Aim: To compare the utility of various admission-based comorbidity indices in men diagnosed with non-metastatic prostate cancer. Methods: The study cohort consisted of men diagnosed with prostate cancer between January 2002 and December 2020 according to the state-wide South Australian Cancer Registry. Comorbid conditions were captured for 11,470 men through linkage to public hospital admission data 5-years prior to prostate cancer diagnosis. The comorbidity indices evaluated included the Charlson Comorbidity Index (CCI), Elixhauser Comorbidity Index (ECI), National Cancer Institute (NCI) comorbidity index, and Cancer, Care and Comorbidity (C3) index. The predictive performance of the four indices for 5-year overall mortality was compared using the C-statistic from Cox proportional hazard models adjusted for age, socioeconomic status, and year of prostate cancer diagnosis. Results: Approximately 31 %, 45 %, 28 % and 47 % of patients had at least one comorbid condition captured by CCI, ECI, NCI and C3, respectively. Regarding the prediction of 5-year overall survival, CCI (c-index = 0.763) slightly higher predictive performance than ECI (0.758), NCI (0.755), and C3 (0.754). Indices in their continuous score resulted in better predictive performance than them being used categorically (0, 1, and 2 +). The NCI (continuous score) showed a stronger association with overall mortality (hazard ratio (HR) 2.47, 95%CI:2.29–2.67) than the other indices, despite its predictive performance being lower than the CCI and ECI. Conclusion: There were only slight differences in the predictive accuracy among the indices, with the CCI having a slightly better prognostic value than the other indices. All four indices demonstrated a strong association with mortality in men diagnosed with prostate cancer.
- Prostate cancer