Decision Aids for Patients With Head and Neck Cancer: Qualitative Elicitation of Design Recommendations From Patient End Users

Introduction

Head and neck cancer (HNC) is the sixth leading type of cancer by incidence worldwide and diagnosed in approximately 4300 Canadians per year [1]. Based on clinical and pathological manifestation, patients with HNC may be provided an option of surgery, radiation, chemotherapy, immunotherapy, and combinations thereof, each influencing morbidity differently [2]. For patients to participate in their care, they must comprehend their disease and treatment options, consider their own preferences, participate in decision-making to the degree they wish, and make a decision consistent with personal preferences. Patients who direct decisions, even if more than anticipated, fare better on all decision-related outcomes, emphasizing the need for oncologists and surgeons to endorse, facilitate, and support patient participation in treatment decision-making [3].

Some patients with HNC may struggle to take part in decision-making regarding treatment options because of the complexity of the information that needs to be accurately conveyed and understood [4,5]. This can be particularly challenging in the context of a globalized world consisting of cultural differences and varying health literacy levels, increasing the need for enhanced transparent communication of risk and outcomes associated with treatments [5,6]. Furthermore, research on risk literacy in medical decision-making shows that across different cultures, people often struggle to grasp the prerequisite concepts necessary for understanding health-related risk information such as numbers, graphs, and basic medical facts. Errors occur because inappropriate information formats complicate and mislead adaptive decision makers [6].

Decision aids (DAs) “are interventions that support patients by making their decisions explicit, providing information about options and associated benefits/harms, and helping to clarify congruence between decisions and personal values” [7]. DAs complement, rather than replace, counseling from a health care practitioner as an interactional strategy to facilitate patient involvement and contribute to patient concordance [4]. DAs are useful when the best treatment strategy depends on a preference for the benefit-harm trade-off inherent in a particular choice.

A scoping review conducted on electronic DAs (eDAs) for patients with HNC returned 12 relevant articles that discussed 5 different patient eDAs [8]. The scoping review confirmed the value of eDAs in this population supporting “further research and development in this area.” The scoping review did not, however, reveal detailed technical features of existing eDAs. The patient eDA developed by Petersen et al [9] remained available online, so features were viewable, but the other 4 eDAs were not available for viewing. The aim of this study is to respond to the gap in the literature on preferred DA architecture by interviewing survivors of HNC on the utility and potential visual designs for an eDA for patients with HNC to encourage informed and collaborative decision-making. Through phenomenological inquiry, the goal was to answer the following questions: (1) What do survivors of HNC think about the utility of electronic decision-support tools utilizing visuals for patients with HNC? and (2) What suggestions do survivors of HNC have on a potential design for visual features and core components of a prototype eDA?

Results will be used to inform next steps in the development and integration of eDAs in HNC care.

Methods

Prototype Visual Development

A point-of-reference (ie, example screen visualizations) mock-up of a DA was developed for the interviews (Figure 1), containing a graph used in the symptom management clinic at BC Cancer-Victoria (BCC-Vic). A line for surgery is included in the graph, serving as a placeholder only, as surgery is managed by surgeons external to BCC-Vic. The graph was displayed to participants in the interviews. Two symptoms were selected for visual demonstration, oral mucositis (Figure 2; photo credit [10]) and radiation dermatitis (Figure 3; photo credit [11]), noted by a shape on the curve. Hovering the cursor over these shapes triggers a pop-up image of the side effect, increasing in severity from left to right in accordance with the Common Terminology Criteria for Adverse Events (CTCAE) [12] grade. The information used in the design and graphs shown to participants was based on the evidence-based literature on HNC [13-15]. A key for these shapes was not provided, as they were meant to provide a demonstration of the mouse-over effect, where information could be obtained by moving the cursor over areas of the screen.

‎

‎

‎

Recruitment and Materials

Two phases of recruitment were conducted through convenience and purposive sampling, respectively, with a target of 6-12 participants or until no new knowledge was obtained [16]. Inclusion and exclusion criteria are described in Textbox 1. Upon consent, demographic, diagnosis, and treatment plans were extracted from BCC’s electronic medical record.

Interviews and Analysis

One-hour long, semistructured interviews, including field notes and reflective questions [17], were conducted following an interview guide. Interview questions were open ended and fell under the categories of introduction, background, experience, e-tools, evaluation, and conclusion. The interview questions were designed to elicit participants’ experience in learning about side effects of treatment along with their impression of the visuals presented to them (in terms of how well they helped explain treatment and side effects; see Multimedia Appendix 1 for the full interview script). For consistency, interviews were conducted by a single researcher (ES), who is most familiar with the subject area and experienced in patient interviews. No other researchers participated in the interviews to remove potential power dynamics that may disempower or affect participant opinions. Participants were introduced to the interviewer including their motivations and background behind this research.

