The DISCO App: Evaluating an Electronic Patient Intervention to Reduce The Financial Burden of Cancer Through Improved Cost Communication In A Non-Randomized Pilot Trial

Background: Financial toxicity affects 30-50% of people with cancer in the US. Although experts recommend patients and oncologists discuss treatment cost, cost discussions occur infrequently. Using a non-randomized pre-test post-test design, we pilot-tested the feasibility, acceptability and preliminary effectiveness of the DIScussions of COst (DISCO) App, a patient communication intervention designed to improve the frequency of cost discussions and related outcomes. Methods: While waiting to see their oncologist, patients (n=32) used the DISCO App on a tablet. Clinic visits were video recorded and patients completed pre- and post-intervention measures of self-ecacy for managing treatment costs, self-ecacy for interacting with oncologists, cost-related distress, and perceptions of the DISCO App. Coders observed the recordings to determine the presence of cost discussions, initiators, and topics. Results: Most patients (84%) reported needing <15 minutes to use the DISCO App. Patients reported the DISCO App made it easier to ask cost-related questions. Findings showed increased self-ecacy for managing treatment costs (p=.02) and for interacting with oncologists (p=.001). All clinic visits included a cost discussion. Conclusion: The DISCO App is feasible, acceptable, and effective in prompting cost discussions and improving outcomes. Prompting patients to discuss costs before meeting with an oncologist may improve treatment cost discussions and related outcomes.

of-pocket costs [15]. Another study of 164 patients with solid tumors found that 45% were non-adherent to treatment due to cost [28]. A study of 1556 cancer survivors found that those who reported nancial problems were more likely to delay (18.3% vs. 7.4%) or forgo treatment (13.8% vs. 5.0%) compared to respondents without nancial problems [29].
Addressing nancial toxicity requires policy changes at the national, state, and hospital levels. In the meantime, however, increasing the frequency and quality of patient-oncologist treatment cost discussions early in diagnosis and treatment may help alleviate nancial toxicity [10,14,[30][31][32][33] by improving patients' knowledge, self-e cacy, and ability to manage potential costs, and by connecting patients with vital economic support [34]. Most patients (80%) and oncologists (80%) want to discuss treatment costs, [35,36] and professional organizations such as the American Society of Clinical Oncology (ASCO) encourage oncologists to discuss treatment cost with patients [14]. However, research has found that these discussions are infrequent [36,37]. In our previous observational study of videorecorded treatment discussions between patients with cancer and their oncologists (n = 103), we found that cost discussions occurred in only 45% of treatment discussions [38]. When cost was discussed, it was mostly patient-initiated (63%) and focused more on indirect costs (e.g., time off work) than on direct costs (e.g., copayments; [39]. Without a cost discussion early in treatment decision making, patients are unlikely to be referred for guidance or assistance in a timely manner, thereby missing out on early nancial and psychological support, which are critical steps in reducing longer-term nancial toxicity [15,34,40] and improving treatment adherence [10,28,41]. Research shows that, in the short term, patientoncologist treatment cost discussions can increase referrals for support (e.g., social work; [34] and reduce cost-related distress [10]. Longer-term effects include improved nancial toxicity [10] and treatment adherence [41]. Another bene t of holding cost discussions early in treatment planning is that they can improve patient self-e cacy, or the expectation that one can successfully perform a behavior [42], for managing cost [43,44]. Similarly, researchers have demonstrated the positive in uence of improving another aspect of selfe cacy-activating patients to manage their own health, and more speci cally, actively participating in clinical interactions by asking questions, stating concerns, and making assertions [45][46][47][48]. Research on clinical communication in many medical settings shows patient active participation plays an important role in short-, intermediate-, and long-term outcomes [49,50]. Patient active participation in uences the amount of information physicians provide [46,51,52], the information exchange process (Barton et al., 2020), the treatment physicians recommend [48], topics patients and physicians discuss [53,54], patient healthcare decisions [55], and patient psychosocial and physical health outcomes [56,57]. Increasing patient active participation during clinical interactions has been shown to change the content and quality of patient-physician interactions and outcomes, and thus has the potential to improve the frequency and quality of patient-oncologist treatment cost discussions. It is also possible that self-e cacy may be the primary mechanism through which patient-oncologist treatment cost discussions mitigate the burden of nancial toxicity.
Question prompt lists (QPLs) are simple communication tools that have been shown to improve patient active participation in cancer treatment discussions and prompt discussion about speci c topics during clinical interactions. QPLs are comprised of a list of questions provided to patients to encourage them to prepare for visits by considering questions they would like to ask their healthcare provider [58][59][60]. QPLs have been shown to improve communication quality (e.g., patient active participation in interactions, [47]; patient-oncologist information exchange, [61]; topics discussed [53,54]; patient psychological and cognitive outcomes (e.g., satisfaction, anxiety; information recall; [58] and patient role in treatment decisions and trust in their oncologist [58,60,62].
In the current research, we build upon current QPLs in two ways. First, while most current QPLs have few if any questions regarding treatment costs, we developed a QPL that speci cally addresses this topic.
Second, while most current QPLs are paper-based and static, we developed a way to tailor the QPL content to patients' speci c needs and clinical setting through the use of an electronic QPL in the form of an application or "app" provided to patients in the clinic prior to a patient's scheduled visit with an oncologist to discuss treatment.
Building on our experience with testing QPLs in oncology outpatient clinics [47,[63][64][65] we designed and built a novel communication tool, the DIScussion of COst Application (DISCO App; Fig. 1; [66]. We designed the DISCO App to be app-based because using an electronic format allows the tool to be tailorable to the individual [58] enhances eventual scalability to other populations, and has the potential to be integrated into EMRs and patient portals. In response to concerns that physicians may be unwilling or unable to respond appropriately to patient questions about treatment cost [67,68], we also developed a treatment cost discussion "tip sheet" for oncologists. The tip sheet emphasizes oncologists' role in cost discussions (as recommended by ASCO) and provides ways to overcome identi ed barriers to cost discussions [36,37,67,68].

