Methods

In December 2013, a steering committee of 15 physicians (6 cardiologists, 3 neurologists, 1 hematologist, and 5 primary care physicians) and 1 pharmacist met to develop the protocol. Participating primary care physicians, known to be managing patients with AF, completed data collection from February to April 2014. Eligible patients were required to be aged ≥18 years and with documented intermittent or permanent AF. Patients were excluded if they had a significant heart valve disorder (i.e., prosthetic valve, hemodynamically significant valve disease), clinically significant hepatic disease (e.g., active hepatitis), or reversible cause of AF (e.g., recent cardiac surgery, pulmonary embolus, and untreated hyperthyroidism).

Information was collected using a 1-page data collection form completed on paper or online. Collected information included patient demographics, AF history, medical history, stroke and bleed risk assessment and how each was calculated, and details of current treatment, including type of antithrombotic, who initiated treatment, most recent INR values for subjects taking warfarin and rationale for not initiating anticoagulation, if appropriate. Details on data management can be found in the Appendix .

Descriptive analysis of variables pertinent to AF management was performed. Calculations were done to estimate the stroke and bleed risk; CHADS ₂ (congestive heart failure, hypertension, age >75 years, diabetes, previous stroke/transient ischemic attack/noncentral nervous system thromboembolism [doubled]) and HAS-BLED (hypertension, abnormal renal/liver function, stroke, bleeding history, labile INR, elderly, drugs/alcohol) scores are reported in this study. Calculated values were compared to the physician reported estimates, using the κ statistic, and data on how frequently these values were in agreement were reported. The use of OAC in accordance with Canadian guidelines and product monograph was examined to determine appropriate use and dosing. Estimates of the time in TTR, using the Rosendaal method, were determined based on ≤10 of the most recent INR values for each patient on warfarin.

Results

A total of 735 physicians participated in the program and completed 7,019 patient forms. The median patient age was 78 and median time since diagnosis of AF was 5.0 years. 41.9% of patients were women and 57.9% were over age 75. Detailed patient demographics, and medical history, are found in Tables 1 and 2 .

The mean CHADS ₂ score in the patient group was 2.0; the mean HAS-BLED score was 1.4. Physicians tended to underestimate stroke and overestimate bleed risk compared with calculated risk scores ( Table 3 ). Details of anticoagulant use by CHADS ₂ score can be found in Table 4 .

Of the 89.4% of patients on OAC therapy, 53.6% were on warfarin, 18.5% on dabigatran, 21.3% on rivaroxaban, and 5.6% on apixaban ( Table 5 ). Canadian approved dosing for dabigatran, rivaroxaban, and apixaban (which differ from US and European labels) are listed in Supplementary Table 1 ; creatinine clearance (CrCl) minimums for the use are 30, 30, and 25 ml/min, respectively.

Of the patients on NOACs, 7.4% were underdosed, whereas overdosing was seen in 4.3%. CrCl below recommended was seen in 3.2% of patients on NOACs. Of those, 16.5% were on the higher available dose. Inappropriate use by agent is seen in Table 6 . Lower use of the NOAC, apixaban, is likely due to its approved for use in AF almost 2 years later than dabigatran and rivaroxaban and at the time of the audit did not have reimbursement in most provinces.

Warfarin was used in 53.6% of patients on anticoagulants, a reduction of 3.9% from the 2013 audit (p <0.001) ; 8.2% of patients on warfarin had a CrCl <30 ml/min. Prescribers listed provincial reimbursement as a factor in anticoagulant choice in 29.5% and adherence in 9.9% of these patients. Although only 3.4% of patients were managed at AF clinics, no difference was noted in the use of anticoagulation for patients treated in that setting compared with the overall population; (Relative risk [RR] of no anticoagulation 0.82, 95% CI 0.54 to 1.25; p = 0.36) however, greater use of warfarin was noted (RR of warfarin vs NOAC use 1.41, 95% CI 1.30 to 1.53; p <0.0001, Supplementary Table 2 ). We compared the bleeding and stroke risks of NOAC versus warfarin patients to assess if risk level influenced choice of warfarin or NOAC. Supplementary Table 3 compares the HAS-BLED and CHADS ₂ scores in both groups. Both higher CHADS ₂ and HAS-BLED scores were associated with the use of warfarin over NOAC. A higher HAS-BLED score (Phi = 0.202, p <0.001) was more strongly associated than a higher CHADS ₂ score (Phi = 0.072, p <0.001) with the use of warfarin over the NOAC. In the warfarin group, the median INR was 2.5, with a median TTR of 75.9%. Although the median TTR was above 70, 19.6% of patients had a TTR of 50% to 69%, and 18.6% had a TTR below 50. We examined antithrombotic use by the type of AF. Paroxysmal AF was associated with less use of any anticoagulant compared with persistent AF (RR of no anticoagulation 2.62, 95% CI 2.21 to 3.10; p <0.001, Supplementary Table 4 ).

We identified 475 patients (7.6% of those on any anticoagulant) on acetylsalicylic acid (ASA) plus an anticoagulant. Of these, 29.3% had no history of ischemic stroke, transient ischemic attack, peripheral arterial disease, or coronary artery disease.

A total of 10.4% of all patients were not receiving anticoagulant therapy. Of these, 15.7% refused therapy and 51.1% were receiving ASA only; 4.4% of the total population received no antithrombotic therapy, of which 61.3% were CHADS ₂ 1 or greater; 5.3% of the total population was receiving ASA only, of which 36.2% were CHADS ₂ ≥2. The use of ASA monotherapy according to CHADS ₂ scores are found in Supplementary Table 5 .

Physicians indicated that compliance concerns affected their choice of treatment in 15.9% of patients. In patients taking warfarin, 9.9% were prescribed the drug because of adherence concerns. Adherence was also a concern in 6.7% of patients on no anticoagulant treatment.