Applications in human-centered design: Shared-Decision Making for mental health treatment in primary care

Abstract

Objective:

Shared Decision Making (SDM) is heralded as a standard for patient-centered care, but implementation of SDM in routine mental health practice has proven difficult to achieve. Human-centered design (HCD) may hold promise for improving SDM implementation in busy clinical settings. This study describes applying HCD to develop an SDM documentation support tool intended to encourage successful use of SDM by mental health clinicians.

Methods:

This descriptive, proof-of-concept study utilized the Discover-Design-Build-Test HCD framework to simplify a comprehensive SDM protocol for mental health decision making. Implementation was piloted within multiple primary care clinics. The study consisted of three phases: information gathering (interviewing clinicians), solution generation and prototyping, and testing a final prototype in routine care settings.

Results:

Our project proceeded through eight cycles of user design and feedback. Clinicians pilot tested the final product, a documentation note template incorporating SDM prompts and explanations. It is currently available for clinical use.

Conclusions:

Clinicians were able to use the HCD-redesigned SDM documentation note template intuitively, i.e., without explicit instruction. Leveraging buy-in from users throughout the entirety of the process (from problem investigation to solution discussions) created opportunities to tailor implementation strategies and may support ownership of the end-product by primary stakeholders.

Practice implications:

HCD may be a promising methodology for streamlining the adoption of complex clinical tasks like SDM.

1. Introduction

Rates of mental health conditions are high among military veterans [1]. The Veterans Health Administration (VHA) is one of the largest healthcare providers in the United States and operates a robust, nationwide integrated mental healthcare system focused on patient-centered and evidence-based care [2]. A cornerstone of patient-centered mental health care is shared-decision making (SDM) [3], defined as a patient-centered approach where “clinicians and patients share the best available evidence and patients are supported to consider options to achieve informed preferences.” [4–7]. SDM improves communication outcomes, engagement, and retention in mental health treatment [8,9]. VHA clinical practice guidelines encourage clinicians to adopt SDM as part of mental healthcare decision-making [10]. VHA policy also utilizes peer review to monitor clinical documentation of SDM in initial mental health visits as part of ongoing quality improvement, although there is no standardized approach to SDM documentation so the quality of these peer reviews is unknown. A systematic review of randomized clinical trials and non-randomized cross-sectional studies identified inconsistencies in how SDM is measured or operationalized (e.g., audio recordings, patient-reported involvement in decision making, patient decision-making scales) as a major barrier to assessing SDM’s effects [11].

While SDM is regarded as central to patient-centered care [12] and a standard of care within VHA, implementing SDM in routine practice has been a challenge [13]. Implementation barriers have included a lack of organizational resources to support SDM, a lack of time in patient encounters, and a dearth of high-quality resources that tailor information about the condition and/or treatment options to the patient [13,14]. Specific to the mental health setting, a prior study of VHA mental health providers found that providers supported the integration of high-quality decision aids into mental health visits, but they expressed low confidence that they could do SDM within the time constraints of existing visits [15,16].

Human-centered design (HCD) may hold promise for improving the implementation of SDM [17–19]. Human-centered design is the study of developing solutions by investigating users, their requirements, interacting systems, and competing needs within an identified environment [17,20,21]. HCD, also known as user-centered design, is a problem-solving framework that develops solutions by engaging users (the people directly impacted by the problem) in every phase of the project. Work in HCD has three distinct phases: 1) determine the user needs and problem characteristics, 2) develop prototypes of solutions based on ideas generated in the initial phase, and 3) test solution prototypes and refine based on user feedback experience [20–22].

HCD can be used to tailor SDM and implementation strategies for different clinical environments. In healthcare, HCD has been used to reduce no-shows for scheduled appointments [23], craft better decision support tools for practitioners [24], develop new patient decision aids and tools [25], and improve implementation strategies in adolescent health [26], psychotherapy [17], and clinical documentation systems [27]. While HCD shows promise for improving the implementation of evidence-based practices, few studies have employed the design process to implement SDM in routine care.

