Article Text

PDF

Original research article
The treatment of paroxysmal atrial fibrillation in UK primary care
  1. Andrea Isaew,
  2. Nicola Jaime Adderley,
  3. Ronan Ryan,
  4. David Fitzmaurice,
  5. Tom Marshall
  1. Institute of Applied Health Research, University of Birmingham, Birmingham, UK
  1. Correspondence to Dr Nicola Jaime Adderley, Institute of Applied Health Research, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK; n.j.adderley{at}bham.ac.uk

Abstract

Objective To determine whether patients with paroxysmal atrial fibrillation (AF) are less likely to be treated with anticoagulants than patients with persistent/permanent AF and to investigate trends in treatment between 2000 and 2015. UK and European guidelines recommend that anticoagulants are offered to all patients with AF at increased risk of stroke, irrespective of AF type.

Methods Sixteen sequential cross-sectional analyses from 2000 to 2015 were carried out with index dates on 1st of May each year. The data source was primary care data from 648 practices across the UK contributing to The Health Improvement Network database. All patients with a diagnosis of AF aged ≥35 years and registered for at least 1 year were included. The main outcome measure was prescription of anticoagulant medication.

Results The proportion of patients with AF with a diagnosis of paroxysmal AF increased from 7.4% (95% CI 7.0 to 7.8) in 2000 to 14.0% (95% CI 13.7 to 14.3) in 2015. Among patients with a CHADS2 score of ≥1, between 2000 and 2015 the proportion prescribed anticoagulants increased from 18.8% (95% CI 16.4 to 21.4) to 56.2% (95% CI 55.0 to 57.3) and from 34.2% (95% CI 33.3 to 35.0) to 69.4% (95% CI 68.9 to 69.8) in patients with paroxysmal and other (persistent/permanent) AF, respectively; RR for treatment of patients with paroxysmal AF compared with patients with other AF increased from 0.48 (95% CI 0.42 to 0.55) to 0.76 (95% CI 0.74 to 0.77). Adjusting for age, sex, Townsend score and presence or absence of contraindications had little effect on the results.

Conclusions In 2000, eligible patients with paroxysmal AF were half as likely to be treated with anticoagulants as patients with other AF; this has improved over time, but in 2015, eligible patients with paroxysmal AF were still around 20% less likely to be prescribed anticoagulant medication.

  • Atrial fibrillation
  • Cardiac arrhythmias and resuscitation science
  • DISEASES
  • PRIMARY CARE
  • QUALITY OF CARE AND OUTCOMES
View Full Text

Statistics from Altmetric.com

Introduction

Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia and a major global public health problem. It is associated with a fivefold increase in risk of stroke, increased incidence of congestive heart failure and higher mortality.1

Recurrent AF is either labelled ‘paroxysmal’ when it is self-terminating or ‘persistent’ when it lasts more than 7 days or is terminated earlier using either pharmacological means or direct current cardioversion; AF is defined as permanent when cardioversion fails to restore normal heart rhythm.2 Evidence varies regarding the similarity in stroke risk between patients with paroxysmal, persistent and permanent AF: some studies have shown comparable stroke risk, while others suggest that risk may be lower in patients with paroxysmal AF.3–8 Nevertheless, patients with paroxysmal AF remain at an elevated risk of stroke relative to patients without AF. Within 1.5 years, over a fifth of patients with paroxysmal AF progress to permanent AF.9 10

Prophylactic treatment of AF with anticoagulants reduces risk of stroke by approximately two-thirds.11–15 Current UK and European guidelines recommend that anticoagulants are offered to all patients with AF with a CHA2DS2-VASc score of ≥2 and are considered for men with a CHA2DS2-VASc score of 1, regardless of whether the pattern of AF is paroxysmal, persistent or permanent, or the duration of AF.1 16–18

Recent data from two international studies, the GARFIELD Registry and the GLORIA-AF Registry, indicate that anticoagulant therapy is not consistently prescribed across AF categories, with lower rates of use in patients with paroxysmal AF than in those with persistent or permanent AF.19–21

The aim of this analysis was to determine whether patients with paroxysmal AF are less likely to be treated with anticoagulants than patients with persistent or permanent AF in the UK and to investigate trends in treatment between 2000 and 2015. To date, this question has not been addressed in a UK primary care setting.

