Computer Simulation of Microbiological Sampling of Liquid Foods

Abstract

A computer program that simulates the process of randomly selecting sample packages from a large lot and examining a portion of the package contents for microorganisms was developed. The program can be used to evaluate the theoretical effects on results (mean, standard deviation, and detection or nondetection) of various choices for the number of sample packages selected, the sample volume withdrawn for examination, the package size, and the behavior of the organisms at different microorganism concentrations. The results can be used to compare the theoretical effectiveness of and the risk inherent in various sampling schemes. The simulation program was used to systematically study the effects of varying the number of sample packages selected, the distribution of samples between packages, and the sample volume examined. The results indicate, for example, that finding no counts when 5-ml volumes are drawn from each of 16 sample packages is no assurance of sterility, but only indicates about a 50% chance that the average microbial concentration is less than 1 cell per 100 ml. Increasing the number of packages examined from 16 to 32 should decrease the likelihood of finding a nonzero result to approximately 25%. For 16 samples of 100 ml and the same organism concentration, however, a zero result at this organism loading is very unlikely. The total volume examined is the most important factor for most practical sampling situations. This finding implies that taking fewer samples of larger volume is the more cost effective strategy to reduce risk.