Interviews were held virtually through Zoom (Zoom Video Communications, Inc.), to allow for screen sharing while also ensuring participants safety during COVID-19; all participants attended from their home. Participants were provided a demonstration of the eDA but did not interact with it. Participants were encouraged to ask for further demonstration if needed and were welcomed to use Zoom’s annotate feature. All interviews were video-recorded, transcribed, and analyzed using Taguette software [18,19]. Reflexive thematic analysis was performed on annotated transcripts, field notes, and reflections, where an open and iterative process was applied to coding following the 6-step approach of Braun et al [16]. A set of tagged codes (see Multimedia Appendix 2 for the codes) was used to identify concepts in the transcripts related to the research questions. These codes were used to develop concepts or “domain summaries at the start of the analytic process.” [16] Upon further engagement and critical reflection, related concepts were then grouped into themes [16]. Coding was completed by the researcher ES who was most familiar with the data and then verified by AWK. Data validation was conducted by sharing preliminary themes with participants for correction, modification, or confirmation.

Ethics Review

Review Ethic Board (REB) approval was granted on January 18, 2021, from the University of Victoria Human REB (University of Victoria Study #BC20-0546) and UBC/BC Cancer Agency REB (H20-02307). All participants provided written informed consent to take part in the study.

Results

Demographics

A total of 12 participants were interviewed: 6 recruited from the Head and Neck Support Group at BCC-Vic and 6 identified by an oncologist. Participant ages ranged from 34 to 76 years, with half (n=6) of the participants diagnosed with stage III cancer and nearly half (5/12, 42%) undergoing surgery prior to radiation. As many as 7/12 (58%) participants were offered adjuvant chemotherapy: 3 accepted cisplatin and the remaining 4 declined chemotherapy; 9/12 (75%) participants accessed the symptom management team on a regular basis, while 10/12 (83%) connected with a counselor through treatment (Table 1).

Thematic Categories

A total of 61 codes were developed and applied, pointing to 17 concepts related to eDAs and the prototype. These 17 concepts were organized into 9 themes, separated into 2 categories (Table 2).

Category 1: eDAs and Decision Support

The term “decision support tools” or “decision aids” was new to 75% (9/12) of participants, while 3/12 (25%) were familiar due to their line of work (Table 3). Regardless of eDA experience, there was strong support for the concept of using visual aids to enhance the explanation of treatment options, their potential side effects, and timeline, demonstrated by SUPPORT OF CONCEPT being the most frequently applied code (see Multimedia Appendix 2).

All participants expressed the usefulness and value of photographs, videos, and graphs. Learning styles appeared to influence the degree of value placed on graphics, with visual learners demonstrating the most enthusiasm. Those whose information needs were adequately met through their oncologist’s verbal explanation still suspected images would reinforce and amplify their message.

The codes USEFUL and VALUE were the most frequently tagged alongside PHOTOS and VIDEOS, often describing statements on how viewing photos of side effects would have clarified and adjusted treatment expectations. Lastly, several participants felt that viewing severe side effects prior to initiating treatment would serve as a catalyst, motivating prophylactic, intensive therapy and self-care strategies to optimize outcomes and quality of life. For example, one participant felt that viewing images of radiation dermatitis prior to treatment would have improved their compliance with skin care guidelines.

An unexpected theme was the potential versatility of eDAs in this area. Participants envisioned the eDA serving not only as an educational tool, but also as a communication tool to use both within and outside clinical encounters. Of the 12 participants, 3 (25%) felt that an eDA would prepare them for oncologist appointments by prompting the creation of informed questions. Participants also described their interest in using an eDA with the interdisciplinary care team, friends, and family. Two participants, for example, described the exhaustion that resulted from the draining task of updating loved ones on their day-to-day well-being.

Category 2: Evaluation of Core Visual Components of a Prototype eDA

Initial reactions to the screens and visualizations that were presented to participants during the interviews were positive, with several participants exuding immediate enthusiasm. All participants described their favorite features in a desired prototype eDA. The visualizations presented were most praised for their simplistic design using a combination of pictures, colors, graphs, and text that offered a snapshot of treatment (Table 4).

