Who represents me? A patient‐derived model of patient engagement via patient and family advisory councils (PFACs)

Abstract Background Despite increasing attention to patient and family advisory councils (PFACs), what patients who are not PFAC members expect of PFACs remains understudied. Understanding their expectations is critical if PFACs are to help health systems achieve certain outcomes (eg increased patient satisfaction with health systems). Objective To obtain rich insights about what patients who are not PFAC members expect of PFACs. Design From July to September 2018, we conducted a qualitative study using focus groups. Setting and participants We recruited patients and caregivers who receive their care from the Johns Hopkins Medicine Alliance for Patients (JMAP), LLC, a Medicare accountable care organization that in 2014 established a PFAC, the Beneficiary Advisory Council. Approach Using grounded theory, we analysed field notes, analytic memos and transcripts to develop a theoretical model of patient engagement via PFACs. Results Forty‐two patients and caregivers participated in five focus groups that included individuals of different ages, races, health statuses and socio‐economic statuses. Participants were largely unaware of PFACs. Participants wanted to know who represented them (interpreted as a form of political representation) and emphasized the need for representatives’ diversity. Who mattered because who could affect what PFACs do. Participants expected that all patients should be able to communicate with PFACs and that meaningful engagement could enhance perceptions of health systems. Conclusions Eliciting views about patient representation from patients who have not been engaged as advisors or representatives has the potential to inform PFACs’ activities. Attention should be given to improving and measuring patients’ awareness of, and interactions with, their patient representatives.


| INTRODUC TI ON
Engaging patients in health-care system design and quality improvement efforts is a critical part of patient-centred care. [1][2][3][4] One way this is happening is via patient and family advisory councils (PFACs).
PFACs are groups of patients, family members and family caregivers who serve as advisors in diverse health-care settings, from individual clinics and hospitals to entire health systems. PFACs can be seen as part of a broader global movement over the past two decades towards patient and public involvement in health-care planning. [5][6][7] In the United States, PFACs are prevalent and may be increasing in number and diversity. 8  Toolkits exist to help facilitate the creation of PFACs. [9][10][11][12] However, PFACs' potential may be underutilized. While case studies suggest how PFACs could improve the safety and quality of patient care, systematic and structured reviews have repeatedly demonstrated a lack of rigorous evaluation and evidence of such impact. 7,[13][14][15] Much work remains to ensure PFACs can further patientand family-centred change in diverse health-care settings. [16][17][18][19] Prior studies of PFACs and related participatory bodies have focused on the perspectives of PFAC members or members of the health-care administrative teams with whom they liaise. 16,17,20,21 To our knowledge, few studies have assessed how a general patient population (ie non-PFAC members) views PFACs. Engaging the broader patient population via PFACs may be an opportunity to improve patient-centred care delivery, but there remains uncertainty about whether PFACs can or should play this role. [22][23][24][25] For instance, if PFACs are to facilitate achieving certain outcomes, such as those related to patient satisfaction with, or trust in, health-care systems, 26 PFACs may need to operate in ways that align with general patients' expectations about these councils. To help fill this gap, we conducted a qualitative study investigating what patients who are not PFAC members might expect from a PFAC.

| Setting
This focus group study was conducted within the Johns Hopkins CMS requires its ACOs to include a patient representative, known as the 'beneficiary representative', on the ACOs' governing boards. This must be someone who receives care from the ACO and who has no conflicts of interest. In collaboration with its first patient representative, JMAP additionally created a PFAC, known as the Beneficiary Advisory Council (BAC). The BAC is a volunteer group which is meant to reflect the diversity of JMAP's Medicare patient population. The BAC reviews policies, programmes and other initiatives, such as quality improvement efforts and patient outreach messages, with the goal of elevating the patient voice in JMAP decision making. This setting presents an opportunity to evaluate what patients who have not been engaged as representatives or advisors think of the concept of a PFAC in a real-world scenario.

