The Receiver Operating Characteristics Curve in the Evaluation of a Random Urine Specimen as a Screening Test for Diabetic Nephropathy

• DOI: 10.2337/diacare.20.4.516
• Publication Date: 2007-03-05T22:51:27Z
• Authors: Themis Zelmanovitz; Jorge L Gross; Jarbas R Oliveira; Alexandre Paggi; Mariana Tatsch; Mirela J Azevedo
• Keywords: Adult; Male; Urinary albumin concentration; Diabetic nephropathy; Sensitivity and Specificity; Statistics, Nonparametric; Specimen Handling; Random Allocation; 03 medical and health sciences; Diabetes mellitus; 0302 clinical medicine; Reference Values; Albuminuria; Humans; Mass Screening; Diabetic Nephropathies; Aged; Reproducibility of Results; Middle Aged; 6. Clean water; Albuminas; Nefropatias; Diabetes Mellitus, Type 2; Creatinine; Creatinina; Female; Urinary albuminto- creatinine ratio

OBJECTIVE To assess the performance of measurements of urinary albumin concentration (UAC) and urinary albumin:creatinine ratio (UACR) in a diurnal random urine specimen (RUS) for the screening of diabetic nephropathy.

RESEARCH DESIGN AND METHODS A total of 95 ambulatory NIDDM patients (49 women, ages 40–75 years) collected 123 RUSs during the morning after completing a timed 24-h urine collection. Albumin was measured by immunoturbidimetry. According to timed urinary albumin excretion rate (UAER) measured in the 24-h collection (criterion standard), samples were classified as normoalbuminuric (UAER < 20 μg/min; n = 54), microalbuminuric (UAER 20–200 μg/min; n = 44), and macroalbuminuric (UAER > 200 μg/min; n = 25). The receiver operating characteristics (ROC) curve approach was used. The ROC curves of UAC and UACR in RUS for screening of microalbuminuria (normo- and microalbuminuric samples; n = 98) and macroalbuminuria (micro- and macroalbuminuric samples; n = 69) were plotted.

RESULTS Spearman's coefficients of correlation of 24-h UAER vs. UAC and UACR were 0.91 and 0.92, respectively (P < 0.001). The calculated areas (± SE) under the ROC curves to screen microalbuminuria for UAC (0.9766 ± 0.015) and UACR (0.9689 ± 0.014) were similar (P > 0.05) as were the corresponding areas for macroalbuminuria (0.9868 ± 0.0094 and 0.9614 ± 0.0241, respectively; P > 0.05). The first point with 100% sensitivity and the point of intersection with a 100%-to-100% diagonal for microalbuminuria were as follows: 16.9 and 33.6 mg/l for UAC and 15.0 and 26.8 mg/g for UACR; for macroalbuminuria 174.0 and 296.2 mg/l for UAC and 116.0 and 334.3 mg/g for UACR, respectively.

CONCLUSIONS Albumin measurements (UAC and UACR) in an RUS presented almost perfect accuracy for the screening of micro- and macroalbuminuria and UAC measured in an RUS is simpler and less expensive than UACR and UAER. It is suggested as a valid test for use in screening for diabetic nephropathy.