Decreased glomerular filtration rate (GFR) is an important cardiovascular risk factor, but estimated GFR (eGFR) may differ depending on whether it is based on creatinine or cystatin C. A combined creatinine/cystatin C equation has recently been shown to best estimate GFR; however, the benefits of using the combined equation for risk prediction in routine clinical care have been less studied. This study compares mortality risk prediction by eGFR using the combined creatinine/cystatin C equation (CKD-EPI), a sole creatinine equation (CKD-EPI) and a sole cystatin C equation (CAPA), respectively, using assays that are traceable to international calibrators.Methods and results
All patients analysed for both creatinine and cystatin C from the same blood sample tube (n = 13,054) during 2005-2007 in Uppsala University Hospital Laboratory were divided into eGFR risk categories>60, 30-60 and <30 mL/min/1.73 m2 by each eGFR equation. During follow-up (median 4.6 years), 4398 participants died, of which 1396 deaths were due to cardiovascular causes. Reduced eGFR was significantly associated with death as assessed by all eGFR equations. The net reclassification improvement (NRI) for the combination equation compared with the sole creatinine equation was 0.10 (p < 0.001) for all-cause mortality and 0.08 (p < 0.001) for cardiovascular mortality, indicating improved reclassification. In contrast, NRI for the combination equation, compared with the sole cystatin C equation, was -0.06 (p < 0.001) for all-cause mortality and -0.02 (p = 0.032) for cardiovascular mortality, indicating a worsened reclassification.Conclusions
In routine clinical care, cystatin C-based eGFR was more closely associated with mortality compared with both creatinine-based eGFR and creatinine/cystatin C-based eGFR.