Development of a decision aid to help dually enrolled veterans make informed choices on medication sourcing

Abstract

BACKGROUND:

Veterans enrolled in both Veterans Affairs (VA) health care and Medicare Part D can choose to obtain medications through the VA, Medicare Part D, or both, with each option differing in cost, coverage, and coordination of care. Poorly informed choices can lead to veteran frustration when their expectations are not met, delays in medication access, and increased risks. This study aimed to develop a decision aid (DA) to help veterans with diabetes make informed choices about medication sourcing (ie, whether to fill medications through VA health care only, Part D only, or both).

METHODS:

DA development was guided by the International Patient DA Standards and the Ottawa Decision Support Framework. Interviews with veterans and care partners informed the prototype design. Alpha testing with 18 end users (mostly veterans) and 12 stakeholders (pharmacists, doctors, payers, Medicare counselors, and others) assessed comprehensibility, usability, and acceptability. During beta testing, the feasibility of the revised DA was assessed during interviews with 20 end users and 8 stakeholders. For end-user interviews, a survey assessing decisional conflict, satisfaction, and knowledge was provided before and after respondents filled out the DA.

RESULTS:

Alpha testing feedback led to simplifying the cost and formulary comparison chart and expanding the medication list template to include more medications, fill location, and prescriber contact information. Based on beta testing responses (n = 16), the mean system usability scale score for the DA was 77.5 (SD = 14.4), suggesting usability. Beta testers also reported the DA to be acceptable in length (94%), balance (88%), and amount of information (81%). Based on pre- vs post-DA survey responses, decisional conflict was reduced, as indicated by an increase in the mean Sure of myself; Understand information; Risk-benefit ratio; Encouragement (SURE) score (pre-DA: 3.1, SD = 1.4 to all 16 respondents reporting a maximum SURE score of 4.0). Knowledge about VA and Medicare coverage of diabetes medications also improved: the proportion who answered all 5 comprehension questions correctly increased from 57% to 81%. Last, the proportion of respondents who reported being “very satisfied” with how they were currently filling their diabetes medicines (VA only, Part D only, or both) improved from 64% pre-DA to 81% post-DA.

CONCLUSIONS:

The DA developed iteratively was usable and acceptable and showed potential in reducing decisional conflict, increasing knowledge, and increasing satisfaction, so it may help Part D–enrolled veterans with decisions about medication sourcing.

Plain language summary

We developed a decision aid to help veterans with diabetes who are enrolled in Veterans Affairs (VA) health care and a Part D plan choose between filling their medications through VA health care only, their Part D plan only, or both. After using the decision aid, veterans seemed more aware of differences between filling their medications through VA health care vs their Part D plan and were more certain and satisfied about how to fill their medications.

Implications for managed care pharmacy

The decision aid in this study was developed for a population of veterans with diabetes who are dually enrolled in VA health care and Medicare Part D. Early evidence suggested that the decision aid would improve knowledge, reduce decisional conflict, and improve satisfaction, but further testing is needed. More resources like this should be developed to help patients better understand their prescription insurance.

Nearly 30% (n = 70,000) of Medicare-eligible veterans with diabetes choose to fill their medications through both Veterans Affairs (VA) health care and Medicare Part D (prescription drug insurance).¹ VA health care and Medicare Part D are independent of one another. To use either one, veterans must meet individual eligibility requirements and then enroll separately. A prior survey found that, despite already being enrolled in VA health care, veterans most commonly reported enrolling in a Medicare Part D plan to save money, cover costs of prescriptions from non-VA doctors, and get medications not available through the VA.² When making subsequent medication-sourcing decisions (ie, whether to fill their medications using VA health care, their Part D plan, or both), cost was often a major driver, likely because half of Medicare-eligible older veterans report income levels lower than $35,000.^3,4 When asked in interviews, veterans currently making medication-sourcing decisions requested information that would allow them to easily compare differences in their copays and find out which of their medications were available through VA health care vs their Part D plan, such as by using a side-by-side chart.⁴

