A total of 612 elderly patients with a measured glomerular filtration rate (mGFR) by the dual plasma sample clearance method with Technetium-99 m-diethylenetriamine-pentaacetic acid (Tc-99 m-DTPA) were divided into four subgroups based on age, sex, mGFR, and whether combined with diabetes. The performance of GFR was assessed while considering bias, precision, accuracy, and root-mean-square error (RMSE). Bland-Altman plots, concordance correlation coefficients (CCCs), and correlation coefficients were applied to evaluate the validity of eGFR.
The median age of the 612 participants was 73 years, and 386 (63.1%) were male. Referring to mGFR (42.1 ml/min/1.73 m2), the CKD-EPI, LMR, BIS1, FAS, and EKFC equations estimated GFR at 44.4, 41.1, 43.6, 41.8 and 41.9 ml/min/1.73 m2, respectively. Overall, the smallest bias was found for the BIS1 equation (- 0.050 vs. range - 3.015 to 0.795, P<0.05, vs. the CKD-EPI equation). Regarding P30, interquartile range (IQR), RMSE, and GFR category misclassification, the BIS1 equation generally performed more accurately than the other eqs. (73.9%, 12.7, 12.9, and 35.3%, respectively). Nevertheless, no equation achieved optimal performance for the mGFR≥60 ml/min/1.73 m2 subgroup. Bland-Altman analysis showed the smallest mean difference (- 0.3 ml/min/1.73 m2) for the BIS1 equation when compared to the other equations.
This study suggested that the BIS1 equation was the most applicable for estimating GFR in Chinese elderly patients with moderate to severe renal impairment.