Anticoagulation Optimization Through a Specialized Center for Atrial Fibrillation

Abstract

Atrial fibrillation is a prevalent condition, which is an independent risk factor for ischemic stroke. Anticoagulation is widely used to mitigate stroke risk but remains underprescribed in certain circumstances. The authors evaluated the performance of a health record database query in identifying patients who may benefit from anticoagulation who did not have an active prescription for anticoagulation. Patient charts were reviewed by primary care providers and reasons for deferral of anticoagulation were provided. A subset of the patients identified were found to not have an appropriate reason for deferring anticoagulation. This initiative ultimately led to an increase in referrals to the center for specialty care of atrial fibrillation.

Introduction

Atrial fibrillation (AF) is a highly prevalent condition among the aging population, with prevalence estimates increasing with advances in digital health.¹ Presently, AF is estimated to affect 59 million individuals worldwide.¹ The population affected by AF is estimated to carry a 4–5-fold increase in the risk of stroke.² Despite this, underutilization of systemic anticoagulation in patients with AF at elevated risk for stroke has been well described.³ Lack of comfort with anticoagulation in the presence of bleeding risk factors and lack of understanding of anticoagulation indications may contribute to this finding. Database queries of electronic health records (EHRs) can identify patients who may not be adequately treated with anticoagulation, with some inherent limitations. A specialist review of clinical data is required to determine the clinical significance of a charted AF diagnosis as well as the validity of factors considered to be contraindications or barriers to anticoagulation.⁴

As EHR capabilities expand, many platforms offer a suite of clinical decision support features, which can improve value metrics and efficiency. Clinical decision support systems to identify AF patients at elevated risk for stroke have been found to perform well in settings where AF is clinically diagnosed acutely.⁵ However, these systems may lack sensitivity in the outpatient care setting if documentation of AF is incomplete. Increasingly, outpatient providers in the primary care system are also faced with decisions for device-detected AF of variable clinical significance, which may or may not be reflected in the EHR. Even in the specialty setting, there is known to be wide practice variation in anticoagulation recommendations for this population.⁶ This finding is likely due to nuanced decision-making encompassing patient preference, bleeding risk, comorbidities, and arrhythmia burden. Such patients are well suited to dedicated specialist visits allowing for undivided focus on this decision.

AF is generally diagnosed initially in the primary care or emergency care setting. A minority of these patients are expeditiously referred to specialty care. However, early specialist involvement is associated with improved outcomes. This effect is likely attributable in part to early initiation of anticoagulation.⁷ Centers for AF are designed to improve access to early specialist evaluation. The authors’ center model includes 24/7 access to specialist consultation for patients admitted to the emergency department with a first diagnosis of AF. These patients are offered a visit in the outpatient setting within 24 hours with an advanced practice provider supervised by an electrophysiologist. This model has been demonstrated to reduce the cost of care, emergency care visits, and inpatient admissions.⁸ Specifically, specialty centers can raise awareness of and facilitate referral for left atrial appendage occlusion and routinely implement shared decision-making tools for pharmacologic anticoagulation, which can improve patient satisfaction and adherence.^7,9

Despite improvement in clinical outcomes, center referrals may lag behind demand. Challenges with the current practice model include variability in clinical practice among primary care providers, lack of awareness of specialty center resources or referral pathways, and lack of familiarity with alternative options to pharmacologic anticoagulation.

Within the authors’ practice setting, an integrated provider–payor network that encompasses 40 hospitals and 700 clinical locations, improving access to more centralized specialty care is a key focus for quality improvement. The aim of this project was to develop an EHR-driven AF care pathway to increase adherence to guideline-directed anticoagulation in the primary care setting through targeted guidance provided by practitioners representing the Center for AF. This project was part of a larger collaborative initiative between the Heart Rhythm Society and the Bristol-Myers Squibb/Pfizer Alliance to design and implement a quality improvement framework for improved AF care and outcomes. The resulting framework and complementary resources may be found at www.CardiQ.org.

In this study, the authors provided an introductory session with primary care practices, which may benefit from an expanded relationship with the authors’ specialty care center. The authors measured referral rates before and subsequent to this intervention.

