Determining sample size and power in clinical trials: the forgotten essential.

Estimation of the sample size is a fundamental but usually ignored requirement of the randomized controlled trial (RCT). Indeed, the publication of small trials without consideration of sample size is worrisome from both medical and ethical viewpoints. Type II errors are common, and readers and investigators may reject worthwhile treatments and interventions. Before embarking on an RCT, the investigator must choose an alpha, a beta, and the rates of outcomes anticipated in both treatment groups. This should reflect the characteristics of the condition and its treatment. If limited sample size or available resources pose a problem, the use of continuous outcome measures, paired before-after measurements, and more common outcome measures can minimize sample size requirements. If these approaches are not satisfactory, then a multicenter trial may be in order.