Take on transplant: human-centered design of a patient education tool to facilitate informed discussions about lung transplant among people with cystic fibrosis

Abstract

Objective

Lung transplant (LTx) saves lives in cystic fibrosis (CF). However, many potential candidates express uncertainty about LTx and die before receiving this treatment. CF guidelines recommend LTx education and clinical discussions well before the need for LTx arises, but limited patient resources exist.

Materials and Methods

We engaged people with CF and CF physicians in human-centered design of “Take On Transplant” (TOT), a web-based education tool to prepare patients for LTx discussions. Across 3 phases, needs assessment, design groups, and iterative user testing of TOT, we refined TOT from wireframe prototypes, to an interactive website, to a fully functional intervention ready for clinical trials.

Results

Fifty-five people with CF and 105 physicians identified information needs to prepare for LTx discussions. Design groups (n = 14 participants) then established core requirements: didactic education (“Resource Library”), patient narratives (“CF Stories”), frequently asked questions (“FAQ”), and self-assessment to tailor content (“My CF Stage”). Iterative usability testing (n = 39) optimized the design of CF Stories and prototype layout. We then developed the TOT website and demonstrated feasibility and preliminary efficacy of use through 2-week field testing (n = 9).

Discussion

Our human-centered design process provided guidance for educational tools to serve the evolving needs of potential LTx candidates. Our findings support the process of patient deliberation as a foundation for shared decision-making in CF, and inform educational tools that could potentially translate beyond LTx.

Conclusion

TOT fills a critical gap in preparing people with CF for shared decision-making about LTx and may serve as a model for educational tools for other preference-sensitive decisions.

INTRODUCTION

Cystic fibrosis (CF) is a rare, life-limiting condition caused by mutations in the CF transmembrane conductance regulator (CFTR) gene, with most deaths resulting from progressive respiratory failure.¹ Lung transplantation (LTx) can improve survival and quality of life,^2,3 but nearly 40% of US CF patients with advanced lung disease die before receiving this life-saving treatment.⁴ Although timely LTx referral is critical, non-referral has been attributed to “patient preference” in up to 40% of non-referred CF patients meeting LTx criteria (ie, patient wishes not to be referred when recommended by their physician).^5,6 While LTx is a guideline-recommended therapy for advanced CF, the Cystic Fibrosis Foundation (CFF) LTx referral guidelines recognized that LTx is a preference-sensitive decision and recommended early referral to allow patients and families time to consider the potential risks and benefits of the procedure.⁷

Several issues make LTx a complex decision for people with CF who could benefit. First, lack of knowledge about risks, benefits, and optimal LTx timing lends itself to uncertainty. The resulting ambivalence is associated with decisional conflict and psychological discomfort.⁸ LTx carries significant risk of morbidity and 10% mortality within the first year post-transplant, but median post-transplant survival is 10 years for adults with CF.⁹ Second, understanding prognosis of one’s CF-related lung disease is key to LTx decision-making. Many patients avoid initiating transplant discussions, instead preferring physician-initiated discussions often hindered by time constraints or physicians’ unfamiliarity with nuanced LTx statistics.¹⁰ These challenges are magnified with the recent advent of CFTR modulators (eg, elexacaftor/tezacaftor/ivacaftor) that can significantly improve lung function and delay LTx.^7,10–12 Yet therapeutic effects vary between patients, there is limited prognostic data, and some patients remain ineligible for modulators based on CFTR mutation type. Limited educational resources exist for people with CF to address complex LTx decision-making. Most educational resources for LTx are not specific to CF (eg, unos.org), are decision aids that have become outdated,^13,14 or are transplant program-specific handouts typically provided after LTx referral. Third, the burdens may not be consistent with a patient’s preferences or goals of care. LTx requires substantial financial commitments and potentially leads to changed social dynamics (eg, requiring the assistance of caregivers). After LTx, CF patients may experience an increased reliance on daily medications and increased frequency of engagement with the medical system. Some CF patients report having lived “long enough” and do not have a desire to undergo the intensive process involved with obtaining and living with LTx.^7,11

Given the complexities of LTx, people with CF need ample time and support to learn about and prepare for informed treatment decisions. Guidelines emphasize educating CF patients for shared decision-making well before the need for transplant arises.^7,15,16 Many CF patients desire more LTx information early, even when lung function is near-normal.¹¹ However, there is limited guidance on how to design educational tools for people with CF, some of whom may be years away from LTx. Existing decision aids for LTx have significant limitations. One decision aid that highlights pros (survival statistics) and cons (complications) of LTx reduced decisional conflict,^13,14 but is not aligned with current CFF guidelines and contains outdated statistics. Another example, the “Informed Choices CF” decision aid, was designed from interviews among critically ill, end-stage CF patients and caregivers, but is not publicly available and has not been shown to improve patient outcomes.^17–19