| Sample
We recruited focus group participants from a sample of 429 JMAP patients who had indicated their willingness to provide deeper insights about their responses on a prior survey that assessed their awareness of patients representatives and how important patient representation is to them. 27 Of these 429, 169 received their care within JMAP's Greater Baltimore region. Focusing on these 169 was thought to improve the likelihood of participation, because focus groups were conducted in Baltimore City. However, because the Greater Baltimore region is defined by provider location (not patients' addresses), our sample included participants from Baltimore City, Baltimore county (and neighbouring counties) and other outlying regions (eg individuals living in Pennsylvania who nevertheless receive care in Baltimore City). The sampling goal was to create diverse focus groups in terms of age, race, ethnicity, health and socio-economic status. To do this, we collected demographic information during recruitment telephone calls and invited participants to join particular focus groups so that all groups would have variation across demographic characteristics.

| Data collection
We held five focus groups between July and September 2018 at Johns Hopkins Berman Institute of Bioethics in Baltimore, Maryland.
One member of the research team (MD) moderated, and another (VD) took detailed field notes. Each focus group lasted approximately 100 minutes and included 8-9 participants. Focus groups were audio-recorded and transcribed. For approximately two-thirds of this time, we elicited participants' views of patient advisory councils, the findings of which are reported here.
The research team created a focus group guide, informed by the research team's prior research, 4,28 to elicit participants' expectations K E Y W O R D S health systems, participation, patient engagement, public involvement, representativeness regarding PFACs. The guide underwent two rounds of substantive feedback from the BAC, resulting in a final focus group guide (see Appendix S1). Probing follow-up questions (which could differ between groups, based upon each group's unique responses) elicited additional rich insights, and we modified the guide slightly over time to allow reflections and memos from earlier focus groups to inform later ones.

| Data analysis
Data analysis utilized constant comparative techniques of grounded theory. Our approach drew upon Charmaz's constructivist version 29 of Glaser and Strauss's classical grounded theory 30 (without the procedural rigidity that many consider evident in Strauss and Corbin 31 ). This approach recognizes that the researchers' participation and own vantage points can affect the resulting theory and that the resulting theory is not abstract but tied to particularities of time and place. Analysis began with the field notes and memos created immediately after each focus group. Based on these notes and memos, it appeared that by the fifth focus group no significant new content was being elicited. Therefore, we placed recruitment on hold.
After the interviews were transcribed, open coding commenced.
Two researchers (MD and VD) reviewed the transcripts and field notes, writing research memos and potential themes to explore during analysis. Next, one researcher (VD) reviewed the text lineby-line and developed a comprehensive set of codes, as close to the transcripts as possible, using both descriptive and evaluative coding approaches. 32 We grouped related codes into preliminary categories based on emergent patterns in the data. A second researcher (MD) used this preliminary codebook to code transcripts independently, modifying some codes and reorganizing others. To help ensure intercoder reliability, the two researchers met and discussed code interpretation and code application discrepancies in detail, resolving them through discussion, and reviewed half of the transcripts lineby-line, for coding accuracy.
This process confirmed that no significant new content was being elicited by the fifth focus group. Having achieved reasonable thematic saturation, no additional sampling or recruiting occurred. Axial coding then commenced by re-reviewing transcripts, field notes and memos using a constant comparative technique to clarify categories of codes, and to begin postulating relationships among them. Group-to-group validation, where a topic seemed important or of interest to all groups, was specifically sought.
Specialized computer software was not needed. The full codebook, with codes, categories and example quotations, is available as Appendix S2.
The goal of our approach was to allow a core category and a theoretical model to emerge from the data. A core category is a central concept or main theme that applies to the research; it must appear frequently in the data, be abstract and help explain other categories and relationships. 31 To arrive at a theoretical model, the researchers reviewed notes and memos, analysed the codes and categories from Appendix S2 in relation to the core category, and drew multiple diagrams of these relationships until arriving at a model that was most logically consistent with the data. 31 To ensure reliability, the researchers employed reflexivity techniques and engaged in member checking by sharing both the full code book and the theoretical model with 20 focus group participants and with the members of the BAC. We used the received comments to update and finalize the codebook categories and the model.

| Ethics review
This study was reviewed and approved by the Johns Hopkins Medicine Institutional Review Board on 26 March 2018 (IRB00167922).

