Comparison of CKD-EPI and MDRD to estimate baseline renal function in HIV-positive patients

Background. Renal dysfunction is common in HIV-positive patients, and guidelines suggest regular monitoring of renal function with estimated glomerular filtration rate (eGFR) and urinalysis. It is unknown whether Chronic Kidney Disease Epidemiological Collaboration (CKD-EPI) or Modification of Diet in Renal Disease (MDRD) provide better estimates of glomerular filtration rate (GFR) in this population. Methods. We compared the CKD-EPI and MDRD equations to estimate GFR at baseline in 20 132 HIV-positive individuals in the UK CHIC cohort. Kappa statistics and Bland-Altman plots were used to assess agreement between the two estimates and Kaplan-Meier plots and Cox regression analysis to describe mortality patterns. Results. At baseline, median eGFR was 100 (87, 112) (CKD-EPI) and 94 (83, 108) (MDRD) (mL/min/1.73m(2)). Good overall agreement between CKD-EPI- and MDRD-defined eGFR bands was observed (Kappa = 0.71, 95% confidence interval: 0.70-0.72). Of the 367 patients with eGFR MDRD 30-59, 57 (15.5%) were categorized as eGFR 60-89 by CKD-EPI. After adjustment for covariates, eGFR <60 (CKD-EPI), eGFR <30 (MDRD) and eGFR >= 105 (both formulae) were significantly associated with an increased risk of death. Mortality in patients classified as having eGFR 60-89 by CKD-EPI and eGFR 3059 by MDRD more closely resembled mortality of patients who had eGFR 60-89 by both formulae. Conclusions. MDRD and CKD-EPI equations showed a high degree of agreement in stratifying patients by baseline eGFR. CKD-EPI estimates of GFR <60 at baseline are more strongly associated with mortality than MDRD estimates of GFR <60, supporting the concept that MDRD may have overestimated the severity of renal impairment in these patients. Our findings support the use of CKD-EPI in HIV-positive individuals.