Estimating the frequency of high microbial counts in commercial food products using various distribution functions

Industrial microbial count records usually form an irregular fluctuating ti me series. If the series is truly random or weakly autocorrelated, the fluc tuations can be considered as the outcome of the interplay of numerous fact ors that promote or inhibit growth. These factors usually balance each othe r, although not perfectly, hence, the random fluctuations. If conditions ar e unchanged, then at least in principle the probability that they will prod uce a coherent effect, i.e., an unusually high (or low) count of a given ma gnitude, can be calculated from the count distribution. This theory was tes ted with miscellaneous industrial records (e.g., standard plate count, coli forms, yeasts) of various food products, including a dairy-based snack, fro zen foods, and raw milk, using the normal, log normal, Laplace, log Laplace , Weibull, extreme value, beta, and log beta distribution functions. Compar ing predicted frequencies of counts exceeding selected levels with those ac tually observed in fresh data assessed their efficacy. No single distributi on was found to be inherently or consistently superior. It is, therefore, s uggested that, when the probability of an excessive count is estimated, sev eral distribution functions be used simultaneously and a conservative value be used as the measure of the risk.