Application of Glomerular Filtration Rate Prediction Formulae in Diabetic Patients

Objective To compare the predictive performances of glomerular filtration rate (GFR) prediction formulae based on serum crea level in diabetic patients and to determine their clinical application values. Methods Fasting serum crea of 125 diabetic patients in Chengdu were detected.^99mTc-DTPA clearance rate was used as the reference test (rGFR). The eGFR was obtained using different formulae (aMDRD and CKD-EPI formula in this study). The bias, precision, accuracy and diagnostic sensitivity of the two formulae were compared. ROC curve analyses were performed with different diagnostic boundary pointsrGFR<90 or <60 mL/(min·1.73²). Results Significant differences between eGFR and rGFR were found with both formulae (P<0.05). There were no significant differences in the predict performance between the two formulae in terms of bias, precision and accuracy within 30% and 50%. Two cases with higher than 90 mL/(min·1.73²) rGFR were identified as outliers using aMDRD formula according to the eGFR-rGFR scatterplot analysis. The ROC curve analysis showed that the AUC were in the range between 0.8 and 0.9. The AUC approached 0.9 when proteinuria was taken into consideration for both formulae, and this involved a significant improvement (P<0.05). Conclusion Both aMDRD and CKD-EPI formulae predict eGFR accurately in diabetic patients. However, the CKD-EPI formula may have better stability in predicting eGFR in patients at an early stage of renal injury. Combined use of proteinuria may improve the diagnostic performance of eGFR prediction formulae.