Risk Factors for Antimicrobial Resistance in Escherichia coli in Pigs Receiving Oral Antimicrobial Treatment: A Systematic Review

Abstract

The aim of this literature review was to identify risk factors in addition to antimicrobial treatment for antimicrobial resistance (AMR) occurrence in commensal Escherichia coli in pigs. A variety of studies were searched in 2014 and 2015. Studies identified as potentially relevant were assessed against eligibility criteria such as observation or experiment (no review), presentation of risk factors in addition to (single dosage) antimicrobial use, risk factors for but not resulting from AMR, and the same antimicrobial used and tested. Thirteen articles (nine on observational, four on experimental studies) were finally selected as relevant. It was reported that space allowance, production size/stage, cleanliness, entry of animals and humans into herds, dosage/frequency/route of administration, time span between treatment and sampling date, herd size, distance to another farm, coldness, and season had an impact on AMR occurrence. Associations were shown by one to four studies per factor and differed in magnitude, direction, and level of significance. The risk of bias was unclear in nearly half of the information of observational studies and in most of the information from experimental studies. Further research on the effects of specific management practices is needed to develop well-founded management advice.

Introduction

Antimicrobials are necessary for the treatment and prevention of infectious diseases to protect the health and welfare of pigs in the current pig production system.¹ In pig housing, antimicrobials have commonly been administered orally (>90% of all antimicrobial treatments²) through the feeding and drinking system. Having been the subject of ongoing debates, the antimicrobial consumption has been reduced in recent years, but still differs between (European) countries.³ In Germany, the treatment incidence rate per pig and half-year decreased fivefold between 2011 and 2014⁴ following the new legislation making registration of treatment incidence compulsory (German Medicines Act).⁵ Current incidence levels and possible changes in administration routes used are not available yet.

Oral administration of antimicrobials was found to increase antimicrobial resistance (AMR) of commensal Escherichia coli in swine.⁶ Resistant E. coli in swine can be a reservoir for resistance genes transferable to other pathogenic and zoonotic bacteria⁷ in humans. AMR can cause higher healthcare costs due to treatment failure and increased mortality.^8,9 For these reasons, it is important to identify risk factors relating and nonrelating to the oral antimicrobial treatments as a way to find key issues to limit resistance levels and hence to protect the therapeutic effectiveness of drugs in the treatment of infections in animals and humans. In the literature, several management and other factors are studied in terms of their relevance for AMR occurrence in E. coli in pigs receiving antimicrobial treatment.^10,11

The question arises for which (management) factors a significant association with AMR in fecal E. coli in pigs has already been confirmed by peer-reviewed articles. Therefore, based on a systematic review of the scientific literature, this study aimed to identify the specific risk factors for AMR occurrence in commensal E. coli in pigs receiving oral antimicrobial treatment.

Materials and Methods

Study population and exposure/outcome of interest

The study population considered relevant for this review was swine in the (farrow) wean-to-finish phase in settings of commercial and research farms. (Production stages without impact of the dam and closer to slaughtering/impact on consumption were aimed for, but studies including the farrowing/suckling stage were not excluded).

The exposure of interest was any factor that was not (single dosage) antimicrobial treatment, assessed for its impact on AMR (e.g., management and treatment practice, geographic, climatic, and temporal factors). We call these factors risk factors in this review, regardless of whether we are discussing exposure from observational or experimental studies. The effect of each risk factor was evaluated in the way that AMR occurrence was compared with the conditions before exposure (initial) or to nonexistence (control) of the risk factor.

The outcome of interest was the extent of AMR in E. coli from pig feces collected rectally or (pooled) from the barn floor.

Identifying the relevant literature

The literature on risk factors for AMR in E. coli in pigs following oral antimicrobial treatment was systematically reviewed. Relevant scientific articles published in peer-reviewed journals were identified using the keyword (stem) combinations (resist OR suscept OR sensit) AND (antibiot OR antibact OR antimicrobi) AND (sow OR swine OR pig OR piglet OR farrow OR wean OR porcin) AND coli AND (oral OR feed OR water OR drink) AND risk.

The online electronic database of the Deutsches Institut für Medizinische Dokumentation und Information (DIMDI 2014) was used (including databases defined in DIMDI presented at www.dimdi.de/dynamic/de/db/dbinfo/index.htm: ZT00; CC00; CCTR93; CDSR93; DAHTA; CDAR94; AR96; GA03; GM03; INAHTA; MK77; NHSEED; ED93; ME60; CV72; CB85; AZ72; IA70; BA70; EM47; DH64; EA08; DD83; II78; IS74). After standard adjustment in the DIMDI search system, duplicates were eliminated by selecting the article from the cheapest database. Additionally, PubMed and Web of Science were searched separately with the keywords (stems). The identified list of articles was then compared with the list of articles from DIMDI. Duplicate articles between DIMDI, PubMed, and Web of Science were attributed to the results from DIMDI.

Moreover, our previous review⁶ on the effects of oral antimicrobial administration on AMR in fecal E. coli in pigs was searched for relevant articles presenting risk factors in addition to (single dosage) antimicrobial use (AMU).

All searches were performed during May and June 2014 and were updated in July 2015. In terms of the date of publication, no restrictions were imposed. Citations of all identified studies were downloaded into a reference bibliography (EndNote X.7) and managed by means of a Microsoft Excel 2010 database.

Screening for key words

The relevance of the studies was, in a first screening step, determined from the title and the abstract, which was then compared to the keywords given. In a second screening step, articles were scanned for their discussion of oral administration and risk factors.

Eligibility assessment

Articles that passed the initial screening were further assessed against eligibility criteria. Only explorative (experimental and observational), but no review studies, were selected. In addition, the risk factor had to describe more than (single dosage) AMU. Studies presenting risk factors resulting from AMR (e.g., carryover to environment) rather than risk factors for AMR were not eligible.

The relevant studies were reviewed to obtain information on the following aspects: administered and tested antimicrobials; type of administration (route, individual, or by group); sample size; method and results from resistance testing; analysis level (descriptive or analytical); study type (observational or experimental); and details of the specific risk factor(s) investigated. All this information was fed into a Microsoft Excel 2010 database.

Articles that reported on studies or single trials, where the same or closely linked antimicrobials were used and tested (e.g., enrofloxacin used for treatment and ciprofloxacin tested), were considered to be relevant. Where studies tested more (and different) antimicrobials than those administered and vice versa, the results (and trials) without any direct links were not taken into account in the review. Moreover, an eligible study had to define and describe tested risk factors and evaluate potential association with AMR.

