Realist Synthesis
- Introduced by
- Ray Pawson (2005)
- Year
- 2005
- Type
- Framework
- Family
- Theory-Driven Synthesis
What does it do?
Realist synthesis is a theory-driven, interpretive method for evidence synthesis developed by Ray Pawson (2005) that focuses on understanding HOW and WHY interventions work, rather than WHETHER they work.
Grounded in realist philosophy, realist synthesis examines Context-Mechanism-Outcome (CMO) configurations: how specific contextual conditions activate mechanisms that produce outcomes. Unlike traditional systematic reviews, which typically answer 'Does intervention X reduce outcome Y?', realist synthesis asks 'Under what conditions, through what mechanisms, for which populations does X work?' This approach is particularly valuable for complex interventions (policies, programs, multi-component treatments) where effectiveness varies dramatically across contexts, and for understanding why interventions succeed in some settings but fail in others.
The intuition behind the method
Traditional evidence synthesis often concludes 'Drug X reduces depression in adults' based on meta-analysis of RCTs. But implementation evidence shows the drug works brilliantly in some clinics and fails in others. Why? Realist synthesis says: outcomes depend on not just the intervention, but the context (availability of psychiatrist supervision, patient socioeconomic status, healthcare system capacity) activating specific mechanisms (side effects tolerated differently by different patients, adherence influenced by cultural beliefs, effectiveness depends on matching patient to specific drug based on prior response pattern). Realist synthesis is detective work: it synthesizes evidence to build a causal theory of what makes interventions work in which contexts. This theory is far more useful for policy and practice than a simple 'Yes, it works' because it identifies conditions under which implementation will succeed or fail.
History
Ray Pawson developed realist synthesis in the early 2000s as a critique of conventional systematic reviews. He argued that systematic reviews focusing narrowly on 'Did it work?' miss critical insights about how it works and for whom. Pawson's realist philosophy (grounded in critical realism) posits that causal mechanisms are real and discoverable through empirical investigation, but they operate within contexts that shape their activation and effects. Wong and colleagues at the University of Toronto advanced realist synthesis methods, particularly for healthcare and implementation research. The Cochrane Collaboration now recognizes realist synthesis as a legitimate review methodology complementary to quantitative systematic reviews. Realist synthesis has become increasingly popular in implementation science, health systems research, and policy evaluation, where context-dependent effectiveness is common.
When to use it
Use realist synthesis when (1) you need to understand how and why complex interventions work, not just whether they work, (2) evidence suggests the intervention's effectiveness varies dramatically across contexts or populations, (3) mechanisms of change are unclear and stakeholders need insight into causal pathways, (4) you are designing or implementing a program and need to understand what conditions enable success, (5) prior systematic reviews concluded 'the evidence is mixed' and you want to understand why, or (6) you aim to identify leverage points for policy or practice improvement (e.g., which aspects of context must change to enable intervention success). Common applications include implementation science, health systems strengthening, education policy, criminal justice programs, and complex social interventions.
Limitations and considerations
- Requires deep substantive knowledge and theoretical sophistication. Realist synthesis is not a straightforward checklist; it demands judgment about mechanisms and causal theory.
- More interpretive than quantitative systematic reviews, introducing subjectivity. Different reviewers may construct different theories from the same evidence. Transparency and stakeholder engagement reduce but do not eliminate this risk.
- Inclusion criteria are looser than traditional systematic reviews (includes diverse study types, some of lower quality). Studies are judged on relevance to program theory, not standardized inclusion criteria. This can lead to less rigorous data if not carefully managed.
- Time-consuming and resource-intensive due to iterative approach and need to understand mechanisms deeply. May not be feasible under tight timelines.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Addresses complexity explicitly: acknowledges that interventions' effects depend on context and mechanisms, not just intervention components alone.
- Synthesizes diverse evidence types: RCTs inform effectiveness, qualitative studies illuminate mechanisms, implementation research reveals context barriers—all contribute to understanding.
