SEARCH

SEARCH BY CITATION

Keywords:

  • Salmonella;
  • antimicrobial resistance;
  • risk factors;
  • swine;
  • multinomial regression

Summary

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

A multinomial logistic regression method was used to investigate the risk factors for antimicrobial resistance (AMR) in Salmonella isolated from faecal samples collected on 80 Ontario swine farms in Canada. The samples were classified into three groups including Salmonella-negative samples (S−), Salmonella-positive samples without AMR (S+AMR−) and Salmonella-positive samples with AMR (S+AMR+). The samples collected directly from pigs had a greater chance to be positive for Salmonella with AMR compared to those samples collected from the pen floor. The odds of culturing Salmonella with or without AMR was higher if pelleted feed was used compared with mash or liquid feed (< 0.001). The faecal samples collected on farrow-to-finish farms had a significant lower chance of testing positive for Salmonella with multidrug resistance than the samples from grow-finisher farms (= 0.004). The chance of culturing Salmonella without AMR on farms with a continuous system was higher than on farms with an all-in/all-out system (= 0.009). However, there was no significant association between the flow system and recovery of Salmonella with AMR. The larger farms were more likely to be in S+AMR+ group than in S− group (P < 0.001) whereas herd size did not appear as a risk factor for being in S+AMR− group compared with S− group. These findings indicate that although on-farm antimicrobial use is one component of resistance, there might be other farm management factors that also affect the development of emerging resistant bacterial foodborne pathogens on swine farms. Finding different risk factors for shedding Salmonella with or without antimicrobial resistance would help to take the appropriate approach to each group if a control programme were to be implemented or an intervention applied.


Impacts

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  • • 
    Development of resistance against antimicrobial agents in Salmonella on swine farms is a multi-factorial phenomenon.
  • • 
    Multinomial approach helps to differentiate the risk factors associated with shedding susceptible and drug-resistant Salmonella sp. and thus the appropriate intervention can be applied to each group.
  • • 
    Using mash or liquid feeding as well as screening commingling animals and replacement gilts for Salmonella can be considered possible solutions to limit Salmonella on swine farms.

Introduction

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

Salmonella enterica subspecies enterica are important foodborne pathogens (Mead et al., 1999), particularly serovars demonstrating multiple antimicrobial resistance (AMR) (Poppe et al., 1998; Threlfall et al., 2000). Food-producing animals including but not limited to pigs and pork products are claimed as an important source for antimicrobial resistant Salmonella (Nielsen and Wegener, 1997; Berends et al., 1998; Wegener et al., 2003). Multidrug-resistant Salmonella serovars are associated with increased hospitalization, mortality and consequent economic cost compared with susceptible strains (Travers and Barza, 2002; Martin et al., 2004) and these serovars are emerging as a concern for the global pork industry. In addition, antimicrobial resistance is reported in other bacteria that cause diseases in pigs (Hendriksen et al., 2008). This may cause serious problems in the treatment and control of infectious diseases on swine farms resulting in a significant economic impact. Antimicrobial resistance among bacterial populations of pigs has frequently been associated with mass medication with antimicrobials in feed and the isolates from conventionally reared animals are more frequently resistant compared with those from the antibiotic-free farms (Jacob et al., 2008). In-feed medication of post-weaned pigs has been associated with the antimicrobial resistance in Escherichia coli (Akwar et al., 2008) and Salmonella (Varga et al., 2009) in finisher pigs. However, multidrug-resistant Salmonella have been prevalent in antibiotic-free production systems (Thakur et al., 2007). Gebreyes et al. (2006) have shown that although prevalence of antimicrobial resistance was higher on conventional farms than antimicrobial-free farms, a penta-resistance pattern was significantly associated with antimicrobial-free operations. Inclusion of sub-therapeutic chlortetracycline (Funk et al., 2007) or apramycin and carbadox (Edrington et al., 2001) was not associated with the level of antimicrobial resistance in Salmonella in pigs. The contradictory findings regarding impact of antibiotic usage on development of resistance in bacteria may be explained by limited information available about antimicrobial usage in animals (McEwen and Fedorka-Cray, 2002). However, there might be other farm management factors that also affect the development of resistant bacterial foodborne pathogens on swine farms. The objective of this study was to investigate the associated management risk factors for shedding Salmonella with or without antimicrobial resistance on Ontario swine farms using a multinomial logistic regression method. Multinomial analysis will allow us to distinguish the risk factors associated with shedding drug-resistant Salmonella strains from those risk factors associated with shedding susceptible strains. This will help to apply the appropriate intervention to each category.

