Factors influencing the prevalence of Salmonella spp. in swine farms: a meta-analysis approach.

A systematic review and meta-analysis was conducted to identify study-level variables that could explain the variation in apparent Salmonella spp. prevalence estimates. Electronic and non-electronic literature searches from 1990 until 2005 were carried out to identify all studies related to the prevalence of subclinical Salmonella infection in swine. The searches were restricted to studies published in English, Spanish, and French. Clinical trials or any other study where an intervention was evaluated were excluded from this analysis. A template was designed to retrieve the most relevant variables and data abstraction was performed in duplicate. A total of 98 papers containing 82 animal-level and 156 farm-level studies were used in the analyses. The median farm-level and animal-level prevalences were 59% and 17%, respectively. Meta-regression analyses were carried out on both farm and animal-level data. Diagnostic procedure, sample size, and country where study was conducted were the three most important predictors in explaining the differences in Salmonella prevalences between studies. When compared to a farm with a apparent prevalence of 50% determined by the blood ELISA, prevalences based on culture of fecal samples were 39% lower and prevalences based on cecum and tissue cultures were 16% and 19% lower, respectively. Similar to farm-level models, animal-level models did not show any difference among serological tests and prevalence values based culture procedures were, on average, 9% lower than those from serological tests. Sample size was negatively associated with prevalence estimates. In conclusion, the methodology was useful for identifying and quantifying sources of variation in Salmonella apparent prevalence among studies and for establishing prevalence distributions that could be used as input parameters in risk assessment and decision models. The analysis provides some guidelines when interpreting and comparing apparent Salmonella prevalence results from studies using different study designs.