Automated Physician Notifications to Improve Guideline-Based Anticoagulation in Atrial Fibrillation

The overall goal is to improve outcomes among patients with atrial fibrillation (AF) by preventing stroke. The investigators propose to implement an automated algorithm using electronic medical record (EMR) data to alert physicians in a large primary care practice network at the Massachusetts General Hospital (MGH) of their patients with AF and elevated stroke risk that are not taking an anticoagulant for stroke prevention. The investigators hypothesize that interventions to notify physicians of such individuals may prompt reassessment for the need for anticoagulation, and thereby increase guideline-indicated anticoagulation rates. Additionally, in a survey component, physicians will characterize reasons for not pursuing anticoagulation in AF patients at elevated risk for stroke.

The overall goal is to improve outcomes among patients with atrial fibrillation (AF) by preventing stroke. The investigators propose to implement an automated algorithm using electronic medical record (EMR) data to alert physicians in a large primary care practice network at the Massachusetts General Hospital (MGH) of their patients with AF and elevated stroke risk that are not taking an anticoagulant for stroke prevention. Using a medical record algorithm, there are an estimated 2,000 to 3,000 such patients in the primary care practices at MGH. The investigators hypothesize that interventions to notify physicians of such individuals may prompt reassessment for the need for anticoagulation, and thereby increase guideline-indicated anticoagulation rates. The investigators define the following three Specific Aims to address the primary study hypothesis: Specific Aim 1: Implement an automated alerting system within the EMR to notify primary care physicians (PCPs) of patients with AF at elevated stroke risk that are not being treated with anticoagulants. To maximize efficiency of the effort, the alert will not require a concurrent clinic visit, but rather will occur with existing data aggregated from the medical record independent of any clinical encounter. Specific Aim 2: Test whether the alerting system increases the rate of anticoagulation at 3 months after implementation. Specific Aim 3: Characterize the reasons for not pursuing anticoagulation in AF patients at elevated risk for stroke. Specific Aim 4: Assess how PCPs want to be alerted about anticoagulation status in the future and what types of support they find helpful

For patients randomly selected for the baseline alert arm, their physicians will be alerted via email that a patient(s) under their care has atrial fibrillation, is at high risk of stroke, and is not currently anticoagulated. Physicians will also be asked to complete a survey related to anticoagulation for each patient and will be provided with educational resources and consultation services.

For patients randomly selected for the 3-month alert arm, their physicians will not be notified during the 3-month study follow-up period. Instead, PCPs will be sent alerts after 3-months via email for these patients.

PCP notification at baseline that patient has atrial fibrillation, high stroke risk, and is not anticoagulated.

Difference in the proportion of patients with AF at high risk of stroke taking an anticoagulant at 3-months in the baseline alert arm , compared to the proportion taking an anticoagulant at 3-months in the 3-month alert arm

Difference in the proportion of patients with AF at high risk of stroke taking an anticoagulant at 3-months in the baseline alert arm , compared to the proportion taking an anticoagulant at 3-months in the 3-month alert arm, stratified by type of anticoagulant (direct oral anticoagulants vs. warfarin)

Physician characteristics derived from the questionnaire and hospital databases that are associated with having a low proportion of the physician's panel of atrial fibrillation patients on oral anticoagulants

Physician-level characteristics associated with having anticoagulation rates of the physician's panel of atrial fibrillation patients in the lowest quartile among all eligible physicians.

Patient-level characteristics derived from the questionnaire and the electronic health record that are associated with not being prescribed an anticoagulant

Patient-level characteristics that may influence decision making for a physician to not prescribe an anticoagulant

Inclusion Criteria: Adult patients 18 and older seen in Massachusetts General Hospital primary care practices in the past 3 years Diagnosed with atrial fibrillation Increased risk of stroke (CHA2DS2VASc score ≥ 2) Not currently taking an anticoagulant Exclusion Criteria: Patients who are subsequently identified as having died prior to or during the course of the study intervention using the Social Security Death Index Listed in the Massachusetts General Hospital system as having a PCP outside of the network

Ashburner JM, Atlas SJ, Khurshid S, Weng LC, Hulme OL, Chang Y, Singer DE, Ellinor PT, Lubitz SA. Electronic physician notifications to improve guideline-based anticoagulation in atrial fibrillation: a randomized controlled trial. J Gen Intern Med. 2018 Dec;33(12):2070-2077. doi: 10.1007/s11606-018-4612-6. Epub 2018 Aug 3.