We developed VaxDA-C19 using an iterative, user-centered design approach. Our overall design process followed the International Patient Decision Aids Standards [32]. The patient decision aid was structured according to the Ottawa Decision Support Framework [33]. The web-based design and development approach combined human-centered design principles [34] and a Scrum framework [35]. Human-centered design was implemented via early and ongoing involvement of users and stakeholders, iterative prototyping, and cycle-by-cycle usability feedback to enhance content layout, navigation, and overall user experience (UX) [34]. Scrum served as a structure for delivering small, incremental updates through brief development cycles, with backlog prioritization and periodic review sessions to ensure that implementation aligned with new evidence and user input [35]. We tailored Scrum for this project by synchronizing review points with usability testing rounds. These rounds included checkpoints to validate content and evidence for modifications to vaccine information or recommendations. We also used streamlined remote meetings to accommodate the team and respond to evolving guidelines.

Throughout the development process, we involved a citizen panel, health care professionals, UX designers, and scientific experts to guide refinements. We also conducted usability testing sessions with adults in Canada, using semistructured interviews, comparative testing, and think-aloud protocols with thematic analysis. We ultimately conducted 4 design cycles in total with adults in Canada and expert reviewers. Cycle 1 involved 9 people (all prospective users), cycle 2 involved 22 people (all prospective users), cycle 3 involved 9 people (3 prospective users and 6 experts), and cycle 4 involved 9 people (9 experts).

To help guide the overall project, a citizen panel of 6 individuals, comprising people from a range of age groups, languages, educational backgrounds, health statuses, and geographic regions (Alberta, Saskatchewan, Ontario, and Québec) provided feedback about the design and development of the patient decision aid and other interventions (reported separately) as well as the user testing methods and results. In the context of COVID-19 restrictions in place at the beginning of the project, the citizen panel was assembled from existing contacts with patient and citizen partners across a range of prior projects and life circumstances. This group met by Zoom (Zoom Video Communications) for 2 hours at a time to discuss ongoing project design, development, and study planning. This group served as an advisory panel, distinct from user testing with previously unknown members of the general public, described below [34].

Health care professionals and research experts, including infectious disease researchers, epidemiologists, primary care providers, and 2 UX designers (internal and external experts), participated in the iterative process as expert reviewers. The designers’ expertise covered interaction design and usability evaluation (UX research and testing) as well as visual and information design. The internal designer was part of the senior author’s (HOW) human-centered digital health research team and supported participatory design activities, including planning and conducting user testing in field settings or in a human-computer interaction laboratory. The external designer is a UX lead specializing in design systems, including the development and evolution of interface components. Depending on their exact area of expertise, the designers’ roles were to support content presentation, interface coherence, and usability of the web-based patient decision aid [34].

This study received ethics approval from the “Comité d’éthique de la Recherche en Sciences de la Santé” of the Université Laval (approval: 2020‐222). All participants received detailed study information and provided written informed consent before participation. We offered user testing participants US $28.80 for their time and to help defray internet costs. We removed all identifying details of participants from reports and ensured the confidentiality of data according to institutional policies.

In light of COVID-19 restrictions in place during the bulk of this project, we used online tools to conduct our user tests. Specifically, we recruited user testing participants through tailored Facebook advertisements, then conducted user testing sessions using Zoom, with participants sharing their screen with the interviewer. In line with the objectives of human-centered design, the recruitment process involved a nonprobability convenience sample to support rapid, iterative refinement. The findings were intended to inform design improvements rather than statistical generalization [36]. Eligible participants for user testing were adults (≥18 years) residing in Canada, able to communicate in English or French, and possessing basic computer skills. Individuals without internet access or unable to provide informed consent were excluded. We therefore selected Facebook options to advertise to people who were 18 years and older across Canada and who indicated an education level of either an associate degree (eg, 2-year postsecondary diploma), high school graduate, or unspecified. We did not exclude people who had other educational statuses but used tailored recruitment in an effort to include people across a range of backgrounds, as our previous experiences suggested that people who respond to such advertisements are often more highly educated. We stopped recruiting for each cycle after reaching thematic saturation in comments and usability issues identified.

