Sample size estimation for clinicians.

Abstract

The purpose of this paper is to describe the importance of an adequate sample size in clinical research and to enable clinicians to estimate their own sample size requirements for the more common types of studies (comparison of two means or two proportions). Four pieces of information are required to determine sample size: the desired level of statistical power, the level of statistical significance, the variability of the data, and the smallest difference between the study groups that is considered to be of clinical significance. Worked examples from the literature are used to illustrate how clinicians may easily do their own sample size calculations using published tables or available computer software, or both. The consideration of sample size and power during the planning stages of clinical research is crucial to the subsequent interpretation of study results, especially if the study is negative, and yet this point is often neglected in the medical literature. Attention to these simple guidelines will help ensure that research results lead to valid conclusions.