Sample size calculations for intervention trials in primary care randomizing by primary care group: an empirical illustration from one proposed intervention trial.