Abstract: In many cases, measured contamination levels of Listeria monocytogenes in bulk cooked meat are characterized by zero-inflation and over-dispersion features. The effect of different left-censored data processing methods on quantitative exposure assessment was discussed in this work. The contamination level of L. monocytogenes in bulk cooked meat products was determined using 254 samples collected from supermarkets, farmers’ markets and catering links in a certain district of Shanghai from February to December 2017. As an illustration, traditional statistical distributions and their zero-inflated models were used to assess the concentration of L. monocytogenes in bulk cooked meat products. The models were compared and validated for their performance using R 2.0 software. The L. monocytogenes concentration data consisting of a considerable number of left-censored data could be appropriately represented by zero-inflated lognormal distribution or zero-inflated Poisson lognormal distribution. The parameter estimates of the zero-inflated models indicated that the prevalence of L. monocytogenes in bulk cooked meat products was 2%. This work provides a theoretical basis for the statistical analysis of left-censored L. monocytogenes concentration data and serves as the essential starting point for a complete risk assessment framework.

Key words: bulk cooked meat, Listeria monocytogenes, zero-inflated model, exposure assessment