From Theory to Patient Care: a model for the development, adaptation, and testing of psychosocial interventions for patients with serious illness.

Abstract

Psychosocial and supportive care interventions are a cornerstone of palliative care science, yet there is little published guidance regarding how to develop, test, adapt, and ultimately disseminate evidence-based interventions. Our objective was to describe the application of a single intervention-development model in multiple populations of patients with serious illness. Specifically, we use the “Promoting Resilience in Stress Management” (PRISM) intervention as an exemplar for how the Obesity Related Behavioral Intervention Trials (ORBIT) intervention-development model may be applied to: (1) create an initial palliative care intervention; (2) adapt an existing intervention for a new patient-population; (3) expand an existing intervention to include new content; and, (4) consider dissemination and implementation of a research-proven intervention. We began by identifying key psychological and social science theories and translating them a testable clinical hypothesis. Next, we conducted observational studies and randomized trials to design, refine, and standardize PRISM within unique patient-populations. We moved backwards in the ORBIT model when necessary to adapt or expand PRISM content and delivery-strategies to meet patient-reported needs. Finally, we began to explore PRISM’s effectiveness using Dissemination and Implementation research methods. Key lessons include the need to ground intervention-development in evidence-based theory; involve patient, clinician, and other stakeholders at every phase of development; “meet patients where they are at” with flexible delivery strategies; invest in the time to find the right scientific premise and the right intervention content; and, perhaps most importantly, involve an interdisciplinary research team.

INTRODUCTION

Psychosocial, communication, and symptom-targeting interventions are critical tools for achieving the broad scope of palliative care goals of alleviating suffering and improving quality of life. Thus, intervention science is a cornerstone of palliative care research. While gold standards for intervention development and testing have been described,¹ there are few examples of how to apply those standards in palliative care research. Our objective was to describe the application of a single intervention-development model in multiple populations of patients with serious illness. Specifically, we use the “Promoting Resilience in Stress Management” (PRISM) intervention² as an exemplar for how the Obesity Related Behavioral Intervention Trials (ORBIT) intervention-development model¹ may be applied to: (1) create a novel psychosocial or palliative care intervention; (2) adapt an existing intervention for a new patient-population; (3) expand an existing intervention to include new content; and, (4) consider dissemination and implementation of a research-proven intervention.

METHODS

In 2015, intervention-scientists from the National Institutes of Health Office of Behavioral and Social Sciences Research, the National Cancer Institute, the National Heart, Lung, and Blood Institute, the National Institute of Child Health and Human Development, the National Institute of Diabetes, Digestive, and Kidney Diseases, and the Obesity Related Behavioral Intervention Trials (ORBIT) Consortium created a cross-disciplinary collaboration to develop a now widely-used model to accelerate the development and implementation of evidence-based programs designed to improve well-being among patients with chronic illness.¹ The ORBIT model was created to fill a gap in intervention science; at the time, there were no established pathways for the translation of behavioral/social science theory to health-related interventions that could be tested in reproducible, rigorous ways.¹ In other words, there was no equivalent translational pathway from pre-clinical to phase 1, 2, 3, and 4 clinical research as exists in pharmacologic development. Additionally, most published clinical intervention studies focused on feasibility, acceptability, preliminary- and definitive-efficacy; comparatively little described the earlier phases of intervention-development, nor how to implement interventions thereafter.

The ORBIT model provides a framework for flexible, progressive, and iterative steps, with pre-specified milestones for a new translational science pathway (Figure 1).¹ Steps include: (1) translating psychological and/or social science theory to a scientific clinical question (the “pre-clinical” phase of the pathway); (2) defining and refining intervention content (“Phase 1”); (3) establishing feasibility, acceptability, and proof of concept (“Phase 2”); (4) establishing efficacy (“Phase 3”); and, (5) establishing effectiveness (“Phase 4”). Built into the model are opportunities to move backwards, as needed, to meet newly-identified patient needs or expand/contract intervention-content before re-testing it in successive phases of study.

