The unrecognized prevalence of chronic kidney disease in diabetes

Background. Diabetes mellitus and chronic kidney disease (CKD) are common and exhibit synergistic associations with premature mortality. Current diabetes guidelines in the UK recommend annual urinary albumin and serum creatinine determinations to screen for diabetic kidney disease. The aim of this study was to estimate the burden of CKD in patients with diabetes and examine the ability of serum creatinine and albuminuria to detect clinically meaningful CKD compared with estimated glomerular filtration rate (eGFR). Methods. All adults known to have diabetes in primary and secondary care in Salford, UK, alive with independent renal function on 1 January 2004 were included in this observational study (n = 7596). Demographic and laboratory parameters were obtained from the Electronic Patient Record. eGFR was determined using the 4-variable modification of diet in renal disease (MDRD) formula. Clinically meaningful CKD was defined as an eGFR < 60 ml/min/1.73 m(2). Results. Creatinine and albuminuria were measured in the preceding 2 years in 82.3 and 55.2% of subjects, respectively. In patients with CKD, normoalbuminuria was present in 48.8%, and serum creatinine was normal (<= 120 mu mol/l) in 54.7%. An abnormal serum creatinine (>= 120 mu mol/l) had a sensitivity and specificity of 45.3 and 100%, respectively, to identify CKD. The combination of abnormal creatinine and albuminuria had an improved performance but still failed to detect a large number with CKD (sensitivity 82.4%, specificity 75.4%). Serum creatinine failed to identify CKD more often in females (OR 8.22, CI 6.56-10.29). Conclusions. Undiagnosed CKD is common in diabetes. Current screening strategies, based on creatinine or albuminuria, fail to identify a considerable number of subjects with CKD. Incorporating eGFR into screening for CKD would identify individuals earlier in the natural history of the disease and enable early effective treatment to delay progression of CKD.