The objectives of this national chart audit (January to June 2013) of 6,346 patients with atrial fibrillation (AF; ≥18 years without a significant heart valve disorder) from 647 primary care physicians were to (1) describe the frequency of stroke and bleed risk assessments in patients with nonvalvular AF by primary care physicians, including the accuracy of these assessments relative to established predictive indexes; (2) outline contemporary methods of anticoagulation used; and (3) report the time in the therapeutic range among patients prescribed warfarin. An annual stroke risk assessment was not undertaken in 15% and estimated without a formal risk tool in 33%; agreement with CHADS 2 score estimation was seen in 87% of patients. Major bleeding risk assessment was not undertaken in 25% and estimated without a formal risk tool in 47%; agreement with HAS-BLED score estimation was observed in 64% with physician overestimation in 26% of patients. Antithrombotic therapy included warfarin (58%), dabigatran (22%), rivaroxaban (14%), and apixaban (<1%). Among warfarin-treated patients, the median international normalized ratio was 2.4 and time in therapeutic range (TTR) was 73%; however, the TTR was <50% in 845 (25%), 50% to 69% in 674 (20%), and ≥70% in 1,827 (55%) patients. In conclusion, we describe a contemporary real-world elderly population with AF at important risk for stroke. There is apparent overestimation of bleeding risk in many patients. Warfarin was the dominant stroke prevention treatment; however, the suggested TTR target was achieved in only 55% of these patients.
The likelihood of stroke in nonvalvular patients with AF is not homogeneous. Established clinical risk factors have been used to formulate stroke risk stratification schemes, which in turn can identify those patients who are at highest risk and would, therefore, benefit from antithrombotic therapy. The most commonly used tool is the CHADS 2 (congestive heart failure, hypertension, age >75 years, diabetes, prior stroke/transient ischemic attack/noncentral nervous system thromboembolism [doubled]) score. Additionally, the risk of major bleeding has been increasingly recognized as an important factor when determining the optimal stroke prevention strategy. Several bleeding risk stratification schemes have been developed, most notably the HAS-BLED (hypertension, abnormal renal/liver function, stroke, bleeding history, labile international normalized ratio, elderly, drugs/alcohol) score. Although current Canadian AF guidelines recommend that all patients with AF be stratified using a predictive index for stroke and risk of bleeding, the frequency with which such risk stratification (with or without the use of a validated predictive score) is undertaken in clinical practice is uncertain. Thus, as part of the Co-ordinated National Network to Engage Physicians in the Care and Treatment of Patients With Atrial Fibrillation (CONNECT AF) quality enhancement initiative and knowledge translation program, we sought to (1) describe the frequency of stroke and bleed risk assessments by primary care physicians, including the accuracy of these assessments relative to established predictive indexes; (2) outline contemporary methods of anticoagulation used for nonvalvular patients with AF; and (3) report the time in therapeutic range (TTR) among patients prescribed warfarin.
Methods
The goals of the CONNECT AF program were to (1) help identify practice patterns and compare performance in AF care with both Canadian primary care physician peers and guideline recommendations; (2) provide knowledge and practice-based content through educational activities to address self-identified needs and practice-identified care gaps; and (3) offer tools to evaluate and potentially alter stroke prevention management in clinical practice.
The CONNECT AF was a quality assurance and educational initiative targeting Canadian primary care physicians and their nonvalvular patients with AF. Although CONNECT AF was not a formal study, central Research Ethics Board approval was sought to waive the need for individual patient consent and to allow for publication of aggregate data.
A steering committee comprising 12 physicians (8 cardiologists, 1 nephrologist, and 3 primary care physicians) and 1 pharmacist was convened through the Canadian Heart Research Center (CHRC). The committee developed patient chart audit forms, educational tools, and assisted with the provision of relevant resources for use by participating physicians.
Participating physicians submitted data for 10 patients with AF from their practice. Participation in the interactive continuing medical education/knowledge translation aspect of the program followed the data submission component.
