The design of a low literacy decision aid about rheumatoid arthritis medications developed in three languages for use during the clinical encounter

Given that the majority of patients newly diagnosed with RA are started on methotrexate [14],[15], we created our tool for the clinical context of an established RA patient with moderate to high disease activity who, at a minimum, has failed a trial of methotrexate. A 2013 study reported on the non-inferiority of triple combination synthetic DMARD therapy compared to adding a biologic [15] to methotrexate such that there is no one algorithm to direct medication choice in RA. At the time of the decision aid development, twelve DMARDs (4 synthetic and 8 biologic DMARDs) had been approved by the FDA and were commonly used in practice. While there are more than 12 FDA-approved DMARDs for RA, several are very rarely used in clinical practice.

The research protocol for this study was approved by the UCSF Committee on Human Research (IRB Number: 10-02339). All patient participants gave their written informed consent to be part of the study. All individuals whose images appear in the guide (Figure 1) signed written consent for permission to use their images for the purposes of the guide (research and education).

Using the design of the diabetes issue cards as a guide, we convened a diverse group of individuals with expertise in RA (clinicians and patients), decision aid research (VMM), design (TM, MB, DR), health literacy (JA, DS), clinical and qualitative research methods (LT, EY, CJK). We adopted the design approach described by Breslin et al. and encouraged all members of the team to bring methodologies and ideas from their respective fields to the table. This approach, with foundations in the design discipline, draws upon elements of field ethnography (which involves direct observation in clinic about how patients and clinicians discuss treatments) as well as participatory action research where both patients and clinicians contribute to the research by allowing participant observers (or third party researchers) to observe actual clinic visits. In addition, patients contributed by informing the researchers about content of the decision aid and evaluating prototypes both on an individual basis in cognitive interviews and as a larger group in a patient advisory board. Results from the cognitive interviews fed back directly into decisions regarding design, number of issue cards, and as a factor in how many cards were ultimately included in the final deck. Content from the interviews was discussed in team meetings and phone calls with consultants and designers.

Patients participated in two ways. The first was to invite patients from several previously conducted focus groups (one each in English, Spanish or Cantonese led by a bilingual, trained focus group moderator using a common interview guide that had been translated) to sit on a patient advisory board which met approximately every two months during the 18-month-long development process. The purpose of the initial focus groups was to gather data to inform the content and delivery of an adapted medication summary guide for RA medications as well as to explore patients’ perspectives on participation in decision making with their clinicians. The Grounded Theory Method [16] using constant comparison to determine similarity and differences within and between groups was used to analyze the transcripts [17]. We recruited heavily from the county hospital clinic for the focus groups, of whom over half had high school education or less. We then selected members for the board who had a range of literacy and education levels and continually emphasized the low literacy principles for the tool development. Of 9 board members, one-third had limited health literacy. The common language was English; however, several members were bilingual (either in Spanish or Cantonese). Cantonese, not Mandarin, is the predominant language spoken by Chinese speakers in San Francisco. Patients also participated during clinic visits at one of two clinics, the university-hospital based arthritis clinic or the county-hospital based RA clinic, when the prototypes were being tested. Clinical rheumatologists provided input as they tested the prototypes in clinic and offered feedback in research meetings.

In addition to using the diabetes issue cards as a template, we followed six steps for the development of low literacy materials outlined by Seligman et al. [18] where the authors describe a process whose main purpose is to create materials which will increase knowledge as well as activate low literacy patients toward healthier behaviors. The six steps are presented in Table 1. Seligman et al. underscore the importance of mapping concepts onto a behavioral theory, such as the social cognitive theory, which suggests that materials should improve knowledge, positively influence outcome expectations, emphasize facilitators of behavior change, address barriers to behavior change, and facilitate the creation of goals [19].

The funder of this project, the Agency for Healthcare Research and Quality (AHRQ), published a summary of their comparative effectiveness review of RA medications for patients and clinicians [20],[21]. Results from this review were used as the basis of the literature review for the decision aid. Additional systematic reviews of DMARDs and classes of DMARDs [22]-[25] were used to supplement the AHRQ review. In addition to gathering evidence-based reviews on DMARD safety, tolerability and comparative effectiveness, we also collected data on how medication discussions were taking place in clinics, as well as eliciting patient concerns about starting a new medication in focus groups, cognitive interviews, and from the advisory board. The process of combining data from the evidence synthesis and direct observations with feedback from various stakeholders is illustrated in Figure 2.

The development team reviewed and discussed issues that emerged from the literature review, clinical observations, and patient comments, and began to create prototypes of the cards. Each card was devoted to a particular issue, such as “side effects” or “cost” as based on the diabetes card design. The number, content, and design of the cards varied throughout their development. Clinicians were asked to use prototypes in clinical encounters where a third party (GEY) observed and recorded how the tool was used, the content of conversations around medications, and the level of patient involvement in the encounter. Prototypes were evaluated in recorded one-on-one interviews with a member of the research staff (LT, JB) and RA patients, as well as non-RA patients (in role play) to gather insights into the content and design of the tool and its ability to generate a conversation. Mock clinic visits and actual clinic visits using the prototypes were then observed by the designer (TM) and other team members (LT, GEY) after which they provided feedback on the nature of the conversations. Health literacy experts also reviewed the prototypes.

