Chapter 5 - Statistical aspects of sampling for microbiological analysis

Abstract

Sampling of product quality and safety implies a need to draw random but representative samples of product. Internationally approved schemes are described, together with definitions of stratified and other approaches, but most are unsuitable for microbiological examination of foods. The principles of attribute sampling are discussed in relation to the differences between attributes and variables sampling including concepts such as acceptance and rejectable quality levels (AQL and RQL) and use of operating characteristics curves. Unless large numbers of samples are tested, criteria lack statistical precision and schemes for detection of pathogens in food are unlikely to reject unsatisfactory product unless contamination is very high. Similarly, sampling plans based on colony count numbers waste much valuable information if the results are considered purely as an index of quality. Alternative approaches use quantitative data more efficiently through implementation of variables sampling plans.