糖心vlog

Key Points

Question听 Does a multipronged physician-directed intervention with clinical decision support improve anticoagulation initiation for eligible adult emergency department patients with atrial fibrillation on discharge or within 30 days of discharge?

Findings听 In this stepped-wedge cluster randomized clinical trial conducted among 13 US community emergency departments (in 9 clusters), anticoagulation initiation improved slightly (from 63% to 68%) but this finding was not statistically significant. Decision support use was modest (27%).

Meaning听 Clinical decision support availability was not associated with higher rates of anticoagulation initiation in this trial; modest use of decision support may have contributed to its lack of effectiveness.

Importance听 Oral anticoagulation for adults with atrial fibrillation or atrial flutter (AFF) who are at elevated stroke risk reduces the incidence of ischemic stroke but remains underused. Efforts to increase anticoagulation initiation on emergency department (ED) discharge have yielded conflicting results.

Objective听 To evaluate the effectiveness of a multipronged intervention supporting anticoagulation initiation for eligible adult ED patients.

Design, Setting, and Participants听 The Clinical Decision Support to Optimize Care of Patients With Atrial Fibrillation or Flutter in the Emergency Department (O鈥機AF脡) pragmatic, stepped-wedge cluster randomized clinical trial was conducted from July 1, 2021, through April 30, 2023, at 13 community medical centers (in 9 clusters) of an integrated health system in Northern California. The study included adult ED patients with primary AFF eligible for anticoagulation initiation when discharged home. Clusters were randomly assigned to staggered dates for 1-way crossover from the control phase (usual care) to the intervention phase.

Intervention听 Physician education, facility-specific audit and feedback, and access to decision support, which identified eligible patients and recommended shared decision-making, anticoagulation initiation (if suitable), and timely follow-up.

Main Outcomes and Measures听 The main outcome was a composite of anticoagulation on discharge or within 30 days. A primary intention-to-treat analysis (decision support access regardless of use) and a secondary per-protocol analysis (decision support use) were performed. Multivariable analyses adjusted for intervention and exposure months with random effects, accounting for clustering by facility and patient.

Results听 A total of 3388 eligible patients with atrial fibrillation were discharged home: 2185 (64.5%) were receiving pre-ED arrival anticoagulation and 1203 (35.5%) were eligible for anticoagulation. Among the 1203 patients with an initiation-eligible encounter, the median age was 74.0 (IQR, 68.0-82.0) years and approximately half (618 [51.4%]) were men. Among the 387 patients with an initiation-eligible control encounter, 244 (63.0%) received anticoagulation (190 [49.0%] at discharge and 54 [14.0%] within 30 days). Among the 816 patients with an initiation-eligible intervention encounter, 558 (68.4%) received anticoagulation (428 [52.5%] on discharge and 130 [15.9%] within 30 days). There was no statistically significant change in initiation of anticoagulation associated with the intervention (adjusted odds ratio, 1.33 [95% CI, 0.75-2.35]; P鈥�=鈥�.13). Decision support was used for 217 eligible case patients (26.6%) (per protocol) and was associated with a statistically significant change in anticoagulation initiation when compared with 599 patients for whom decision support was not used (164 [75.6%] vs 394 [65.8%]; P鈥�=鈥�.008).

Conclusions and Relevance听 In this trial, a multipronged intervention to facilitate thromboprophylaxis among eligible ED patients with AFF did not significantly increase anticoagulation initiation. Opportunities exist to further improve stroke prevention among ED patients with primary AFF.

Trial Registration听 ClinicalTrials.gov Identifier:

Atrial fibrillation (AF) and atrial flutter are common cardiac arrhythmias that have a substantial negative effect on individuals and public health.^1,2 Both contribute to impaired quality of life and increase risk of ischemic stroke, heart failure, disability, and death.³ Oral anticoagulants (OACs) can reduce ischemic stroke risk by two-thirds and mortality by 25%.^4-6 Despite recommendations in clinical practice guidelines,^3,5-7 an estimated 30% to 50% of eligible adults with AF or flutter (hereinafter, AFF) remain untreated with OACs, even following widespread availability of direct OACs that are highly effective for stroke prevention and safer than vitamin K antagonists (eg, warfarin).^8-11