Access to specific medications was another common driver for veteran medication-sourcing decisions, because the same medications may not be available through VA health care vs each Part D plan (eg, nonformulary) or to the same degree (eg, different prior authorization requirements).⁴ This is especially true for more expensive, brand-name-only medications when a specific drug product may not be, or remain, the preferred product. When making medication-sourcing decisions, veterans have also reported considering other factors, such as a preference for a VA or non-VA health care provider or convenience such as travel time and costs to routine clinical appointments, although these factors were reported less frequently than medication cost and access.^2,4 Veterans living in rural areas may be more impacted by convenience factors, although the ability to use mail-order pharmacies through both VA health care and Part D plans has helped for chronic medications.⁴

Although there are advantages for veterans who are dually enrolled in VA health care and Medicare Part D, such as increased access to medications and potential cost savings, there are also important disadvantages, and these are not always considered. For example, in interviews with 24 veterans and 12 care partners, none considered the potential risks of adverse drug events caused by poor care coordination.⁴ Given that there is limited sharing between electronic health record systems used by most VA health care providers and non-VA health care providers, the risks associated with fragmented care and, specifically, health care providers not being aware of what the other is prescribing for the same patient, are important for veterans to consider. Prior work has shown that dual use of VA health care and Part D to fill medications is associated with increased risks of hospitalization,⁵ emergency department visit,⁶ and unsafe medication use,^7–9 and in the case of diabetes, poor glycemic control.¹⁰ Prior work has also highlighted veteran and care partner difficulties related to navigating VA vs non-VA health care systems and unrealistic veteran expectations such as being able to fill any medication through either VA health care or their Part D plan, which can lead to delays in medication access.^1,7,10–12

Ultimately, care coordination is often the responsibility of the veteran and/or care partner, who express frustration navigating multiple complex health care systems, concerns around managing potentially overlapping or interacting medications from different systems, and the desire to streamline information into 1 place.¹² At present, there are no resources to aid Part D–enrolled veterans in their medication-sourcing decisions. To address this gap, the study sought to develop a decision aid (DA) to support Part D–enrolled veterans with diabetes when making choices on whether to fill their medications through VA health care only, Part D only, or both.

Methods

To guide the DA development process, we followed the International Patient DA Standards (IPDAS) process for systematic development of patient DAs (Figure 1),¹³ and we reported on user involvement via the DEVELOPTOOLS checklist (Supplementary Table 1 ^{(3.6MB, pdf)} , available in online article).¹⁴

FIGURE 1.

Decision Aid Development Process

SCOPING, DESIGN, AND PROTOTYPE

In the scoping phase, the purpose and target audience of the DA was shaped based on a review of the literature on veterans eligible for Medicare Part D (prescription drug insurance), discussion among research team members, and early feedback from the Veteran Research Engagement Panel at the Durham VA. For example, to ensure the scope of the DA was manageable, we chose to focus on a single chronic condition with high out-of-pocket spending (diabetes) and to narrow the target audience to veterans who were already enrolled in Medicare Part D.

In the design phase, we assembled a Stakeholder Advisory Board (SAB), which included a veteran, a care partner, a VA primary care physician, a VA endocrinologist, a VA pharmacist, a pharmacist from the VA Community Care Program, a representative from VA Pharmacy Benefits Management Services, and a director of a community-based program responsible for Medicare counseling. The SAB provided feedback on the interview guides for the design phase and participated in alpha testing of the DA via group calls, individual calls, and/or e-mail feedback.