The present study’s goal was to create solutions that promote the adoption of SDM by mental health clinicians without creating more workload or burden [17]. The team applied HCD methods to mental health visits in primary care settings because (1) this is the first point of contact with mental health for most patients and often when treatment is initiated, and (2) HCD may be particularly needed to implement new practices in busy primary care settings. The present project serves as a small-scale use case for other HCD applications in the VHA or similar integrated health systems.

2. Methods

While there are several models detailing HCD, this project used a framework called Discover-Design-Build-Test (DBBT) [18] (Fig. 1). In the Discover phase, the HCD team engages users to explore all facets of the problem from the user’s point of view. Discussions with users narrow the scope of potential solutions based on what is most important to them. Observations of users in their clinical workflow helps identify needs that users could not articulate. Since design solutions are tailored for specific problems and environments, the Discover phase seeks to clearly define the problem characteristics and constraints.

Fig. 1. The Discover-Design-Build-Test framework.

[18]. Each phase of the DDBT process has its own goals.

In the Design-Build phase, the HCD team develops prototypes of solutions and works closely with users in a back-and-forth manner to refine these prototypes. The goal at this phase is to provide many possible solutions to users to find the ones that best reduce work burden and evoke the least resistance to change.

In the Test phase, users test the most promising prototype(s) from the Design-Build phase in routine care settings and continue to provide feedback. The HCD team further modifies the prototype(s) based on users’ “test drive” experiences. While the DDBT framework describes HCD in a linear fashion, the reality is that design is cyclical and iterative. After one cycle of prototype, feedback, and testing, additional cycles generally follow.

Our project was divided into 3 waves of participation. Our focus was to understand the workflow of mental health practitioners during an initial mental health visit, a 20-minute intake called the Functional Assessment. Functional Assessments were ideal for our project due to several factors. In a Functional Assessments, clinicians must complete a variety of screening and assessment tasks within a limited amount of time, including an evaluation of patient history, the reason for the visit, standard mental health measurements, suicide screening, and a treatment plan. This form of assessment is common for integrated care settings (mental health in primary care), and the Functional Assessment is based on a standardized protocol across the VHA [28]. Most importantly, the purpose of the Functional Assessment is to decide on a treatment direction, so it naturally centers on decision-making, an ideal fit for our use case of implementing SDM.

The HCD team recruited participants from a single, large healthcare system comprised of 1 tertiary care center, 1 secondary care hospital, and 9 community outpatient clinics. Within this healthcare system, there were up to 25 eligible clinicians who conducted Functional Assessments in integrated mental health (primary care) settings. Participants were recruited in three waves corresponding to the 3 primary HCD phases (Discover, Design-Build, Test), with a goal of recruiting 3–4 clinicians per wave: one set for discovery interviews, one set for designing/building, and one set for testing. For each wave, the team sought to interview providers with differing perspectives and utilized purposive sampling to invite providers across job classifications (e.g., psychiatrists, psychologists, nurse care managers, social workers) and clinic types (e.g., women’s health, rural clinics).

Eligible clinicians working in integrated mental health were contacted via email, sent a brief description of the study purpose and time commitment, and asked to reply if they were willing to participate in a 30-minute HCD interview conducted via video teleconference at a time of their choosing. Prior to the interview, participants were provided with an information statement, and at the time of the interview, they were asked to provide verbal consent and permission to audio record the interview. In addition to recruited providers, our team included two subject matter experts, also clinicians in integrated mental health settings, who provided guidance on interpreting findings and advised on design iterations. All procedures were approved by the local Institutional Review Board.

3. Results

Overall, 9 clinicians consented and participated: 4 providers in Wave 1, 4 providers in Wave 2, and 2 providers (1 returning) in Wave 3. The 2 subject matter experts provided input in all waves.

