Development of a <em>Salmonella</em> source-attribution model for evaluating targets in the turkey meat production
A Salmonella source attribution model based on a microbial-subtyping approach was developed to estimate the public health effect of setting a new target for the reduction of Salmonella in fattening turkey flocks in the European Union. The model considers the quantitative contribution and relevance of different Salmonella serovars found in turkeys to human salmonellosis and includes 25 Member States, four animal-food sources of Salmonella (turkeys, broilers, laying hens and pigs) and 23 Salmonella serovars. This turkey-target Salmonella attribution model (TT-SAM) employs prevalence and serovar distribution data from the EU statutory monitoring and EU-wide Baseline Surveys on Salmonella in animal-food sources, data on incidence and serovar distribution of human salmonellosis, and food availability data. It is estimated that around 2.6 %, 10.6 %, 17.0 % and 56.8 % of the human salmonellosis cases are attributable to turkeys, broilers, laying hens (eggs) and pigs, respectively. Of the turkey-associated human salmonellosis cases, around 63 % is estimated to be due to serovars other than the currently regulated S. Enteritidis and S. Typhimurium. Four serovars (S. Kentucky, S. Saintpaul, S. Senftenberg and S. Kottbus) had turkeys as the most important reservoir for human infections. Different scenarios are presented showing changes in the percentage of turkey-associated human salmonellosis cases under different prevalences of Salmonella in fattening turkey flocks. Comparing the situation in 2010 with a theoretical combined prevalence of 1 % for S. Enteritidis and S. Typhimurium (i.e. the current target), the expected reduction in number of turkey-associated cases is very small. Since, all MSs except one have already met the transitional target, this result is not unexpected. However, when adjusting the combined prevalence of all serovars to 1 %, a large reduction in the percentage of turkey-associated cases compared to the situation in 2010 is achieved. Uncertainty and data limitations are discussed thoroughly and a number of recommendations are provided.