Patient eligibility criteria included age ≥18 years with documented paroxysmal, permanent, or persistent AF and without (1) a significant heart valve disorder (i.e., prosthetic valve, hemodynamically significant, or severe valve disease), (2) clinically significant hepatic disease (e.g., active hepatitis), or (3) a reversible cause of AF (e.g., recent cardiac surgery, pulmonary embolus, untreated hyperthyroidism).
Information collected for each patient included demographics, AF history, medical history, stroke risk assessment and method by which it was determined (including, but not restricted to, CHADS 2 and CHA 2 DS 2 -VASc [congestive heart failure, hypertension, age >75 years, diabetes mellitus, stroke, vascular disease, age 65 to 74 years, sex category] criteria ), bleed risk assessment and method by which it was determined (including, but not restricted to, HAS-BLED criteria ), current antithrombotic therapy, and most recent INR values (if on warfarin).
This information was gathered by means of a single-page chart audit form. The completed forms were scanned using TELEform (version 10.0; Cardiff Software Inc., San Diego, California), and data were stored in an electronic database at the CHRC.
The steering committee trained 100 physicians with experience and expertise in stroke prevention management of patients with AF (cardiologists, hematologists, neurologists, internists, and family doctors) to serve as regional faculty for both live and Web-based education and feedback sessions. At these sessions, which followed the chart review, participating physicians were educated about risk stratification and appropriate antithrombotic use in patients with AF through a case-based learning approach. Thereafter, aggregate data for local and national practice were reviewed.
As part of the chart audit, descriptive analysis of demographic variables, risk score calculations, and other variables pertinent to AF management was performed. Point prevalence data are presented. Continuous variables are summarized as median (25th and 75th percentiles), and discrete variables are reported as counts and percentages of known nonmissing cases.
Calculations were performed to determine the estimated stroke and bleed risk, using the CHADS 2 , CHA 2 DS 2 -VASc, and HAS-BLED indexes. These values were then compared with the physician-reported risk estimates using the kappa statistic, and data on how frequently these values were in agreement were reported.
Using up to 6 of the most recently available INR values, estimates of the TTR were established using the method by Rosendaal et al, with the therapeutic range defined as an INR range from 2 to 3. All analyses were performed using SAS software, version 9.2 (SAS Institute Inc., Cary, North Carolina).
Results
A total of 647 Canadian physicians provided data on 6,346 patients with AF from 10 provinces ( Supplementary Table 1 ) from January to June 2013. Patient demographics, AF details, and medical history are presented in Table 1 . The typical patient studied was an elderly men with hypertension who had AF for the past 5 years.
| Variable | |
|---|---|
| Age (years) ∗ | 78 (70, 84) ∗ |
| >75 years | 3888 / 6316 (62%) |
| Female | 2622 / 6310 (42%) |
| Weight (kg) ∗ | 80 (68, 92) ∗ |
| ATRIAL FIBRILLATION – TYPE AND PRIOR TREATMENT | |
| Years since first diagnosed ∗ | 5 (3, 10) ∗ |
| Paroxysmal | 1897 / 5894 (32%) |
| Persistent | 1789 / 5894 (30%) |
| Permanent / Accepted | 2226 / 5894 (38%) |
| Prior Cardioversion | 904 (14%) |
| Ablation | 235 (3.7%) |
| Prior Anti-Arrhythmic Drug(s) | 2119 (33%) |
| CURRENT RHYTHM | |
| Atrial Fibrillation | 4198 / 5726 (73%) |
| Atrial Flutter | 113 / 5726 (2%) |
| Sinus Rhythm | 1354 / 5726 (24%) |
| Other | 61 / 5726 (1.1%) |
| MEDICAL HISTORY | |
| Prior Stroke | 717 (11.3%) |
| Hemorrhagic | 52 / 717 (7.3% of those with prior stroke) |
| Prior Transient Ischemic Attack | 719 (11%) |
| Prior non-central nervous system thromboembolism | 83 (1.3%) |
| Hypertension | 4098 (65%) |
| Diabetes Mellitus | 1781 (28%) |
| Coronary Artery Disease | 1883 (30%) |
| Peripheral Arterial Disease | 501 (8%) |
| Prior Heart Failure / Left Ventricular Dysfunction | 1287 (20%) |
| Prior Bleed | 393 (6.2%) |
| Major | 166 / 393 (42%) |
| Minor | 215 / 393 (55%) |
| Liver Disease | 74 (1.2%) |
| Creatinine > 200 umol/L | 86 / 5873 (1.5%) |
| Estimated glomerular filtration rate < 60 ml/min | 1972 / 5464 (36%) |
| Dialysis Dependent | 23 (0.4%) |
| Alcohol Consumption (> 10 drinks / week for females, > 15 drinks / week for males) | 461 (7.3%) |
The median CHADS 2 score was 2 (1, 3), and the median CHA 2 DS 2 -VASc score was 4 (3, 5). The median HAS-BLED score was 2 (1, 3).