3.1. The discover phase

The initial project phase sought to achieve the following tasks: 1) understand the visit workflow and requirements of the Functional Assessment, 2) determine the active components of SDM as a protocol and lastly, 3) outline the needs of an ideal solution to support quick adoption of SDM into clinical practice. The first round of interviews (wave 1) gathered clinician perspectives of the visit: what information clinicians need before and during the visit, the context of the visit, and its goals. Clinicians also provided a walkthrough of the clinical work-flow, from referral to the actual visit to wrap up activities such as clinical documentation. As the HCD team probed clinicians on the structure of the visit, the team also presented a specific protocol developed in VHA for conducting SDM for mental health treatment, the Evidence-based Psychotherapy Shared-Decision Making Toolkit for Mental Health Providers and its included Provider Checklist [3], to see how it fit or did not fit within clinicians’ current practice. In discussing one specific approach to conducting SDM, clinicians revealed what level of experience they had with SDM generally and whether they believed SDM was occurring within each Functional Assessment. Interviews during the Discover phase were approached as open-ended questions with an emphasis on contextual inquiry. This technique allowed participants to explain their workflow in their own words, while the research team was able to probe further on granular details such as timing, logistics, etc. At the end, each interviewee was asked to share their documentation template or visit note so the HCD team could compare approaches and see where individual preferences aligned and differed.

After the HCD team analyzed the data and conversations from the Discover phase, it became apparent that clinicians valued the idea of SDM in Functional Assessments because they believed that patients should be informed about treatment options and should have a role in making treatment decisions. At the same time, few clinicians could clearly define what components comprise an SDM conversation, and even fewer could tell us with certainty that they consistently completed SDM within Functional Assessments. When presented with a specific protocol for SDM from the Shared-Decision Making Toolkit for Mental Health Providers [3], there was pushback from integrated care clinicians about its applicability to their settings. The suggested protocol was based on a 60- to 90-min visit, whereas a Functional Assessment visit is only 30-minutes long, with only 3–4 min available for SDM at the end. Through a series of discussions within the HCD team and with subject matter experts, the goal became to simplify the toolkit into a short visit framework and then focus on the “must-have” components of SDM. The eventual solution needed to educate providers about SDM along the way while fitting into the narrow timeframe allotted to patient discussion. Ideally, the solution required little to no tutorial and work within the routine workflows these clinicians had already established.

3.2. The design phase

In the Design phase, the HCD team built upon what was learned in the prior phase about SDM in practice – the lack of knowledge, the visit constraints – and generated solutions to address these issues. The Design phase is the most generative phase with multiple potential solutions brainstormed, discarded and/or refined at once. It was important that solutions did not increase work burden on a provider population with limited time and resources. Therefore, the team began to look at what tasks clinicians were already doing as part of the Functional Assessment to see if there was a step that could be easily modified. In design, intuitive solutions are understandable to users without additional instruction or training, and they typically involve the smallest change possible within established workflows to facilitate adoption and decrease resistance to the modified intervention.

During the Design phase, the HCD team considered multiple potential formats for education, including some technology-driven ideas (e.g., integrating new software, creating an SDM education curriculum). But, given the rigidity of the existing technological systems in this healthcare organization and the time constraints of providers, these solutions were not feasible to build and deploy on a large scale. The team ultimately settled on clinical documentation as the solution to introduce and educate providers about SDM. As most clinicians used their documentation templates to guide their visits, the team identified an opportunity to guide clinicians through SDM conversations by building prompts into documentation templates. Among all the solutions considered, leveraging the requirement of visit documentation was the least burdensome path forward. A schematic summary of the template development can be found in Appendix 1.

3.3. The build phase

Moving into the Build phase, the team built out the Functional Assessment in an electronic health record template format. The first step was to develop a Structured Agenda, an outline of all the tasks and informational sections that need to be addressed, start to finish, from the clinician’s perspective. Once a Structured Agenda was developed, the HCD team added in sections for SDM.

