Writing in the journal Statistical Modelling, the scientists said risk assessments related to food safety over more than one step of the production line can be hampered by a lack of detailed quantitative data.
A Bayesian hidden variable model was developed to overcome the problem of limited data and was put to work to evaluate the prevalence of microbes in the production of pork and any changes from one stage of the production chain to another.
The study used data on the occurrence three bacteria; Listeria monocytogenes, Yersinia enterocolitica and Yersinia pseudotuberculosis. Animal specific data from conventional and organic pig farms as well as abattoirs was available, along with data at meat cutting farms, but this was merely farm specific.
Accommodating gaps and variations
Results of different tests of different types from these various points in the production chain were computed into the model. The scientists said the model makes allowance for variations – in particular it allows for the fact that microbes, although present, may not be detected in differently targeted samples taken at different points of the production chain, and that some observations may also be missing.
The output of the model is obtained in the form of a probability distribution, which describes the remaining uncertainty. The magnitude of the microbial risk is measured by calculating probabilities and the results can then be used to inform decision making.
Assessing steps to reduce risks
The model can also be used to evaluate the effectiveness of steps to reduce or eradicate food safety risks or other background variables that may affect risks.
In the case of pork production, the test results and the model estimated that the prevalence of one or more of the three bacteria tracked in Finnish pork is 1 to 11 per cent. In the slaughterhouse, if the head were removed and the rectum sealed off, then the scientists said this figure could be reduced to 0 to 2 per cent.
Looking forward, the Finnish Food Safety Authority Evira said: “The output of the model can be utilised in determining the ideal monitoring point for microbes and in planning risk control action for the production chain.”
The model was developed within the framework of a research project in which the Department of Food Hygiene and Environmental Health at the University of Helsinki participated. Funding was provided by the Ministry of Agriculture and Forestry, Finland.
Source: Statistical Modelling
2010; 10(1): 69-87
Causal hidden variable model of pathogenic contamination from pig to pork
Authors: Ranta, J., Siekkinen, K.-M., Nuotio, L., Laukkanen, R., Hellström, S., Korkeala, H., Maijala, R.