Background: Health-care systems across the globe are facing increased pressures to balance the efficient use of resources and at the same time provide high-quality care. There is greater requirement for services to be evidence based, but practices that are of limited clinical effectiveness or cost- effectiveness still occur.
Objectives: Our objectives included completing a concept analysis of de-implementation, surfacing decision-making processes associated with de-implementing through stakeholder engagement, and generating an evidence-based realist programme theory of ‘what works’ in de-implementation.
Design: A realist synthesis was conducted using an iterative stakeholder-driven four-stage approach. Phase 1 involved scoping the literature and conducting stakeholder interviews to develop the concept analysis and an initial programme theory. In Phase 2, systematic searches of the evidence were conducted to test and develop this theory, expressed in the form of contingent relationships. These are expressed as context–mechanism–outcomes to show how particular contexts or conditions trigger mechanisms to generate outcomes. Phase 3 consisted of validation and refinement of programme theories through stakeholder interviews. The final phase (i.e. Phase 4) formulated actionable recommendations for service leaders.
Participants: In total, 31 stakeholders (i.e. user/patient representatives, clinical managers, commissioners) took part in focus groups and telephone interviews.
Data sources: Using keywords identified during the scoping work and concept analysis, searches of bibliographic databases were conducted in May 2018. The databases searched were the Cochrane Library, Campbell Collaboration, MEDLINE (via EBSCOhost), the Cumulative Index to Nursing and Allied Health Literature (via EBSCOhost), the National Institute for Health Research Journals Library and the following databases via the ProQuest platform: Applied Social Sciences Index and Abstracts, Social Services Abstracts, International Bibliography of the Social Sciences, Social Sciences Database and Sociological Abstracts. Alerts were set up for the MEDLINE database from May 2018 to December 2018. Online sources were searched for grey literature and snowballing techniques were used to identify clusters of evidence.
Results: The concept analysis showed that de-implementation is associated with five main components in context and over time: (1) what is being de-implemented, (2) the issues driving de-implementation, (3) the action characterising de-implementation, (4) the extent that de-implementation is planned or opportunistic and (5) the consequences of de-implementation. Forty-two papers were synthesised
to identify six context–mechanism–outcome configurations, which focused on issues ranging from individual behaviours to organisational procedures. Current systems can perpetuate habitual decision- making practices that include low-value treatments. Electronic health records can be designed to hide
or remove low-value treatments from choice options, foregrounding best evidence. Professionals can
be made aware of their decision-making strategies through increasing their attention to low-value practice behaviours. Uncertainty about diagnosis or patients’ expectations for certain treatments
provide opportunities for ‘watchful waiting’ as an active strategy to reduce inappropriate investigations and prescribing. The emotional component of clinician–patient relationships can limit opportunities for de-implementation, requiring professional support through multimodal educational interventions. Sufficient alignment between policy, public and professional perspectives is required for de-implementation success.
Limitations: Some specific clinical issues (e.g. de-prescribing) dominate the de-implementation evidence base, which may limit the transferability of the synthesis findings. Any realist inquiry generates findings that are essentially cumulative and should be developed through further investigation that extends the range of sources into, for example, clinical research and further empirical studies.
Conclusions: This review contributes to our understanding of how de-implementation of low-value procedures and services can be improved within health-care services, through interventions that make professional decision-making more accountable and the prominence of a whole-system approach
to de-implementation. Given the whole-system context of de-implementation, a range of different dissemination strategies will be required to engage with different stakeholders, in different ways,
to change practice and policy in a timely manner.