Xu Jin, Yu Menggang
Department of Statistics and Actuarial Science, East China Normal University, Shanghai, 200241, China.
Biom J. 2013 May;55(3):430-43. doi: 10.1002/bimj.201100231. Epub 2013 Apr 2.
Motivated by a recent symptom management trial to simultaneously assess multiple binary endpoints for cancer chemotherapy, we extend the univariate McNemar test to multivariate cases for doubly blinded clinical trials with matched pairs. We propose a general method to test noninferiority or equivalence. The method employs the intersection-union principle on the marginal score statistics to obtain an asymptotic α-level test. Power formula and sample size calculation are provided by a simple numerical method that accounts for the correlation structure among the endpoints. We further consider sample size re-estimation through internal pilot study. To avoid the need of unblinding for doubly blinded trials, we also propose a blinded approach for nuisance parameter estimation. The effectiveness of the proposed methods is demonstrated by simulation studies. Application to the cancer chemotherapy trial is illustrated.
受近期一项针对癌症化疗同时评估多个二元终点的症状管理试验的启发，我们将单变量麦克尼马尔检验扩展到多变量情形，用于配对双盲临床试验。我们提出了一种检验非劣效性或等效性的通用方法。该方法在边际得分统计量上采用交并原则以获得渐近α水平检验。通过一种考虑终点间相关结构的简单数值方法提供了功效公式和样本量计算。我们进一步考虑通过内部预试验进行样本量重新估计。为避免双盲试验中揭盲的需要，我们还提出了一种用于干扰参数估计的盲法。模拟研究证明了所提方法的有效性。文中还举例说明了该方法在癌症化疗试验中的应用。