Figure 1. Key questions of intervention development.

Legend: Adapted from the ORBIT Model for the development and implementation of evidence-based programs to improve well-being among patients with serious illness (Czajkowski SM, Health Psychology, 2015)

We applied the ORBIT model to develop and test the Promoting Resilience in Stress Management (PRISM) intervention.^2–10 PRISM is a brief, disease-agnostic, skills-based intervention designed to promote “resilience resources” – positive psychological and coping skills such as stress-management, goal-setting, cognitive-reframing, and meaning-making – in turn, promoting self-perceived resilience, reducing distress, and improving quality of life (Table 1).² PRISM follows gold standards of intervention science, including a detailed manual for content and delivery by trained interventionists, and procedures for monitoring intervention-fidelity.¹¹ It is founded on resilience and stress-and-coping theories,^12–14 has been designed and refined for various populations (ORBIT Phase 1), has been tested in proof-of-concept and pilot efficacy studies (ORBIT Phase 2), is the subject of several ongoing large multi-site clinical trials (ORBIT Phase 3), and is being positioned for effectiveness testing (ORBIT Phase 4). Here, we describe the methods involved in each of these steps, including key points for rigor and reproducibility.

Table 1.

PRISM intervention content

Session Focus	Details	Format
1. Managing Stress	Relaxation strategies (e.g., deep breathing, guided imagery); Mindfulness techniques (e.g., being present in the moment, becoming aware of stressors without judgement)	1:1 (in-person or video)
2. Goal-setting	Setting Specific, Measurable, Actionable, Realistic, and Time-Dependent (SMART) goals, planning for roadblocks, exploring what is controllable and what is not	1:1 (in-person or video)
3. Positive Reframing	Recognizing negative self-talk (e.g., thoughts that keep us up at night, self-destructive, or catastrophic thinking), reframing perspectives to make them realistic, more manageable, and (if possible) optimistic	1:1 (in-person or video)
4. Meaning Making	Identifying gratitude, meaning, and/or purpose from illness experience	1:1 (in-person or video)
“Coming Together”	Discussion about what was learned, what helped, what other loved ones can do to help	Family meeting (in-person or video)
Boosters	Reflections on skills, opportunities for refreshers and further development of skills	1:1 (in-person or video)
Skill Practice	Between-session exercises to practice, further develop and track skills	Digital App (or paper/pencil)

Legend: PRISM = Promoting Resilience in Stress Management. Sessions are delivered by certified lay-coaches. The program manual details scripts for delivery, and all PRISM-coaches receive at least 8 hours supervised training, including role-plays, to ensure competency and fluency, plus ongoing mentoring and fidelity-monitoring throughout delivery of the program. Sessions schedules vary by illness-population and patient needs, generally every other week for approximately 30 minutes to 1 hour per session. Sessions may be delivered in person (e.g., at bedside) or via video-conference, depending on circumstance and recipient preference. In research studies, all interventions are monitored for fidelity using a standardized tool.

RESULTS

Question #1: How do I translate theory to an initial design?

When we set out to create the PRISM program, we observed that most clinical interventions target pathology. Healthcare providers are trained to see the negative and exhorted to “fix” it. We wondered what would happen if we turned that paradigm upside-down? What if we also tried to bolster individual resilience and buffer the adversities of illness? Would doing so alleviate suffering and improve quality of life?

Step 1: Find the right theory.

Successful interventions must be grounded in, and informed by, empirical theory or conceptual frameworks (Figure 1).¹ In our case, decades of work in behavioral and social sciences suggested that theories of individual resilience were inconsistent.^12,15–17 Resilience had been defined as a (perhaps immutable) trait, such as grit or hardiness.^18,19 It had been defined as an outcome, or relatively positive state of being, after a particular stressor.^17,20–22 Here, resilience was often dichotomized as the presence of a (perhaps surprising) positive outcome, such as post-traumatic growth, or the absence of a (perhaps expected) negative outcome, such as depression. It had been defined as a process of (usually positive) adaptation, where resilient people were better able to withstand the stressors of adversity.^12,23–25 Defining resilience as a trait, outcome, or process had implications for intervention development. If resilience were an immutable trait, then interventions might involve risk-stratifying people based on personal characteristics like grit. If resilience were an outcome, then we could only wait to see where people ended up, and then target pathology directly. In contrast, defining resilience as a process provided opportunities for intervention; if we determined how and when to optimize that process, then an intervention might have impact.

