But this may also happen with non-RAR randomisation as well. Small centres that are low recruiters with a different demographic may have different outcomes to larger (metropolitan) centres. Most ICU trials are like this. It rarely gets spoken about, though.

Agree. Yes, every pt should be in a trial. In which case is it incumbent on trialists to construct good trials, adequately powered with interventions that matter. We don't really talk about the paucity of many things. We do keep investigating haloperidol, though.