We used Figma design software to iteratively design the overall structure and appearance of VaxDA-C19. As described further in the Results section, the patient decision aid ultimately included an introductory section in which the patient decision aid is presented, and the user selects the age group of the person for whom a decision is required (themself or their child), their province or territory of residence (which determines which vaccines are available), and the decision they are making (whether or not to get a primary series of vaccines, whether or not to get a booster dose, which vaccine to get, or which booster to get). The user is then invited to create an avatar representing themself or their child. This avatar is used in the next section, which describes the individual-level benefits and risks of the vaccine. VaxDA-C19 offers layered, straightforward summaries of evidence tailored to the user’s selected age, group, province or territory, and decision, including vaccine availability and eligibility based on current authorizations and guidelines. It presents individual benefits and harms in both text and visual risk formats, with links to public health sources and peer-reviewed evidence. The top of the page provides a summary, and the user can click to view more detailed explanations below the summary. On the next page, the user can choose to create up to 8 more avatars representing people around them or their child or to have additional avatars auto-generated. They may then view or skip a brief (2.5-minute) animation showing how vaccination may protect those around us [37]. This optional brief animation explains community immunity and can be viewed or skipped. At the same time, core evidence on individual risks and benefits, along with the rest of the patient decision aid, remains accessible at all times. Following this optional component, the user is invited to reflect on what matters to them, relevant to the decision, using a values clarification method previously developed and tested in the context of treatment decisions and influenza vaccination decisions [38]. Finally, the user has the option to read answers to frequently asked questions (FAQs) and to learn more about our team. A short walk-through video is available on YouTube [39].

Our multidisciplinary team followed a human-centered design over 4 iterative cycles, summarized in Figure 1 [34]. We reported our patient decision aid development process using the DEVELOPTOOLS reporting checklist, provided in Checklist 1 [40]. As per the International Patient Decision Aids Standards [32], our design process was structured in multiple testing cycles, each involving small to medium user samples (3‐31 participants, no overlap between cycles). This aligns with cognitive ergonomics standards [41-43], which state that iterative testing with small samples efficiently identifies approximately 85% of usability issues per cycle. After each cycle, the team met to review findings and make decisions about modifications. The citizen panel met separately throughout the project to offer advice from their viewpoints.

We used Zoom to conduct user testing interviews. We evaluated whether VaxDA-C19 met its communication goals, and then applied a rapid thematic analysis [44,45] to qualitative data from interviews, comparative testing, and think-aloud sessions. Two trained research associates (EP and HH) independently coded transcripts, first extracting functional requirements (ie, needed features, content adjustments, and navigation flows), and then flagging any usability issues. We resolved disagreements through discussion with the lead author (DE) and principal investigator (HOW) until we reached a consensus. Finally, DE, EP, and HOW organized the agreed functional requirements and identified bugs into a clear, detailed plan to guide the development team. We reported the qualitative components in line with the SRQR (Standards for Reporting Qualitative Research) and provided the completed checklist in Checklist 2 [46]. We documented cycle-by-cycle decisions and changes to support transparency.

Across all cycles, we presented participants with increasingly higher-fidelity versions of VaxDA-C19 prototypes. We asked participants to summarize the information, identify points of confusion or unanswered questions, and react and respond to wording.

Across the cycles, 45 participants responded to our user testing recruitment efforts, of which 34 completed one iterative user testing phase (cycle 1: n=9, cycle 2: n=22, and cycle 3: n=3). Prospective participants who did not complete the testing did so because they did not wish to finish the study (n=2), were unavailable for any of the times scheduled for testing (n=4), responded after we had already reached saturation in findings (n=2), or for other reasons (n=3). Among the 34 participants, the mean age was 49 (SD 16) years. The majority of our study participants were women, Canadian-born, and English speakers. In total, 11/34, 32% of participants reported an ethnocultural identity other than White. Detailed demographics are shown in Table 2.

Following the cycle 1 findings summarized in Table 3, we implemented several targeted modifications to improve the patient decision aid. We explicitly clarified that each icon represented 1 individual in a given population to address confusion around probabilistic visualization. Considering mixed feedback on icon-array layouts, we retained both designs for further testing.

We revised the health care provider message to a more neutral tone, addressing concerns about pressure or discomfort. Finally, to improve the vaccine-confidence content, the structure was reorganized to enhance clarity. We simplified technical language, clarified the opening sentence, and incorporated more transparent information about potential side effects, in line with participants’ expectations regarding tone, accessibility, and completeness.