Materials and Methods

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

Study design and sample collection

Eighty commercial swine farms representative of the Ontario swine industry were visited between January and July 2004. The farms had originally been chosen from 5000 Ontario pig producers as a stratified random sample based on herd size. To have balanced geographic representation, the sampling strategy included farms in eastern Ontario and the Niagara region where there are relatively few pig farms. On each farm, one grower-finisher barn was chosen and within the barn five pens with the pigs closest to market weight were selected for sampling. Faecal samples were obtained from two pigs per pen (800 samples) and an additional pooled sample was collected from the fresh manure from five spots on the pen floor (397 samples). By sampling 10 pigs and five pooled samples in finisher barns holding 1000 pigs each on average, there is a 95% confidence that at least one positive sample would be identified if the minimum prevalence of disease was 15%. It assumes that the test has 100% specificity and 60% sensitivity and that the intracluster correlation coefficient is equal to 0.3 (Farzan, 2004).

Salmonella isolation

Faecal samples were cultured using non-selective pre-enrichment and selective enrichment based on protocol MFHP-20 (D’Aoust and Purvis, 1998) in the Laboratory Service Division (LSD), University of Guelph. Briefly, 25 g of faeces were added to 225 ml of buffered peptone water, incubated at 35°C for 18–24 h, and cultured either in Rappaport Vassiliadis broth or Tetrathionate Brilliant Green broth. After overnight incubation at 42.5°C, a loopful of each selective enrichment broth was plated on Brilliant Green Sulfa and Bismuth Sulfite agar. The Salmonella suspect colonies were streaked onto MacConkey agar, one presumptive Salmonella colony from each sample was tested on Triple Sugar Iron agar, Lysine Iron agar, Christensen’s Urea agar, and submitted to the Reference Laboratory for Salmonellosis, Laboratory for Foodborne Zoonoses, Public Health Agency of Canada, Guelph for serotyping. The isolates were stored at −70°C.

Antimicrobial susceptibility testing

Frozen Salmonella isolates were re-cultured on nutrient agar plates and tested for antimicrobial susceptibility by the agar dilution method described before (Poppe et al., 2001). Briefly, the isolates were cultured in Muller Hinton (MH) broth to obtain 0.5–1.0 McFarland density and using a Cathra Replicator plated onto MH agar plates containing antimicrobials (Sigma-Aldrich, St Louis, MO, USA). Susceptibility breakpoint levels and the reference strains were those described in the Clinical and Laboratory Standards Institute (CLSI) guidelines (CLSI, 2004, 2005). After 24-h incubation at 37°C, the plates were examined for bacterial growth and isolates that grew were considered to be resistant. The reference strains employed were Escherichia coli ATCC 25922, Escherichia coli ATCC 35218 and Pseudomonas aeruginosa ATCC 27853. We also employed the bovine strain R1022 possessing the aac(3)IV gene and resistant to apramycin, gentamicin and tobramycin and other antimicrobials.

Data collection

Data regarding risk factors at farm-level were extracted from two questionnaires. A questionnaire was completed at the time of the farm visit and information included pig flow (continuous or all-in/all-out), obvious presence of other animals (cats, dogs, wild birds, domestic poultry, cattle and rodents), number of finisher pigs at the site (herd size), type of operation (grower-finisher/farrow-to-finish), feed form (pelleted/mash/liquid), biosecurity and cleaning procedures when finisher pens were emptied. A biosecurity score was computed for each farm. The producers were asked about their boots, coveralls, bootdip, biosecurity sign, shower and downtime. One score was given to each question and a total biosecurity score was computed by adding the scores for all six questions. Also producers were asked questions regarding cleaning procedures (scraping, pressure washing with hot/cold water, using detergent, and/or disinfection after cleaning) when finisher pens are emptied. The possible responses to cleaning questions were: never (coded as 0), very rarely (coded as 1), sometimes (coded as 2), very often (coded as 3), or always (coded as 4) and a total cleaning score was obtained for each farm by adding the scores of all questions. To collect data for the presence of other animals, responses for each animal were scored as 1 (if seen on the farm), 2 (if seen on the site, not inside the barn), 3 (some evidence was in the barn), 4 (seen some animals in the room) and 5 (seen many in the room). The total score for ‘presence of other animals’ was calculated by adding the score for all animals. The data on in-feed and water antimicrobial usage was extracted from another questionnaire which had been administered 1 year prior to the study. Feed tags and drug-use records were used to confirm the questionnaire responses. A farm was classified as positive if an antimicrobial was added to feed and/or water.