Next, as previously reported,⁴ we conducted 36 semistructured phone interviews with veterans (n = 24) and care partners (n = 12) to identify the factors considered, challenges faced, and informational needs when making choices to fill medications through VA health care only, Part D only, or both. Care partners were defined as relatives, friends, or anyone who assisted the veterans with medication-sourcing decisions. In brief, the cost and availability of desired medications within each system’s formulary were key factors. Additional factors included convenience, such as streamlined care (eg, all-in-one system), travel time for clinical appointments (especially for veterans in rural locations), and trust in the provider or health care system. The interview guide and study findings were mapped back to decisional needs, based on the Ottawa Decision Support Framework.¹⁵ Decisional needs are deficits that can negatively influence a decision and include inadequate knowledge, unrealistic expectations, inadequate support/resources, decisional conflict, unclear values, complex decision characteristics, and personal and clinical needs.¹⁵

During these interviews, we also asked participants about the preferred information and format (eg, online vs paper copy) in a hypothetical guide. Participants desired information comparing costs and what medications were available through the VA vs Medicare Part D, alongside a phone number to call to clarify any questions. In terms of format, all veterans wanted a hard-copy version, and approximately half also wanted a digital version. We drafted an initial DA prototype based on interview findings and SAB feedback. Approval was obtained from the institutional review board of Durham VA Health Care System. The procedures used in this study adhere to the tenets of the Declaration of Helsinki.

PROTOTYPE

The DA prototype (Supplementary Figure 1 ^{(3.6MB, pdf)} ) was designed to address the decisional needs per the Ottawa Decision Support Framework that were identified in design interviews. The first part of the DA focused on general education (addressing inadequate knowledge) about filling medications through VA health care vs a Part D plan, such as enrollment and costs, formulary coverage, and prior authorization. Because we found in design interviews that veterans did not consider potential risks associated with dual use,⁴ part 1 of the DA also described these potential risks arising from providers not being kept up to date on medication changes other providers were making.

The design interviews confirmed that cost, formulary coverage, and convenience-related factors could influence veteran decisions when filling their medications.⁴ Part 2 of the DA listed reflection questions and then a values clarification exercise to help veterans consider which of these factors mattered most to them. Some veterans mistakenly thought that they could switch between filling through VA health care vs their Part D plan and the same medications would be available.¹² However, this is often not the case, as formulary coverage, utilization management (eg, prior authorization or criteria for use in the VA), and cost-sharing amounts differ between VA health care and each Part D plan. To address this incorrect assumption, part 3 of the DA provided a template that would allow for a side-by-side comparison of the veteran’s medication costs and availability through VA health care vs their Part D plan. Part 1 of the DA also provided key differences between VA health care and Part D plans in general, such as that VA copays are always fixed and have an annual cap of $700 whereas Part D plans often have both copays and coinsurance that may vary by pharmacy, and an annual cap of $2,000 began 2025. To help reduce decisional conflict and any additional unrealistic expectations, the end of part 3 contained tables comparing dual-use benefits and drawbacks as well as general differences when filling through VA health care vs Part D.

To address the inadequate resources and/or support reported during design interviews, we included links to online resources and relevant contact information for the VA and Part D throughout the DA at relevant points as well as aggregated in the resources section at the end of the DA. The DA also aimed to guide users to where they could get specific support for their personal and clinical needs by encouraging users to speak to their VA or non-VA health care providers, including spaces for the veteran to input the provider’s contact information. This information could also then be relayed easily to the veteran’s health care providers so that they can contact one another easily if needed.

ALPHA TESTING

We provided the initial DA to 12 stakeholders and conducted 30-minute virtual interviews with everyone to collect feedback between September and November 2024. Stakeholder participants—veterans, care partners and clinical stakeholders (VA pharmacists, nurses, and physicians who cared for Medicare Part D–enrolled veterans taking diabetes medications)—were recruited from the SAB as well as through convenience sampling.