In case publications from identical authors evaluated samples and presented results of high overlap, the publication with the most relevant result(s) and with the lowest risk of bias was chosen for the final reference list. If necessary, however, additional references by the same authors were used and cited for the purpose of extracting relevant supplementary information. The lists of references of the selected articles were reviewed for additional eligible studies.

Once all eligible studies had been selected, their descriptions of factors and summary measures of the effect on resistance occurrence were surveyed for a qualitative synthesis. From the summary measures, test values in the form of odds ratios (ORs) or estimates were extracted from the articles where possible. Estimates from observational studies/logistic analyses were converted to ORs. Alternatively, if no test values were presented, prevalence values (e.g., percentage of isolates showing resistance to the antimicrobial according to a certain cutoff value) stated in the article were used instead. If an article did not present any result values, only information on the data analysis used in the study was taken into account.

For studies that evaluated AMU as risk factor for AMR and further variables as potential confounders, the specified quantitative effect of the (confounding) variables (potential risk factors from the review's point of view) on AMR was, if presented, obtained.

Alternatively, if the potential risk factors’ effect was not presented, qualitative information on the observed confounding effects was used. Conditions necessary for confounding include inter alia variables’ association with both AMU and AMR, and variables to be no intermediary step in the causal pathway from AMU to AMR. Therefore, potential risk factors stated to act as confounders were evaluated as being associated with AMR in this review. In contrast, for potential risk factors stated not to act as confounders, it was unclear whether they were associated with AMR and were hence described as being no identified confounders for AMU and AMR in the review.

Risk of bias

Two observers evaluated the risk of bias for each one of the selected studies using methods for experimental and observational studies developed by Higgins et al.¹² and Kim et al.¹³ As part of these methods, the following criteria were scored for low, high, or unclear risk of bias: selection, performance, detection, attrition, and reporting (Table 1).

Table 1.

Evaluated Risk of Bias for Six Bias Criteria in Nine Observational and Four Experimental Studies (Final Selection)

	Observational (n = 9)^a			Experimental (n = 4)^b
	Risk of bias			Risk of bias
Bias criteria	High	Low	Unclear	High	Low	Unclear
Selection
Selection of participants	0	2	7
Random sequence generation				0	0	4
Confounding variables	1	6	2
Allocation concealment				0	0	4
Performance
Measurement of exposure	0	3	6
Blinding of exposure allocation				0	0	4
Detection
Blinding of outcome assessments	0	3	6	0	0	4
Attrition
Incomplete outcome	0	7	2	1	2	1
Reporting
Selective reporting	0	8	1	0	2	2
Risk of bias total score	1	29	24	1	4	19

Bias evaluation of observational studies modified after.¹³

Bias evaluation of experimental studies modified after.¹²

After first screening and reading the articles, the observers met to agree more detailed bias definitions and subsequently used those for individual evaluation. A final meeting was held on scoring to discuss results that differed to the extent where one observer had evaluated the risk of bias of an article as high and the other observer as low in the same criterion. Both observers deliberated on their results until agreement on one classification was reached. If one observer had assigned a high or low bias score, whereas the other opted unclear bias, the more critical class was finally adopted.

Results

Selection of eligible studies

The process of the literature review is presented in a flow diagram (Fig. 1). In total, 257 records were initially identified (a list is accessible through the corresponding author). After removal of duplicates and screening for keywords in the title and abstract, 27 studies were assessed for their eligibility based on the complete document.

FIG. 1.

Flow diagram on process and numbers of articles in the literature review. DI, Deutsches Institut für Medizinische Dokumentation und Information; WOS, Web of Science; PM, PubMed; RP, review paper by Burow et al.⁶; REF, References from reference lists.

Of the 27 studies, 4 were ineligible for not being explorative. Another two studies were excluded because they did not evaluate additional risk factors besides (single dosage) AMU. Further reasons for exclusion were: no investigation of risk factors for resistance in pigs (but carry over of resistance from pigs to the environment or to other creatures; four studies); no registration of antimicrobial administration (one study); and not studying the resistance effect at all (one study). Four studies were not included in the final selection due to an overlap with other selected publications in terms of their subject of investigation and author. Three additional relevant articles were identified from the selected reference lists. From these, two studies (two articles with an overlap between study and author) were assessed as eligible.

Overall, 13 studies, 6 from DIMDI, 2 from Web of Science, 3 from former reviews, and 2 from the reference lists of selected studies met the eligibility criteria and were finally included.

Characteristics of the selected studies

Of the 13 studies, 9 were of the observational and 4 of the experimental type (Tables 1 and 2). All the finally selected studies (Table 2) were either conducted in North America (n = 9) or Europe (n = 4). The observational studies took place, where indicated, between 1999 and 2004. None of the experimental studies specified a time period. For most studies, pigs at the production stages of weaners and finishers were sampled by taking voided and pooled samples. In the observational studies, multiple antimicrobials were administered to the herds. Most of these studies also tested for multiple antimicrobials, except for two studies, one focusing on tetracycline¹⁴ and the other on fluoroquinolone^10,15 resistance only. The experimental studies concentrated on tetracycline administration and tetracycline or multiple resistances as well as apramycin administration and resistance.

Table 2.

