Characterisation of Campylobacter contamination in broilers and assessment of microbiological criteria for the pathogen in broiler slaughterhouses

Campylobacter is considered an important foodborne pathogen, having a high impact on public health burden. Source allocation studies identified broiler meat as the most important food transmission vehicle of Campylobacter. In this regard, microbiological limits are under development in the EU and characterisation of microbial contamination seems to be an essential step towards implementing sampling schemes. The aim of this study was to analyse broiler batches processed at three conventional slaughterhouses in Germany for their Campylobacter load at the end of processing. Microbial data of positive and negative batches from the studied slaughterhouses were used to fit different statistical distributions (i.e., lognormal, Poisson-lognormal and negative binomial) by means of the maximum likelihood estimation method. Finally, the performance of microbiological criteria was assessed by calculating the probability of accepting batches using two microbiological limits recently discussed by the European Food Safety Authority (500 and 1000 cfu/g). Samples of broiler neck skin and caecal content (356 batches) were collected in three German slaughterhouses (A, B and C) for fresh meat production from conventionally reared broilers during the period from July 2013 to June 2016. Microbial analyses of five pooled neck skin samples and one pooled caecal sample per batch were performed. In this study, the prevalence of Campylobacter caecum positive broiler batches processed at slaughterhouses A, B and C was 42.7 (n = 117), 21.7 (n = 120) and 47.9% (n = 119), respectively. The Campylobacter mean count in neck skin samples of positive batches was highest in slaughterhouse A, with 2.9 ± 0.6 log10 cfu/g compared to slaughterhouse B with 2.7 ± 0.7 log10 cfu/g and slaughterhouse C with 2.7 ± 0.7 log10 cfu/g. The results showed that distributions skewed to the right, thus indicating a high proportion of low microbial counts in the samples. Negative binomial regression provided a better fit at low contamination levels (<100 cfu/g), while the Poisson-lognormal distribution described both within- and between-batch variability better at higher microbial counts. The fitted statistical distributions were further used to evaluate the performance of different microbiological criteria based on an acceptance sampling approach in relation to the defective units based on different limits (mean counts) set at each slaughterhouse. Results obtained will help food business operators to evaluate the status of microbiological hygiene and safety in relation to Campylobacter contamination, also assisting them in setting their own limits of acceptance for process improvement under consideration of existing legal requirements.