From November 2024 to January 2025, we conducted a second round of 1-hour virtual interviews on a revised DA with 18 end users, consisting of 16 veterans and 2 care partners. Veterans and care partners were identified and recruited through the same approach described above.⁴ In brief, veterans were eligible if they met the following inclusion criteria based on self-report in a final stage of a phone screen: (1) enrolled in Medicare Part D in the past year, (2) filled at least 1 prescription (any medication) through the VA and at least 1 prescription through Medicare Part D in the past year, (3) have type 2 diabetes, (4) filled at least 1 prescription for a diabetes medication through VA or Part D in the past year, and (5) resided in 1 of 2 geographic regions (ie, Veterans Integrated Service Network 6, which covers part of North Carolina and Virginia, and 19, which covers parts of Colorado, Utah, Montana, Wyoming, Idaho, Nebraska, Oklahoma, and Texas).

Care partners were identified through their care-recipient veteran and were eligible to participate regardless of if their veteran participated in an interview. End users were emailed the DA, and we employed a read aloud–think aloud approach to assess user comprehension of the content, overall readability, usability, and acceptability during interviews. During the read aloud, we encouraged interviewees to stop at any point to share their questions, thoughts, and feedback. End-user interviewees were compensated $25 for their participation. Across the end user and stakeholder interviews, the DA was iteratively refined based on feedback. A priori alpha testing ended when we were no longer receiving feedback that led to major changes to the DA. Toward the end of the alpha testing phase, we were making fewer changes and were ready to begin beta testing of a near-final version of the DA. After the alpha testing phase, we switched from software more suitable for prototyping (Figma) to publishing (Adobe InDesign) to allow for a hard-copy version to be mailed to participants for beta testing.

BETA TESTING

From February to June 2025, we conducted 8 virtual interviews with stakeholders who had not previously served on the SAB. From March to June 2025, we conducted virtual interviews with 17 veterans and 3 care partners using the same approach as in alpha testing.

Our goal in beta testing interviews was to gather feedback on feasibility and to inform future implementation of the DA. In these interviews, we probed to learn how end users (veterans and care partners) would use the DA in practice (eg, what information they find helpful and which interactive elements they engage with), and how stakeholders (VA health care providers) might use information from the DA should a veteran they care for bring the DA to an appointment. End-user interviewees received the DA via mail and were asked to mail the DA back to allow for assessment of how the DAs were filled out.

In conjunction with these interviews, we asked participants to complete a survey before viewing the DA and a separate survey after working through the DA (Supplementary Figure 2 ^{(3.6MB, pdf)} ). Questions in both the pre- and post-DA surveys included (1) a 4-question measure of decisional conflict, called Sure of myself; Understand information; Risk-benefit ratio; Encouragement (SURE), where higher scores represent lower decisional conflict¹⁶; (2) 5 true/false questions assessing knowledge about the differences in filling medications in the VA compared with filling through Part D; and (3) 3 questions on satisfaction with medication filling, access, and costs. The post-DA survey also included (1) an adapted version of the system usability scale (SUS), where higher values are associated with greater usability¹⁷ (a tool is typically considered usable based on a SUS score of ≤68)¹⁸; (2) an adapted acceptability questionnaire, which included some free-text fields and asked about balance and amount of information, length, and whether the tool would be useful¹⁹; and (3) questions about how often the end user would use the DA and what next steps they would take after going through the DA. Responses were summarized using descriptive statistics. Given the small sample size, we did not assess the significance of the change in scores. The goal of beta testing was to provide a preliminary estimate of effect size, and it was not powered to detect a significant change. All study activities were approved by the institutional review board at the Durham VA, and verbal informed consent was obtained before the surveys and interviews.

Results

ALPHA TESTING: COMPREHENSIBILITY, USABILITY, AND ACCEPTABILITY

Among the 18 end-user alpha testing interviews (16 veterans and 2 care partners), 7 currently filled their diabetes medications through VA health care only, 4 Part D only, and the remaining 7 reported currently filling through both VA health care and their Part D plan. The mean age of veterans was 72 years. Among the 12 stakeholder alpha testing interviews, 7 were members of our SAB.