Characteristics of the 13 Finally Selected Studies

Study type	Time of sampling	Sampled animals (no. and type of herds)	Feces sampling and Escherichia coli isolates	Antimicrobial use	Antimicrobial tested	State, reference
Observational	March–September 1999 (Ontario), May–August 2000 (British Columbia), twice per farm	Weaners, finishers (47 farrow-to-finish with ≥50 sows)	10 samples from weaners pooled per farm visit, in total 94 pooled weaner samples (same for finishers), 5 isolates per pooled sample	Multiple	Multiple^a	Canada¹⁹ see also²⁶
Observational	January 2004–December 2006, monthly per farm	Piglets, growers, finishers, boars (5 farrow-to-finish, 7 grow-to-finish)	Floor sample, pooled per pen (∼140 samples) per month, in total 1,239 isolates from wastewater and fecal samples	Multiple	Multiple^a	Texas,²⁸ see also²⁴
Observational	March–October 2003, once per herd	800 piglets at 72 days, 800 growers at 125 days, 800 finishers at 186 days of average age (50 herds with ≥150 sows, ≥600 fattening pigs)	1 individual sample pooled per pen (4 pigs from 4 pens of piglets, growers, and finishers), 2 isolates per pooled sample	Tetracycline (documented)	Multiple^a	Belgium²⁵
Observational	2 samplings per herd: 1 month before and shortly before slaughter	Finishers (34 farrow-to-finish)	20 samples per herd and visit, 8,119 tested isolates	Part of the herds: multiple	Multiple^a	Canada²⁰
Observational	Once per farm	Finishers (54 finish, each ∼960 pigs)	Voided samples from floor of a pen, maximal 36 samples per barn	Ceftiofur plus multiple	Multiple^b	North Carolina¹⁶
Observational	May–September 2004, once per herd	100 clinically healthy growers, finishers (20 herds with ≥100 sows)	Voided sample from 5 pigs of 20 pens, pooled per pen, 405 samples, on average 63 isolates per herd, in total 1,439 isolates	Multiple	Multiple^a	Canada¹¹
Observational	December 2002– October 2003, once per farm	Finishers (59 breeding-to-finish with ≥100 females, 49 grow-to-finish with ≥100 places)	8–30 pooled samples per farm, number of isolates not presented	Multiple, part of the farms: fluoroquinolone	Fluoroquinolone^a	Great Britain,¹⁰ see also¹⁵
Observational	May–September 2000, three times per farm at 4–6 weeks intervals	Finishers (76 farrow-to-finish, multisite, 6 grow-to-finish, 2 finishing)	Pooled sample by pen from 5 pens per visit, in total 269 farm-visit-pools, ∼1,349 samples, 1,147 isolates	Multiple	Multiple^a	Canada²⁷
Observational	January 2003– December 2005, once, twice, three times or four times per herd	Slaughter pigs (558 unknown kind)	1 sample of cecal content per herd and sampling occasion, in total 593 studied isolates (26 herds with 2, 3 with 3, 1 with 4 isolates)	Multiple	Tetracycline^a	Danmark¹⁴
Experimental	7-days-treatment, sampled on 5 days during and on 8 days posttreatment	18 weaners at 7 weeks of age (from 3 litters)	Individual sample, 6× during treatment, 8× posttreatment, in total 252 samples, 423 isolates	Chlortetracycline (Aureomycin)	Multiple^a	Great Britain²¹
Experimental	3× during 21-day-adjustment, 5× during 14-day-therapy, 5× during 71-days post-therapy	36 weaners at 7–9 weeks of age (from 2 herds, one herd without antimicrobial use for 8 years)	Voided sample from each pig in a pen, pooled per pen, number of isolates not presented	Mainly chlortetracycline	Chlortetracycline^a	Kentucky¹⁸
Experimental	5–141-days-treatment, sampled 2-days pre-, 3–5-days during, 1–4-days post- treatment	70 weaners at 25 days of age (from 1 source)	Individual swab sample, 1× pre-, 1–6× during, 1–4× posttreatment, ≥4 confirmed E. coli per sample chosen	Apramycin	Apramycin^a,b	Tennessee¹⁷
Experimental	2 samplings in ∼9 weeks interval	Feeders at beginning of feeding period ∼9 weeks before market weight (from 1 source)	6–7 voided samples from ∼⅓ of pigs per pen in 21 barns, in total 1,296 isolates tested, for continuous, pulse, and no chlortetracycline treatment 113, 121, 118 isolates from start-feeding period and 126, 146, and 128 isolates from preslaughter period tested	Chlortetracycline, tylosin^a	Multiple	North Colorado²²

The minimum inhibitory concentration was used to evaluate resistance occurrence.

Polymerase chain reaction was used to evaluate resistance gene occurrence.

Minimum inhibitory concentration was used to evaluate resistance occurrence in all studies, with the exception of two investigations,^16,17 which (also) used polymerase chain reaction for resistance gene detection (Table 2).

Most studies used a significance level of 5% as the inclusion criterion for exposure in the modeling procedure. However, Taylor et al. described a significance level of 10%,¹⁰ whereas Langlois et al.¹⁸ and Lutz et al.¹⁶ did not present any significance criterion.

Effect of risk factors

A range of risk factors for AMR was identified from the 13 articles. The risk factors were grouped into the following categories: farm/herd management (Table 3); treatment management (Table 4); and unmodifiable aspects (geographic, climatic, and temporal aspects; Table 5). A detailed list of information on identified risk factors is presented in Supplementary Tables S1–S3 (Supplementary Data are available online at www.liebertpub.com/mdr).

Table 3.

Factors of Farm/Herd Management for Antimicrobial Resistance in Escherichia coli from Swine (Presented in the Finally Selected Studies)

	Resistance
Factor and its direction/categorical comparison	Increase	Decrease	Equal	(Direction) unclear
Herd size increased	(Carbadox)²⁰	¹⁴		(No confounder)¹¹, (confounder)¹⁹, (Most antimicrobials, no confounder)²⁰
Density increased, space reduced	¹⁷			(No confounder)¹¹
Flooring system slatted vs. (partly) nonslatted			²⁵	(No confounder)¹¹
Feeding presented in feeder vs. on floor				(No confounder)¹¹
Age/production stage increased		^19,24		(No confounder)¹¹
Selling for consumption vs. breeding				¹¹
Weaning age increased	(Tendency of significance)²⁵
Length of stay in pen/production phase				(No confounder)¹¹
Building, farm separated vs. nonseparated		(Most antimicrobials)²⁰		(No confounder)¹¹, (Some antimicrobials, no confounder)²⁰
Restocking system with increased biosecurity			²⁵	(No confounder)¹¹
Cleaning, disinfection, hygiene increased	²⁵		¹⁷	(No confounder)¹¹
Relocation recently yes vs. no			²⁵
New animal introduction yes vs. no/less	¹⁷, (Most antimicrobials)²⁰			(Some antimicrobials, no confounder)²⁰
Visitors without pig contact within last 2 days		¹⁰

Table 4.

Factors of Treatment Management for Antimicrobial Resistance in Escherichia coli from Swine (Presented in the Finally Selected Studies)

	Resistance
Factor and its direction/categorical comparison	Increase	Decrease	Equal	(Direction) unclear
Dosage increase	During/shortly after treatment^18,21,23	Longer after treatment^18,23
Pulse vs. continuous		²³	²²
Frequency low vs. high		¹⁶
Interval between last treatment and sampling increased		¹⁴
Administration route: individual/parenteral vs. oral in-feed (in groups)		^20,27

Table 5.