Data analysis

Data were entered into an excel spreadsheet (Microsoft® Excel 2003) and imported into Stata (Stata 9.1 Intercooled for Windows XP; StataCorp, College Station, TX, USA). A Spearman rank correlation was applied to investigate the correlation between resistances to different antimicrobials. The hypothesis is that presence of resistance to any of tested antimicrobials increases resistance to others. It was also used to determine the herd-level correlation between serotypes and number of antimicrobial resistance on farm. The hypothesis tested is that shedding of Salmonella Typhimurium increases on-farm presence of antimicrobial resistance. A multinomial logistic regression modelling method described by Dohoo et al. (2003) was used to investigate the risk factors associated with Salmonella-positive samples without antimicrobial resistance (S+AMR− group) and Salmonella-positive samples with antimicrobial resistance (S+AMR+ group). This approach may be useful in understanding whether the risk factors for S+AMR− group are identical to those risk factors associated with S+AMR+ group. The model was fitted as Generalized Linear Latent and Mixed Model (GLLAMM) procedure. GLLAMMs are multilevel latent (random) variables models which can be fitted for different type of outcomes (continuous, categorical, multinomial, etc.). Our data had three hierarchical levels such as pig, pen and farm. Another advantage of the GLLAMM models is that more than one random effect can be included into model. Thus, pen and farm variables were included as latent variables (random effects). The probability of being in S+AMR− group and S+AMR+ group was obtained by fitting two separate logistic models whereas S− group served as the base group. The predicted probability for a sample being in S− group (P1), S+AMR− group (P2) or S+AMR+ group (P3) is computed as:

  • image
  • image
  • image

The univariable association between the risk factors and outcome was performed using a single multinomial logistic regression. A likelihood ratio test was used to test the significance of the coefficients in the model. The independent variables with a < 0.2 in the univariable analysis were selected for initial inclusion in the final multinomial model. The models were manually built using a backward elimination process and variables with a P > 0.05 in the likelihood ratio test were excluded. The likelihood ratio test was used to determine the overall significance of the model. The colinearity between the independent variables including ‘herd size’, ‘biosecurity’, ‘cleaning’ and ‘other animals’ variables was evaluated by means of a Pearson product moment coefficient of correlation. The presence of confounding was investigated by looking at the effect of each predictor on the coefficients of other variables as they entered and were removed from the model. Variables were considered confounders if the change in coefficient was greater than 25%. Two-way interactions were tested between the main effects, i.e. ‘feed’, ‘site’, ‘flow’, ‘herd size’ and ‘sample’ variables in the final model. To assess the assumption of linearity between the outcome and the continuous variable (herd size) predicted probability of being in S− group, S+AMR− group or S+AMR+ group was plotted against the herd size. Also the square-transformation of the herd-size variable was created and its significance in the model was assessed by use of P < 0.05 in the likelihood ratio test. In another approach, herd sized was categorized and three categories were included into model. For feed variable, mash and liquid were collapsed in one category and pelleted group was used as reference. Also a model was fitted with three dummy variables for feed, that is, pelleted, mash, and liquid. An Akaike’s Information Criterion (AIC) was used to determine the best model.

Results

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

Antimicrobial resistance

Salmonella was cultured from 164 (13.7%) of 1197 samples. Thirty-eight (23.2%) of the isolates displayed no antimicrobial resistance whereas the remaining 126 isolates were resistant to 1–10 antimicrobials including tetracycline (69.5%), streptomycin (65.2%), sulfisoxazole (64%), ampicillin (61%), spectinomycin (59.1%), chloramphenicol (57.9%), florfenicol (57.9%), kanamycin (14%), neomycin (14%) and nitrofurantoin (1.2%). All isolates were susceptible to amikacin, amoxicillin/clavulanic acid, apramycin, carbadox, cefoxitin, ceftiofur, ceftriaxone, cephalothin, ciprofloxacin, gentamicin, nalidixic acid, sulfamethoxazole/trimethoperim and tobramycin.

Fifteen antimicrobial resistance patterns were observed among Salmonella isolates (Table 1). The ‘ACFSpSSuT’ and ‘ACFKNSpSSuT’ were the most common resistance carried by S. Typhimurium var. Copenhagen and S. Typhimurium. The two patterns were determined on nine and six farms, respectively. One, two and three antimicrobial resistance patterns were determined on 20, four and one farms, respectively. Of 164 isolates, 101 of these were S. Typhimurium (including var. Copenhagen) and had resistance to 3–10 antimicrobials whereas the 18 S. Infantis were susceptible to all tested antimicrobials.

