The effect of reducing numbers of Campylobacter in broiler intestines on human health risk

• Two approaches for translating numbers of Campylobacter in broiler intestinal content into human health risk are comparted.
• One approach uses linear regression of concentration data from caeca and skins, the other a risk assessment model
• No generally valid simple relation exists between intestinal concentrations and concentrations on meat.
• A one to two log reduction in intestinal Campylobacter concentration has a large effect on the human health risk.
• The differences in effect of reduction of concentrations on risk between different European countries are small.

One option for Campylobacter control in broiler chickens is to reduce the concentration in the intestinal content of the birds prior to slaughter, for example by vaccination or phage therapy. It is however unsure how such a reduction in concentration can be translated into a reduction in concentration on the meat and a reduction in the human health risk of campylobacteriosis. In this study, two methods are presented and compared. The first is a linear regression model, based on count data from caecal contents and skin sample data, obtained after processing from the same flocks. Alternatively, a previously published risk assessment model is used, that describes the dynamics of transfer and survival of Campylobacter during broiler processing at the slaughterhouse. Data from five European countries are used as inputs for the models. For both approaches the analyses show that a one to two log reduction in concentration in the intestinal content has a large impact on the risk of campylobacteriosis due to the consumption of chicken meat: a relative risk reduction between 44% and 95%. Therefore it seems promising to aim interventions at a reduction of the concentration of Campylobacter in the broiler intestines. However, it is not possible to derive a generic rule that can be used to relate a reduction in concentration in broiler intestines into a reduction in human health risk. Regression models based on different data sets predict different relationships between bacterial count data from caeca and skins, whereas the risk assessment model requires data on contamination of the birds’ exterior, which is not sufficiently available in combination with caecal concentration data. Simulations performed with the risk assessment model show that it can pretty well adequately describe the observed correlation in the data and the variation in regression lines obtained.