DISCO App intervention
The DISCO App ( Fig. 1; [66] is displayed on an iPad provided to patients just prior to their second interaction with their oncologist, in which they discuss and nalize treatment plans. The DISCO App opens and the QPL is introduced with text that explains that the DISCO App includes a short survey, which will lead to some cost-related questions the patient can consider asking the oncologist. This section asks patients to enter their demographic information and their nancial circumstances. Speci cally, patients respond to 17 questions, such as "How much do you know about your insurance coverage?"; "Are you currently employed?"; "Is there anyone who helps you when you're sick or need help of any kind?". Based on patient responses, an individually-tailored QPL with up to 18 cost-related questions in 7 categories is generated (Table 1). For example, patients who indicate they are employed will be prompted to ask: "Can I schedule my treatment around my job?"; patients who indicate transportation concerns will be prompted to ask: "Are services available if I can't nd someone to drive me?" Patients who indicate they are unfamiliar with their insurance coverage will be prompted to ask: "Is there someone I can talk to about my insurance and treatment cost questions?" All patients, regardless of their responses, are provided with four general questions about their diagnosis and have the option of adding in any of their own questions. Once they have completed the questions and received their individualized QPL, they can take the iPad or a printed question list into the meeting with the oncologist.

Methods
Ethical approval for this study was granted by the Wayne State University (WSU)/Karmanos Cancer Institute (KCI) institutional review board (IRB # 115117B3E(M)). The extension of CONSORT 2010 checklist speci cally for pilot trials was used for reporting the study. [69] Research aims A large-scale and longitudinal randomized controlled trial is needed to determine if the DISCO App will lead to the anticipated short-and longer-term patient outcomes, as compared to usual care. However, because this is a newly-developed technology-based intervention, we must rst conduct preliminary testing. The purpose of this study was to pilot-test the DISCO App in an oncology clinic setting to determine its feasibility, acceptability, and preliminary effectiveness on short-term outcomes including: patient-oncologist treatment cost discussions during treatment discussions, patients' self-e cacy for managing treatment costs, patients' self-e cacy for interacting with physicians, and patients' treatment cost-related distress.

Design
For this pilot trial, we employed a non-randomized pre-test post-test design to test the feasibility, acceptability and preliminary effectiveness of the DISCO App. This was a one-arm trial in which all recruited patients were assigned to the intervention arm.

