1983 Volume 14 Issue 3 Pages 437-452
The usefulness of an approximation formula to calculate the power of analysis of variance for bioequivalence tests in a two-way crossover design was examined by comparison with the power estimated from the upper probability integrals of the noncentral F-distribution. The approximation formula was shown to be useful for the calculation of the power for usual bioequivalence tests . The calculation of the power for bioequivalence tests in a multi-way crossover design was attempted using the approximation formula.
Further, data in previously published reports on bioequivalence between drug preparations were reviewed from the standpoint of the power. Only 24 of 86 bioequivalence analyses conducted with 25 different drugs gave a power higher than 80% to detect a 20% difference in bioavailability with α=0.05, and the power of about a half of the analyses was lower than 50%.
