Use of thromboprophylaxis guidelines and risk stratification tools in atrial fibrillation: A survey of general practitioners in Australia

Abstract Rationale and Objectives Clinical guidelines produced by cardiology societies (henceforth referred to simply as ‘clinical guidelines’) recommend thromboprophylaxis with oral anticoagulants (OACs) in patients with atrial fibrillation (AF) who have moderate‐to‐high stroke risk. However, deviations from these recommendations are observed, especially in the primary healthcare setting. The primary aims of this study were to evaluate the self‐reported use of AF clinical guidelines and risk stratification tools among Australian general practitioners (GPs), and their perceptions regarding the available resources. Method We conducted an online survey of Australian GPs. Descriptive statistics were used to summarise the findings. Results Responses from 115 GPs were included for analysis. Respondents reported various ways of accessing thromboprophylaxis‐related information (n = 113), including clinical guidelines (13.3%), ‘Therapeutic Guidelines©’ (37.2%) and Royal Australian College of General Practitioners websites (16.8%). Of those who reported reasons against accessing information from clinical guidelines (n = 97), the most frequent issues were: too many AF guidelines to choose from (34.0%; 33/97), different guidelines for different diseases (32.0%; 31/97), time‐consuming to read guidelines (21.6%; 21/97), disagreements between different guideline recommendations (20.0%; 19/97), conflict with criteria for government subsidy (17.5%; 17/97) and GPs' busy schedules (15.5%; 15/97). When assessing patients' risk of stroke (n = 112) and bleeding (n = 111), the majority of the respondents reported primarily relying on a formal stroke risk (67.0%) and bleeding risk (55.0%) assessment tools, respectively. Respondents reported using formal stroke and bleeding risk assessment tools mainly when newly initiating patients on therapy (72.4%; 76/105 and 65.3%; 65/101, respectively). Conclusion Among our small sample of Australian GPs, most did not access thromboprophylaxis‐related information directly from AF‐specific clinical guidelines developed by cardiology societies. Although the majority reported using formal stroke and bleeding assessment tools, these were typically used on OAC initiation only. More focus is needed on formal risk reassessment as clinically indicated and at regular review.

developed by cardiology societies. Although the majority reported using formal stroke and bleeding assessment tools, these were typically used on OAC initiation only. More focus is needed on formal risk reassessment as clinically indicated and at regular review.

| INTRODUCTION
Atrial fibrillation (AF) is the most frequently diagnosed cardiac arrhythmia in clinical practice. 1 According to the Global Burden of Diseases estimates, the global prevalence of AF was 38 million in 2017. 2 AF is one of the most commonly managed chronic illnesses in general practice in Australia. 3 It is associated with an increased burden of stroke and systemic embolism. 2 The risk of stroke associated with AF, especially in the absence of moderate-to-severe mitral stenosis or mechanical heart valve, depends mainly on the patient's age and the presence or absence of other commonly concomitant diseases. [4][5][6] Oral anticoagulants (OACs) are typically used in patients with AF for the prevention of thromboembolic events, mainly stroke. They are associated with a 70% relative reduction in the risk of stroke. 4 Clinical guidelines produced by cardiology societies worldwide, including the National Heart Foundation of Australia (NHFA) and the Cardiac Society of Australia and New Zealand (CSANZ), recommend thromboprophylaxis with OACs in patients with moderate-to-high risk of stroke. [4][5][6] In addition, assessment of the risk of bleeding is recommended with the aim of identifying and addressing modifiable risk factors for bleeding. [4][5][6] Thromboprophylaxis that adheres to recommendations based on a formal stroke risk assessment tools such as the CHA 2 DS 2 -VASc score is associated with better treatment outcomes. 7 However, deviations from these recommendations, primarily undertreatment, are observed. 8 Recent Australian studies have reported that 19%-37% of hospitalised AF patients at high risk of stroke did not receive OAC therapy, 9,10 while the rate of nonprescribing in high-risk patients in general practices has been reported at 35%-45%. 11,12 Several factors contribute to OAC undertreatment in AF. Most of the previous studies that reported factors associated with thromboprophylaxis undertreatment in AF were based on findings from retrospective studies and largely outside of the general practice setting. 7,10,[13][14][15] Nonetheless, a previous review article identified that prescriber-related factors, including their beliefs and practice patterns, were among the major contributors. 8 Previous qualitative studies reported that more emphasis is given to bleeding than stroke prevention when prescribing OACs, although with limited use of formal bleeding risk assessment tools by prescribers. [16][17][18] Also, the proportion of prescribers, including general practitioners (GPs, known elsewhere as 'primary care physicians'), cardiologists and neurologists, who use formal bleeding risk assessment tools seems to correspond with the proportion of those who use formal stroke risk assessment tools. 19 Uncertainties in prescribers' knowledge and skills in calculating and applying stroke and bleeding risk assessment in AF have been documented elsewhere. 19 Findings from a study conducted in nine European countries indicate that GPs use different local/national and international guidelines in management of their patients with AF, yet 21% of the small sample of 212 GPs reported not following any specific clinical guidelines. 20 A recent exploratory qualitative study that conducted semi-structured interviews among GPs in Western Australia identified the decision-making process as a key reason for deviations from thromboprophylaxis guidelines by the study participants, with limited use of clinical guidelines and complexities in balancing risk versus benefit of thromboprophylaxis. 16 In addition to the stroke and bleeding risks, older age, dementia, frailty and falls risk were reported to influence decisions leading to deviations from guideline-recommended thromboprophylaxis. 16 Supporting data on where Australian GPs access information regarding thromboprophylaxis in AF and how different factors contribute to their decision-making process are limited. Therefore, understanding GPs' sources of information upon which they base decisions and the different weights GPs ascribe to various factors in their thromboprophylaxis decision-making process is important.
Our primary aims were to evaluate the self-reported use of AF clinical guidelines produced by cardiology societies and risk stratification tools among Australian GPs, and their perceptions regarding the available resources. Our secondary aim was to assess the weightings ascribed by GPs to factors affecting the thromboprophylaxis decision-making process in patients with AF in Australian general practice.

