Quality of Warfarin Management in Canadian Primary Care

Student Researcher:
Sharon (Yi Hsuan) Liu

Supervisor / Principle Investigator:
Dr. Scott Garrison

Additional Authors:
Dr. Michael Allan
Neil Drummond, PhD
Dr. Michelle Greiver, MSc, MD, CCFP
Michael Kolber, MD, MSc
Christina Korowynk
Donna P Manca, MD, MCLSC
Finlay McAlister
Bill Peeler
Dr. Alexander Singer

MD Class of 2020


It has been suggested that well controlled warfarin patients are unlikely to stay well controlled, and that warfarin control is suboptimal in primary care. Objective: To determine the quality and stability of warfarin anticoagulation in a diverse population of Canadian primary care patients and providers. Design: Prospective cohort study. Setting: 935 community primary care practices contributing electronic medical records data to the Canadian Primary Care Sentinel Surveillance Network (CPCSSN). Participants: 18,303 adult warfarin users with ≥ 7 INR readings within 8 weeks of another reading excluding the first 5 (to avoid the period of initial warfarin titration). Patients were 46.6% female and 71.0 years of age on average. Where indication was known, atrial fibrillation was present in 89.7%, venous thromboembolism in 13.5%, and mechanical valves in 5.0%. Oustcome Measures: Proportion of time in therapeutic INR range (TTR) using Rosendaal’s method and assuming a 2.0-3.5 INR target range spanning all indications and matching the range of lowest risk from observational studies. Proportion of TTR > 75% (good control). Proportion of TTR < 60% (poor control). Proportion of patients with good control during their first year on warfarin who have poor control the following year. Proportion of INR readings < 2.0, > 3.0, and >3.5. Results: Using the 2.0–3.5 target INR range, median TTR was 77.4% (IQR 64.6%-86.4%). TTR was > 75% in 56.0% of patients and < 60% in 19.3%. Of the 63.1% with good control in their first year of anticoagulation, only 10.2% had poor control the following year. The proportion of readings potentially out of range were <2 (27.9%),> 3.0 (19.4%), and >3.5 (8.6%). Conclusions: Warfarin users managed in primary care have INR values within therapeutic range on a median of 77.4% of the time, comparable to TTR in randomized trials where INR is considered well controlled. Furthermore, good INR control predicts future INR stability. These results meaninfully inform decisions to switch established warfarin users onto newer agents.