Methods

Patients with a charted diagnosis of AF and a guidelines-based class I recommendation to receive anticoagulation (CHA₂DS₂-VASc score ≥ 2 in males and ≥ 3 in females)¹⁰ were included in the study. Diagnosis of AF and thromboembolic risk factors were determined via a query of the EHR (EpicCare). Rates of appropriate anticoagulation were stratified by charted primary care providers and grouped by individual practice. The authors then targeted 5 sites within Community Medicine, Inc. (CMI), the largest primary care group within the authors’ system, which encompasses >500 individual providers across 130 practice sites. CMI has an infrastructure in which there is designated physician leadership at each site to support quality improvement interventions with metrics, dashboards, and value incentives. The 5 selected practices were high-volume sites (defined as practice size >200 patients) with relatively lower reported rates of anticoagulation use for high-risk AF patients compared with other sites of similar size.

Electrophysiology practitioners from the system’s Center for AF met with representatives from these primary care practices to review a standardized educational presentation. The presentation reviewed current society guidelines for the prescription of anticoagulation as well as the role of the center for AF in providing advanced specialty care. Specifically, consideration of specialist referral was recommended in high-risk patients with contraindication to oral anticoagulation for consideration of left atrial appendage occlusion, as well as in intermediate-risk patients (CHA₂DS₂-VASc score 1 in males and 2 in females) for shared decision-making conversation.

Providers were given a list of patients identified via the EHR query as not receiving anticoagulation despite eligibility and asked to review their charts and determine the rationale to better understand the performance of the EHR-driven pathway in correctly identifying target patients as well as the diversity of reasons for not prescribing anticoagulation in this patient population. Measured outcomes included: baseline rates of anticoagulation as determined by the EHR query, reasons for not prescribing anticoagulation and revised estimate of patients who may be appropriate for intervention, and number of AF center referrals preceding and following the educational session.

Results

The overall baseline rate of active anticoagulation prescriptions among eligible patients was estimated by initial database query to be 83% across all CMI sites. Among the 5 targeted primary care practices, rates of patients identified as not appropriately anticoagulated within each individual practice ranged from 17% to 31%, with an overall average of 23%. A total of 172 patients in these practices were identified as not on recommended anticoagulation according to CHA₂DS₂-VASc score by the initial query.

Documented reasons for the lack of prescribed anticoagulation according to primary care provider review are shown in Figure 1. The most common provider-documented reasons for patients not appearing on anticoagulation were: the patient was on oral anticoagulation that was not correctly reflected in the EHR (32 patients, 19%), prohibitive bleed risk (26 patients, 16%), no clear clinical diagnosis of AF (26 patients, 16%), a cardiology provider had recommended against anticoagulation (22 patients, 13%), the patient was on aspirin (22 patients, 13%), the patient only had one AF episode (13 patients, 8%), and the patient had postoperative AF (7 patients, 4%).

Figure 1.

Distribution of reasons for lack of anticoagulation prescription as determined by providers reviewing medical record information for individual patients. Abbreviations: AF, atrial fibrillation; ASA, aspirin; LAAO, left atrial appendage occlusion; OAC, oral anticoagulation; OP, outpatient.

There was a modest increase in referrals to the AF center from these groups following the educational session. In the 3 months before the session, providers from the targeted practices referred 17 patients to the center. In the 3 months following the session, these providers referred 21 patients.

Discussion

In this study, multiple reasons were cited for the lack of documented anticoagulation among patients considered to be eligible based on CHA₂DS₂-VASc score. The majority of these cases were either clinically appropriate contraindications to anticoagulation or misidentified as an artifact of the EHR query. Approximately 1 in 4 cases from the raw data analyzed may be appropriate for an actionable recommendation to start anticoagulation based on clinically appropriate criteria. In 13% of cases, the provided rationale was the use of aspirin for thromboembolic prophylaxis, despite guideline recommendations for oral anticoagulation in the population studied. All of these patients would be appropriate candidates for anticoagulation. A portion of the other documented reasons (single AF episode, postoperative AF, no documented AF since increase in CHA₂DS₂-VASc score) may under certain conditions warrant reconsideration of anticoagulation, also contributing to the fraction of patients who would be appropriate candidates for anticoagulation.

