Quantitative microbiological risk assessment using individual data on food storage and consumption (Part 2): A comparison with traditional QMRA approaches

In a previous study, we integrated data from individual consumers collected in a dietary survey in France in a multi-food quantitative microbiological risk assessment (QMRA) for listeriosis. Here, we compared the “individual-based” modelling approach applied in that study with several other approaches where the data are treated as in more “traditional” QMRA methods, for example by assuming independent randomly sampled variables from distributions fitted through the data, instead of the observed individual data themselves. We found that assigning randomly sampled storage times instead of the reported individual storage times resulted in a higher risk estimate than the baseline, expressed as expected annual number of cases in the population. Assigning randomly sampled storage temperature and point estimates for portion size and frequency of consumption, slightly increased the estimated risk. Statistical analysis did not show dependency between portion size, frequency of consumption, storage temperature and storage time in the data set, which can be explained by the fact that only a few individuals had a large impact on the final population risk. Analysis of expected numbers of cases per age class, sex and food group showed small differences between approaches. Our analysis was challenged by the difference between a model structure where the risk is calculated per individual (when based on a dietary survey with individual data) and one where it is calculated per serving, as in “traditional” QMRA. We showed that an “individual-based” QMRA is more resource-demanding but can give fundamentally different risk estimates, which are potentially more accurate. The application of tools for efficient knowledge exchange and integration is needed to facilitate the usage of this type of QMRA.