Archive of Featured Articles
Systematic review of studies that combine CFIR with Theoretical Domains Framework
Constructs within the CFIR are primarily (though not exclusively) focused on organization-level constructs. The Individual Characteristics domain within CFIR is purposefully underdeveloped because of the decades of science and models that define determinants of individual-level behavior change. The Theoretical Domains Framework (TDF), developed by Susan Michie and colleagues, is comprised of 84 constructs organized across 14 domains. Constructs within the CFIR and TDF do overlap but their predominant focus varies. For example, within the Inner Setting domain in CFIR, the Goals & Feedback construct is defined as, “The degree to which goals are clearly communicated, acted upon, and fed back to staff and alignment of that feedback with goals.” This construct relates to goals, their communication, alignment, monitoring, and action planning at the organization-level. Within the TDF, several constructs may be related to these concepts but operate at the individual level, including constructs within the Goals domain and additional constructs within the Behavioral Regulation domain including self-monitoring and action planning.
This review (based on articles published through October 2016) identified 5 study protocols and 7 completed studies that combined use of CFIR and TDF. The authors found the single most common rationale for combined use of CFIR + TDF was the perception that they addressed different conceptual levels: organization and individual, respectively. Combining use of these frameworks has value but also adds considerable complexity. The authors call out the need for tools and guidance for how and when to combine use of multiple frameworks to help ensure that each adds value.
Use of CFIR in published articles through January 2015
Colleagues at University of North Carolina led a systematic review of all published articles that cite the CFIR. In early 2015, only 26 of 429 articles that cited the CFIR, actually used it in a meaningful way (i.e., applied it in an implementation study). Nearly all studies used qualitative or mixed methods, but two studies developed quantitative items to assess a subset of CFIR constructs. Over half of articles did not identify implementation outcomes which limited their ability to identify determinants of implementation. Four recommendations were made to further advance rigorous use of the CFIR (or any framework): 1) apply the framework across multiple phases of implementation; 2) be transparent about how/why subsets of constructs were chosen for evaluation; 3) identify and assess implementation outcomes so that determinants (based on constructs’ association with those outcomes) can be identified; 4) more deeply integrate the framework into evaluation work, using it to guide data collection, analysis, and interpretation. Also, the authors reflect on the CFIR’s usefulness and they recommend refinements so that it can continue to improve and lead to robust development and testing of theory and models.
Using correlation to identify determinants of implementation
A Telephone Lifestyle Coaching (TLC) program was piloted in 25 sites in Veterans Health Administration (VHA) medical centers. Our team conducted an in-depth evaluation of implementation in 11 of those pilot sites. This article shows how the CFIR was used to qualitatively code themes. Analysts also independently rated each construct to indicate whether it was a barrier or facilitator to implementation and how strongly it manifested (e.g., strong facilitator, weak barrier). The rate of referrals made by each pilot site was used to indicate penetration, which is an implementation outcome. Pearson correlation was computed to assess the association of each construct with the implementation outcome measure; the p-value was used to identify key determinants of implementation. 7 CFIR constructs within the Inner Setting and Process domains were identified as determinants of implementation. Appendices provide detail about the context of this evaluation including the strategies used to implement and qualitative descriptions of non-significantly correlated constructs. This method assumes each construct is independently correlated with the outcome. Other methods, such as Qualitative Comparative Analysis (QCA) can be used to identify multiple pathways and combinations of constructs that may lead to desired outcomes.