Skip to main content
NCBI home page
Canadian Journal of Veterinary Research logoLink to Canadian Journal of Veterinary Research
. 2016 Apr;80(2):95–105.

Farm-level prevalence and risk factors for detection of hepatitis E virus, porcine enteric calicivirus, and rotavirus in Canadian finisher pigs

Barbara Wilhelm 1,✉, Danielle Leblanc 1, David Leger 1, Sheryl Gow 1, Anne Deckert 1, David L Pearl 1, Robert Friendship 1, Andrijana Rajić 1, Alain Houde 1, Scott McEwen 1
PMCID: PMC4836045  PMID: 27127336

Abstract

Hepatitis E virus (HEV), norovirus (NoV), and rotavirus (RV) are all hypothesized to infect humans zoonotically via exposure through swine and pork. Our study objectives were to estimate Canadian farm-level prevalence of HEV, NoV [specifically porcine enteric calicivirus (PEC)], and RV in finisher pigs, and to study risk factors for farm level viral detection. Farms were recruited using the Canadian Integrated Program for Antimicrobial Resistance Surveillance (CIPARS) and FoodNet Canada on-farm sampling platforms. Six pooled groups of fecal samples were collected from participating farms, and a questionnaire capturing farm management and biosecurity practices was completed. Samples were assayed using validated real-time polymerase chain reaction (RT-PCR). We modeled predictors for farm level viral RNA detection using logistic and exact logistic regression. Seventy-two herds were sampled: 51 CIPARS herds (15 sampled twice) and 21 FoodNet Canada herds (one sampled twice). Hepatitis E virus was detected in 30/88 farms [34.1% (95% CI 25.0%, 44.5%)]; PEC in 18 [20.5% (95% CI: 13.4%, 30.0%)], and RV in 6 farms [6.8% (95% CI: 3.2%, 14.1%)]. Farm-level prevalence of viruses varied with province and sampling platform. Requiring shower-in and providing boots for visitors were significant predictors (P < 0.05) in single fixed effect mixed logistic regression analysis for detection of HEV and PEC, respectively. In contrast, all RV positive farms provided boots and coveralls, and 5 of 6 farms required shower-in. We hypothesized that these biosecurity measures delayed the mean age of RV infection, resulting in an association with RV detection in finishers. Obtaining feeder pigs from multiple sources was consistently associated with greater odds of detecting each virus.

Introduction

The RNA viruses hepatitis E virus (HEV), norovirus (NoV), and rotavirus (RV), have all been hypothesized over the past decade to infect humans zoonotically (1,2,3). A recent scoping review identified a small number of zoonotic cases [HEV n = 3; NoV n = 0; RV n = 40 (zoonoses n = 3; human-animal re-assortants n = 37)] categorized as “likely” zoonoses (4). Several researchers have assayed Canadian pigs for detection of HEV: a survey of 70 Quebec farms reported 34% categorized as HEV-infected (5); the prevalence of HEV RNA shedding in 51 close-to-market pigs on one infected farm in Québec was estimated at 42.2% (6). Hepatitis E virus isolates obtained from 2 locally acquired Canadian Hepatitis E cases in the province of Quebec demonstrated close (> 97% nt) sequence identity between case isolates and those collected from pigs in the same province over part of the ORF2 gene (7). Canadian pig farms have been sampled for NoV, which was detected in 30 of 120 samples from 10 Ontario farms (8), and in Quebec, in which NoV RNA was detected in samples from 4 of 20 farms (9). Rotavirus RNA was detected on 9 of 10 Ontario pig farms (10) and more recently in 9 of 10 Quebec finisher farms (11). A recent Canadian study reported detection of HEV on retail pork livers (12). However, a Canadian national survey of finisher pigs to estimate the prevalence of these viruses has not been conducted to date. Our study objectives, therefore, were to estimate farm-level prevalence of HEV, porcine enteric calicivirus (PEC), and RV in Canadian finisher pigs, and to investigate potential associations between farm level detection and farm management practices.

Materials and methods

Recruitment and sample collection

Expected farm-level and individual-level prevalence of HEV in finisher pigs were estimated by a meta-analysis (MA) of 12 published North American investigations of HEV detection in fecal samples in pigs aged 4 to 6 mo using the Comprehensive Meta-analysis 2 software (Biostat; Englewood, New Jersey, USA) and the method of moments (13). Hepatitis E virus was used in preference over PEC and RV for sample size calculations, as this was the virus for which we were able to obtain the most comprehensive dataset. Heterogeneity, or across study variation in viral prevalence, was investigated by calculation of Higgins’ I2 (14) and T2, an estimate of τ2 or true variance in prevalence across studies (15). Sample size calculations to estimate the number of pooled fecal samples to detect HEV RNA on-farm, using the MA summary estimates of within-farm and farm-level prevalence (desired precision +/− 10%), were conducted in EpiTools (16).

Two sampling platforms were used to recruit pig farms: the Canadian Integrated Program for Antimicrobial Resistance Surveillance (CIPARS) sampling platform for finisher pigs (17), and the FoodNet Canada (FC) sampling platform (8,10,18). The CIPARS maintains a national network of participating veterinarians, each of whom invite client farms to participate in the CIPARS program by completing annual questionnaires regarding management practices and herd performance, and allowing collection of pooled fecal samples from finisher pigs (17). For this project, each CIPARS veterinarian was sent an electronic invitation to participate. Those agreeing received a package containing invitation letters for the clients they recruited, and a consent form for the participating farmers. During the farm visit, the CIPARS farm program questionnaire capturing herd-level data on demographics, current management practices, biosecurity status, pig inventory, and herd performance, was completed by the producer with the help of the herd veterinarian. Data volunteered by the farmer were then validated by the veterinarian and entered on the questionnaire. Missing or incomplete data, or those entries falling outside pre-set logical limits, were further investigated by a telephone call or e-mail to the veterinarian when completed questionnaires were received by CIPARS.

Similarly, during a routine visit for FC sample collection, each pig farm participating in the FC sentinel pig farm program was invited to participate. The FC farms were a subset of 100 farms located in a single province, originally selected as a random sample of 5000 pig farms, stratified by size (18). During the farm visit, a questionnaire identical to the CIPARS farm program questionnaire with the exception of the CIPARS antimicrobial use questions, which were not administered to the FC farmers was completed by the producer with the help of the FC sample collection technician.