Table 1.   Antimicrobial resistance pattern among Salmonella serovars isolated on Ontario swine farms
AMR patternSerovar (no. isolates/no. farms)
  1. A, ampicillin; C, chloramphenicol; F, florfenicol; K, kanamycin; N, neomycin; Nit, nitrofurantoin; Sp, spectinomycin; S, streptomycin; Su, sulfisoxazole; T, tetracycline.

ACFKNNitSpSSuTTyphimurium var. Copenhagen (2/1)
ACFKNSpSSuTTyphimurium var. Copenhagen (30/6), I:4,12:-:- (1/1)
ACFSpSSuTTyphimurium var. Copenhagen (50/8), Typhimurium (6/3), I:4,12:i:- (1/1)
ACFSSuTTyphimurium var. Copenhagen (4/1)
AKNSSuTTyphimurium var. Copenhagen (2/1)
CFSpSSuTDerby (1/1)
AKNSTTyphimurium (2/1)
SpSSuTAgona (5/1), Derby (3/2), Typhimurium var. Copenhagen (2/1)
ASSuTTyphimurium var. Copenhagen (2/2)
SpSSuTyphimurium (1/1)
KNDerby (2/1)
SuTHavana (1/1)
NitDerby (5/1), I: Rough-O:fg:- (1/1)
TLondon (3/1)
AHavana (2/1)
NoneInfantis (18/8), Rough (5/2), Agona (4/2), Havana (3/1), Putten (2/1), Brandenberg (2/1), Senftenberg (1/1), Ohio (1/1), I: 6,7 (1/1), I:28 (1/1)
Total164/37

There was strong correlation among resistance to tetracycline, streptomycin, sulfisoxazole, ampicillin, spectinomycin and chloramphenicol at the farm-level (Spearman’s rank correlation coefficient = 0.84–0.99, < 0.001) (Table 2). There was also a farm-level correlation between resistance to those six antimicrobials and the isolate being a S. Typhimurium var Copenhagen (Spearman’s rank correlation coefficient = 0.81–0.91, < 0.001).

Table 2.   Spearman’s rank correlation between the six most frequent antimicrobial resistance and S. Typhimurium var. Copenhagen (< 0.001)
 TSSuASpC
Tetracycline (T)      
Streptomycin (S)0.91     
Sulfisoxazole (Su)0.910.99    
Ampicillin (A)0.890.960.97   
Spectinomycin (Sp)0.840.920.910.86  
Chloramphenicol (C)0.840.900.920.930.90 
S. Typhimurium var. Copenhagen0.830.910.900.900.810.81

Multinomial logistic regression model for risk factors

Salmonella was not cultured from any sample collected on 43 farms. Susceptible Salmonella were isolated from at least one sample on 12 farms and drug-resistant Salmonella were recovered from at least one sample on 25 farms. The distribution of risk factors among those farms is shown in Table 3. The samples were classified into three groups such as 1033 S− group, 38 S+AMR− group and 126 S+AMR+ group. In the single multinomial logistic regression analysis, six independent variables including feed, type of site, type of sample, pig flow, antimicrobial usage and herd size were selected (< 0.2 in the likelihood ratio test) for inclusion in the final multinomial logistic model.

Table 3.   Distribution of risk factors among 80 Ontario swine farms
Risk factors No. farms (percent)
S− (43 farms)S+AMR− (12 farms)S+AMR+ (25 farms)Total (80 farms)
  1. SD, standard deviation.

  2. aNo. pigs on site.

Feed formMash35 (81.4)9 (75)12 (48)56 (70)
Pellet2 (4.7)2 (16.7)9 (36)13 (16.2)
Liquid6 (13.9)1 (8.3)4 (16)11 (13.8)
Type of siteGrower-finisher11 (25.6)3 (25)17 (68)31 (38.7)
Farrow-to-finish32 (74.4)9 (75)8 (32)49 (61.3)
Antimicrobial usageYes34 (79.1)10 (83.3)20 (80)64 (80)
No9 (20.9)2 (16.7)5 (20)16 (20)
Herd sizeaMean93397913161060
SD665734967791
FlowContinuous22 (51.2)8 (66.7)13 (52)43 (53.8)
All-in/all-out21 (48.8)4 (33.3)12 (48)37 (46.2)

