Diagnostic accuracy of urine dipsticks for detection of albuminuria in the general community

Background: Urine dipsticks, an inexpensive accessible test for proteinuria, are widely advocated for mass screening; however, their diagnostic accuracy in the general community is largely unknown. Study Design: Evaluation of diagnostic test accuracy in a cross-sectional cohort. Setting & Participants: AusDiab, a representative survey of Australian adults 25 years and older (conducted in 1999/2000). Stratified cluster random sampling from 11,247 individuals participating in the biomedical examination; complete urinalysis data available for 10,944. Index Test: Urine dipsticks (Bayer Multistix), with a positive result defined as <1+ or trace or higher protein. Reference Test: Albumin-creatinine ratio (ACR), measured on a random spot urine sample. Reference test positivity was defined as ACR <30 mg/g or ACR <300 mg/g. Results: Numbers of participants with ACR <30, 30-300, and <300 mg/g were 10,219 (93.4%), 634 (5.8%), and 91 (0.8%), respectively. The area under the receiver operating characteristic curve (AUC) for dipstick detection of ACR <30 mg/g was 0.8451 ± 0.0129 (SE) in men and 0.7775 ± 0.0131 in women (P < 0.001). The AUROC for dipstick detection of ACR <300 mg/g was 0.9904 ± 0.0030 in men and 0.9950 ± 0.0016 in women (P = 0.02). Dipstick result <1+ identified ACR <30 mg/g with 57.8% sensitivity (95% CI, 54.1%-61.4%) and 95.4% specificity (95% CI, 95.0%-95.8%) and identified ACR <300 mg/g with 98.9% sensitivity (99% CI, 92.1%-100%) and 92.6% specificity (99% CI, 92.0%-93.3%). A dipstick result of trace or higher identified ACR <30 mg/g with 69.4% sensitivity (95% CI, 65.9%-72.7%) and 86.8% specificity (95% CI, 86.1%-87.4%) and identified ACR <300 mg/g with 100% sensitivity (99% CI, 94.3%-100%) and 83.7% specificity (99% CI, 82.8%-84.6%). A negative dipstick result (less than trace) had a negative predictive value of 97.6% (95% CI, 97.2%-97.9%) for ACR <30 mg/g and a negative predictive value of 100% (99% CI, 99.9%-100%) for ACR <300 mg/g. The probability of an ACR <30 mg/g confirmed on laboratory investigation was 47.2% (95% CI, 43.9%-50.5%) based on a dipstick result <1+ and 27.1% (95% CI, 25.1%-29.2%) based on a trace or higher result. Limitations: Isolated urine samples precluded assessment of test reproducibility. Urine specific gravity and pH were not recorded; therefore, the effect of urine concentration on test performance was not assessed. Conclusions: A dipstick test result <1+ or less than trace has a high negative predictive value in the general community setting, with minimal risk of a missed diagnosis of macroalbuminuria. High false-positive rates emphasize the need for laboratory confirmation of positive results.