Methods

Data source

Analysis was performed using data from The Health Improvement Network (THIN), an anonymised database of electronic medical records from UK general practices using Vision software. The version of the THIN database used in this study (THIN1505) included primary care data for approximately 14.0 million patients at 648 practices across the UK. General practices were eligible for inclusion in the study from the latest of: the practice acceptable mortality recording date,22 the Vision installation date and the study start date (1 year prior to the first index date).

Study design

Sixteen sequential cross-sectional analyses were carried out with index dates on 1st of May each year from 2000 to 2015. All analyses were conducted using StataIC V.13. Data were not used prior to 2000 as there were fewer practices and patients contributing data, and data accuracy is better in more recent years.

The study population included all patients with a diagnosis of AF aged 35 years and over on the index date who were registered at least 1 year prior to the index date. The exposure was a diagnosis of paroxysmal AF, with all other types of AF (persistent or permanent AF, hereafter referred to as ‘other AF’) as the comparator group. Outcome was treatment with anticoagulants.

Analysis

The proportions of patients with paroxysmal AF and other AF prescribed anticoagulants on each index date were calculated, with 95% CIs; p values for trends in treatment between 2000 and 2015 were calculated using χ2 tests. Patients were stratified according to eligibility for anticoagulant treatment: in primary analysis, this was defined according to CHADS2 score (≥1), since CHADS2 score has been in use for a longer period of time than CHA2DS2-VASc score23 24; in sensitivity analysis, eligibility was defined according to CHA2DS2-VASc score (≥1). Crude and adjusted risk ratios (RR) were calculated on each index date. RRs were calculated using Cox regression, setting follow-up time to 1 for all patients, using the Breslow method to break ties and using the robust variance estimator25; RRs were adjusted for age and sex only, and for age, sex, Townsend score, CHADS2 score and presence of any contraindication.

Definitions of variables

AF was defined as a clinical code for atrial fibrillation or atrial flutter recorded ever prior to the index date, excluding patients with a clinical code indicating AF resolved recorded on or after the last recorded AF clinical code and prior to the index date. Patients were categorised as having paroxysmal AF if the last AF clinical code prior to the index date indicated paroxysmal atrial fibrillation or paroxysmal atrial flutter; otherwise patients were categorised as having other (persistent or permanent) AF. Current anticoagulant treatment was defined as a record of a prescription for any anticoagulant drug (including warfarin, parenteral anticoagulants, other vitamin K antagonists and new oral anticoagulants) within 90 days prior to the index date or a clinical code indicating provision of anticoagulant therapy within 365 days prior to the index date.

CHADS2 scores were calculated by adding one point for a history of congestive heart failure, hypertension, age ≥75 and diabetes, and two points for a history of stroke or transient ischaemic attack (TIA). CHA2DS2-VASc scores were calculated by adding one point for a history of congestive heart failure, hypertension, diabetes, vascular disease, age 65–74 years and female sex (if another risk factor was present, otherwise 0), and two points for age ≥75 and history of stroke, TIA or thromboembolism. History of congestive heart failure, stroke, TIA, thromboembolism, vascular disease and diabetes were defined by a relevant clinical code recorded ever prior to the index date, excluding patients with a clinical code indicating diabetes resolved recorded after the last recorded diabetes code and prior to the index date; hypertension was defined as a current (previous 90 days) prescription of antihypertensive drugs or the mean of the three most recent systolic blood pressures in the last 3 years ≥160 mm Hg.

Contraindications to anticoagulants were defined as a clinical code within the 2 years prior to the index date of peptic ulcer, intracranial, intraocular or retroperitoneal bleeding, bleeding disorders, haemorrhagic stroke, oesophageal varices, aneurysm, or proliferative retinopathy; a clinical code recording allergy or adverse reactions to anticoagulants ever prior to the index date; a clinical code indicating pregnancy in the 9 months prior to the index date; or severe hypertension with a mean (of the three most recent measures in the last 3 years prior to the index date) systolic blood pressure of >200 mm Hg or diastolic blood pressure of >120 mm Hg.

