2013. Atlanta: U.S. Department of Health andHuman Services. 2013. https://www.cdc.gov/drugresistance/pdf/ar-threats-2013-508.pdf.
2. McEwen SA, Fedorka-Cray PJ. Antimicrobial use and resistance in animals.Clin Infect Dis. 2002;34(Supplement 3):S93–S106.
3. Marshall BM, Levy SB. Food animals and antimicrobials: impacts on humanhealth. Clin Microbiol Rev. 2011;24(4):718–33.
4. Codex Alimentarius. Code of Practice to Minimize and Contain AntimicrobialResistance. List of Standards- FAO. CAC/RCP 61–2005. www.fao.org/input/download/standards/10213/CXP_061e.pdf. Accessed 10 Apr 2017.
5. Casewell M, Friis C, Marco E, McMullin P, Phillips I. The European ban ongrowth-promoting antibiotics and emerging consequences for human andanimal health. J Antimicrob Chemother. 2003;52(2):159–61.
6. World Health Organization. Critically important antimicrobials for humanmedicine. 4th revision. 2016. p. 1–31.
7. European Commission. Ban on antibiotics as growth promoters in animalfeed enters into effect. European Commission. 2005. http://europa.eu/rapid/press-release_IP-05-1687_en.htm. Accessed 28 June 2017.
8. Food and Drug Administration. Guidance for Industry #152. Center forVeterinary Medicine. 2003. p. 1–35. https://www.fda.gov/downloads/AnimalVeterinary/GuidanceComplianceEnforcement/GuidanceforIndustry/UCM052519.pdf.
9. Moulin G, Góchez, D, Lasley J, Erlacher-Vindel E. OIE Annual report on theuse of antimicrobial agents in animals. OIE. 2016. p. 1–65. http://www.oie.int/fileadmin/Home/fr/Our_scientific_expertise/docs/pdf/AMR/Survey_on_monitoring_antimicrobial_agents_Dec2016.pdf.
10. Food and Drug Administration. FDA’s CVM Key Initiatives for AntimicrobialStewardship. https://www.fda.gov/AnimalVeterinary/SafetyHealth/
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 35 of 38
AntimicrobialResistance/JudiciousUseofAntimicrobials/ucm535158.htm.Accessed 11 April 2017.
11. MacGowan AP. Clinical implications of antimicrobial resistance for therapy. JAntimicrob Chemother. 2008;62(suppl 2):ii105–14.
12. Levy SB, Marshall B. Antibacterial resistance worldwide: causes, challengesand responses. Nat Med. 2004;
13. Parascandola M, Weed DL. Causation in epidemiology. J EpidemiolCommunity Health. 2001;55(12):905–12.
14. Davies J, Davies D. Origins and evolution of antibiotic resistance. MicrobiolMol Biol Rev. 2010;74(3):417–33.
15. Tavares LS, Silva CS, de Souza VC, da Silva VL, Diniz CG, Santos MO.Strategies and molecular tools to fight antimicrobial resistance: resistome,transcriptome, and antimicrobial peptides. Front Microbiol. 2013;4:412.
16. Rothman KJ. Causal inference in epidemiology. Modern Epidemiology.Boston: Little, Brown and Company. 1986;7–21.
17. Tlaskalová-Hogenová H, Štěpánková R, Hudcovic T, Tučková L, Cukrowska B,Lodinová-Zádníková R, et al. Commensal bacteria (normal microflora),mucosal immunity and chronic inflammatory and autoimmune diseases.Immunol Lett. 2004;93(2–3):97–108.
18. Huddleston JR. Horizontal gene transfer in the human gastrointestinal tract:potential spread of antibiotic resistance genes. Infect Drug Resist. 2014;7:167–76.
19. Casey JA, Curriero FC, Cosgrove SE, Nachman KE, Schwartz BS. High-densitylivestock operations, crop field application of manure, and risk ofcommunity-associated methicillin-resistant Staphylococcus aureus infectionin Pennsylvania. JAMA Intern Med. 2013;173(21):1980–90.
