Comparative Performance of Four Creatinine-based GFR Estimating Equations
10.13471/j.cnki.j.sun.yat-sen.univ（med.sci）.2022.0413
- VernacularTitle:验证四种基于肌酐的肾小球滤过率估算方程的性能
- Author:
Pei-jia LIU
1
;
Hong-quan PENG
2
;
Xing-hua GUO
3
;
Lei-le TANG
4
;
Shao-min LI
1
;
Jia FANG
1
;
Xun LIU
1
Author Information
1. Department of Nephrology， The Third Affiliated Hospital of Sun Yat-sen University， Guangzhou 510630， China
2. Department of Nephrology， Kiang Wu Hospital， Macau 999078， China
3. Department of Rheumatology， The Third Affiliated Hospital of Sun Yat-sen University， Guangzhou 510630， China
4. Department of Cardiology， The Third Affiliated Hospital of Sun Yat-sen University， Guangzhou 510630， China
- Publication Type:Journal Article
- Keywords:
estimted glomerular filtration rate （eGFR）;
predictive performance;
chronic kidney disease epidemiology collaboration （CKD-EPI） equation;
Xiangya equation;
European kidney function consortium （EKFC） equation
- From:
Journal of Sun Yat-sen University(Medical Sciences)
2022;43(4):621-630
- CountryChina
- Language:Chinese
-
Abstract:
ObjectiveTo assess the predictive performance of four creatinine-based equations for estimated glomerular filtration rate （eGFR）： 2012 chronic kidney disease epidemiology collaboration （CKD-EPIcr） equation ， 2021CKD-EPIcr equation， Xiangya equation and European kidney function consortium （EKFC） equation. MethodsA total of 198 patients with chronic kidney disease from the Third Affiliated Hospital of Sun Yat-sen University and the Kiang Wu Hospital in Macau were enrolled. We compared the GFR measured （mGFR） by iohexol plasma clearance and the eGFR calculated by four equations. The agreement between mGFR and eGFR was analyzed by Bland-Altman plots， concordance correlation coefficient （CCC）， coverage probability （CP） and total deviation index （TDI）. The performance of eGFR equations， including their bias， precision， root square mean error （RSME）， and percentage of estimates within 30% deviation of measured GFR （P30）， were evaluated. Bootstrap method （2 000 samples） was used to calculate bias， interquartile range （IQR）， RSME， and 95% confidence intervals （CI） for P30. After selecting the optimal eGFR equation as the reference， we statisticlly tested other equations by ① Wilcoxon signed-rank test for bias； ② McNemar-Bowker test for P30； ③ comparing RMSE and IQR with independent samples t test after 2 000 bootstrap samples were obtained. ResultsThe median mGFR and four eGFR equations （EKFC， 2012CKD-EPIcr， 2021CKD-EPIcr and Xiangya equation） in the overall population were 56.2 mL·min-1·（1.73m2）-1， 67.1 mL·min-1·（1.73m2）-1， 73.0 mL·min-1·（1.73m2）-1， 66.9 mL·min-1·（1.73m2）-1 and 63.8 mL·min-1·（1.73m2）-1， respectively. The Bland-Altman plots showed that EKFC equation had the lowest mean difference and the narrowest 95% limit of agreement. The EKFC equation had the optimal performance on CCC， TDI and CP with values of 0.90， 24.41 and 0.50， respectively. Overall， the bias， accuracy， P30 and RSME from the EKFC equation was -0.99， 14.64， 0.80， and 14.68， respectively， with 95% CI ranging from -2.53 to 0.94， 11.82 to 17.35， 0.73 to 0.85， and 12.69 to 17.35， respectively， which were superior to those values from other three eGFR equations. The differences were statistically significant （all P < 0.05）. The results in the mGFR subgroups were basically consistent with the overall trend. ConclusionsOf the four eGFR equations validated in this study， the EKFC equation comprehensively surpasses 2012CKD-EPIcr equation， 2021CKD-EPIcr equation， and Xiangya equation. With P30>75%， the EKFC equation can meet clinical diagnostic needs. Therefore， the EKFC equation is recommended for estimating GFR in a Chinese population， but more participants need be included to further support this conclusion.
