In Reply: I agree with the comments from Drs Bosland and McCormick regarding the importance of null results from preclinical animal models and support their suggestion for reducing publication bias by increasing journal space devoted to well-conducted null studies.
Dr Pinsky is correct that the exclusion from SELECT of men with a prior diagnosis of prostate cancer would decrease the expected number of prostate cancer deaths based on general population rates. His estimate of the mortality deficit in SELECT compared with the SEER population is undoubtedly closer to the truth than the back-of-the-envelope calculation in my Editorial. My goal was to draw attention to the extremely low rates of advanced disease and mortality in SELECT compared with the rates one would expect in a similar population with far less PSA screening. In that regard, the most appropriate direct comparison might be the PLCO trial participants with normal baseline PSA and rectal examinations who were assigned to the nonscreening arm.
Gann PH. Antioxidant Supplementation and Cancer Prevention—Reply. JAMA. 2009;301(18):1877-1879. doi:10.1001/jama.2009.630