While existing decision aids^{13,14,17–19} focus on prognostic estimates and preference elicitation at point of care consultations after a patient is eligible for LTx, they are not designed to support “deliberation”—the process patients take in “considering information about the pros and cons of their options, to assess their implications, and to consider the range of possible futures, practical as well as emotional”.(20, p.1365) Rather than occurring once at the point of care (during a LTx consultation in clinic), deliberation is an iterative process in which patients explore what matters most. As the foundation for shared decision-making, deliberation involves many questions and answers, often requiring patient collaboration over time with health professionals and social networks. Point of care decision aids are not designed to support deliberation, but instead assume patients are already sufficiently educated for shared decision-making at consultation. A broader view of these interventions, including those used in preparation before and used during a consultation, can support patients by making their decisions explicit, providing information about options and associated benefits/harms, and helping clarify congruence between decisions and personal values.²¹ Although some LTx decisions aids have been rigorously designed^14,19 and tested for usability,¹⁸ translation of educational support from point of care (one time, directed by providers during a visit) to deliberation (over time, directed by patients before consultation) contexts is unexplored in the CF setting.

There is a critical need for patient education tools that provide the breadth of LTx information CF patients seek over time to prepare for discussions with providers well before LTx is indicated, as guidelines recommend.^7,15,16 Yet it is unclear how new tools should be designed to support deliberation among CF patients learning about and preparing for shared LTx decisions. To address this gap, we engaged CF stakeholders in a multi-phase, human-centered design of “Take On Transplant” (TOT), a web-based patient education tool designed for people with CF to learn about LTx and prepare for LTx discussions with CF care teams.

OBJECTIVE

This project aimed to design and establish the feasibility of TOT for CF patients. In partnership with CF patients and physicians, we addressed the following research questions (RQ):

1. What are the LTx information needs of people with CF?

2. What design requirements do people with CF recommend for patient LTx education tools?

3. How can we iteratively refine TOT to meet user requirements?

   • 3.1 What design features do people with CF find most useful and usable?

   • 3.2 What is the feasibility of using TOT?

STUDY DESIGN

We employed human-centered design (HCD),²² an iterative approach to: (1) understand user needs and context, (2) specify user requirements, (3) generate design solutions, and (4) evaluate designs against requirements.²²Figure 1 shows HCD activities in our 3 phase project, including (1) needs assessment, (2) design groups, and (3) usability and field testing. University of Washington Institutional Review Board approved study procedures.

Figure 1.

Human-centered design of TOT: HCD is an iterative methodological approach to: (1) understand user needs and context, (2) specify user requirements, (3) generate design solutions, and (4) evaluate those designs against requirements. In this 3 phase project, we accomplished these steps of HCD to answer specific research questions (RQ) through a needs assessment (RQ1), design groups (RQ2), and iterative cycles of prototyping to generate design solutions and evaluate designs against requirements, which included lab-based usability testing and field testing (RQ3).

STAKEHOLDER ENGAGEMENT

People with CF

Non-transplanted (ie, pre-transplant) individuals participated in needs assessment, usability testing, and field testing. Transplanted individuals (ie, transplant recipients), who could share expert perspectives from having experienced LTx, participated in needs assessment and design groups. We recruited participants in-person, by phone, via collaborators from the University of Washington CF Center, 6 other CF Centers, and through the CFF’s Community Voice program (a national organization of patients with CF and their caregivers). Of the 117 patients who participated across the 3 phases in total, 109 were unique. The remaining 8 were transplanted individuals who participated in both the needs assessment and design groups.

CF physicians

Using a CFF listserv with active email addresses for all practicing CF physicians in the United States, we invited the 1088 physicians at CFF-accredited centers to complete an online REDCap survey.^23,24Supplementary Material S1 summarizes characteristics of respondents (n = 105, 10% response rate) and their responses.

PHASE 1: NEEDS ASSESSMENT

Methods

The goal of phase 1 was to understand user needs and context (HCD step 1). We conducted interviews with CF patients and surveys of CF physicians to answer RQ1: What are the LTx information needs of people with CF?

Interviews with CF patients (June 2019–September 2020) examined knowledge about LTx and perspectives on communication about LTx with CF providers among transplanted and non-transplanted individuals. Non-transplanted individuals were sampled purposively for representation across varied lung function levels (FEV1) with balance across genders and were recruited from the University of Washington Adult CF Center. Transplanted individuals were identified based on time since LTx and to incorporate diverse experiences with the transplant process (eg, variety of comorbidities and complications). Transplanted individuals were recruited from the University of Washington and the CFF Community Voice.