| Study design
We conducted an online survey among GPs practising in general practices in Australia. A questionnaire was developed based on a review of the literature 8,21 and a previous exploratory qualitative study by the researchers. 16  Multiple strategies were used to recruit respondents, including advertising via professional websites, newsletters and social media (LinkedIn, Twitter and Facebook), and through direct contact, medical practices and professional organisations. In addition, we contracted a commercial company to recruit respondents through a targeted approach (i.e. only those potential respondents who work in healthcare/medical industry). Anonymous data were collected from May 2021 to November 2021 using an online survey management platform, Qualtrics (Qualtrics International Inc.). We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement in reporting this study (Supporting Information). 23

| Eligibility
To take part in the survey, respondents had to be a GP practising in a general practice setting in Australia, and consent to take part in the survey. Respondents who accessed the survey from outside Australia [based on their internet protocol (IP) address], or who only completed the sociodemographic questions, or whose responses were illogical (e.g., respondents with zero years of experiences as a GP; and respondents aged 30-39 years with a 31 years' experience as a GP) were excluded.

| Data analysis
Descriptive statistics were used to summarise the findings. Continuous variables were expressed as median and interquartile range (IQR) and categorical variables as frequencies and percentages. A χ 2 was performed to investigate whether there was association between primarily (i.e., either 'entirely' or 'mainly') relying on formal stroke risk assessment tools and primarily (i.e., either 'entirely' or 'mainly') relying on formal bleeding risk assessment tools. All statistical analyses were performed using SAS version 9.4 (SAS institute Inc.).

| Ethics
The study was approved by the University of Western Australia Human Research Ethics Committee (RA/4/20/6366).  (Table 1).     | 487 stroke risk and bleeding risk assessment tools, with those who primarily relied on formal tools to assess stroke risk more likely to also rely on formal tools to assess bleeding risk (χ 2 = 46.1, 1 DF, p < 0.001).  and bleeding risk, as they change over time because of patients' age and other risk factors. 28 Recent data from the Australian general practice setting indicated that one-third of patients whose stroke risk changed from low-to-moderate to high were not prescribed OAC therapy. In the remaining patients who received OAC therapy, OAC initiation was reported to be delayed by a median of 2 years, which suggested a need for more frequent stroke risk reassessments. 12 This is also true with bleeding risk reassessments. In addition to identifying and addressing modifiable bleeding risk factors and reducing bleeding events, more frequent bleeding risk assessment could help with higher prescription of OACs in eligible patients with AF. 29 The proportion of respondents that reported relying on formal bleeding risk assessment tools as the primary means of assessing patients' risk of bleeding was numerically lower than those who reported using formal stroke risk assessment tools to assess patients' risk of stroke (55.0% vs.

| The weight of different factors in thromboprophylaxis prescribing
66.9%). This is consistent with previous studies, which have shown that GPs tend to use stroke risk stratification tools more often than they use bleeding risk assessment tools. 8,16 Importantly, not primarily relying on formal bleeding risk assessment tools was more common among those who also do not primarily rely on formal stroke risk assessment tools. The inconsistent use of stroke and bleeding risk stratification tools means that prescription of OACs in eligible patients may vary among GPs. There is a recognised disparity in OAC prescribing between general practices in Australia: after ranking prescribing rates into quintiles (five equal parts), prescribing in patients with AF who have moderate-to-high risk of stroke was 65.6% in the highest practice site quintiles while this figure is only 38.6% in the lowest practice site quintiles. 11 Therefore, any efforts to improve the use of formal bleeding risk assessment tools should also take improving the use of formal stroke risk assessment tools into consideration. Understanding the reason for these differences may assist in improving the use of OAC prescribing in Australian general practice.
Less than one in six respondents ( Therefore, updating the PBS criteria to be in line with the guideline recommendations is warranted to minimise another potential source of confusion for GPs.

| Strengths and limitations
Interpretation of the findings of the study should consider the potential limitations of this study. Considering more than 37,000 GPs practise across Australia, 24 the sample size was small making any generalisations and inferences difficult. Also, because this was an anonymous, online survey, it was not possible to calculate the response rate of respondents, as the denominator was not able to be determined. Despite these limitations, the study provides insight into sources of information used by GPs when prescribing thromboprophylaxis in AF, their reasons for not accessing such information from AF clinical guidelines, and the weight they give to different important factors in their thromboprophylaxis decisionmaking, which may prove useful in developing future strategies to ensure consistent, high-quality thromboprophylaxis for all Australians with AF.