We analysed interview data from a process evaluation in a hybrid trial, to ascertain the patient perspective on implementation determinants. The process evaluation design was CFIR-informed; we used CFIR throughout our process of data collection and analysis. We undertook a data driven, inductive, thematic analysis and a theory-based, deductive CFIR analysis and compared these analyses against one another.
Our thematic approach allowed analysis to be driven by the patient experience and resulted in the fullest account of what mattered to patients, showing a greater range of implementation-relevant factors than were included in CFIR. Through thematic analysis we discerned three themes: (i) living with high cholesterol, (ii) the treatment, (iii) accessing and receiving the treatment. The findings showed that patients had unmet needs and suggested what a new treatment would need to offer, to be taken up by patients. Inclisiran was seen as an efficacious, acceptable treatment that would meet patient needs and the importance of general practice as place was key.
Whilst we had not originally planned to undertake a deductive, CFIR-based analysis, we were prompted to do so by the good fit between our data and several CFIR constructs that became apparent during the early stages of our analysis, the debate around the utility of CFIR for the patient perspective and the coincidental publication of the updated CFIR, with an increased emphasis on patients, during our data analysis period. These factors together led to our interest in comparing the two versions of CFIR.
In our CFIR analyses, 11 constructs across five domains in the original CFIR[1] and 16 constructs across four domains in the updated CFIR[11] were relevant. With some existing constructs moved between domains and others added, the updated version of CFIR has better utility for the patient perspective. For our first theme, ‘living with high cholesterol’, the original CFIR accommodated all our categories except ‘negative societal image’ and ‘media coverage’. In relation to our second theme, ‘views about the treatment’, the original CFIR accommodated all except one category, and, with the addition of the ‘societal pressure’ sub construct that includes media influence, the updated version accommodated all. For the third theme ‘accessing and receiving the treatment’, CFIR could not accommodate our key category ‘importance of place’, or the code ‘familiar setting’. The original version included no interpersonal skills and whilst the updated version includes ‘interpersonal skills’ that can be applied to any role, other social skills including empathy and building relationships have been introduced but are assigned to certain roles, being included within the definition for the ‘innovation facilitator’ role, which includes empathy and building relationships, but not the ‘innovation deliverer’ role. This was problematic for our analysis, as these were skills that patients perceived and valued in deliverers.
Comparing our findings to previous analyses of CFIR informed studies of the patient perspective, we found a similarity in that the intervention/innovation domain had the greatest number of relevant constructs.[2–6] Only three studies found the ‘patient needs and resources’ construct relevant to their analysis. However, overall, the analyses reported lacked detail and apply CFIR at domain, not construct level. In comparison, our analysis is more detailed and also provides the first application of the updated CFIR to the patient perspective.
Comparing the CFIR analyses with our thematic analysis overall, the original CFIR accommodated everything except media influence, the innovation deliverer role, interpersonal skills, general practice as setting type, soft skills for deliverers and the importance of place and familiarity of place. The updated version now accommodates the first four, but not soft skills for deliverers, the familiarity of the setting, or the importance of place. The updated ‘individuals’ domain now lists more roles than were in the original CFIR and was comprehensive when identifying all the roles mentioned by our interviewees. More skills have also been added, but some are ascribed to particular roles which seemed to preclude us from applying them where we needed them. We recommend that all personal attributes are grouped together, in a construct(s) within the ‘characteristics’ subdomain, following the way that the COM-B based constructs are presented and not assigned to any particular role. This would enable users to select and apply the constructs to any role or interaction as required. ‘Familiarity of place’ could potentially be added to the ‘inner setting’ domain as one of the attributes that researchers are now prompted to document at the top level of that domain.