| Characteristics of participants
Of 169 potential people called in random order, 33 agreed to participate, 26 refused to participate, and 110 were not immediately reachable via the contact information left. In addition, we recruited seven family members and two friends of participants, which helped us meet our goal of including family caregivers. This recruitment yielded 42 total participants, comprised of both patients and caregivers who were not PFAC members.
Characteristics of the focus group participants are shown in Table 1. The focus groups were diverse in including individuals of various ages, races, levels of educational attainment, yearly personal income, self-reported health statuses, self-reported difficulty doing errands and statuses as caregivers or patients. Most participants were unaware of patient and family advisory councils in general and the BAC in particular. Additional details about participants, including each group's composition, are available as Appendix S3. Quotes below include individual's self-identified gender and age.

| Theoretical model
The primary product of our analysis is the model shown in Figure 1.

| Who: The core category
The core category of central, unifying importance among all focus groups was that of who. This was often the first response participants had when prompted to express their gut reactions to a patient advisory council. The word 'who' itself appeared on average more than once per page of the transcripts, and member checking with the BAC and several focus group participants confirmed the centrality of who. To illustrate: I think who they are is very important.
(female, 36) Related to the core topic of 'who' was a dominant expressed desire that patients who are not PFAC members should have a way to be aware of who is representing them on a council. Participants brainstormed ways of being informed, including mailed newsletters, websites and materials made available in medical practice offices, among others. As some participants said: I would have to assume there would be willingness on the part of the advisory council is to have a pamphlet that talks about who these people are. One reason participants valued a PFAC interacting with the broader patient population appeared to be because they wanted to be sure the council was informed by others.

| Who: How should a PFAC be comprised?
When turning to the ideal PFAC composition, participants believed that individual PFAC members need not have every characteristic of an ideal member. Here, the emphasis was on diversity and balance.
As participants said: I don't think one patient probably could categorize everybody's illness, disability and all of that, I think we probably have to go a little deeper than one person.

| Who: The broader patient population should help determine the 'who'
The arrows in the model from the general patient population also reflect an expressed desire that the non-PFAC members should help to determine who the PFAC is via the recruitment process. Some However, most believed it would be necessary to screen potential PFAC members in some way, perhaps to ensure PFAC members had the characteristics previously discussed. Options discussed included selection by a physician who knows the patient, an interview process or even an election by the general patient population.

| Why 'who' matters
As illustrated in Figure 1 Still, other participants thought that transparently disclosing the resources given to PFAC members could help mitigate the risk of being perceived as not independent: Well you say that in the newsletter, 'Hey, we need support to put our newsletter out,' and transportation maybe, I don't know, whatever you need, right and you say …that doesn't make us not representing you, okay?

| PFACs and political representation
The centrality of the concept of who to participants suggests their views could be interpreted through the theoretical concept of political representation. This was evident in the words used by participants (eg 'term limits' and, 'Because when you have boards, or whatever you want to call them, representing people, that's just like the government. They're supposed to represent us'.) At the same time, we cannot conclude that participants endorsed any particular view of representation. Pitkin's seminal work on political representation describes how representation aims to make the voices of the public 'present again' in policy processes. 33  It is possible that participants quickly associate with representativeness because regulations sometimes use the term 'patient representative'. We attempted to avoid that language by repeatedly using 'advisory councils'. Still, our findings reinforced the importance of and need for additional clarity regarding representativeness. The need to understand legitimate representation is decades old, seen, for example, in health maintenance organizations in the 1970s 34 and managed care organizations in the 1990s. 35 One well-studied context has been US Federally Qualified Health Centers (FQHCs) that provide care in underserved areas and have long required a majority of consumers on their governance boards. 17,[36][37][38] Studies have shown that consumers on these boards are not representative of FQHC patients in the descriptive sense and, as a result, may fail to meet all patients' needs equally; there is also evidence that increasing consumer representation may harm FQHC financial performance.
Discussions about 'who' should be involved in health planning thus continue to be a major area of inquiry. 24,39 At a high level, however, our findings suggest participants endorsed a delegate model of representation (where representatives act on the stated preferences of those represented) as compared to a trustee model (where representatives follow their own thoughts regarding the right action). 40 A delegate model requires PFACs to connect with their constituency, at least enough to learn their preferences. 16,37 In prior studies, PFAC members have themselves expressed the desire to learn more about the preferences of other patients. 19

