Abstract
BACKGROUND: Estimated GFR by serum creatinine (eGFRcreatinine) is a pivotal measure of kidney function in clinical practice but can be affected by several non-GFR determinants, resulting in misclassification. Combining multiple kidney markers to predict risk is an area of substantial interest. DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS: This study followed 9489 adults from visit 4 (1996-1998) of the Atherosclerosis Risk in Communities Study for a median of 11.2 years, and assessed joint association of eGFRcreatinine, eGFRcystatin, and urinary albumin/creatinine ratio (ACR) with mortality, coronary heart disease, heart failure, AKI, and ESRD using Cox proportional hazards models. The predictive ability of ACR and eGFRcystatin beyond eGFRcreatinine was also investigated. RESULTS: Lower eGFRcreatinine and eGFRcystatin as well as elevated ACR were independently associated with risk for all outcomes. eGFRcreatinine <60 was not associated with risk of mortality, coronary heart disease, or heart failure if eGFRcystatin ≥60 with ACR <30 mg/g compared with those with all three markers above CKD cutoffs (i.e., eGFRcystatin ≥60, eGFRcreatinine ≥60, and ACR<30), whereas risk association with kidney outcomes remained: Hazard ratio (95% confidence interval), 0.96 (0.66, 1.39) for mortality, 0.85 (0.55, 1.31) for coronary heart disease, 0.99 (0.60, 1.63) for heart failure, 1.61 (0.92, 2.82) for AKI, and 3.53 (1.06, 11.68) for ESRD. Adding ACR to the fully adjusted model with eGFRcreatinine or adding eGFRcystatin to both eGFRcreatinine and ACR improved risk classification for all outcomes (P ≤ 0.01). CONCLUSIONS: eGFRcystatin can be a useful confirmatory marker in those with eGFRcreatinine <60 and whose ACR is <30 mg/g. This approach improves risk classification, and provides reassurance to a large group of individuals with eGFRcreatinine <60.