精確方法決定對等性檢定樣本數之研究

A Study on an Exact Method for the Sample Size Determination in Equivalence Test

林耘亘(Yun-Hsuan Lin)

指導教授 ： 劉仁沛

國立臺灣大學/生物資源暨農學院/農藝學研究所/碩士(2017年)

https://doi.org/10.6342/NTU201701320

摘要

在一般藥品及臨床試驗上，經常使用對等性檢定(Equivalence Test)來評估原廠藥與專利藥之間的差異是否顯著，透過雙單尾檢定(Two one-sided tests procedure, TOST)，以檢定兩種處理的對等性，而當兩個處理之間差異不為0且在對等界線內時，其檢定力函數(Power function)為非對稱型分配，故過去所提出的方法只能算出近似的樣本數，使其所計算出來的檢定力發生不足或是過多的現象，雖然過去曾提出簡單的迭代方法來改善此問題，但其檢定力函數的積分上下限包含了樣本標準差(s)，為一個隨機變數，透過給定其為常數，所得之樣本數亦為近似的結果。
本論文所提出的方法考慮了樣本標準差的分配，為一精確方法來決定樣本數，此外，討論兩個藥物動力學反應值所需的樣本數，透過正交對角化共變異矩陣，得到一個新的分配使得兩個變數之間為獨立，再應用所提出的方法於兩變數的樣本數決定，並呈現計算出來的樣本數進行比較與分析。

關鍵字

對等性檢定 ； 雙單尾檢定 ； 樣本數 ； 檢定力

並列摘要

The equivalence test is a hypothesis to confirm whether the new test product conforms to the standard reference product. It has many applications such as evaluation of GMO crop and its conventional crop, or generic drugs and their innovative drugs. The equivalence hypothesis can be decomposed into the non-inferiority (NI) and non-superiority (NS) hypothesis. The two one-sided tests (TOST) procedure is usually proposed to test the difference between two treatments. When the difference in population means between two treatments is not 0 but within the equivalence limits, the proportion of the type II error rate allocated to each of the two tails of the central t-distribution cannot be analytically determined. Hence, no close form of the exact sample size for the equivalence hypothesis is available.
Current methods provide the sample sizes by assuming the sample standard deviation as a constant. In fact, the lower and upper limits for the power calculation contains s, which is a random variable. By integrating the probability density function of s, the power can be computed. We suggest a method with consideration of type II error rates for both one-sided hypotheses to determine the sample size for the equivalence hypothesis, by treating the sample standard deviation as a random variable.
In addition, the covariance matrix of multiple responses was transformed by orthogonal diagonalization. We obtained a new distribution in which multiple responses are independent of each other. Consequently, we apply the proposed method to the determination of the sample size for multiple responses. Numerical examples illustrate the applications of the proposed method.

並列關鍵字

Equivalence ； Hypothesis ； Sample size ； Type I error ； Type II error ； Power

參考文獻

Berger, R.L. (1982) Multiparametric hypothesis testing and acceptance sampling, Technometrics, 24, 433-440.

Berger, R.L., Hsu, J.C. (1996) Bioequivalence trials, intersection union tests and equivalence confidence sets. Statistical Science, 4, 283-319.

Chow, S.C., Wang, H. (2001) On sample size calculation in bioequivalence trials. Journal of Pharmacokinetics and Pharmaceutics, 28,155-169.

Chow, S.C., Shao, J., Wang, H. (2002) A note on sample size calculation for mean comparisons based on noncentral t-statistics, Journal of Biopharmaceutical Statistics, 12, 441-456.

Chow, S.C., Liu, J.P. (2010) Design and Analysis of Bioavailability and Bioequivalence Studies, 3rd Ed. Chapman & Hall/CRC, New York, 119-121.