Figure 5

Operating within the confines of cure rates which should exceed 90%, and thus trials which must have the statistical power to exclude, with 95% confidence, that the cure rate of the new antimalarial does exceed this value, the relationship between sample size (y axis) and cure rate of the established treatment (x axis) is illustrated. For superiority trials with 2α= 0.05 and β = 0.2, the dashed line shows the lower limit of possible sample sizes; i.e. when the new treatment has an estimated efficacy of 100%. For non-inferiority trials the solid line shows the sample size required to ensure that the 95% CI for treatment efficacy of the new treatment exceeds 90%; i.e. the non-inferiority margin δ is (cure rate of the standard treatment-0.9), assuming that the true cure rates are the same. These assume 1:1 randomization.