Participants and Setting
Data were collected in two outpatient community clinics a liated with WSU/KCI located in Southeast, Michigan from August 2018 to December 2019.
Patients were eligible to participate if they were recently diagnosed with breast, lung, prostate, or colorectal cancer and were scheduled to see a participating radiation oncologist for an initial treatment discussion. Oncologists were eligible if they routinely treated patients for breast, lung, prostate, or colorectal cancer. Patients were identi ed by a clinic member of the research team who reviewed participating oncologists' schedules and contacted eligible patients prior to their appointment to determine interest in participating. Interested patients were consented on the day of their appointment and prior to meeting with their oncologist.

Procedure and Measures
Upon recruitment into the study, oncologists completed a one-time baseline questionnaire where they provided information on their sex and race/ethnicity, and then received the oncologist treatment cost discussion tip sheet. After patients were recruited, they completed a baseline questionnaire that assessed sociodemographic characteristics and other personal attributes.
Patients arrived up to one hour before their scheduled appointment to complete self-report measures on an iPad through an electronic survey service (Qualtrics). After completing pre-interaction measures, a member of the research staff showed patients the DISCO App on an iPad and provided a brief explanation on how to use it. The iPad was connected to a printer on-site, allowing patients to print their individually-tailored QPL created from the DISCO App for note-taking.
Each examination room was equipped with unobtrusive digital audio and video devices that recorded the examination room during the clinical interaction. This recording system has been used by the study team for more than 15 years, [28,41] and research has strongly suggested that video recording has little impact on participants' verbal or nonverbal behaviors [58] and provides enhanced validity compared with audio recording alone [59].
After the patient-oncologist treatment discussion, patients completed post-discussion self-report measures. All patients and oncologists provided consent as participants, which included speci c permission to be video recorded.

Feasibility
To assess the feasibility of the DISCO App, patients reported how long it took them to use the DISCO App and research staff collected a copy of the printed question list to determine how many questions patients selected from the DISCO App to be printed for a discussion with their oncologist. We also tracked if any patient needed assistance with using the DISCO App and if any participant refused to be video recorded and/or left the study early.

Acceptability
To assess the acceptability of the DISCO App, after the clinical interactions, patients completed measures of their perceptions of the DISCO App ("the questions in the DISCO App were easy to understand"; "the DISCO App made it easier to ask my doctor cost-related questions"; "some of the questions in the DISCO App were useful to me as I was talking with my doctor"; "some of the questions in the DISCO App made me uncomfortable") on a Likert-type scale (1 = strongly disagree to 5 = strongly agree).

Preliminary Effectiveness Self-report Measures
To assess the preliminary effectiveness of the DISCO App, patients completed pre-and post-interaction measures of nancial toxicity-and cost discussion-related outcomes including self-e cacy for managing treatment costs (e.g., how con dent are you in your ability to nd out how to pay for direct costs that arise with treatment?; 5 = very con dent, 1 = not at all con dent [70], self-e cacy for interacting with physicians (e.g., how con dent are you in your ability to know what questions to ask a doctor?; 5 = very con dent, 1 = not at all con dent; [71], and cost-related distress (e.g., I am concerned about my ability to afford to pay for my cancer treatment; 5 = strongly agree, 1 = strongly disagree).

Oncologist and Patient Interaction Coding
To further assess the preliminary effectiveness of the DISCO App, we analyzed video-recorded patientoncologist treatment cost discussions using methods from our previous research [38]. Patient-oncologist cost discussions were broadly de ned as verbal discussion of topics related to a monetary expense for the patient for cancer treatment. This includes direct costs such as appointment copayments and indirect costs such as loss of income due to time off from work.
Two trained coders observed all video-recorded interactions to identify the presence of cost discussions using a validated coding system and, when these occurred, the initiator (patient or oncologist) and the cost topics raised, from a list identi ed from previous research, including insurance/copayment, out-ofpocket costs, time off from work, transportation/parking, lodging, social work/ nancial navigation, and scheduling treatment around patient's schedule [38].
Interrater reliability was assessed using 20% of the sample of video-recorded interactions. First, the coders identi ed the presence of cost discussions and reliability was determined using percent agreement, which was 87.5%. Second, the coders labeled each cost discussion's initiator and topic. Reliability for this phase was determined by Cohen's kappa ( K = 1.0). Because the high K value suggested high intercoder reliability, the remaining video-recorded interactions were coded by one coder each.