Building on participant feedback in cycle 2, we made several targeted changes to the next iteration of the VaxDA-C19 decision aid to enhance its clarity, transparency, and trustworthiness. The photos and names of scientists were moved to a dedicated “About Us” section to reduce perceptions of promotional content and clarified their roles to reduce confusion. While keeping the explanatory text detailed, we simplified complex terms to increase accessibility. In response to recurring concerns and unanswered questions about vaccine technology, safety, variants, and side effects, we also expanded the FAQs section.

Following cycle 3, we implemented targeted modifications to improve navigation, content clarity, and overall UX in VaxDA-C19. We enhanced button visibility and repositioned key navigation elements, adding clear user instructions to support a smoother transition. To increase content accessibility and transparency, we simplified medical terminology and added comparative data on vaccine efficacy and side effects directly within relevant sections. To streamline interactivity, we introduced an avatar bypass option and clarified the functionality of values clarification sliders through revised instructions and enhanced visual design. External expert recommendations were also applied to improve mobile navigation, refine spacing, structure vaccine menus more clearly, and integrate contextual guidance through informative pop-ups.

VaxDA-C19 evolved from an initial Google Slides mock-up into a fully functional, bilingual web-based application (Figures 4 and 5). Key technical improvements include transitioning to a responsive HTML5 and JavaScript architecture, with real-time content synchronization (Multimedia Appendix 5). Major features include automated visualizations for herd immunity, refining comparative interactive features, and enhanced icon arrays. These refinements ensure that VaxDA-C19 remains human-centered, responsive to evolving vaccination guidelines, while improving usability and user trust.

This study aimed to iteratively develop and refine a bilingual (English and French) web-based patient decision aid (VaxDA-C19) to help people in Canada make evidence-informed, values-congruent decisions about COVID-19 vaccination. VaxDA-C19 includes interactive components such as customizable avatars, personalized icon arrays constructed out of those avatars, an optional, animated, theory-based visualization of community immunity, and a values clarification method. We included these features based on their capacity to facilitate effective communication, optimize cognitive load, enhance empathy, and provide clear, evidence-based information on vaccine options, benefits, and risks [32,48,49,52,61-63]. We present 3 principal findings from this work.

First, a recurrent theme in participant feedback was the need for more detailed and transparent information about vaccine efficacy, variant coverage, immunity duration, and the inclusivity of clinical trials. To address these concerns, enhance transparency, and hopefully reinforce trust in the information presented, rather than offering a brief list of references in a separate section at the end of the document or website, as is commonly done in patient decision aids, we integrated explicit links to peer-reviewed studies and health authority documents immediately following each statement. We used the word “citation” for each reference and linked directly to open-access resources whenever possible. We also designed visual strategies, such as including images of scientists, to increase credibility [65-67]; however, responses to this approach varied. Some participants found this approach reassuring, while others perceived it as manipulative or politically motivated. To ultimately mitigate these concerns, we removed images and prioritized emotional neutrality and transparent, nonjudgmental communication [1,4,15,68-70]. This approach aligns with the trust determination theory, which emphasizes how credibility, inclusivity, and clarity foster public confidence and reduce skepticism [71]. Similarly, the extended parallel process model highlights how fear, distrust, and perceived governmental or institutional biases influence message reception, potentially diminishing the effectiveness of health communication [72].

Second, rather than seeking an elusive one-size-fits-all information presentation, presenting a simple overview of essential points and allowing people to easily access more detail when they want it may better serve a wider range of information needs. People with lower health literacy, who often express greater vaccine hesitancy, find more transparent explanations of risks and benefits particularly helpful [68,73-75]. Some participants in our study similarly emphasized the importance of clear, simplified explanations, particularly for individuals with varying levels of health literacy and also for people who lack the time or interest to delve deeply into information about vaccines. However, as noted earlier in our first point, other participants wanted far more detail. These tensions between some people’s desires for comprehensive information and others’ desires for more succinct information highlighted the complexity inherent in balancing simplicity and visual appeal with informational accuracy [76,77]. To address these competing priorities, we revised medical terminology, restructured content for better readability, and ultimately presented a summary of key evidence to all users, followed by accordion menus, which allow users to expand sections to read more about a given topic. By giving people control over the amount of detail they receive, we hope VaxDA-C19 will serve the needs of people who want a summary of essential points as well as people whose questions are not answered by simplified public health messages. Future research should continue to examine optimizing content presentation to provide sufficient detail without overwhelming users [78].

