Written and maintained by CASRAI Editorial Board
Last updated
CFIR (the Consolidated Framework for Implementation Research) is a determinant framework: a menu of the contextual factors — organised into five domains — that research has repeatedly found to help or hinder whether an intervention actually takes hold in a real setting. It was developed by Laura Damschroder and colleagues and published in the journal Implementation Science in 2009 as a synthesis of 19 existing implementation theories, then substantially revised in 2022. Most people meet CFIR as a coding scheme: researchers use it to code interview transcripts, field notes, and documents for the specific barriers and facilitators a given implementation ran into, then use those codes to explain why an intervention succeeded, stalled, or failed.
That coding use is where CFIR is genuinely different from two frameworks it gets confused with constantly. This guide covers the five domains as they stand after the 2022 update, works through an actual coding pass on a sample excerpt, and is explicit about where CFIR stops and where a framework like RE-AIM or the Iowa Model picks up.
What CFIR is for — and what it isn’t
CFIR answers one question: which contextual factors, present in this specific implementation, plausibly explain the outcome? It doesn’t tell you whether the outcome was good (that’s evaluation) and it doesn’t tell you what sequence of decisions to follow to get an intervention into practice (that’s a decision pathway). Confusing these is the single most common misuse of CFIR in the literature, so it’s worth stating plainly, since two of CFIR’s closest neighbours are already live on this site:
- CFIR is a determinant/diagnostic framework. It organises the “why” — the barriers and facilitators a specific implementation actually encountered — and it’s used before an implementation (to plan) or after (to explain results), most often via qualitative coding.
- RE-AIM is an evaluation framework. It scores what happened along five outcome dimensions (Reach, Effectiveness, Adoption, Implementation, Maintenance) with a specific denominator for each. It doesn’t explain why those numbers came out the way they did — that’s the gap CFIR-coded data fills.
- The Iowa Model is a decision pathway. It’s a sequence of go/no-go gates that routes a clinical trigger through problem identification, evidence appraisal, piloting, and adoption. It tells a team what to decide next; it doesn’t supply a vocabulary for coding why a pilot ran into trouble.
In practice, the three are often used together on the same project: CFIR identifies and explains the contextual determinants, a decision-pathway model like Iowa governs the go/no-go sequence, and RE-AIM scores the outcome once something is actually running. None of the three substitutes for either of the others, and a study that says it “used RE-AIM” to identify barriers, or “used CFIR” to report reach and effectiveness, has usually misapplied one of them.
The five CFIR domains (2022 revision)
The 2022 update renamed three of the five domains and reorganised the construct list underneath them; researchers holding CFIR-coded data from before 2022 should treat this as a domain relabelling and a reorganisation, not a different framework — the original domains map onto their current names as follows: Intervention Characteristics became Innovation, Characteristics of Individuals became Individuals, and Process became Implementation Process. Outer Setting and Inner Setting kept their names. The current domains and their constructs, per the CFIR Research Team’s own published list:
1. Innovation
The “thing” being implemented — a clinical protocol, a piece of software, a service model. Constructs: Innovation Source, Innovation Evidence-Base, Innovation Relative Advantage, Innovation Adaptability, Innovation Trialability, Innovation Complexity, Innovation Design, Innovation Cost.
2. Outer Setting
The environment the implementing site sits inside — the health system, the district, the regulatory context. Constructs: Critical Incidents, Local Attitudes, Local Conditions, Partnerships & Connections, Policies & Laws, Financing, and External Pressure (with societal, market, and performance-measurement sub-pressures).
3. Inner Setting
The setting where the innovation is actually implemented — the unit, the clinic, the department. Split into characteristics that persist regardless of what’s being implemented (Structural Characteristics, Relational Connections, Communications, Culture) and characteristics specific to this particular innovation (Tension for Change, Compatibility, Relative Priority, Incentive Systems, Mission Alignment, Available Resources, Access to Knowledge & Information).
4. Individuals
The people involved, split into two subdomains. Roles: High-Level Leaders, Mid-Level Leaders, Opinion Leaders, Implementation Facilitators, Implementation Leads, Implementation Team Members, Other Implementation Support, Innovation Deliverers, and Innovation Recipients — the last of these added in the 2022 revision, formally recognising the people on the receiving end of the innovation as a distinct role rather than folding them into “deliverers.” Individual Characteristics: Need, Capability, Opportunity, Motivation.
5. Implementation Process
The activities and strategies used to actually implement the innovation: Teaming, Assessing Needs, Assessing Context, Planning, Tailoring Strategies, Engaging, Doing, Reflecting & Evaluating, and Adapting. Assessing Context — explicitly separated from Assessing Needs in the 2022 revision — is the construct most directly relevant to a barriers/facilitators coding pass: it’s defined as identifying barriers and facilitators before or during implementation, which is the coding work this guide walks through next.
Using CFIR as a coding framework
Most CFIR-based studies use the framework deductively: the five domains and their constructs become a codebook applied to qualitative data (semi-structured interviews, focus groups, field notes, meeting minutes) collected from people involved in an implementation. The mechanics are close to any deductive qualitative coding pass — see CASRAI’s worked-example guide to coding qualitative interview data and content analysis guide for the general method — with two things specific to CFIR:
- Constructs, not domains, are the coding unit. A transcript segment gets coded to a specific construct (e.g. “Compatibility” or “Relative Priority”), not just to “Inner Setting” — the domain is the grouping the construct belongs to, not the code itself.