Data Analysis
Data included patient and oncologist self-report sociodemographics, patient perceptions of the DISCO App, patient self-reported outcome variables, and coder ratings of the video-recorded clinical interactions.
We used descriptive statistics to describe the patient and oncologist participant sociodemographic characteristics, feasibility, acceptability and preliminary effectiveness. To describe feasibility, we determined how long patients reported using the DISCO App and how many questions patients selected from list the DISCO App created for them. We also determined how many patient and oncologist participants agreed to have their visits video recorded and if any patients dropped out of the study. To assess acceptability, we determined patients' perceptions of the DISCO App. To assess preliminary effectiveness, we determined presence, initiator, and topics of any cost discussions that occurred during the video-recorded interactions. We used two-tailed paired samples t tests to determine any pre-to postintervention changes in patient self-report outcomes of self-e cacy for managing treatment costs, selfe cacy for interacting with oncologists, and cost-related distress. For all analyses α was set at p < .05 (two-tailed).

Results
Thirty-two recently-diagnosed patients of three participating oncologists agreed to participate. Most (n = 30, 94%) interactions were video recorded (in two cases technical di culties prevented recording). The sociodemographic characteristics of patients and oncologists are reported in Table 2. Most patients had breast cancer (84%). nished using the DISCO App without assistance during the time they were in the exam room and waiting to see their oncologist. On average, patients selected 6.5 questions from the individualized question prompt list to print, with a range of 1-18 selected questions. We emphasize that the list of questions was individually-tailored so not every patient was presented with all 18 possible questions.

Acceptability
Patients reported that the questions in the DISCO App were easy to understand (M = 4.5; SD = .8); the DISCO App made it easier to ask their doctor cost-related questions (M = 3.8; SD = .8); and that some of the questions in the DISCO App were useful to them as they were talking with the doctor (M = 3.8; SD = .7). They also reported that the questions in the DISCO App did not make them feel uncomfortable (M = 4.1; SD = 1.1).

Preliminary Effectiveness Observed Cost Discussions
Cost discussions occurred in all 30 of the video-recorded interactions. The number of cost discussions per interaction ranged from 1-6, for a total of 97 individual cost discussions across all interactions.
Patients and oncologists were equally likely to initiate the rst cost discussion during a clinical interaction, and the topic most-frequently discussed rst was insurance. Oncologists were more likely to initiate any subsequent cost discussions. The most frequent patient-initiated topics were insurance followed by time off from work. The most frequent oncologist-initiated topics were social work/ nancial navigation followed by insurance (Table 3).

Discussion
To our knowledge, this study is among the rst to investigate whether a patient-focused intervention can successfully prompt patient-oncologist treatment discussions, and, in turn, improve nancial toxicityrelated outcomes for patients. Findings from this pilot test demonstrated that the DISCO App is both feasible to implement in a clinical oncology setting and acceptable to patients. Moreover, it appears to be effective in improving short-term outcomes, including prompting treatment cost discussions and increasing patient self-e cacy for interacting with physicians and managing costs. These promising ndings suggest that through prompting cost discussions and improving short-term outcomes, the DISCO App may lead to improved longer-term outcomes, including the burden of nancial toxicity and its consequences.
Treatment cost discussions have been identi ed as one way to help mitigate the deleterious experience of nancial toxicity due to cancer treatment by increasing patients' self-e cacy in managing their treatment costs [10,14,[30][31][32][33][34]. In a previous observational study we conducted also using video-recorded treatment discussions, we observed that only 45% of interactions had a treatment cost discussion [38].
Here, we observed that 100% of video-recorded treatment visits had at least one treatment cost discussion, and 70% of treatment discussions included at least three different treatment cost topics. We also observed improvements in other important patient outcomes, including self-e cacy for managing treatment costs and self-e cacy for interacting with physicians. We anticipate that patient self-e cacy may be the mechanism through which the DISCO App will improve outcomes. Speci cally, we think patient self-e cacy will in uence both the DISCO App's direct short-term intent -to prompt patientoncologist treatment cost discussions -and its indirect longer-term intent, to mitigate the experience of nancial toxicity and improve treatment adherence. Of course, our pilot study design does not allow us to conclude with certainty that the DISCO App or cost discussions it prompted led to observed improvements in patient self-e cacy for managing their treatment cost or interacting with their physicians. However, these promising ndings warrant further investigation into the more nuanced relationships between the short-and longer-term effects of the DISCO App and treatment cost discussions for patients.

