Well, unless you’re going to test lots and lots of samples that is. Simply put, it’s unlikely that enough samples could be tested to have any reasonable confidence that the lot is safe. For example, you’d need to test 299 samples to be 95% confident of detecting contamination in 1% of the units in a lot. Or, to put it another way, testing 10 samples from a lot gives a 90% probability of failing to detect 1% contamination. So, how confident would you be when n=5? Such an approach should be used for due diligence purposes and to confirm that your HACCP plan is working (verification).
Marcel Zwietering and friends provide some examples and the relevant highlights from their paper were:
- Finished product testing is often not very effective for controlling food safety.
- Food safety management systems and preventive approaches are more effective and can reduce finished product testing.
- Finished product testing may be used for process and product verification.
Marcel H. Zwietering, Liesbeth Jacxsens, Jeanne-Marie Membré, Maarten Nauta, Mats Peterz, Relevance of microbial finished product testing in food safety management, Food Control, 60, (2016), Pages 31-43, http://doi.org/10.1016/j.foodcont.2015.07.002.
The subject is given extensive treatment by FAO/WHO .