Based on feedback from the interviews regarding comprehensibility, usability, and acceptability, we made numerous changes to the DA (Table 1). Some of the more notable updates made to the DA to increase usability included replacing the points allocation exercise with a ranking exercise and simplification of the medication cost and formulary comparison chart. To increase acceptability, we also expanded the medication list to include a larger scope (eg, diabetes supplies in addition to medications, expanding beyond antihyperglycemic medications), added a “Current as of:” date field, emphasized the importance of keeping the medication list up to date, and added columns to provide the prescriber contact information and indicate whether the medication was filled through VA health care.

TABLE 1.

Example Feedback and Resulting Modifications from Alpha Testing

Category	Feedback	Resulting modifications
Comprehensibility	•Recommended we clarify the specific “adverse events” – VA endocrinologist	•Changed “chance of adverse events” to “risks to blood sugar levels”
Usability	•Most reported that they would not take the time to fill out the dual-use columns in the comparison chart – veterans •Stated they did not find the entirety of the comparison chart helpful and did not think the dual-use columns added value – care partner	•Removed dual-use columns from comparison chart. Instead, kept the focus of the comparison chart on differences in cost and formulary coverage in VA vs Medicare Part D
	•Found the points allocation exercise to be the most difficult part of the decision aid. Recommended that the exercise be replaced with a simpler exercise – veteran •Found the points allocation exercise to be a bit confusing. Recommended a ranking exercise instead – veteran •Could not decide how they would allocate their points. Would prefer a ranking exercise instead – veteran	•Changed points allocation exercise to ranking exercise
	•Reported writing the most recent date on his medication list – veteran •Stressed the importance of keeping the medication list current – VA endocrinologist	•Added a space to write the current date on the medication list pages
	•Advised that we add a column to the medication list to indicate where med is getting filled – VA primary care physician	•Replaced how often med is taken column with where is med filled column
	•Reported always keeping their medication list on their person. Recommended other veterans do the same – veteran •Stated her husband keeps his current medication list in his billfold – care partner	•Added recommendation that veterans carry their medication list in their wallet
Acceptability	•Shared concern that the medication list does not include enough space to write all of their patients’ medications – VA endocrinologist •Recommended including diabetes supplies on medications list – veteran •Suggested veterans add their over-the-counter medication to their medication list – care partner •Recommended medical equipment like sensors, insulin pumps, and other supplies be included in the list – VA endocrinologist	•Removed space to write notes in favor of extending the medication list •Added note that medical supplies and over-the-counter medications can be listed on the medication list
	•Recommended we add a statement to address the time it should take to complete the decision aid – care partner •Expressed worry about potential users’ bandwidth to go use the decision aid in its entirety. Appreciated an acknowledgment of the time it takes to complete the decision aid – VA endocrinologist	•Added statement to acknowledge the time it takes to complete decision aid
	•Stated decision aid should not use the term “controlled” for blood sugar. Instead, the phrases “blood sugar within target ranges” or “meets health care goals” should be used – VA nurse	•Changed language from “blood sugar control” to “within target ranges”
	•Advised we add a note about drug-drug interactions – VA pharmacist	•Added a note about drug-drug interactions
	•Advised that we remove mention of UHC, CVS Health, and Humana to avoid any perception of bias or marketing – Medicare counselor representative	•Removed list of Medicare Part D insurance companies to avoid any potential bias from tool

UHC = United Healthcare; VA = Veterans Affairs.

BETA TESTING: USABILITY AND ACCEPTABILITY

Among the 20 end-user beta testing interviews, 17 were veterans and 3 were care partners. Six represented veterans who filled their diabetes medications through VA health care only, 4 through Part D only, and the remaining 10 reported filling through both. The mean age of veterans was 69 years. The 8 stakeholder interviews consisted mostly of VA pharmacists as well as a nurse, social worker, and endocrinologist.

