Bioequivalence: An overview of statistical concepts

Abstract

Bioequivalence (BE) means the absence of a greater-than-allowable\ndifference between the systemic bioavailability of a test product and\nthat of a reference product. Studies to test the BE of drug products,\nand the statistical basis for their design, analysis and\ninterpretation, have evolved over the last two decades. A crossover\ndesign is preferred over a parallel-group design as it segregates the\ninter-subject variation (which is not product-dependent) from the\nintra-subject variation (which is product-dependent). The value of\ntesting two one-sided null hypotheses of non-equivalence at a\nsignificance level of 0.05, and the importance of estimating a 90%\nconfidence interval of the ratio (test/reference) of mean AUC and Cmax\nvalues, and of the difference between mean Tmax values, are now\nrecognized and form the current standards for BE. The number of\nsubjects required for a BE study with the desired power (at least 0.80)\nand significance level (0.05), depends on the expected deviation of the\ntest product from the reference product and the error variance\nassociated with the bioavailability parameters (AUC, Cmax, Tmax etc.)\nof the drug substance. At present, according to the Indian regulatory\nauthority, the number of subjects required to conduct a BE study is 12\nwhich is inadequate for most drug substances by the current\ninternational standards and criteria. This review expounds the\nforegoing principles of BE testing.