Validation of the Lund-Malmö, Chronic Kidney Disease Epidemiology (CKD-EPI) and Modification of Diet in Renal Disease (MDRD) equations to estimate glomerular filtration rate in a large Swedish clinical population.

Abstract Objective. The aim of this study was to validate externally the Swedish Lund-Malmö revised creatinine-based glomerular filtration rate (GFR) equations (LM Revised) in a Swedish cohort in comparison with the North American Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology (CKD-EPI) equations. Material and methods. The study included 1397 examinations [median age 61 years, median body mass index (BMI) 26 kg/m(2)] in 996 patients referred for iohexol clearance (median 44 ml/min/1.73 m(2)). Bias, precision [interquartile range (IQR)], accuracy expressed as percentage of estimates ± 10% (P (10)) and ± 30% (P (30)) of measured GFR, and classification ability for five GFR stages (<15, 15-29, 30-59, 60-89 and ≥90 ml/min/1.73 m(2)) were compared. Results. Overall, all three equations performed satisfactorily: LM Revised, MDRD, CKD-EPI showed, respectively, a median bias of -5.8%, -2.2% and 1.7%, IQR 11.9, 12.3 and 11.7 ml/min/1.73 m(2), P (10) 35%, 34% and 38%, P (30) 84%, 79% and 79% and correctly classified GFR stages 68%, 65% and 69%. LM Revised was at least as accurate in terms of P (30) as the other equations at GFR intervals <90, while CKD-EPI was the only unbiased and the most accurate equation at ≥90 ml/min/1.73 m(2). LM Revised was more stable in terms of bias and accuracy across age and BMI groups than MDRD and CKD-EPI. Both MDRD and CKD-EPI overestimated measured GFR among elderly patients and in the small group of underweight men. Conclusion. The ideal all-purpose GFR prediction equation does not exist. LM Revised should be preferred in patients with suspected or known renal insufficiency, while CKD-EPI is most useful in settings where patients with no a priori suspicion of renal impairment are evaluated. Differences in creatinine measurements between laboratories may limit the generalizability of the present validation.