How to calculate clinical study sample size

Sample size calculation is required in your protocol and IRB submission. This guide covers common cases without requiring specialized software.

1. 1

   Define your hypothesis and primary variable type

   State H0 and H1. Identify if the primary variable is continuous, binary, or time-to-event.

2. 2

   Estimate the minimum clinically relevant effect

   Not the expected effect, but the minimum that would be clinically useful.

3. 3

   Choose significance level and power

   Conventionally alpha = 0.05, power = 0.80. Confirmatory studies may require 0.90.

4. 4

   Estimate variance or baseline proportion

   For continuous variables, estimate SD. For binary, baseline proportion in the control group.

5. 5

   Apply the appropriate formula

   Mean difference, proportion difference, log-rank for survival — each has its formula. Use G*Power or R (pwr package).

6. 6

   Add attrition margin

   Increase final sample size by 10-20% for dropouts and incomplete data.

7. 7

   Document the calculation in the protocol

   IRB requires formula, assumptions, sources, and software used.