| Power and control
We can also interpret our findings in relation to questions about how much power PFACs should have. Arnstein's ladder of participation describes eight levels of citizen participation over a policy process, ranging from manipulative non-participation to consultation to full citizen control. 41 As the model reflects, our participants appeared not to demand or expect full patient control. A minority expressed a concern that PFACs are tokenistic, using words such as 'public relations' or 'window dressing'.
The reason for this could not be definitively inferred. Perhaps this resulted from perceived knowledge barriers; participants noted that having health-care organization administrators involved with PFACs is essential for translating the complexities of modern health care to members. Alternatively, perhaps patients are conditioned not to expect such control. Our findings may cohere with recent scholarship suggesting that the top rung of the ladder, citizen control, may be an ideal that is difficult to be achieved. 42

| The impact of patient councils
Lastly, our findings intersect with recent efforts at improving evaluation of patient and public involvement in health care. 28,43 Although participants discussed elements of fair and inclusive decision-making processes, as the model reflects, their primary interest was in PFACs' ability to improve health-care delivery. This included the patient experience of care delivery and concrete changes to policies and programmes.
Consistent with other work, our study suggests that PFACs could influence patient satisfaction and trust in the health system. 26 Existing resources and toolkits emphasize the potential of engagement to improve trust, 10 but real evidence is lacking. 44 This is another critical knowledge gap, and it could be particularly important for vulnerable patients. In the United States (as elsewhere), distrust of the health-care system may be associated with poor health as much as or even more than distrust of individual health-care professionals, 45 and evidence suggests that trust differs according to certain demographic variables, such as race. 46

| Implications
This study, by investigating the expectations of a general patient population (ie non-PFAC members), adds important elements to the existing literature on patient engagement in health-care governance. First, the components of the model can be the subjects of formal evaluation. In particular, this study underscores the important of metrics of communication between PFACs and the patients they represent.
These have not been prominent in prior evaluation approaches. 28,47 Second, consistent with prior findings from a survey of more than 3000 Medicare beneficiaries, 27

| Limitations
Like all studies, ours has limitations. First, although rich insights were obtained, the number of participants and focus groups was rather small and came from one specific geographic region. However, we were able to achieve thematic saturation and, likely because of participants' lack of familiarity with any specific PFACs, also to obtain perspectives on an idealized version of PFACs. Future studies are needed to collect insights from patients who are aware of PFACs and to compare those study's findings to ours. Second, as a qualitative study, there is inherent subjectivity in analysis despite employed reflexivity and member checking; the findings may not generalize to all health-care organizations or study populations. Third, by sampling and organizing participants into focus groups based on diversity, there is the possibility that findings might have changed if, for example, we conducted more homogenous groups. Fourth, although our groups included participants with diverse baseline views about patient representation via PFACs, there remains the possibility that participants' views systematically differ compared to non-participants.
We are careful not to conclude that this study's findings are the only correct theoretical model regarding what a PFAC should be or do. The fact that 'who' was the core category, rather than 'why', came somewhat as a surprise to the researchers (whose previous work had emphasized the 'why' question 4 ). We acknowledge that this finding could have resulted from a circumstance where few patients know of PFACs' existence. We must also acknowledge that some may view PFAC members not as representatives at all, but instead as lending their own individual experiences with care to organizational decisions. In fact, we observed this perspective while conducting semi-structured interviews with ACO leaders and beneficiary representatives before the present study (unpublished data).
Consistent with our constructivist approach to grounded theory, our findings represent a model created from the insights of a particular non-PFAC member population.

| CON CLUS ION
This study represents one of the first in-depth examinations of what a broader patient population expects of the patients who volunteer to serve as advisors at the health-care system where they receive care. Patients who are not PFAC members expect that PFACs should expend significant efforts in communicating with all patients about who those councils are and what they do. By doing so, PFACs may be better situated to realize their potential to improve patient-centred care and to improve general patients' views of, and trust in, the health systems these councils advise.

ACK N OWLED G EM ENTS
The

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.