此篇研究提出新的三臂生物等效性試驗以及三臂非劣性試驗的統計設計方法，本研究在兩個試驗中所用到的皆是二元型態的臨床指標。三臂生物等效性試驗通常包含三個假設檢定，其中兩個檢定實驗藥物和參考藥物分別對於安慰劑的優越性；而另一個檢定實驗藥物和參考藥物之間的等效性。在此項研究中，我們提出只需要透過一個假設檢定就可以證明兩個藥之間的生物等效性，其中的檢定程序是透過比率差異的比值來完成。在二元型態指標的三臂非劣性試驗中，最常使用比率差異的檢定程序。此篇文章提出透過勝算比的檢定程序，並比較在最大概似估計以及有限制的最大概似估計下的樣本數表現。除此之外，我們也比較了在兩種檢定程序下的樣本數表現。;This study developed new methodology of statistical design on non-inferiority trial and bioequivalence trial. The three-arm bioequivalence trial usually consists of three hypothesis tests in practice, where two hypothesis tests demonstrate the superiority of experimental drug and reference drug against placebo, and the last one demonstrates the equivalence of experimental drug and reference drug. We derived a single set of null and alternative hypotheses for three arm equivalence trial with binary endpoints which is based on the ratio of proportion differences. For the thee-arm non-inferiority trial with binary endpoints, the test procedure via rate difference is the most common used. An alternative test procedure via odds ratio was proposed for the three-arm non-inferiority trial. We compared the sample size performance of our test procedure based on the MLE and RMLE. In addition, we compared the sample size performance of our test procedure with rate difference procedure.