Primary care physicians identified that 4,969 specialist consultants were also involved in the AF/stroke prevention management in a total of 4,254 (67%) patients, including cardiology 3,368 (53%), internal medicine 1,012 (16%), anticoagulation clinic 409 (6.5%), neurology 134 (2.1%), and hematology 46 (0.7%). The current stroke prevention treatment approach was either initiated or recommended by a specialist in 4,033 (64%) patients.
Primary care physicians estimated a 2% per year or greater risk of stroke in 4,545 patients (72%), <2% per year risk in 858 patients (14%), and did not know the risk in 850 patients (13%). In 93 cases (1.5%), the data were missing. In contrast, a CHADS 2 stroke risk of ≥2% (i.e., CHADS 2 score 1 to 6) was calculated for 5,830 patients (92%) and <2% per year (i.e., CHADS 2 score of 0) for 516 patients (8.1%; Table 2 ). However, in those patients with a CHADS 2 score of 0, 118 (23%) had a CHA 2 DS 2 -VASc score ≥2, which translates to a ≥2% risk of stroke per year.
| Median Score (25 th , 75 th percentiles) | 2 (1, 3) |
|---|---|
| CHADS 2 Score | N |
| 0 | 516 (8.1%) |
| 1 | 1484 (23%) |
| 2 | 2130 (34%) |
| 3 | 1230 (19%) |
| 4 | 679 (11%) |
| 5 | 260 (4.1%) |
| 6 | 47 (0.7%) |
The physician-established stroke risk was calculated, documented, or on file in 3,356 patients (68%), whereas stroke risk was estimated without using a formal risk score in 1,617 patients (32%). In those patients whose CHADS 2 score was unknown by the physician, a calculated risk of <2% per year was determined for 72 patients (8.5%) and a calculated risk of ≥2% per year was determined in 778 patients (92%). In those patients who had a missing risk estimate response, 7 patients (7.5%) fell into the low-risk group (<2% per year) and 86 patients (93%) fell into the higher risk group (≥2% per year).
When comparing physician-estimated stroke risk with the CHADS 2 score estimates, there was agreement in 4,680 cases (87%; kappa 0.37 [95% CI 0.34 to 0.41]), apparent underestimation (compared with CHADS 2 calculated risk) occurred in 572 (11%), and overestimation in 151 patients (2.8%).
