Reproducibility of albuminuria in type 2 diabetic subjects. Findings from the Renal Insufficiency And Cardiovascular Events (RIACE) study

Background. Measurement of urinary albumin excretion (UAE) shows important intra-individual variability suggesting the need for multiple assessments. This study aimed at investigating the reproducibility of UAE in type 2 diabetes. Methods. UAE was obtained from two to three samples collected in a 3- to 6-month period from 4062 of the 15 773 type 2 diabetic subjects participating in the Renal Insufficiency And Cardiovascular Events (RIACE) Italian Multicentre Study in 2007-08. UAE was assessed as albumin excretion rate (AER) in 24-h urine collections from 833 subjects and albumin: creatinine ratio (A/C) in early-morning urine samples from 3229 patients. Albuminuria was measured by immunonephelometry or immunoturbidimetry. Results. The median coefficient of variation (CV) was 32.5% (interquartile range: 14.3-58.9). Concordance rate between a single UAE and the geometric mean of multiple measurements was 94.6% for normoalbuminuria, 83.5% for microalbuminuria, 91.1% for macroalbuminuria and 90.6% for albuminuria (micro + macro). CV was significantly higher (P < 0.01) for AER measurement than for A/C and with immunoturbidimetry than with immunonephelometry, whereas concordance rates were similar between the two modalities of urine collection and the two assay methods. Receiver-operating characteristic (ROC) plots demonstrated a good performance of single UAE in predicting the geometric mean of multiple measures at the cut-off level of both microalbuminuria (ROC(AUC) 0.926; 95% confidence interval: 0.915-0.937) and macroalbuminuria (ROC(AUC) 0.950; 95% confidence interval: 0.927-0.973). Conclusions. Data from this large cohort indicate that, in type 2 diabetic subjects, a single UAE value, thought to be encumbered with high intra-individual variability, is an accurate predictor of nephropathy stage for clinical and epidemiological purposes.