#3568 THE ERROR OF ESTIMATED GFR IN PREDIALYSIS CARE: IMPACT ON CLINICAL DECISION MAKING

Abstract

Abstract Background and Aims The error of estimated glomerular filtration rate (eGFR) and its consequences in predialysis are unknown. Method We analysed 315 predialysis patients who underwent measured GFR (mGFR) by the clearance of iohexol and eGFR by 52 formulas. We evaluated the agreement between eGFR and mGFR by the concordance correlation coefficient (CCC), total deviation index (TDI) and coverage probability (CP). In a sub-analysis we investigated the impact of the error of eGFR in decisions like (i) starting dialysis, (ii) preparation for renal replacement therapy (RRT) and (iii) continuing with clinical follow-up. For this sub-analysis patients who started dialysis due to uremia or fluid overload were excluded (n = 18). Results Patients: 315 patients were included, 70% male and mean age was 66±13 years. 38% had diabetic nephropathy. Mean mGFRwas 22 ± 8 ml/min. Mean eGFR was 22 ± 8 ml/min (CKD-EPI), 22 ± 8 ml/min (MDRD) or 26 ±10 ml/min (24-h creatinine clearance). Cystatin C measurement was available in 266 patients (84%) and averaged 2.8 ± 1.08 mg/L. Over 70% of the patients had proteinuria 1588±2204 mg/24 h. Agreement between eGFR and mGFR eGFR had very low precision and accuracy in reflecting mGFR as reflected by average CCC, TDI and cp of 0.60, 70% and 22%, respectively. The error was comparable between creatinine- and cystatin-based formulas. Extreme variations -larger than 10 ml/min- between mGFR and eGFR were frequent. Differences in clinical decision making: Concerning clinical decisions, the error of formulas (mainly GFR overestimation) would have suggested (a) early unnecessary preparation for RTT in 14% of clinically stable patients evaluated by mGFR; (b) to continue with clinical follow-up in 59% of the subjects in whom RTT was indicated by low mGF and (c) to delay the starting of dialysis in all patients (n = 6) in whom RRT was indicated based on very low mGFR without uremic symptoms or fluid overload. Conclusion The error of formulas in predialysis care was frequent and large and may have consequences in clinical care.