The semi-structured interviews lasted 60–90 min, were recorded, and transcribed. Interviews and analyses took place in 2 steps. First, we analyzed transcripts from interviews with non-transplanted individuals using qualitative content analysis to inductively identify emergent themes^25,26 reflecting patients’ LTx information needs. Data were collected until thematic saturation was achieved. Four coders (KJR, MRH, PJS, and LEB) blind-coded 6 transcripts in Dedoose (dedoose.com) as interviews were being conducted. Codes were discussed during investigator meetings to achieve consensus on a codebook of topics reflecting information needs. Thirteen transcripts were then co-reviewed by 2 coders (MRH and LEB) to ensure reliability of the coding and achieved agreement >85%.²⁷ MRH coded the remaining transcripts. The coders met weekly to discuss code consistency, thematic patterns, preliminary findings, and exemplar quotes on topics that reflect patients’ information needs. Second, we conducted interviews with transplanted individuals after interviews with non-transplanted patients were completed. Drawing upon topics that emerged from non-transplant interviews, 3 coders (KJR, MRH, and LEB) applied the codebook deductively to transcripts with transplanted individuals using directed content analysis.²⁵ We contrasted the information needs between groups by comparing the prevalence of coding topics.

Surveys with CF physicians (May–June 2020) examined experiences discussing LTx with CF patients and important topics to discuss in those conversations. We summarized responses with descriptive statistics, including prioritizing topics most relevant to physicians for LTx discussions with patients.

Results

Fifty-five patients (35 non-transplanted and 20 transplanted participants) completed interviews (Table 1). The 105 physician survey respondents (10% response rate) were 58% male with a broad range of years in practice from both adult (44%) and pediatric (49%) centers (see Supplementary Material S1).

Table 1.

Characteristics of people with CF^a across studies

Phase	Phase 1	Phase 2	Phase 3	Phase 3	Phase 3
Study	Patient interviews	Design groups	Usability testing Study A	Usability testing Study B	2-week Field testing
	(n = 55)	(n = 14 participants)	(n = 25)	(n = 14)	(n = 9)
Age mean (SD)	35.9 (11.4)	38.5 (9.2)	41.7 (11.8)	36.1 (8.8)	41.7 (9.5)
Gender, n (%)
Woman	27 (49%)	9 (64%)	10 (40%)	11 (79%)	6 (67%)
Man	27 (49%)	5 (36%)	14 (56%)	3 (21%)	3 (33%)
Another gender	1 (2%)	0 (0%)	1 (4%)	0 (0%)	0 (0%)
Race, n (%)
Black/African American	1 (1.8%)	1 (7.1%)	1 (4%)	0 (0%)	0 (0%)
White	52 (94.5%)	13 (92.9%)	24 (96%)	14 (100%)	9 (100%)
More than 1 race	1 (1.8%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Unknown	1 (1.8%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Ethnicity, n (%)
Hispanic or Latino	1 (1.8%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Not Hispanic or Latino	54 (98.2%)	14 (100%)	25 (100%)	14 (100%)	9 (100%)
Education
High school diploma/GED	2 (3.6%)	0 (0%)	0 (0%)	1 (7.1%)	2 (22.2%)
Some college/trade school	14 (25.5%)	0 (0%)	7 (28%)	4 (28.6%)	2 (22.2%)
4-year college degree	14 (25.5%)	0 (0%)	10 (40%)	9 (64.3%)	4 (44.4%)
Graduate degree	5 (9.1%)	0 (0%)	8 (32%)	0 (0%)	1 (11.1%)
Not assessed	20 (36.4%)	14 (100%)	0 (0%)	0 (0%)	0 (0%)
Insurance status, n (%)
Private	35 (63.6%)	6 (42.9%)	17 (68%)	8 (57.1%)	8 (88.9%)
Medicaid	12 (21.8%)	2 (14.3%)	4 (16%)	1 (7.1%)	0 (0%)
Medicare	8 (14.5%)	5 (35.7%)	2 (8%)	4 (28.6%)	1 (11.1%)
TriCare/military health plan	0 (0%)	0 (0%)	1 (4%)	1 (7.1%)	0 (0%)
State special needs program	0 (0%)	0 (0%)	1 (4%)	0 (0%)	0 (0%)
Transplant status, n (%)
Non-transplanted	35 (63.6%)	0 (0%)	25 (100%)	14 (100%)	9 (100%)
Transplanted	20 (36.4%)	14 (100%)	0 (0%)	0 (0%)	0 (0%)

^aPeople with CF were recruited from multiple adult CF centers including the University of Washington, University of Minnesota, University of Pittsburgh, University of Michigan, National Jewish Health, Indiana University, and Maine Health.

Note: Eight transplanted individuals took part in patient interviews and design groups.

Supplementary Material S1 lists physician survey results regarding their experiences discussing LTx with CF patients and topics prioritized for LTx discussions with patients. All physicians reported having discussed LTx with a patient and were willing to do so. The majority (92%) reported that having discussions long before the actual need for transplant referral would be useful. The majority reported feeling moderately or very prepared to initiate a LTx conversation (97%), respond to LTx questions, (92%) and engage in shared LTx decision-making (90%) with patients. The majority were moderately or very interested (89%) in using an educational tool for patients designed to facilitate LTx discussions in their clinical practice. Physicians’ highest priority topics (ie, top 3) for LTx discussions were early referral (67%), barriers/contraindications to LTx that impact eligibility (58%), and adherence to therapy (51%).

