Power analysis also indicated that a sample size of 52 would be needed for accurate power for the treatment group, and an N = 32 would be required to appropriately assess site differences. Figure 10.