- Produces actionable theory: Rather than 'This program works (p <0.05),' synthesis produces 'This program works when X, Y, Z conditions are met; mechanisms are A, B, C; in other contexts, mechanisms are blocked by barriers 1, 2, 3.'
- Iterative and reflexive: theory evolves as evidence is examined; researchers remain transparent about how theory changed during synthesis.
- Particularly powerful for explaining heterogeneity: if studies show different effects, realist synthesis asks why (context/mechanism differences) rather than treating heterogeneity as a nuisance.
Applications
- Implementation research: Understanding why a clinical intervention adopted in one hospital succeeds, while the same intervention fails in another. Realist synthesis identifies context factors and mechanisms enabling success.
- Health system strengthening: Evaluating complex policy interventions (e.g., primary care restructuring, telemedicine rollout) by examining CMO configurations across different healthcare settings.
- Education and training programs: Synthesis of evidence on which teacher training programs work best, in which school contexts, through what mechanisms (e.g., improved teacher knowledge and confidence, school leadership support, resource availability).
- Social interventions: Understanding mechanisms through which community health worker programs reduce maternal mortality—mechanisms depend on trust, cultural context, referral pathways.
- Criminal justice programs: Synthesis of evidence on recidivism reduction programs, examining how context (staff training, participant motivation, community reentry support) and mechanisms (cognitive change, social support, employment) interact.
How it works
How it works
Define your research question in terms of what works for whom, in what circumstances, and how: 'What mechanisms explain why cognitive behavioral therapy for depression works better for some patients than others? Under what conditions does CBT engagement increase, and how do these mechanisms interact with patient/provider/organizational contexts?' Develop a preliminary program theory—an articulation of how you EXPECT the intervention to work, based on literature, stakeholder consultation, and logic models. This theory is hypothesis-like but is grounded in substantive knowledge, not just hunches. Conduct a systematic search (similar to traditional systematic reviews) for evidence about the intervention, but cast a wider net: include RCTs, observational studies, implementation research, qualitative studies, and realist evaluations that illuminate mechanisms and context effects. Screen studies for relevance to your program theory (Does this study speak to the mechanisms or contexts in my theory?). Unlike traditional systematic reviews, realist synthesis is iterative: as you extract data, your understanding of mechanisms and contexts may evolve, and you may re-search for studies addressing refined theory.
Extract data from included studies focusing on CMO configurations: What are the Context features (patient characteristics, provider factors, organizational setting, policy environment, cultural context)? What Mechanisms (cognitive, behavioral, social, organizational processes) are activated? What Outcomes result? Create a matrix mapping contexts to mechanisms to outcomes. Look for patterns: Are certain mechanisms active in certain contexts? Do outcomes differ predictably based on context-mechanism combinations? Synthesize evidence thematically, organized around key mechanisms and contexts. For example: 'Mechanism 1: Therapeutic Adherence depends on Context (older patients, stable housing, previous therapy success); when this context activates this mechanism, Outcome is sustained engagement. In contrast, Mechanism 2: Adverse side effects in patients with comorbid conditions undermines adherence despite otherwise favorable context.' Refine your program theory as evidence accumulates: update hypothesized CMO configurations, discard mechanisms unsupported by evidence, and elaborate mechanisms supported by multiple studies. Document assumptions: what aspects of theory remain speculative (unsupported by included evidence) vs well-evidenced?
Present refined program theory as refined CMO configurations: explicit, testable statements about how the intervention works in which contexts. Example: 'Cognitive-behavioral therapy for depression is effective when: [Context: patient has moderate depression, supportive social network, access to trained CBT therapist] activates [Mechanism: cognitive restructuring of negative thoughts, behavioral activation to increase positive reinforcement] producing [Outcome: sustained reduction in depressive symptoms]. However, when [Context: severe depression with suicidality, unstable housing, lack of therapist availability], mechanisms may be blocked or counteracted by [Context: overwhelming environmental stressors, limited cognitive capacity for cognitive work, lack of continuity in treatment], and outcomes may be poor.' Use evidence from all study types to build this theory: RCT data shows what works on average; qualitative studies explain why; implementation research reveals context barriers. Report confidence in different CMO configurations: which are well-supported by multiple studies of different types, and which remain hypothetical? Discuss implications: for practitioners, identify conditions likely to support success; for policy, identify investments or changes needed to activate key mechanisms.