Unmodifiable Aspects for Antimicrobial Resistance in Escherichia coli from Swine (Presented in the Finally Selected Studies)

	Resistance
Factor and its direction/categorical comparison	Increase	Decrease	Equal	(Direction) unclear
Another pig farm in the vicinity	¹⁰
Poultry farm in the vicinity		¹⁰
Province (region)			¹⁹
Coldness	¹⁷
Heat			¹⁷
Summer vs. autumn/winter season	¹⁰
Season/year			²⁴
Year			¹⁹	¹⁴

For most factors, an association between a factor and AMR was reported either in one or two studies. Only for one factor (production/herd size) was such an association reported in four studies. Where more than one study reported on a specific factor, the studies differed in terms of the magnitude, significance level, and direction of association (as was the case, e.g., for production size^11,14,19,20 and dosage^18,21–23).

Several studies evaluated AMU as the main risk factor for AMR and further risk factors as potential confounders^11,20,24 (Supplementary Tables S1 and S3).

Farm/herd management

In the context of farm/herd management, several studies examined production-related factors (Table 3 and Supplementary Table S1).

Herd size differed in significance level and direction of association with (AMU and) AMR between the relevant studies.^11,14,19,20 Vieira et al., who studied a relatively large sample of Danish pig herds (n = 558; Table 2), found that the more slaughter pigs (1–200, 201–1,000, 1,001–3,000 vs. >3,000) a Danish herd contained, the lower was the risk of finding a pig with resistant E. coli.¹⁴ However, the mean number of slaughter pigs in a herd in a year was similar for a herd evaluated as resistant (n = 3,762.20) or as susceptible (n = 3,717.30).¹⁴ Moreover, Dewulf et al. did not find any significant effect by the number of animals per herd or per compartment (univariable analysis).²⁵

Dunlop et al. found higher carbadox resistance rates in larger herds (examined farrow-to-finish farms with 31–90, 91–120, 121–150, or >150 sows). However, resistance to six additionally examined antimicrobials did not show any clear direction of association with herd size.²⁰ The number of sows was no confounder (range 274–1,042, median 456 sows among 20 observed herds) in relation to a range of antimicrobials.¹¹ Akwar et al. found an association between the number of pigs on a farm (range: 50–1,400; mean± standard deviation: 393 ± 9) and AMR (OR = 1).^18,28

Dewulf et al. (univariable analysis) and Rosengren et al. also examined the factor density of animals per square meter and per pen and did not find any effect²⁵ and no confounding effect for AMU and AMR.¹¹ Mathew et al. found a long-lasting increase of resistance in pigs housed in a limited space.¹⁷ Ventilation type and fully slatted versus partly slatted floor did not affect AMR.²⁵ In the study by Rosengren et al., flooring type and feed presentation did not act as confounders for AMU and AMR.¹¹

Concerning production (age) groups, two relevant studies found younger pigs (nursery piglets, weaners) to be at higher risk for carrying resistant E. coli compared with older rearing pigs (slaughtered pigs, finishers).^19,24 Dewulf et al. found the same, but insignificant result (univariable analysis).²⁵ Rosengren et al. discovered that the difference between a herd/production type that sells pigs for breeding (breeding stock) as opposed to slaughtering (commercial) had an effect on AMR. However, the direction of association was inconsistent between the text and table in the article.¹¹ Increased weaning age was close to being significantly associated with increased AMR.²⁵ The number of days and weeks pigs spent in each production phase was no confounder for AMU and AMR.¹¹

A range of biosecurity-related factors were studied (Table 3 and Supplementary Table S1). Building separation inconsistently reduced resistance to a range of tested antimicrobials.²⁰ Farm separation was no confounder for AMU and AMR.¹¹ Neither the restocking system used nor cleaning with increased biosecurity showed any effect on AMR.^17,25 Nor were these factors confounders for AMU and AMR, and the same applies to disinfection.¹¹ Poor inside pen hygiene was associated with lower AMR.²⁵ Recent relocation did not show any effect.²⁵ New animal introduction increased AMR,¹⁷ whereas purchase of only boars, but not gilts led to lower resistance prevalence to most antimicrobials.²⁰ With regard to visitors’ conditions when entering a barn, visitors free from pig contact over the 2 days before the visit were associated with a lower risk of resistance.¹⁰

Treatment management

In the literature identified as relevant, the following treatment management factors were evaluated for their impact on AMR: dosage, frequency, interval to last treatment, and administration route (Table 4 and Supplementary Table S2).

Higher dosages were associated with higher AMR during and up to 2 weeks after treatment.^18,21,23 This association reversed (2–4 weeks) posttreatment.^18,23 The findings were similar for pigs from herds that had not been treated for 8 years and those that had previously been treated.¹⁸ Pulse treatment led to either low resistance occurrence²³ or, in a way similar to dosage regimes, differences in resistance, with prevalence between pulse treatment and low-level continuous treatment decreasing after about 9 months.²² Rare use of ceftiofur was found to be linked with lower resistance odds than frequent ceftiofur treatment.¹⁶ In addition, the longer the interval between the last treatment and the sampling date, the lower the risk of resistance.¹⁴

Concerning the administration route, some studies observed individual and parenteral treatment in addition to the common oral in-feed application. Individual and parenteral treatments were found to be associated with a lower risk of resistance. Finally, in-feed antimicrobial treatment was more consistently associated with an increased risk of resistance than individual animal treatment.^20,27

Unmodifiable (geographic, climatic, and temporal) aspects

The group of unmodifiable aspects consisted of geographic, climatic, and temporal aspects (Table 5 and Supplementary Table S3). Whereas the existence of another pig farm in the vicinity increased the risk of resistance, a poultry farm nearby decreased the risk.¹⁰ Province (region) in Ontario did not show any significant effect on resistance.¹⁹

Compared with optimal temperatures, the climatic factor, coldness, led to a long-lasting increase in resistance, whereas heat did not show any effect.¹⁷ Conversely, the summer season was found to be associated with a substantially higher resistance rate than the autumn/winter season in a study conducted in Great Britain.¹⁰ In contrast, a study carried out in Texas did not find a common seasonal resistance trend over a 3-year period.^24,28 The year a farm visit took place (1999 vs. 2000) did not show any difference in resistance in a study undertaken in Ontario.¹⁹ In the findings by Vieira et al. (Denmark), resistance differed over the 3 observed years 2003–2005, but without any consistent trend across the years.¹⁴

Risk of bias

Of the observational studies, one half of the assessed bias criteria were evaluated to be at low and the other half at unclear risk (Table 1). The selection of participants and the blinding (of exposure allocation and outcome assessments), in particular, were unclear in the majority of the studies. The experimental trials were evaluated as being at unclear risk of bias for most criteria (Table 1).

