Sample size calculations for the ALIGN cluster randomised trial

Justification for the original sample size calculation is made in the trial protocol. In this additional file, we provide details of the assumptions and calculations in estimating the likely width of the 95% confidence interval (CI) for the estimators, the risk difference and the log odds ratio, for the primary practitioner outcome, x-ray referral. Table 1 contains the likely widths of the 95% CIs for these estimators assuming a range of values of the sample size parameters (cluster (practice) size, coefficient of variation (CV), design effect, and control and intervention group x-ray referral rates). Justification for the selected values of the parameters follows.