On modeling the irregular fluctuations in microbial counts.

Abstract

Daily or other periodic microbial counts in many foods, particularly ground meats, poultry, and raw milk, show an irregular fluctuating pattern. The cause of the fluctuations is the interplay of many random factors that tend to promote or inhibit microbial growth. Therefore, the actual size of a microbial population can vary randomly around a typical level or around a trend determined, for example, by seasonal variations or changing sanitary conditions. Fluctuations around a fixed level can often be modeled as a sequence of independent counts having a lognormal or other parametric distribution. The independence of the counts and the type of distribution can be established by standard statistical tests. Once selected, the distribution function can be used to estimate the probability of encountering a population in any given size range. The model can be modified to describe sequences that include zero counts, which are the result of the organisms' absence, or of the failure of the method to detect them. In the case of fluctuations around a trend, a modified version of the model can be used to estimate the probability of deviations from the trend. A more thorough modification is required in order to account for fluctuating patterns that include outbursts of appreciable duration, like those caused by massive contamination of a water reservoir.