Cost Conversations About Anticoagulation Between Patients With Atrial Fibrillation and Their Clinicians

Key Points Question What factors contribute to cost conversations about anticoagulation treatment between patients with atrial fibrillation and their clinicians, and what outcomes are associated with these conversations? Findings In this cohort study of 830 audiovisual recordings of encounters and participant surveys from a randomized trial comparing atrial fibrillation care with and without a shared decision-making (SDM) tool, cost conversations were associated with the use of an SDM tool, with middle-income patients, and with consultations conducted by female primary care staff clinicians. Cost conversations were associated with patients’ decision-making processes but not final treatment choice. Meaning These findings suggest that SDM tools may inform efforts to promote cost conversations in practice, an important consideration when increasing costs of care are being passed on to patients.


Introduction
Atrial fibrillation (AF) affects over 5 million people in the US. 1,2AF may reduce the quality and duration of life, particularly by increasing the risk of devastating thromboembolic strokes, and it accounts for over $25 billion a year in health care costs, mostly from preventable thromboembolic strokes. 3Use of oral anticoagulants, warfarin, and non-vitamin K antagonists or direct oral anticoagulants (DOACs) prevents strokes in patients with AF.Yet up to half of all patients with AF do not receive oral anticoagulants, 4 and of those who do initiate treatment, 30% to 50% will discontinue therapy within the first 12 months. 5,6eatment costs burden patients directly, whether through out-of-pocket costs such as deductibles, copayments, and coinsurance for those who are insured, or the total cost of medications for those that are not insured.With 1 in 3 patients having trouble paying their medical bills, 7 delays, avoidance of care, and nonadherence expose patients to the risk of preventable strokes. 4,8ill, cost conversations occur infrequently in practice: only 1 in 3 clinicians report ever having such conversations with their patients. 9Cost conversations should be part of the work patients and clinicians do to co-create sensible and feasible treatment programs, work that is often called shared decision-making (SDM).Addressing the cost of treatment options at the point of prescription can result in feasible care, lessening the need for future discovery and remediation of cost-related adherence issues.These cost conversations during the clinical encounter may enable patients to make choices more closely aligned with their personal financial circumstances and preferences.
Except for 2 studies, one in surgical patients 10 and another in patients with several medical conditions, 11 there is limited research on the ability of SDM tools to promote and support conversations about treatment costs.
Current guidelines suggest the use of DOACs over warfarin 12 because DOACs are somewhat safer 13 and do not need regular blood draws for International Normalized Ratio (INR) monitoring.
7][18] On the other hand, warfarin entails indirect costs such as time and travel for INR testing, costs of laboratory tests, etc.
In the SDM4AFib multicenter trial, we demonstrated that using an SDM conversation tool designed for use during the clinical encounter improved aspects of SDM quality and clinician satisfaction. 19This SDM tool included information on the direct annual cost estimates of each treatment option presented to patients.To the extent that the tool highlighted cost (among other patient important issues such as bleeding risk, etc), the SDM tool may have encouraged cost conversations.The present secondary analysis sought to assess the association of using this effective SDM tool with (1) the incidence of cost conversations between patients with AF and their clinicians and (2) patient, clinician, and encounter factors associated with their occurrence; we also sought to estimate the degree to which these cost conversations were associated with the choice of anticoagulation agent.

Study Design
This is a secondary analysis of audiovisual recordings of clinical encounters and patient surveys obtained during the conduct of the Shared Decision Making for Atrial Fibrillation (SDM4AFib) trial, an encounter-randomized multicentered trial comparing standard care with and without the use of an SDM encounter tool.The trial protocol and its primary outcomes have been published elsewhere (trial protocol available in Supplement 1). 19,20Institutional review boards at the coordinating center (ie, the Mayo Clinic) and at participating sites approved all trial procedures.Specifically, recordings took place when both patients and clinicians gave written informed consent.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for observational studies.

Setting and Participants
The trial took place in 5 (academic, community, and safety-net) medical centers in the Midwest (3), Alabama (1)

Intervention
The Anticoagulation Choice SDM encounter tool 21 first provides the patient's tailored stroke risk at 1 or 5 years with and without anticoagulation using the CHA 2 DS 2 -VASc score.Then, the tool compares treatment options by patient-important issues, such as medication use, risk of bleeding (using the HAS-BLED [hypertension, abnormal kidney or liver function, stroke, bleeding, labile international normalized ratio, elderly age {over 65 years}, and drug of alcohol use] estimator), 22 the need for periodic monitoring, reversibility, and estimated out-of-pocket costs.Data from a public web service were used to offer cost estimates for each treatment option. 23A systematic review was the source of estimates of the cost of monitoring the INR to adjust warfarin doses 24 (eFigure in Supplement 2).

Outcomes
The outcome for this secondary analysis was the occurrence of a cost conversation in the encounter.

