CFIR: Consolidated Framework for Implementation Research

The Consolidated Framework for Implementation Research, known as CFIR, is a widely used tool in implementation science that organizes the factors influencing whether an evidence-based practice actually takes hold in a real-world setting. Published in 2009 and updated in 2022, the framework groups these factors into five broad domains and dozens of specific constructs, giving research and clinical teams a shared vocabulary for diagnosing why an initiative succeeds or fails. What makes CFIR distinctive is not that it invented new ideas about implementation but that it pulled together constructs scattered across existing theories into a single, organized structure that teams can apply systematically.

How the Framework Was Built

CFIR did not emerge from a single new study. Its developers used a snowball sampling approach to identify published implementation theories, then evaluated the constructs within those theories based on empirical support, definitional consistency, and measurability. Where different theories used different labels for what was essentially the same concept, the developers merged them. Where a single construct in one theory actually conflated two distinct ideas, they split it apart.¹ The result was a meta-framework: a curated menu of factors, drawn from the best of what already existed, that any implementation team could pick from depending on their project.

This origin story matters because it explains both CFIR’s greatest strength and a recurring criticism. The strength is comprehensiveness. Because it synthesized so many prior frameworks, CFIR covers territory that narrower models miss. The criticism is that the framework is descriptive rather than prescriptive. It tells you what to look at but does not, on its own, tell you what to do about what you find. That gap has driven much of the follow-up work, including the strategy-matching tools discussed later.

The Five Domains

CFIR organizes the influences on implementation into five domains. In the original 2009 version these were labeled Intervention Characteristics, Outer Setting, Inner Setting, Characteristics of Individuals, and Process. The 2022 update relabeled and restructured some of these, but the basic architecture remains. Think of the domains as five different lenses, each pointing at a different part of the implementation landscape.

The Innovation domain (originally “Intervention Characteristics”) focuses on the thing being implemented. Is it adaptable enough to fit local conditions? Is it compatible with existing organizational values and workflows? Studies in aged care settings, for instance, have highlighted that an innovation’s adaptability to specific care environments and its compatibility with existing systems are critical factors for uptake.² An intervention that works perfectly in a research trial but cannot flex to local realities will stall.

The Outer Setting domain captures forces outside the organization doing the implementing: policies, funding environments, peer pressure from other organizations, and the needs of the people being served. Measuring these external factors reliably has proven tricky. A systematic review of measurement tools for outer-setting constructs found that many instruments had strong content norms but weak internal consistency, reflecting how difficult it is to quantify messy, multi-layered external pressures.³

The Inner Setting domain zeroes in on the organization itself: its culture, leadership, available resources, communication networks, and readiness for change. This domain tends to surface the most frequently in applied studies. In a mixed-methods analysis of a school-based health initiative, leadership engagement correlated with both how acceptable and how appropriate the program was perceived to be, while implementation climate correlated with acceptability, appropriateness, and feasibility.⁴ Tangible resources like curriculum materials and gardens enabled the work, while intangible shortfalls like lack of time hindered it. Survey instruments designed to capture inner-setting constructs have generally shown good reliability and validity, making this one of the more measurable parts of the framework.⁵

The Individuals domain (called “Characteristics of Individuals” in the original, broadened in the update) examines the people involved. This includes not just their knowledge and attitudes but their roles and influence. Research on implementation champions, for example, found six key attributes that mattered: influence, ownership, physical presence at the point of change, persuasiveness, grit, and a participative leadership style.⁶ These attributes helped champions overcome institutional silos, build networks, and cultivate the kind of learning climate that made people receptive to change.

The Process domain captures how implementation unfolds over time: planning, engaging stakeholders, executing, and reflecting on progress. A study of PrEP delivery in rural South Africa illustrated what happens when this domain breaks down. Shifts in national policy, funding pressures, and inconsistent communications from the national health department led to fragmented planning, engagement, and monitoring at clinics that were already stretched thin.⁷

How Teams Actually Use CFIR in Practice

The most common way teams apply CFIR is through qualitative research, typically interviews or focus groups with people involved in an implementation effort. The framework serves as a codebook: analysts tag sections of interview data to specific CFIR constructs, then look for patterns. An early landmark application was a large-scale evaluation of the VA’s MOVE! weight management program, which used CFIR to code qualitative data across multiple sites and identify the constructs most strongly linked to implementation effectiveness. That study also laid out a step-by-step approach for rating constructs in ways that could be compared across sites and studies.⁸

Traditional qualitative analysis using CFIR is thorough but slow. A rapid approach developed by a VA research team shortens the process considerably. Instead of transcribing every interview word for word, the primary analyst takes notes during the interview, codes those notes into a CFIR construct-by-site matrix immediately afterward, and flags spots where more detail is needed. A second analyst then listens to the recordings and builds on the initial notes. The codebook uses CFIR constructs as the deductive backbone but leaves room for inductive codes that capture things the framework does not.⁹ This rapid approach makes CFIR more practical for time-pressed teams evaluating programs in real time.