Of the 20 interviewees, 16 (80%) responded to a post-interview survey. In terms of usability, the mean and median SUS scores were 77.5 (SD = 14.4) and 78.8 (25th-75th percentile = 69.4-83.8), respectively. For acceptability, when asked to rate how information was presented in each section of the DA, most responses were “good” or “excellent” (Supplementary Figure 3 ^{(3.6MB, pdf)} ). In particular, respondents highly rated how the information was presented in parts 3 and 4 (comparison chart, weigh your options & tables, medication list, and resources; Supplementary Figure 3 ^{(3.6MB, pdf)} ). All 16 respondents agreed that there was sufficient information in the DA to help with medication-sourcing decisions, with 15 respondents (94%) answering that they would find the overall DA useful. Fifteen respondents (94%) found that DA length to be “just right,” 13 respondents (81%) found the amount of information to be “just right,” and 14 respondents (88%) found the presentation to be “balanced” (Supplementary Figures 4-7 ^{(3.6MB, pdf)} ).

BETA TESTING: FEASIBILITY

Survey respondents reported taking a mean of 36.3 minutes (SD = 26.5) and a median of 30 minutes (25th-75th percentile = 15-46.3 minutes) to read and fill out the DA. Four of 16 respondents (25%) reported that they did not complete the medication list and 2 of 16 (12.5%) reported that they did not complete the comparison chart or ranking question (categories not mutually exclusive; Supplementary Figure 8 ^{(3.6MB, pdf)} ).

After completing the DA, 7 of 16 respondents (44%) reported they would take further action (Supplementary Figure 9 ^{(3.6MB, pdf)} ). Four respondents (25%) each reported that they would check out a new resource (eg, website provided in the DA), update a medication list they already had, or add information from the DA’s medication list to one they already had. These categories were not mutually exclusive, meaning that respondents could select more than 1 action. Three respondents (19%) each reported that they would use the medication list template from the DA or change what they would do at their health care provider’s office. Only 2 (13%) respondents indicated they would change how they filled their nondiabetes medications. No respondents reported that they would change how they filled their diabetes medications, change their Part D plan, contact a Medicare counselor, or call their local VA pharmacy. During the end-user interviews, veterans and care partners recommended that the DA be sent to either all Medicare-eligible diabetic veterans and/or all veterans new to either the VA or Medicare Part D. Regarding the desired DA delivery method, mail and then e-mail were ranked highest and second highest, respectively (Supplementary Figure 10 ^{(3.6MB, pdf)} ). Mail had a mean rank of 2.3 (SD = 1.5) and a median rank of 2 (25th-75th percentile = 1-3), whereas e-mail had a mean rank of 2.6 (SD = 1.2) and a median rank of 2 (25th-75th percentile = 2-4), where lower numbers represent higher rank (Supplementary Table 2 ^{(3.6MB, pdf)} ).

From stakeholder interviews, some VA pharmacists found the DA helpful for themselves because they were not familiar with Medicare Part D. They would struggle finding Medicare Part D costs and formulary coverage, and so would desire further training in these areas. Some clinicians noted that they would not have time to help the veteran go through and fill out the DA during a single appointment, so they preferred that the DA already be filled out. They also noted that a veteran coming into their appointment with the medication list filled out would be helpful, and they especially liked having the information about where the medication was filled and the name and contact information for each medication’s prescriber. Another suggestion that arose was the possibility to eventually have a program that would train veterans to help each other to fill and use the DA.