All fecal samples and questionnaires used farm herd codes as identifiers; the identities of the participating farms were known only to the CIPARS herd veterinarians or the FC program technician. This project received approval from the Research Ethics Board of the University of Guelph.

All sample collection, including those samples collected from FC farms, followed the CIPARS sampling protocol. Briefly, 5 fresh fecal samples were collected into one specimen jar and mixed with a stir stick. One jar was collected from each of 6 close-to-market pens, and shipped by courier to the Public Health Agency of Canada (PHAC) diagnostic laboratory in St. Hyacinthe, Quebec. Samples were then divided, and part of each pooled fecal sample originating from a farm participating in this project was forwarded to the Agriculture and Agri-Food Canada laboratory (AAFC) in St. Hyacinthe, where they were held frozen at −80°C prior to batch processing and testing. Samples from FoodNet Canada farms were stored in a freezer at −80°C at the University of Guelph until completion of sampling, when they were forwarded by courier to AAFC St. Hyacinthe.

Viral RNA extraction from feces

Fecal samples were thawed and a 1 g portion of sample was further diluted 1:5 (p/v) in PBS pH 7.4 (Life Technologies, Burlington, Ontario). After vigorous shaking, mixtures were centrifuged at 4000 × g for 20 min and supernatants were considered as viral suspensions. A volume of 13 μL of SDS 10%, 0.69 μL of proteinase K (20 mg/mL) and 4.8 × 103 PFU of feline calicivirus used as sample process control were added to 130 μL of viral suspension. The mixture was incubated at 37°C for 1 h. The RNA was extracted from 140 μL of the treated viral suspension with QIAampViral RNA mini (Qiagen, Toronto, Ontario) and the robotic workstation system QIAcube for automated sample purification of RNA; the QIAcube manufacturer’s protocol was used. The extracted RNA was recovered in 60 μL volume of AVE solution and frozen at −80°C until further use.

Viral RNA detection

Detection of HEV, PEC, and RV was done using the quantitative real-time polymerase chain reaction (qRT-PCR) detection systems previously described (12) adapted in Brilliant II RT-PCR Core reagent kit, 1-Step (Agilent Technologies Canada, Mississauga, Ontario). Briefly, a sample of 2.5 μL of RNA template was used in a final volume of 25 μL of core RT-PCR buffer containing 2.5 μL of MgCl2, 1.0 μL of dNTP, 1.25 μL of bovine serum albumin at 20 mg/mL, each of the forward and reverse primers and the TaqMan probe, 1 μL of the reference dye diluted 1:500, 1 μL of reverse transcriptase diluted 1:10, and 1.25U of SureStart TaqDNA polymerase (Qiagen). Porcine enteric calicivirus, NoV G II, and RV were detected in a monoplex assay and HEV/FCV in duplex assay.

Viral quantification

Extracted nucleic acids showing amplification inhibition in the first detection step were treated with OneStep PCR Inhibitor Removal kit (Zymo Research, Irvine, California, USA) before retesting. Addition of 2 μL of Hepatitis A virus RNA (3.98 × 104 Tissue culture infectious dose) to each assay was used as amplification control. For each detection system, a calibration curve was generated using 10-fold serial dilutions of the plasmid constructs to the corresponding virus in a 5 ng/mL salmon sperm solution. Arithmetic mean load was determined using the MX-pro V4.0 software (Stratagene, La Jolla, California, USA).

Data storage, cleaning, and analysis

Data pertaining to detection of viral RNA in samples, and questionnaire responses from each participating farm were captured in Excel spreadsheets (Microsoft, Mississauga, Ontario) and forwarded to a central repository. Data were cleaned in Excel and exported to Stata IC 13 (StataCorp, College Station, Texas, USA) for descriptive statistics and modeling. Missing data received their own unique code. However, we found that a small group (n = 3) of FC respondents had not answered the entire group of biosecurity questions (Table I). Therefore, we deleted this subset of farms from the “biosecurity” variable analyses. We checked the correlation among predictor variables to avoid issues associated with collinearity. Where collinear variables were eligible for inclusion in the full model, the variable with the most complete set of observations was selected.

Table I.

Description of variables and participating farms

Variable Categories CIPARS Number of samples (%) FC Number of samples (%)
General and Farm Descriptors
 Herd veterinarian Unique identifier for each veterinarian n = 17 veterinarians N = 1 technician
 Herd identifier n = 51 herds N = 21 herds
 Number of days sampled 1 = 1 day 1 = 36 herds 1 = 20 herds
2 = 2 days 2 = 15 herds 2 = 1 herd
 Province Province A 108 samples, 18 herds
Province B 72 samples, 12 herds
Province C 114 samples, 19 herds 132 samples, 22 herds
Province D 102 samples, 17 herds
 Production system used Continuous 192 (48.5%) 96 (72.8%)
All-in-all-out 204 (51.5%) 18 (13.6%)
Missing observations 18 (13.6%)
 Month collected January 18 0
February 6 0
March 48 0
April 12 0
May 12 0
June 12 0
July 18 0
August 48 0
September 42 36
October 36 30
November 96 54
December 48 12
 Season Winter 72 (18.2%) 0 (0%)
Spring 36 (9.1%) 0 (0%)
Summer 108 (27.3%) 36 (27.3%)
Fall 180 (45.5%) 96 (72.7%)
 Finishers onsite 0–2000 240 (60.6%) 120 (90.9%)
2001–4000 150 (37.9%) 12 (9.1%)
≥ 4000 6 (1.5%) 0 (0%)
Mean (SD) Mean (SD)
 Finishers onsite, numerical value 1706 (SD 1172) 810 (SD 1175)
Biosecurity predictors
 Multiple source No 342 (86.4%) 114 (86.4%)
Yes 54 (13.6%) 6 (54.5%)
Not captured 0 (0%) 12 (9.1%)
 Boots provided No 66 (16.7%) 6 (4.5%)
Yes 330 (72.3%) 114 (86.4%)
Not captured 0 (0%) 12 (9.1%)
 Coveralls provided No 72 (18.2%) 48 (36.4%)
Yes 324 (81.8%) 66 (16.7%)
Not captured 0 (0%) 18 (13.6%)
 Boot dip provided No 288 (72.7%) 96 (72.7%)
Yes 108 (27.3%) 12 (9.1%)
Not captured 0 (0%) 24 (18.2%)
 Biosecurity sign No 78 (19.7%) 0 (0%)
Yes 318 (80.3%) 114 (86.4%)
Not captured 0 (0%) 18 (13.6%)
 Shower-in No 216 (54.5%) 90 (68.2%)
Yes 180 (45.5%) 24 (18.2%)
Not captured 0 (0%) 18 (13.6%)
 Downtime required No 120 (30.3%) 96 (72.7%)
Yes 276 (69.7%) 12 (9.1%)
Not captured 0 (0%) 24 (18.2%)
Open in a new tab

CIPARS — Canadian Integrated Program for Antimicrobial Resistance Surveillance; FC — FoodNet Canada.

