ABSTRACT
Sample size is a critical design speci‚cation for a clinical trial to render a statistically meaningful interpretation of the trial’s results. To plan the sample size or broadly the amount of statistical information for a randomized clinical trial to meet its objective, the concept of detecting a postulated treatment effect with a suf‚cient statistical power and the concept of estimating a treatment effect with a suf‚cient level of precision are commonly entertained. The key assumption behind these concepts is that the treatment effect parameter used in the hypothesis testing is constant and representative of the treatment effect in every subset of the study patient population. Under the global clinical trial strategy in which the same study protocol is implemented in two or more geographical regions, this assumption may be doubtful in the sense that the treatment effect may be heterogeneous across the geographical regions for a variety of plausible reasons. Such interregion heterogeneity will impact the sample size planning. The bridging trial strategy as described in the International Conference on Harmonisation (ICH, 1998) E5 document relies largely on the comparisons between the treatment effects in the foreign regions and the treatment effects in the new region. The across-trial comparisons will further impact the sample size planning.
