Sample size nomograms for interpreting negative clinical studies

Abstract

In recent years there has been increasing attention to the appropriate interpretation of a clinical study. One special concern has been the difficulty inherent in interpreting studies that were not statistically significant: Was the sample size sufficient to detect a clinically important effect if, in fact, it existed? This concern is further complicated because readers may have differing opinions of what size effect is clinically important. A pair of sample size nomograms has been developed, using common levels of statistical significance, to assist in this interpretation. The nomograms are intended to provide the clinician with a handy and easy-to-use reference for ascertaining whether an apparently negative study has a sample size adequate to detect reliably any difference between treatment groups that the clinician believes is clinically important. Examples are provided to show these principles and the use of the nomograms in interpreting negative studies.