Supplementary Material S2 lists information needs that surfaced from patient interviews. Whereas transplanted individuals focused more on topics related to the LTx procedure and recovery (eg, surgery, recovery, post-LTx survival, quality of life), non-transplanted individuals also discussed preparing for LTx (eg, LTx eligibility, evaluation, waitlist). Overall, patient participants expressed needs spanning a spectrum of clinical information about LTx (eg, surgery, recovery, survival, quality of life). When asked “What three things would you want to know from a CF clinician about lung transplant”, the top 3 needs among non-transplanted participants were LTx eligibility, recovery from surgery, and post-transplant survival. For example, one non-transplanted participant wanted to better understand their LTx eligibility given their risks and prognosis: “Right now I have no idea when, what triggers that kind of conversation. So just knowing some different thresholds that you would need to cross, sort of, to enter in the conversation, might be good to know ahead of time.”

As a major component of preparing for LTx, patient participants also expressed the need for experiential information based on experiences of LTx recipients. Many transplanted participants reflected on the value of first-hand experience as they embarked on the process. Desired information included aspects of the day-to-day lived experience ranging from preparing for and recovering from surgery, to finances, moving, work/school, and caregiver support. For example, one non-transplanted participant inquired: “What can I expect, and what is it like after? And how is my life going to go? Anything I might need to prepare for that.”

Finally, transplanted participants described a range of information sources they found useful (eg, clinicians, social workers, Internet, peers). Some described tracking questions and answers from different sources, such as a transplant Facebook group. Participants expressed varied preferences for learning about treatment options and discussing LTx with providers. Many wanted to moderate information by controlling how much they received and when, based on the urgency of the LTx discussion.

Findings revealed 3 key themes. First, people with CF need a range of clinical information throughout the LTx journey that could be supported through a “Resource Library” of didactic education about LTx. Second, they need experiential information from other patients that could be supported through “CF Stories” about experiences of LTx recipients as well as those who chose to forgo LTx. Third, they need information tailored to their current situation that could be supported by a “My CF Stage” self-assessment that recommends content specific to a patient’s LTx knowledge, readiness, and disease severity.

PHASE 2: DESIGN GROUPS

Methods

The goal of phase 2 was to specify user requirements (HCD step 2). We conducted design groups with transplanted people with CF who could reflect upon their LTx experience as they responded to wireframe prototypes. This work answered RQ2: What design requirements do people with CF recommend for patient education tools for LTx?

Wireframe prototypes: Based on the needs assessment, we created low-fidelity wireframe prototypes with different design features to support provision of didactic content through a Resource Library, experiential information through narrative patient stories, and tailored information through a knowledge assessment. We explored design features for tracking “My Questions” that could prompt LTx conversations with providers. Because the timing of LTx decisions includes assessment of one’s prognosis, we included a “Risk Calculator” based on a 3-year prognostic model^28,29 to address patients’ needs for clinical information about prognosis and LTx eligibility.

Design groups (July–September 2020) included a series of 3 2-h sessions. We conducted 2 participant tracks of 3 design group sessions in parallel. Participants in each track completed all 3 sessions. Each session included 4–7 participants, was moderated remotely through Zoom by an experienced lead (AH) and research team members who took notes (KJR, MRH, and LEB), and were recorded for review and synthesis through qualitative memoing.²⁷ This sample size is consistent with similar group design methods, such as co-design groups and workshops.^30–33 We presented participants with wireframes grounded in user personas of fictitious CF patients—“Jack” who is not yet considering LTx and “Jill,” who is sicker and approaching LTx. Group discussion elicited participant feedback on design features. Supplementary Material S3 shows personas and sample wireframes.

After each session, participants rated the wireframes by completing the System Usability Scale (SUS).³⁴ SUS produces an Overall score and subscale scores for Usability and Learnability. The instrument has excellent reliability (r = 0.9) and validity (r = 0.82).^35,36 Scores range from 1 to 100 (higher scores are better), which we summarized with descriptive statistics. After each session team members (AH, KJR, MRH, and LEB) debriefed to review notes for qualitative memoing³⁷ through which we summarized design feedback and corresponding design improvements in an executive summary of results Between parallel sessions 1, 2, and 3, we compared and synthesized design feedback across participant tracks, and then iteratively refined the wireframes for subsequent sessions. We used SUS results to benchmark design improvements across the 3 parallel sessions (6 sessions in total), which established user requirements for TOT.

Results

Across the 2 tracks, 14 transplanted individuals completed the series of 3 design group sessions (Table 1). The mean time since LTx was 1.7 years (SD = 1.8).

