Beyond HERS: some (not so) random thoughts on randomized clinical trials.

Science is the process of discovering truth, and "truth" is sampled each time we do a study. The results from all of our studies will be distributed around the truth, and different study designs give different amounts and different qualities of sampled material. Truth is ascertained only when sufficient numbers of appropriate studies are conducted, and no one study or one study design has a monopoly on truth. Currently, the randomized clinical trial is considered the penultimate study design and the ultimate test of the hypothesis, but only if it is double-blinded, placebo-controlled, and analyzed by an intention-to-treat protocol. The study design most similar to the randomized controlled trial is the prospective cohort study. In this observational approach, a cohort (group of individuals) is assembled and followed in real time while end points (e.g. breast cancers, heart attacks, fractures) accrue. This is contrasted to the randomized controlled trial, where a group of individuals is assembled, intervened upon, and followed in real time while end points accrue. The major advantage of the randomized controlled trial over an observational study is that the randomization process should eliminate any "bias" in the exposure of interest. However, the randomized controlled trial, like all study designs, has other limitations. Major limitations of the randomized controlled trial include significant financial and other costs, problems with external generalizability, the placebo effect, external monitoring, multi-center differences, and the (frequently problematic) intention-to-treat analysis rule. Many of these limitations do not occur in prospective cohort studies. For example, since a placebo is not administered in an observational study, there is no placebo effect, and since the study is not monitored by a data and safety monitoring board, abrupt truncation of the study duration is not usually seen in observational cohort studies. These limitations of randomized controlled trials are discussed, with specific references to several recently published randomized controlled trials in women (HERS, NSABP P-1, and the Royal Marsden Hospital trials). The HERS trial is significant because despite overwhelming observational evidence that menopausal estrogen therapy prevents heart disease, HERS found no overall difference in heart disease events in women assigned to an estrogen-plus-progestin intervention. The NSABP P-1 and the Royal Marsden Hospital trials are significant in that they were testing the same hypothesis (whether tamoxifen can prevent breast cancer), but came to entirely different conclusions. Two major questions will be posed from this specific review: One: Given conflicting evidence by study design (observational vs. randomized clinical trial), does menopausal estrogen therapy protect against heart disease? Two: Given conflicting evidence within study design (conflicting randomized clinical trials), does tamoxifen prevent breast cancer?