Abstract
BACKGROUND: Renal impairment is a major risk factor for mortality in various populations. Three formulas are frequently used to assess both glomerular filtration rate (eGFR) or creatinine clearance (CrCl) and mortality prediction: body surface area adjusted-Cockcroft-Gault (CG-BSA), Modification of Diet in Renal Disease Study (MDRD4), and the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation. The CKD-EPI is the most accurate eGFR estimator as compared to a "gold-standard"; however, which of the latter is the best formula to assess prognosis remains to be clarified. This study aimed to compare the prognostic value of these formulas in predicting the risk of cardiovascular mortality (CVM) in population-based, cardiovascular risk, heart failure (HF) and post-myocardial infarction (MI) cohorts. METHODS: Two previously published cohorts of pooled patient data derived from the partners involved in the HOMAGE-consortium and from four clinical trials - CAPRICORN, EPHESUS, OPTIMAAL and VALIANT - the high risk MI initiative, were used. A total of 54,111 patients were included in the present analysis: 2644 from population-based cohorts; 20,895 from cardiovascular risk cohorts; 1801 from heart failure cohorts; and 28,771 from post-myocardial infarction cohorts. Participants were patients enrolled in the respective cohorts and trials. The primary outcome was CVM. RESULTS: All formulas were strongly and independently associated with CVM. Lower eGFR/CrCl was associated with increasing CVM rates for values below 60 mL/min/m(2). Categorical renal function stages diverged in a more pronounced manner with the CG-BSA formula in all populations (higher χ(2) values), with lower stages showing stronger associations. The discriminative improvement driven by the CG-BSA formula was superior to that of MDRD4 and CKD-EPI, but remained low overall (increase in C-index ranging from 0.5 to 2 %) while not statistically significant in population-based cohorts. The integrated discrimination improvement and net reclassification improvement were higher (P < 0.05) for the CG-BSA formula compared to MDRD4 and CKD-EPI in CV risk, HF and post-MI cohorts, but not in population-based cohorts. The CKD-EPI formula was superior overall to MDRD4. CONCLUSIONS: The CG-BSA formula was slightly more accurate in predicting CVM in CV risk, HF, and post-MI cohorts (but not in population-based cohorts). However, the CG-BSA discriminative improvement was globally low compared to MDRD4 and especially CKD-EPI, the latter offering the best compromise between renal function estimation and CVM prediction.