I am puzzled by a phenomenon that I have occasionally noticed in randomized trials in which the number of patients enrolled in a trail seems to be far smaller than the number expected based on epidemiologic information about the prevalence of the disease.
For example, in the study of the efficacy and safety of recombinant human activated protein-C for severe sepsis (Bernard, et.al. NEJM 2001: 344,699-709) the authors state that there are approximately 750,000 cases of sepsis each year in the United States of which at least 225,000 are fatal. Assuming that there are 5,000 hospitals in the United States makes the average number of cases of sepsis at any one hospital about 300 over 2 years.
The study was carried out in 168 hospitals. Granted it was worldwide and not only in the USA. A total of 1,690 patients were randomized in the study. However, during this time period at a rate of 300 patients per 2 years one would expect 50,400 potential patients to have been entered into the study, approximately 4% of the (likely) eligible patients.
Could this be a subtle (or not so subtle) form of selection bias in which the authors/researchers have used some kind of a filter (not described) to find a small population in which the therapy might be more beneficial than if given to all patients with severe sepsis?
I am very curious to hear what others think about this. If you think it is a problem why haven't we (people in general) been talking about it more?
Dan
****************************************************************************
Dan Mayer, MD
Professor of Emergency Medicine
Albany Medical College
47 New Scotland Ave.
Albany, NY, 12208
Ph; 518-262-6180
FAX; 518-262-5029
E-mail; [log in to unmask]
****************************************************************************
|