Abstract:
|
Non-inferiority tests are well developed for randomized parallel group trials where the control and experimental groups are independent. However, these tests may not be appropriate for non-inferiority matched study designs where many control subjects are matched on baseline covariates to each experimental subject. These tests may require an adjustment to account for the correlation among matched subjects. We propose a method that extends Farrington-Manning's (FM) non-inferiority test to the case of correlated "many-to-one" matched data. We conducted a Monte Carlo simulation study to compare the size and power of the proposed test statistic with tests developed for clustered matched pair data. In the presence of intra-class correlation (ICC), the sizes of tests developed for clustered matched pair data are inflated when applied to "many-to-one" matched data. The size of the proposed method is close to the nominal level for a variety of correlation patterns. As the ICC between the experimental and control groups increases, the proposed test becomes more powerful than the FM test. The proposed method is a good alternative to the FM test for non-inferiority matched studies.
|