MON-264 Comparison of performance of various estimated glomerular filtration rate equations with measured creatinine clearance from 24-hour urine

Abstract

There are several equations to estimate glomerular filtration rate (GFR), but each of the methods have specific limitations. We compared various creatinine-based GFR with measured GFR by 24-hour urine collection (24u-GFR). A total of 731 24-hours urine samples from 245 patients with CKD Stages 1-3 were analyzed by following equations: Cockcroft-Gault formula (CG); 4 variable abbreviated MDRD equation (abbMDRD); 4 variable IDMS-MDRD equation (IDMS-MDRD); Chronic Kidney Disease Epidemiology Collaboration equation (CKD-EPI); and novel 4 variable equations for Korean (KOR-MDRD and KOR-IDMS-MDRD). The adequacy of 24-hour urine samples was analyzed by calculating the predicted daily creatinine excretion. Of 731 urine samples, 455 urine samples were well collected as measured urine creatinine/estimated urine creatinine between 0.8-1.2. The IDMS-MDRD GFR showed the best overall correlation with 24u-GFR (R=0.936, P<0.001) and best reliability by intraclass correlation coefficient (ICC, 0.914, 95% CI 0.897-0.928, P<0.001). In subgroup analysis, IDMS-MDRD showed the highest ICC values in CKD stage 2 and 3 (0.622 in CKD stage 2, 0.820 in CKD Stage 3, all P<0.001). In CKD Stage 1, KOR-MDRD showed the highest ICC value (0.764, P<0.001) and the best correlation with 24u-GFR (R=0.773, P<0.001). Fifteen to thirty percent accuracy (P15, P30) of KOR-MDRD was the highest in overall (P15: 63.7, P30: 97.4), and CKD-EPI and KOR-IDMS-MDRD showed higher P30 than other equations in CKD stage 1 (92.4 and 94.9). The estimated GFR with IDMS-MDRD equation was well calibrated with 24u-GFR in CKD stage 2 and 3. In CKD stage I, KOR-MDRD equation showed good performance with 24u-GFR.