Predicting Renal Replacement Therapy and Mortality in CKD
Received 18 December 2006; accepted 5 July 2007.
Background
Prognostic risk scores can help clinicians intervene on higher risk patients and counsel them. Our objective is to identify characteristics that predict the rate of progression to renal replacement therapy (RRT) and evaluate how those characteristics predict mortality and a composite end point (RRT and mortality).
Study Design
Retrospective cohort study.
Setting & Participants
We conducted the study at Kaiser Permanente Northwest, a health maintenance organization. We followed up members with an estimated glomerular filtration rate (eGFR) that indicated chronic kidney disease (2 eGFRs < 60 mL/min/1.73 m2 [<1.0 mL/s/1.73 m2] at least 90 days apart).
Predictors
We measured baseline clinical characteristics between January 1997 and June 2000 by using electronic medical records and patients’ histories of hospitalization.
Outcomes & Measurements
We calculated adjusted hazard ratios and concordance statistics for progression to RRT, mortality, and the composite by using Cox regression.
Results
Patients (n = 6,541) were followed up for up to 5 years. We observed 1.6 progressions to RRT/100 person-years and 11.4 deaths/100 person-years. The 6 characteristics of age, sex, eGFR, diabetes, hypertension, and anemia predicted RRT effectively (c statistic, 0.91). However, hypertension and age predicted in the opposite direction for mortality and its composite end point. The c statistic decreased: mortality (0.70), mortality and RRT (0.71).
Limitations
Characteristics were measured without a protocol; extensive missing data prevented the evaluation of known risk factors (eg, proteinuria).
Conclusions
Predicting RRT effectively requires a separate risk score. Predicting the composite end point would favor characteristics that predict mortality because it is 7 times as common as RRT.
Address correspondence to David H. Smith, RPh, PhD, Center for Health Research, Kaiser Permanente Northwest, 3800 N Interstate Ave, Portland, OR 97227.
ABSTRACT