Impact of the prevalence of different pathogens on the performance of sampling plans in lettuce products.

Prevalence and concentration of Listeria monocytogenes, Salmonella spp. and enteric pathogenic viruses (namely Hepatitis A-HAV, and noroviruses genogroup I-NoVGI and genogroup II-NoVGII) were determined in raw and RTE lettuce from a Spanish processing premise. Fifteen samplings were made from September 2010 to February 2012 (n=600 samples). Sampling strategies for pathogen detection were suggested by the characterization of the uncertainty in prevalence associated with the performance of two-class attributes sampling plans (c=0). A probabilistic model was run (1000 iterations) using a Bayesian approach with a conjugate beta distribution considering the impact of taking different number of samples on the proportion of positive samples and lots (within- and between-lot prevalence). No enumeration results were obtained for the pathogens tested. Presence of L. monocytogenes and NoVGII in RTE lettuce (10%) and NoVGI and NoVGII in unprocessed lettuce (10%) was obtained in the tested lots during cold season. Results evidenced that, as the number of samples increased, the probability of rejecting a contaminated lot became higher, yielding right-skewed distributions with values close to 1. According to our results, 25 samples would result in 80% of rejected lots, while 95% confidence level would be reached with n>100. However, although those levels would imply a unrealistic high number of samples making the application of the sampling plan unfeasible, these results might be useful for food operators and risk managers to know the underlying distributions of microbial contamination together with potential control measures to be applied to assure a safer production of minimally processed vegetables.