There were several established resilience-as-a-process theories, and all had merit. We selected one that both resonated with what we observed clinically and could be operationalized. Specifically, we leveraged a theory that resilience is a process of harnessing resources to sustain well-being in the face of adversity.^12,26 We liked this construct for 4 reasons. First, it was grounded in cultural anthropology, with established generalizability to diverse individuals and communities. Second, it was responsive to other theories of resilience; an individual’s resources could include both grit (a trait) and learned coping-skills (a process), thereby integrating previously exclusive theories. Third, this conceptualization recognized and embraced individual differences. Rather than a one-size-fits-all definition, it acknowledged that some resources work for some people, while others work for others. Last, it suggested a discrete opportunity to identify and promote common resilience resources in a manualized intervention.

Next, we needed a theoretical foundation to support the idea that promoting common resilience resources would, in fact, change outcomes. Here, we leveraged Stress-and-Coping theory, which posits that personal appraisals influence outcomes.^13,14 If a person appraises an illness as catastrophic and overwhelming, for example, their subsequent emotional, physical, and functional outcomes tend to be poor. If the same person appraises their illness as manageable, then they tend to be more functional in the long-run. We hypothesized that self-perceived resilience would work the same way: if someone believes they have the resources to be resilient in the face of adversity, does that appraisal translate to better health and psychosocial outcomes?

Step 2: Formulate a precise scientific clinical question.

We leveraged our two identified theories (resilience theory: “resilience is a process of harnessing resources to sustain well-being,” and stress-and-coping theory: “changing appraisals of one’s resilience impacts outcomes”) to formulate a question that could be answered in subsequent, hypothesis-driven research: Does a program that teaches common resilience resources change self-perceptions of resilience, and in turn, improve downstream emotional, physical, and functional outcomes?

Step 3: Collect data to define associations and intervention content.

After formulating a scientific question, the ORBIT model directs intervention-developers to Phase 1 of the translational pathway, where they design a preliminary program (Figure 1).¹ This first involves defining associations of interest and intervention content (“Phase 1a”). We conducted cross-sectional cohort studies among parents of children with cancer,¹⁶,²⁷ patient-survivors of cancer,²⁸ and patients with diabetes^29,30 to confirm that self-perceived resilience was associated with outcomes like psychological distress, quality of life, functional status, and health behaviors. We also conducted qualitative studies with parents of children with cancer¹⁶ and adolescents/young adults (AYAs) with cancer³¹ to explore their own perceptions of resilience, how they identified that resilience, and what their illness-experiences had taught them about facilitators and inhibitors of resilience. We confirmed our selected theories and identified 4 commonly endorsed resilience resources: skills in stress-management, goal-setting, positive reframing, and meaning-making.^2,31

Step 4: Refine program content.

Once intervention theory, associations, and content are defined, intervention-developers refine content (“Phase 1b,” Figure 1).¹ We did this in a stepwise process. First, we reviewed the literature for existing interventions targeting any of our target resilience resources. We studied successful and unsuccessful methods and delivery strategies. We considered if and how successes might be adapted for our populations of interest, and we noted barriers to success including time-commitments and human resources. We decided we wanted to create a simple, scalable, cost-effective, disease-agnostic program that could be delivered to multiple illness-populations by trained lay-staff.

