Therefore, we are unable to determine the independent effect of physician reminders and patient mailings on quality, and the observed reduction in racial disparities might be caused by factors other than quality improvement, including regression to the mean or changes in the characteristics of patients in the rolling cohort.