Discussion

Characteristics of the selected studies

Thirteen observational and experimental studies were finally selected that present risk factors for AMR in E. coli in pigs receiving oral antimicrobial treatment. The observational studies were conducted 5–10 years ago (Table 2), thus reflecting risk factors in still current pig production systems. All selected studies took place either in Europe or North America (Table 2). Conclusions from the findings may therefore be drawn mainly for pig production systems of the same or similar conditions.

The focus was predominantly on young pigs, especially weaners (Table 2). In the period after birth and weaning, when the animals are challenged by physical and behavioral adaptation, the risk for diseases and treatments is highest.²⁹ Hence, a focus on risk factors for AMR in young pigs may be crucial. However, given the nature of consumer concerns, the inclusion of older pigs (close to slaughter) may be highly relevant.

Effects of the various risk factors

Magnitude, significance level, and direction of association between a given factor and AMR differed between different studies. One reason for this could be that the relations between the various factors and AMR are rather complex. Due to the challenging nature of these relations, the possibilities of presenting a full picture in explorative studies are (so far) limited.

Factors may be associated directly with AMR occurrence, and/or indirectly through production and health issues affecting AMU.^27,30 The route of association with AMU may often be neither direct nor indirect, but a mixture, in the same way that the association can, to a certain extent, be direct and indirect with AMR. For instance, an improved biosecurity measure can directly reduce transmission of AMR, but also indirectly reduce transmission of diseases, hence lead to a decrease in further AMU and finally result in a reduction in AMR. In this review, we simplified matters to focus on the association of risk factor's with AMR, but also discuss and compare these factors with reported risk factors for AMU.

Farm/herd management

In terms of farm/herd management, the literature identified the following main factors as having an impact on AMR: herd size, space/density, production group/age, cleanliness, animal purchase/introduction and visitor's precondition (Table 3 and Supplementary Table S1).

Although herd size is commonly assumed to be positively linked with AMR, it did not show any clear direction of association across the reviewed studies.

Vieira et al. in fact found a negative link in a relatively large number and range of herd sizes.¹⁴ The authors explain this negative link thus: “This can be correlated to the larger treatment incidence rates associated with these smaller herds where, even when few pigs are medicated, they represent a larger proportion of the total number of slaughter pigs produced in that herd.”¹⁴ It might be that this effect applies to large herd sizes of 3,000 and more pigs as studied in the Danish condition, although Vieira et al. examined only one isolate per herd (Table 2). The chance of finding a resistant isolate may be relatively lower in large herds when the animal prevalence is low. However, according to Hering et al., few fecal samples were sufficient to determine the prevalence of cefotaxime-resistant E. coli for risk factor analysis in pig farms.³¹

In contradiction, Dunlop et al. found a positive link between herd size and carbadox resistance in farrow-to-finish herds with up to 150 and more sows.²⁰ Moreover, herd size and AMU were also reported to be positively associated, even though to some extent they were described as being indirectly associated through biosecurity.^29,32,33 Since the remaining relevant studies and findings did not show any significant or clear effect,^11,19,20,25 no obvious conclusion can be drawn from the reviewed studies that would allow a specific management recommendation on herd size for the purpose of reducing AMR.

Reduced space was found to increase AMR in the longer term,¹⁷ whereas animal density did not show any significant effect²⁵ or was unclearly reported.¹¹ Whether and to what extent space was directly (spread of AMR) or indirectly (e.g., through pathogenic pressure, health, AMU) associated with AMR cannot be determined from the literature. Further knowledge on the association and causal pathway between management factors and AMR is needed.

Younger rearing pigs were at higher risk of carrying resistant E. coli than older rearing pigs according to most of the relevant studies.^19,24,25 This higher AMR in young pigs may be related to the similarly higher disease incidence and AMU found in younger pigs.^11,29,34

Regarding biosecurity-related risk factors, inside pen hygiene was associated with AMR in the same way that poor hygiene was linked with lower AMR occurrence.²⁵ The opposite is perhaps to be expected.

Surprisingly, further biosecurity-related factors such as cleaning, washing, disinfection, restocking, rodent control, ventilation type, and flooring system, all of which could reasonably be assumed to impact on E. coli occurrence and resistance risk, did not show any significant effect on AMR.^10,17,25 Moreover, the study of risk factors for cefotaxime resistance conducted by Hering et al. found hygiene (related) measures such as separation of sick pigs, ventilation location, fly control, disinfection, and stable-assigned clothes to be associated with significantly increased AMR.³¹ (This study did not fulfill the eligibility criteria of the search procedure and was therefore not included in the final review list).

These unexpected results may reflect that a long persistence of resistant bacteria or a higher intake of susceptible E. coli from dirty environments may contrast the impact of increased biosecurity activities. It may also be that, in the way of reversed causation, managers of herds with increased incidence of health problems and consequently AMU, have taken increased biosecurity action. Hence, these biosecurity measures may not have caused increase of AMR, but are a consequence of increased AMR (and may not yet have led to an observable decrease of AMR).

New animal introduction increased¹⁷ and less animal purchase decreased²⁰ AMR prevalence. Congruently, an increased number of animal origins per herd was found to be associated with an increased number of extended-spectrum beta-lactamase-producing Enterobacteriaceae in cattle and dairy cows.^35,36 More animal movements may increase the risk of disease introduction, AMU, and thereby AMR. It is not surprising, then, that integrated herds (with sows and finishers) were reported to use less antimicrobials than conventional herds (with only finishers) that purchase more pigs from different herds with different disease status.³³ Moreover, farms whose visitors had not visited other pig herds in the 2 days before their visit have a lower risk for AMR.¹⁰

Sternberg Lewerin et al. modeled the effect of different biosecurity factors on the transmission of an enteric disease and found measures such as quarantine for new animals and protective clothes for visitors to halve the risk in fattening pig herds and reduce it by five times in farrow-to-finish herds.³⁷ Awareness of the importance of external biosecurity issues was identified as being already high in Swedish farmers who, in a questionnaire, assigned high scores to animal purchase and visitor protocols.³⁸ All these findings underline the importance of preconditions and restrictions on the entry of external animals and humans into herds and reveal the need for further knowledge on the link between internal biosecurity measures and AMR occurrence.