The feedback from Wave 2 participants was used to refine and slim down the agenda to fit within the timeframe of the visit. The Build phase is iterative; the prototype templates went through different stages, including several structured agendas and word documents formatted to “look like templates” (i.e., mock-ups). Initial versions included explanatory sections on SDM and options for dialogue; however, these explanatory sections made the template very long and clinicians were unlikely to read through the information. In subsequent rounds, the HCD team slimmed the SDM sections down into prompts. Our participants were provided a walkthrough of each prototype template for feedback, and their feedback was incorporated into the next prototype version. Each participant reviewed a new prototype and had not seen previous templates. This allowed for design feedback to stay fresh and uninfluenced. Given these rounds of user feedback and ongoing discussion, the HCD team reached a consensus regarding the essential template components, clinical information, length, flow and format. In all, 8 iterations of the SDM documentation template were refined with users before the final prototype was digitized into VHA’s electronic health record (EHR) and catalogued. The final template prototype married the requirements of the Functional Assessment with SDM visit goals. An overview of the HCD process for this project can be found in Table 1.

Table 1.

DDBT phases, waves of participant recruitment, tasks, and results from the HCD process.

Phase	Participants	Tasks	Results of the Design Process
Discover	Wave 1	• Assess the SDM protocol • Evaluate Functional Assessment workflow	Developed Visit Framework, a schematic flow of the Functional Assessment
Design	Wave 2	• Define constraints and parameters of the problem and solution criteria (boundaries) • Brainstorm potential solutions • User engagement	Identified Barriers: (1) a knowledge gap about SDM and (2) a perceived lack of time to have an SDM conversation Constraints: only 3–4 min available for treatment decision-making Parameters: avoid increasing work burden of clinicians (i.e., clinicians not available for additional training, which would take away from clinical care). Need to work within the parameters of the visit and existing technology. HCD goal: Utilize provider-centered design to support SDM with veterans. Solution: Build a template for visit documentation that guides clinicians through the SDM process with their patients
Build	Wave 2	• Build template prototypes and iterate based on participant [user] feedback. • The iterative design phase concludes when user feedback reaches a consensus on acceptability of the template into clinical practice.	Worked with users to: 1. Streamline the clinical workflow in the Functional Assessment to ensure 3–4 min for decision-making conversations. 2. Refine SDM prompts. Originally, the team included scripts to guide clinicians in SDM conversations. These scripts were removed due to user feedback stating they were too lengthy and unnecessary. 3. Technical capabilities were added based on user feedback. Specifically, the team added checkboxes for faster documentation, nested treatment options (e.g., after selecting one mental health condition, a dropdown list of evidence-based treatment options for that condition), and active links for useful resources such as condition-specific patient decision aids.
Test	Wave 3	• Participants test in environment	Participants used the new template for visit documentation. They provided usability feedback and parting thoughts. All agreed that the final prototype template could be added to the electronic health record catalogue and would be appropriate for wide use.

3.4. The test phase

During our final phase, the HCD team invited back participants that had reviewed a prototype in the build phase as well as new participants (Wave 3). Participants were guided to save the EHR template in their own personal library and use it for one week in the clinical setting. They could choose to use the template within real visits or test drive it with a mock patient. Participants were scheduled for debriefing sessions after a week to review their experience and thoughts. The HCD team purposefully did not provide a tutorial of the template prior to participant testing. This ensured that each participant generated their most important points for feedback without having a pre-conceived idea of what areas were important to the research team.

The final prototype included sections focusing on connecting to the patient, presenting tailored treatment options, discussing the options, exploring values and preferences, and informed decision-making. Our testing group (Wave 3) consisted of 4 active clinicians who each used the template in their current practice. Final feedback from this small group was overall positive. All individuals understood the SDM elements of the template without requiring additional training or instruction, which the team considered a success given the knowledge barrier identified during the Discover phase. Each tester had different preferences in terms of tweaking the format, length, technical functionalities (for example check box vs. free text), and how the template aligned with their established clinical routine. In terms of willingness to use the SDM template in place of their current documentation practice, 3 clinicians believed they could switch over easily (despite having comments of additional personal preferences) and 1 tester was not interested in replacing their current template with our prototype but would be interested in a short SDM-only addendum they could be added to their current template. Given that every clinician had their own preferences for clinical documentation and practice routines, the template was built to be editable and adapt to these individual preferences. It is expected that participants will want to further tweak the template to suit their preferred routine. Ultimately, all participants agreed the template was ready for clinical use and could be added to the existing EHR catalogue for wide availability. Fig. 2 displays the evolution of the SDM template from prototype to clinical use.

