The effects of dietary additives, vaccinations and processing aids as control measures for Salmonella spp. in chicken meat: A systematic review and meta-analysis

Salmonellosis is a leading cause of foodborne disease and is often associated with contaminated chicken meat. Research to reduce the risk of Salmonella contamination of chicken products has led to the development of various through-chain control measures. This review aimed to statistically investigate and evaluate the effects of dietary additives, vaccines, and processing aids on Salmonella contamination in chickens. Eligibility criteria was set to include randomised controlled trials (published 2009-2019 in English) that treated meat chickens or products with a dietary additive, vaccine or processing aid and measured Salmonella in the ceca or product within 24 hours. The confidence in cumulative evidence was evaluated based on the GRADE method. Meta-analyses were performed for the effects of each intervention on Salmonella using standardised mean differences (SMD). Additional sub-categories were also meta-analysed and included in a random effects regression model to investigate interactions between intervention effects and subgroups. Many eligible studies included multi-arm interventions, where a total of 20 dietary additive (from 13 studies), 12 vaccination (from 9 studies) and 70 processing aid (from 27 studies) outcomes were synthesised. Most of the outcomes were judged as having an unclear risk of bias, mostly due to no mention of treatment allocation concealment or blinding. The random effects meta-analysis estimated that the average effect of dietary additives (SMD = -1.99; 95% CI: -2.66 to -1.33; P value < 0.001), vaccinations (SMD = -1.21; 95% CI: -1.74 to -0.69; P value < 0.001) and processing aids (SMD = -1.27; 95% CI: -1.57 to -0.97; P value <0.001) significantly reduced Salmonella. The Chi² test for heterogeneity (P value < 0.001 for all three analyses) and the I² statistic (72.5%, 82.8% and 85.8%, respectively) indicated high levels of heterogeneity across the intervention effects. The certainty of the gathered evidence was very low, mainly due to consistently unclear or high risk of bias assessments, high levels of heterogeneity in treatment effects that could not be explained by subgroup analyses, unit of analysis errors and publication bias identified in funnel plots. Although the meta-analyses found large reductions in Salmonella, it is likely that the true effects were smaller due to limitations present. Further research and transparency in methods is required to identify causes of heterogeneity and provide reliable recommendations to the industry.