A Community-Engaged Process for Adapting a Cardiovascular Health Intervention for Persons with Serious Mental Illness Open Access

Cardiovascular disease (CVD) is the leading cause of death in the United States.¹  People with serious mental illnesses—including schizophrenia, bipolar disorder, or major depressive disorder—experience one of the largest disparities in CVD-related mortality of any group, dying 10 to 20 years earlier than the overall US population.²  Underlying this wide disparity is the alarmingly high prevalence of all CVD risk behaviors and risk factors in this population,³  compounded by low rates of CVD risk factor control.⁴ 

To address the high burden of CVD in persons with serious mental illness, there is an urgent need to scale-up effective interventions that have been tailored to meet the specific needs of this population.⁵  Two such interventions, IDEAL⁶  and Life Goals,⁷  have led to clinically significant CVD risk reduction in clinical trials; however, the uptake of these interventions into real-life clinical settings remains limited. To facilitate wider scale-up, a crucial step is to collaboratively engage with end-users to adapt the intervention in a way that maintains fidelity to the intervention’s “core elements” (i.e., the active ingredients that lead to the intervention’s intended effect on outcomes) while enhancing the intervention’s acceptability by tailoring its “non–core elements” and implementation strategies (i.e., the methods or techniques used to improve adoption, implementation, and sustainment of interventions⁸ ) to better meet the needs of community partners.⁹ 

Increasing evidence suggests that interventions that are adapted to a new context, patient population, or group of providers tend to be more effective.^10,11  Research also suggests that when adapting an intervention, engaging with community partners during the process can increase the relevance of research, trust between research and community partners, and adoption of the intervention into routine practice.¹²  Guidance on adapting complex health interventions now routinely recommends following a community-based participatory approach,¹³  which seeks to promote shared decision-making and ownership by equitably involving community members, researchers, and other key partners in all phases of the research project.¹⁴ 

Although the imperative to engage with community partners in adapting interventions seems clear, guidance on how to engage with stakeholders is often missing or poorly described.^15,16  This lack of specification is meaningful because it limits the ability of researchers and practitioners to effectively plan, conduct, and evaluate community engagement activities. This gap in knowledge can be particularly acute for communities facing health inequities,¹⁷  which often have limited resources to make and document appropriate adaptations,¹⁸  potentially undermining intervention uptake even further. The relative lack of detail on the operationalization of community engagement activities also has important implications for measurement. A more rigorous approach to establishing the link between community engagement activities and implementation outcomes is needed,¹⁵  not only to help quantify the value of these activities, but also to identify where community engagement activities may have greatest impact.

In this article, we describe a novel, theory-informed process of garnering input from community partners to adapt IDEAL Goals, a combination of 2 complementary evidence-based interventions (IDEAL and Life Goals) for improving CVD risk factors in persons with serious mental illness. We begin by situating our approach in the Replicating Effective Programs (REP) framework,⁹  which offers a systematic process for translating evidence-based interventions into community-based settings. We then systematically describe 2 community engagement activities—a needs assessment and community working groups—using the recommendations of Concannon et al¹²  for specifying and reporting stakeholder engagement in research. By clearly delineating how we conceptualized and operationalized these community engagement activities, we aim to contribute to the evidence base of how to better ascertain, integrate, and measure community-engaged processes in the adaptation of evidence-based practices.

This study is funded by the National Heart, Lung, and Blood Institute as part of the “Disparities Elimination through Coordinated Interventions to Prevent and Control Heart and Lung Disease Risk” (DECIPHeR) Alliance, a 7-year cooperative agreement consisting of a 3-year planning phase (UG3 phase) and 4-year implementation phase (UH3 phase). Our project, “Achieving Cardiovascular Health Equity in Community Health: Optimizing Implementation Strategies,” is led by a research team at Johns Hopkins University and the University of Michigan.

In phase 1 (UG3 planning phase), the focus of this article, we engaged with community partners to adapt a CVD risk reduction intervention and implementation strategies to fit the needs of community mental health organizations. In phase 2 (UH3 trial phase), we plan to test the implementation strategies in Maryland and Michigan community mental health programs by using a cluster-randomized type 3 hybrid implementation-effectiveness trial.¹⁹  The overall study aims to inform which combination of implementation strategies will lead to optimal uptake of the intervention in community-based settings, a critical step in reducing CVD disparities and achieving health equity in this population.