In the final model, feed and type of sample were associated with both S+AMR− and S+AMR+ group classification (Table 4). The samples collected from the pen floor had a lower chance of being positive for drug-resistant Salmonella compared with those samples which were collected directly from pigs (< 0.001). The odds of culturing Salmonella with or without AMR was higher if pelleted feed was used compared to mash or liquid feed (< 0.001). The faecal samples collected on farrow-to-finisher operations had a significant lower chance of testing positive for Salmonella with multidrug resistance than the samples from grower-finisher farms (= 0.004). The chance of culturing Salmonella without AMR on farms with a continuous system was higher than on farms with an all-in/all-out system (= 0.009). However, there was no significant association between the flow system and recovery of Salmonella with AMR. The larger farms were more likely to be in S+AMR+ group than in S− group (P < 0.001) whereas herd size did not appear as a risk factor for being in S+AMR− group compared with being in S− group. The shape of relationship between herd size and predicted probability of being in either S+AMR− or S+AMR+ is compared with shape of relationship between herd size and predicted probability of being in Salmonella negative group (Fig. 1). The ‘feed’ variable changed the significance level of ‘herd size’ variable indicating that the association between ‘type of feed’ and probability of being in S− group, S+AMR− group or S+AMR+ group is confounded by herd size. There was no evidence of interaction at the level of < 0.05.

Table 4.   Multinomial logistic regression for risk factors associated with Salmonella with or without AMR
GroupaRisk factor CoefficientbStandard errorConfidence interval (95%)P-value
  1. aS− group : 1033 Salmonella-negative samples (reference); S+AMR− group: 38 Salmonella-positive samples without AMR; S+AMR+ group: 126 Salmonella-positive samples with AMR.

  2. bCoefficient is controlled for other variables in the model.

  3. cNo. pigs on site.

S+AMR− groupFeedPellet3.210.921.4, 5.00.001
Mash or liquidRef   
SamplePig0.920.380.17, 1.660.016
PenRef   
SiteGrower-finisher1.400.710.009, 2.780.049
Farrow-to-finishRef   
FlowContinuous1.761.360.47, 3.080.01
Al-in/all-outRef   
Herd sizec 0.00050.0004−0.0003, 0.0010.2
S+AMR+ groupFeedPellet3.580.841.92, 5.24<0.001
Mash or liquidRef   
SamplePig1.140.270.60, 1.67<0.001
PenRef   
SiteGrower-finisher1.860.640.60, 3.110.004
Farrow-to-finishRef   
FlowContinuous1.160.96−0.51, 2.740.1
Al-in/all-outRef   
Herd sizec 0.0010.00030.0005, 0.002<0.001
Log likelihood (null) = −449.5; LR (full) = −415.70; LR χ= 67.6, df = 10, < 0.001
image

Figure 1.  The relationship between number of pigs on farm and antimicrobial resistance in Salmonella.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

The results of multinomial logistic analysis indicate that development of resistance against antimicrobial agents in Salmonella in swine populations may be a multi-factorial phenomenon. In the present study, in-feed and water antimicrobial usage was not associated with Salmonella shedding either susceptible (S+AMR− group) or resistant (S+AMR+ group) isolates. The data on drug use in this study were collected 1 year prior to the faecal collection and the drug-use policy may have changed on certain farms. This potential misclassification bias is a possible explanation for the lack of association between antimicrobial usage and Salmonella shedding. Surveillance programmes have shown some limitations to provide data on the antimicrobial usage in animals (McEwen and Fedorka-Cray, 2002). Pig flow, type of site and herd size did not appear to impact the level of on-farm drug use and this might be due to the limitations to collect drug-use data. However, lack of true association or sample size issue may be considered as other explanations.

Multi-drug resistant Salmonella isolates were less likely present in the pooled samples collected from the pen floor compared with those samples collected directly from pigs. Although some antimicrobial agents may be present in manure (Loke et al., 2002) and lead to development of antimicrobial resistance, it is possible that drug-resistant Salmonella isolates die earlier than susceptible isolates. However, pooled samples were collected from different spots on the pen floor and each pooled sample may represent more than one pig and this might increase the probability of being positive for Salmonella isolates without AMR.

The faecal samples collected on farrow-to-finish farms had a significantly lower chance of testing positive for Salmonella with multidrug resistance than the samples from grow-finisher farms. This may be explained in part by the fact that commingling pigs from different sources is more likely on grower-finisher operations than on farrow-to-finish farms. By introducing pigs from multiple sources, the producers may increase the likelihood of introducing resistant strains of Salmonella.

