Table 1. steps in the REAL realist review process.

Step 1a: Locating existing theories, development of initial programme theory	•   Identified 28 papers that were used to identify key causal dynamics and contextual influences on CE, with input from scholar/practitioner experts.    •   Initial programme theory developed at inception workshop with core team and context experts.
Step 1b: (iterative throughout) Conceptual resources	•   Substantive theories and concepts relevant to our programme theory were drawn on to inform development of Context Mechanism Outcome configurations (CMOCs) – realist causal explanations that combine into an overall programme theory.
Step 2: Searching for evidence	•   Our main search focused on malaria research trials.    •   Potential additional searches on other health research paradigms were not conducted due to the volume of literature available from the main search.    •   Citation tracking and drawing on ‘sibling’ and ‘kinship’ papers identified additional papers where our programme theory needed strengthening.
Step 3: Document selection	•   Documents were screened by BA using titles and abstracts and then by full text. 10% random sample were reviewed by RV.    •   Full text documents were selected based on their ability to provide relevant data for the review; including relevant empirical detail or explanatory accounts of engagement outcomes.
Step 4: Data extraction	•   Selected documents were coded in Nvivo 12 1 .    •   Coding frameworks were developed independently by two reviewers and iteratively refined with input from the team.
Step 5: Data analysis and synthesis	•   Working across and within coded data extracts, CMOCs were iteratively developed and refined by two reviewers with input from the wider team.
Step 6: Refine programme theory	•   The overall programme theory was iteratively refined and finalised with input from the review team and content expert advisors, including at a ‘validation workshop’ towards the end of the review