Assessing glomerular filtration rate in healthy Indian adults: a comparison of various prediction equations

Abstract

Background: Accurate measurement of renal function is important for the diagnosis, stratification and management of kidney disease. As the use of recommended filtration markers is limited by cumbersome and costly techniques, renal function is typically estimated by using various specifically derived prediction equations. Most of these equations have been derived from Caucasian patients suffering from varying degrees of chronic kidney disease. This study considers the validity of these equations in an Indian population without known kidney disease.

Methods: One hundred and twenty-two consecutive renal donors who had undergone 99mTc-diethylenetriaminepentaaceticacid (DTPA) glomerular filtration rate (GFR) estimation were enrolled. The predictive capabilities of the Cockcroft and Gault equation for creatinine clearance corrected for body surface area (CG-CrCl), CG-CrCl corrected for GFR (CG-GFR), modification of diet in renal disease (MDRD) 1, MDRD 2 and 24-hr urinary creatinine clearance (urine-CrCl) were evaluated with DTPA GFR as measured GFR.

Results: The mean age of the study population was 44.7 yrs with 72.2% being female). The mean measured DTPA GFR was 83.42 ml/min with a range of 61-130 ml/min. The median % absolute difference between the calculated and measured GFR was 19.7, 15.4, 19.3, 20.8 and 25.5, respectively, for CG-CrCl, CG-GFR, MDRD 1, MDRD 2 and urine-CrCl. Pearson's correlation between the measured and estimated GFR varied from 0.09-0.27. The precision as reflected by R2 value was 0.05 for CG-CrCl and CG-GFR, 0.06 for MDRD 1 and MDRD 2 and 0.01 for urine-CrCl. The bias was -14.14, 1.46, 11.89, 17.70 and -2.80 for CG-CrCl, CG-GFR, MDRD 1, MDRD 2 and urine-CrCl, respectively. The accuracy within 30% was 71.3, 85, 86, 76 and 69% for CG-CrCl, CG-GFR, MDRD 1, MDRD 2 and urine-CrCl, respectively.

Conclusions: Our results from a healthy Indian population suggest that of all the predictive equations, MDRD 1 and MDRD 2 were the most precise, MDRD 1 the most accurate and CG-GFR the least biased. However, the poor correlation and error level exhibited by these equations makes them sub-optimal for clinical use.