임상적 동등성 평가를 위한 임상시험에서 동등성범위의 설정과 가설검정

Equivalence Margin and Hypothesis Testing in the Equivalence Trial

Interest in equivalence trials has been increasing for many years because ethical requirements preclude the use of a placebo. Equivalence trials aim to show that two treatments have equivalent therapeutic effects. The design of a clinical trial to test the equivalence of two treatments differs from that of testing efficacy. The conventional null hypothesis of no difference between two treatments is not appropriate for equivalence trials. So, we must use non-zero (inequivalence) null hypothesis to test clinical equivalence between two treatments. It's very important but difficult to specify a threshold value $({\delta})$ which is clinically insignificant in advance for defining equivalence. Statistical analysis should be based on confidence intervals. When we have predefined a range of equivalence, we can check whether the confidence interval obtained on the observed difference lies entirely within the range. If it does, equivalence is demonstrated. In this paper, the several issues of equivalence clinical trials are discussed: choice of equivalence margin, testing statistical hypothesis, and estimating confidence intervals.