Third and finally, specific challenges associated with developing a patient decision aid for COVID-19 vaccination included a novel pathogen that launched a global pandemic, novel vaccine technology, continuously evolving evidence about vaccine effectiveness and adverse events following immunization, changing vaccine products and authorizations, and evolving recommendations for use. The fast-evolving COVID-19 vaccination landscape required proactive content management. We used flexible structures like JSON templates to enable timely updates aligned with new evidence and public health recommendations. This technical adaptability, combined with sufficient time commitments on the part of team members, should help us ensure that the VaxDA-C19 will remain scientifically accurate, while also facilitating version control, scalability, and potential future reuse for patient decision aids about other vaccine-preventable diseases.

The COVID-19 pandemic led to the rapid development of digital patient decision aids aimed at addressing vaccine hesitancy, which intensified due to misinformation and information overload (infodemic) [79]. Several countries, including France [80], the Netherlands [81], and Australia [82], introduced digital tools to support informed decision-making. Compared to static informational approaches, these interactive decision aids improved user knowledge, reduced decisional conflict, and increased satisfaction with the decision-making process [79,81]. Despite these benefits, major challenges persist, including potential biases in content presentation, variations in complexity, and inconsistent accessibility, all of which affect usability and effectiveness [79,81].

International evaluations identified key factors that contribute to the success of digital patient decision aids. Co-design approaches that involve health professionals and target populations enhance relevance and usability. Regular updates ensure content accuracy and alignment with evolving public health guidelines. User-friendly interfaces and personalization based on individual risk assessments further support engagement and informed decision-making [79-82]. Certain digital patient decision aids have also integrated practical features, such as direct scheduling links for vaccination appointments, to facilitate action [79,81]. Similarly, in our study, multiple cycles of qualitative usability testing within an iterative, human-centered approach helped identify and address navigation, comprehension, and usability issues early in development. Small-scale usability tests effectively detected key issues, allowing for rapid refinements. Cognitive ergonomics literature supports this strategy, indicating that explanatory texts provide more explicit contextual cues and facilitate information processing, particularly for individuals with lower health literacy [83-85].

Limited linguistic and cultural adaptability restricts the reach of many existing digital patient decision aids, particularly among culturally and linguistically diverse populations [79,81]. VaxDA-C19 aims to address this limitation initially through its bilingual content in English and French, the official languages of Canada. Following evaluation in an online randomized controlled trial, we also plan to translate it into other major languages spoken in Canada.

Although digital patient decision aids may influence vaccine intentions, evidence directly linking these tools to increased vaccine uptake remains incomplete. Ongoing randomized controlled trials, such as SMART-DA in Korea [86], aim to assess their impact on key outcomes, including vaccine intentions, knowledge retention, decisional conflict, and emotional responses. Findings suggest that patient decision aids improve preventive health behaviors by enhancing decision quality rather than by applying persuasive messaging [87,88].

Our study aligned with previous research, suggesting that effective communication in patient decision aids such as VaxDA-C19 requires careful consideration of emotions, trust, and perceived biases [1,4,10,89]. Emotional responses play an important role in how individuals interpret vaccine-related messages [1,15]. In our study, some participants expressed skepticism, anxiety, or distrust, particularly if they perceived wording as being politically charged, overly promotional, ambiguous, or institutionally biased. This finding aligns with other studies [90], and these reactions overall align with Sandman’s risk=hazard + outrage model, which posits that public outrage or mistrust can overshadow objective risk assessments, leading to defensive responses such as rejection or disengagement [4,18,90,91]. Preexisting attitudes toward vaccines may also influence people’s reactions to interventions such as VaxDA-C19. Future research should incorporate such considerations into study design.

Our study also confirmed that intuitive navigation, visual clarity, and balanced content presentation played essential roles in usability. These findings align with existing literature, which demonstrates that interactive and personalized elements improve engagement and comprehension in digital health interventions [52,92,93].