20. Rule AM, Evans SL, Silbergeld EK. Food animal transport: a potential sourceof community exposures to health hazards from industrial farming (CAFOs).J Infect Public Health. 2008;1(1):33–9.
21. McEachran AD, Blackwell BR, Hanson JD, Wooten KJ, Mayer GD, Cox SB,Smith PN. Antibiotics, bacteria, and antibiotic resistance genes: aerialtransport from cattle feed yards via particulate matter. Environmental HealthPerspectives. 2015;123(4):337.
22. Ward MJ, Gibbons CL, McAdam PR, Van Bunnik BAD, Girvan EK, Edwards GF,et al. Time-scaled evolutionary analysis of the transmission and antibioticresistance dynamics of Staphylococcus aureus clonal complex 398. ApplEnviron Microbiol. 2014;80(23):7275–82.
23. Pantosti A. Methicillin-resistant Staphylococcus aureus associated withanimals and its relevance to human health. Frontiers in Microbiology. 2012;3:127.
24. van Rijen MML, Bosch T, Verkade EJM, Schouls L, Kluytmans JAJW. On behalfof the CAMSG. Livestock-associated MRSA carriage in patients withoutdirect contact with livestock. PLoS One. 2014;9(6):e100294.
25. Barza M. Potential mechanisms of increased disease in humans fromantimicrobial resistance in food animals. Clin Infect Dis. 2002;34(Supplement3):S123–5.
26. Eliopoulos GM, Cosgrove SE, Carmeli Y. The impact of antimicrobialresistance on health and economic outcomes. Clin Infect Dis. 2003;36(11):1433–7.
27. Travers K, Michael B. Morbidity of infections caused by antimicrobial-resistant bacteria. Clin Infect Dis. 2002;34(Supplement 3):S131–4.
28. Kaye KS, Engemann JJ, Mozaffari E, Carmeli Y. Reference group choice andantibiotic resistance outcomes. Emerg Infect Dis. 2004;10(6):1125–8.
29. Jones TF, Ingram LA, Cieslak PR, Vugia DJ, Tobin-D'Angelo M, Hurd S, et al.Salmonellosis outcomes differ substantially by serotype. J Infect Dis. 2008;198(1):109–14.
30. Oxman AD, Group GW. Grading quality of evidence and strength ofrecommendations. BMJ. 2004;328(19):1490–4.
31. GRADE Working Group. http://www.gradeworkinggroup.org/. Accessed 12July 2016.
32. Ferguson JK, Dalton CB, McGettigan P, Hill S. Antimicrobial resistance inanimal enteric bacteria and human disease—a review of the scientificliterature. Commissioned Report Jt Expert Tech Advis Comm Antibiot Resist(JETACAR) Canberra. 1998;
33. Australian Government Department of Health. The Use of Antibiotics inFood-Producing Animals: Antibiotic-Resistant Bacteria in Animals andHumans. Commonwealth of Australia. 2014. http://www.health.gov.au/internet/main/publishing.nsf/content/health-pubs-jetacar-cnt.htm. Accessed28 June 2017.
34. Andreoletti O, Budka H, Buncic S, Colin P, Collins JD, De Koeijer A, Griffin J,Havelaar A, Hope J, Klein G, Kruse H. Foodborne antimicrobial resistance asa biological hazard. EFSA J. 2008;765:2–87.
35. Chauvin C, Querrec M, Perot A, Guillemot D, Sanders P. Impact ofantimicrobial drug usage measures on the identification of heavy users,patterns of usage of the different antimicrobial classes and time-trendsevolution. J Vet Pharmacol Ther. 2008;31(4):301–11.
36. Jensen VF, Jacobsen E, Bager F. Veterinary antimicrobial-usage statistics basedon standardized measures of dosage. Prev Vet Med. 2004;64(2–4):201–15.
37. Timmerman T, Dewulf J, Catry B, Feyen B, Opsomer G. Kruif ad, Maes D.Quantification and evaluation of antimicrobial drug use in group treatmentsfor fattening pigs in Belgium. Prev Vet Med. 2006;74(4):251–63.
