GFR prediction using the MDRD and Cockcroft and Gault equations in patients with end-stage renal disease

Background. Although prediction equations are recommended to determine GFR and creatinine clearance (CrCl), neither the MDRD equations nor the Cockcroft and Gault formula have been validated for the low levels of GFR present in end-stage renal disease (ESRD). The accuracy of the MDRD equations and the Cockcroft and Gault formula in predicting GFR and CrCl, respectively, was examined in patients with ESRD and its relationship to the basal GFR and two markers of malnutrition, urinary creatinine and body fat determined. Methods. Inulin clearance (Cin) was measured in 26 non-diabetic patients with ESRD and the 24 h CrCl determined. GFR was predicted using three equations derived from the MDRD study population containing four to six variables. Both CrCl and GFR were predicted from the Cockcroft and Gault formula. Estimates of bias and precision were obtained and Bland and Altman analysis performed. Body fat was measured by DEXA scan. Results. The predicted GFR (MDRD) was 10% lower than Cin (8.83±0.71 ml/min/1.73 m2) with all three MDRD equations, showing a similar degree of precision and bias. Cin gave a negative correlation with the difference between the predicted GFR (MDRD) and the measured GFR. The predicted GFR (MDRD) underestimated GFR when Cin >8 ml/min/1.73 m2 but overestimated GFR when Cin <8 ml/min/1.73 m2. The Cockcroft and Gault formula overestimated CrCl by 14% and overestimated Cin by 35%. Cin gave a negative correlation with the difference between the predicted GFR (Cockcroft and Gault) and measured GFR, overestimating GFR when Cin <13 ml/min/1.73 m2. The overestimation of GFR by the MDRD equation was not associated with urinary creatinine excretion. However, both Cockcroft and Gault and the MDRD predictions showed a positive, but weak, correlation with body fat. Conclusion. The MDRD equations were more accurate in predicting the group mean GFR in patients with ESRD than the Cockcroft and Gault formula. However, the predicted GFR using either formula was related to the basal GFR and percentage body fat.