Failure of prediction of liver function test abnormalities with the urine urobilinogen and urine bilirubin assays

A prospective observational study of 229 cases was conducted in a busy ambulatory care setting to evaluate the sensitivity, specificity, predictive values, and accuracy of spot urine urobilinogen and urine bilirubin assays as screening tests for serum liver function test (LFT) abnormalities. Both urine tests exhibited remarkably similar characteristics overall once they were adjusted to maximize accuracy and predictive values (occurring at a normal or abnormal "threshold," respectively, of 3.4 or 5.07 mumol/d for urobilinogen and 0 or 1+ for urine bilirubin). The percentage of cases correctly identified were 81% to 83% for serum bilirubin assays, 68% to 72% for other LFTs, but only 62% to 63% for screens for cases with at least one abnormal LFT finding. Poor sensitivities (47% to 49%) limited the detection of abnormal findings by the screen; both screens were reasonably specific (79% to 89%), but negative predictive values were suitable (89%) for serum bilirubin results only and were prohibitively lower (49% to 50%) in predicting all patients without LFT abnormalities. We conclude that spot urine urobilinogen and urine bilirubin determinations, although good screens for isolated serum bilirubin elevations, have unacceptable statistical properties as predictors of other LFT results due to a high proportion of false-negative results.