Treatment management

Most studies found a positive link between treatment management criteria (dosage, frequency, consistency interval, and route) and AMR (Table 4 and Supplementary Table S2). However, the use of less antimicrobials does not seem to lead to less AMR under all circumstances, since AMR decreased faster in therapeutically treated pigs after finalizing treatment than in subtherapeutically treated pigs.^18,23 Preconditions in terms of treatments administered previously, even to pigs’ dams,³⁹ play a role in this context and may have affected the observed associations. Unfortunately, most of the reviewed studies did not give a description of the animal's origin nor of previous treatment routines in breeding herds.

Treatment of animals from one herd routinely fed on antibiotic supplements led to higher AMR compared with a herd that had not received treatment over the last 8 years.¹⁸ An increase in the time span between the last treatment and sampling was associated with a decrease in AMR.¹⁴ The findings of Vieira et al. suggest that a prolonged waiting time between the last treatment and slaughtering of animals should be considered as a protective measure for reducing the transfer of resistant bacteria into the food chain. Overall, since every dosage and increased frequency of treatment was found to increase AMR, AMU should generally be avoided if possible.

Individual, parenteral treatment was found to be beneficial compared with feed (group) treatment^20,27 in being more consistently associated with lower risks of resistance. This finding would support an approach whereby pigs are preferably treated individually. It also suggests that pig farmers should consider parenteral treatment and that they restrict the treatment of diseased pigs. Thereby, fewer animals would come into direct contact with antimicrobials and the risk for carry over ⁴⁰ (this study was rejected in the review's screening procedure and therefore not included in the final review list) may be reduced.

However, since the effects of the administration route were not compared with the same age group of pigs nor for the same antimicrobial in this study, specific knowledge on this aspect is still lacking. In case of unavoidable treatment, the individual parenteral administration route can be advantageous.

Unmodifiable (geographic, climatic, and temporal) aspects

The existence of another pig farm nearby increased the risk of AMR (Table 5 and Supplementary Table S3),¹⁰ probably through emission/transmission of disease pathogens or commensals carrying resistance genes through fecal or airborne transfer^41–44 (von Salviati et al.⁴³ was rejected in the review's screening procedure and, therefore, not included in the final review list). In this context, biosecurity measures, even though not all showed a significant effect in the reviewed studies, have great importance in preventing infections and reducing the risk of resistance carry over.

The existence of poultry farms in the neighborhood was linked with less AMR (Table 5),¹⁰ which may indirectly be associated with a lower chance of pig farms in the neighborhood. Although, in principle, there could be the same risk for transmission of resistant bacteria or resistance genes from poultry farms to pig farms, “the local environment may have been seeded with avian strains of E. coli, which may have had a reduced ability to colonize pigs.”¹⁰ According to Taylor et al., a distance of at least 1.6 km should be kept when planning a new pig farm.¹⁰

Whereas coldness increased AMR occurrence, heat did not have any effect (studied in Tennessee, North America, of Mathew et al.¹⁷). Conversely, in a study conducted in Great Britain, Taylor et al. found higher AMR in summer compared with autumn/winter time.¹⁰ The pigs in the trials of Mathew et al. were exposed to Salmonella typhimurium, which could have impacted AMR occurrence. Taylor et al. explained the higher resistance risk in summer with the seasonality of enteric disease in British pigs.¹⁰ Annual differences in AMR prevalence were reported from Denmark, but without any consistent trend across the years.¹⁴

Seasonal and annual effects on AMR occurrence was only found in the European,^10,14 but not the North American^19,24 observational studies (Tables 2 and 5 and Supplementary Table S3). One reason for this difference may be that climatic conditions and/or housing management differ between the countries of the continents. However, the climate did have an effect in that optimal and balanced climatic conditions appear to have the potential to limit AMR occurrence in pig production systems.

Risk of bias

Most of the relevant finally selected studies were evaluated as having an unclear risk of bias in most of the evaluated criteria (Table 1). This unclear risk of bias stems from missing or unclear information leading to a limitation in the amount of information that can be gleaned from the studies. In consequence, some of the presented risk factors may be biased to some degree. This finding calls for higher quality standards in both explorative study presentation and in the peer-reviewing process of future publications.

Conclusion

In the reviewed studies, a range of risk factors were investigated. Associations between a specific factor and AMR occurrence often differed across studies or were based on only one study. The quality of the studies (and their presentation) partly limited their interpretation. Therefore, we agree with Murphy et al. who concluded from their review of modifiable nonantimicrobial factors for AMR in pigs, cows, chicken, and dogs³⁰ that more comprehensive research is needed before any full conclusions can be drawn and recommendations made. Currently, the complex topic of risk factors for AMR in E. coli in pigs from different farming systems in different countries is not adequately investigated. More global and specific surveillance on AMU, non-AMU factors, and AMR in individual pig production farms is needed.

Multiple antimicrobials were used and tested in most of the reviewed studies. Further research is necessary that additionally considers the more complex relations of possible cross-resistance and coselection.

However, the present review points to important individual factors (space allowance, production size and stage, hygiene, entry of animals and humans into herds, dosage/frequency and route of administration, time span between the last treatment and sampling, distance to another farm, and climate control). These appear to be promising key issues worth investigating further due to their potential to limit AMR occurrence in commensal E. coli in pigs.

Footnotes

Acknowledgments

The authors wish to thank librarian Antje Doerendahl for helping with the literature search. This study was funded by the Federal Institute for Risk Assessment, Grant No. 1322-532.

Disclosure Statement

No competing financial interests exist.

References

Ungemach

F.R.

, Mueller-Bahrdt

, and Abraham

2006. Guidelines for prudent use of antimicrobials and their implications on antibiotic usage in veterinary medicine. Int. J. Med. Microbiol., 296:33–38.

Merle

, Hajek

, Kasbohrer

, Hegger-Gravenhorst

, Mollenhauer

, Robanus

, Ungemach

F.R.

, and Kreienbrock

2012. Monitoring of antibiotic consumption in livestock: a German feasibility study. Prev. Vet. Med., 104:34–43.

European Medicines Agency, European Surveillance of Veterinary Antimicrobial Consumption. 2015. Sales of veterinary antimicrobial agents in 26 EU/EEA countries in 2013. Fifth ESVAC report. EMA/387934/2015. Available at http://www.ema.europa.eu/docs/en_GB/document_library/Report/2015/10/WC500195687.pdf (Online.)