Although there was unanimous agreement on the usefulness and value of visuals shown, the least consistent responses regarded the complexity of the graph. All participants felt they could read the graph without an explanation; some even read it “instantly.” While some felt the eDA visualizations may be too simplistic, one-third of participants (n=4) expressed a concern that the graph is overly complicated for the public, which includes individuals of varying education levels, ages, and health literacy. One participant expressed concern that the use of technology in medical care may be disconcerting to the elderly population, but suggested that if the difficulty is technology navigation, then the oncologist can operate the device.

Although there was strong support for the basic features of the prototype, every participant provided suggestions for improvement that pointed to their desire for customized information. Some mentioned features to consider customizing including treatment type (option to view a single treatment only), radiation dose (total radiation grays or number of fractions), filtering by side effect, adjustable timelines, results stratified by sex, detail within photos, option to display information source/reference, and associated statistics.

There were mixed opinions on the appropriateness of medically explicit images that become increasingly gruesome as the CTCAE grading score advances. Every participant agreed to view these images, but several appreciated the invitation to skip. Those in favor of including vivid detail argued that it is an accurate and honest display of treatment realities, while those hesitant to include graphic photographs were concerned that it would instill fear and possibly dissuade the pursuit of treatment. Two participants suggested offering the option of viewing an artistic sketch with less detail instead of a photograph. Another participant suggested emphasizing that symptoms are manageable with guidance from the health care team, using this as a segue to discuss supportive care measures such as medications and self-care techniques. Furthermore, they suggested it is “really important” to make clear that these strategies will “not cure it but will help them.”

There were mixed responses on which type of device (eg, desktop/laptop, tablet, phone) a prototype should be designed for, discussing the advantages and limitations of each device. Those with strong opinions based their response on how they envisioned themselves using an eDA. For example, one participant suggested a mobile phone interface for the benefit of portability, as they envisioned using it on public transport. A different participant preferred a computer to view details on a larger screen. Lastly, another participant thought a tablet to be the most practical for clinic use as it could be passed between the patient and health care provider while using a screen large enough for comfortable viewing.

SUGGESTIONS FOR IMPROVEMENT was the third most frequently applied code. Most recommendations fit into the following 7 categories:

• Additional information to add: Participants offered creative suggestions on information and features to add to the prototype, summarized in Table 5
• Legend: As previously described, 2 shapes dotted the graph representing 2 different side effects. These shapes were used for demonstration purposes only and would not be part of the final prototype, but participants found this confusing and were concerned that symbols would complicate the graph. They liked the idea of selecting the side effects of interest from a list on the side.
• Links to additional resources: Two participants highlighted the opportunity to link to supportive care resources, such as Inspire Health, BCC Patient and Family Counseling, and meditation exercises.
• List of side effects: Several participants recommended expanding the list of side effects to encompass rare side effects, such as fatigue, headaches, hair loss, neuropathy, tinnitus, vision problems, and dermatological changes.
• Interactive tracking: There was great enthusiasm for building out interactive tracking options. Some ideas were options to track diet intake, speech language pathologist–prescribed exercise routine, mental health, and symptoms such as pain and nausea.
• Add videos: Several participants suggested adding videos, with 1 participant particularly adamant on the value video could add, including “video clips of people who were able to speak to their experience.”
• Late side effect/life after treatment: Several participants requested additional information on life after treatment.

Discussion

Principal Findings

This study contributes to the literature on the utility and patient-recommended requirements and design of eDAs. Participant feedback was overwhelmingly supportive of using visuals to support explanations of treatment and potential side effects. All participants could imagine themselves using an eDA as an education and communication tool and agreed to contribute to the next phase in the full development of eDA, further demonstrating their endorsement.

Use and Value of Visual Aids

Literature demonstrates that patients prefer images to demonstrate benefit and harm trade-offs in health as they are perceived as easier to understand [21]. Verbally relayed numerical facts were least effective in encouraging a specific health decision while graphical information was the most preferred, demonstrating that “consideration should be given to developing visual aids to support shared clinical decision making” [21]. This is consistent with our findings: every participant agreed that the use of images alongside verbal explanations would complement or augment an oncologist’s message. Furthermore, “using transparent information formats enhances risk comprehension, communication, and recall and helps people make better decisions about their health” [22].

Malleability of the eDA

Participants were interested in adding interactive options; most were intrigued by symptom tracking to auto-populate a personalized curve on the graph. Here lies an opportunity to incorporate patient-reported outcome measures, standardized instruments used to capture patients’ perceptions of their health status, functional status, or health-related quality of life [23]. Integrating patient-reported outcome measures within the eDA would support the gold standard of patient self-administration without interviewer interpretation [24]. Furthermore, literature demonstrates the benefit of real-time symptom tracking during cancer care, improving health outcomes and communication with health care providers [25,26].

