Meir J. Stampfer, Jaquelyn L. Jahn and Peter H. Gann
Affiliations of authors: Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA (MJS, JJ); Departments of Epidemiology and Nutrition, Harvard School of Public Health, Boston, MA (MJS); Department of Pathology, University of Illinois at Chicago, Chicago, IL (PG).
Correspondence to: Meir J. Stampfer, MD, DrPH, Harvard School of Public Health and Brigham and Women’s Hospital, Departments of Epidemiology and Nutrition, Channing Division of Network Medicine, 677 Huntington Ave, 3rd Fl, Boston, MA 02115 (e-mail: email@example.com).
Prostate-specific antigen (PSA) screening remains a focus of intense controversy, despite decades of clinical experience and several randomized trials. In this issue of the Journal, Stattin et al. (1) take advantage of a natural experiment in Sweden and use sophisticated statistical modeling techniques to provide further persuasive evidence for the benefits of PSA-based detection to reduce prostate cancer–specific mortality. Sweden has not endorsed a PSA screening program, but considerable “opportunistic” PSA screening has caused an apparent increase in rate of overall prostate cancer detection. This increase, of course, is primarily due to the identification of latent tumors, which are age-related, and in autopsy studies, are prevalent in about 36% of US and European men aged 70 to 79 years (2). Stattin et al. (1) used this effect of PSA screening to identify counties in Sweden with higher rates of PSA screening by categorizing the counties according to the increase in overall prostate cancer rates in recent decades after introduction of the PSA test. As expected, these high-incidence counties had greater usage of PSA testing, and the case patients had characteristics tending to be slightly more typical of screened populations, with somewhat lower levels of PSA at diagnosis and earlier-stage disease (3). Over the time period studied, the authors noted a 19% reduction in prostate cancer–specific mortality in the counties with more PSA screening as compared with less.
The authors note that this is not a randomized trial, and the conclusions, as in any observational study, rest on the assumption that the control counties with lower screening are truly comparable and represent the counterfactual—what would …