The failure of glomerular filtration rate estimating equations among obese population

Obesity is a major public health with increasing numbers of obese individuals are at risk for kidney disease. However, the validity of serum creatinine-based glomerular filtration rate (GFR) estimating equations in obese population is yet to be determined.

We evaluated the performance of the reexpressed Modification of Diet in Renal Disease (MDRD), reexpressed MDRD with Thai racial factor, Thai estimated GFR (eGFR) as well as Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations among obese patients, defined as body mass index (BMI) ≥25 kg/m2 with the reference measured GFR (mGFR) determined by 99mTc-diethylene triamine penta-acetic acid (99mTc-DTPA) plasma clearance method. Serum creatinine levels were measured using standardized enzymatic method simultaneously with GFR measurement. The statistical methods in assessing agreement for continuous data including total deviation index (TDI), concordance correlation coefficient (CCC), and coverage probability (CP) for each estimating equation were compared with the reference mGFR. Accuracy within 10% representing the percentage of estimations falling within the range of ±10% of mGFR values for all equations were also tested.

A total of 240 Thai obese patients were finally recruited with mean BMI of 31.5 ± 5.8 kg/m2. In the total population, all eGFR equations underestimated the reference mGFR. The average TDI values were 55% indicating that 90% of the estimates falling within the range of -55 to +55% of the reference mGFR. The CP values averaged 0.23 and CCC scores ranged from 0.75 to 0.81, reflecting the low to moderate levels of agreement between each eGFR equation and the reference mGFR. The proportions of patients achieving accuracy 10% ranged from 23% for the reexpressed MDRD equation to 33% for the Thai eGFR formula. Among participants with BMI more than 35 kg/m2 (n = 48), the mean error of all equations was extremely wide and significantly higher for all equations compared with the lower BMI category. Also, the strength of agreement evaluated by TDI, CCC, and CP were low in the subset of patients with BMI ≥35 kg/m2.

Estimating equations generally underestimated the reference mGFR in subjects with obesity. The overall performance of GFR estimating equations demonstrated poor concordance with the reference mGFR among individuals with high BMI levels. In certain clinical settings such as decision for dialysis initiation, the direct measurements of GFR are required to establish real renal function among obese population.