Accuracy of the MDRD (Modification of Diet in Renal Disease) study and CKD-EPI (CKD Epidemiology Collaboration) equations for estimation of GFR in the elderly

Background: Glomerular filtration rate (GFR) is a measure of kidney function, commonly estimated using equations that adjust serum creatinine concentration for age, race, and sex. The Modification of Diet in Renal Disease (MDRD) Study equation is widely used, but underestimates GFR at higher levels. The serum creatinine-based Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI(cr)) equation generally provides more accurate estimation at GFR >60 mL/min/1.73 m(2). Newer equations have been reported using cystatin C concentration either alone (CKD-EPI(cys)) or in combination with creatinine concentration (CKD-EPI(cr-cys)). None of these equations has been well validated in older people. We tested the accuracy of these equations in people 74 years or older compared with GFR measured by a reference method.

Study design: Diagnostic test evaluation in a prospective cohort.

Setting & participants: Participants (n = 394; median age, 80 [range, 74-97] years) recruited from nephrology clinics and the community.

Index test: GFR estimated using the MDRD Study, CKD-EPI(cr), CKD-EPI(cys) and CKD-EPI(cr-cys) equations.

Reference test: GFR measured using an iohexol clearance method.

Results: Median measured GFR was 53.4 (range, 7.2-100.9) mL/min/1.73 m(2). MDRD Study-, CKD-EPI(cr)-, and CKD-EPI(cr-cys)-estimated GFRs overestimated GFR (median differences of 3.5 [P< 0.001], 1.7 [P < 0.001], and 0.8 [P = 0.02] mL/min/1.73 m(2), respectively); the CKD-EPI(cys) equation was unbiased. Accuracy (percentage of estimates within 30% of measured GFR [P(30)]) was 81%, 83%, 86%, and 86% for the MDRD Study, CKD-EPI(cr), CKD-EPI(cys), and CKD-EPI(cr-cys) equations, respectively. Accuracy of the MDRD Study equation was inferior (P = 0.004) to the CKD-EPI(cr) equation at GFR >60 mL/min/1.73 m(2).

Limitations: Those of non-European ancestry were not included. For practical reasons, only a 4-hour sampling protocol was used for iohexol clearance.

Conclusions: The CKD-EPI(cr) equation appeared less biased and was more accurate than the MDRD Study equation. No equation achieved an ideal P(30) in the overall population. Our data suggest that GFR estimation is as satisfactory in older people of European ancestry as it has been reported to be in younger individuals.