I have a pre-post study measuring change in quality of life scores with a treatment cohort and control cohort. However, the treatment cohort is at "high risk" so regression to the mean is a real concern here. I'd like to use linear mixed modeling and I saw another cross validated contributer mention that linear mixed models account for regression to the mean. Is this true? I have always thought that linear mixed models were vulnerable to regression to the mean just like paired t-tests and repeated measures ANOVA. And that you should use analysis of covariance instead. Does anyone have any thoughts or references on this?