In assessing the effect of a predictor variable in univariable analysis, we included random intercepts for herd and sampling event using the “xtmelogit” command (19) in Stata 13 to account for the hierarchical data structure. We estimated the proportion of variance occurring at each level of the model, assuming that the lowest level variance on the logit scale was π2/3, based on use of the latent variable technique (20). All variables for which P < 0.20 in univariable analysis were considered for inclusion in the multivariable model. However, where full multivariable models failed to converge, precluding multivariable analysis, only single fixed effect mixed models are presented.

For sparse datasets exact logistic regression was employed, with farm-level detection of viral RNA as the outcome of interest, and a farm was categorized “positive” if viral RNA were detected in one or more of 6 pooled samples collected on-farm.

Results

Meta-analysis of 12 published North American surveys of finisher swine yielded a summary estimate of within-farm HEV prevalence in finisher pigs of 23.0% [95% confidence interval (95% CI): 13.2%, 35.8%] with Higgins’ I2 value of 89.31% and T2 of 1.18, suggesting marked heterogeneity across studies. Calculation of the number of pooled samples (consisting of 5 samples from individual animals) required to demonstrate freedom from viral presence, using the meta-analysis estimate of 23.0% prevalence on-farm, confirmed that 3 pooled samples would be sufficient to identify freedom from disease. The standard operating procedure of the CIPARS on-farm swine surveillance system (i.e., collection of 6 pooled samples per farm), was adopted for the purposes of this study, on all participating farms including the FC farms, given the heterogeneity in the dataset supporting the MA estimate of prevalence. Meta-analysis of 5 published North American surveys of finisher pigs yielded a summary estimate of farm-level HEV prevalence in finisher pigs of 30.0% (95% CI: 14.7%, 51.6%) with Higgins’ I2 value of 71.9% and T2 of 0.92, suggesting marked heterogeneity across studies. Calculation of the number of farms required to estimate a prevalence of 30% with desired precision yielded a target of 75 farms participating.

Farm recruitment and sample collection

All 26 CIPARS veterinarians were approached to participate in this study in the summer of 2011. Nine veterinarians enrolled and they recruited 17 finisher herds in total. In the fall of 2011 the FC farms were invited by the FC program technician to participate, and 20 of 28 enrolled with one farm being tested twice. At the end of 2011, CIPARS veterinarians were again invited to participate. In 2012, 15 veterinarians participated, enrolling 34 new farms, and re-testing 15 farms.

A total of 528 samples from 72 herds from 4 provinces (A, B, C, and D) with 6 pooled samples per herd were collected over a 17-month period and stored at −80°C until further use. A summary of the characteristics of participating farms in both sampling platforms is presented in Table I. Overall, relative to CIPARS farms, FC farms fed fewer finisher pigs; all were located in province C, and most practiced continuous pig flow (72.7%) as opposed to all-in-all-out flow. Compared with CIPARS farms, more FC farms provided boots (86.4% versus 72.3%), but fewer provided coveralls prior to entry (16.7% versus 81.8%); fewer required shower-in (18.2% versus 45.5%), and fewer required downtime after visiting another pig farm, prior to entry (9.1% versus 69.7%). Sampling was only conducted September to December, for FC farms. “Retest” was included as a categorical variable in the dataset as 16 farms were sampled twice.

Data were missing from a sub-group of FC farms for the following biosecurity variables: provision of boots or coveralls, use of a boot dip, posting a biosecurity sign, requirement for shower-in, or downtime, and obtaining pigs from multiple sources.

Hepatitis E virus

Hepatitis E virus RNA was detected from one or more samples collected from 19 of 66 CIPARS farms (estimated herd-level prevalence 28.8%; 95% CI: 19.3%, 40.6%), and 11 of 22 FC farms (estimated herd-level prevalence 50.0%, 95% CI: 30.7%, 69.3%). Hepatitis E virus RNA was detected from 35/396 individual pooled samples collected from CIPARS farms (estimated sample-level prevalence 8.8%; 95% CI: 6.4%, 12.2%), and 38/132 pooled samples from FC farms (estimated sample-level prevalence 28.8%; 95% CI: 21.8%, 37.0%). Hepatitis E virus prevalence by province is presented in Table II. Mean viral load in positive samples was 4 360 583 genomic copies (gc)/g (SD = 15 000 000 gc/g).

Table II.

Prevalence and mean load of hepatitis E virus, porcine enteric calicivirus, and rotavirus in Canadian finisher pigs