Next, we developed a curriculum to teach each of the 4 intervention modules, relying on our interdisciplinary team’s expertise in cognitive behavioral therapy, skills-training, and coaching. We then presented the preliminary program script to interdisciplinary stakeholders, including physicians, nurses, psychologists, social workers, health services researchers, patient caregivers, and AYA patients. With their feedback, we refined the content and drafted a full curriculum “manual.”

In the final phase of refinement, we conducted an iterative, cognitive think-aloud with consecutive small (N=5) cohorts of AYAs with cancer (Figure 2). In a one-on-one session with each AYA, we delivered all PRISM modules and then presented the full manual. AYAs provided “free-thoughts” with each section of the manual, reflecting on the scripted words and their experience receiving the intervention. Investigators recorded all AYA comments in an annotated manual. After 5 AYAs completed the process, we reviewed feedback qualitatively for acceptability, appropriateness, and suggestions for major revisions. Consistent and significant suggestions prompted revisions and a new draft manual. We then repeated the process with another N=5 AYAs and continued until feedback was minor. In our case, we completed the process after 2 cohorts (N=10) AYAs.

Figure 2.

Framework for Iterative Patient-Centered Iterative Refinement

Step 5: Establish procedures for reproducibility, including an intervention “manual” with instructions for training, delivery, and fidelity-monitoring.

Finally, we expanded the manual to include details for program delivery and certifying PRISM coaches. The latter involves at least 8 hours of supervised training, including role-play scenarios, to demonstrate competency and fluency. The manual includes procedures for monitoring program fidelity,¹¹ procedures for continued education, staff-support, and trouble-shooting guidance from a senior coach throughout the intervention, suicide prevention training, resources for additional skills-training and practice, plus references to research supporting the intervention-design. This, too, was shared with stakeholders in psychology, social work, medicine, and nursing. The training manual and fidelity monitoring tool were pilot tested with 5 college graduates and 3 PhD psychologists to confirm understandability and content-completeness.

Proof-of-Concept and Pilot Studies.

There is a rich literature describing the “how-to” and challenges of proof-of-concept and/or preliminary efficacy pilot studies.^32–35 Because these are more common within palliative care research communities, we elected not to focus on this ORBIT Phase 2 here (Figure 1).¹ Briefly, to confirm PRISM’s feasibility, acceptability, and proof-of-concept, we conducted a single-arm cohort study among N=24 AYAs with either cancer or type 1 diabetes.² We found PRISM to be feasible (80% of those approached enrolled and completed the intervention) and valuable (100% of recipients recommended it to other patients like them, feedback was universally positive). Before- and after-scores in patient-reported outcomes of resilience and distress suggested resilience went up and distress went down with PRISM.

An important observation was that AYAs with cancer preferred the intervention during planned hospital stays, while AYAs with diabetes preferred modules delivered via video-conference when they were well and at home. Thus, we adapted the manual to indicate that different populations might have different delivery-preferences, and that coaches should explore preferences throughout the intervention in order to “meet patients where they are.”

Next, we conducted a Phase 2 randomized clinical trial (RCT) of N=92 AYAs with cancer and found that PRISM was associated with improved patient-reported resilience, hope, quality of life and distress at 6-months, compared to usual care.^4,9 Finally, we proceeded to launch and are currently conducting Phase 3 randomized trials testing PRISM’s efficacy in populations of AYAs with cancer or diabetes (Clinicaltrials.gov NCT03847194, NCT03640325).

Question 2a: How do I adapt existing content for a new population?

During the Phase 2 RCT, local interest in PRISM sparked two new initiatives: PRISM-adaptations for AYA patients with Cystic Fibrosis (CF) and for parents of children with cancer. In such cases, the ORBIT model prompts investigators to move backwards and revisit Phase 1 to confirm (or redefine/re-refine) associations and content.¹ In the case of AYAs with CF, for example, we confirmed that our disease-agnostic modules and content were still feasible, acceptable, and appropriate.⁸ Because the AYA CF population was similar to the AYA oncology population and because the program was relatively unchanged, we proceeded directly to the design of a Phase 3 clinical trial testing PRISM’s efficacy among AYAs with CF (trial-design currently in progress).