38. Gow SP, Waldner CL. Antimicrobial drug use and reason for treatment in203 western Canadian cow-calf herds during calving season. Prev Vet Med.2009;90(1–2):55–65.
39. Merle R, Mollenhauer Y, Hajek P, Robanus M, Hegger-Gravenhorst C,Honscha W, et al. Monitoring of antibiotic consumption in cattle onagricultural farms. Berliner und Munchener tierarztliche Wochenschrift. 2013;126(7–8):318–25.
40. Thomson K, Rantala M, Hautala M, Pyörälä S, Kaartinen L. Cross-sectionalprospective survey to study indication-based usage of antimicrobials inanimals: results of use in cattle. BMC Vet Res. 2008;4(1):1.
41. Zwald AG, Ruegg PL, Kaneene JB, Warnick LD, Wells SJ, Fossler C, et al.Management practices and reported antimicrobial usage on conventionaland organic dairy farms. J Dairy Sci. 2004;87(1):191–201.
42. Carrique-Mas JJ, Trung NV, Hoa NT, Mai HH, Thanh TH, Campbell JI, et al.Antimicrobial usage in chicken production in the Mekong Delta of Vietnam.Zoonoses Public Health. 2015;62:70–8.
43. National Animal Health Monitoring System (NAHMS). https://www.aphis.usda.gov/aphis/ourfocus/animalhealth/monitoring-and-surveillance/nahms.Accessed 12 July 2016.
44. Kim DP, Saegerman C, Douny C, Dinh TV, Xuan BH, Vu BD, et al. First surveyon the use of antibiotics in pig and poultry production in the Red RiverDelta region of Vietnam. Food Public Health. 2013;3(5):247–56.
45. Agricultural Resource Management Survey (ARMS). http://www.ers.usda.gov/data-products/arms-farm-financial-and-crop-production-practices/arms-data.aspx. Accessed 12 July 2016.
46. McDermott PF, Bodeis SM, English LL, White DG, Walker RD, Zhao S, et al.Ciprofloxacin resistance in campylobacter jejuni evolves rapidly in chickenstreated with fluoroquinolones. J Infect Dis. 2002;185(6):837–40.
47. Delsol AA, Anjum M, Woodward MJ, Sunderland J, Roe JM. The effect ofchlortetracycline treatment and its subsequent withdrawal on multi-resistant salmonella enterica serovar Typhimurium DT104 and commensalEscherichia coli in the pig. J Appl Microbiol. 2003;95(6):1226–34.
48. Mathew AG, Jackson F, Saxton AM. Effects of antibiotic regimens onresistance of Escherichia coli and salmonella serovar Typhimurium in swine. JSwine Health Prod. 2002;10(1):7–14.
49. Jacobs-Reitsma WF, Kan CA, Bolder NM. The induction of quinoloneresistance in campylobacter bacteria in broilers by quinolone treatment.Lett Appl Microbiol. 1994;19(4):228–31.
50. Inglis GD, McAllister TA, Busz HW, Yanke LJ, Morck DW, Olson ME, et al.Effects of subtherapeutic administration of antimicrobial agents to beefcattle on the prevalence of antimicrobial resistance in campylobacterjejuni and campylobacter hyointestinalis. Appl Environ Microbiol. 2005;71(7):3872–81.
51. Mathew AG, Garner KN, Ebner PD, Saxton AM, Clift RE, Liamthong S. Effectsof antibiotic use in sows on resistance of E. coli and salmonella entericaTyphimurium in their offspring. Foodborne Pathog Dis. 2005;2(3):212–20.
52. Delsol AA, Sunderland J, Woodward MJ, Pumbwe L, Piddock LJ, Roe JM.Emergence of fluoroquinolone resistance in the native campylobacter colipopulation of pigs exposed to enrofloxacin. J Antimicrob Chemother. 2004May 1;53(5):872–4.
53. Ebner PD, Mathew AG. Effects of antibiotic regimens on the fecal sheddingpatterns of pigs infected with salmonella Typhimurium. J Food Prot. 2000Jun;63(6):709–14.
