Randomized clinical trials enjoy the view from the top of the hierarchy of "truth seeking epidemiologic mechanisms". Medical students are taught that the randomization process serves to immunize the study against the dreaded selection biases. Clinicians generally feel more intellectually confident in reading about a RCT.
Case control studies are another matter.The vagaries of analyzing potential confounders in case control studies elude many physicians and our distrust of case control studies is intensified by the number of dueling or contradictory non -randomized studies over which we have puzzeled.I never get tired of mentioning the juxtaposed, non-randomized, 1985 NEJM articles in which we were told by one group of prestigious researchers that post-menopausal hormone replacement therapy decreased the risk of coronary heart disease (by about one half) and another group of researchers, from an equally prestigious department, told us that the risk increased by a factor of two.These were not rookie epidemiologists or young aspirating medical authors with freshly minted MPH degrees data dredging to bolster their CVs. We are talking about Harvard and the Framingham study.A well respected epidemiologist, Dr. John C. Bailar, has commented that either study taken alone would have been convincing. An editorial by him in that issue seemed to conclude that both articles seemed to be sound in their methodology and that a reason(s) for the discrepancy was not apparent.Bailar suggests that observational studies are subject to a great deal more variation than is usually captured by the statistical tests that are used and that differences such as these may well be due to confounders that are either unrecognized or have larger effects that anyone recognized.
So it is with no surprise that an editorial in the Jan.9,2005 issue of the archives of Internal Medicine ( The PSA Conundrum, Arch. Intern Med/vol 166,Jan 9,2006 pg 7-8)struggles to explain why one case control study concluded PSA screening is efficacious and another concludes it is not. I certainly do not feel competent to assess the relative merits of varying methodological approaches in conflicting case-control studies and I doubt if most physicians do either let alone have the time or energy to attempt to do so. Dr. Michael J. Barry , the editorialist tells us 78% of male primary care physicians and 95% of male urologists reported having a PSA on themselves suggesting physicians seem to believe that on balance PSA screening is a good thing. I am not sure if it is or not but I am sure that another case control study purporting to show either positive or negative results is not going to convince many physicians to change their minds.Two large randomized trials are underway.There is the PLCO screening trial in the U.S. and the ERSPC trial in Europe addressing the issue of PSA screening.Results from both are due in 2009.Let us hope that the two trials point in the same direction.