Can you please help me with sample size calculation of a clustered RCT where the outcome is a rate and there is a baseline rate? That is at the beginning of the study both groups will have an outcome rate of 76/1000 days. We will have 23 clusters in each arm and 104 participants per cluster (SD=33) and expecting to follow-up the groups for 1 years. We need to know the minimum detectable difference for 80% power at 5% level with an ICC of 0.01.
I am using the following Stata command but the effect size looks very high.
clustersampsi, rates detectabledifference r1(.076) m(37960) k(23) rho(.01) size_cv(.45) alpha(0.05) beta(0.8) base_correl(0)
when I am doing the calculation for a stepped wedged RCT with 3 steps (3 months duration) it is coming much lower.
steppedwedge, rates detectabledifference complete(1) vartotal(1) r1(.076) m(9360) k(12) rho(.01) alpha(0.05) beta(0.8) steps(3)
The difference is very high. I want to cross-check this with a SAS program.