54. Ray KA, Warnick LD, Mitchell RM, Kaneene JB, Ruegg PL, Wells SJ, et al.Antimicrobial susceptibility of salmonella from organic and conventionaldairy farms. J Dairy Sci. 2006;89(6):2038–50.
55. Sato K, Bartlett PC, Kaneene JB, Downes FP. Comparison of prevalence andantimicrobial susceptibilities of campylobacter spp. isolates from organicand conventional dairy herds in Wisconsin. Appl Environ Microbiol. 2004;70(3):1442–7.
56. Luangtongkum T, Morishita TY, Ison AJ, Huang S, McDermott PF, Zhang Q.Effect of conventional and organic production practices on the prevalence
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 36 of 38
and antimicrobial resistance of campylobacter spp. in poultry. Appl EnvironMicrobiol. 2006;72(5):3600–7.
57. Griggs DJ, Johnson MM, Frost JA, Humphrey T, Jørgensen F, Piddock LJV.Incidence and mechanism of ciprofloxacin resistance in campylobacter spp.isolated from commercial poultry flocks in the United Kingdom before,during, and after fluoroquinolone treatment. Antimicrob Agents Chemother.2005;49(2):699–707.
58. Avrain L, Humbert F, L’Hospitalier R, Sanders P, Vernozy-Rozand C,Kempf I. Antimicrobial resistance in campylobacter from broilers:association with production type and antimicrobial use. Vet Microbiol.2003;96(3):267–76.
59. Price LB, Leila GL, Vailes R, Silbergeld E. The persistence of Fluoroquinolone-resistant campylobacter in poultry production. Environ Health Perspect.2007;115(7):1035–9.
60. Klare I, Badstübner D, Konstabel C, Böhme G, Claus H, Witte W. Decreasedincidence of VanA-type vancomycin-resistant enterococci isolated frompoultry meat and from fecal samples of humans in the community afterdiscontinuation of avoparcin usage in animal husbandry. Microb DrugResist. 1999;5(1):45–52.
61. Endtz HP, Ruijs GJ, van Klingeren B, Jansen WH, van der Reyden T, MoutonRP. Quinolone resistance in campylobacter isolated from man and poultryfollowing the introduction of fluoroquinolones in veterinary medicine. JAntimicrob Chemother. 1991;27(2):199–208.
62. Desmonts M-H, Dufour-Gesbert F, Avrain L, Kempf I. Antimicrobial resistance incampylobacter strains isolated from French broilers before and after antimicrobialgrowth promoter bans. J Antimicrob Chemother. 2004;54(6):1025–30.
63. Dutil L. Ceftiofur Resistance in Salmonella enterica Serovar Heidelberg fromChicken Meat and Humans, Canada-Volume 16, Number 1—January 2010-Emerging Infectious Disease journal-CDC.
64. Boerlin P, Wissing A, Aarestrup FM, Frey J, Nicolet J. Antimicrobial growthpromoter ban and resistance to Macrolides and Vancomycin in Enterococcifrom pigs. J Clin Microbiol. 2001;39(11):4193–5.
65. Alexander TW, Yanke LJ, Topp E, Olson ME, Read RR, Morck DW, et al. Effectof subtherapeutic administration of antibiotics on the prevalence ofantibiotic-resistant Escherichia coli bacteria in feedlot cattle. Appl EnvironMicrobiol. 2008;74(14):4405–16.
66. Berge ACB, Epperson WB, Pritchard RH. Assessing the effect of a single doseflorfenicol treatment in feedlot cattle on the antimicrobial resistancepatterns in faecal Escherichia coli. Vet Res. 2005;36(5–6):723–34.
67. Kanwar N, Scott HM, Norby B, Loneragan GH, Vinasco-Torres J, Cottell JL, etal. Impact of treatment strategies on cephalosporin and tetracyclineresistance gene quantities in the bovine fecal metagenome. Sci Rep. 2014Jul 23;4(Article number: 5100):1–0.
68. Aarestrup FM, Carstensen B. Effect of tylosin used as a growth promoter onthe occurrence of macrolide-resistant enterococci and staphylococci in pigs.Microb Drug Resist. 1998;4(4):307–12.
