Box 1 Definitions of key concepts used in this paper

Context

In realist terms, context is any system, structure or condition that affects outcomes, including individuals’ attributes and social interactions [3]

Mechanism

Mechanisms are what makes an intervention work: “They are not the observable machinery of program activities, but the response that interaction with a program activity or resource triggers (or does not trigger) in the reasoning and behaviour of participants” [70]

Process effects

These are proximal impacts that influence intervention outcomes or are of evaluative interest for other reasons (e.g. they help explain unexpected variation in implementation); others use the term ‘formative outcomes’ [84]; Desired process effects are those the investigators consider to be prerequisites for a successful intervention

Programme theory

This is, “An explicit theory or model of how an intervention contributes to a set of specific outcomes through a series of intermediate results” [85]; programme theory should be plausible, useful and consistent with the evidence

Proposition

Propositions are generalised theoretical statements grounded in the data [86]; in realist evaluation, they link and condense information about contexts, mechanisms and outcomes; propositions are refined through empirical testing but remain fallible [87]

Realist process evaluation

Process evaluation helps explain how an intervention had its effects [7]; realist process evaluation applies realist principles to this process and investigates causal patterns (known as demi-regularities) to show how intervention strategies may be operating under what conditions to generate process effects for which groups [3]

Retroduction

This is a form of analysis that “involves constant shuttling between theory and empirical data, using both inductive and deductive reasoning” [88]