A Systematic Review Evaluating Estimated GFR Performance in South Asian Populations

Estimated glomerular filtration rate (eGFR) equations are widely used to measure kidney function; however, their performance in South Asian populations is variable. We performed a systematic review and meta-analysis to assess the performance of creatinine-based (eGFR_Cr), cystatin-C-based (eGFR_CysC) and combined (eGFR_Cr-CysC) eGFR equations compared with measured GFR (mGFR) in South Asians.

Pubmed/Medline, Embase (Ovid), Scopus, Web of Science, and Cochrane Library for Systematic Reviews and Clinical Trials were searched for relevant studies between January 1, 1976 and November 30, 2024. Eligible studies compared eGFR to mGFR in South Asian adult populations. Meta-analyses were performed to assess bias (standardized mean difference [SMD]) and 30% accuracy (meta proportion/percentage of eGFR results within 30% of the corresponding mGFR result [P30]) for eGFR equations compared with mGFR with multivariable meta-regression for key moderators.

Thirty-five cohorts (n = 4725) were included. eGFR_Cr equations overestimated mGFR with the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI-2009_Cr) equation showing the highest standardised mean difference (0.66; 95% confidence interval [CI]: 0.35-0.97). eGFR_CysC had the least standardized mean difference (-0.03 to 0.15). However, accuracy for eGFR_Cr and eGFR_CysC equations were < 75%. eGFR_Cr-CysC had the best P30, ranging from 79% to 82%. Meta-regression identified age, sex, ethnicity, and cohort type as having significant impacts on bias effect size.

eGFR equation accuracy in South Asians is suboptimal, consistently below the accepted clinical P30 threshold of 75%. eGFR_CysC equations had reduced bias compared with eGFR_Cr or eGFR_Cr-CysC; however, accuracy was better with eGFR_Cr-CysC. Emerging equations, such as the CKD-EPI-Pakistan variant and European Kidney Function Consortium (EKFC) equations warrant further investigation in diverse South Asian cohorts with population-specific calibration.