In discussions with the patient advisory board, we sought to identify any issues that the cards did not cover and therefore expand the repertoire of issues. Most importantly, we examined whether the cards accomplished the goal of creating a conversation between patient and clinician. Iterations of the tool continued until the main patient issues and concerns were covered and the cards were consistently generating an exchange between clinician and patient about RA medications. This entire process took approximately eighteen months.

AHRQ funded this project and played no other role in this work.

The development of a low literacy medication decision aid (RA Choice) in three languages was feasible using a collaborative process with patient and clinician input at every step. There is a clear need for increased support and structure to promote and facilitate shared decision-making in RA, especially among more vulnerable populations with poorer health outcomes and barriers to communication like literacy and language. Decision aids are one way to promote patient-centered care [41]. There are several decision aids for patients with RA [42]-[44], however none have been developed for patients with limited health literacy or limited English language proficiency for use in the clinic encounter.

During our development process, certain decisions had to be made to respect clinician concerns about consultation time, inability to have comprehensive information on the cards and meet standards for a low literacy tool, and be cognizant of not overwhelming patients with too much information. We based our tool development on a process described for a diabetes decision aid, the overarching goal of which was to create a conversation between patients and clinicians. Both diabetes and RA are chronic conditions with multiple options for therapy requiring a high degree of self-management. Given the parallels between the two conditions and choices for therapy, the diabetes issue card design was an excellent template for the RA decision aid.

Our process was not without limitations. While we created the decision aid and translated (and back-translated) it into Spanish and Chinese, we found it difficult to fully involve Cantonese-speaking patients and research staff in the process. Due to time constraints we were unable to thoroughly pilot the tools with Cantonese speakers, interpreters, and clinicians in the clinics. There are both language and cultural adaptions which may enhance the use and efficacy of RA Choice given the opportunity to observe a broader number of encounters and this is a goal for future projects. More specifically, lessons learned for future work include a focus on a single language during development, more in depth cognitive interviews with Cantonese speakers, and to determine the best way, without using names or characters, to share information on medications (of note, only one of 7 Cantonese focus group participants could recognize the written name of their RA medications and relied on color and size of pill). In addition, background research on shared decision making preferences in these populations is needed. We chose to create a print tool (as opposed to web-based) which may limit its usability in clinics which are fully electronic, as well as limit, if effective, our ability to distribute it and keep it up-to-date without having to recall distributed cards. It may also make it more difficult to keep the tool updated with new medications or comparative effectiveness data. A web-based version of the tool may be more appropriate once exam rooms are better equipped for full participation of patients and providers and the investment justified by evidence of efficacy of the cards. It is also a challenge to create a design that appeals to all users, therefore an electronic tool tailored to each individual may have greater reach as well as be easier to update. The state of the comparative effectiveness literature made it impossible to accurately depict benefits (lack of head to head trials for all biologics) and the research team ultimately concluded that using a measure such as the ACR 50 misled patients (i.e., depicting one medicine as more effective, leading to more patients achieving an ACR 50, when head-to-head trials are either not available or suggest no difference). This led to a decision not to include an issue card on efficacy. The complete experience of the decision aid tool in the clinical encounter relies on the communication skills of the clinician to engage the patient in conversation and of the patient to engage in conversation and for both to use the tools as intended. In contrast to the development of the diabetes decision aid, we were not able to pilot test the cards in actual clinical encounters in high numbers consistently across the tool development (more so in the beginning stages) due to the fact that clinics would be sites for a pilot trial.

The tool, in and of itself, does not meet all International Patient Decision Aid Standards (IPDAS) Collaboration criteria, which were designed for standalone tools directed at patients. However, cumulative, including through this publication, the tool satisfies criteria for balanced evidence-based and up-to-date information, depiction of relevant issues using state-of-the-art approaches, and designed for purpose and setting [45]. The categories of the IPDAS criteria that were not met include: a description of the positive features (as discussed above a benefits issue card is not included in the deck), presentation of probabilities of outcomes (to do so would have added significant complexity to the design of the tool which in turn would create excess burden on patients with limited health literacy). In terms of values clarification, this takes place upon the introduction of the issue cards by the clinician when he or she asks the patient to “pick which issue you would like to discuss first” about a new medication. The act of allowing the patient to hold and select an issue card is a concrete way to allow the patient to clarify and express his or her values. By providing a copy of the decision aid in the form of a trifold pamphlet as well as the medication summary guide to take home, patients are also encouraged to share with family or friends what may matter most to them in choosing a medication – be it cost, mode of administration, side effects, etc.

The development of a low literacy decision aid for RA medications in three languages was feasible using a collaborative process with patient, clinician and design input. Our decision aid, which is intended to be used in the context of an RA patient with moderate to high disease activity despite being on at least one DMARD, promoted a conversation and exchange of experience on the part of the clinician and values and preferences on the part of the patient. It is our expectation that the use of RA Choice in the clinical encounter will foster more patient-centered care and enhance shared decision-making for all RA patients.

A clinical trial to test the final prototype of RA Choice is needed to determine its impact on patients and clinicians, its feasibility for use in usual care settings, and its impact on process and outcomes of care.

Role of funding: Funding for this project was from the Agency for Healthcare Research and Quality (R18 HS019209-01, PI Yelin). AHRQ had no role in study design; in collection, analysis and interpretation of data; in the writing of this manuscript; or in the decision to submit this paper for publication. Dr. Yelin, Ms. Trupin and Ms. Evans-Young also received funding from the NIH (NIAMS P60 AR053308).