Assessment of the efficacy and quality of evidence for five on-farm interventions for Salmonella reduction in grow-finish swine: A systematic review and meta-analysis

Five on-farm practices for reduction of Salmonella shedding or sero-prevalence in grow-finish swine were selected through scoping study and expert consultation. Specific examples were selected based on supporting evidence from at least one controlled trial (CT), and availability to Canadian swine producers. Efficacy was evaluated using systematic review and meta-analysis (SR-MA) methodology. A modified Grading of Recommendations, Assessment, and Evaluation (GRADE) approach was applied to assess the quality of evidence for each intervention, and a ‘summary of findings’ table was developed to present findings ‘at-a-glance’. MA of the small dataset of CTs investigating feeding meal, and measuring serology, yielded a significant summary estimate of efficacy (odds ratio (OR)=0.21; 95% confidence intervals (CI): 0.14, 0.31) with non-significant heterogeneity (P>0.10). MA of the dataset investigating inclusion of organic acids in the ration, measuring serology, yielded a significant summary estimate with significant heterogeneity across studies (P<0.001, I²=91%) precluding presentation of a single summary estimate; a range of results were reported (OR Range: 28 (1.6, 498); 0.07 (CI: 0.042, 0.33)). Pen disinfection between batches of finishers was studied in one large CT measuring both fecal culture (OR 0.84 (0.68, 1.1)) and serology (OR 0.48 (0.40, 0.58)) outcomes. The dataset investigating Salmonella spp. vaccination contained inconsistent findings (OR Range: 4.5 (1.3, 15); 0.07 (0.008, 0.68)), with significant heterogeneity across studies (P=0.005, I²=82), assessed measuring fecal culture. MA of the dataset investigating inclusion of in-feed tetracyclines yielded a significant OR indicating a potential harmful effect, measuring fecal culture, (OR Range: 14 (1.9, 108); 1.0 (0.43, 2.5)) with significant heterogeneity (P=0.003, I²=82%) across studies, suggesting some potential for withdrawal of in-feed tetracyclines to reduce Salmonella shedding. Therefore our ranking of intervention efficacy is: feeding meal>inclusion of acids in ration, feeder pen disinfection or Salmonella spp. vaccination>in-feed tetrayclines. Study design characteristics increasing risk of bias, including failure to justify sample size (19 of 31 studies) and failure to report random or systematic sampling (13 of 31studies), resulted in modified GRADE evidence rankings of ‘low’ for these interventions. This suggests that further research is likely to affect our findings. Field CTs investigating herd-level interventions with measurements at the herd- and individual-levels are recommended. Overall, SR-MA was a useful approach for ranking efficacy, and GRADE offered a transparent method for ranking quality of evidence, although both were limited by the small number of comparable studies available.