Urinary dipstick proteinuria testing: does automated strip analysis offer an advantage over visual testing?

Abstract

To compare the diagnostic test properties of automated and visually read urine dipstick screening for detection of a random protein:creatinine ratio (PrCr) ≥ 30 mg/mmol. Urine samples were collected prospectively from 160 women attending high-risk maternity clinics at a tertiary care facility. Samples were divided into two aliquots; one aliquot was tested using two different urine test strips, one read visually and one by an automated reader. A second aliquot of the same urine was analyzed for urinary protein and creatinine. Performance of visual and automated dipstick results (proteinuria ≥ 1+) were compared for detection of PrCr ≥ 30 mg/mmol using non-dilute urine samples (urinary creatinine ≥ 3 mmol/L). Both urine test strips showed low sensitivity (visual 56.0% and automated 53.8%). Positive likelihood ratios were 15.0 for visual dipstick testing (95% CI 5.9 to 37.9) and 24.6 for automated (95% CI 7.6 to 79.6). Negative likelihood ratios were 0.46 for visual dipstick testing (95% CI 0.29 to 0.71) and 0.47 for automated (95% CI 0.31 to 0.72). Automated dipstick testing was not superior to visual testing for detection of proteinuria in pregnant women in a primarily outpatient setting. Sensitivity may depend on the test strips and/or analyzer used.