Major bleed risk was determined by physicians to be ≥2% per year in 2,038 patients (32%) and <2% per year in 2,700 patients (43%); the risk was unknown by the physician in 1,442 patients (23%). In 166 patients (2.6%), the data were missing. The HAS-BLED–derived major bleed risk was ≥2% per year (i.e., HAS-BLED score 3 to 9) for 1,679 patients (27%) and <2% per year (i.e., HAS-BLED score 0 to 2) for 4,667 patients (73%; Table 3 ).
| Median Score (25 th , 75 th percentiles) | 2 (1, 3) |
|---|---|
| HAS-BLED Score | N |
| 0 | 407 (6.4%) |
| 1 | 2069 (33%) |
| 2 | 2191 (35%) |
| 3 | 1198 (19%) |
| 4 | 398 (6.3%) |
| 5 | 69 (1.1%) |
| 6 | 12 (0.2%) |
| 7 | 1 (0.02%) |
| 8, 9 | 1 (0.02%) |
In 4,407 cases, major bleed risk was established by the physician using a calculated or documented risk stratification score estimate in 2,359 (54%) and estimated or guessed in 2,048 cases (46%).
In those patients whose HAS-BLED score was unknown, a calculated risk of <2% per year was determined for 1,068 patients (74%) and a calculated risk of ≥2% per year was determined in 374 patients (26%). In the missing data set, 119 patients (72%) were in the low-risk group (score 0 to 2) and 47 patients (28%) were in the high-risk group (score 3 to 9).
When comparing physician-estimated bleeding risk with the HAS-BLED score estimates, there was agreement in 3,040 cases (64%; kappa 0.23 [0.21 to 0.26]) with apparent overestimation compared with HAS-BLED calculated risk in 1,239 (26%) and underestimation in 459 patients (9.7%).
Physicians reported that 5,925 patients (93%) were currently receiving anticoagulation ( Table 4 ): warfarin (58%), dabigatran (22%), rivaroxaban (14%), or apixaban (<1%).
| Anticoagulation Therapy | N |
|---|---|
| Warfarin | 3656 (58%) |
| Apixaban (mg BID) | 18 (0.3%) |
| 5 | 14/18 (78%) |
| 2.5 | 4/18 (22%) |
| Dabigatran (mg BID) | 1413 (22%) |
| 150 | 666/1361 (48.9%) |
| 110 | 688/1361 (50.6%) |
| 75 | 7/1361 (0.5%) |
| Rivaroxaban (mg OD) | 860 (14%) |
| 20 | 555/815 (68%) |
| 15 | 260/815 (32%) |
| Patient Not Treated with anticoagulation therapy | 421 (6.6%) |
Of the 3,656 patients (58%) on warfarin, the median INR was 2.4 (2.0, 2.7) and the median TTR was 73% (49, 92). The TTR was <50% in 845 (25%), 50% to 69% in 674 (20%), and ≥70% in 1,827 patients (55%).
Among the patients on apixaban 2.5 mg twice daily (22% of patients on this agent), dabigatran 110 mg twice daily (50%), or rivaroxaban 15 mg daily (32%), most had a potential reason identified by their primary care physician for lower dose therapy such as older age (e.g., ≥75 years), moderate renal dysfunction (e.g., estimated creatinine clearance 30 to 45 ml/min), anemia, or concomitant antiplatelet therapy ( Supplementary Table 2 ). Furthermore, 20% to 30% of these patients had a HAS-BLED score of ≥3.
Antiplatelet use is listed in Table 5 : aspirin (20%), clopidogrel (2.1%), and dual antiplatelet therapy (0.8%). Combination anticoagulation and antiplatelet therapy was used in 1,033 patients (16%).
| Antiplatelet Therapy | N |
|---|---|
| Aspirin | 1250 (20%) |
| Clopidogrel | 130 (2.1%) |
| Prasugrel | 1 (0.02%) |
| Ticagrelor | 2 (0.03%) |
| Combination Antiplatelet therapy | 49 (0.8%) |
Physicians indicated that, according to their assessment of patient stroke risk, the appropriate stroke prevention strategy would be an oral anticoagulant in 5,665 (89%), antiplatelet therapy in 587 (9.3%), and no therapy in 114 patients (1.8%).
Tables 6 and 7 show the use of antithrombotic and anticoagulation therapy, respectively, according to calculated CHADS 2 score.