In design group session 1, wireframes conveyed the overall TOT design concept which included a knowledge assessment survey that guided the user to 4 modules: Resource Library, CF Stories, My Questions, and Risk Calculator. Participants expressed enthusiasm for the design concept—they verified the importance of didactic resources and narrative patient stories from “someone like me.” However, one African American participant reported not seeing themself represented in the stories, which prompted us to create more inclusive personas (revised names/images) and solicit more diverse narratives. Participants recommended adding resources for caregivers and expressed interest in frequently asked questions (FAQ) rather than tracking questions. The risk calculator alarmed several participants who expressed concern about a non-transplanted patient reading that they may be at a “high risk” for dying or undergoing LTx in the next 3 years without a clinician present to explain the nuances.

In design group session 2, we revised the wireframes to reflect this feedback by replacing the Risk Calculator with “My CF Stage” that included questions about disease severity and generated a tailored report indicating “Lung transplant discussion urgency” and content recommendations. We replaced My Questions with FAQs answered by patients and clinicians. Although participants found the revisions favorable, tension surfaced between preferences for the “siloed” design (content separated into modules) versus a more integrated user experience that integrated Resource Library content with CF Stories and FAQ. Other feedback focused on health literacy and accessibility, such as clearly defining medical terms (eg, FEV₁, CFTR).

In design group session 3, we revised the wireframes to explore preferences for the siloed design versus an alternative “integrated” design that lists related content of all 3 types (didactic education, patient stories, FAQ) on the same page. Participants were split, which we investigated further during iterative user testing. Despite these mixed preferences, consensus was reached on the core design features for TOT: didactic education (“Resource Library”), patient narratives (“CF Stories”), “FAQ” answered by patients and clinicians, and tailored content recommended based on a self-assessment (“My CF Stage”). Figure 2 shows improved SUS scores with design refinements across sessions (Multimedia 3 shows SUS scores by track). These findings established user requirements for TOT (Table 2), which we used to create interactive prototypes for user testing.

Figure 2.

Improved SUS Scores across the 3 design groups sessions: Mean SUS scores (±SE) for Overall SUS scores, Usability sub scores, and Learnability sub scores showing improvement with design refinements across the 3 design group sessions. Supplementary Material S3 shows the SUS scores by track.

Table 2.

User requirements for TOT

LTx information need	User requirement	Prototype design feature
Clinical information	Didactic education	“Resource Library” contributed by CF clinical experts
Experiential information	Patient narratives	“CF Stories” about the experiences of LTx recipients and those who chose to forgo LTX
Clinical and experiential information	Answers to common questions	“FAQ” answered by CF patients and physicians
Control the amount and timing of information	Tailored content	“My CF Stage” self-assessment survey

PHASE 3: ITERATIVE PROTOTYPING AND USER TESTING

The goal of phase 3 was to generate design solutions (HCD step 3) and then evaluate those designs against user requirements (HCD step 4). We built, evaluated, and refined high-fidelity interactive prototypes through iterative prototyping and user testing, including lab-based usability testing (3.1) and 2-week field testing (3.2). Sample sizes for user testing were based on HCD guidelines that suggest at least 8 representative users for formal user evaluations²² and best practice guidelines that suggest 10 ± 2 participants identify the majority of usability problems.³⁸ Additional participants were recruited to achieve thematic saturation regarding qualitative usability feedback. This work addressed RQ3. How can we iteratively refine TOT to meet user requirements?

Usability testing

Usability testing (December 2020–March 2021) investigated useful profile attributes of CF Stories (Study A) and usable architectural layouts for TOT (Study B). This work answered RQ 3.1 What design features do people with CF find most useful and usable?

Study A: Useful CF stories profile attributes

Because of user interest in finding stories about “someone like me”, we created a limited version of the interactive TOT prototype that presented only CF Stories (Figure 3). We adapted stories as narratives presented in patients’ own words from interviews. CF Stories displayed in a standard format detailing experiences during “pre-transplant and evaluation,” “the call, the transplant, and the recovery” (ie, peri-transplant), “post-transplant,” and “My take on transplant” offering the patient’s lessons learned. Stories of individuals who opted to forgo Ltx (“not transplanted”) were inherently limited to pre-transplant and evaluation experiences. Each story included a profile with photo, first name, content tags, “story highlights,” and the following attributes: gender, age at transplant (or at transplant evaluation for those not transplanted), FEV₁ at transplant (or at transplant evaluation for those not transplanted), current FEV₁, CFTR genotype, and CFTR modulator status at transplant. Study A investigated which attributes participants found most useful.

Figure 3.

Interactive CF Stories prototype: The interactive “CF Stories” prototype allows participants to browse a list of story profiles (A), each displaying a photo, patient first name, content tags, profile attributes, and story highlights. Clicking on the “View Story” button leads to the full-text narrative for that patient’s CF Story (B).

Methods

We recruited non-transplanted participants for individual Zoom sessions lasting 60–90 min in which they explored CF Stories. Participants were asked to scroll through profiles to select stories. After reading, participants described reasons for story selection and what they learned to help prepare for LTx discussions. We summarized sessions with field notes to describe the number of stories selected, reasons for selection including attributes of interest, and what they found helpful about the stories. We recruited and conducted sessions until reaching saturation in qualitative themes regarding attributes of interest and helpful features of CF Stories.