Pelleted feed has been reported before as a common risk factor for increasing Salmonella occurrence in swine (von Altrock et al., 2000; Kranker et al., 2001; Lo Fo Wong et al., 2004) and the use liquid feed has the potential to decrease prevalence of Salmonella (Farzan et al., 2006). This is a potential for pelleted feed to be contaminated during transport (Fedorka-Cray et al., 1997). This form of feed also decreases the lactic acid bacterial population in the gut (van Winsen et al., 2002) which may explain the association between pelleted feed and Salmonella prevalence in pigs.

The impact of herd size and pig flow on Salmonella on swine farms has been investigated before and some studies found those as risk factors for Salmonella (Carstensen and Christensen, 1998; van der Wolf et al., 2001; Farzan et al., 2006), whereas some researchers reported no association between herd size and Salmonella (Lo Fo Wong et al., 2004). The variation in findings regarding the association between herd size and Salmonella might be explained by the fact that herd size is a complex risk factor per se and includes other related parameters. In the current study, for example, the ‘herd size’ was recorded as only the number of finisher pigs at the site, yet it may represent other related parameters (i.e. the number of sows, nurseries and gilts, the number of unit and the size of production). This may result in a higher risk of introduction of Salmonella into farms and its transmission between pigs in the larger farms.

All-in/all-out pig flow is shown as a practice to limit spread of Salmonella (Farzan et al., 2006). Continuous flow, however, did not appear to be a risk factor for samples testing positive for Salmonella with AMR (S+AMR+ group) although it had a significant effect on increasing the probability of classifying a sample as susceptible (S+AMR- group). As antimicrobial resistance was associated with serovar, the impact of pig flow and herd size on occurrence of Salmonella might be serovar-dependent. However, sample size considerations may account for apparent lack of association between continuous flow and recovery of multi-drug resistant Salmonella.

The variation in findings among different studies regarding association between risk factors and Salmonella on swine farms might be attributed to whether prevalence of Salmonella was considered as outcome or the outcome was categorized into resistant/susceptible categories. In this study, for example, when refining outcome into three distinct categories, herd size seemed to increase the risk of Salmonella with antimicrobial resistance but was not associated with risk of Salmonella without antimicrobial resistance. In the present study, AMR was correlated with S. Typhimurium var. Copenhagen. The correlation between antimicrobial resistance Salmonella serotypes and AMR has been reported before (Rajić et al., 2004; Nollet et al., 2006). It has been shown that members of serogroup B are more likely to exhibit resistance than other serogroups (Edrington et al., 2001) and AMR is mostly reported in association with S. Typhimurium DT104 (Gebreyes et al., 2004). Therefore, the risk factors for being in S+AMR+ group may represent the risk factors for S. Typhimurium var. Copenhagen on swine farms.

To our knowledge, this is the first study investigating the risk factors for shedding Salmonella with or without antimicrobial resistance. This approach allowed us to include Salmonella-negative samples in the analysis and thus increased the statistical power of the study compared with using a binomial logistic regression analysis. Also we determined the risk factors associated with each category and this may help to apply the appropriate intervention to each group. These findings indicated that using mash or liquid feeding as well as screening commingling animals and replacement gilts for Salmonella can be considered possible solutions to limit Salmonella on swine farms.

Acknowledgements

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

We would like to acknowledge the Public Health Agency of Canada, the Ontario Ministry of Agriculture, Food and Rural Affairs, Ontario Pork, the Canadian Research Institute for Food Safety (CRIFS) for the financial and technical support. We also thank the research technicians and producers who participated in the project.

