Using the Johns Hopkins Aggregated Diagnosis Groups (ADGs) to predict mortality in a population-based cohort of adults with schizophrenia in Ontario, Canada

Administrative health care databases are increasingly used for health services and comparative effectiveness research. When comparing outcomes between different treatments, interventions and exposures, the ability to adjust for differences in the risk of the outcome occurring between treatment groups is important. There is a paucity of validated methods to ascertain comorbidities for risk-adjustment in ambulatory populations of subjects with schizophrenia using administrative health care databases. Our objective was to examine the ability of th\e Johns Hopkins' Aggregated Diagnosis Groups (ADGs) to predict 1-year mortality in a population-based cohort of subjects with schizophrenia. We used a retrospective cohort constructed using population-based administrative data that consisted of all 94,466 residents of Ontario, Canada between the ages of 20 and 100 years who were alive on January 1, 2007 and who had been diagnosed with schizophrenia prior to this date. Subjects were randomly divided into derivation and validation samples. A logistic regression model consisting of age, sex, and indicator variables for 14 of the 32 ADG categories had excellent discrimination: the c-statistic (equivalent to the area under the ROC curve) was 0.845 and 0.836 in the derivation and validation samples, respectively. Furthermore, the model demonstrated very good calibration.