The evidence-based intervention we sought to adapt, IDEAL Goals, is a multicomponent behavioral lifestyle intervention that is based on the collaborative care model.²⁰  The IDEAL Goals intervention was formulated iteratively through the combination of 2 evidence-based CVD risk reduction interventions, IDEAL and Life Goals, that were designed for people living with serious mental illness. In the IDEAL trial, which centered on tailored behavioral counseling and care coordination and care management activities, participants experienced a relative reduction in the 10-year probability of a cardiovascular event (as measured by the global Framingham Risk Score).⁶  The overarching concept of the Life Goals intervention, which had both a self-management and care coordination component, was that behavioral management of psychiatric symptoms, in combination with care coordination, can promote healthier lifestyles for cardiovascular health.²¹  Life Goals participants saw improvements in blood pressure, cholesterol, and physical health quality of life.⁷  As a combination of the 2 interventions, IDEAL Goals incorporates the core elements of each, namely psychoeducation and care management.

The study context includes outpatient community mental health programs in Maryland and Michigan that have implemented behavioral health homes, a model of care that provides enhanced care support to people living with both physical and mental illnesses. Multiple state Medicaid programs have implemented behavioral health homes as part of the Affordable Care Act Medicaid Health Home Waiver,²²  which permits qualifying mental health programs to be paid for physical health care coordination and management. While behavioral health homes have improved access to care and preventive screening for persons with serious mental illness, they have largely failed to reduce CVD risk factors in this population,²²  likely owing to limited use of evidence-based practices.²²  In light of the evident need for intervention, as well as the intervention’s alignment with existing goals and workflows, behavioral health homes represent an optimal setting to implement IDEAL Goals.

Our approach to garnering input from community stakeholders is based on the REP framework.⁹  As an implementation framework, REP offers a systematic approach to translating evidence-based interventions into community-based settings that maximizes fidelity to the interventions’ core elements while incorporating stakeholders’ input to enhance fit in new populations and organizations. First developed by the US Centers for Disease Control and Prevention to package and disseminate HIV behavioral and treatment interventions for implementation in community-based settings,²³  the approach has since been used to guide a variety of translation efforts, including adapting interventions that target chronic disease prevention and management in underserved communities.²⁴ 

The theoretical underpinnings of REP include Rogers’ diffusion of innovation theory,²⁵  which posits that innovations pass through particular communication channels among members of a social system over time, and Bandura’s social learning theory,²⁶  which describes the relationship between behavior change and persons’ beliefs about their ability to change and the results of the change. These theories support REP’s foundational approach of directly engaging with end-users (eg, providers, patients, policy-makers) in order to tailor the intervention to fit local needs, increase stakeholder buy-in through the process of partnering, and promote equity in delivery across groups.

The 4 phases of REP include (1) preconditions (e.g., identifying need, target population, and suitable intervention), (2) preimplementation (e.g., community input and intervention packaging), (3) implementation (e.g., package dissemination, training, technical assistance, and evaluation), and (4) maintenance and evolution (e.g., preparing the intervention for sustainability). This article focuses on the preimplementation phase of REP.

To engage with our community partners during the preimplementation phase, we used 2 community engagement activities: (1) a needs assessment, which serves to benchmark usual care and identify potential implementation barriers and facilitators; and (2) community working groups, which aim to collaboratively engage with end-users in adapting the intervention and implementation plans (Figure 1). The 2 activities were conducted sequentially, so that learnings from the needs assessment could inform the composition and content of the community working groups. To describe how we operationalized these activities, we used elements of the Stakeholder Engagement Reporting Questionnaire of Concannon et al¹²  (2014) to guide our description of the participants and recruitment process; the intensity, methods, and modes of engagement; and the impact of engagement activities.

An institutional review board at the Johns Hopkins University School of Medicine approved the protocol. All participants were read a waiver of documentation of consent and verbally confirmed their understanding of the study.

Our first step in engaging with community partners was to conduct a needs assessment to identify CVD risk factor levels and anticipated barriers and facilitators to implementing the IDEAL Goals intervention. Informed by the ADAPT-ITT framework for adapting evidence-based practices,²⁷  our needs assessment leveraged a mixed methods approach that included site-level surveys and in-depth interviews with community mental health organization staff.

Participants included administrative leaders and managers, nurses, frontline staff (eg, case managers, community health workers, service coordinators, rehabilitation specialists), and other relevant staff working at community mental health programs. The pool of eligible organizations included 31 sites in Maryland and Michigan that had implemented a behavioral health home and expressed a willingness to participate in the project. For each site, our target was to recruit 1 participant (usually the administrative lead) to complete an infrastructure survey and at least 2 participants (1 frontline staff and 1 health home nurse or organizational leader) to participate in the in-depth interviews.

To recruit participants, our study team set up introductory meetings with site leaders to provide an overview of the project and describe the anticipated time commitment. The study team then recruited clinicians and staff directly through information sheets, presentations at staff meetings, and when permitted by site leadership, direct emails to staff.

