I have come across a study which presents in itself two, “identical” RCTs (one in one country, one in another). I’m a little curious as to why this would be beneficial to conduct two identical trials instead of a single, larger trial. Each “trial” is multicenter, so unless there is a regulatory burden that makes multinational studies too tricky, I am curious as to what benefits accrue in this approach.
Here’s what I think is actually true. If different groups all do different experiments, then the variety of apparatus/setups will help average out bias in the measurement error. In this context the several replications are an improvement on the alternative where a single group uses a single apparatus/setup to do a single experiment with a large number of replications.
Senn dedicates much of a chapter of Statistical Issues in Drug Development to this topic (note that he is not advocating the practice, merely discussing why it seems to be commonplace). It does seem to be more of a regulatory-driven practice than rooted in statistical justification, though perhaps in some cases there’s a legitimate reason to believe the treatment’s effectiveness will actually vary according to locale. That still isn’t really a great reason to advertise as “2 separate trials” but it may be “why” this is done sometimes.
I once reviewed a paper for a journal that was a pooling of 2 separate non-inferiority trials done in different countries. The first had already been reported independently and had successfuly met its noninferiority target. The second had never been reported independently - instead, the authors were pooling the results of the second trial with the first and submitting them for publication. A very close review of the data made it quite apparent that most likely the second trial had failed to meet the non-inferiority conclusion (quite badly, in fact), and I suspected that the authors were trying to be coy about this by publishing the second trial as part of a pooled analysis rather than publishing the second trial as a standalone publication and then reporting the pooled analysis afterwards. This feels a bit like silly semantics - why not simply look at the totality of the data? - but I felt that by registering the two trials separately and reporting the first one’s results, they were committing to the identity as “two separate trials” and were obligated to report the results of each independently, even if they wanted to pool the data in the end,