Kreienbrock

, Hemme

, and Hartmann

2015. Aktueller Stand der Untersuchungen zum Antibiotikaeinsatz in der Tierhaltung. Abstract presented at the 2nd BfR-Symposium Antibiotikaresistenz in der Lebensmittelkette, Berlin, November. 2–3. Abstract no. 3.8. (In German.)

Bundesministerium für Ernährung und Landwirtschaft (BMEL). 2013. Pressemitteilung Nr. 278 vom 16.10.13. Weniger Antibiotika in der Tierhaltung: Novelliertes Arzneimittelgesetz verkündet. www.bmel.de/SharedDocs/Pressemitteilungen/2013/278-KL-Verkuendung-16_AMG-Novelle.html (Online.).

Burow

, Simoneit

, Tenhagen

B.A.

, and Kasbohrer

2014. Oral antimicrobials increase antimicrobial resistance in porcine E. coli—a systematic review. Prev. Vet. Med., 113:364–375.

Hunter

J.E.B.

, Shelley

J.C.

, Walton

J.R.

, Hart

C.A.

, and Bennett

. 1992. Apramycin resistance plasmids in Escherichia coli—possible transfer to Salmonella typhimurium in calves. Epidemiol. Infect., 108:271–278.

Ciorba

, Odone

, Veronesi

, Pasquarella

, and Signorelli

2015. Antibiotic resistance as a major public health concern: epidemiology and economic impact. Ann. Ig., 27:562–579.

Cohen

M.L.

1992. Epidemiology of drug-resistance—implications for a postantimicrobial era. Science, 257:1050–1055.

10.

Taylor

N.M.

, Clifton-Hadley

F.A.

, Wales

A.D.

, Ridley

, and Davies

R.H.

2009. Farm-level risk factors for fluoroquinolone resistance in E. coli and thermophilic Campylobacter spp. on finisher pig farms. Epidemiol. Infect., 137:1121–1134.

11.

Rosengren

L.B.

, Waldner

C.L.

, Reid-Smith

R.J.

, Dowling

P.M.

, and Harding

J.C.

2007. Associations between feed and water antimicrobial use in farrow-to-finish swine herds and antimicrobial resistance of fecal Escherichia coli from grow-finish pigs. Microb. Drug Resist., 13:261–269.

12.

Higgins

J.P.T.

, Altman

D.G.

, Gotzsche

P.C.

, Juni

, Moher

, Oxman

A.D.

, Savovic

, Schulz

K.F.

, Weeks

, and Sterne

J.A.C.

; Cochrane Bias Methods Group and Cochrane Statistical Methods Group. 2011. The Cochrane Collaboration's tool for assessing risk of bias in randomised trials. Br. Med. J., 343:9.

13.

Kim

S.Y.

, Park

J.E.

, Lee

Y.J.

, Seo

H.J.

, Sheen

S.S.

, Hahn

, Jang

B.H.

, and Son

H.J.

2013. Testing a tool for assessing the risk of bias for nonrandomized studies showed moderate reliability and promising validity. J. Clin. Epidemiol., 66:408–414.

14.

Vieira

A.R.

, Houe

, Wegener

H.C.

, Wong

, and Emborg

H.D.

2009. Association between tetracycline consumption and tetracycline resistance in Escherichia coli from healthy Danish slaughter pigs. Foodborne Pathog. Dis., 6:99–109.

15.

Taylor

N.M.

, Davies

R.H.

, Ridley

, Clouting

, Wales

A.D.

, and Clifton-Hadley

F.A.

2008. A survey of fluoroquinolone resistance in Escherichia coli and thermophilic Campylobacter spp. on poultry and pig farms in Great Britain. J. Appl. Microbiol., 105:1421–1431.

16.

Lutz

E.A.

, Mc Carty

M.J.

, Mollenkopf

D.F.

, Funk

J.A.

, Gebreyes

W.A.

, and Wittum

T.E.

2011. Ceftiofur use in finishing swine barns and the recovery of fecal Escherichia coli or Salmonella spp. Resistant to Ceftriaxone. Foodborne Pathog. Dis., 8:1229–1234.

17.

Mathew

A.G.

, Arnett

D.B.

, Cullen

, and Ebner

P.D.

2003. Characterization of resistance patterns and detection of apramycin resistance genes in Escherichia coli isolated from swine exposed to various environmental conditions. Int. J. Food Microbiol., 89:11–20.

18.

Langlois

B.E.

, Dawson

K.A.

, Stahly

T.S.

, and Cromwell

G.L.

1984. Antibiotic-resistance of fecal coliforms from swine fed subtherapeutic and therapeutic levels of chlortetracycline. J. Anim. Sci., 58:666–674.

19.

Akwar

H.T.

, Poppe

, Wilson

, Reid-Smith

R.J.

, Dyck

, Waddington

, Shang

, and McEwen

S.A.

2008. Associations of antimicrobial uses with antimicrobial resistance of fecal Escherichia coli from pigs on 47 farrow-to-finish farms in Ontario and British Columbia. Can. J. Vet. Res., 72:202–210.

20.

Dunlop

R.H.

, McEwen

S.A.

, Meek

A.H.

, Clarke

R.C.

, Black

W.D.

, and Friendship

R.M.

1998. Associations among antimicrobial drug treatments and antimicrobial resistance of fecal Escherichia coli of swine on 34 farrow-to-finish farms in Ontario, Canada. Prev. Vet. Med., 34:283–305.

21.

Delsol

A.A.

, Anjum

, Woodward

M.J.

, Sunderland

, Roe

J.M.

, Delsol

A.A.

, Anjum

, Woodward

M.J.

, Sunderland

, and Roe

J.M.

2003. The effect of chlortetracycline treatment and its subsequent withdrawal on multi-resistant Salmonella enterica serovar Typhimurium DT104 and commensal Escherichia coli in the pig. J. Appl. Microbiol., 95:1226–1234.

22.

Wagner

B.A.

, Straw

B.E.

, Fedorka-Cray

P.J.

, and Dargatz

D.A.

2008. Effect of antimicrobial dosage regimen on Salmonella and Escherichia coli isolates from feeder swine. Appl. Environ. Microbiol., 74:1731–1739.

23.

Mathew

A.G.

, Jackson

, and Saxton

A.M.

2002. Effects of antibiotic regimens on resistance of Escherichia coli and Salmonella serovar Typhimurium in swine. J. Swine Health Prod., 10:7–13.

24.

Alali

W.Q.