Hepatitis E virus Porcine enteric calicivirus Rotavirus
Farms Number positive/number sampled (%) Samples Number positive/number sampled (%) Farms Number positive/number sampled (%) Samples Number positive/number sampled (%) Farms Number positive/number sampled (%) Samples Number positive/number sampled (%)
National 30/88 (34.1%) 73/528 (13.8%) 18/88 (20.5%) 38/528 (7.2%) 6/88 (6.8%) 8/528 (1.5%)
 Province A 3/18 (16.7%) 3/108 (2.8%) 2/18 (11.1%) 3/108 (2.8%) 2/18 (11.1%) 2/108 (1.9%)
 Province B 6/12 (50.0%) 14/72 (19.4%) 3/12 (25.0%) 7/72 (9.7%) 1/12 (8.3%) 1/72 (1.4%)
 Province C 14/41 (34.1%) 42/246 (17.1%) 6/41 (14.6%) 13/246 (5.3%) 3/41 (14.3%) 5/246 (2.0%)
 Province D 7/17 (41.2%) 14/102 (13.7%) 7/17 (41.2%) 15/102 (14.7%) 0/17 0/102
Month
 January 1/3 (33.3%) 1/18 (5.6%) 1/3 (33.3%) 1/18 (5.6%) 0/3 (0%) 0/18 (0%)
 February 0/1 (0%) 0/6 0/1 (0%) 0/6 (0%) 0/1 (0%) 0/6 (0%)
 March 2/8 (25.0%) 7/48 (14.6%) 1/8 (12.5%) 1/48 (2.1%) 0/8 (0%) 0/48 (0%)
 April 1/2 (50.0%) 2/12 (16.7%) 0/2 (0%) 0/12 (0%) 0/2 (0%) 0/12 (0%)
 May 0/2 (0%) 0/12 0/2 (0%) 0/12 (0%) 0/25 (0%) 0/12 (0%)
 June 1/2 (50.0%) 2/12 (16.7%) 1/2 (50.0%) 3/12 (25.0%) 0/2 (0%) 0/12 (0%)
 July 0/3 (0%) 0/18 (0%) 1/3 (33.3%) 6/18 (33.3%) 1/3 (33%) 1/18 (6%)
 August 1/8 (12.5%) 1/48 (2.1%) 1/8 (12.5%) 1/48 (2.1%) 1/8 (12.5%) 1/48 (2.1%)
 September 3/13 (23.1%) 3/78 (3.8%) 1/13 (7.7%) 2/78 (2.6%) 1/13 (7.7%) 2/78 (2.6%)
 October 8/11 (72.7%) 20/66 (30.3%) 2/11 (18.2%) 3/66 (4.5%) 1/11 (9.1%) 1/66 (1.5%)
 November 8/25 (32.0%) 26/150 (17.3%) 5/25 (20.0%) 10/150 (6.7%) 2/25 (8.0%) 3/150 (2.0%)
 December 5/10 (50.0%) 11/60 (18.3%) 5/10 (50.0%) 11/60 (18.3%) 0/10 (0%) 0/60 (0%)
GM/g (all samples) 619 401 gc/ga 31 644 gc/g 275 277 gc/g
SD 5 819 952 gc/g 442 402 gc/g 6 051 227 gc/g
GM/g (‘positive’ samples) 4 360 583 gc/g 388 556 gc/g 18 200 000 gc/g
SD 15 000 000 gc/g 1 521 130 gc/g 48 800 000 gc/g
Open in a new tab
a

gc/g = genomic copies per gram of sample.

GM — geometric mean; HEV — hepatitis E virus; PEC — porcine enteric calicivirus; RV — rotavirus; SD — standard deviation.

We performed mixed logistic regression modeling HEV RNA detection in a sample as the outcome measure, with a single fixed effect and random intercepts for farm and sampling event (Tables III, IV). The following predictors were eligible for inclusion in the full multivariable model (i.e., P < 0.20): sampling platform — with FC farms having significantly (P < 0.05) greater odds of HEV RNA detection on-farm relative to CIPARS farms; province — with province A farms having reduced odds of HEV detection relative to the other provinces; year of sampling, with farms sampled in year 2 having reduced odds of HEV detection relative to year 1 sampling; season — with samples collected in spring or fall having greater odds of HEV detection relative to winter; and the number of finisher pigs on-farm — with larger farms having reduced odds of HEV detection. Requiring shower-in was associated with significantly (P < 0.05) reduced odds of HEV detection.

Table III.

Single fixed effect logistic regression modeling predictors of viral detection in Canadian finisher pigs: general and farm descriptors

Hepatitis E virus HEV Porcine enteric calicivirus PEC Rotavirus RV
Variable ORa (95% CI) OR P-value Model P-value OR (95% CI) OR P-value Model P-value ORb (95% CI) OR P-value Model P-value
Sampling platform 0.01 0.29 0.0007
 CIPARS Referent Referent Referent
 FC 14.58 (1.82, 116.93) 0.01 0.12 (0.003, 5.89) 0.29 0.05c (0, 0.24) 0.0001
Province 0.20 0.56 0.01
 A Referent Referent Referent
 B 16.65 (0.62, 447.63) 0.09 3.50 (0.06, 188.94) 0.54 0.73 (0.31, 1.34) 0.74
 C 8.92 (0.73, 109.04) 0.09 1.58 (0.06, 44.15) 0.79 0.63 (0.28, 1.50) 0.33
 D 10.82 (0.50, 232.98) 0.26 10.82 (0.24, 485.40) 0.22 0.06c (0.35) 0.0005
Year of sampling 0.08 0.58 < 0.0001
 Year 1 Referent Referent Referent
 Year 2 0.19 (0.03, 1.19) 0.08 1.95 (0.18, 21.05) 0.58 48.35c (8.26, Inf) < 0.0001
Re-test 0.67 0.15 0.002
 No Referent Referent Referent
 Yes 2.06 (0.07, 58.72) 0.67 8.45 (0.45, 158.54) 0.15 0.08 (0, 0.46) 0.001
Season 0.07 0.93 0.003
 Winter Referent Referent Referent
 Spring 1.58 (0.03, 74.37) 0.82 1.42 (0.005, 402.51) 0.90 1.0 (0, Inf)
 Summer 0.31 (0.02, 5.47) 0.42 1.05 (0.02, 70.60) 0.98 14.34 (2.41, Inf) 0.0009
 Fall 5.45 (0.42, 72.20) 0.20 2.37 (0.05, 102.61) 0.65 7.04 (1.19, Inf) 0.03
Production type 0.34 No convergenced 0.01
Referent Referent Referent
Continuous
 All-in-all-out 2.29 (.31, 16.80) 0.42 2.87 (1.30, 6.26) 0.007
 Not captured 184.26 (0.09, 3.87e + 5) 0.18 1.41 (0, 9.48) < 0.0001
Finishers on-farm 0.99 (0.99, 1.00) 0.07 0.07 1.00 (0.99, 1.00) 0.80 0.78 No convergence
Farm size No convergence No convergence 0.0003
0 to 2000 finishers Referent Referent Referent
2 to 4000 finishers 0.58 (0.19, 1.51) 0.34
> 4000 finishers 6.93 (1.96, 22.18) 0.003
Open in a new tab
a

Mixed logistic regression with random intercepts for “farm” and “sampling event” with “sample positive” as the outcome measure.

b

Exact logistic regression with “farm positive” as the outcome measure.

c

Mean unbiased estimate.

d

Maximum memory allocated in Stata = 2 Megabytes.

CIPARS — Canadian Integrated Program for Antimicrobial Resistance Surveillance; FC — FoodNet Canada; HEV — hepatitis E virus; Inf — infinity; OR — odds ratio; PEC — porcine enteric calicivirus; RV — rotavirus; 95% CI — 95% confidence interval.