Practice Implications
ASCO's Value of Cancer Treatment Options Framework [14] and its Patient-Clinician Communication Guideline [72] both encourage oncologists to discuss treatment costs with their patients. Despite these guidelines, however, research has shown that cost discussions infrequently occur and that physician engagement around cost concerns is an unmet patient need [73]. Many of the cost discussions we observed in the video-recorded data addressed cost-related topics beyond ASCO's more narrow de nition of treatment costs. ASCO's tools focus exclusively on direct out-of-pocket costs for patients such as copayments or insurance costs [72]. In this study, both patients and oncologists initiated discussions about indirect costs for patients (e.g., taking time off from work, transportation to treatment appointments), which can often be just as detrimental to patients [74][75][76][77]. This may be evidence that oncologists are well positioned to address the indirect cost of care, even if the precise direct cost of treatments remain elusive, and that ASCO may need to expand its de nition of treatment costs.
With regard to the concern that physicians may not be willing or able to respond to patients' cost-related questions, we observed that, once an initial cost discussion occurred in an interaction, oncologists were more likely to initiate subsequent cost discussions later in the interaction. Perhaps this demonstrates that oncologists are prepared to help patients with many of their cost-related concerns, and patients' prompts

Conclusion
In conclusion, we found the DISCO App to be feasible and acceptable in a clinic setting and effective at prompting treatment cost discussions between patients and oncologists and improving other patient outcomes related to nancial toxicity. We note here that these ndings provided preliminary evidence supporting a larger test of the DISCO App. The effectiveness of an enhanced version of the DISCO App on short-and longer-term patient outcomes, including patient-oncologist treatment cost discussions, is currently being tested in a randomized controlled trial with a diverse patient population (RSG-20-026-01-CPHPS, Hamel, PI).

Declarations
Ethics approval and consent to participate Ethical approval, including patient and physician consent to participate, for this study was granted by the Wayne State University (WSU)/Karmanos Cancer Institute (KCI) institutional review board (IRB # to discuss cost encourage oncologists to further the discussion. Surprisingly, we also observed that oncologists sometimes asked patients for their list of questions and responded to their questions, rather than waiting for patients to verbalize their questions. This was not part of our planned analyses, but future research should identify unintended effects of the cost-related question prompt list, whether positive or negative.
Since research has shown that, despite the guidelines from ASCO, treatment cost discussions do not occur regularly, perhaps these guidelines coupled with an individualized prompt from an intervention like the DISCO App just prior to meeting with an oncologist may help these discussions occur more frequently for patients. Lack of knowledge of the cost of treatment has been identi ed as a barrier to providers discussing treatment cost with patients [68]. Our ndings, albeit preliminary, could be demonstrating that oncologists may have a better understanding of the costs of treatment than previously thought. Of course, additional work is needed to better understand how the presence of the question list produced by the DISCO App in uences both the content and quality of cost discussions and patients' satisfaction with those discussions.

Limitations
Findings must be considered within the limitations of the study. This was a descriptive, one-arm pilot test using pre-post patient-reported measures and video-recordings to assess the DISCO App's feasibility, acceptability, and preliminary effectiveness on short-term outcomes. Also, most patient participants were women with breast cancer, which limits the generalizability of the ndings [77, 78]. Further study with a randomized controlled trial design is needed to assess these and longer-term outcomes, and this research is needed in a diverse patient population with regard to racial/ethnic background, gender, and cancer type. 115117B3E(M)).

Consent for publication
Not applicable Availability of data and materials The data that support the ndings of this study are available from the corresponding author, [LMH], upon reasonable request.