Beyond coding data, analysts can use relationship coding and causation coding to capture how CFIR constructs interact with one another. Rather than treating each construct as an isolated checklist item, this approach maps out potential causal links and bidirectional relationships between constructs within a project.¹⁰ An organization might find, for instance, that weak leadership engagement is not just a standalone barrier but also the upstream cause of poor communication networks and low implementation climate, creating a cascade of problems.

Matching Barriers to Strategies with the CFIR-ERIC Tool

One of the most practical developments to come out of CFIR-related work is the CFIR-ERIC Matching Tool. ERIC stands for Expert Recommendations for Implementing Change, a compilation of named implementation strategies (things like “audit and provide feedback,” “use train-the-trainer,” or “develop an implementation blueprint”). The matching tool links specific CFIR-identified barriers to the ERIC strategies most likely to address them. The idea is that once you diagnose what is going wrong using CFIR, you can look up evidence-informed tactics to fix it.¹¹

How well does this work in practice? A study in the Veterans Health Administration compared the tool’s recommended strategies against real-world data on which strategies actually improved hepatocellular carcinoma care. The tool’s top twenty recommended strategies had a higher effectiveness rate than the broader set of strategies facilities actually used: about 70% of the top recommended strategies proved effective, compared with about 48% of all strategies facilities deployed on their own.¹² The tool did not perfectly predict what would work, but it meaningfully outperformed an unguided approach. That is a useful finding for teams deciding how to allocate limited implementation budgets.

The 2022 Update and Equity

The original CFIR was not static. Over more than a decade of use, researchers flagged recurring gaps, and the framework’s developers solicited structured feedback. The resulting 2022 update involved revisions to existing domains and constructs alongside the addition, removal, and relocation of constructs. Two changes stood out. First, the update better centered innovation recipients, the people on the receiving end of whatever is being implemented. Second, it added constructs related to equity, formally recognizing that implementation processes can widen or narrow health disparities.¹³

The equity addition was overdue. An intervention can be effective on average but reach marginalized populations less effectively, or it can be implemented in a way that assumes resources and infrastructure not equally available across communities. By embedding equity constructs directly into the framework, the update pushes teams to consider distribution of benefit and access from the start of their planning, not as an afterthought. For teams already using the original CFIR’s constructs in surveys or interview guides, the transition required mapping old constructs onto the new structure, a process that has been documented to help researchers maintain continuity across studies while incorporating the expanded domains.¹⁴

CFIR in Low- and Middle-Income Countries

CFIR was developed primarily with high-income-country health systems in mind, and adapting it to low- and middle-income settings has required some rethinking. A systematic review evaluating the framework’s fit for these contexts found broad agreement that constructs like culture and stakeholder engagement transferred well. But two constructs, “patient needs and resources” and “individual stages of change,” were commonly identified as incompatible with how implementation works in resource-limited environments. Reviewers also called for a new domain capturing characteristics of the broader health system, including architecture-level features like how care is financed and organized.¹⁵

In practice, inner-setting constraints tend to dominate in these environments. A systematic review of interventions to improve antibiotic use across low- and middle-income countries found that the most commonly reported barriers fell within the inner setting domain, particularly resource limitations and facility infrastructure.¹⁶ That does not mean the other domains are irrelevant, but when basic staffing, supplies, and physical infrastructure are stretched, they tend to overshadow subtler factors like organizational culture or individual attitudes.

Some teams have addressed these limitations pragmatically by adapting CFIR rather than using it wholesale. In a multi-country project across sub-Saharan Africa aimed at improving health data quality, researchers narrowed the framework to just two domains and ten constructs most relevant to their aim, identifying these through discussions with key informants from each country’s project.¹⁷ This selective approach avoided the framework feeling unwieldy or mismatched while still providing a common analytical structure across very different health systems.

