Risk Factors Associated with Salmonella enterica Prevalence in Three-site Swine Production Systems in North Carolina, U.S.A.
Julie A. Funk, Peter R. Davies, Wondwossen A. Gebreyes
Introduction
In the USA, there is increasing consolidation and vertical integration of the pork industry (Anonymous, 1997b). A growing proportion of swine are produced in large scale, multiple-site systems that segregate production into 2 or 3 sites. These systems were initially designed with control of production impairing diseases as the primary goal (Harris, 1988). In North Carolina, USA, the predominant systems for swine production consist of 3 sites, segregated into breeding sows/boars and suckling pigs; nursery pigs; and finisher pigs.
Previous data from our laboratory (Davies et al., 1997) and others (Anonymous, 1997a) indicated that there was a relatively high prevalence of SE positive swine farms in North Carolina. In addition to a large number of positive farms, previous research (Davies et al., 1999; Davies et al., 2000; Funk et al., 2001) indicated that there was substantial temporal variability in fecal shedding prevalence within and among groups of pigs raised in the same 3-site systems. This suggested that risk factors associated with the cohort level, in addition to farm or company level factors may be important for control of SE fecal shedding.
We conducted an intensive, long-term investigation to identify potential management and environmental risk factors associated with SE prevalence in market age pigs reared in large 3-site swine production systems.
Materials and Methods
Two swine production companies in North Carolina, USA were purposively selected based on three criteria: 1) 3-site production systems 2) All-in/all-out (AIAO) pig flow and 3) Willingness to cooperate.
The study design is fully nested within company. For each company, three nurseries, and for each nursery three finishing sites were selected for monitoring. For every finisher site, three cohorts of finishing pigs were surveyed. Therefore, 54 cohorts (2 companies X 3 nurseries X 3 finishers X 3 cohorts) of pigs were intended for investigation. Due to changes in pig flow, 49 cohorts were completed.
For each cohort of pigs, 96 (10 g) fecal samples were collected from randomly selected individual pigs within 1 month of marketing. Bacteriological culture was conducted with standard methods previously described (Funk et al., 2000).
Data collection regarding management practices was recorded at two levels, finishing site and cohort (Table 1). The high and low daily temperature was recorded from the nearest National Weather Service Station for each day of the finishing period from placement until
sampling.
Table 1. Data collection.
Finisher site characteristics | Growth performance (nursery and finisher) |
---|---|
Number of barns | Number placed |
Age of barns | Date of placement |
Target placement | Average weight at placement |
Ventilation system | Average daily gain |
Area per Pen (m2) | Feed conversion |
Number of water spaces/pen | Mortality (%) |
Distance from nursery | Average weight at close |
Type of feeders | Number and date of pigs sold pre- sampling |
Waste disposal and handling | Number of pigs sold |
Presence of a gate or fence | Medication Usage (nursery and finisher) |
Water source | Medications used and means of delivery |
Prerequisites for farm entry | Cost of medication used per pig |
Use of boot baths | Disinfectant used prior to pig placement |
Clothing and boots used only at site | Weather (finisher) |
Bird-proof buildings | Daily high (°C) for each day pre- sampling |
Other domestic species at site | |
Number of people in houses daily | |
Presence of toilet/shower |
Two logistic regression models were constructed for risk factor analysis as described by Hosmer and Lemeshow (1989) using Egret software (Cytel Corp., Cambridge, MA, USA). For Model 1 the prevalence cut-off value to define the outcome variables corresponds to the 75th percentile (18.75%) prevalence for all cohorts. For Model 2, the cut-off value was set at the median (10%) prevalence (range 0-71.9%, mean 16.1%). For both models, the lower prevalence category was the referent group. All explanatory variables were dichotomously categorized based on graphical representation and descriptive statistics. Inclusion of a random effects term for both nursery source and finisher site was attempted after selection of the final model. There was no significant explanation of excess variation in either model (p > 0.10).
Results
SE infection was common, as only 4/49 cohorts were SE negative pre-market. Seven hundred and fifty-four SE isolates were identified, representing 32 different serotypes. Typhimurium (including the Copenhagen variant) was the most frequently isolated serotype (28%).