We found most of what mattered to our participants could be accommodated within CFIR. In this regard, it functioned as a useful coding frame, with constructs that matched well with our data, especially during the earlier stages of thematic analysis. Once we moved on to grouping codes and discerning categories, for our thematic analysis, we needed to move and rearrange data, using our matrix and then other categories, shifting beyond the CFIR structure of constructs and domains. So whilst CFIR can provide a good coding frame, for thematic analysis, it is too rigid for the process of developing the more analytical categories and themes. As frameworks are inherently static, this is not a criticism of the nature of CFIR per se, but identifies a limitation for its application in thematic analysis. Previously, others commented that determinant frameworks can lead researchers to ‘accumulate and catalogue constructs without offering additional analytical purchase’[23] and this matches our experience with the analysis described here and our need to reorganise and structure summaries of data, differently the CFIR structure, in order to develop categories and themes. Similarly, none of our themes could be accommodated within or matched entirely with any one CFIR domain, all our themes cut across two or more domains. This fits with recent research comparing CFIR with thematic analysis, where themes cut across domains.[24]
Previous commentary on the original CFIR, identified that the framework did not sufficiently capture the patient perspective. Criticism focussed on the location of patients, in the ‘patient need and resource’ construct, in the ‘outer setting’ domain and suggestions were made to create a dedicated ‘patient perspective’ domain and separate constructs for knowledge of and prioritisation of, patient need.[9, 10] In setting out their intentions for the updated CFIR, the authors emphasise patients, stating their intentions of ‘better centering innovation recipients’. We agree that placing patients and their needs in the ‘outer setting’ seemed marginalising. Previously ‘patient need and resource’ was underpinned theoretically by the PRISM, now this underpins ‘patient-centredness’ within the inner setting. PRISM was key in interpreting the data for this study and we consider the relocation of these essential patient-focussed concepts an important improvement. The newly-added construct ‘innovation recipients’ was relevant to our data on patient need, so was another useful amendment and its location alongside all other roles, in the ‘individuals’ domain, is inclusive.
The current study compared the utility of the original and updated CFIR for the patient perspective, but was situated within a broader process evaluation that generated multiple perspectives on the implementation of the same new service. Several issues raised by patients were also important to professionals. Patient preference for general practice as a place for accessing the service was recognised by GPs and nurses, who also agreed that it was, in principle a good setting for such a service. Patients concerns about costs and uncertainty around future financing of the treatment were also important barriers for professionals, who thought that lack of secure long term funding for the treatment could limit willingness to prescribe and therefore impact wider spread. Patients were also able to give views on what a future service might look like, describing the potential for future nurse led services, views that fit closely with those of GPs and nurses.[25]
Overall, we found the updated version of CFIR useful in determining patient perceived factors that can affect implementation. Considering this and the CFIR authors’ assertion that they wished to make the framework more patient-centred, we were surprised by the CFIR authors’ response to debate on the position of patients in CFIR and their assertion that CFIR should not currently be used with patients. The CFIR authors assert that current power imbalances and inequity preclude the patient perspective on implementation determinants from having any influence. The authors are critical of the situation, stating that patient involvement, engagement, co-design initiatives should result in more power-sharing and that they hope that patients will increasingly hold roles as decisions makers/deliverers and have greater ‘influence, authority and power in healthcare systems’.
Most implementation research targets the ‘agents of implementation’, such as health and care professionals, service commissioners and policy makers.[26] In this regard we agree that patients can lack the power to influence change in healthcare systems, but there are also many circumstances where patients have agency and ability to influence the uptake of interventions. This can include research where the primary research focus is associated with strategies to incorporate interventions into everyday clinical practice. In these instances, such as our own VICTORION-Spirit study, patient experience data is crucial to inform efforts to promote wider adoption and spread. Therefore, we argue for a different approach to that of the CFIR authors - applying CFIR to patient data, including the patient perspective, in the creation of knowledge and thereby reducing some of the in-built inequality in the implementation science evidence base.
Strengths of the study are that it provides a more detailed application of CFIR to the patient perspective than has been available previously. In doing so, it offers a perspective that is often underrepresented in implementation science research. This research took place in primary care, future research should apply CFIR to the patient perspective in other settings. Limitations are that all participants took part in a trial and may not be typical of the eligible population for inclisiran. We only interviewed those who agreed to receive the treatment; trial team facilitators informed us that some patients had declined the treatment because they were scared of needles, but we did not obtain from patients their reasons for declining or stopping inclisiran. Future CFIR-informed research should include views from patients who decline or drop out of treatment to understand reasons for this.