Farm management practices that affect the prevalence of Salmonella in pastured poultry farms

Farm practices can affect the prevalence of Salmonella in the final product when poultry are exposed to the outside environment. Pastured poultry farms in the Southeastern United States were investigated in this study. Farm practice and processing variables that may affect the presence of Salmonella were determined by developing predictive models using the random forest method. Important variables affecting the prevalence of Salmonella in preharvest (feces and soil), and postharvest (whole carcass rinses) samples were identified. Predictive models were generated with each type of sample, and the models were tested with the corresponding test set. The model performances were measured by the area under curve (AUC) values from the receiver operating characteristic (ROC) curve. All models developed in this study were robust in predicting Salmonella presence, with AUC values above 0.83. It was found that as the number of years of operation increased, there was increase in predicted probability of finding Salmonella. The first three ingredients in the brood and pasture feed were identified as the top predicting variables for both preharvest and postharvest variables. These models and data can help to inform pastured poultry producers about management practices that can reduce Salmonella prevalence within their production systems.