Comparing comorbid-illness indices assessing outcome variation

Abstract

OBJECTIVE: We investigated and compared the effects of three different comorbid indices on selection of procedure and outcome variation to determine which, if any, could be used for interpreting outcomes data.

DESIGN: Retrospective cohort study.

SETTING: Large multispecialty group practice.

PATIENTS: Patients (aged 55–85 years) with residence in the United States who underwent a first-time prostatectomy for benign prostatic hyperplasia: 302 total; 253 transurethral procedures (TURP) versus 49 open procedures (OP).

MEASUREMENTS AND MAIN RESULTS: The following indices were used to assess comorbid disease: Charlson index (CI), index of coexistent disease (ICED), and Kaplan-Feinstein index (KFI). The main outcome measure was the five-year mortality rate. The unadjusted five-year mortality rates were 16% (40/ 253) for TURP and 4% (2/49) for OP; survival analysis revealed this difference to be marginally significant at thep=.05 level. In an effort to control for the effect of comorbidity, CI, ICED, and KFI were independently assessed; together with age, they each had similar effects in rendering the risk of death associated with procedure type insignificant. However, comorbidity, as derived with ICED (not CI or KFI), was identified as a confounding variable when assessing the five-year mortality rate after prostatectomy as ICED was associated with the procedure type (predictor variable) and the five-year mortality (outcome variable).

CONCLUSION: Differences in the composition and scoring of comorbid indices may have important implications for interpreting outcomes data. Nevertheless, these results, together with those of previous studies, suggest that the reported increased mortality for patients undergoing TURP is probably due to case-mix differences.