Walsh [1] describes two populations in which the prostate-specific antigen (PSA) test is used: an asymptomatic group, in which use of the test constitutes screening, and a symptomatic group, in which use of the test constitutes diagnosis. He appropriately points out that the PSA test is likely to have a different yield in patients with complaints due to prostatic enlargement than in asymptomatic persons. Patients with symptoms, however, can present in one of two ways and, by so doing, place themselves into one of two groups: a case-finding group or a diagnosis group. Case finding [2] describes a patient who goes to the physician for one reason (for example, for evaluation of headaches) and on review of systems responds positively to questions designed to elicit evidence of a different problem (for example, urinary obstruction). Diagnosis describes the evaluation of the primary problem (for example, difficulty in urinating). Dr. Walsh does not mention a case-finding group in his analysis.

The prevalence of clinically important prostate cancer is likely to be higher in the diagnosis group than in the case-finding group, and, because of this, the positive predictive value of PSA will differ. In addition, the sensitivity and specificity of PSA are likely to be higher in the diagnostic group than in the case-finding group because of the phenomenon of spectrum bias, which explains how the sensitivity and specificity of a test can be substantially higher in a group of patients with more advanced disease [3]. Patients in the diagnosis group are likely to have more advanced disease.

Because of the probability that PSA will have different test characteristics in all three populations (screening, case-finding, and diagnosis), further evaluation of this test in all three groups may be profitable.