BETA TESTING: DECISIONAL CONFLICT, KNOWLEDGE, AND SATISFACTION

Before receiving the DA, participants were sent an online survey with questions related to decisional conflict, knowledge, and satisfaction. These same questions appeared in the post-DA survey. Fourteen (70%) beta-testing participants responded to the pre-DA survey, and 16 (80%) responded to the post-DA survey. Decisional conflict was reduced, as indicated by an increase in the mean SURE score (pre-DA = 3.1, SD = 1.4 to all 16 respondents reporting a maximum SURE score of 4.0; Figure 2). Knowledge about VA and Medicare coverage of diabetes medications also improved—the proportion of those who answered all 5 questions correctly increased from 57% to 81% (Figure 3). Satisfaction with accessing medications and cost of medications did not change substantially, but the proportion who reported being “very satisfied” with how they were currently filling their diabetes medicines (VA only, Part D only, or both) improved from 64% pre-DA to 81% post-DA (Figure 4). Results were similar when restricted to the 14 respondents who answered both the pre-DA and post-DA surveys (Supplementary Figures 11-13 ^{(3.6MB, pdf)} ). The final DA is provided as Supplementary Figure 14 ^{(3.6MB, pdf)} .

FIGURE 2.

Decision Conflict Improved (Was Reduced) After DA

The SURE score ranges from 0 to 4, with higher scores reflecting lower decisional conflict.

DA = decision aid; SURE = Sure of myself; Understand information; Risk-benefit ratio; Encouragement.

FIGURE 3.

Knowledge Improved After DA

DA = decision aid.

FIGURE 4.

Satisfaction With How Medications Were Being Filled Improved After DA

DA = decision aid; VA = Veterans Affairs.

Discussion

Approximately one-third of veterans enroll in VA health care and more than three-quarters of veterans enrolled in VA health care have dual insurance, with Medicare being one of the most common sources of additional coverage.²⁰ Medication cost and coverage may be very different through VA health care vs a Medicare Part D plan. Understanding these options is necessary for veterans to make well-informed choices, given that insurance can be quite complicated and veterans often do not have sufficient resources to help them with their choices.⁴ To begin to address decisional support needs related to dual insurance, we developed a DA for a specific use case: to help the 70,000 Medicare Part D–enrolled veterans with diabetes with their medication-sourcing decision (ie, VA health care only, Part D only, or both).

Although there have been many studies published on the development of patient DAs that focus on treatment choice,^21–29 there are relatively few publications on the development process for insurance-related DAs or educational content.^30,31 To make well-informed choices on filling through VA only, Part D only, or both, veterans must become familiar with their VA prescription benefit, Medicare Part D insurance plan, and how to determine differences in costs and formulary coverage. The DA helped veterans acquire this knowledge about comparative benefits, which reinforced the reality that veterans need to coordinate their care to reduce any risks related to filling medications in both the VA and Medicare, as different health care prescribers are often involved.

Based on feedback gathered from interviews with veterans, a few care partners, and clinical stakeholders (pharmacists, other clinicians, and payer and Medicare counselor representatives), the DA was comprehensible, usable, and acceptable. The average SUS score was 77.5, which is above the threshold of 68 that suggests usability,¹⁸ and most end users also found the length (94%), amount of information (81%), and balance (81%) of information in the DA to be acceptable. As insurance-related information is particularly complicated (eg, jargon or fine print), we were pleased to find that end users generally found the DA to be usable and acceptable.

The DA was also found to reduce decisional conflict, increase knowledge, and increase satisfaction with how medications were being filled. Prior insurance-related DAs have shown mixed results on some of these outcomes.^32–36 In 1 study, a DA created to help individuals choose a plan in the federal health insurance marketplace increased knowledge, decision self-efficacy, confidence in their choice, and health insurance literacy.³² A second study found that an online decision support tool helping Medicare-eligible individuals choose a Part D plan did not impact knowledge, satisfaction, decisional conflict, or plan switching.³³ Only when combined with expert recommendation did the tool lead to increased plan switching and increased satisfaction.³³ Several studies comparing different approaches to presenting insurance information (eg, visual vs narrative, simplifying content, or adding quick links) have also had mixed results.^34–36 In the context of these mixed results, it is encouraging to see that our DA reduced decisional conflict, increased knowledge, and increased 1 of the satisfaction measures.