The final logistic regression models are shown in Tables 2 and 3.
Discussion
Finisher level factors
The absence of a toilet was associated with finisher prevalence greater than 18.75% (Model 1). Typically, finisher sites are not located at the same location as the producer’s home. Although a human-pig transmission cycle is possible, it is more likely that the decision to incur the cost of putting in washroom facilities serves as a proxy for opinions regarding hygiene. Hygiene has been stressed as important for SE control (Berends et al., 1996). Lo Fo Wong et al. (2001) described routine hand-washing (in concert with batch production) as a practice associated with decreased S. enterica seroprevalence at slaughter in Danish herds.
Cohorts with prevalence greater than 10% (Model 2) were at increased odds for high human traffic. It has been suggested that the number of visitors to a farm is a component of hygiene risk for SE on swine farms (Berends et al., 1996) but no association was found between visitor restriction protocols and SE status of broiler chickens (Renwick et al., 1992).
The presence of other domestic animals on site was associated with finisher prevalence greater than 10% (Model 2). Bahnson et al. (2001) also found that contact with other domestic animals was moderately associated with increased seroprevalence But other authors have not shown an association with the presence of domestic animals other than the target species and SE status (Renwick et al, 1992; Henken et al., 1992;Rose et al., 1999; Kabagambe et al., 2000).
Cohort level factors
Above median finisher feed conversion was positively associated with the high prevalence group in Model 1(>18.75%). No clinical salmonellosis was diagnosed in this study. Management practices that result in poor feed efficiency may promote transmission and shedding of SE. The association between growth performance and SE prevalence agrees with the results of Baum et al. (1998).
Above median pig density (decreased m2/ pig) at the time of sampling was associated with increased SE prevalence (Models 1 and 2). The variation in stocking density in this study was a function of the number of pigs marketed prior to sampling. Although there are known impacts related to stocking density for growth performance in swine (Randolph et al., 1981), the data regarding animal density on SE shedding in swine is sparse. Linton et al. (1970) identified higher rates of infection in pens with higher pig density, but this result was not repeatable on subsequent sampling in the same herd. Low stocking density at sampling was equivalent to sampling later marketing groups. Morrow et al. (1999) described that pigs in later marketing groups had a decreased prevalence of isolation of SE from cecal contents at slaughter. Conversely, Bahnson and Fedorka-Cray (1997) reported that the last group sold from a batch of pigs was at greater risk to be categorized as high prevalence based on lymph node culture at slaughter.
High prevalence cohorts (>18.75%, Model 1) were at greater odds to have been sampled in winter or spring. This result is similar to that reported by Christensen and Rudemo (1998) who reported higher SE seroprevalence in swine in winter compared to summer. On the other hand, Baum et al. (1998) found that herds tested in summer and fall had higher seroprevalence than herds tested in winter and spring. Cohorts with prevalence >10% had increased odds of being sampled when the high temperature was below the median (<21.7°C) the day of sampling (Model 2). This result is in agreement with the seasonal association of increased prevalence in cooler seasons.
Above median variability of the daily high temperature during the finishing period was positively associated with increased SE prevalence (Model 1, >18.75%). Periods of wide temperature variation are a challenge for ventilation management. All finishing barns were naturally ventilated, curtain-sided buildings.
Conclusions
The results of this study suggest that both finisher site and cohort level management and environmental factors were associated with the prevalence of pigs shedding SE at the end of the finishing phase in 3-site systems in North Carolina, USA. The implication for design of epidemiological investigations is that considerable resources will be necessary to identify factors that vary with cohort. The association of prevalence with timing of sampling in relationship to marketing groups and pig density is problematic for design of investigations, and presents considerable difficulty in using fecal culture to determine pre-harvest risk.
Despite these limitations, there is indication that improvement in farm biosecurity and management practices may help control SE shedding. Fortunately, these practices are associated with decreased risk for introduction of production-impairing diseases and improved growth performance. These benefits should help offset intervention costs, as SE control in itself does not result in immediate economic benefit for most producers in the US pork industry.