Fig. 2. The evolution of the SDM template prototype through HCD project phases.

A diagram showcasing the digital transformation roadmap, detailing the progression of the SDM Template Prototype through HCD project phases.

4. Discussion and conclusion

4.1. Discussion

The SDM template garnered overall positive feedback and provides a useful example of employing the HCD process to develop an implementable tool, while also leaving room to continue to innovate.

While the cyclical nature of HCD can feel daunting at first, users usually respond positively since it allows for their voices to be heard, without judgement or repercussions, throughout the entirety of the project [19,22]. Since HCD is ultimately founded on the principle of engagement first, this approach can help mitigate known issues found in traditional methods of implementation science, such as the mismatch of published strategies with identified barriers and obstacles [22]. Deploying HCD as an alternative to or in tandem with implementation science may help bypass some of the frustration felt by users. By engaging users in the development of new tools and solutions, the initial ideas are primed to match the needs of those working in the problem environment, reducing the chance that a solution will fail during implementation.

The present application of HCD to clinical workflow demonstrates a pathway for developing better and more innovative approaches to patient care. This can be as small as re-working a documentation process or as large as developing new services and care delivery methods. While it is well known that clinician behavior is not easy to change, HCD offers a universally adaptable framework to help address complex problems that do not have one-size-fits-all solutions but instead require solutions that are tailored to specific environments. The present study is a small, proof-of-concept that demonstrates the process of HCD and shows how groups can create a strategy for SDM that is tailored to their unique clinical environment.

Recently, the required adoption of VHA video visits to expand access to clinical care, particularly specialty care (during the pandemic and in underserved communities) met with similar resistance to change, echoing comparable required modifications to routine clinical practice akin to SDM adoption [29,30]. Predictably, the VHA saw a variety of strategies implemented to help clinicians adjust to the new visit types [31]. The success of each strategy depended entirely on the degree of best fit with a particular clinical environment. It is possible that if HCD methods were in play when the policy shift was conceived, ineffective strategies could have been eliminated earlier, conserving VHA resources. This offers yet another example of how HCD can help clinicians adapt to changes quickly to innovate towards better patient-centered care and improve learning health systems for the future.

4.2. Limitations

There are considerable limitations to the present study, which was a small, proof-of-concept pilot. One major limitation is that the study engaged only frontline clinicians and not patients, although patients are equally important end-users in the SDM process. The scope and funding of this pilot study did not allow for additional testing with patients, but this is strongly recommended as a future direction. Another limitation was the relatively small sample size of clinicians in this study (9 out of 25 eligible clinicians participated). Due to workplace regulations, the study team was not allowed to offer compensation to providers or alter their clinical schedule to create time for study participation. This may have limited provider recruitment because participation required providers to volunteer their time during existing breaktimes. Finally, due to the lack of standardized documentation of SDM in VHA and the confidential nature of peer review ratings, it was not possible to externally validate clinicians’ reports about whether they documented SDM in their visits. The study team relied on clinician self-report of their SDM documentation practices, which may have been biased.

4.3. Conclusion

While the application of HCD is still fairly new in health sciences research, recent publications have demonstrated how the methodology can be applied to a multitude of patient care issues [19]. The pattern of identifying user needs, exploring the current environment, and using iterative cycles to refine solutions so they fit into practice as seamlessly as possible, presents a continuously applicable methodology to address a range of healthcare system problems [22]. Through the course of our study, participants and expert clinicians were necessary for the HCD team to develop a full visit template incorporating SDM. Further work could expand upon the idea of EHR documentation templates as a means for helping clinicians learn and implement new skills in routine practice.

4.4. Practice implications

The present project showcases the human-centered design process to develop a strategy for adopting SDM into clinical practice. Findings provide support for the utility of HCD in implementation science, and its use as an important set of tools for improved patient-centered care.

Appendix 1. Visual summary of SDM template development