69. Linton AH, Hedges AJ, Bennett PM. Monitoring for the development ofantimicrobial resistance during the use of olaquindox as a feed additive oncommercial pig farms. J Appl Bacteriol. 1988;64(4):311–27.
70. Dunlop RH, McEwen SA, Meek AH, Clarke RC, Black WD, Friendship RM.Associations among antimicrobial drug treatments and antimicrobialresistance of fecal Escherichia coli of swine on 34 farrow-to-finish farms inOntario, Canada. Prev Vet Med. 1998;34(4):283–305.
71. Akwar HT, Poppe C, Wilson J, Reid-Smith RJ, Dyck M, Waddington J, et al.Associations of antimicrobial uses with antimicrobial resistance of fecalEscherichia coli from pigs on 47 farrow-to-finish farms in Ontario and BritishColumbia. Canadian J Vet Res = Revue Canadienne De RechercheVeterinaire. 2008;72(2):202–10.
72. Chantziaras I, Boyen F, Callens B, Dewulf J. Correlation betweenveterinary antimicrobial use and antimicrobial resistance in food-producing animals: a report on seven countries. J AntimicrobChemother. 2014;69(3):827–34.
73. Pantosti A, Del Grosso M, Tagliabue S, Macri A, Caprioli A. Decrease ofvancomycin-resistant enterococci in poultry meat after avoparcin ban.Lancet. 1999;354(9180):741–2.
74. Zilberberg MD, Chen J, Mody SH, Ramsey AM, Shorr AF. Imipenemresistance of pseudomonas in pneumonia: a systematic literature review.BMC Pulm Med. 2010;10:45.72.
75. Bliziotis IA, Samonis G, Vardakas KZ, Chrysanthopoulou S, Falagas ME. Effectof aminoglycoside and β-lactam combination therapy versus β-lactam
monotherapy on the emergence of antimicrobial resistance: a meta-analysisof randomized, controlled trials. Clin Infect Dis. 2005;41(2):149–58.
76. Costelloe C, Metcalfe C, Lovering A, Mant D, Hay AD. Effect of antibioticprescribing in primary care on antimicrobial resistance in individual patients:systematic review and meta-analysis. BMJ. 2010;340
77. Carmeli Y, Troillet N, Eliopoulos GM, Samore MH. Emergence of antibiotic-resistant Pseudomonas Aeruginosa: comparison of risks associated withdifferent antipseudomonal agents. Antimicrob Agents Chemother. 1999;43(6):1379–82.
78. Harbarth S, Samore MH, Lichtenberg D, Carmeli Y. Prolonged antibioticprophylaxis after cardiovascular surgery and its effect on surgical siteinfections and antimicrobial resistance. Circulation. 2000;101(25):2916–21.
79. El Amari EB, Chamot E, Auckenthaler R, Pechère JC, Van Delden C.Influence of previous exposure to antibiotic therapy on thesusceptibility pattern of Pseudomonas Aeruginosa Bacteremic isolates.Clin Infect Dis. 2001;33(11):1859–64.
80. Pechere JC, Marchou B, Michea-Hamzehpour M, Auckenthaler R. Emergenceof resistance after therapy with antibiotics used alone or combined in amarine model. J Antimicrob Chemother. 1986;17(suppl A):11–8.
81. Tan C-H, Li J, Nation RL. Activity of colistin against heteroresistantAcinetobacter Baumannii and emergence of resistance in an in vitropharmacokinetic/pharmacodynamic model. Antimicrob Agents Chemother.2007;51(9):3413–5.
82. Goossens H, Ferech M, Vander Stichele R, Elseviers M. Outpatient antibioticuse in Europe and association with resistance: a cross-national databasestudy. Lancet. 365(9459):579–87.
83. Olofsson SK, Cars O. Optimizing drug exposure to minimize selection ofantibiotic resistance. Clin Infect Dis. 2007;45(Supplement 2):S129–36.
