1. ¹United States Renal Data System, Minneapolis, Minnesota, USA
2. ²Clinical Statistics, GPRD, Abbott Laboratories, Abbott Park, Illinois, USA
3. ³Department of Medicine, University of Minnesota, Minneapolis, Minnesota, USA

Correspondence: Jiannong Liu, United States Renal Data System Coordinating Center, 914 South 8th Street, Suite S-253, Minneapolis, Minnesota 55404, USA. E-mail: jliu@usrds.org

Received 2 January 2009; Revised 25 August 2009; Accepted 1 September 2009; Published online 11 November 2009.

Abstract

Since comorbid conditions are highly prevalent among patients with end-stage renal disease, indexes measuring them have been widely used to describe the comorbidity burden and to predict outcomes as well as adjust for their roles as confounders. The current comorbidity indexes, however, were developed for general populations or on small patient cohorts. In this study we developed a new index for mortality analyses of dialysis patients based on the 2000 US incident dialysis population, and validated this using the 1999 and 2001 incident and 2000 prevalent dialysis patient populations. Numerical weights were assigned to the comorbid conditions of atherosclerotic heart disease, congestive heart failure, cerebrovascular accident/transient ischemic attack, peripheral vascular disease, dysrhythmia, other cardiac diseases, chronic obstructive pulmonary disease, gastrointestinal bleeding, liver disease, cancer, and diabetes. A patient's comorbidity score was the sum of the weights corresponding to the individual conditions present and could be used as a continuous variable in analyses. Our index performance was almost identical to the individual comorbid conditions regarding model fit, predictive ability, and effect on inference, and it outperformed the widely used Charlson Comorbidity Index.