References

  1. Top of page
  2. Summary
  3. Impacts
  4. Introduction
  5. Materials and Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  • Akwar, H. T., C. Poppe, J. Wilson, R. Reid-Smith, M. Dyck, J. Waddington, D. Shang, and S. A. McEwen, 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, 202210.
  • Von Altrock, A., A. Schutte, and G. Hildebrandt G, 2000: [Results of the German investigation in the EU Project “Salmonella in Pork (Salinpork)”-2. Investigations in a slaughterhouse]. Berl. Munch. Tierarztl. Wochenschr.; 113, 225233.
  • Berends, B. R., F. Van Knapen, D. A. Mossel, S. A. Burt, and J. M. Snijders, 1998: Impact on human health of Salmonella spp. on pork in the Netherlands and the anticipated effects of some currently proposed control strategies. Int. J. Food Microbiol. 44, 219229.
  • Carstensen, B., and J. Christensen, 1998: Herd size and sero-prevalence of Salmonella enterica in Danish swine herds: a random-effects model for register data. Prev. Vet. Med. 34, 191203.
  • CLSI (Clinical and Laboratory Standards Institute), 2004: Performance Standards for Antimicrobial Disk and Dilution Susceptibility Tests for Bacteria Isolated from Animals, Information Supplement. Updated tables for the NCCLS antimicrobial susceptibility testing standard M31-A2. Clinical and Laboratory Standards Institute, Wayne, PA.
  • CLSI (Clinical and Laboratory Standards Institute), 2005. Performance Standards for Antimicrobial Susceptibility Testing, Fifteenth Informational Supplement M100-S15, pp. 25, 102–106. Clinical and Laboratory Standards Institute, Wayne, PA.
  • D’Aoust, J. Y., and U. Purvis 1998: Isolation and identification of Salmonella from foods, HPB Method MFHPB-20. Research Division, Health Protection Branch, Public Health Agency of Canada, Ontario, Ottawa, Full document. Available at: http://www.hc-sc.gc.ca/fn-an/res-rech/analy-meth/microbio/volume2/mfhpb20-01_e.html.
  • Dohoo, I., W. Martin, and H. Stryhn, 2003: Veterinary Epidemiologic Research. AVC Inc., University of Prince Edward Island, PE, Canada.
  • Edrington, T. S., R. B. Harvey, L.A. Farrington, and D. J. Nisbet, 2001: Evaluation of subtherapeutic use of the antibiotics apramycin and carbadox on the prevalence of antimicrobial-resistant Salmonella infection in swine. J. Food. Prot. 64, 20672070.
  • Farzan, A. 2004: The difference of Salmonella prevalence in grower-finisher pig operations with liquid and dry feeding. MSc thesis, University of Guelph, Guelph, ON.
  • Farzan, A., R. M. Friendship, C. E. Dewey, K. Warriner, C. Poppe, and K. Klotins, 2006: Prevalence of Salmonella spp. on Canadian pig farms using liquid or dry-feeding. Prev. Vet. Med. 73, 241254.
  • Fedorka-Cray, P. J., A. Hogg, J. T. Gray, K. Lorenzen, J. Velasquez, and P. Von Behren, 1997: Feed and feed trucks as sources of Salmonella contamination in swine. J. Swine Health Prod. 5, 189193.
  • Funk, J., T. E. Wittum, J. T. Lejeune, P. J. Rajala-Schultz, A. Bowman, and A. Mack, 2007: Evaluation of stocking density and subtherapeutic chlortetracycline on Salmonella enterica subsp. enterica shedding in growing swine. Vet. Microbiol. 24, 202208.
  • Gebreyes, W. A., P. R. Davies, P. K. Turkson, W. E. Morrow, J. A. Funk, C. Altier, and S. Thakur, 2004: Characterization of antimicrobial-resistant phenotypes and genotypes among Salmonella enterica recovered from pigs on farms, from transport trucks, and from pigs after slaughter. J. Food. Prot. 67, 698705.
  • Gebreyes, W. A., S. Thakur, and W. E. Morrow, 2006: Comparison of prevalence, antimicrobial resistance, and occurrence of multidrug-resistant Salmonella in antimicrobial-free and conventional pig production. J. Food. Prot. 69, 743748.
  • Hendriksen, R. S., D. J. Mevius, A. Schroeter, C. Teale, E. Jouy, P. Butaye, A. Franco, A. Utinane, A. Amado, M. Moreno, C. Greko, K. D. Stärk, C. Berghold, A. L. Myllyniemi, A. Hoszowski, M. Sunde, and F. M. Aarestrup, 2008: Occurrence of antimicrobial resistance among bacterial pathogens and indicator bacteria in pigs in different European countries from year 2002–2004. Acta Vet. Scand. 13, 50.