The infrastructure survey was sent via REDCap to 31 participants. The response rate was 83.9% (26/31 surveys). The survey, which took approximately 60 minutes to complete, included questions related to the clinic’s infrastructure (eg, number of employees, staff turnover), health information technology (eg, functionalities of data tracking systems used to capture physical health information), existing general medical services (eg, care management activities), and consumer demographics (including estimates of CVD risk factors).

The in-depth interviews, which lasted approximately 60 minutes, were conducted by a team of 7 trained study team members using a videoconferencing platform (Zoom). We interviewed a total of 94 participants (mean, 3.2 per site) across 29 sites between December 2021 and June 2022. The types of roles represented included 26 nurses, 44 frontline staff, 17 administrative leads, and 7 other professional roles. The semistructured interview guide was based on the Consolidated Framework for Implementation Research²⁸  and iteratively modified during the data collection process. Key topics included anticipated barriers and facilitators to implementing the IDEAL Goals intervention, current workflows and practices related to clients’ physical health, health information technology, and providers’ perceptions of the feasibility of implementation strategies. Interviews were audio-recorded and transcribed.

Engaging with stakeholders through the needs assessment served 3 main purposes (Table 1). First, a rapid analysis of the data identified several key barriers (eg, challenges in communicating with primary care providers, high amounts of staff turnover) and potential solutions (eg, protocols for communication with primary care providers, clear data collection standards) that could be further discussed in community working groups. Second, preliminary insights into implementation barriers and facilitators, as well as current workflows and practices around CVD risk factors, informed the intervention experts’ initial adaptations of the IDEAL Goals intervention. For example, participants across organizations voiced issues with poor communication within their own team and with primary care physicians and their offices; as such, the intervention now incorporates methods to improve communication between site nurses and frontline staff as well as between mental health and primary care providers. Analysis of the qualitative interviews has also informed adaptations to the implementation strategy approaches. In response to feedback regarding time constraints for training, IDEAL Goals training will include options for synchronous and asynchronous learning as well as a “test out” option. Additionally, interviewees repeatedly shared challenges related to high staff turnover; to ameliorate this issue, we plan on developing online training modules to rapidly onboard new staff, designating a champion at each site, and training the maximal amount of staff.

We also plan to conduct additional data analysis to more fully understand organizations’ capacity and willingness to implement the program. For the survey data, this will involve generating descriptive statistics of organizational characteristics and client demographics. For the interview data, a multidisciplinary coding team will use a hybrid inductive-deductive approach to synthesize recurrent themes related to perceived implementation barriers and facilitators.

We next engaged in community working groups, a form of structured focus groups, to collaboratively work with end-users to adapt the IDEAL Goals intervention and the implementation strategies planned for the upcoming trial. Based on principles of community-engaged research and cross-functional teams, the overarching aim of community working groups is to provide a structured opportunity to promote information sharing across end-users and shared decision-making between researchers and community stakeholders. Consistent with prior recommendations to include a comprehensive range of stakeholders in the process,⁹  we convened community working groups with the intended group of implementers (i.e., clinicians and staff from community mental health organizations) and with the intended recipients of the intervention (i.e., consumers with serious mental illness). Our goal was to ensure both the implementers and recipients were actively involved in tailoring the intervention and implementation strategies.

Community working groups were held as structured focus groups with (1) clinicians and staff representatives from 22 community mental health programs in Maryland and Michigan, and (2) consumers with serious mental illness from a subset of 5 of those programs.

For the clinician and staff community working groups, we invited stakeholders who participated in the in-depth interviews and expressed an interest in being involved in future community working groups. We also invited staff nominated by site leadership who were considered opinion leaders at their sites. Our study team recruited participants via emails, information sheets, and virtual meetings related to the project. Our target was to involve 1 or more representatives from each site.

For the consumer-focused community working groups, we identified 5 geographically diverse sites (2 in Maryland, 3 in Michigan) who expressed a willingness to participate. In recruiting consumer participants, our study team communicated about the community working groups with site leadership and then distributed study information to the sites for them to share with potentially interested consumers.

The community working groups for clinicians and staff were 1-hour, virtual meetings hosted on the Zoom platform to accommodate stakeholders across multiple organizations and 2 states. In an effort to start developing long-lasting relationships with and between stakeholders, we planned a series of 5 meetings, held once a month between November 2022 and March 2023. To promote attendance, the meetings were scheduled between 12 and 2 pm, a time window that largely aligned with frontline staff’s workflows (recognizing that administrative leaders and nurses tended to have more control of their schedules). To provide as much advance notice as possible, we also sent a timeline and calendar invites for all 5 sessions at the start of the community working groups.