Patients with AFF often present to the emergency department (ED) for treatment. These encounters provide an opportunity to identify patients eligible for OAC initiation.¹² Initiating OACs on ED discharge home is safe and associated with a mortality reduction.¹³ In a large study, ED patients with AF who received OACs on discharge were more likely to be taking anticoagulants 1 year later compared with those only referred for treatment.¹⁴ Like other clinicians, emergency medicine physicians often underprescribe OACs to eligible patients with AFF.^15-19 Several ED studies have used clinical decision support systems (CDSSs) to increase OAC prescribing for eligible patients, with mixed results.^20-24

We designed a low-intensity multipronged intervention to improve care of ED patients with primary AFF. The intervention promoted evidence-based best practices across the breadth of ED AFF clinical issues, providing recommendations on rate reduction, cardioversion, stroke prevention, and trigger identification and avoidance (eTable 1 in Supplement 1).²⁵ The intervention included serial physician education, monthly facility-specific audit and feedback, and access to a web-based CDSS called RISTRA-AF (Risk Stratification-AF).

Emergency physicians hold a range of opinions on their role in long-term stroke prevention in AFF, particularly with regard to whether to inform patients of their stroke risk and either (1) defer prescribing to others or (2) initiate OACs.^12,26 We designed the CDSS to accommodate this variation. The RISTRA-AF system prepopulated the CHA₂DS₂-VASc (congestive heart failure, hypertension, age 75 years or older, diabetes, stroke, vascular disease, age 65-74 years, and female sex) score²⁷ to help physicians identify patients with AFF eligible for OACs. Physicians were encouraged to have a shared decision-making conversation with at-risk patients about stroke prevention and to facilitate timely outpatient follow-up.

To test the effectiveness of the intervention, we undertook this pragmatic, stepped-wedge cluster randomized clinical trial across 13 community EDs, examining OAC initiation among eligible patients on ED discharge and in the subsequent 30 days. We performed both (1) a primary intention-to-treat analysis of the intervention that included CDSS access (irrespective of use) and (2) a secondary per-protocol analysis of CDSS use vs nonuse during the intervention phase. We hypothesized that the intervention would increase the proportion of eligible adults with AFF who were prescribed OACs on ED discharge or within the following 30 days, and we also hypothesized that OAC initiation would increase with CDSS use.

The Clinical Decision Support to Optimize Care of Patients With Atrial Fibrillation or Flutter in the Emergency Department (O鈥機AF脡) pragmatic, stepped-wedge cluster randomized clinical trial of a multipronged intervention was undertaken over 22 months from July 1, 2021, through April 30, 2023.²⁵ We used an open cohort design (eMethods 1 in Supplement 1). The O鈥機AF脡 trial had 2 primary aims: to increase OAC initiation in eligible ED patients with primary AFF discharged home (the aim described herein) and to reduce hospitalization of the larger ED primary AFF population (which will be reported separately).²⁵ We selected a stepped-wedge approach because the educational program of a staggered rollout in our system was easier to implement than with a traditional parallel-group design.^25,28 This design also expanded intervention exposure across all study EDs, which was desirable because the intervention was thought to be an improvement over usual care. In this pragmatic trial, we designed the intervention to be low intensity with minimal disruption of customary clinical patterns of care, closer to usual than ideal care.²⁹ The trial protocol is presented in Supplement 2. The Kaiser Permanente (KP) Northern California Institutional Review Board approved the study with a waiver of informed patient consent because it was determined to be minimal risk. This study followed the Consolidated Standards of Reporting Trials () guideline extension for stepped-wedge cluster randomized clinical trials.³⁰

The O鈥機AF脡 trial was undertaken in KP Northern California, an integrated health care delivery system that provides comprehensive care for more than 4.5 million members who are representative of the ethnoracial and socioeconomic diversity of surrounding local and state populations.^31,32 Kaiser Permanente Northern California is a learning health care system with an applied research agenda and is supported by a comprehensive integrated electronic health record (EHR) (Epic; Epic Systems), which includes inpatient, outpatient, emergency, pharmacy, laboratory, and imaging data.^33,34

Emergency departments were staffed by emergency medicine residency-trained, board-certified (or board-eligible) physicians. During the study period, treating physicians had access to a standard discharge order set for AFF OACs, which recommended dabigatran, on the system鈥檚 pharmacy formulary, as the first-line OAC if not contraindicated. Care of patients taking OACs was managed by a regional pharmacy-led, telephone-based anticoagulation service using structured protocols.³⁵

The criteria for ED trial selection were as follows: (1) having an on-site emergency physician study champion and (2) not having participated in the pilot study. Pairs of EDs served by 1 shared staff of emergency physicians were treated as 1 site, leaving 9 study site clusters. The principal investigator allocated clusters to 1 of 9 sequences using a computer-generated randomization sequence. After an initial study period of 3 months in which all clusters were in the control phase, the intervention was implemented in 1 cluster per step at 1-month intervals, starting in October 2021 (Figure 1). The first 2 months of implementation were a transition period. The staggered rollout occurred over 9 months, followed by 11 months during which all clusters were in the transition or intervention phase.