Results

Twenty-five non-transplanted participants (Table 1) selected and read an average of 3.2 CF Stories (SD = 0.91, range = 2–6). Several participants appreciated the granular detail of stories. Some participants reflected on the helpfulness of narrative suggestions and depictions of the “whole process.” However, others found stories less helpful when content was unrelated to their personal experiences or when they “already knew” the information. Reflecting those with milder lung disease (further out from needing LTx), many commented that it “made sense” to consider what the process would be like if they were in a “similar situation.” One participant emphasized “If I ever got closer [to transplant] I for sure would read more [CF Stories]—Knowledge is power.” Participants appreciated stories that highlighted LTx conversations with loved ones, and found stories provided peer support.

Participants were particularly drawn to CF Stories from peers in similar situations, highlighting the importance of profile attributes. The profile attributes that influenced participant story selection most included: age at transplant, gender, FEV₁, CFTR genotype, and CFTR modulator status. Some participants used photos to determine whether the people in the CF Stories were “like them.” Most participants wondered whether transplant recipients in the stories had access to newer, highly effective CFTR modulators. Based on these findings, we revised the TOT prototype to include the profile attributes participants found most useful.

Study B: Usable architectural layouts

The mixed findings from design groups on siloed (content types separated into sections) versus integrated (related content types interwoven) designs led us to more deeply investigate architectural layouts. The design group findings indicated what content and features to include in a web-based educational tool, but did not inform how the content and features should be laid out together to assist an individual in preparing for LTx discussions. To further inform the design of the tool, we conducted usability study B to explore alternative architectural layouts. An unanswered question was how much agency a person living with CF might prefer when using this tool. Some interactive tools adopt “Author-Driven” designs that guide readers in a linear path through an integrated set of recommended content, whereas “Reader-Driven” designs present content in full to allow readers to choose their own path through different types of content.³⁹ This mirrors how “system-driven” and “user-driven” designs can be distinguished in personal informatics applications.⁴⁰

After integrating improvements from Study A (profile attributes), we generated 2 versions of the interactive TOT prototype (Figure 4): The “Author-Driven” prototype recommends an integrated list of tailored content based on responses to “My CF Stage” self-assessment, and the “Reader-Driven” prototype siloes content by type for participants to freely browse. Study B investigated which layout participants found most usable.

Figure 4.

Interactive Reader-Driven and Author-Driven prototypes: Interactive prototypes illustrating “Author-Driven” (left) and “Reader-Driven” (right) layouts. Using the Author-Driven prototype begins by the participant completing the “My CF Stage” self-assessment survey (A), then based on their responses, a tailored page of recommended content displays an integrated mix of the 3 types of content (resource articles, CF stories, FAQs) (B)—a sample resource article about “Transplant referral” is shown (C). Using the Reader-Driven prototype begins by selecting one of the 3 types of content (ie, Resource Library, CF Stories, FAQ) displayed in separate siloes (D), which allows participants to freely browse to each type, such as all patient stories (E)—a sample patient story about “Amy” is shown (F).

Methods

We recruited non-transplanted participants for user testing to compare prototypes through Zoom. During each 90-min recorded session, participants interacted with both the Author-Driven and Reader-Driven prototypes to complete scenario-based usability tasks. For each task, the participant was given up to 15 min to search for information they would find useful while playing the role of 2 user personas: “Marco” (healthier patient) and “Tamika” (sicker patient) (Supplementary Material S3). Afterward, they rated SUS for each prototype and described their preferred architectural layout in an exit interview. We reviewed usability tasks from recordings to determine which content types were viewed (Resource Library, CF Stories, FAQ) in each prototype. We compared SUS scores between prototypes with paired t-tests. We summarized qualitative comments to help explain SUS scores. We recruited and conducted sessions until reaching saturation in qualitative themes regarding preferred architectural layouts.

Results

Fourteen non-transplanted participants completed Study B (Table 1). Participants first viewed more varied content types with the Reader-Driven prototype (Table 3) than the Author-Driven prototype. The Reader-Driven prototype led to higher overall SUS (90 vs 82, P = .001) and usability sub scores (88 vs 77, P = .001), while learnability was equally high (Table 3). During exit interviews, participants reported wanting greater control over finding information about LTx. Although most participants found the tailored content recommended by the Author-Driven prototype useful as a starting point, they also found that prototype overly restrictive. They expressed concern about missing “hidden gems” of potential interest. Many participants expressed frustration with didactic and experiential information shown together, and instead wanted to focus on either finding “facts” or experiences. Based on these findings, we developed the TOT website for field testing (Figure 5). The website includes the “My CF Stage” feature of the Author-Driven prototype to recommend content on a personalized home page, while maintaining access to all content through the siloed design of the Reader-Driven prototype.

Table 3.

