Estimation of Microbial Contamination of Food from Prevalence and Concentration Data: Application to Listeria monocytogenes in Fresh Vegetables

Abstract

A normal distribution and a mixture model of two normal distributions in a Bayesian approach using prevalence and concentration data were used to establish the distribution of contamination of the food-borne pathogenic bacteria Listeria monocytogenes in unprocessed and minimally processed fresh vegetables. A total of 165 prevalence studies, including 15 studies with concentration data, were taken from the scientific literature and from technical reports and used for statistical analysis. The predicted mean of the normal distribution of the logarithms of viable L. monocytogenes per gram of fresh vegetables was ؊2.63 log viable L. monocytogenes organisms/g, and its standard deviation was 1.48 log viable L. monocytogenes organisms/g. These values were determined by considering one contaminated sample in prevalence studies in which samples are in fact negative. This deliberate overestimation is necessary to complete calculations. With the mixture model, the predicted mean of the distribution of the logarithm of viable L. monocytogenes per gram of fresh vegetables was ؊3.38 log viable L. monocytogenes organisms/g and its standard deviation was 1.46 log viable L. monocytogenes organisms/g. The probabilities of fresh unprocessed and minimally processed vegetables being contaminated with concentrations higher than 1, 2, and 3 log viable L. monocytogenes organisms/g were 1.44, 0.63, and 0.17%, respectively. Introducing a sensitivity rate of 80 or 95% in the mixture model had a small effect on the estimation of the contamination. In contrast, introducing a low sensitivity rate (40%) resulted in marked differences, especially for high percentiles. There was a significantly lower estimation of contamination in the papers and reports of 2000 to 2005 than in those of 1988 to 1999 and a lower estimation of contamination of leafy salads than that of sprouts and other vegetables. The interest of the mixture model for the estimation of microbial contamination is discussed.Quantitative microbial risk assessment (QMRA) is in rapid development in the area of food safety. To obtain a quantitative exposure assessment or a quantitative risk characterization for a given food and a given food-borne pathogen, statistical distributions of microbial concentrations are used as input values (such as bacterial concentrations in raw food materials subjected to further process) and/or as output values (predicted contamination in foods after processing and/or storage) (35,75). Microbial contaminations in foods are expressed in two forms: (i) prevalence data (percentage of positive samples in a given study, i.e., growth/no growth of a target pathogen after enrichment in an appropriate broth of an aliquot of a food sample) and (ii) concentration data expressed as CFU per gram or CFU per milliliter. Data on the concentration of pathogenic bacteria in foods, while scarce, are essential to QMRA.Previous works on QMRA did not consider low levels of concentrations, i.e., concentrations below the threshold of detection of microbiological methods (6,15,44...

Listeria innocua growth in fresh cut mixed leafy salads packaged in modified atmosphere

“…The samples analyzed in this study were not naturally contaminated with Listeria spp. when tested by both approaches, in agreement with a recent study that showed the low incidence of L. monocytogenes (À2.63 viable Log per gram) in leafy vegetable salads (Crépet, Albert, Dervin, & Carlin, 2007).…”

A comparison between the discrete Poisson-gamma and Poisson-lognormal distributions to characterise microbial counts in foods

“…and observed distributions (PG, PLN, etc.). It should be noticed that in the heterogeneous Poisson framework used in this study, the zero counts or non-detections and the positive counts arise from the same trials (production batches), unlike the meta-analytical method proposed by Crepet et al (2007) whereby the prevalence (zero counts) component and the counts component originated from a number of different studies. However, with their methodology the integration of the prevalence studies in the counts model was attained at expenses of an overestimation in bacterial concentration since a positive value (i.e., a positive sample) was imputed in the prevalence data consisting only of zero counts in order to estimate the concentration from a prevalence value.…”

“…Finally, as both measures of dispersion were estimated using data pooled from different production batches, the dispersion factor (1/k) and the standard deviation values are expected to be higher than their respective measures representing only the heterogeneity within a batch. Although the assumption of data normality of bacterial counts has been applied for long time (Crepet, Albert, Dervin, & Carlin, 2007;Kilsby, Aspinall, & Baird-Parker, 1979;Kilsby & Pugh, 1981;Legan et al, 2001), and in many cases, supported by experimental evidence (Gill, McGinnis, & Badoni, 1996;Peleg & Horowitz, 2000), some concerns have been lately raised in that it does not allow complete absence of microorganisms and that the presence of zero counts in more than 15% of the samples complicates the fitting method (Corradini et al, 2002;Gill, Deslandes, Rahn, Houde, & Bryant, 1998), which is of significance when modelling microorganisms present in low concentrations such as pathogens. Ignoring non-detects, substituting them with the limit of enumeration or limit or quantification, or half of them (a rather common practice known as 'imputation'), are typical sources of bias in the analysis, and normally tend to overestimate the mean values (Hirano et al, 1994;Hornung & Reed, 1990).…”

Prevalence and quantification of pathogenic Vibrio parahaemolyticus during shrimp culture in Thailand

“…The concentration variable was described by lognormal distribution (μ, σ). This probability is assumed to follow a Poisson distribution (Crépet et al 2007): P c = 1 -e (-NC) Eq. (2) The prevalence (P s ) of contaminated sample depends on numbers of both contaminated samples (s) and sample size (k).…”