Comorbidity and its change predict survival in incident dialysis patients

Background: Few studies have performed a comprehensive comparison of the prognostic importance of comorbidity to that of other case-mix factors influencing incident dialysis patients' survival. Longitudinal change in the comorbid illness burden of incident dialysis patients has not been measured. Comorbidity severity and its change may serve as important prognostic markers of survival, independent of other case-mix factors. Methods: The Choices for Healthy Outcomes in Caring for End-Stage Renal Disease Cohort Study used the Index of Coexistent Disease (ICED) to assess comorbidity at the initiation of chronic dialysis treatment (1,039 incident patients) and during follow-up (733 patients). Using proportional hazards regression analyses, the relationship to survival of baseline ICED level and change in ICED level was examined. Results: At the initiation of chronic dialysis treatment, 36% of patients were at ICED level 0 to 1 (least comorbidity severity); 35%, level 2; and 29%, level 3. After multivariable adjustment, baseline ICED level was the strongest predictor of subsequent mortality. Compared with ICED level 0 to 1, relative risks for mortality were 1.9 (95% confidence interval, 1.3 to 2.6) for ICED level 2 and 2.8 (95% confidence interval, 2.0 to 3.9) for ICED level 3. The prevalence and severity of most comorbid conditions increased during follow-up. After controlling for baseline ICED level and other factors, change in ICED level over time was significantly associated with mortality (P = 0.01). Conclusion: Indexing comorbidity when patients begin chronic dialysis therapy and recording the evolution of index scores yields a predictor of mortality risk that is independent of other case-mix factors.