Evaluation of the diagnostic performance of the creatinine‐based Chronic Kidney Disease Epidemiology Collaboration equation in people with diabetes: A systematic review

• Published in: Diabetic medicine 2021-01, Vol.38 (1), p.e14391-n/a
• Main Authors: Zafari, N., Churilov, L., Wong, L. Y.‐L., Lotfaliany, M., Hachem, M., Kiburg, K. V., Kong, L., Torkamani, N., Baxter, H., MacIsaac, R. J., Ekinci, E. I.
• Format: Article
• Language: English
• Subjects: Bias; Collaboration; Creatinine; Creatinine - analysis; Diabetes; Diabetes mellitus (insulin dependent); Diabetes mellitus (non-insulin dependent); Diabetes Mellitus, Type 2 - epidemiology; Epidemiology; Glomerular Filtration Rate; Humans; Kidney diseases; Kidney transplantation; Medical diagnosis; Organ donors; Quality control; Renal Insufficiency, Chronic - diagnosis; Renal Insufficiency, Chronic - epidemiology; Systematic review
• ISSN: 0742-3071; 1464-5491
• DOI: 10.1111/dme.14391

Aims GFR estimated with the creatinine‐based Chronic Kidney Disease Epidemiology Collaboration (CKD‐EPICr) equation is used to screen for diabetic kidney disease and assess its severity. We systematically reviewed the process and outcome of evaluating CKD‐EPICr in estimating point GFR or GFR decline over time in adults with type 1 or type 2 diabetes. Methods In this systematic review, MEDLINE, Embase and Cochrane Central Register of Controlled Trials were searched up to August 2019. Observational studies comparing CKD‐EPICr with measured GFR (mGFR) in adults with diabetes were included. Studies on people with kidney transplant, non‐diabetes related kidney disease, pregnancy, potential kidney donors, and those with critical or other systematic illnesses were excluded. Two independent reviewers extracted data from published papers and disagreements were resolved by consensus. Risk‐of‐bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies‐2 tool. (PROSPERO registration number: CRD42018108776). Results From the 2820 records identified, 29 studies (14 704 participants) were included. All studies were at risk of bias. Bias (eight different forms) ranged from −26 to 35 ml min−1 1.73 m−2; precision (five different forms) ranged between 9 and 63 ml min−1 1.73 m−2; accuracy (five different forms) ranged between 16% and 96%; the correlation coefficient between CKD‐EPICr and mGFR (four different forms) ranged between 0.38 and 0.86; and the reduced major axis regression slope ranged between 0.8 and 1.8. Conclusions Qualitative synthesis of data suggested CKD‐EPICr was inaccurate in estimating point GFR or GFR decline over time. Furthermore, a lack of consistency in the methods and processes of evaluating the diagnostic performance of CKD‐EPICr limits reliable quantitative assessment. The equation needs to be improved in adults with diabetes.