Sample size calculations for linear mixed models of rate of change based on lmer, lme, or gee "placebo" pilot estimates.