How to design Sample Size; A Practical Guide for Clinical Research

Abstract

In clinical trials, it is necessary to design the sample size in advance from scientific, ethical, and practical aspects. This paper describes key points and practical methods for designing sample size. First, the sample size must be designed without calculation in some cases, such as in observational studies using only existing information or studies on rare diseases. On the other hand, when calculating sample size, there are three factors to consider: alpha, beta, and effect size. Once these three points are determined, the sample size is calculated using a calculation tool. There is little room for consideration of α and β since they are often set at 0.05 and 0.2, respectively. Therefore, the sample size calculation effect size should be fully considered. In determining the effect size, it is recommended to refer to values from previous literature and similar studies or to adopt clinically meaningful values. It is also important to note that the elements that should be determined as the effect size varies depending on the study design and primary endpoint. Furthermore, there are more matters to consider in complex study designs（e.g., studies with matching, survival time analysis, non-inferiority trials, etc.）, and sample size design can be complicated if you are unfamiliar with it. In such cases, it is recommended that biostatisticians or epidemiologists be consulted to employ appropriate calculation methods.