Computerized alert system confers more anticoagulation use, better outcomes in AF

Issue: December 2018

CHICAGO — An alert-based computerized decision support system for patients with atrial fibrillation was associated with more oral anticoagulant prescriptions and fewer major adverse CV events.

Perspectives from Mark Link, MD; Khaldoun Tarakji, MD, MPH

For the AF-ALERT trial, researchers randomly assigned 458 patients hospitalized for AF not already on anticoagulation (mean age, 73 years; 55% men) to receive the alert-based computerized decision support system or usual care.

“In a prior study of hospitalized patients with AF, antithrombotic therapy was omitted in nearly 40% of those at risk,” Gregory Piazza, MD, MS, cardiovascular medicine specialist at Brigham and Women’s Hospital and assistant professor of medicine at Harvard Medical School, said during a presentation at the American Heart Association Scientific Sessions. “Failure to prescribe antithrombotic therapy in these patients resulted in an excess 36% increase in major adverse events, including stroke and myocardial infarction. This low rate of adherence to guideline recommendations for stroke prevention in AF highlights a critical gap in best implementation of clinical practice by providers.”

The support system sends an alert to a clinician with the patient’s CHA2DS2-VASc score and the annual stroke rate associated with that score. The clinician is then prompted to do one of three things: initiate an order for anticoagulation, review the evidence-based practice guidelines or provide a reason for not proceeding with an anticoagulation order. For the latter, the doctor can select the following: bleeding risk too high, stroke risk not high enough, patient is at risk for falls, patient refuses anticoagulation or other.

Piazza noted a similar system demonstrated success for venous thromboembolism prophylaxis.

An alert-based computerized decision support system for patients with atrial fibrillation was associated with more oral anticoagulant prescriptions and fewer major adverse CV events.

Source: Adobe Stock

The primary efficacy endpoint, rate of anticoagulation prescription during hospitalization, at discharge and at 90 days, was 19.4% in the alert group compared with 7.1% in the usual-care group (P < .001).

Patients in the alert group had a lower occurrence of the secondary efficacy endpoint of major adverse CV events at 90 days after enrollment: 11.3% vs. 21.9% (P = .002). MACE was defined as stroke/transient ischemic attack, systemic embolism, MI and all-cause mortality. The HR for freedom from MACE at 90 days was 0.5 (95% CI, 0.31-0.81). Major bleeding or clinically relevant nonmajor bleeding at 90 days, the primary safety outcome of the trial, occurred in 4.4% of the alert group compared with 7.6% of the usual-care group (P = .15). The alert group also had reduced risk for all major adverse events (death, MI, stroke/TIA/systemic embolism or major/clinically relevant nonmajor bleeding) at 90 days (HR = 0.48; 95% CI, 0.31-0.73).

PAGE BREAK

The cumulative incidence of MI at 90 days was lower in the alert group (Gray’s test P = .0002), as was cumulative incidence of stroke/TIA/systemic embolism at 90 days (Gray’s test P = .01), Piazza said.

After receiving an alert, 35.4% of doctors opened an anticoagulation order set, 0.8% read guidelines and 63.7% declined to open an anticoagulation order set. The most common reason provided for not ordering anticoagulation was bleeding risk (50%), according to the researchers.

The system “reduced major adverse cardiovascular events beyond what we expected to see with the increase of anticoagulation,” Piazza said. “We saw something similar in the trial of VTE prophylaxis. [Computerized decision support] has the potential to be a powerful tool in prevention of cardiovascular events in patients with AF.” – by Erik Swain

Reference:

Piazza G, et al. LBS.03 – Late Breaking Clinical Trial: Harnessing Technology and Improving Systems for Global Health. Presented at: American Heart Association Scientific Sessions; Nov. 10-12, 2018; Chicago.

Disclosures: The trial was investigator-initiated and sponsored by Daiichi Sankyo. Piazza reports he received research grants from Bristol-Myers Squibb, Daiichi Sankyo, EKOS/BTG and Janssen and consulted or served on an advisory board for Bayer and Portola.

Perspective

Mark Link, MD

We know that people with AF are underprescribed anticoagulation for a variety of reasons: worry about fall risk, bleeding and issues with cost. Part of the reason patients may be underprescribed is because physicians don’t realize that they are at risk for stroke.

That is where this study comes in. It evaluated a best-practices alert on EPIC, which is an electronic medical record system widely used throughout the United States. The alert looks at the CHA₂DS₂-VASc score, which is a score from 0 (very little) to 9 (very high) which estimates risk for stroke, and a diagnosis of AF. If it sees that there is a CHA₂DS₂-VASc score of at least 1 and a diagnosis of AF, it alerts the physician to say the patient should be anticoagulated. Now the physician has to think about whether there is a reason not to anticoagulate them.

The nice thing about this trial is the design of a randomized approach to see if the alert changed practices or not. The researchers found pretty dramatic results. Patients who had best-practice alerts were far more likely to be prescribed anticoagulation in the hospital, at discharge and at 90 days. But the real interesting result is that the alert resulted in a 55% relative risk reduction in the composite of death, MI, stroke/TIA and systolic embolism. That is a huge, impressive difference.

The difference in outcomes was more than I would have expected given the difference in anticoagulation. That may have been a bit of a fluke. But we know anticoagulation significantly decreases the risk for stroke, and may decrease risk for some of the other outcomes, though that is not nearly as clear.

This tells us that these computerized programs to alert physicians about anticoagulation for patients hospitalized with AF are going to become more common, as they should be.

Mark Link, MD

Professor of Medicine

Director, Cardiac Electrophysiology

UT Southwestern Medical Center

Disclosures: Link reports no relevant financial disclosures.

Perspective

Khaldoun Tarakji, MD, MPH

There is much frustration felt by physicians about electronic medical records. They have not provided the expected benefits of making our daily lives easier and more efficient. Keeping up with current guidelines and treating chronic conditions have become difficult for all of us, and our colleagues in primary care.

This new technology attempts to fill this gap by providing computerized decision support about patients with AF who are at risk for stroke but not on anticoagulation. The software provides badly needed information such as an automated CHA₂DS₂-VASc score calculation. Not only did it lead to better use of anticoagulation upon discharge, but it also led to significant reductions in outcomes like death, MI, stroke and TIA.

It is remarkable that not only is this system feasible and led to better compliance with guideline-recommended anticoagulation practices, but also it led to better clinical outcomes.

This is just the beginning. We already know that systems like this work for other initiatives such as deep vein thrombosis prophylaxis. This is a combined effort by many institutions and teams. We can all learn from successful initiatives like this one, and we can all learn from our mistakes so that we don’t repeat them. We cannot afford to repeat the same mistakes. It is time for technology to help us provide better care for patients.

Khaldoun Tarakji, MD, MPH

Cardiology Today Next Gen Innovator

Staff Physician, Section of Electrophysiology and Pacing

Robert and Suzanne Tomsich Department of Cardiovascular Medicine

Sydell and Arnold Miller Family Heart & Vascular Institute

Cleveland Clinic

Disclosures: Tarakji reports he serves on medical advisory boards for AliveCor and Medtronic.