Testing and refining middle-range theory in evaluations of public-health interventions: evidence from recent systematic reviews and trials.

Evaluations of public-health interventions might potentially be used to test and refine middle-range theory (ie, theory about the mechanisms, which generate outcomes that is analytically generalisable enough to span a range of contexts, interventions or outcomes, but specific enough to be salient in a given application). This approach has been suggested as one means of developing more informed assessments of how different interventions work and whether mechanisms might transfer across contexts. However, we have noticed that studies included in some of our recent systematic reviews are not oriented towards helping test middle-range theory because interventions draw on multiple middle-range theories (so that it is difficult to draw any conclusions about each middle-range theory based on their results) and these middle-range theories are insufficiently clear (with vague constructs) or parsimonious (with too many constructs) to be readily testable. Some studies might in future better contribute to testing and refining middle-range theory via focusing on interventions informed by one middle-range theory and focused on one mechanism at a time. Such 'proof-of-principle' studies should draw on middle-range theory that is sufficiently clear and parsimonious to allow such testing. These evaluations might facilitate more rigorous testing of middle-range theory and hence refinement of scientific knowledge. They might inform broader assessments of how mechanisms transfer across contexts aiding the development of future public-health interventions. Such studies would be a complement not an alternative to pragmatic studies of scalable complex interventions, often informed by more than one middle-range theory.