- Code for valence, not just presence. The same construct can be coded as a facilitator, a barrier, or mixed, depending on what the participant actually said. “Compatibility” isn’t a barrier or a facilitator by default — whether a given implementation’s workflow fit was reported as smooth or disruptive is what determines the valence, and that valence is what makes the coded data usable for explaining an outcome rather than just cataloguing which constructs came up.
Interview questions are typically written to probe one or a few constructs directly (a semi-structured guide organised by domain works well — see CASRAI’s guide to designing semi-structured interviews), but real answers rarely stay inside one construct’s boundary, which is exactly why a full codebook pass across the transcript, rather than pre-sorting answers by the question that prompted them, is the standard approach.
A worked coding example
The excerpt and coding below are an illustrative composite, written to demonstrate the coding process — not a transcript from a real study, institution, or participant.
Composite scenario: a hospital unit is six weeks into implementing a new fall-prevention screening protocol. A staff nurse is asked what’s helped and what’s gotten in the way.
“Honestly the tool itself is fine — it’s quicker than the one we used before, so nobody’s really fighting it on that front. The problem is we’re still charting it on paper and then re-entering it into the EHR at end of shift, because IT couldn’t get the build finished before rollout. That’s maybe five extra minutes a patient, and on a bad night that adds up. Our unit manager has been really good about backing us up when we push back on it in huddle, though — she’s made it clear this is a priority for her, not just something corporate handed down.”
A deductive CFIR pass on this excerpt produces at least four codes:
| Segment | Domain | Construct | Valence |
|---|---|---|---|
| “quicker than the one we used before” | Innovation | Innovation Relative Advantage | Facilitator |
| “charting it on paper and then re-entering it… IT couldn’t get the build finished” | Inner Setting | Compatibility (workflow fit) / Structural Characteristics (IT infrastructure) | Barrier |
| “five extra minutes a patient… on a bad night that adds up” | Innovation | Innovation Complexity (added burden) | Barrier |
| “unit manager has been really good about backing us up… made it clear this is a priority for her” | Inner Setting | Relative Priority (leadership signalling) / Individuals: Mid-Level Leaders | Facilitator |
Two things this example is meant to show: first, a single sentence can legitimately carry two construct codes (the charting problem is simultaneously a Compatibility issue and a Structural Characteristics issue — the EHR build gap is the underlying cause, the workflow mismatch is the lived experience of it, and a rigorous codebook usually captures both rather than forcing a single label). Second, the same construct (Innovation Relative Advantage, Innovation Complexity) can pull in opposite directions within one interview — the tool is simultaneously “quicker” and “five extra minutes,” and both are true, which is exactly the kind of nuance a pure outcome measure like RE-AIM’s Effectiveness dimension can’t capture on its own.
From coded barriers to implementation strategies
Coding is a diagnostic step, not an endpoint. Once barriers and facilitators are coded to specific CFIR constructs, the next step in most implementation projects is selecting strategies to address the barriers that actually showed up — rather than applying a generic implementation “toolkit” regardless of what the coded data says. The implementation-science literature has built structured matching resources specifically for this construct-to-strategy step, most built around the Expert Recommendations for Implementing Change (ERIC) compilation of 73 discrete implementation strategies. The practical takeaway for a CFIR coding project: treat the coded barrier list as the actual input to strategy selection, not as a report that ends the project once it’s written up.
Frequently asked questions
What are the five domains of CFIR?
Innovation, Outer Setting, Inner Setting, Individuals, and Implementation Process, per the 2022 revision. The 2009 original used different names for three of these: Intervention Characteristics (now Innovation), Characteristics of Individuals (now Individuals), and Process (now Implementation Process).
Is CFIR a theory?
No — CFIR is a determinant framework, not a theory. It doesn’t propose or test causal mechanisms for how change happens; it provides a common taxonomy of the contextual factors implementation researchers have found relevant across many prior theories, so studies can report and compare determinants using shared terminology instead of each inventing its own.
Can CFIR be used for quantitative research, or only qualitative?
CFIR constructs can be operationalised as quantitative survey items (several validated CFIR-based surveys exist), but the framework originated from, and is most commonly applied through, qualitative coding of interviews and documents — which is the use this guide focuses on.
What’s the actual difference between CFIR and RE-AIM?
CFIR explains why an implementation succeeded or struggled, by coding the contextual barriers and facilitators present. RE-AIM measures what happened, across five outcome dimensions with an explicit denominator for each. A study can use both without redundancy: CFIR for the “why,” RE-AIM for the “what happened” — see CASRAI’s RE-AIM guide for the outcome side.
Does CFIR tell you which implementation strategy to use?
Not directly. CFIR identifies and codes the determinants; strategy selection is a separate step that maps coded barriers onto a compilation of implementation strategies (most commonly the ERIC compilation) — CFIR supplies the diagnostic input to that step, not the strategy list itself.
For the broader field this guide sits in, see CASRAI’s implementation science dictionary entry and the research methods pillar. For related determinant and evaluation work, see the Knowledge-to-Action framework and CASRAI’s guide to hybrid effectiveness-implementation trial designs.