Common pitfalls
- Imposing a predetermined theory uncritically. Realist synthesis should be iterative and data-responsive, not wedded to initial theory. Be prepared to discard or substantially revise initial CMO configurations if evidence contradicts them.
- Including studies of poor methodological quality without acknowledging limitations. Just because diverse evidence is included does not mean all studies are equally trustworthy. Weight evidence quality explicitly.
- Overstating the explanatory power of refined theory. Realist synthesis produces plausible theories grounded in evidence, but it does not 'prove' mechanisms. Acknowledge remaining uncertainty, especially about mechanisms poorly represented in available literature.
- Neglecting quantitative evidence. Realist synthesis values qualitative insights about mechanisms, but quantitative evidence (from trials, surveys) about effect sizes and population characteristics is equally important. Balance depth of mechanism understanding with breadth of outcome evidence.
Frequently asked questions
How is realist synthesis different from a traditional systematic review?
Traditional systematic reviews ask 'Does the intervention work?' and synthesize evidence quantitatively, aiming for a single pooled effect estimate. Realist synthesis asks 'How and why does it work, for whom, in what circumstances?' and synthesizes diverse evidence iteratively to build causal theory (CMO configurations). Traditional reviews use tight inclusion criteria and assess study quality uniformly. Realist reviews include diverse evidence types judged on relevance to understanding mechanisms and contexts. Both are systematic and transparent, but realist synthesis is more theory-focused and interpretive.
What is a CMO configuration, and how do I develop it?
CMO stands for Context-Mechanism-Outcome. A configuration is a causal claim: 'In context X (specific conditions of patient, provider, organization, society), mechanism M (causal process) is activated, producing outcome O.' Example: 'In context of experienced, supervised therapists working with motivated patients in well-resourced clinics, the mechanism of cognitive restructuring is activated, producing sustained anxiety reduction.' Develop configurations by synthesizing evidence: use RCTs and observational studies to identify outcomes, qualitative and implementation studies to identify mechanisms, and compare across studies to identify context variations.
Can I include low-quality studies in realist synthesis?
Yes, realist synthesis includes diverse evidence types and quality levels if relevant to understanding mechanisms and contexts. However, acknowledge quality: for questions about effect size, prioritize RCTs and strong observational studies. For understanding mechanisms, qualitative and implementation studies are valuable even if not RCTs. Explicitly rate quality and weight your interpretations accordingly.
How do I balance theory development with empirical evidence?
Start with preliminary program theory based on substantive knowledge, then iteratively update it as evidence emerges. Do not force evidence into a predetermined theory; be prepared to substantially revise. Distinguish between: (1) Mechanisms well-supported by multiple studies (describe as established); (2) Mechanisms supported by one or two studies (describe as likely); (3) Aspects of theory not addressed in literature (describe as speculative). This transparency allows readers to judge theory credibility.
Related methods
Sources
Pawson, R., Greenhalgh, T., Harvey, G., & Walshe, K. (2005). Realist review—a new method of systematic review designed for complex policy and programme evaluation. Journal of Health Services Research & Policy, 10(S1), 21–35.
Open DOI referenceCopied.Pawson, R. (2013). The Science of Evaluation: A Realist Manifesto. SAGE Publications.
Open publication referenceCopied.Wong, G., Westhorp, G., Pawson, R., & Greenhalgh, T. (2013). Realist synthesis: Introduction and some practical guidance. Cochrane Database of Systematic Reviews, 12, CD012032.
Open publication referenceCopied.