How to overcome the hurdles of running a clinical trial?

Abstract

The statistical method is an integral part of clinical trials, because clinical trials are scientific experiments involving human subjects and are to produce objective inferences as to the benefit of intervention – including more narrowly defined medical regiment – under evaluation. Many interventions with reputed anecdotal evidence of plausible benefit turn out to be not so useful if not outright harmful. Furthermore, a treatment good for one group of patients may not be necessarily good for another group of patients receiving a placebo. Clinical trial data are complex and rarely complete. Recently, behavior modifications such as smoking cessation and diet are being evaluated by clinical trial methods. People believed beta-carotene and vitamin C would prevent cancer, but clinical trials have shown no benefit. This points out that the medical plausibility should not be accepted blindly. I will explain the role of statisticians in planning, designing and analyzing clinical trials mainly from my experience.