동등성 시험을 신뢰구간을 사용하여 검정하는 경우 왜 신뢰도 90%인 신뢰구간을 사용하는가? = Why is 90% Confidence Interval Used When Confidence Interval Approach is Used for Testing Equivalence?

다국어 초록 (Multilingual Abstract)

It is a convention to use 5% significance level when a statistical test is employed for clinical data. But when a confidence interval is used for testing equivalence, 90% confidence interval has often been used. When $1-{\alpha}$ confidence interval is used for hypothesis testing, its significance level is often ${\alpha}$. So it makes a confusion that the significance level is 10% if 90% confidence interval is employed for testing equivalence. In this paper I will clarify this issue by reviewing relevant papers and conducting simulation studies. I hope that it will be beneficial to statisticians in pharmaceutical companies, CROs, university hospitals.

참고문헌 (Reference)

1 권광옥, "생물학적 동등성의 이해" 신일상사 2006

2 식품의약품안정청, "고시 제 2005-31호 생물학적 동등성시험기준"

3 Lehmann, E.L., "Testing Statistical Hypotheses" John Wiley & Sons 1986

4 Westlake, W.J., "Symmetric con¯dence intervals for bioequivalence trials" 32 : 741-744, 1976

5 Chow, S.C., "Statistics in Drug Research: Methodologies and Recent Developments" Marcel Dekker 2002

6 ICH, "Statistical principles for clinical trials" E9-, 1998

7 FDA, "Guidance on statistical procedures for bioequivalence studies using a standard two-treatment crossover design, O±ce of Generic Drugs, Center for Drug Evaluation and Research" Food and Drug Administration 1992

8 Chow, S.C., "Design and Analysis of Bioevailability and Bioequivalence Studies" Marcel Dekker 2000

9 Liu, H.K., "Con¯dence intervals in bioequivalence assessment" 51-54, 1990

10 Hsu, J.C., "Constrained simultaneous con¯dence intervals for multiple comparisons with the best" 12 : 1136-1144, 1984

11 Berger, R.L., "Bioequivalence trials, intersection union-tests and equivalence con¯dence sets (with discussion)" 11 : 283-319, 1996