Comparison of content types viewed and usability of Author-Driven versus Reader-Driven prototypes

	Author-driven	Reader-driven
First content type viewed—n (%)
Resource Library	12 (86%)	7 (50%)
CF Stories	1 (7%)	5 (36%)
FAQ	1 (7%)	2 (14%)
All content types viewed—n (%)
Resource Library	14 (100%)	14 (100%)
CF Stories	10 (71%)	14 (100%)
FAQ	11 (79%)	13 (93%)
SUS—Mean (SD), range
Overall	82 (10), 68–98	90 (9), 70–100
Usability	77 (12), 59–97	88 (11), 63–100
Learnability	99 (3), 88–100	99 (3), 86–100

Figure 5.

Take on Transplant (TOT) patient education tool evaluated during field testing: The website tailors content based on “My CF Stage” self-assessment survey (A). Based on the survey responses, content is displayed on a personalized home page (B) along with a visual meter indicating the urgency of discussing transplant with one’s care team. Tailored content recommended on the home page consists of patient-generated CF Stories about personal experiences of LTx recipients as well as those who chose to forgo LTx, didactic Resource Library articles covering clinical aspects of LTx, and FAQ answered by CF patients and CF physicians. The version of TOT field tested included 17 CF Stories, 28 Resource Library articles divided across 5 sections, and 24 FAQs.

Field testing

Field testing (July–September 2021) evaluated the feasibility of using TOT over 2 weeks. This work addressed RQ 3.2 What is the feasibility of using TOT?

Methods

We recruited non-transplanted individuals for field testing to assess the feasibility of methods and explore website use and properties of outcomes measures. Participants completed a baseline survey including demographics and 3 survey instruments: (1) 10-item Decisional Conflict Scale (reliability 0.81; validity 0.78–0.92),^41,42 (2) 14-item investigator-designed LTx knowledge assessment, and (3) investigator-designed Likert rating of preparedness for LTx discussions with one’s CF provider from 0 “Don’t know” to 4 “Very prepared.” Supplementary Material S4 provides the investigator-designed instruments. The knowledge assessment included 14 confidence-weighted True/False statements,⁴³ with the following response options that were scored for a maximum sum of 28 points and a minimum score of −28: “I am sure this statement is true.”(+2 points), “I think the statement is true, but I am unsure.” (+1 point), “I think the statement is false, but I am unsure.” (−1 point), “I am sure this statement is false.” (−2 points), and “I don’t know” (0 points). The 8 items with a false statement were reverse scored (eg, “I am sure this statement is true.” (-2 points)).

After introducing participants to TOT, we asked participants to use the website as desired for 2 weeks, during which we tracked site interactions in usage logs. From usage logs we calculated time spent interacting with CF Stories, Resource Library, and FAQ. After 2 weeks, participants completed a follow-up survey to rate TOT with SUS,³⁴ Feasibility Intervention Measure (FIM),⁴⁴ and Acceptability Intervention Measure (AIM).⁴⁴ We summarized survey data with descriptive statistics. Our primary outcome was feasibility as assessed by time spent with the website, FIM, AIM, and qualitative feedback. To explore preliminary efficacy of TOT, we re-assessed decisional conflict, LTx knowledge, and preparedness for pre-post comparison with paired t-tests.

Results

Nine non-transplanted participants completed field testing (Table 1). They spent an average of 4.0 h (SD = 2.5 h, range = 0.6–7.5 h) using TOT over 2 weeks. Most time was spent interacting with CF Stories (M = 1.9 h, SD = 1.4 h), followed by Resource Library (M = 0.8 h, SD = 0.7 h), and FAQ (M = 0.4 h, SD = 0.4 h). Table 4 shows outcome measures. Overall SUS ratings were high, corresponding with the highest possible rating of “Best imaginable.”³⁴ FIM and AIM ratings were also high. There was a significant 2-week improvement in decisional conflict (36.7 vs 9.4, P = .008), LTx knowledge (10.2 vs 15.3, P = .034), and preparedness for LTx discussions (2.9 vs 3.8, P = .026).

Table 4.

Two-week outcome measures from field testing for SUS, FIM, AIM, and short-term preliminary efficacy

Outcome	PRE	Post
	Mean (SD), range	Mean (SD), range
System Usability Scale (SUS)		94.4 (6.1), 80–100
Overall score	—	93.4 (7.6), 75–100
Usability subscale	—	98.6 (4.2), 88–100
Learnability subscale	—
Feasibility of Intervention Measure (FIM)	—	4.9 (0.4), 3.8–5.0
Acceptability of Intervention Measure (AIM)	—	4.8 (0.3), 4.3–5.0
Preliminary short-term efficacy	Pre	Post
Decisional Conflict Scale*	36.7 (25.7), 0.0–75.0	9.4 (13.1), 0.0–40.0
LTx knowledge*	10.2 (6.8), 1.0–19.0	15.3 (6.3) 2.0–23.0
Preparedness for LTx discussions*	2.9 (0.9), 1.0–4.0	3.8 (0.4), 3.0–4.0

^* P < .05.