References
Anonymous (1997a): Shedding of Salmonella by finisher hogs in the U.S. Info Sheet N223.196, United States Department of Agriculture, Animal and Plant Inspection Service, Veterinary Services, National Animal Health Monitoring System.
Anonymous(1997b): Part III: 1990-1995 Changes in the U.S. pork industry. Report N248.1097, United States Department of Agriculture, Animal and Plant Inspection Service, Veterinary Services, National Animal Health Monitoring System.
Bahnson, P.B. and Fedorka-Cray, P.J. (1997): The associations between herd characteristics and Salmonella in slaughter age pigs. Proceedings of the Second International Symposium of Epidemiology and Control of Salmonella in Pork, Copenhagen, Denmark, 145-147.
Bahnson, P.B., Mateus-Pinella, N., Omran, L.M., Grass, J., Fransen, L., Fedorka-Cray, P.J., 2001b. Risk factors for high levels of antibody to Salmonella spp. Among market weight pigs. Proceedings of the 4th International Symposium on the Epidemiology and Control of Salmonella and other Food borne pathogens in pork, Leipzig, Germany, pp. 250-252.
Baum, D.H., D.L. Harris, B. Nielsen, P.J. Fedorka-Cray (1998): Risk Factors Associated with Increased Seroprevalence of Salmonella in Finishing Swine. Proceedings of the 15th International Pig Veterinary Congress, Birmingham, England, pp. 79-79.
Berends, B.R., H.A.P. Urlings, J.M.A. Snijders, F. Van Knapen, (1996): Identification and quantification of risk factors in animal management and transport regarding Salmonella spp. in pigs. Int. J. Food Microbiol. 30, 37-53.
Christensen, J. and M. Rudemo (1998): Multiple change-point analysis applied to the monitoring of Salmonella prevalence in Danish pigs and pork. Prev. Vet. Med. 36, 131-143.
Davies, P.R., J.A. Funk, M. Morrow (1999): Fecal shedding of Salmonella by a cohort of finishing pigs in North Carolina. Swine Health. Prod. 7, 231-234.
Davies, P.R., J.A. Funk, M. Morrow (2000). Fecal shedding of Salmonella by gilts before and after introduction to a swine breeding farm. Swine Health. Prod. 8, 25-29.
Davies, P.R., W.E.M. Morrow, F.T. Jones, J. Deen, P.J. Fedorka-Cray, I.T. Harris (1997): Prevalence of Salmonella in finishing swine raised in different production systems in North Carolina, USA. Epidemiol. Infect. 119, 237-244.
Funk, J.A., P.R. Davies, M.A. Nichols (2001): Longitudinal study of Salmonella enterica in two, three-site production systems. (submitted)
Funk, J.A., P.R. Davies, M.A. Nichols (2000): The effect of fecal sample weight on detection of Salmonella enterica in swine feces. J. Vet. Diagn. Invest. 12, 412-418.
Harris, D.L. (1988): Alternative approaches to eliminating endemic diseases and improving performance of pigs. Vet. Rec. 123, 422-423.
Henken, A.M., K. Frankena, J.O. Goelema, E.A.M. Graat, J.P.T.M. Noordhuizen (1992): Multivariate epidemiological approach to salmonellosis in broiler breeder flocks. Poultry Sci. 71, 838-843.
Hosmer, D.W. and S. Lemeshow (1989): Applied Logistic Regression. John Wiley & Sons, New York.
Kabagambe, E.K., S.J. Wells, L.P. Garber, M.D. Salman, B. Wagner, P.J. Fedorka-Cray (2000): Risk factors for fecal shedding of Salmonella in 91 US dairy herds in 1996. Prev. Vet. Med. 43, 177-194.
Linton, A.H., T.W. Heard, J.J. Grimshaw, P. Pollard (1970): Computer-based analysis of epidemiological data arising from salmonellosis in pigs. Res. Vet. Sci. 2, 523-532.
Lo Fo Wong, D.M.A., 2001. Epidemiology and control options of Salmonella in European pig herds, PhD Thesis, Royal Veterinary and Agricultural University, Copenhagen, Denmark.