When considering participants’ desired features alongside cumulative feedback, customizability is required to suit the range of learning styles, needs, and preferences. A potential solution to varying information needs may include suggestions from Table 5 and then allowing users to filter and sort the viewable data. For example, participants unanimously supported the idea of selecting side effects of interest from an extensive list of potential side effects. Taken further, graphics could be shown as real-life photos or sketched images to better meet participants comfort level with medically explicit images. Embedding customizability may also address the discourse regarding the suitability of the graph. The mixed responses align with the literature on graph literacy, which “reveal[s] that people, regardless of their numeracy skills, differ substantially in their ability to understand graphically presented quantitative information about health” [22]. Furthermore, Nayak et al [27] tested graphical interpretation of a visual dashboard on patients with prostate cancer and despite 78% of participants having a college education, variation remained in graph literacy results.

Prognosis Versus Quality of Life

Our visualizations designed and presented to participants did not include information on prognosis or survival outcomes to simplify the inceptive eDA. The information presented was designed to be used in the presence of an oncologist who could personalize the message and discuss prognosis on a case-by-case basis. Previous research, however, demonstrated that comprehension of medical information on survival and cancer treatment options is equal or better than when patients are shown the same information with the addition of mortality statistics [28]. In a viewpoint paper on his personal experience with advanced stage tongue cancer, Kushniruk [29] argues for “patients to be more informed about choices and statistics, including the meaning of survival curves in relation to different treatment options.” Although the desire for incorporation of information about prognosis in DAs was not a prominent theme in our findings, the importance of survival cannot be overlooked and should be considered for inclusion in future iterations.

Limitations

Selection bias was introduced through the convenient sampling method. Recruitment initially targeted patients at BCC-Vic’s Monthly Head and Neck Support Group, comprising those closely matching the inclusion criteria. This group, however, may more likely represent survivors who experienced challenges through treatment, thus introducing a source of bias. To help offset this, an equal number of participants (n=6) were recruited through purposive sampling to diversify the demographics. Because of the nonprobability sampling methods, the results cannot be generalized to a wider population. Future work should consider including the perspective of those without cancer (ie, cancer naïve) and survivors along the cancer trajectory (ie, never diagnosed through to long-term survivorship). Furthermore, race was not included in the demographics and future work should include diverse racial representation, including historically underrepresented groups.

There are limitations to the mock-up design method used for presenting design ideas and visualizations that did not adhere to a formal development process, such as the Ottawa Patient Decision Aid Development eTraining or the International Patient Decision Aid Standards Collaborations [30]. It was instead based on the research teams’ phenomenological and personal experiences with HNC; graphs used with patients at BCC-Vic; informal, preliminary feedback from patients; and results of the scoping review. The designs and visualizations were sufficient for this study purpose of applying a user-centered design [31] to this innovation with plans to build through the co-design methodologies of Kushniruk and Nøhr [32] and Kushniruk and Patel [33].

Repeat interviews to improve data richness were not conducted as these interviews were intended to be preliminary, setting the stage for the next phase of the project that will include in-depth workshops with participants. Additionally, rich data were collected from the single interviews and additional interviews were not yet deemed necessary.

Lastly, due to the specificity of the target population, results cannot be generalized to other tumor types. Further requirements analysis and testing will be required prior to expanding to other populations.

Conclusions

This research highlights the value of incorporating visuals into technology-based innovations to support patient decision-making in oncology care. All participants felt an eDA, used with their oncologist, would support an enhanced and transparent explanation of treatment and augment the quality of the consult. Participants liked the simple design of the prototype visualization but desired customizability to adapt the eDA to their individual information needs. This research highlights the value of user-centered design, rooted in acceptability and utility, in medical health informatics, recognizing cancer survivors as the ultimate knowledge holders.

The next steps include applying the co-design methodologies of Nayak et al [27] and Kushniruk [29] to develop a fully functioning eDA to be tested with patients with cancer, incorporating visualizations and the feedback described in this paper. This will consider implications to clinical workflow, including physician’s time, and improve accessibility by ensuring visualization is amendable to printing for those without digital access. Usability testing of the eDA will be conducted in a clinical setting at BCC-Vic and the University of Victoria through a newly funded follow-up study (ie, the Head and Neck Cancer Application for Patients and their Partners [HANC APP Study]).