Comparison of the Sensitivity of Sterility Tests Based on the Analysis of 2x2 Contingency Tables

Abstract

The work is devoted to statistical methods for determining the comparative sensitivity of sterility analyzers with a binary scale: 1/0, +/-. The paper considers both traditional statistical test methods namely the direct binomial test, the McNemar test, and the new, highly sensitive Yefimov Method, and Pearson-Yefimov statistical test. First, the comparative effectiveness of statistical methods for assessing the sensitivity of testers is demonstrated on a specially developed model system (the Test Bench), and then on real experimental data. The data were obtained during the validation of a new tester and a new method for analyzing samples for sterility. The new method is compared with the old one, the certified method. The sensitivity of both new and old methods (testers) was already determined earlier by a direct non-statistical method, which allowed us to compare the results of the two approaches. The factors influencing the efficiency of statistical tests are revealed and described.