Our study has 2 main limitations. First, while recruiting user testing participants through tailored Facebook advertisements allowed us to reach people living in different parts of Canada with diverse backgrounds and who are online and therefore potential users of a web-based patient decision aid, no recruitment platform can perfectly represent the broader population. We note that our user testing participants were, as a group, slightly older than the broader group of Facebook users in Canada. Specifically, among Facebook users in Canada, the largest age group (around 25%) is between 25 and 34 years, with the next largest group (19%) being 35 to 44 years [94]. We may have recruited more people older than these groups due to increasing interest in COVID-19 vaccines with age. We note that the age group we recruited was approximately representative of the desired population for a patient decision aid about COVID-19 vaccines. Older people in Canada were already receiving strong recommendations to obtain the vaccine, whereas people in the age range of our user testing participants were likely to be both candidates themselves and may also be parents of children who were candidates for COVID-19 vaccination at the time of the study. Our study participants also included more people with higher levels of education than the general population in Canada. People with more education may seek more information about health compared to people with less education. Additionally, we note that, as a group, participants included a substantial proportion of people without a college or university degree. In other words, although there were many people with higher levels of education in our study, there were also many people without such education. No one intervention can be everything to everyone; our goal with this intervention was to provide a middle ground between the simplified information available on public health websites and the detailed information buried in technical documents and academic literature. Because vaccination campaigns are most successful when they reach as many people as possible, an approach with different components for different subpopulations (eg, an online patient decision aid for people with enough motivation to seek more information, community-based education meetings for people who trust community leaders, and simple information for people who have no questions and simply want to know where to get their vaccine) is likely to achieve the best population-level protection. Second, the challenge of balancing comprehensive content with cognitive load remains potentially unresolved, particularly in adapting information for users with varying levels of health literacy. While brief, simple messaging can be most effective in some contexts and for some people, the resulting lack of detail and missing explanations can be a problem for others. Variability in patient decision aid formats, development methodologies, and user preferences suggests that no single approach fits all contexts [25,95]. We therefore chose to prioritize user control over information presentation, trusting that many people will know what they need. Nevertheless, additional research should investigate these aspects further to refine content presentation and improve the effectiveness of digital patient decision aids.

Our study also has 2 main strengths. First, we used a rigorous iterative design based on human-centered methodologies, a multidisciplinary approach, bilingual validation, and structured engagement of a range of interested parties, including user testing, a citizen panel, UX interviews, and expert reviews. The collaboration between a diversity of researchers, UX specialists, members of the public, and health care professionals helped refine the patient decision aid, contributing to its scientific reliability and usability. User testing participants from different backgrounds offered different responses and reactions, providing a range of feedback to which we were able to respond. Complementing the contributions of these members of the public who participated in user testing and saw VaxDA-C19 at a single time point, members of the citizen panel saw VaxDA-C19’s evolution over time and were able to comment at a higher level on how well or poorly the design was responding to its original goals as well as the evolving pandemic context. A wide range of experts brought insights from related projects, literature, and guidelines and also lent their expertise to verify VaxDA-C19’s content and design. Systematically integrating all these people’s feedback on an ongoing basis allowed for continuous adjustments. Second, we believe that VaxDA-C19 fills an unmet need by bridging the gap between highly technical vaccine information and highly simplified public health messaging. As COVID-19 vaccines continue to be offered to people in Canada, we hope to help support high-quality decision-making, thus contributing to the health of people in Canada and to public health more broadly. The platform itself may also be adapted to other vaccines.

We plan to assess the effects of VaxDA-C19 in an online randomized controlled trial (ISRCTN91743566). Specifically, participants who are eligible to receive COVID-19 vaccines will be randomly assigned to 1 of 3 arms: a control group receiving no additional information, a usual information group directed to the Public Health Agency of Canada’s COVID-19 vaccination web pages, or an intervention group receiving access to VaxDA-C19. We will assess effects on the primary outcome of vaccination intentions and secondary outcomes: knowledge, emotions, decisional conflict, vaccine uptake, and decisional regret. We will also examine effects on trust in the information provided, comparing only between the usual information and VaxDA-C19. Finally, we will conduct a follow-up study 3‐6 months later to ascertain whether or not participants received COVID-19 vaccines.

VaxDA-C19 is an online patient decision aid designed and developed to help people in Canada make evidence-informed, values-congruent decisions about COVID-19 vaccines on an ongoing basis. An iterative, human-centered design approach and a multidisciplinary team bringing diverse perspectives led to a flexible application that we hope will prove useful for years to come.