Modelling homogeneous and heterogeneous microbial contaminations in a powdered food product.

The actual physical distribution of microorganisms within a batch of food influences quantification of microorganisms in the batch, resulting from sampling and enumeration by microbiological tests. Quantification may be most accurate for batches in which microorganisms are distributed homogeneously. However, when the distribution is non-homogeneous, quantification may result in an under-, or overestimation. In the case of pathogens being non-homogeneously distributed, this heterogeneity will impact on public health. Enumeration data are commonly modelled by the Lognormal distribution. Although the Lognormal distribution can model heterogeneity, it does not allow for complete absence of microorganisms. Studies that validate the appropriateness of using Lognormal or other statistical distributions are scarce. This study systematically investigated laboratory and industrial scale batches of powdered infant formula, modelled the enumeration data using a range of statistical distributions, and assessed the appropriateness of individual models. For laboratory scale experiments, batches of milk powder were contaminated by distributing similar numbers of cells of Cronobacter sakazakii either homogeneously throughout a batch of milk powder or by distributing the cells in a localised part of the batch. Each batch was then systematically sampled and the distribution determined by enumerating the samples. By also enumerating the remainder of the batch, a balance could be made of the total number of microorganisms added and of the number retrieved from a batch. Discrete, as well as continuous statistical distributions, were fitted to enumeration data and the parameters estimated by Maximum Likelihood. The data were fitted both as censored and uncensored data. Enumeration data obtained for an industrial batch of powdered infant formula were investigated in this way as well. It was found that Normal, Poisson and Zero-Inflated Poisson distributions fitted the data sets very poorly. In case of homogeneous contamination, there was not a notable difference between the ability of Negative Binomial, Poisson-Lognormal, Weibull, Gamma, and Lognormal distributions to model the data. Overall, either the Negative Binomial distribution or the Poisson-Lognormal distribution fitted the data best in the 10 batches studied, especially when part of a data set contained zeros and/or the numbers were low. The Negative Binomial fitted the laboratory batches best and the Poisson-Lognormal fitted the industrial batch best.

Actual distribution of Cronobacter spp. in industrial batches of powdered infant formula and consequences for performance of sampling strategies.

The actual spatial distribution of microorganisms within a batch of food influences the results of sampling for microbiological testing when this distribution is non-homogeneous. In the case of pathogens being non-homogeneously distributed, it markedly influences public health risk. This study investigated the spatial distribution of Cronobacter spp. in powdered infant formula (PIF) on industrial batch-scale for both a recalled batch as well a reference batch. Additionally, local spatial occurrence of clusters of Cronobacter cells was assessed, as well as the performance of typical sampling strategies to determine the presence of the microorganisms. The concentration of Cronobacter spp. was assessed in the course of the filling time of each batch, by taking samples of 333 g using the most probable number (MPN) enrichment technique. The occurrence of clusters of Cronobacter spp. cells was investigated by plate counting. From the recalled batch, 415 MPN samples were drawn. The expected heterogeneous distribution of Cronobacter spp. could be quantified from these samples, which showed no detectable level (detection limit of -2.52 log CFU/g) in 58% of samples, whilst in the remainder concentrations were found to be between -2.52 and 2.75 log CFU/g. The estimated average concentration in the recalled batch was -2.78 log CFU/g and a standard deviation of 1.10 log CFU/g. The estimated average concentration in the reference batch was -4.41 log CFU/g, with 99% of the 93 samples being below the detection limit. In the recalled batch, clusters of cells occurred sporadically in 8 out of 2290 samples of 1g taken. The two largest clusters contained 123 (2.09 log CFU/g) and 560 (2.75 log CFU/g) cells. Various sampling strategies were evaluated for the recalled batch. Taking more and smaller samples and keeping the total sampling weight constant, considerably improved the performance of the sampling plans to detect such a type of contaminated batch. Compared to random sampling, stratified random sampling improved the probability to detect the heterogeneous contamination.

Factors influencing the accuracy of the plating method used to enumerate low numbers of viable micro-organisms in food.

This study aims to assess several factors that influence the accuracy of the plate count technique to estimate low numbers of micro-organisms in liquid and solid food. Concentrations around 10CFU/mL or 100CFU/g in the original sample, which can still be enumerated with the plate count technique, are considered as low numbers. The impact of low plate counts, technical errors, heterogeneity of contamination and singular versus duplicate plating were studied. Batches of liquid and powdered milk were artificially contaminated with various amounts of Cronobacter sakazakii strain ATCC 29544 to create batches with accurately known levels of contamination. After thoroughly mixing, these batches were extensively sampled and plated in duplicate. The coefficient of variation (CV) was calculated for samples from both batches of liquid and powdered product as a measure of the dispersion within the samples. The impact of technical errors and low plate counts were determined theoretically, experimentally, as well as with Monte Carlo simulations. CV-values for samples of liquid milk batches were found to be similar to their theoretical CV-values established by assuming Poisson distribution of the plate counts. However, CV-values of samples of powdered milk batches were approximately five times higher than their theoretical CV-values. In particular, powdered milk samples with low numbers of Cronobacter spp. showed much more dispersion than expected which was likely due to heterogeneity. The impact of technical errors was found to be less prominent than that of low plate counts or of heterogeneity. Considering the impact of low plate counts on accuracy, it would be advisable to keep to a lower limit for plate counts of 25 colonies/plate rather than to the currently advocated 10 colonies/plate. For a powdered product with a heterogeneous contamination, it is more accurate to use 10 plates for 10 individual samples than to use the same 10 plates for 5 samples plated in duplicate.

Optimization of a monitoring network for sulfur dioxide.

In this study, we develop and apply a methodology to reduce an existing monitoring network to find an optimal configuration of a smaller network. We use a criterion based on locally weighted regression with two different weight functions. The methodology is applied to the Dutch national SO2 network and offers the possibility to include different politically relevant options in the model by weight criteria. Because full enumeration of all monitoring networks is impossible, a combinatorial search algorithm is applied to find a (sub)optimal solution. If different (optimal) monitoring networks result from different criteria, then the best can be selected.