Comparison of Elixhauser and Charlson Methods for Discriminative Performance in Mortality Risk in Patients with Schizophrenic Disorders

Abstract

Although Charlson Comorbidity Index scores (CCIS) and Elixhauser comorbidity index scores (ECIS) have been used to assess comorbidity in patients with schizophrenia, only CCIS, not ECIS, have been used to predict mortality in this population. This nationwide retrospective study investigated discriminative performance of mortality of these two scales in patients with schizophrenia. Exploiting Taiwan’s National Health Insurance Research Database (NHRID), we identified patients diagnosed with schizophrenia discharged from hospitals between Jan 1, 1996 and Dec 31, 2007. They were followed up for subsequent death. Comorbidities presented one year prior to hospital admissions were identified and adapted to the CCIS and ECIS. Discriminatory ability was evaluated using the adjusted hazard ratio and Akaike information criterion (AIC) and Harrell’s C-statistic. We identified 58,771 discharged patients with schizophrenic disorders and followed them for a mean of 10.4 years, 16.6% of whom had died. Both ECIS and CCIS were significantly associated with mortality, but ECIS had superior discriminatory ability by a lower AIC and higher Harrell’s C-statistic (201231 vs. 201400; 0.856 vs. 0.854, respectively). ECIS had better discriminative performance in mortality risk than CCIS in patients with schizophrenic disorders. Its use may be encouraged for risk adjustment in this population.