Table IV.

Single fixed effect logistic regression modeling predictors of viral detection in Canadian finisher pigs: biosecurity practicesa

Hepatitis E virus Porcine enteric calicivirus Rotavirus
Variable ORb (95% CI) OR P-value Model P-value ORb (95% CI) OR P-value Model P-value ORc (95% CI) OR P-value Model P-value
Finishers from multiple sources 0.57 0.77 0.0003
 No Referent 0.57 Referent 0.77 Referent 0.0006
 Yes 2.14 (0.16, 29.34) 1.48 (0.10, 21.63) 4.42 (1.89, 9.89)
Boots provided 0.63 0.01
 No Referent 0.63 Referent 0.04 Referent 0.01
 Yes 1.82 (0.16, 21.02) 0.08 (0.007, 0.83) 0.04 9.14d (1.64, Inf)
Coveralls provided 0.62 0.60 0.003
 No Referent Referent Referent 0.0001
 Yes 0.38 (0.03, 4.51) 0.44 0.26 (0.02, 3.51) 0.31 17.32d (3.07, Inf)
Shower-in 0.05 0.67 < 0.0001
 No Referent Referent Referent
 Yes 0.16 (0.03, 0.96) 0.05 0.68 (0.10, 4.26) 0.70 8.59d (3.43, 25.74) < 0.0001
Biosecurity sign 0.74 0.20 0.006
 No Referent 0.74 Referent 0.20 Referent 0.004
 Yes 1.51 (0.13, 17.44) 0.20 (0.02, 2.22) 10.06d (1.77, Inf)
Downtime required 0.57 0.44 0.16
 No Referent 0.57 Referent 0.44 Referent 0.22
 Yes 1.83 (0.23, 14.22) 0.43 (0.05, 3.56) 0.59d (0.27, 1.34)
Open in a new tab
a

The three FoodNet Canada farms for which data for all biosecurity predictors were missing, were dropped from the dataset for these analyses.

b

Mixed logistic regression with random intercepts for “farm” and “sampling event” with “sample positive” as the outcome measure.

c

Exact logistic regression with “farm positive” as the outcome measure.

d

Mean biased estimate.

HEV — hepatitis E virus; Inf — infinity; PEC — porcine enteric calicivirus; RV — rotavirus; 95% CI — 95% confidence interval.

In an intercept-only model, with random intercepts for herd and sampling event, 45.1% of the variance occurred at the individual sample level (i.e., the lowest hierarchical level of the data), with 32.6% of the variance occurring at the herd level, and 22.3% at the sampling event. Similar proportions were observed for each single fixed effect model.

The full multivariable model failed to converge, as did most models incorporating a subset of the eligible variables, precluding investigation of interaction or confounding.

Porcine enteric calicivirus

Porcine enteric calicivirus RNA was detected from one or more samples collected from 17 of 66 CIPARS farms (estimated farm-level prevalence 25.8%; 95% CI: 16.8%, 37.4%), and 1 of 22 FC farms (farm-level prevalence 4.6%; 95% CI: 0.80%, 21.8%). Porcine enteric calicivirus RNA was detected from 36 of 396 individual pooled samples collected from CIPARS farms (estimated sample-level prevalence 9.1%; 95% CI: 6.6%, 12.3%), and 2 of 132 pooled samples from FC farms (estimated sample level prevalence 1.5%; 95% CI: 0.4%, 5.9%). Porcine enteric calicivirus prevalence by province is presented in Table II. Mean viral load in positive samples was 388 556 gc/g (SD = 1 521 130 gc/g). In mixed logistic regression with single fixed effect and random intercepts for farm and sampling event, providing boots on entry to the unit was associated with significantly (P < 0.05) reduced odds of PEC detection. Samples collected during re-test of a farm were associated with increased (P = 0.15) odds of PEC detection. However, a multivariable model including both of these predictors failed to converge.

In an intercept-only model, with random intercepts for herd and sampling event, 53.9% of the variance occurred at the individual sample level (i.e., the lowest hierarchical level of the data), with negligible variance occurring at the herd level, and 46.1% at the sampling event. Similar proportions were observed for each single fixed effect model.

Rotavirus

Rotavirus RNA was detected from one or more samples collected from 6/66 CIPARS farms (estimated farm-level prevalence 9.1%, 95% CI: 4.2%, 18.5%), and 0 of 22 FC farms. Rotavirus RNA was detected from 8 of 396 individual pooled samples collected from CIPARS farms (estimated sample level prevalence 2.0%, 95% CI: 1.0%, 3.9%), and 0 of 132 pooled samples from FC farms. Rotavirus prevalence by province is presented in Table II. Mean viral load in positive samples was 18 200 000 gc/g (SD = 48 800 000 gc/g). In univariable exact logistic regression with “farm positive” as the outcome measure (Table III), both sampling platform and province were significant (P < 0.05) predictors for RV detection, with farms from province A having greatest odds of RV detection, and FC farms reduced odds of RV detection as compared with CIPARS farms (OR = 0.05, 95% CI: 0, 0.24). Production type was a significant predictor for RV detection, with farms practicing all-in-all-out pig flow having significantly increased odds relative to farms using continuous flow (OR = 2.87, 95% CI: 1.30, 6.26). Farms obtaining grow-finish pigs from multiple sources had greater odds of RV detection (OR = 4.42, 95% CI: 1.89, 9.89) as compared with those only using one source. Provision of boots and coveralls for visitors, requiring a shower prior to entering the unit, and posting a biosecurity sign at the entry to the site, all were significantly (P < 0.05) associated with increased odds of RV detection in finisher pigs.

Overall, FC farms had greater odds of HEV RNA relative to CIPARS farms, and reduced odds of RV RNA detection. Province A had lower odds of HEV relative to the other provinces, and province D had lower odds of RV RNA detection relative to province A.

Discussion

The overall farm-level prevalence of HEV in finishers in our survey (34.1%, 95% CI: 25.0%, 44.5%) is comparable with published North American estimates ranging from 25% to 68% (6,21). The FC farms tend to be smaller multi-species farms as compared with CIPARS farms (Deckert & Friendship, 2015, University of Guelph, personal communication). Other species, including cats, dogs, cattle, sheep, goats, and rabbits have been reported to be capable of HEV sero-conversion and/or shedding (3), and it is possible that HEV shedding of other species on-farm may have contributed to the introduction of HEV to the pig herd. The significant association between requiring shower-in and reduced odds of HEV RNA detection on-farm (OR = 0.16, 95% CI: 0.03, 0.96) is consistent with the possibility that asymptomatic human shedders could be a potential exposure source of HEV for pigs, as has been hypothesized for pork (7). Conditional estimates of effect would be required to further investigate this finding, which was not possible in this dataset.