84. Melnyk AH, Wong A, Kassen R. The fitness costs of antibiotic resistancemutations. Evol Appl. 2015;8(3):273–83.
85. Cometta A, Baumgartner JD, Lew D, Zimmerli W, Pittet D, Chopart P, et al.Prospective randomized comparison of imipenem monotherapy withimipenem plus netilmicin for treatment of severe infections in nonneutropenicpatients. Antimicrob Agents Chemother. 1994;38(6):1309–1313.83.
86. Geli P, Laxminarayan R, Dunne M, Smith DL. “one-size-fits-all”? Optimizingtreatment duration for bacterial infections. PLoS One. 2012;7(1):e29838.
87. Gupta A, Fontana J, Crowe C, Bolstorff B, Stout A, Duyne SV, et al.Emergence of multidrug-resistant salmonella enterica serotype Newportinfections resistant to expanded-Spectrum Cephalosporins in the UnitedStates. J Infect Dis. 2003;188(11):1707–1716.85.
88. Varma JK, Marcus R, Stenzel SA, Hanna SS, Gettner S, Anderson BJ, et al.Highly resistant salmonella Newport-MDRAmpC transmitted through thedomestic US food supply: a FoodNet case-control study of sporadicsalmonella Newport infections, 2002–2003. J Infect Dis.2006;194(2):222–30.
89. Karon AE, Archer JR, Sotir MJ, Monson TA, Kazmierczak JJ. Humanmultidrug-resistant salmonella Newport infections, Wisconsin, 2003-2005.Emerg Infect Dis. 2007;13(11):1777.87.
90. Spika JS, Waterman SH, Hoo GWS, St. Louis ME, Pacer RE, James SM, et al.Chloramphenicol-resistant salmonella newport traced through hamburgerto dairy farms. N Engl J Med. 1987;316(10):565–570.88.
91. van Loo I, Huijsdens X, Tiemersma E, de Neeling A, van de Sande-BruinsmaN, Beaujean D, et al. Emergence of methicillin-resistant Staphylococcusaureus of animal origin in humans. Emerg Infect Dis. 2007;13(12):1834–9.
92. Graveland H, Wagenaar JA, Heesterbeek H, Mevius D, van Duijkeren E,Heederik D. Methicillin resistant Staphylococcus aureus ST398 in veal calffarming: human MRSA carriage related with animal antimicrobial usage andfarm hygiene. PLoS One. 2010;5(6):e10990.
93. Denis O, Suetens C, Hallin M, Catry B, Ramboer I, Dispas M, et al. Methicillin-resistant Staphylococcus aureus ST398 in swine farm personnel, Belgium.Emerg Infect Dis. 2009;15(7):1098–101.
94. Mølbak K, Baggesen DL, Aarestrup FM, Ebbesen JM, Engberg J, FrydendahlK, et al. An outbreak of multidrug-resistant, Quinolone-resistant salmonellaenterica serotype Typhimurium DT104. N Engl J Med. 1999;341(19):1420–5.
95. Holmberg SD, Osterholm MT, Senger KA, Cohen ML. Drug-resistantsalmonella from animals fed antimicrobials. N Engl J Med.1984;311(10):617–22.
96. Fey PD, Safranek TJ, Rupp ME, Dunne EF, Ribot E, Iwen PC, et al.Ceftriaxone-resistant salmonella infection acquired by a child from cattle. NEngl J Med. 2000;342(17):1242–9.
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 37 of 38
97. Huijsdens XW, van Dijke BJ, Spalburg E, van Santen-Verheuvel MG, Heck ME,Pluister GN, et al. Community-acquired MRSA and pig-farming. Ann ClinMicrobiol Antimicrob. 2006;5(1):1–4.
98. Smith TC, Male MJ, Harper AL, Kroeger JS, Tinkler GP, Moritz ED, et al.Methicillin-resistant Staphylococcus aureus (MRSA) strain ST398 ispresent in Midwestern U.S. swine and swine workers. PLoS One.2009;4(1):e4258.
99. Juhasz-Kaszanyitzky E, Janosi S, Somogyi P, Dan A, van der Graaf-van BlooisL, van Duijkeren E, et al. MRSA transmission between cows and humans.Emerg Infect Dis. 2007;13(4):630–2.
