Interventions to reduce non-typhoidal Salmonella in pigs during transport to slaughter and lairage: Systematic review, meta-analysis, and research synthesis based infection models in support of assessment of effectiveness.

A systematic review of the effectiveness of interventions to reduce Salmonella prevalence or concentration in pork was undertaken. A broad search was conducted in two electronic databases. Each citation was appraised using screening tools designed and tested a priori. Level 1 relevance screening excluded irrelevant citations; level 2 confirmed relevance and categorized. Data were then extracted, and intervention categories were descriptively summarized. Meta-analysis was performed to provide a summary estimate of treatment effect where two or more studies investigated the same intervention in comparable populations. The Grading of Recommendation, Assessment, Development and Evaluation (GRADE) approach was used to assess the confidence in the estimated summary measures of intervention effect for each data subgroup. Data were also extracted from the control groups of 25 challenge trials captured by the review, to fit logistic regression models of Salmonella infection in pigs, using odds of infection as the outcome measure. The only intervention captured by the review which was significantly associated with reduced risk of Salmonella in field settings, was elimination of lairage, which is not currently feasible commercially. The logistic regression model for fecal Salmonella shedding in pigs with a random intercept for trial yielded the following predictors significantly associated with increased odds of infection: oral challenge route relative to intra-nasal, log increase in challenge dose, and elapsed time post-challenge. Univariable exact logistic regression modeling lymph node contamination post-challenge yielded the following predictors significantly associated with increased odds of Salmonella infection: younger animals relative to older ones; intra-nasal challenge route relative to oral route; and animals sampled within the first 7days post-challenge relative to those sampled at 14 or 21days. We hypothesize that the presence of absence of one or more of these predictors across studies could help to explain the inconsistent and/or non-significant findings reported for some interventions applied at lairage.