Overall farm-level prevalence of PEC was 20.5% (95% CI: 13.4%, 30.0%). In comparison, a Dutch study reported PEC RNA detected in 3- to 9-month-old pigs on only 2 of 100 farms sampled (22) and a US investigation detected PEC RNA in 9 of 9 farms sampled (23). The significant association estimated between provision of boots upon entry to the unit, and reduced odds of PEC RNA detection on-farm (OR = 0.08, 95% CI: 0.007, 0.83), is consistent with an agent transmitted via the fecal-oral route.

Rotavirus had the lowest farm-level prevalence of the 3 viruses of interest in this survey. A recent Vietnamese survey reported a farm-level RV prevalence of 74% (77/104) (24). However, with a smaller sample size, an Italian survey reported a numerically lower prevalence (3 of 8 herds) surveyed (25). In comparison with HEV and PEC, RV had a different geographic distribution, with province “A” having greater odds of RV detection on-farm, and no RV detected in province “D” farms.

While few of the predictor variables from this study, with the exception of geographic region, have been investigated in other published research, we generally expected that the “biosecurity” variables (providing boots/boot dip/coveralls, requiring shower-in/downtime) would be associated with reduced odds of detection, and sampling platform and province could be significant predictors, for viral RNA detection on-farm. For example, investigators in France have reported a significant association between lack of hygiene measures such as failure to provide boots, and detection of HEV in the livers of slaughter pigs (26). In our study, obtaining finishers from multiple sources was the one predictor variable consistently associated with increased odds of viral RNA detection. This is similar to other research, which reported that obtaining pigs from more than one source was a significant (P < 0.05) predictor for the presence of Salmonella spp. on-farm (27,28) in cross-sectional studies of 109 and 359 farms, respectively.

However, the measures of association for some predictor variables differed in magnitude and, in some cases, direction, across the 3 viruses studied, although the primary transmission route for all 3 is the fecal-oral route. The varying relationships between predictor variables and outcome across the 3 viruses may reflect differences in viral biology and relative importance of various possible methods of introduction/transmission. On an infected farm, incident HEV infection occurs most frequently in the nursery (21), RV infection occurs most frequently in suckling pigs, and mean age of PEC infection has varied across studies (29,30). In our study, RV detection in finishers was significantly associated with adoption of widely accepted biosecurity measures such as provision of boots and coveralls (Table III). Possibly, given the tendency for RV infection to occur early in the pig’s life, the application of these biosecurity measures may functionally delay the mean age of RV infection. This could result in biosecurity measures such as provision of boots being significantly associated with RV detection in finishers, particularly in farrow-to-finish herds.

The fact that sampling platform was a significant predictor for HEV and RV RNA was noteworthy. However, it highlights the potential importance of stratified sampling in surveillance programs; some farms (e.g., small scale/mixed species), although perhaps not producing a large proportion of market animals, may be important biologically in pathogen transmission and control.

Our study has several limitations. The farmers who participated in this survey are volunteers, suggesting the potential for selection bias, by functionally using an incomplete definition to define the eligible population (31). There is no national sampling frame of finisher farms with which our study group might be compared with regards to important demographic or management parameters. However, CIPARS also operates an abattoir-level sampling platform that includes cecal sampling of pigs. The close similarity between CIPARS farm level antimicrobial resistance estimates, and those obtained from the national abattoir component of CIPARS (which randomly samples slaughter pigs and is therefore unaffected by producer or veterinarian willingness to participate), indicates that the convenience sample of farms obtained by the CIPARS on-farm sampling platform is likely representative of Canadian pig production (17). Unfortunately, however, no similar mechanism exists for estimating the similarity of the FC farms to the national herd.

Additionally, participants self-reported farm management and biosecurity practices. A minority (n = 20) of respondents did not complete some survey questions, including those pertaining to farm biosecurity practices, with 3 FC respondents not answering any of these questions. We hypothesize that the missing biosecurity data could be missing not at random, perhaps reflecting respondents’ reluctance to report that fairly widely endorsed biosecurity measures (32) were not practiced. If the 3 farms for which data were missing were, in fact, not using these measures, their inclusion in the analysis would have increased the observed protective effect of these biosecurity measures.

While the diagnostic sensitivity and specificity of the RT-PCR assays used in this study have been investigated in fecal samples collected from other populations (e.g., humans, younger pigs), the diagnostic sensitivity and specificity of these assays in fecal samples from finisher pigs are not known. Additionally, the RV assay targets the NSP3 gene of group A RVs, and its performance in detecting RVs of other serotypes such as group C RVs, which are also reported in finisher pigs (33) is also unknown. A Brazilian survey of diarrheic pigs 1 to 4 weeks old identified 144 fecal samples inconclusively categorized using polyacrylamide gel assay. Of these, 19 and 5 were identified as group B and C RVs, respectively, using group-specific RT-PCR, highlighting both the potential for their detection on-farm, and the requirement for appropriate assays (34).

Biosecurity practices in pig farms, and farmers’ attitudes towards them, have been increasingly investigated in North America and Europe, with sow replacement policy, and protocols around trucking of livestock being consistently recognized as important components of a farm biosecurity protocol (35). These practices were not captured by our questionnaire, which was originally designed to capture data pertinent to development of antimicrobial resistance. The extent to which missing predictors/confounders may have affected our estimates is therefore unknown.

Our recruitment and ultimately our sample size was limited by both the lack of a complete sampling frame of pig farms in Canada, and the relatively recent swine flu pandemic of 2009, which participating veterinarians reported decreased farmer willingness to participate in this project. It is possible that a larger sample size, and therefore increased study power, would have resulted in more of the predictor variables being significantly associated with viral detection. Additionally, we sampled the finisher pigs at only one point in their lives. It is possible that some of the predictors investigated (e.g., all-in-all-out pig flow) might be significantly associated with delaying, but not preventing, viral infection. Longitudinal studies would be helpful in understanding viral kinetics, and the effects of predictors (i.e., biosecurity/management practices) on mean age of infection. We identified knowledge gaps in addition to the lack of a Canadian sampling frame. Detection of HEV, PEC, and RV on Canadian finisher farms suggests the potential for occupational exposure to these viruses amongst swine workers. Field surveys and research synthesis have supported the hypothesis that swine workers and veterinarians have increased odds of exposure to HEV (36,37). However, the extent of Canadian illness attributable to occupational exposure (if any) remains unknown. This reflects both the challenges in data collection (HEV infection is not federally notifiable in Canada) and possibly also incomplete understanding of the causal complement, or combination of risk factors, required to produce clinical disease after viral exposure.