100. Ojeniyi AA. Direct transmission of Escherichia coli from poultry to humans.Epidemiol Infect. 1989;103(03):513–22.
101. Levy SB. Emergence of antibiotic-resistant bacteria in the intestinal Flora offarm inhabitants. J Infect Dis. 1978;137(5):688–90.
102. Levy SB, FitzGerald GB, Macone AB. Changes in intestinal Flora of farmpersonnel after introduction of a tetracycline-supplemented feed on a farm.N Engl J Med. 1976;295(11):583–8.
103. Nijsten R, London N, Bogaard AVD, Stobberingh E. Resistance in faecalEscherichia coli isolated from pigfarmers and abattoir workers. EpidemiolInfect. 1994;113(01):45–52.
104. Van den Bogaard AE, London N, Driessen C, Stobberingh EE. Antibioticresistance of faecal Escherichia coli in poultry, poultry farmers and poultryslaughterers. J Antimicrob Chemother. 2001;47(6):763–71.
105. Marshall B, Petrowski D, Levy SB. Inter- and Intraspecies spread ofEscherichia coli in a farm environment in the absence of antibiotic usage.Proc Natl Acad Sci U S A. 1990;87(17):6609–13.
106. Linton AH, Howe K, Bennett PM, Richmond MH, Whiteside EJ. Thecolonization of the human gut by antibiotic resistant Escherichia coli fromchickens. J Appl Bacteriol. 1977;43(3):465–9.
107. Kruse H, Sørum H. Transfer of multiple drug resistance plasmids betweenbacteria of diverse origins in natural microenvironments. Appl EnvironMicrobiol. 1994;60(11):4015–21.
108. Blake DP, Hillman K, Fenlon DR, Low JC. Transfer of antibiotic resistancebetween commensal and pathogenic members of the Enterobacteriaceaeunder ileal conditions. J Appl Microbiol. 2003;95(3):428–36.
109. Moubareck C, Bourgeois N, Courvalin P, Doucet-Populaire F. Multipleantibiotic resistance gene transfer from animal to human enterococci in thedigestive tract of gnotobiotic mice. Antimicrob Agents Chemother. 2003;47(9):2993–6.
110. Dahl KH, Mater DDG, Flores MJ, Johnsen PJ, Midtvedt T, Corthier G, et al.Transfer of plasmid and chromosomal glycopeptide resistance determinantsoccurs more readily in the digestive tract of mice than in vitro andexconjugants can persist stably in vivo in the absence of glycopeptideselection. J Antimicrob Chemother. 2007;59(3):478–86.
111. Lester CH, Frimodt-Møller N, Sørensen TL, Monnet DL, Hammerum AM. Invivo transfer of the vanA resistance gene from an Enterococcus faeciumisolate of animal origin to an E. Faecium isolate of human origin in theintestines of human volunteers. Antimicrob Agents Chemother.2006;50(2):596–9.
112. Schjørring S, Struve C, Krogfelt KA. Transfer of antimicrobial resistanceplasmids from Klebsiella pneumoniae to Escherichia coli in the mouseintestine. J Antimicrob Chemother. 2008;62(5):1086–93.
113. Hirsh DC, Wiger N. The effect of tetracycline upon the spread of bacterialresistance from calves to man. J Anim Sci. 1978;46(5):1437–46.
114. Oppegaard H, Steinum TM, Wasteson Y. Horizontal transfer of a multi-drugresistance plasmid between coliform bacteria of human and bovine originin a farm environment. Appl Environ Microbiol. 2001;67(8):3732–4.
115. Simonsen GS, Haaheim H, Dahl KH, Kruse H, Løvseth A, Olsvik Ø, et al.Transmission of VanA-type vancomycin-resistant enterococci and vanAresistance elements between chicken and humans at avoparcin-exposedfarms. Microb Drug Resist. 1998;4(4):313–8.