Pairing CFIR with Other Frameworks

CFIR is good at answering “what factors are helping or hindering implementation?” but is less designed to answer “how well is the implemented program reaching people and producing outcomes?” That is where complementary frameworks come in. One common pairing is CFIR with RE-AIM, a framework that evaluates Reach, Effectiveness, Adoption, Implementation, and Maintenance. Used together, RE-AIM provides the outcome metrics for judging implementation success, while CFIR identifies the modifiable factors that explain why those outcomes are what they are.¹⁸ A team might use RE-AIM to discover that a diabetes prevention program has poor reach in rural clinics, then turn to CFIR to figure out that the problem is rooted in outer-setting policies combined with inner-setting resource shortfalls. Neither framework alone gives the full picture.

Common Pitfalls and Limitations

A systematic review of how researchers actually used CFIR revealed patterns that suggest the framework is often applied superficially. Few studies justified why they selected certain CFIR constructs over others, meaning teams tended to use the entire framework as a broad coding guide rather than tailoring it to their specific implementation question. The majority of studies used CFIR only for data analysis rather than also for planning or guiding the implementation itself. And few studies connected CFIR constructs to any measurable outcomes.¹⁹

In other words, teams often used CFIR as a retrospective labeling system rather than a prospective planning tool. That is better than no framework at all, but it undersells the framework’s potential. The most effective applications tend to involve selecting specific constructs in advance based on what the team knows about its context, collecting data targeted to those constructs, and then linking findings to specific implementation strategies through tools like the CFIR-ERIC approach described earlier.

Another limitation is sheer size. The updated CFIR includes dozens of constructs across five domains, which can overwhelm teams unfamiliar with implementation science. Selective use, focusing on the domains and constructs most relevant to a given project, is not a weakness but a practical necessity. The sub-Saharan African data quality project mentioned earlier is a good example of principled narrowing.

Using CFIR to Remove Low-Value Practices

Most implementation work focuses on getting people to adopt something new. But a growing area of interest is de-implementation: getting clinicians and organizations to stop doing things that are ineffective, wasteful, or harmful. CFIR has proven useful here because the same organizational dynamics that help or hinder adoption also affect removal. A scoping review of determinants for low-value care and its de-implementation found that the factors identified broadly mapped onto the same domains that CFIR uses: patients, professionals, inner context, outer context, process, and the practice itself.²⁰

Practical applications have followed. CFIR-informed interview guides have been used to study the transition from Pap smear-based cervical cancer screening to primary HPV testing, a case where clinicians needed to stop one practice and start another simultaneously.²¹ Similarly, a study on reducing low-value imaging used the framework to identify multilevel determinants of why clinicians order imaging tests that guidelines do not support.²² The appeal of using CFIR for de-implementation is that the framework already captures the organizational inertia, professional identity, and process fragmentation that keep outdated practices alive. The constructs do not need to be reinvented; they just need to be read in reverse.

CFIR and Telehealth

The COVID-19 pandemic forced a rapid expansion of telehealth, and CFIR became one of the most commonly used frameworks for studying how that expansion went. A scoping review of CFIR applications to telehealth implementation found that most of the published work appeared during or after 2020, and about three-quarters of the studies were qualitative or mixed-method efforts using the framework to identify barriers and facilitators.²³ Telehealth is a particularly interesting test case for CFIR because it stresses multiple domains simultaneously. The innovation itself must be usable by patients and providers, the outer setting includes wildly varying broadband infrastructure and reimbursement policies, the inner setting requires IT support and workflow redesign, individual clinicians range from tech-savvy to deeply resistant, and the process of switching from in-person to virtual care requires new planning, training, and evaluation mechanisms.

The surge in telehealth-related CFIR studies has also drawn attention to cost. A commentary applying CFIR’s inner- and outer-setting constructs to telemedicine cost analysis argued that factors like location, practice type, and clinical specialty substantially affect whether telehealth saves money or costs more. The implication is that blanket statements about telehealth being cheaper ignore the implementation context, exactly the kind of nuance CFIR is designed to surface.

Whether applied to telehealth, antibiotic stewardship, cancer screening, or school health programs, CFIR functions less as a theory that makes predictions and more as a diagnostic tool that prevents teams from overlooking important factors. Its value is proportional to how deliberately it is used. Teams that select constructs thoughtfully, collect data targeted to those constructs, and connect their findings to specific strategies tend to get the most out of it. Teams that bolt it on as a post hoc labeling exercise get a well-organized description of what happened, but not much guidance on what to do next.