We question the relevance of the study by Lachs and colleagues [1] to the issue of spectrum bias and the predictive value of the dipstick test for diagnosis of urinary tract infection (UTI) because it is based on a misinterpretation of the utility of the leukocyte esterase test, a surrogate marker for pyuria. Pyuria, although usually present in patients with UTI, is not specific for bacterial infection but rather indicates an inflammatory process for which one of many conditions may be responsible. Thus, the value of a positive leukocyte esterase test should be judged not on its ability to detect UTIs but on the additional information it provides in evaluating the clinical picture.

To determine sensitivity and specificity, there must be a "gold standard" for the disease state—in this case, the bacterial count. Judgments of number, degree, and type of symptoms present were used to categorize patients as having a high or low probability of UTI. The sensitivity and specificity of the leukocyte esterase test differed for these two groups of patients. With spectrum bias, sensitivity and specificity will differ according to variations in clinical presentation of a particular disease. The varying probability of UTI does not necessarily reflect different disease manifestations but the presence of symptoms indicating a higher probability of other diseases. In this case, sensitivity and specificity are lower in certain groups not because of variances in performance across a clinical spectrum of a single disease but rather because of confounding by other diseases.

We believe that it is inappropriate to try to determine sensitivity, specificity, and predictive value for a "gold standard" (bacteriuria) in a cohort with diverse clinical presentations when the screening test surrogate (pyuria) is known to have independent predictive value for other conditions.