The structure of the community working groups for clinicians and staff (Table 2) was designed to accommodate both information sharing at the group level and active discussions and decision-making at the small group level. Each session began by convening the entire group (mean, 24 participants per session). During this time, the study team welcomed participants (∼5 minutes) and provided a brief presentation (∼10 minutes) to provide foundational information on the topic of the day.

The larger group was then divided into smaller Zoom “break-out” rooms, each staffed by a moderator and notetaker from our study team, and an average of 5 community stakeholders. The smaller group size served 2 purposes. First, it is a best practice of focus group discussions to convene between 6 and 8 individuals, in order to be small enough to ensure that everyone has the opportunity to speak and large enough to foster a group dynamic that is most conducive to discussion.²⁹  Second, the smaller groups provided a flexible structure to accommodate a larger number of topics and the ability to adjust the composition of the group based on the topic. For example, if a topic was specific to particular stakeholder group (eg, a nursing workflow), we had the ability to adjust the composition of the break-out room to include those with first-hand knowledge of the topic. For most break-out rooms, however, we intentionally maximized the diversity of roles and sites represented, as much of the intervention relies on interdependent work between all of the professional roles.

The majority of time (∼30 minutes) was spent in break-out rooms, where the “work” of the community working groups was largely done. During this time, there was an overarching theme for the small group discussion, which was explored through a semistructured discussion guide. Questions largely focused on eliciting information not adequately explored during the in-depth interviews, the perceived acceptability and feasibility of the IDEAL Goals intervention and implementation strategies, and ideas for potential adaptations to the intervention’s packaging and delivery. When relevant, we shared sample materials (eg, leaders’ guides, consumer notebooks) to ask for concrete feedback about participants’ preferences.

In the last 10 minutes, we reconvened the entire group to engage in a round of “lightning” report-outs, in which the notetaker from each break-out room provided a brief overview of what the small group discussed. Our goal was to provide an opportunity for information-sharing across the break-out rooms, so that all the stakeholders in attendance could have a general sense of what was discussed. A written summary of highlights was shared after the meeting, so that those unable to attend would be kept informed.

We concluded the meeting with a brief (<5 minute) “pulse survey.” The survey included participant characteristics; the brief version of the Research Engagement Survey Tool,³⁰  a validated survey used to assess the level of partner engagement in research studies; and measures of perceived acceptability, appropriateness, and feasibility of the intervention and implementation strategies.³¹ 

The community working groups for consumers with serious mental illness were 1-hour, in-person meetings held with an average of 8 participants per meeting. Following a brief presentation about the intervention, most of the meeting was devoted to eliciting specific feedback on consumer-facing educational materials and participants’ experiences with the clinic’s current work to improve heart health.

Of note, our decision to hold separate working groups for consumers was rooted in a desire to create a “safe space” for participants to share their candid perceptions about what was, and was not, working well. In the future, we plan to convene community working groups that include both providers and consumers in the same group, with careful attention paid to providing consumers with an appropriate orientation (eg, research and communication training) and establishing norms that foster a sense of psychological safety to speak freely.

In the short-term, we will continue to refine the intervention package based on partner feedback to increase flexibility in how the intervention is delivered (eg, allowing for both group- and individual-based delivery of the psychoeducational content), enhance the acceptability of the intervention’s packaging to both providers and consumers (eg, removing italics to promote ease of reading, using more color), and embed more structured opportunities to strengthen communication between providers, both within and outside of the organization. We will also continue to adapt the implementation strategies, including increasing the level of technical assistance offered to sites wanting to further develop or adapt their data systems to better track and use physical health measures.

In the long-term, we aim to contribute to the evidence based on the impact of engagement activities on implementation outcomes, first by examining the relationship between the level of partner engagement in research and their perceptions of the acceptability, appropriateness, and feasibility of the intervention over time. Second, we plan to systematically describe the proposed adaptations to the IDEAL Goals intervention and implementation strategies, using the Framework for Reporting Adaptations and Modifications – Expanded (FRAME)³²  and FRAME-Implementation Strategies,³³  respectively. We plan to use these frameworks to guide documentation of the content and context of what is being modified and why.

To successfully scale interventions that can reduce CVD risks in persons with serious mental illness, it is important to involve community partners in adapting interventions and implementation strategies to better fit local needs and increase ownership in the implementation process. In this article, we systematically describe the process we used during the preimplementation phase of REP to engage with community partners to adapt IDEAL Goals, a CVD-risk reduction intervention, and implementation strategies that will be used to support intervention uptake in an upcoming hybrid implementation-effectiveness trial. As we look ahead to the trial, we will continue engaging with our community partners through ongoing community working groups focused on refining the program to maximize uptake, and ultimately, reduce cardiovascular risk in persons with serious mental illness.