Sample size for testing a proportion in clinical trials: A 'two-step' procedure combining power and confidence interval expected width

摘要

The power of the statistical test, sample size calculations should consider the precision of the effect estimate. The expected width of the exact confidence interval, based on inverting the binomial test, is an appropriate measure of the latter, together with the probability of obtaining, under the alternative hypothesis, confidence intervals with a width less than the expected one. In testing a proportion against a reference value, the authors devised a 'two-step' procedure. At the first step, the required sample size for the binomial testing is obtained; then at the second step, this sample size is iteratively increased until the probability of obtaining exact confidence intervals with a width less than the expected one exceeds a required threshold. In Section 2, after recalling the exact procedures for calculating the sample size and the confidence interval (CI) for the binomial distribution, the expected width (EW) of CI and the expected probability of CI's with a width less than the expected one are introduced. Section 3 gives the details of the proposed approach with an example of the overall procedure, and the discussion is reported in Section 4. (18 refs.)