Sample size estimation

This was calculated on the requirement to have at least 80% power to detect a statistically significant difference at the 5% level if the true healing rates were 60% and 40% in the experimental and placebo groups.

Before the analysis reported here, a logistic regression analysis was undertaken and this confirmed an absence of any suggestion of interactions between drug, bandage, and dressing on the probability of ulcer healing within 24 weeks. Subsequently in this analysis the effect of drug alone was analysed separately. In accordance with the trial protocol, the primary assessment of the statistical significances of differences in healing rates is on the basis of the c2 test with Yates' correction for continuity applied to healing by 24 weeks.

Flow of participants

figure