A recently published article in the International Journal of Epidemiology discusses sample size requirements for cluster randomized trials when the size of the cluster itself varies. The authors develop an approximation that uses the coefficient of variation (CV) of the distribution of cluster sizes. The authors point out that in many clinical trials, the CV of the cluster size can be large and that 0.65 is a reasonable expection. They also point out that when the CV of cluster size is small (less than 0.23) the effect on sample size calculations is negligible.
You can find an earlier version of this page on my old website.