Objective: To determine the sensitivity, specificity, and positive and negative predictive values of an enzymatic urine screening test for diagnosing bacteriuria in pregnancy.
Methods: Clean-catch midstream urine samples were collected from 383 women who had routine prenatal screening for bacteriuria. Sensitivity, specificity, and positive and negative predictive values for each screening test (enzyme activity, nitrites or leukocytes on dipstick, and bacteria or pyuria on microscopic examination) were estimated using urine culture as the criterion standard. Urine cultures were considered positive if they grew 10(4) colony-forming units of a single uropathogen. Standard deviations used to calculate 95% confidence intervals were based on binomial distribution. A sample of 30 urine specimens was selected to evaluate interrater agreement using Cohen's kappa statistic.
Results: Five of 383 samples were contaminated, leaving 378 samples for evaluation. Thirty of 43 specimens with positive urine culture had positive enzyme activity. Of 335 samples with no growth, 150 had negative enzyme activity. Sensitivity, specificity, and positive and negative predictive values for the Uriscreen enzymatic screening test (Bard Patient Care Division, Murray Hill, NJ) were 70%, 45%, 14%, and 92%, respectively. Sensitivity of the Uriscreen was lower than that of bacteria alone. Interrater agreement for Uriscreen testing was high among the three testers (kappa =.86).
Conclusion: The Uriscreen enzymatic screening test had inadequate sensitivity for rapid screening for bacteriuria in pregnancy.