March 03, 2020
2 min read

Algorithms identify ED oral anticoagulation prescriptions for AF

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact

Clare L. Atzema

Several algorithms identified oral anticoagulation prescriptions given in the ED for patients with atrial fibrillation, according to a study published in JAMA Network Open.

Clare L. Atzema, MD, MSc, scientist of evaluative clinical sciences at Sunnybrook Research Institute in Toronto, staff emergency physician at Sunnybrook Health Sciences Centre in Toronto, associate professor in the division of emergency medicine at University of Toronto and senior scientist at ICES in Toronto, and colleagues analyzed data from 2,015 patients who visited an ED in Ontario with a primary diagnosis of AF from 2009 to 2014.

“Currently most ED doctors do not initiate chronic medications like oral anticoagulation for stroke prevention, because if you initiate a chronic medication for a patient, you are, from a medicolegal and ethical perspective, somewhat obliged to follow them up a few days or weeks later to see if the dosage is correct, if the side effects are manageable, etc,” Atzema told Healio. “But as ED doctors, we rarely see our patients again. And often we can't guarantee that the patient can get in to see their longitudinal provider in a timely way after we initiate the medication, or that they even have one. If we could assess ED oral anticoagulation prescribing rates over time, we could use the rates for feedback to EDs on their prescribing rates, as well as to monitor the effect of interventions that are meant to increase ED oral anticoagulant prescribing, etc. But right now, very few of these huge health databases collect or record ED prescriptions. So, the algorithm allows us to identify ED oral anticoagulant prescriptions for the first time.”

Researchers tested 32 algorithms that fell into one of the following categories:

  • prescriptions filled on the same day that the patient left the ED;
  • the number of days following hospital discharge when a patient filled the prescription;
  • prescription fills that were excluded if they were done after an outpatient visit with a relevant clinician;
  • excluding patients with a history of venous thromboembolism; and
  • patients that were removed for seeing a relevant clinician 30 days before the ED visit.

Sensitivity substantially improved when day 0 was included in the algorithm, or the day the patient left the ED. Algorithm 3 — prescription filled in 1 to 3 days — identified 34% of patients compared with 83% of patients in algorithm 2, which identified patients who filled prescriptions between 0 and 3 days.

Longer follow-up for prescription fills lowered specificity, as shown in algorithm 2 compared with algorithm 10, which focused on prescription fills between 0 and 30 days (91% vs. 74%, respectively) with a slight improvement in sensitivity (83% to 88%).

The number of false positives decreased after restricting patients who filled a prescription before seeing another clinician. As a result, specificity improved from 74% in algorithm 10 to 97% in algorithm 13 (prescription filled 0 to 30 days before any appointment with another clinician). This also led to a decrease in sensitivity from 88% to 68%. There was a further increase in specificity when patients with an outpatient visit before the ED were removed (99%), but this also led to decreased sensitivity (24%).

Algorithm 2 (prescription fill between 0 and 3 days) maximized specificity (91%; 95% CI, 90-92) and maintained sensitivity (83%; 95% CI, 79-87). The positive predictive value was 70% (95% CI, 66-74) with a negative predictive value of 96% (95% CI, 95-97). High specificity was observed in algorithm 13 (97%; 95% CI, 96-98).

“The ultimate goal is to have more patients in the community who should be on stroke prevention therapy, ie, oral anticoagulation, actually on it,” Atzema said in an interview. “The ED is a place these patients go often — as would you if your heart suddenly started racing, you had chest pain or you had shortness of breath, the top three presenting symptoms for patients with AF in the ED — but it is an untapped health care setting in terms of oral anticoagulant initiation. That said, this specific work won't change anyone's clinical practice in the ER tomorrow.” – by Darlene Dobkowski

Disclosures: Atzema reports no relevant financial disclosures. Please see the study for all other authors’ relevant financial disclosures.