FDA's preferred MPN methods for standard, large or unusual tests, with a spreadsheet

The US Food and Drug Administration (FDA) presents its preferred methods for statistical interpretation of serial dilution tests for microbes in the Bacteriological Analytical Manual (BAM) of the FDA and the AOAC. This article explains why the particular methods in the manual were selected and describes the adjustments chosen to create a new, flexible method for calculating most probable numbers (MPNs), confidence intervals, and measures of improbability for up to 18 dilutions and up to 1020 tubes per dilution. The methods have been put into a spreadsheet for easy use, and are offered free in the interest of promoting the microbial safety of food and water. The FDA prefers the MPN, not adjusted for bias, for its point estimate of concentration, with the improbability standards and confidence intervals of De Man for standard three-dilution tables with up to 10 tubes per dilution. For large or unusual test designs, the FDA also prefers the MPN but with a new improbability index and the confidence intervals of Haldane. Some other methods in recent literature are discussed briefly, and reasons are given why they are not preferred. Published by Elsevier Science Ltd.