Examples of cost conversations included any statement or verbal exchange between clinician and
patient or caregiver about cost or other financial considerations surrounding anticoagulation agents, including indirect costs, insurance-related costs and offers of financial assistance.This conversation could be brief (eg, mere mention of costs being important) or lengthy (eg, extensive discussion on patient possibly entering the Medicare Part D "donut hole" if the patient did not have supplemental drug coverage).Cost conversations could take place altogether during the consultation or take place intermittently throughout the encounter.
Two reviewers, after training and documentation of reliability, worked independently to ascertain whether or not a cost conversation occurred.Chance-adjusted inter-rater reliability (using Lin's concordance correlation coefficient) was verified at the start, and again a third and two-thirds into the process of reviewing all recordings; this ranged from 0.84 to 0.96.We also used patients' electronic medical records to note which anticoagulation agent (warfarin vs DOAC, other decision) was selected.In postencounter surveys, patients reported whether "cost did not matter," "cost was one factor in my decision," or "cost was the sole factor in my decision."

Patient and Clinician Factors
We were interested in exploring the extent to which patient age, sex, race/ethnicity (ie, White vs non-White race, Hispanic vs non-Hispanic ethnicity), marital status (married vs other), highest level of schooling (high school graduate or less vs more than high school graduate vs other), stroke risk (low vs high), medication cohort (start vs review), and total number of medications taken were associated with the incidence of cost conversations.Marital status was included as it was assumed to be associated with financial means.Race was included because it was assumed to affect the need and comfort level of cost conversations.
We also explored associations between the incidence of cost conversations and clinician

Statistical Analysis
Our primary analysis was conducted at the encounter level using multilevel mixed-effects models adjusting for study arm and patient characteristics, with random effects allowed for clinics and clinicians within medical centers 25 and in accordance with the intention-to-treat principle, that is, analyzing encounters in the arm to which they were randomized.We also conducted a per-protocol analysis based on whether the SDM tool was actually used in the encounter.As we did not assume that patients and clinicians declined being recorded at random, we compared the participant characteristics of those with available recorded encounters with the characteristics of participants in all encounters.We tested univariate associations between cost conversations and patient, clinician, and encounter characteristics using χ 2 and nonparametric 2-sample Kruskall-Wallis tests.
Multivariable mixed-effects logistic regression included independent variables found to be significant in bivariate analysis and additional variables purported to influence conversations on costs (eg, income, education, marital status). 11,26,27We also tested univariate associations between the occurrence of cost conversation in the encounter and (1) the choice of anticoagulation agent, and (2) whether or not cost was a factor in medication decision-making.We used SAS software version 9.4 (SAS Institute) to conduct the analyses.P < .05 was considered statistically significant in 2sided tests.

Results
The ).The estimates of effect size were not different in the per-protocol analysis (Table 3).Examples of the range of conversations that took place in these encounters is included in the eTable in Supplement 2.
Patients who had cost conversations with their clinicians reported more frequently that cost influenced their decision of which anticoagulation agent to take when compared with patients who did not have cost conversations (patients reporting costs were one factor: OR, 3.66; 95% CI, conversations and the role of cost as a consideration was not associated with patient choice of medication during that visit (Table 4).

Discussion
Within a clinical trial using an SDM tool, cost conversations occurred in 3 of every 4 encounters between patients with AF and their clinicians.Use of an SDM tool and several clinician characteristics were significantly associated with cost conversations, whereas we could not find significant associations between patient characteristics other than income and the occurrence of cost conversations.Cost is influential in the choice of anticoagulant, but cost conversations were not associated with the selection of more or less expensive options.
Use of the SDM encounter tool was associated with a 10-fold increase in the odds of having a cost conversation during the encounter, a finding consistent with prior studies. 10,11While it is unclear how the Anticoagulation Choice SDM encounter tool may have supported cost conversations, it is possible that the cost content included in the tool, which highlighted the difference in out-of-pocket costs between warfarin and DOACs, helped at least identify cost as an important issue for patients with AF.It is also possible that the SDM tool prompted clinicians in general to more frequently ask about patient preferences, of which cost was one factor; patients using the tool also had more  b Risk ratios (or relative risks) in this table refer to the ratio of the probability of a cost conversation occurring in a reference group to the probability of its occurrence in the comparator group (for example female vs male clinicians or family medicine vs cardiology).
c This includes resident physicians, as well as nurses, physicians assistants, and pharmacy students.treatment, and diet/drug interactions. 19While the SDM tool was an important driver of cost conversations, we noticed that the incidence of cost conversations in the standard care arm (63%) was considerably higher than the incidence reported in other series 28,29 ; clinician reports ranged from 15% to 50% of encounters, while patient reports ranged from 15% for newly prescribed medications to 44% for ongoing care visits.This higher-than-usual rate may reflect the decisional setting: anticoagulation to prevent AF-associated strokes involves considering potentially expensive alternatives and implementing treatments with substantial fidelity.
Some clinician characteristics were associated with cost conversations.Female clinicians were nearly 3 times more likely than their male counterparts to discuss costs.This may reflect patients' expectations that female clinicians will engage in greater empathic communication and be more receptive to discuss biomedical and psychosocial issues in greater detail. 30Additionally, consultant clinicians were 4 times more likely to discuss costs than those in training, suggesting that clinician experience in managing AF includes bringing issues of cost to the conversation with patients.
Furthermore, clinicians practicing family medicine, internal medicine, and other specialties were approximately 12, 4, and 5 times more likely than cardiologists to discuss costs, respectively.
Although the reason for this finding is unclear, it may reflect differences in communication styles and attention to context between specialists and primary care clinicians. 31,32The only patient characteristic associated with having a cost conversation was income: patients with higher incomes (over $100 000) and lower incomes (less than $40 000) were less likely to have cost conversations relative to middle-income patients.These results reflect the association of financial capacity (for higher income) and the possible role of Medicare medication subsidies (for lower income) and the support cost conversations, given that very few have been tested. 33nally, we found that the occurrence of cost conversations was associated with patients who reported that cost was an important factor in their selection of anticoagulant.However, the actual choice of medication was not significantly correlated with the incidence of cost conversations, suggesting that either these conversations were not productive or consequential, or that considerations other than drug costs (eg, satisfaction with the existing regiment or the need to avoid diet restrictions or periodic monitoring) were more important in selecting an anticoagulation agent.
Further analysis of the content and quality of these conversations will likely yield greater insight about their role in improving effective patient-centered care.