The O鈥機AF脡 patients were adults (aged 鈮�18 years) receiving care for primary AFF in a participating ED during the study period (Figure 2), regardless of physician CDSS use. Patients with AFF were identified retrospectively using International Statistical Classification of Diseases, Tenth Revision diagnostic codes (I48.xx). Patients with AFF were excluded for any of the following concurrent ED diagnoses: pregnancy, ST-elevation myocardial infarction, acute myo- or pericarditis, acute pneumonia, pulmonary embolism, shock (eg, septic, hemorrhagic, or cardiogenic), recent major thoracic trauma (<48 hours), thyroid storm, or acute toxidrome (eg, sympathomimetic or anticholinergic). Patients with heart failure were not excluded to broaden the utility of the RISTRA-AF system. Criteria for inclusion in this stroke prevention analysis were an elevated CHA₂DS₂-VASc score (鈮�2 in men and 鈮�3 in women, as per 2019 US guidelines⁶), not currently or recently (<90 days) taking anticoagulants, and being discharged home from the ED (Figure 2).

Emergency department clinicians were blinded to the study until the month prior to or concurrent with gaining RISTRA-AF access, at which time they were educated about best practices for ED AFF management (eTable 1 in Supplement 1) and shown how to use the RISTRA-AF system.²⁵ Throughout the intervention phase, standardized emails were distributed monthly to study EDs to acknowledge local CDSS users and highlight elements of recommended ED AFF management. Users who passed a brief training quiz on RISTRA-AF content received small gift cards for their first 3 enrollments, as with other studies.³⁶ In June 2022, we began distributing to intervention site leads their department鈥檚 performance metrics on stroke prevention to share with their physicians. No outreach was undertaken among primary care clinicians.

The RISTRA-AF system was accessible within the ED Navigator of the EHR but needed to be manually activated; no alert within the EHR reminded physicians of CDSS eligibility. Details of decision support leading up to anticoagulation recommendations are explained in eMethods 2, eFigure 1, and eFigure 2 in Supplement 1. The anticoagulation recommendation screen provided the patient鈥檚 CHA₂DS₂-VASc score, their estimated annual stroke risk, and several management options (Figure 3). The screen included an eConsult hyperlink to our health system鈥檚 pharmacy-led anticoagulation management service, which was indicated when prescribing OACs or when patients requested more information on the risks and benefits of anticoagulation. A patient-specific educational handout was available to be printed from the anticoagulation screen (eFigure 3 in Supplement 1). The clinician was also presented with age-based recommended doses of dabigatran, as well as common contraindications, and hyperlinks to a KP patient handout on dabigatran and to a KP table on alternative anticoagulants. The CDSS reminded physicians to consider bleed risk and to defer OACs if indicated (Figure 3). To help physicians estimate bleed risk, we provided a hyperlink to internal CDSS screens about the HAS-BLED (hypertension, abnormal kidney and/or liver function, stroke, bleeding history or predisposition, labile international normalized ratio, elderly [older age], and concomitant drug and/or alcohol use) scoring system, particularly its intent, its variables, and its score-based bleed risk estimates (eFigures 4 and 5 in Supplement 1).^37-39

Our primary outcome was a composite of OAC initiation among eligible patients with AFF, either on ED discharge or in the subsequent 30 days. Although the latter was not directly controlled by the treating emergency physician, the shared decision-making conversation with ED patients about stroke prevention may have facilitated post-ED discharge OAC initiation, as many patients wanted to continue the conversation with their primary care clinician. The study was designed to identify an anticipated 4.9% change in rates of OAC initiation with 80% power (detailed in eMethods 3 in Supplement 1).²⁵

