Building a predictive model for assessing the risk of Salmonella shedding at slaughter in fattening pigs

Abstract

Salmonellosis continues to be a major cause of foodborne outbreaks worldwide, and pigs are one of the main sources of human infection. Salmonella pork contamination is a major concern for abattoirs and is related to the presence of Salmonella in pigs' feces at slaughter. Being able to predict the risk of Salmonella shedding in pigs arriving at the slaughterhouse could help mitigate abattoir and carcass contamination. For this purpose, 30 batches of 50 pigs each were selected from 30 different fattening units. The pigs were tagged and bled for the detection of antibodies against Salmonella approximately one month before slaughter. Pooled floor fecal samples were also collected from 10 pens per unit for Salmonella detection, and a questionnaire on biosecurity was administered to each farm. At the abattoir, colon content was collected from each tagged pig for the Salmonella shedding assessment. A predictive model for Salmonella shedding at slaughter was built with two-third of the pigs by employing random-effects logistic regression analysis, with Salmonella shedding as the dependent variable and pig serology and other farm/environmental characteristics as the independent variables. The model included farm as the grouping factor. Data from the remaining one-third of the pigs were used for model validation. Out of 1,500 pigs initially selected, 1,341 were identified at the abattoir and analyzed. Salmonella was detected in 13 (43.3%; 95%CI = 27.4-60.8) of the fattening units. The mean batch seroprevalence (cut-off OD% ≥40) among the fattening units was 31.7% (95%CI = 21.8-41.0), and a total of 316 pigs (23.6%; 95%CI = 21.4-25.9) shed Salmonella at slaughter. The model predicted reasonably well (Area under the curve = 0.76; P < 0.05) whether a pig would shed Salmonella at slaughter, with estimates of sensitivity and specificity at 71.6% and 73.6%, respectively. Serology, the percentage of Salmonella-positive pens on the farm, and the internal biosecurity score were significantly associated (P < 0.05) with Salmonella shedding at the abattoir, and several scenarios were observed by the model. The study highlighted that although serology may be helpful for identifying batches of pigs at risk of shedding Salmonella upon their arrival at the abattoir, it may not be necessary in some scenarios.

Keywords: abattoir; prediction model; salmonella control; shedding; swine.