Limitations
This study had several limitations.It is possible that the general nature of the costs offered in the SDM tool may have been sufficient to trigger cost conversations but in terms not individualized enough (ie, not able to estimate the exact out-of-pocket costs associated with each option) to support those conversations.Our findings may not apply broadly to the care of patients who would have opted not to participate in the SDM4AFib trial or to not have their encounter recorded, as happened with 10% of the trial's encounters.Also, our results may not apply to other clinical contexts and health systems (eg, European).Our study also does not analyze conversation initiator, quality, length, and content (including direct vs indirect costs) of cost conversations; however, we provide a range of examples of conversations in the eTable in Supplement 2. In contrast to these limitations, our study draws from a rigorous and large randomized trial and is based on direct observations of clinical encounters in which actual treatment decisions were made, from which we made reproducible assessments of the occurrence of cost conversations.

Conclusions
Cost conversations occurred in approximately 3 out of 4 encounters between patients with AF and their clinicians in this trial.These conversations are important as patients considered cost, and the patient's own financial capacity to shoulder it, an important issue in the decision-making process.The use of an SDM tool, several clinician characteristics, and only 1 patient characteristic (income) were associated with the occurrence of a cost conversation.These findings can inform further efforts to promote helpful cost conversations in practice.With increasing costs of health care passed on to patients, these conversations are likely to be more widely relevant as patients co-create treatment programs with their clinicians and implement those programs in their lives with sufficient fidelity for them to be effective and safe.
patients with atrial fibrillation and their clinicians, and what outcomes are associated with these conversations?Findings In this cohort study of 830 audiovisual recordings of encounters and participant surveys from a randomized trial comparing atrial fibrillation care with and without a shared decision-making (SDM) tool, cost conversations were associated with the use of an SDM tool, with middleincome patients, and with consultations conducted by female primary care staff clinicians.Cost conversations were associated with patients' decisionmaking processes but not final treatment choice.Meaning These findings suggest that SDM tools may inform efforts to promote cost conversations in practice, an important consideration when increasing costs of care are being passed on to patients.
consent document.Patients were enrolled into 2 cohorts: a start cohort including patients new to anticoagulation and a review cohort of patients receiving ongoing anticoagulation.All clinicians at the participating medical centers who regularly had conversations with patients with AF about anticoagulation were eligible for participation.Clinicians completed a baseline survey at the time of study enrollment.Data were collected from both patients and clinicians immediately following the clinical encounter: the postencounter survey captured patient demographics, health literacy, subjective numeracy, and medication and treatment outcomes.

Table 4 .
Incidence of Cost Conversations and Cost as a Factor in Patient Decision-Making and Medication Choice Downloaded From: https://jamanetwork.com/ on 09/28/2023 conversations about bleeding, anticoagulation treatment routine, reversing anticoagulation

Table 3 .
Multivariable Association of SDM Tool, Patient, Clinician, and Context Characteristics on Incidence of Cost Conversations need to talk about cost of treatment.Taken together, these findings suggest that interventions to promote cost conversations may need to be preferentially directed at modifying clinician and practice factors rather than focus on patient-only interventions.This hypothesis should be explored further.In addition, more attention should be given to developing evidence-based interventions to JAMA Network Open.2021;4(7):e2116009.doi:10.1001/jamanetworkopen.2021.16009(Reprinted) July 13, 2021 8/12 Downloaded From: https://jamanetwork.com/ on 09/28/2023