In conclusion, all 3 viruses were detected in finishing pigs, suggesting occupational exposure to feeder pigs is a potential source of human exposure to these viruses. Overall, HEV had the greatest farm-level prevalence of the 3 viruses studied; sampling platform was a significant predictor of HEV and RV viral RNA detection in finishers in this dataset. The significance, effectiveness, and direction of effect of the biosecurity practices captured (provision of boots, or coveralls, on entry; requiring shower-in prior to entry; use of boot dip; posting biosecurity sign; requiring downtime prior to entry) varied across the 3 viruses studied. The management procedure consistently associated with greater odds of viral RNA detection for all 3 viruses, was obtaining feeder pigs from multiple sources.

Figure 1.

Figure 1

Open in a new tab

Study recruitment and sample collection process.

AAFC — Agriculture and Agri-Food Canada laboratory; CIPARS — Canadian Integrate Program for Antimicrobial Resistance Surveillance; FC — FoodNet Canada; HEV —Hepatitis E virus; PEC — Porcine enteric calicivirus; RT-PCR — Real time reverse transcriptase polymerase chain reaction; RV — Rotavirus.

Acknowledgments

We gratefully acknowledge the assistance of Bryan Bloomfield, Danielle Daignault, Louise Bellai, Karen Richardson, the CIPARS veterinarians, and the farmers participating in this study, as well as the funding support of OMAFRA — University of Guelph Partnership Research Program (Grant # 027118).

