Furthermore, by using broad enrolment criteria, investigators might have included a very mixed population, only some of whom might be responsive to the intervention, thereby lowering the signal-to-noise ratio to a point where potentially valuable treatments have not been identifi ed.