How many do I need? Basic principles of sample size estimation.

Abstract

BACKGROUND: In conducting randomized trials, formal estimations of sample size are required to ensure that the probability of missing an important difference is small, to reduce unnecessary cost and to reduce wastage. Nevertheless, this aspect of research design often causes confusion for the novice researcher. AIM: This paper attempts to demystify the process of sample size estimation by explaining some of the basic concepts and issues to consider in determining appropriate sample sizes. METHOD: Using a hypothetical two group, randomized trial as an example, we examine each of the basic issues that require consideration in estimating appropriate sample sizes. Issues discussed include: the ethics of randomized trials, the randomized trial, the null hypothesis, effect size, probability, significance level and type I error, and power and type II error. The paper concludes with examples of sample size estimations with varying effect size, power and alpha levels. CONCLUSION: Health care researchers should carefully consider each of the aspects inherent in sample size estimations. Such consideration is essential if care is to be based on sound evidence, which has been collected with due consideration of resource use, clinically important differences and the need to avoid, as far as possible, types I and II errors. If the techniques they employ are not appropriate, researchers run the risk of misinterpreting findings due to inappropriate, unrepresentative and biased samples.