References

  • 1.Bank-Wolf BR, Konig M, Thiel H-J. Zoonotic aspects of infections with noroviruses and sapoviruses. Vet Microbiol. 2010;140:204–212. doi: 10.1016/j.vetmic.2009.08.021. [DOI] [PubMed] [Google Scholar]
  • 2.Martella V, Banyai K, Matthijnssens J, Buonavoglia C, Ciarlet M. Zoonotic aspects of rotaviruses. Vet Microbiol. 2010;140:246–255. doi: 10.1016/j.vetmic.2009.08.028. [DOI] [PubMed] [Google Scholar]
  • 3.Meng XJ. Hepatitis E virus: Animal reservoirs and zoonotic risk. Vet Microbiol. 2010;140:256–265. doi: 10.1016/j.vetmic.2009.03.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Wilhelm B, Waddell L, Greig J, Rajić A, Houde A, McEwen SA. A scoping review of the evidence for health risks of three emerging potentially zoonotic viruses: Hepatitis E virus, norovirus and rotavirus. Prev Vet Med. 2015;119:61–79. doi: 10.1016/j.prevetmed.2015.01.015. [DOI] [PubMed] [Google Scholar]
  • 5.Ward P, Müller P, Letellier A, et al. Molecular characterization of hepatitis E virus detected in swine farms in the province of Quebec. Can J Vet Res. 2008;72:27–31. [PMC free article] [PubMed] [Google Scholar]
  • 6.Leblanc D, Ward P, Gagné MJ, et al. Presence of hepatitis E virus in a naturally infected swine herd from nursery to slaughter. Int J Food Microbiol. 2007;117:160–166. doi: 10.1016/j.ijfoodmicro.2007.03.008. [DOI] [PubMed] [Google Scholar]
  • 7.Wilhelm B, Muellner P, Pearl DL, Rajić A, Houde A, McEwen SA. Preliminary molecular epidemiological investigation of hepatitis E virus sequences from Quebec, Canada. Prev Vet Med. 2015;118:359–369. doi: 10.1016/j.prevetmed.2014.12.011. [DOI] [PubMed] [Google Scholar]
  • 8.Mattison K, Shukla A, Cook A, et al. Human noroviruses in swine and cattle. Emerg Infect Dis. 2007;13:1184–1188. doi: 10.3201/eid1308.070005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.L’Homme Y, Sansregret R, Plante-Fortier E, et al. Genetic diversity of porcine Norovirus and Sapovirus: Canada, 2005–2007. [Last accessed December 4, 2015];Arch Virol. 2009 154:581–593. doi: 10.1007/s00705-009-0344-6. Available from: http://dx.doi.org/10.1007/s00705-009-0344-6. [DOI] [PubMed] [Google Scholar]
  • 10.Lamhoujeb S, Cook A, Pollari F, Bidawid S, Farber J, Mattison K. Rotaviruses from Canadian farm samples. Arch Virol. 2010;155:1127–1137. doi: 10.1007/s00705-010-0700-6. [DOI] [PubMed] [Google Scholar]
  • 11.Lachapelle V, Sohal JS, Lambert MC, et al. Genetic diversity of group A rotavirus in swine in Canada. [Last accessed December 4, 2015];Arch Virol. 2014 159:1771–1779. doi: 10.1007/s00705-013-1951-9. Available from: http://dx.doi.org/10.1007/s00705-013-1951-9. [DOI] [PubMed] [Google Scholar]
  • 12.Wilhelm B, Leblanc D, Houde A, et al. Survey of Canadian retail pork chops and pork livers for detection of hepatitis E virus, norovirus, and rotavirus using real time RT-PCR. Int J Food Microbiol. 2014;185:33–40. doi: 10.1016/j.ijfoodmicro.2014.05.006. [DOI] [PubMed] [Google Scholar]
  • 13.DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–188. doi: 10.1016/0197-2456(86)90046-2. [DOI] [PubMed] [Google Scholar]
  • 14.Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. Br Med J. 2003;327:557–560. doi: 10.1136/bmj.327.7414.557. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduction to meta-analysis. West Sussex, England: John Wiley and Sons; 2009. [Google Scholar]
  • 16.Sergeant ESG. Epitools epidemiological calculators [homepage on the Internet] [Last accessed January 5, 2015]. Available from: http://epitools.ausvet.com.au.
  • 17.Deckert A, Gow S, Rosengren L, et al. Canadian Integrated Program for Antimicrobial Resistance Surveillance (CIPARS) Farm Program: Results from finisher pig surveillance. [Last accessed December 4, 2015];Zoonoses Public Health. 2010 57(Suppl 1):71–84. doi: 10.1111/j.1863-2378.2010.01356.x. Available from: http://dx.doi.org/10.1111/j.1863-2378.2010.01356.x. [DOI] [PubMed] [Google Scholar]
  • 18.Farzan A, Friendship RM, Cook A, Pollari F. Occurrence of Salmonella, Campylobacter, Yersinia enterocolitica, Escherichia coli O157 and Listeria monocytogenes in swine. Zoonoses Public Health. 2010;57:388–396. doi: 10.1111/j.1863-2378.2009.01248.x. [DOI] [PubMed] [Google Scholar]
  • 19.Rabe-Hesketh S, Skrondal A, Picles A. Reliable estimation of generalized linear mixed models using adaptive quadrature. Stata J. 2002;2:1–21. [Google Scholar]
  • 20.Rabe-Hesketh S, Skrondal A. Multilevel and longitudinal modeling using Stata. 2nd ed. College Station, Texas: Stata Press; 2008. [Google Scholar]
  • 21.Huang FF, Haqshenas G, Guenette DK, et al. Detection by reverse transcription-PCR and genetic characterization of field isolates of swine hepatitis E virus from pigs in different geographic regions of the United States. J Clin Microbiol. 2002;40:1326–1332. doi: 10.1128/JCM.40.4.1326-1332.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.van Der Poel WH, Vinjé J, van Der Heide R, Herrera MI, Vivo A, Koopmans MP. Norwalk-like calicivirus genes in farm animals. Emerg Infect Dis. 2000;6:36–41. doi: 10.3201/eid0601.000106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Scheuer KA, Oka T, Hoet AE, et al. Prevalence of porcine noroviruses, molecular characterization of emerging porcine sapoviruses from finisher swine in the United States, and unified classification scheme for sapoviruses. J Clin Microbiol. 2013;51:2344–2353. doi: 10.1128/JCM.00865-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Anh PH, Carrique-Mas JJ, Cuong NV, et al. The prevalence and genetic diversity of group A rotaviruses on pig farms I the Mekong Delta region of Vietnam. Vet Microbiol. 2014;170:258–265. doi: 10.1016/j.vetmic.2014.02.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Monini M, Di Bartolo I, Ianiro G, et al. Detection and molecular characterization of zoonotic viruses in swine fecal samples in Italian pig herds. Arch Virol. 2015;160:2547–2556. doi: 10.1007/s00705-015-2538-4. [DOI] [PubMed] [Google Scholar]
  • 26.Walachowski S, Dorenlor V, Lefevre J, et al. Risk factors associated with the presence of hepatitis E virus in livers and seroprevalence in slaughter-age pigs: A retrospective study of 90 swine farms in France. Epidemiol Inf. 2014;142:1934–44. doi: 10.1017/S0950268813003063. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Correge I, Hemonic A, Gouvars B. Farming conditions and practices linked to Salmonella prevalence in slaughter pigs. Epidemiol Sante Anim. 2009;55:17–26. [Google Scholar]
  • 28.Lo Fo Wong DMA, Dahl J, Stege H, et al. Herd-level risk factors for subclinical Salmonella infection in European finishing-pig herds. Prev Vet Med. 2004;62:253–266. doi: 10.1016/j.prevetmed.2004.01.001. [DOI] [PubMed] [Google Scholar]
  • 29.Cunha JB, de Mendonca MC, Miagostovich MP, Leite JP. Genetic diversity of porcine enteric caliciviruses in pigs raised in Rio de Janeiro State, Brazil. Arch Virol. 2010;155:1301–1305. doi: 10.1007/s00705-010-0695-z. [DOI] [PubMed] [Google Scholar]
  • 30.Shen Q, Ren R, Zhang W, et al. Prevalence of hepatitis E virus and porcine caliciviruses in pig farms of Guizhou province, China. Hepat Mon. 2011;11:459–463. [PMC free article] [PubMed] [Google Scholar]
  • 31.Delgado-Rodriguez M, Llorca J. Bias. J Epidemiol Community Health. 2004;58:635–641. doi: 10.1136/jech.2003.008466. Available from: http://dx.doi.org/10.1136/jech.2003.008466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Canadian Swine Health Board. National swine farm-level biosecurity standard [homepage on the Internet] [Last accessed August 18, 2015]. Available from: http://www.cpc-ccp.com/biosecurity/national_swine_farm-level_biosecurity_standard_bst.php.
  • 33.Moutelíková R, Prodělalová J, Dufková L. Prevalence study and phylogenetic analysis of group C porcine rotavirus in the Czech Republic revealed a high level of VP6 gene heterogeneity within porcine cluster I1. Arch Virol. 2014;159:1163–1167. doi: 10.1007/s00705-013-1903-4. [DOI] [PubMed] [Google Scholar]
  • 34.Medici KC, Barry AF, Alfieri AF, Alfieri AA. Porcine rotavirus groups A, B, and C identified by polymerase chain reaction in a fecal sample collection with inconclusive results with polyacrylamide gel electrophoresis. J Swine Health Prod. 2011;19:146–150. [Google Scholar]
  • 35.Bottoms K, Poljak Z, Dewey C, Deardon R, Holtkamp D, Friendship R. Evaluation of external biosecurity practices on southern Ontario swine farms. Prev Vet Med. 2013;109:58–68. doi: 10.1016/j.prevetmed.2012.08.013. [DOI] [PubMed] [Google Scholar]
  • 36.Wilhelm BJ, Rajić A, Greig J, et al. A systematic review/meta-analysis of primary research investigating swine, pork or pork products as a source of zoonotic hepatitis E virus. Epidemiol Inf. 2011;8:1127–1144. doi: 10.1017/S0950268811000677. [DOI] [PubMed] [Google Scholar]
  • 37.Meng XJ, Wiseman B, Elvinger F, et al. Prevalence of antibodies to hepatitis E virus in veterinarians working with swine and in normal blood donors in the United States and other countries. J Clin Microbiol. 2002;40:117–122. doi: 10.1128/JCM.40.1.117-122.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Canadian Journal of Veterinary Research are provided here courtesy of Canadian Veterinary Medical Association

RESOURCES