Among parents, Phase 1 iterative refinement (Figures 1 and 2) again suggested the program and content were feasible and acceptable.¹⁰ What was different was that parents suggested a half-day symposium delivery-style might be more feasible with caregiving responsibilities. Thus, we adapted the program to include a group workshop delivery option and compared it and our traditional one-on-one platform to usual care in a Phase 2 RCT.⁶ We found that both the one-on-one and group-platforms were highly acceptable, and only one-on-one delivery was efficacious at improving parent-reported resilience compared to usual care.^6,36 In the future, options for continued development of the PRISM-parent program might include moving forward on the ORBIT pathway to test the one-on-one program in a Phase 3 RCT, or moving backwards to re-define/refine the group program to augment its efficacy.

Additional new Phase 1 work for PRISM has continued in populations as diverse as AYAs with chronic musculoskeletal pain, youth with craniofacial anomalies and their parents, adults with congenital heart disease, and both healthcare workers³⁷ and school-aged children navigating the COVID-19 pandemic. In each case, investigators are working directly with their populations of interest to define/redefine associations of interest and content, with longer term plans to test their program in clinical trials. To date, we have found that PRISM’s content remains the same in each population, and there are important nuances in population-preferred language, timing, and approach.

Question 2b: How do I build new content into an existing intervention?

In our work with AYAs with advanced cancer or CF, several suggested opportunities to expand PRISM’s content. “You just helped me figure out what is important to me, and what goals I want to set…now I need help talking about those things with my doctors and family,”³⁸ Recognizing an opportunity to integrate concepts of early Advance Care Planning (ACP), something rare in AYA oncology and AYA CF care,^39,40 we partnered with the team at agingwithdignity.org to develop content based on the AYA-validated ACP guide, Voicing My Choices™. Again revisiting ORBIT Phase 1 (Figure 1)¹ and partnering with patient, parent, and clinician stakeholders, we identified 4 priority pages within Voicing My Choices™ and developed language to demonstrate how they were direct applications of PRISM skills.³⁸ We conducted a Phase 2 feasibility and acceptability study of the expanded program,³⁸ and are now conducting a Phase 3 trial testing its efficacy (Clinicaltrials.gov NCT03668223).

We have continued to identify opportunities to expand PRISM’s content. In all cases, the primary scaffolding of PRISM’s 4 core sessions is a necessary building block for additional population- or need-specific skills. Ongoing work includes modules to support adult patients who are Black, Indigenous, or People of Color who experience racism in healthcare, and AYAs with serious illness who are struggling to maintain social support and normalcy.

Question 3: What happens when I put the intervention into patient hands?

After establishing efficacy in Phase 3 trials, the ORBIT model suggests investigators evaluate effectiveness (Figure 1).¹ In other words, without the controlled parameters of a clinical trial, what happens when an intervention is put into clinician or patient hands? What factors enable its successful integration into clinical care? These questions are important because fewer than 50% of interventions developed in research-settings make it to clinical practice, and the average research-to-practice timeline is 17 years.^41,42 Dissemination and Implementation (D&I) research answers these questions by evaluating and optimizing a program’s reach (e.g., uptake), effectiveness (e.g., impact in uncontrolled settings), adoption (e.g., support from organizations and payors), implementation (e.g., degree of protocol fidelity in real-world settings), and maintenance (e.g., sustainability).^43–45

Recognizing a need to consider multiple strategies for PRISM’s D&I, we designed 2 projects. First, we launched and are currently conducting a pilot implementation study with various clinical programs at Seattle Children’s Hospital where we offer PRISM as standard of care. We are tracking its adoption, delivery, and use, and monitoring barriers and facilitators to its implementation. To date, a key lesson-learned is that clinical programs strongly desire PRISM and lack funds or personnel to deliver it. One of our working conclusions is that PRISM’s adoption, implementation, and sustainability will require cost-effectiveness data to support hospital administration decisions (now being collected in Phase 3 RCTs), scalable coaching opportunities (we created an affordable “pool” of coaches for remote video-coaching across clinical programs), and delivery modes that demand fewer, if any, human resources.