We examined patient characteristics, including demographics, comorbidities, and CHA₂DS₂-VASc scores, as well as treatment variables, including use of the CDSS and OAC initiation within 30 days of ED discharge, using data extracted from EHRs and the RISTRA-AF system. Race and ethnicity was self-reported as Asian, Black or African American (hereinafter, Black), Hispanic or Latinx (hereinafter, Hispanic), White, or other race or ethnicity (including American Indian or Alaska Native or Native Hawaiian or Other Pacific Islander). Race and ethnicity data were collected to demonstrate that the diversity of the cohort reflects the diversity of the population of northern California. We randomly selected for manual review 40 case patients from the intervention phase in which OACs, although indicated, were not prescribed on ED discharge or within the subsequent 30 days. As in prior work, manual medical record review was conducted to examine unstructured documentation in the EHRs to identify treating ED physicians鈥� documentation of stroke risk and explanations for withholding OACs.¹⁷

We present categorical data as frequencies and proportions and continuous variables as medians with IQRs. We report binomial exact 95% CIs where appropriate. We examined within- and between-cluster correlation over time to elucidate possible correlation structures, including possible time decay in the correlation over time. We determined the intraclass correlation and churn rates over time and by cluster. We used descriptive statistics to examine outcome trends over time overall and by intervention status. We used marginal models (eMethods 4 in Supplement 1) to estimate the average treatment effect of the intervention on OAC initiation while allowing for clustering by patient and facility with appropriate correlation structures and adjustment for temporal trends and other time-related effects. We undertook a sensitivity analysis that included only each patient鈥檚 first eligible encounter. We performed both (1) a primary intention-to-treat analysis (decision support access regardless of use) and (2) a secondary per-protocol analysis (decision support use). We also compared 2 groups within the intervention phase: encounters without and with CDSS use, a preplanned analysis. Both groups received the same CDSS access intervention but differed in clinician CDSS use. The latter served as a per-protocol analysis to evaluate the effectiveness of CDSS use. P鈥�<鈥�.05 (2-tailed) was considered statistically significant. All analyses were conducted with SAS, version 9.4 (SAS Institute Inc).

A total of 3388 case patients meeting criteria for OACs were discharged from the ED during the study period: 2185 (64.5%) were receiving OACs before ED arrival and 1203 (35.5%) were eligible for OAC initiation. The 1203 initiation-eligible case patients (585 women [48.6%] and 618 men [51.4%]) had a median age of 74.0 (IQR, 68.0-82.0) years and a median CHA₂DS₂-VASc score of 4.0 (IQR, 3.0-5.0). Initiation-eligible case patients identified as Asian (163 [13.5%]), Black (71 [5.9%]), Hispanic (117 [9.7%]), White (803 [66.7%]), or other race or ethnicity (49 [4.1%]). A total of 387 patients (32.2%) were in the control phase and 816 (67.8%) were in the intervention phase. Patient characteristics were similar between the 2 study phases (Table).

Among the 387 initiation-eligible control patients, 244 (63.0%) received OACs (190 [49.0%] on discharge and 54 [14.0%] within 30 days). Among the 816 initiation-eligible intervention case patients, 558 (68.4%) received OACs (428 [52.5%] on discharge and 130 [15.9%] within 30 days). The increase in OAC initiation from 63.0% (control phase) to 68.4% (intervention phase) was 5.4 percentage points (95% CI, 鈭�0.04 to 11.1; P鈥�=鈥�.07) (Figure 4). When adjusted, there was no statistically significant change in OAC initiation associated with the intervention (adjusted odds ratio, 1.33 [95% CI, 0.75 to 2.35]; P鈥�=鈥�.13). When patients receiving OACs before ED arrival were included, the increase in overall OAC use from 86.4% (control phase) to 88.9% (intervention phase) was 2.5 percentage points (95% CI, 0.07 to 4.92; P鈥�=鈥�.04). We report intraclass correlation and churn rates in eResults in Supplement 1. Sensitivity analysis including only first encounters resulted in minimal change to the odds ratio estimate and no change to the inference.

Among encounters in the intervention phase, patients receiving OACs had a lower prevalence of several comorbidities and a different distribution of CHA₂DS₂-VASc scores compared with untreated patients (eTable 2 in Supplement 1). During the intervention phase, physicians used the RISTRA-AF system with 217 eligible case patients (26.6%). Compared with 599 case patients for whom the CDSS was not used, those with physician CDSS use were dissimilar only with regard to history of prior AFF (eTable 3 in Supplement 1). In the per-protocol analysis, CDSS use was associated with a statistically significant increase in OAC initiation when compared with nonuse (164 [75.6%] vs 394 [65.8%]; P鈥�=鈥�.008) (Figure 4).

