Research training | The teaching of more established research methods in observational epidemiology, randomised controlled trials and meta-analysis in postgraduate training could be complemented with other mixed-method and natural experimental approaches more often taught and used in the social sciences |
Research methodology | Efforts to develop consensus and guidance on the appraisal of natural experimental studies could be expanded, complementing the current emphasis on internal validity with greater consideration of external validity, transferability and utility for informing action |
Research funding | Research funding bodies and their peer reviewers could assess natural experimental studies more closely on their own merits rather than using templates based too closely on the expectations of a typical randomised controlled trial |
Academic publishing | Journals could adopt editorial policies committed to selecting manuscripts based more on the applicability of a given study design to a given research question than on prior assumptions about a hierarchy of study design |
Policymaking | Policymakers could allow more time (and assign funding, if appropriate) to enable adequate theorisation, robust study design and baseline data collection to be undertaken before new policies and other interventions are implemented |
Knowledge exchange | Policy and research communities could establish horizon-scanning or intelligence-sharing networks to bring implementers and potential evaluators into dialogue as early as possible in the process of establishing new interventions |