  • Jacob, M. E., J. T. Fox, S. L. Reinstein, and T. G. Nagaraja, 2008: Antimicrobial susceptibility of foodborne pathogens in organic or natural production systems: an overview. Foodborne Pathog. Dis. 5, 721730.
  • Kranker, S., J. Dahl, and A. Wingstrand, 2001: Bacteriological and serological examination and risk factor analysis of Salmonella occurrence in sow herds, including risk factors for high Salmonella seroprevalence in receiver finishing herds. Berl. Munch. Tierarztl. Wochenschr. 114, 350352.
  • Lo Fo Wong, D. M., J. Dahl, H. Stege, P. J. Van Der Wolf, L. Leontides, A. Von Altrock, and B. M. Thorberg, 2004: Herd-level risk factors for subclinical Salmonella infection in European finishing-pig herds. Prev. Vet. Med. 62, 253266.
  • Loke, M. L., J. Tjørnelund, and B. Halling-Sørensen, 2002: Determination of the distribution coefficient (log Kd) of oxytetracycline, tylosin A, olaquindox and metronidazole in manure. Chemosphere 48, 351361.
  • Martin, L. J., M. Fyfe, K. Dore, J. A. Buxton, F. Pollari, B. Henry, D. Middleton, R. Ahmed, F. Jamieson, B. Ciebin, S. A. McEwen, and J. B. Wilson, 2004: Increased burden of illness associated with antimicrobial-resistant Salmonella enterica serotype Typhimurium infections. J. Infect. Dis. 189, 377384.
  • McEwen, S. A., and P. J. Fedorka-Cray, 2002: Antimicrobial use and resistance in animals. Clin. Infect. Dis. 34, 93106.
  • Mead, P. S., L. Slutsker, V. Dietz, L. F. McCaig, J. S. Bresee, C. Shapiro, P. M. Griffin, and R. V. Tauxe, 1999: Food-related illness and death in the United States. Emerg. Infect. Dis. 5, 607625.
  • Nielsen, B., and H. C. Wegener, 1997: Public health and pork and pork products: regional perspectives of Denmark. Rev. Sci. Tech. 16, 513524.
  • Nollet, N., K. Houf, J. Dewulf, B. Catry, L. De Zutter, A. De Kruif, and D. Maes, 2006: Variability in antimicrobial resistance among Salmonella enterica strains from fattening pigs and sows. Microb. Drug. Resist. 12, 7481.
  • Poppe, C., N. Smart, R. Khakhria, W. Johnson, J. Spika, and J. Prescott, 1998: Salmonella Typhimurium DT104: a virulent and drug-resistant pathogen. Can. Vet. J. 39, 559565.
  • Poppe, C., M. Ayroud, G. Ollis, M. Chirino-Trejo, N. Smart, S. Quessy, and P. Michel, 2001: Trends in antimicrobial resistance of Salmonella isolated from animals, foods of animal origin, and the environment of animal production in Canada, 1994–1997. Microb. Drug. Resist. 7, 197212.
  • Rajić, A., M. E. McFall, A. E. Deckert, R. Reid-Smith, K. Manninen, C. Poppe, C. E. Dewey, and S. A. McEwen, 2004: Antimicrobial resistance of Salmonella isolated from finishing swine and the environment of 60 Alberta swine farms. Vet. Microbiol. 04, 189196.
  • Thakur, S., D. A. Tadesse, M. Morrow, and W. A. Gebreyes, 2007: Occurrence of multidrug resistant Salmonella in antimicrobial-free (ABF) swine production systems. Vet. Microbiol. 125, 362367.
  • Threlfall, E. J., L. R. Ward, J. A. Frost, and W. A. Gebreyes, 2000: Spread of resistance from food animals to man-the UK experience. Acta Vet. Scand. 93, 6374.
  • Travers, K., and M. Barza, 2002: Morbidity of infections caused by antimicrobial-resistant bacteria. Clin. Infect. Dis. 34, 131134.
  • Varga, C., A. Rajić, M. E. McFall, R. J. Reid-Smith, and S. A. McEwen, 2009: Associations among antimicrobial use and antimicrobial Resistance of Salmonella spp. isolates from 60 Alberta finishing swine farms. Foodborne Pathog. Dis. 6, 2331.
  • Wegener, H. C., T. Hald, D. Lo Fo Wong, M. Madsen, H. Korsgaard, F. Bager, P. Gerner-Smidt, and K. Mølbak, 2003: Salmonella control programs in Denmark. Emerg. Infect. Dis. 9, 774780.
  • Van Winsen, R. L., D. Keuzenkamp, B. A. Urlings, L. J. Lipman, J. A. Snijders, J. H. Verheijden, and F. Van Knapen, 2002: Effect of fermented feed on shedding of Enterobacteriaceae by fattening pigs. Vet. Microbiol. 87, 267276.
  • Van Der Wolf, P. J., W. B. Wolbers, A. R. Elbers, H. M. Van Der Heijden, J. M. Koppen, W. A. Hunneman, F. W. Van Schie, and M. J. Tielen, 2001: Herd level husbandry factors associated with the serological Salmonella prevalence in finishing pig herds in the Netherlands. Vet. Microbiol. 78, 205219.