We randomly selected 40 eligible case patients from the intervention phase for whom OACs were not initiated (on discharge or within 30 days) and conducted a manual medical record review. Twelve physicians鈥� clinical notes explained that after a shared decision-making discussion, their patients had declined anticoagulation. Eleven physicians noted elevated bleeding risks (eg, recent gastrointestinal bleed, pancytopenia, or upcoming major surgery). Nine physicians documented that their patients deferred the OAC decision until discussion with their primary care clinician. Six physicians did not mention stroke risk or OACs in their note. Finally, 2 physicians had undercalculated the CHA₂DS₂-VASc score (the method of score calculation was not documented).

In this pragmatic, stepped-wedge cluster randomized clinical trial conducted in a US community-based integrated health care system, a multipronged intervention including physician education, facility-specific audit and feedback, and CDSS access was not associated with a statistically significant change in OAC initiation on or within 30 days of ED discharge among patients with primary AFF. We address these results in light of our power calculations in eDiscussion 1 in Supplement 1.

We adopted a pragmatic trial design following recommendations from the PRECIS-2 (Pragmatic Explanatory Continuum Indicator Summary) tool.²⁹ We sought to avoid major disruptions to usual care and to allow the intervention to be more easily adapted to less integrated care delivery settings. To illustrate, first, we addressed patients with typical AFF (including those with concomitant heart failure). Second, we kept changes to care delivery minimal. Third, stroke prevention recommendations were flexible and easy to implement (eg, making OAC initiation optional). Fourth, we did not embed an automated alert in the EHR. Fifth, CDSS activation was elective and physician driven. Finally, we used facility-specific rather than physician-specific audit and feedback, as we preferred a more sustainable approach and did not want physicians to feel their practice patterns were being scrutinized. These low-intensity characteristics supported the acceptability of the intervention in an ED environment whose baseline complexity was exacerbated by the concurrent COVID-19 pandemic. This streamlined implementation, however, may have reduced the likelihood of CDSS use and, therefore, the potential effectiveness of the intervention bundle on OAC initiation. These pragmatic accommodations may explain, in part, the relatively modest use of the CDSS.

To our knowledge, there has been only 1 other randomized clinical trial published to improve OAC initiation among eligible ED patients with AF discharged home. The Rapid Atrial Fibrillation/Flutter (RAFF-3) stepped-wedge cluster randomized clinical trial was conducted across 11 EDs in Canada, and it had a primary outcome of reducing ED length of stay.²³ Improving OAC use was a secondary aim. In contrast to the O鈥機AF脡 trial, RAFF-3 excluded patients with permanent AF and those who had cardioverted spontaneously before initiation of ED treatment; more than 90% of enrolled patients had presumed recent-onset AF less than 48 hours in duration. Their preintervention OAC use was high: 61% of eligible patients received OACs before ED arrival and an additional 17% were prescribed OACs on discharge. The intervention included multimodal physician and nurse education of Canadian AF best practices^40,41 (supported by a local champion with local action plans based on prior physician interviews that had identified barriers to implementation), a smartphone application, and facility-specific audit and feedback reports. The intervention failed to significantly increase OAC use on discharge. Initiation of OACs following discharge was not reported.

Several pre-to-post studies of ED OAC-focused CDSS interventions have been published, but methodological differences prevent ready comparison with our trial (eDiscussion 2 in Supplement 1).^20-22 Forthcoming trials will be more comparable.²⁴

On manual medical record review, we ascertained that most emergency physicians not prescribing OACs had a clinically acceptable reason for noninitiation (eg, the patient had a recent hemorrhage or upcoming surgery). After a shared decision-making discussion, some patients declined anticoagulation or wanted to continue the discussion with their primary care physician. A minority of patients, however, had their predicted stroke risk either unaddressed or even miscalculated, shortcomings in care that may have been reduced by more widespread CDSS use. Additional methods of enhancing ED physicians鈥� use of CDSS while maintaining a low-intensity approach need to be explored.⁴²