Outcome measures: SUS scores range from 1 to 100 (higher is better)³⁴; FIM and AIM scores range from 1 to 5 (higher is better)⁴⁴; Decisional conflict scores range from 0 to 100 (lower is better)³¹; Lung transplant (LTx) knowledge scores range from −28 to 28 (see Supplementary Material); Preparedness for LTx discussions scores range from 0 to 4 (see Supplementary Material).

DISCUSSION

To address gaps in educational resources promoting a deliberative approach to LTx shared decision-making, as recommended in CFF guidelines,^7,15,16 we employed HCD with CF patients and physicians to develop TOT. Needs assessment identified the value of clinical, experiential, and tailored information based on disease severity. Design groups established design requirements to meet those needs through a Resource Library, CF Stories, FAQ, and recommended content based on My CF Stage. Iterative user testing optimized the design of CF Stories and architectural layout, and then demonstrated feasibility of TOT use over 2 weeks. Benchmark improvements in usability were shown throughout the HCD process. Field testing showed potential to reduce decisional conflict, improve LTx knowledge, and prepare patients for LTx discussions with care teams.

Shared decision-making should start well before point of care consultations, a time when many one-time decision aids are delivered. Informatics support for patient deliberation²⁰ has received comparably less attention. Well-designed educational tools should support deliberation over time by improving knowledge and reducing decisional conflict as patients’ needs evolve while preparing for shared decision-making. Our findings identify content domains and design features to help patients learn about LTx and prepare for LTx conversations. The importance of tools meeting a breadth of interests should also be recognized, which was demonstrated by varied preferences for Reader-Driven and Author-Driven layouts. This variation is intuitive given the varied disease severity in CF, where LTx may be emergent for some, while for others LTx remains years away. Other factors may influence preferences (eg, health literacy, learning styles, coping mechanisms) and thus flexibility of educational tools is critical. In TOT, we retained design features of both layouts for flexibility to meet needs across the illness trajectory, while giving the user control over the amount of LTx information to which they are exposed (eg, showing “highlights” from CF Stories as a preview of the content).

In addition to demonstrating informatics opportunities for patient deliberation in the context of shared decision-making, our findings expand research on peer support for patients with rare conditions.⁴⁵ Findings demonstrate the importance of experiential stories in helping patients engage in healthcare, as was shown to have a pivotal role in other conditions like cancer.⁴⁶ Combining clinical didactics with experiential narratives meets a spectrum of information needs,⁴⁷ which may be valuable for CF in which few educational resources exist and access to peer support is limited.⁴⁵ Offering diverse narrative stories in TOT, indexed by profiles that aid users in locating relevant peer support,^48,49 could improve patient engagement.⁴³ Diverse representation surfaced not only in finding relatable CF Stories, but also in personas to engage research participants. This insight echoes findings of disparities in shared decision-making experiences among CF patients that call for tailored interventions.⁵⁰ Such strategies to improve diversity, equity and inclusion in biomedical informatics research and practice are critical.⁵¹

Although we engaged adults from varied socioeconomic backgrounds, our sample lacked racial and ethnic diversity. The prevalence of CF is highest in White non-Hispanic individuals,⁵² yet CF is present in other groups who experience treatment disparities and worse outcomes.^53,54 Inclusion of diverse participants in future research is critical for improving equity. We prioritized the voices of participants from underrepresented groups, which led us to develop more inclusive study materials (personas) and content. For example, 40% of CF stories were contributed by patients on Medicaid. Although the results from field testing demonstrate feasibility, the sample size was too small to discern generalizable outcomes on preliminary efficacy. Despite these limitations, our iterative stakeholder engagement across different stages of disease led to design refinements that demonstrated improved usability through SUS benchmarks. Future research should further investigate inclusive design affordances that promote TOT engagement among diverse CF patients and caregivers, and engage providers on fitting TOT into clinical workflows.

CONCLUSION

This project describes the use of HCD for a patient education tool to help CF patients prepare for LTx discussions with their care team. Findings provide design guidance for educational tools that meet the diverse and evolving information needs of patients across the illness trajectory, which could translate beyond the context of CF and LTx. Informative profiles help patients select experiential stories for relevant peer support. Combining affordances of Author-Driven and Reader-Driven architectures offers users both guidance and control to meet diverse, evolving information needs. These design contributions fill a gap in informatics support for the process of patient deliberation as a critical foundation for shared decision-making.

FUNDING

This research was supported by the National Institutes of Health (K23HL138154, R03HL158728, and T15LM007442) and the Cystic Fibrosis Foundation (RAMOS17A0, RAMOS20A0-KB).

AUTHOR CONTRIBUTIONS

ALH, KJR, SGK, and MRH conceived of the study, MRH, LEB, NR, JBP, ALH, and KJR collected and analyzed data. All co-authors participated in review and preparation of the manuscript. MRH died prior to completion of the final manuscript.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Journal of the American Medical Informatics Association online.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.