To expand delivery options and minimize human resources, we developed a now award-winning mobile app version of PRISM (mPRISM) which teaches each skill with a scripted interface and embedded practice opportunities (Table 2).⁴⁶ What remains unclear is the degree to which mPRISM also requires human coach-contact. While it is attractive to imagine a stand-alone program that does not rely on human resources, our preliminary (ORBIT Phase 1 app-development) experience is that AYA patients like the app, use it initially, and then gradually stop using it. This is consistent with previous literature suggesting that roughly half of smartphone owners who download health-related apps stop using then due to perceived burden, lost interest, or other reasons.⁴⁷ AYA patients report being more engaged and satisfied when they have a personal coach available in person, by video, or by text. We are designing studies assessing the comparative efficacy of mPRISM with varying degrees of human versus digital interaction. While PRISM’s long-term impact remains a question of interest, one thing is clear: soliciting stakeholder feedback is necessary at every step of intervention-design.

Table 2.

Selected examples of session skills and practice materials embedded in early PRISM “app”

DISCUSSION

Intervention science is critical in palliative care research, and few models exist to guide us in how to rigorously and reproducibly design, test, implement, and disseminate those interventions. For the past decade, our team has relied on the ORBIT model¹ to develop the PRISM intervention. To date, we have worked in every phase of research, and we continue to learn. Key lessons learned are: (1) Invest time to find the right scientific premise. A key to our success is that we grounded our work in evidence-based theory. (2) Patient, clinician, and other stakeholder involvement is critical at every phase. If the recipients don’t like the program, they won’t use it. (3) Meet patients (recipients) where they are. Flexibility to go back and forth on the ORBIT pathway, opportunities to meet emergent unmet needs, and willingness to adapt delivery strategies are key. (4) Strong, well-designed content endures. We have explored the PRISM program with over a dozen different populations, and the core content has not changed. And, (5) Having an interdisciplinary research team is necessary. Each stage of the pathway demands different expertise, and having a collaborative group of physicians, psychologists, social workers, methodologists, digital health designers, and others has been key.

We recognize that our single experience may not be generalizable to other investigators, populations, or programs. Additional threats to generalizability include the facts that we focus on a single construct and have worked with relatively medically stable and cognitively able patients. A key limitation of our work is that it is still on-going; we cannot share the successes or failures of our Phase 3 and Phase 4 endeavors. It may be that our application of the model falls short at these critical phases.

Additional challenges (not addressed in the ORBIT model) are: (1) When can investigators forgo certain steps on the pathway? For example, if we continue to see that PRISM’s content does not change, how much longer must we re-redefine/refine in Phase 1? This question is complicated because grant funders may expect to see preliminary population associations of interest and qualitative work to confirm appropriateness. Such development work is costly, hard to fund, and limits critical clinical investigation and innovation. (2) When is it best to adapt an existing program versus start anew? When interventions are constructed from a single, cohesive group of researchers, there may be a conservative effect; it becomes harder to re-imagine a different set of concepts and tools. Although time and resource intensive, there are times where it may be valuable to start from the basics with brand new perspectives. As the field becomes more experienced with the creation, adaptation, and application of evidence-based interventions in new populations, we hope that these norms and expectations become more nuanced.

For now, we hope to share the breadth of possibilities we have experienced with our program development. By applying the ORBIT model, we have taken PRISM from theory to patients and back again. We look forward to continued learning together in our palliative care research community.

KEY MESSAGE:

Palliative Care intervention-science is critically important. We describe the application of a gold-standard intervention-development model to the creation and study of a psychosocial intervention for patients with serious illness. Lessons-learned include the need to identify the right evidence-based theory, develop content directly with patients, and work with an interdisciplinary team.