This study had several limitations. Unlike some health systems, very high rates of baseline anticoagulation in the control phase of this study (64.5%) reduced the pool of patients with primary AFF eligible for OAC initiation. Thus, a larger sample may have been required, as other studies have found.⁴³ We selected OAC initiation as our outcome because it focused on changing physician behavior for a highly effective preventive therapy. We did not study more patient-focused measures, such as filled prescriptions or longer-term ischemic stroke outcomes. Physicians who wanted assistance addressing stroke prevention may have been more likely to use the RISTRA-AF system, introducing selection bias. A Hawthorne effect was unavoidable: physicians during the intervention period may have increased their adherence to best practices knowing their behaviors were being studied. Stroke prevention is a critical component of comprehensive AFF care but may not be in the forefront of ED AFF management priorities while addressing more urgent issues of heart rate reduction and cardioversion. Fortunately, thromboprophylaxis in AFF is a multidisciplinary responsibility. Many different clinician touchpoints are necessary to optimize patient care. The trial鈥檚 integrated care delivery setting may limit generalizability of these findings, although the relatively low-intensity intervention involves components found in many quality improvement projects, making it more readily adaptable to a broad variety of settings.

In this pragmatic, stepped-wedge cluster randomized clinical trial among 13 US community EDs, a multipronged educational and technical intervention did not significantly increase anticoagulation initiation. Opportunities exist to further improve stroke prevention among ED patients with primary AFF.

Accepted for Publication: September 12, 2024.

Published: November 6, 2024. doi:10.1001/jamanetworkopen.2024.43097

Open Access: This is an open access article distributed under the terms of the CC-BY License. 漏 2024 Vinson DR et al. 糖心vlog Open.

Corresponding Author: David R. Vinson, MD, Department of Emergency Medicine, Kaiser Permanente Roseville Medical Center, 1600 Eureka Rd, Roseville, CA 95661 (drvinson@ucdavis.edu).

Author Contributions: Ms Warton had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Vinson, Warton, Durant, Mark, Ballard, Loyles, Solomon, Go.

Acquisition, analysis, or interpretation of data: Vinson, Warton, Durant, Mark, Ballard, Hofmann, Sax, Kene, Lin, Poth, Ghiya, Ganapathy, Whiteley, Bouvet, Rauchwerger, Zhang, Shan, DiLena, Kea, Pai, Go, Reed.

Drafting of the manuscript: Vinson, Warton, Durant, Hofmann, Sax, Ghiya, Zhang.

Critical review of the manuscript for important intellectual content: Vinson, Warton, Durant, Mark, Ballard, Sax, Kene, Lin, Poth, Ganapathy, Whiteley, Bouvet, Rauchwerger, Zhang, Shan, DiLena, Kea, Pai, Loyles, Solomon, Go, Reed.

Statistical analysis: Warton, Mark.

Obtained funding: Vinson, Ballard, Rauchwerger.

Administrative, technical, or material support: Vinson, Hofmann, Sax, Lin, Poth, Ganapathy, Whiteley, Rauchwerger, Zhang, Shan, DiLena, Kea, Loyles, Reed.

Supervision: Vinson, Mark, Ballard, Poth.

Conflict of Interest Disclosures: Dr Solomon reported receiving grants from Bristol Myers Squibb and Edwards Lifesciences outside the submitted work. Dr Go reported receiving grants from the Bristol Myers Squibb鈥揚fizer Alliance outside the submitted work. Dr Reed reported receiving grants from the Kaiser Permanente Northern California Delivery Science and Applied Research (DARE) program during the conduct of the study. Dr Reed also reported receiving grants from the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute on Aging, the Agency for Healthcare Research and Quality, and the Patient-Centered Outcomes Research Institute outside the submitted work. No other disclosures were reported.

Funding/Support: This study was supported by The Permanente Medical Group DARE program and its Physician Researcher Program.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentation: This study was presented in part at the Annual Meeting of the Society for Academic Emergency Medicine; May 16, 2024; Phoenix, Arizona.

Data Sharing Statement: See Supplement 3.

Additional Contributions: We are grateful to our multidisciplinary Atrial Fibrillation Advisory Panel for their content expertise. We thank the members of The Permanente Medical Group Physician Researcher Program for their wise input on study matters. We are indebted to the hundreds of busy emergency physicians who provided invaluable feedback and did the indispensable work of implementation. We also thank the hospitalists, cardiologists, pharmacists, and nurses who supported our changes in practice. We are grateful to the leadership of The Permanente Medical Group for investing in embedded researchers and applied research. Lastly, we thank the patients of Kaiser Permanente for helping us improve care using information collected through our electronic health record systems.