THE EPISTEMOLOGICAL WEIGHT OF RANDOMIZED-CONTROLLED TRIALS DEPENDS ON THEIR RESULTS

摘要

When evaluating the internal validity of clinical trials, physicians and medical researchers place substantial importance on factors such as blinding and randomization. For a particular randomized-controlled trial (RCT), causal inference and explanation require additional contextual considerations, to which we assign the term "epistemological weight." We argue that one component of epistemological weight, epistemological function, depends on a trial's results. We further note that discordant RCTs often lead to hypothesis generation, an epistemological function that is rare among concordant studies. As an example of methodologically sound but discordant studies, we explore the issue of hormone-replacement therapy and cardiovascular outcomes.