In Reply: Dr Tsai requests clarification regarding the unit of analysis in our study. The initial composite score for each hospital was a crude composite that was calculated simply as the total number of fulfilled opportunities over all patients in the hospital divided by the total number of opportunities. For example, suppose a hospital treated 10 patients who in total had 45 “process opportunities” (not necessarily the same number of opportunities for each patient) and actually gave correct care in 30 of these opportunities: the hospital's crude composite process performance score would be 30 out of 45 (67%). To explore the association with case mix, the outcome for the next stage of the analysis was a dichotomous end point indicating whether a given process-of-care opportunity was correctly fulfilled. We used a hierarchical model to adjust the outcome for the indicators of core performance measures (treatment opportunity mix) and various patient-level covariates (such as patient age and race) as fixed effects and a hospital indicator variable as a random effect.
Mehta RH, Liang L, Peterson ED. Analyzing Patient Case Mix and Hospital Rankings—Reply. JAMA. 2009;301(11):1125. doi:10.1001/jama.2009.305