An additional 16% of patients were documented not to be on anticoagulation due to bleeding risk, and 2% of patients had declined oral anticoagulation. These indications have been found to be particularly high yield for referral to the center AF. Many of these patients are appropriate for shared decision-making interventions to inform the decision to adhere to anticoagulation recommendation or for referral for left atrial appendage closure if bleeding risk is assessed to be truly prohibitive.

A plurality of the patients identified by the EHR query were inappropriately included. About 19% of the patients were determined after provider review to be prescribed anticoagulation, and an additional 16% did not have clear evidence of AF. Notably 1% of patients had a CHA₂DS₂-VASc score below the cutoff for universal recommendation for anticoagulation. This finding highlights the importance of fostering individual provider awareness of current anticoagulation guidelines and available AF specialty resources. The automated algorithm performance was demonstrated to lack precision in correctly identifying at-risk patients. It is likewise possible that a percentage of the population of interest was not detected by the EHR query, either due to incomplete documentation or coding of thromboembolic risk factors or incomplete documentation of AF. When accounting for performance limitations, the purely EHR-based tool may not have correctly identified the most underperforming centers.

The authors found that the educational interventions were productive in establishing relationships between the primary care sites and the AF center. This study measured an absolute number of referrals, and there was a trend towards increased referrals in the interval following the intervention. The baseline rate of referrals was notably high, and the incremental volume may reflect referrals dedicated to anticoagulation management rather than rhythm control or other indications. Despite an initial referral indication for anticoagulation management, many of these patients may also benefit from evaluation for potential rhythm control as well. Recent guidelines have recommended consideration of an early rhythm-control approach in low-risk patients,¹⁰ which intersects with the population faced with equivocal long-term anticoagulation recommendations requiring more complex or shared decision-making as well.

Aside from an increase in the absolute number of referrals, the educational sessions may have had an additional impact on referring physician satisfaction by establishing a dialogue that may lead to ongoing education and support through more informal and virtual channels. In a decentralized system with distance barriers, personal points of contact and visibility remain a challenge. A secondary intention in providing one-time educational sessions is to establish improved communications with primary care providers on a longer-term basis to improve care with evolving patient needs.

This study has several limitations that highlight the importance of further investigation. Notably, patients with intermediate thromboembolic risk (CHA₂DS₂-VASc score of 1 in males or 2 in females) were not included. The query was limited to higher-risk patients to improve specificity in identifying patients who objectively were not receiving guideline-recommended therapy. Conversely, intermediate-risk patients lack absolute guidelines for anticoagulation. Society recommendations focus on the revision of risk based on multiple criteria.¹¹ These decisions are more time-intensive and therefore well suited to a specialty care setting. Further investigation into rates of anticoagulation and referrals for this population may also demonstrate an opportunity for quality improvement.

The query also did not account for the appropriate dosing of anticoagulation. Anticoagulation is not prescribed according to trial dosing or Food and Drug Administration labeling in a high percentage of patients in many settings for a variety of reasons.¹² Under- or over-dosing of anticoagulation has been associated with poor clinical outcomes including increased thrombotic rates and increased bleeding events.¹³ An EHR query to determine rates of adherence to appropriate dosing in this system may be useful to assess performance by another clinically important metric and help target additional settings with educational needs.

In the future, a longitudinal study may yield additional insight into rates of appropriate anticoagulation or alternative such as left atrial appendage occlusion in the patients identified as not appropriately anticoagulated. Further investigation may determine that initial contact with the specialty center may yield a more perpetual attention to appropriate anticoagulation independent of referrals to a specialty center.

Conclusion

Multiple explanations exist for deviation from guideline-based recommendations for anticoagulation in AF. EHR screening can, with some limitations, identify high-yield opportunities to optimize AF anticoagulation through involving a specialized center for AF. More widespread and comprehensive education and establishing a long-term pattern of referrals will be important to fully leverage the impact of this intervention.

Conflicts of Interest

The authors have no conflicts of interest to disclose.

Funding

This project was funded by a quality improvement grant from the Heart Rhythm Society.