Abstract
AIM: Reduced kidney function is a known risk amplifier for atherosclerotic cardiovascular disease (ASCVD) and adverse kidney events. Accurate assessment of kidney function using estimated glomerular filtration rate (eGFR) is therefore essential for evaluating ASCVD risk and kidney prognosis. We aimed to compare the revised 2021 Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) [2021-eGFRcr(AS)] and European Kidney Function Consortium (EKFCcr) with the 2009 CKD-EPI [2009-eGFRcr(ASR)] equations in predicting the risk of hospitalisations for acute myocardial infarction (AMI), acute kidney disease (AKD), and chronic kidney disease (CKD) in a multi-ethnic Asian cohort. METHODS: This was a multi-centre, retrospective cohort study of adults who attended the ambulatory clinics at the Singapore General Hospital and SingHealth Polyclinics. Individuals were included if they had at least one serum creatinine and albuminuria result in 2014 and at least one follow-up visit between 2015 and 2018. Demographic data, comorbidities, biochemistry, and hospitalisations were retrieved from electronic medical records. eGFR was calculated using the 2009-eGFRcr(ASR) and 2021-eGFRcr(AS) and EKFCcr equations. Multivariable logistic regression models for the associations between eGFR categories and hospitalisations for AMI, AKD, and CKD were evaluated for their goodness-of-fit and discrimination. RESULTS: Among 10,137 individuals in the study, the mean age was 65.5 (10.8) years. The mean eGFRs were 85.6 (20.4), 89.3 (20.0), and 79.6 (19.5) mL/min/1.73 m2 according to the 2009-eGFRcr(ASR), 2021-eGFRcr(AS), and EKFCcr equations, respectively. Compared to the 2009-eGFRcr(ASR) equation, 28.8-33.3% of individuals were reclassified to a less severe eGFR category by the 2021-eGFRcr(AS) equation, while 1.6%-36.6% were reclassified to a more severe eGFR category by the EKFCcr equation. Over a mean follow-up of 44.9 (12.6) months, hospitalisations for AMI, AKD, and CKD occurred in 42 (0.4%), 228 (2.4%), and 189 (2.0%) of patients, respectively. More severe eGFR categories were independently associated with all the outcomes. For hospitalisation for CKD, the model with the 2021-eGFRcr(AS) equation had significantly better discrimination (area under the receiver operating characteristic curve difference +0.010 (p = 0.016) and better fit (Vuong Z statistic = -2.175, p = 0.015) compared to the model with the 2009-eGFR(ASR). However, the discrimination and fit of models for predicting AMI and AKI hospitalisations were similar between 2021-eGFRcr(AS) and 2009-eGFR(ASR) equations were similar. The EKFCcr-based models did not demonstrate improved discrimination or fit for hospitalisation for AMI, AKD, and CKD, compared to 2009-eGFRcr(ASR)-based model. CONCLUSION: Lower eGFR ascertained by the 2009-eGFRcr(ASR), 2021-eGFRcr(AS), and EKFCcr equations were independently associated with greater risks of hospitalisation for cardiovascular and kidney disease in a multi-ethnic Asian cohort. Adoption of the race-free 2021-eGFRcr(AS) equation improved prediction of hospitalisation for CKD compared to the 2009-eGFRcr(ASR) and was non-inferior in predicting hospitalisation for AMI and AKD. These findings support the use of the 2021-eGFRcr(AS) equation in clinical practice, to predict health service utilisation for hospitalisations for cardiovascular and kidney disease, aligning with global initiatives for race-neutral kidney function evaluation.