Morrow, W.E.M., P.R. Davies, T. See, J. Eisemann, K. Zering, S. Kihlstrom, K. Karli (1999): Prevalence of Salmonella spp. in the feces on farm and ceca at slaughter for a cohort of finishing pigs. Proceedings of the Third International Symposium on the Epidemiology and Control of Salmonella in Pork, Washington, DC, 155-157.
Randolph, J.H., G.L. Cromwell, T.S. Stahly, D.D. Kratz (1981): Effects of group size and space allowance on performance and behavior of swine. J. Anim. Sci. 53, 922-927.
Renwick, S.A., R.J. Irwin, R.C. Clarke, W.B. McNab, C. Poppe, S.A. McEwen (1992): Epidemiological associations between characteristics of registered broiler chicken flocks in Canada and the Salmonella culture status of floor litter and drinking water. Can. Vet. J. 33, 449-458.
Rose, N. F. Beaudeau, P. Drouin, J.Y. Toux, V. Rose, P. Colin (1999): Risk factors for Salmonella enterica subsp. enterica contamination in French broiler-chicken flocks at the end of the rearing period. Prev. Vet. Med. 39, 265-277.
Table 2. Final Model 1. Cohorts with prevalence > 18.75% compared to cohorts ≤ 18.75%.
Explanatory variable | Value range (median) | Coding | Logistic Regression Final Model Estimates | ||
---|---|---|---|---|---|
Coefficient (std. error) | ORa | 95% CIb (p-value) | |||
Toilet present | NA | 0 = no 1= yes |
-2.37 (1.07) | 0.09 | 0.01-0.77 (0.03) |
Finisher feed conversion | 2.41-3.04 (2.67) | 0 = ≤ 2.67 1 = > 2.67 |
2.60 (1.23) | 13.48 | 1.22-149.46 (0.03) |
Pig density at samplingc | 0.58-2.86 (0.75) | 0 = ≤ 0.75 1 = > 0.75 |
-2.88 (1.20) | 0.06 | .001-0.56 (0.02) |
Season | NA | 0 = winter/spring 1 = summer/fall |
-3.13 (1.52) | 0.04 | 0.002-0.87 (0.04) |
High temperature variance | 23-174 (125) |
0 = ≤ 175 1 = > 175 |
2.08 (1.09) | 8.03 | 0.94-68.34 (0.02) |
Table 3. Final Model 2. Cohorts with prevalence > 10% compared to cohorts ≤ 10%.
Explanatory variable | Value range (median) | Coding | Logistic Regression Final Model Estimates | ||
---|---|---|---|---|---|
Coefficient (std. error) |
ORa | 95% CIb (p-value) | |||
Number of humans at finisher daily | 1-3 (2) |
0 = < 3 1= 3 |
1.57 (0.77) | 4.81 | 1.06-21.77 (0.04) |
Other domestic animal species | NA | 0 = no 1 = yes |
1.55 (0.81) | 4.73 | 0.97-23.20 (0.06) |
Pig density at samplingc | 0.58-2.86, (0.75) |
0 = ≤ 0.75 1 = > 0.75 |
-1.51 (0.72) | 0.22 | 0.05-0.90 (0.04) |
High temperature(C) day of sampling | 1.1-36.1 (21.7) |
0 = ≤ 21.7 1 = > 21.7 |
-1.43 (0.69) | 0.24 | 0.06-0.93 (0.04) |
aOdds Ratio
bConfidence Interval
cm2/pig
dSample variance [n? x2-(? x)2]/[n(n-1)], x = daily high temperature
Dr. Julie Funk
Dr. Funk is originally from Michigan and received her Bachelors of Science and Doctor of Veterinary Medicine degrees from Michigan State University. She was an associate in a predominantly swine veterinary practice in Wolcott, Indiana, before returning to academia and completing a Residency in Production Medicine and receiving a Master’s of Science degree in Veterinary Sciences at The University of Illinois, Urbana-Champaign. She then attended North Carolina State University and completed a Doctor of Philosophy degree in Comparative Biomedical Sciences with a concentration on Production Medicine. She is currently an Assistant Professor of Epidemiology and Food Safety at the College of Veterinary Medicine at The Ohio State University.