How are randomized clinical trials ethically justified? A systematic scoping review and thematic analysis of reasons...

We set out to identify and count the types of reasons that are used in contemporary scholarship about the ethical permissibility of randomized trials, with the goal of developing a finer grained taxonomy of reasons than what is currently used by most participants in this literature. Because of its central role in justifying normative conclusions about randomized clinical trials (RCTs), we paid particular attention to both uses of the keyword "equipoise" and to the different concepts associated with it.

We conducted a scoping review to identify articles that included arguments that were likely to express reasons justifying RCTs. Text excerpts that expressed reasoning about the ethical permissibility of RCTs were extracted from relevant papers, and our data were generated by coding these excerpts using a mixed-methods protocol that fused elements of a grounded analysis and thematic coding. In our study, each theme corresponded to a specific type of reason that was contentful and stable when applied to our corpus of text extracts.

Our search, screening, and text extraction process yielded 1,335 unique text excerpts, which then formed the basis of our coding. Although we found that 16 themes were sufficient to saturate this corpus, slightly less than 100% of our excerpts were covered by just 10 themes. We also tracked uses of 16 keywords in the text excerpts to explore whether there was any relationship between the keywords and our themes and found that keywords frequently did not cooccur with the presence of our themes.

Our data and analysis support the conclusion that there is significant diversity in the types of reasons offered to justify RCTs; 10 themes effectively captured all the text excerpts we analyzed, and these themes cannot be reduced to the occurrence of relevant keywords. This result highlights how individuals and organizations may use different reasons to consider randomized trials to be justified and even when they use similar language the concepts they are referencing may not be consistent.