We read with interest the recent article by Jacobsen et al1 on the predictive properties of prostate-specific antigen (PSA) as a screening test for prostate cancer in a population-based setting. This study demonstrates an important point when considering operating characteristics of screening or diagnostic tests when used with different populations. In the urology practice setting, the prevalence of prostate cancer and benign prostate disease (a primary source of false-positive results) is far higher than in the general population. It should be expected that the specificity of the screening test will be greater in the general population, where there are proportionately fewer men with benign disease and more men free of disease. In community-based and primary care settings, the distributions of PSA levels among men with and without prostate cancer are more disparate than in the specialty setting.
We want to draw attention to 2 additional points that are crucial