116. Dorn CR, Tsutakawa RK, Fein D, Burton GC, Blenden DC. Antibiotic resistancepatterns of Escherichia coli isolated from farm families consuming home-raised meat. Am J Epidemiol. 1975;102(4):319–26.
117. Huddleston JR. Horizontal gene transfer in the human gastrointestinal tract:potential spread of antibiotic resistance genes. Infect Drug Resist. 2014 Jun20;7:167–76.
118. Trobos M, Lester CH, Olsen JE, Frimodt-Møller N, Hammerum AM. Natural transferof sulphonamide and ampicillin resistance between Escherichia coli residing inthe human intestine. J Antimicrob Chemother. 2009 Jan 1;63(1):80–6.
119. McCarthy AJ, Loeffler A, Witney AA, Gould KA, Lloyd DH, Lindsay JA.Extensive horizontal gene transfer during Staphylococcus aureus co-colonization in vivo. Genome Biol Evol. 2014 Oct 1;6(10):2697–708.
120. Nikolich MP, Hong G, Shoemaker NB, Salyers AA. Evidence for naturalhorizontal transfer of tetQ between bacteria that normally colonize humansand bacteria that normally colonize livestock. Appl Environ Microbiol. 1994;60(9):3255–60.
121. Varma JK, Mølbak K, Barrett TJ, Beebe JL, Jones TF, Rabatsky-Ehr T, et al.Antimicrobial-resistant nontyphoidal salmonella is associated with excessbloodstream infections and hospitalizations. J Infect Dis. 2005;191(4):554–61.
122. Krueger AL, Greene SA, Barzilay EJ, Henao O, Vugia D, Hanna S, et al. Clinicaloutcomes of Nalidixic acid, Ceftriaxone, and multidrug-resistantNontyphoidal salmonella infections compared with Pansusceptibleinfections in FoodNet sites, 2006–2008. Foodborne Pathog Dis.2014l;11(5):335–41.
123. Lee LA, Puhr ND, Maloney EK, Bean NH, Tauxe RV. Increase in antimicrobial-resistant salmonella infections in the United States, 1989–1990. J Infect Dis.1994;170(1):128–34.
124. Martin LJ, Fyfe M, Doré K, Buxton JA, Pollari F, Henry B, et al. Increasedburden of illness associated with antimicrobial-resistant salmonella entericaserotype Typhimurium infections. J Infect Dis. 2004;189(3):377–84.
125. Helms M, Vastrup P, Gerner-Smidt P, Molbak K. Excess mortality associatedwith antimicrobial drug-resistant salmonella typhimurium. Emerg Infect Dis.2002;8(5):490–5.
126. Helms M, Simonsen J, Olsen KEP, Mølbak K. Adverse health eventsassociated with antimicrobial drug resistance in campylobacter species: aregistry-based cohort study. J Infect Dis. 2005;191(7):1050–5.
127. Cosgrove SE, Sakoulas G, Perencevich EN, Schwaber MJ, Karchmer AW,Carmeli Y. Comparison of mortality associated with methicillin-resistant andmethicillin-susceptible Staphylococcus aureus bacteremia: a meta-analysis.Clin Infect Dis. 2003;36(1):53–9.
128. Jay KV, Katherine DG, Jessa O, Timothy JB, Felicita M, Frederick JA.Hospitalization and Antimicrobial Resistance in Salmonella Outbreaks, 1984–2002. Emerg Infect Dis J. 2005;11(6):943.
129. ADUFA Reports. http://www.fda.gov/ForIndustry/UserFees/AnimalDrugUserFeeActADUFA/ucm042896.htm. Accessed 12 July 2016.
130. Funk JA, Lejeune JT, Wittum TE, Rajala-Schultz PJ. The effect ofsubtherapeutic chlortetracycline on antimicrobial resistance in the fecal floraof swine. Microb Drug Resist. 2006;12(3):210–8.
• We accept pre-submission inquiries
• Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research
Submit your manuscript atwww.biomedcentral.com/submit
Submit your next manuscript to BioMed Central and we will help you at every step:
Hoelzer et al. BMC Veterinary Research (2017) 13:211 Page 38 of 38