, Scott

H.M.

, Harvey

R.B.

, Norby

, Lawhorn

D.B.

, and Pillai

S.D.

2008. Longitudinal study of antimicrobial resistance among Escherichia coli isolates from integrated multisite cohorts of humans and swine. Appl. Environ. Microbiol., 74:3672–3681.

25.

Dewulf

, Catry

, Timmerman

, Opsomer

, de Kruif

, and Maes

2007. Tetracycline-resistance in lactose-positive enteric coliforms originating from Belgian fattening pigs: degree of resistance, multiple resistance and risk factors. Prev. Vet. Med., 78:339–351.

26.

Akwar

H.T.

, Poppe

, Wilson

, Reid-Smith

R.J.

, Dyck

, Waddington

, Shang

, and McEwen

S.A.

2008. Prevalence and patterns of antimicrobial resistance of fecal Escherichia coli among pigs on 47 farrow-to-finish farms with different in-feed medication policies in Ontario and British Columbia. Can. J. Vet. Res., 72:195–201.

27.

Varga

, Rajic

, McFall

M.E.

, Reid-Smith

R.J.

, Deckert

A.E.

, Checkley

S.L.

, and McEwen

S.A.

2009. Associations between reported on-farm antimicrobial use practices and observed antimicrobial resistance in generic fecal Escherichia coli isolated from Alberta finishing swine farms. Prev. Vet. Med., 88:185–192.

28.

Alali

W.Q.

, Scott

H.M.

, Christian

K.L.

, Fajt

V.R.

, Harvey

R.B.

, and Lawhorn

D.B.

2009. Relationship between level of antibiotic use and resistance among Escherichia coli isolates from integrated multi-site cohorts of humans and swine. Prev. Vet. Med., 90:160–167.

29.

van Rennings

, von Munchhausen

, Ottilie

, Hartmann

, Merle

, Honscha

, Kasbohrer

, and Kreienbrock

2015. Cross-sectional study on antibiotic usage in pigs in Germany. PLoS One, 10:e0119114.

30.

Murphy

C.P.

, Fajt

V.R.

, Scott

H.M.

, Foster

M.J.

, Wickwire

, and McEwen

S.A.

2015. Scoping review to identify potential non-antimicrobial interventions to mitigate antimicrobial resistance in commensal enteric bacteria in North American cattle production systems. Epidemiol. Infect., 144:1–18.

31.

Hering

, Hille

, Fromke

, von Munchhausen

, Hartmann

, Schneider

, Friese

, Roesler

, Merle

, and Kreienbrock

2014. Prevalence and potential risk factors for the occurrence of cefotaxime resistant Escherichia coli in German fattening pig farms—a cross-sectional study. Prev. Vet. Med., 116:129–137.

32.

Laanen

, Persoons

, Ribbens

, de Jong

, Callens

, Strubbe

, Maes

, and Dewulf

2013. Relationship between biosecurity and production/antimicrobial treatment characteristics in pig herds. Vet. J., 198:508–512.

33.

Hybschmann

G.K.

, Ersboll

A.K.

, Vigre

, Baadsgaard

N.P.

, and Houe

2011. Herd-level risk factors for antimicrobial demanding gastrointestinal diseases in Danish herds with finisher pigs: a register-based study. Prev. Vet. Med., 98:190–197.

34.

Jensen

V.F.

, Emborg

H.D.

, and Aarestrup

F.M.

2012. Indications and patterns of therapeutic use of antimicrobial agents in the Danish pig production from 2002 to 2008. J. Vet. Pharmacol. Ther., 35:33–46.

35.

Snow

L.C.

, Warner

R.G.

, Cheney

, Wearing

, Stokes

, Harris

, Teale

C.J.

, and Coldham

N.G.

2012. Risk factors associated with extended spectrum beta-lactamase Escherichia coli (CTX-M) on dairy farms in North West England and North Wales. Prev. Vet. Med., 106:225–234.

36.

Reist

, Geser

, Hachler

, Scharrer

, and Stephan

2013. ESBL-producing Enterobacteriaceae: occurrence, risk factors for fecal carriage and strain traits in the Swiss slaughter cattle population younger than 2 years sampled at abattoir level. PLoS One, 8:e71725.

37.

Sternberg Lewerin

, Osterberg

, Alenius

, Elvander

, Fellstrom

, Traven

, Wallgren

, Waller

K.P.

, and Jacobson

2015. Risk assessment as a tool for improving external biosecurity at farm level. BMC Vet. Res., 11:171.

38.

Backhans

, Sjolund

, Lindberg

, and Emanuelson

2015. Biosecurity level and health management practices in 60 Swedish farrow-to-finish herds. Acta Vet. Scand., 57:14.

39.

Callens

, Faes

, Maes

, Catry

, Boyen

, Francoys

, de Jong

, Haesebrouck

, and Dewulf

2015. Presence of antimicrobial resistance and antimicrobial use in sows are risk factors for antimicrobial resistance in their offspring. Microb. Drug Resist., 21:50–58.

40.

Scherz

, Stahl

, Glunder

, and Kietzmann M

2014. Effects of carry-over of fluoroquinolones on the susceptibility of commensal Escherichia coli in the intestinal microbiota of poultry. Berl. Munch. Tierarztl. Wochenschr., 127:478–485.

41.

Friese

, Schulz

, Laube

, von Salviati

, Hartung

, and Roesler

2013. Faecal occurrence and emissions of livestock-associated methicillin-resistant Staphylococcus aureus (laMRSA) and ESbl/AmpC-producing E. coli from animal farms in Germany. Berl. Munch. Tierarztl. Wochenschr., 126:175–180.

42.

Laube

, Friese

, von Salviati

, Guerra

, and Rosler

2014. Transmission of ESBL/AmpC-producing Escherichia coli from broiler chicken farms to surrounding areas. Vet. Microbiol., 172:519–527.

43.

von Salviati

, Laube

, Guerra

, Roesler

, and Friese

2015. Emission of ESBL/AmpC-producing Escherichia coli from pig fattening farms to surrounding areas. Vet. Microbiol., 175:77–84.

44.

Duan

H.Y.

, Chai

T.J.

, Cai

Y.M.

, Zhong

Z.B.

, Yao

M.L.

, and Zhang

X.X.

2008. Transmission identification of Escherichia coli aerosol in chicken houses to their environments using ERIC-PCR. Sci. China Ser. C Life. Sci., 51:164–173.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB

0.03 MB

0.10 MB