More than 40% of DA end users reported that they would take additional steps after completing the DA. Several were motivated enough to check out a resource that they learned about from the DA, and in interviews some mentioned that they wanted to look at the VA formulary website. Others planned to make updates or additions to their medication list or use the medication list template. None reported that they would change how they filled their diabetes medications, but a couple reported that they would change how they filled their nondiabetes medications. It is not clear whether additional steps were needed for each individual’s situation, such as whether how they were already filling their medications matched the decision they arrived at by the end of the DA. Additionally, if the DA reduced decisional conflict and increased satisfaction with how veterans were already filling their medications, this could still be beneficial to veterans. However, this deserves future exploration because if some veterans wanted to act but were not sure how to go about doing so, then we would want to minimize such a barrier. During interviews, veterans suggested having a phone number that they could call to have their questions answered. In the future, we plan to develop additional decisional support via a pharmacist who can help answer questions for the veteran. Subsequently, we will test in a randomized trial the impact of the DA alongside this decisional support.

LIMITATIONS

A few study limitations should be noted. First, only a few care partners were interviewed in alpha (n = 2) and beta (n = 3) testing, so end-user feedback from these sessions was primarily from veterans. Veterans were the primary target audience of the DA, and although we thought it would be informative to include some feedback from care partners, the study did not have the budget to support further exploration from care partners. A future study is warranted if care partners become the primary user. Second, our research team determined that, because of feasibility, the scope would be limited to after Part D enrollment (ie, choice about medication-sourcing decision as opposed to which Medicare Part D plan to choose). To help veterans optimize choice in their Medicare Part D plan, which can also help veterans save in out-of-pocket costs and allow for access to specific medications, future study is needed. Third, the pre-post design in assessing knowledge is limited; when participants see the same questions before and after seeing the DA, they may learn the content and post-DA knowledge responses may be biased upwards. A future trial will be necessary to test its effectiveness on these measures and to understand whether the DA changes patient knowledge, because this preliminary evaluation was not powered to detect a significant change in the knowledge measures. Fourth, alpha and beta testers were compensated $25 if eligible for compensation (eg, not a VA employee), which could potentially bias results to be more favorable toward the VA. However, $25 is a relatively small amount, and it is considered standard research practice to provide some kind of respondent compensation to aid in recruitment. Fifth, the DA is currently focused on veterans with diabetes and for them to report their diabetes medications. However, veterans could also find such a DA helpful across all medications (not just limited to diabetes medications). In the future, we hope to expand the DA to address all medications, and further study will be required.

Despite these limitations, the study had several strengths, including following the IPDAS systematic process for DA development; 3 rounds of interviews across design, alpha testing, and beta testing phases; a preliminary evaluation of pre- vs post-DA survey responses; and inclusion of end users and stakeholders in interviews and on the SAB.

Conclusions

In conclusion, the DA developed in this study was found to be usable and acceptable, and it resulted in decreased decisional conflict, increased knowledge, and increased satisfaction in medication filling.

Disclosures

Dr Hung reports research grant funding from Abbott, AstraZeneca, the VA, and the National Institutes of Health (NIH); consulting from Genentech/Roche; honoraria and conference support from the Academy of Managed Care Pharmacy; and honoraria from American Medical Association. Dr Reed reports research funding and external relationships at https://scholars.duke.edu/person/shelby.reed/research. Dr Thorpe reports research grant funding from the VA and NIH and royalties from SAGE Publications, Inc.

Ms Upchurch serves on the Medicare Payment Advisory Commission; the views presented here reflect the author’s opinions and not that of the Commission.

This work was supported by IK2 HX003359 and RCS 10-391 from the US Department of Veterans Affairs Health Systems Research Service as well as the Center of Innovation to Accelerate Discovery and Practice Transformation (CIN 13-410) at the Durham VA Health Care System.

Data Availability

Deidentified data will be made available upon reasonable request and as allowable by current VA policies.