Performance of Charlson Comorbidity Index and Elixhauser Comorbidity Index to Predict Prognosis Among Cancer Patients

Objective: To evaluate the performance of the Charlson comorbidity index (CCI) and the Elixhauser comorbidity index (ECI) in predicting 3-year mortality in cancer patients.Methods: We identified patients who were first hospitalized with gastric, lung, liver, or colorectal cancer between 2008 and 2014 in Korean National Health Insurance data.The study population was classified into four age groups (25-54, 55-74, 75-84, and ≥85 years).Patients were categorized according to their CCI scores and ECI counts (0, 1-2, 3-4, and ≥5).Cox proportional hazards models were used to estimate hazard ratios and 95% confidence intervals for 3-year mortality depending on their comorbidity categorized by each index.Predictive discrimination was assessed using the area under the curve (AUC) by adding each index to a baseline model with age and sex.Results: Among 13,579 patients with cancer, 21.3%, 54.2%, 19.8% and 4.8% were aged 25-54, 55-74, 75-84, and ≥85 years, respectively.Adding the CCI or ECI to the baseline model did not significantly improve mortality prediction among patients aged 25-54 years or ≥85 years.Significant discrimination was observed only for colorectal cancer after adding the CCI or ECI to the baseline model (CCI AUC 0.686, p = 0.008; ECI AUC 0.682, p = 0.024).Conclusion: Adding the CCI or ECI to the baseline model produced statistically significant but modest improvements in discrimination only among patients aged 55-84 years and those with colorectal cancer.Although the baseline model for lung cancer already showed acceptable discrimination (AUC >0.70), adding either comorbidity index provided no significant additional improvement.Only severe comorbidity burden (≥5) was significantly associated with mortality.(PeRM 2026;18:147-158)