Results

A total of 179 343 of the 4 419 659 patients eligible for inclusion in the study between 2000 and 2015 had AF. As an individual could contribute to the analysis in more than 1 year, the analyses consisted of a total of 848 852 records of patients with AF. The proportion of patients with AF with a diagnosis of paroxysmal AF increased from 7.4% (95% CI 7.0 to 7.8) in 2000 to 14.0% (95% CI 13.7 to 14.3) in 2015, an increase in prevalence from 0.14% (95% CI 0.13 to 0.15) to 0.43% (95% CI 0.42 to 0.44). Baseline characteristics of patients with paroxysmal AF and patients with other AF (persistent/permanent) are shown in table 1. Across all years, in patients with paroxysmal AF compared with patients with other AF, the mean age was 4.5 years (95% CI 4.4 to 4.6) lower; 6.4% (95% CI 6.1 to 6.7) fewer were males; mean CHADS2 score was 0.43 (95% CI 0.42 to 0.44) points lower; 1.6% (95% CI 1.4 to 1.8) fewer were in the most deprived Townsend quintile; and 0.5% (95% CI 0.4 to 0.7) fewer had one or more contraindications to anticoagulants.

Table 1

Baseline characteristic of patients with paroxysmal AF or other AF, 2000–2015

Over the 15-year period studied, the proportion of patients with paroxysmal AF prescribed anticoagulants increased from 16.0% (95% CI 14.0 to 18.2) to 50.7% (95% CI 49.6 to 51.8), while the proportion of patients with other AF prescribed anticoagulants increased from 33.5% (95% CI 32.7 to 34.3) to 67.1% (95% CI 66.6 to 67.5). Among eligible patients only, defined as those with a CHADS2 score of 1 or more, the proportion of patients with paroxysmal AF prescribed anticoagulants increased from 18.8% (95% CI 16.4 to 21.4) to 56.2% (95% CI 55.0 to 57.3), and the proportion of patients with other AF prescribed anticoagulants increased from 34.2% (95% CI 33.3 to 35.0) to 69.4% (95% CI 68.9 to 69.8) (figure 1).

Figure 1

Proportion of patients with paroxysmal AF and patients with other AF prescribed anticoagulant treatment by CHADS2 score, 2000–2015. AF, atrial fibrillation.

In 2000, patients with paroxysmal AF were half as likely to receive anticoagulant treatment as patients with other AF, RR 0.48 (95% CI 0.42 to 0.55); this disparity has declined, particularly since 2011 and 2012, but in 2015, patients with paroxysmal AF are still less likely to be prescribed anticoagulants, RR 0.76 (95% CI 0.74 to 0.77) (table 2). Adjusting for age, sex, Townsend score, CHADS2 score and presence or absence of contraindications made only a small difference to the results: fully adjusted RR increased from 0.41 (95% CI 0.36 to 0.48) in 2000 to 0.77 (95% CI 0.75 to 0.79) in 2015.

Table 2

RRs for anticoagulant treatment of patients with paroxysmal AF relative to patients with other AF, 2000–2015

Among only those patients with AF eligible for anticoagulant treatment, RR increased from 0.55 (95% CI 0.48 to 0.63) to 0.81 (95% CI 0.79 to 0.83) between 2000 and 2015. Adjusting for age, sex, Townsend score and presence or absence of contraindications had little effect on the results in recent years but caused an increase in effect size in earlier years: adjusted RR increased from 0.46 (95% CI 0.40 to 0.53) in 2000 to 0.80 (95% CI 0.79 to 0.82) in 2015 (table 2); the slightly increased effect of adjustment in earlier years is likely to be due to the greater difference in mean age between patients with paroxysmal AF and patients with other AF, particularly between 2000 and 2006. In sensitivity analysis, eligibility for anticoagulant treatment was defined according to CHA2DS2-VASc score (1 or more); this made only a marginal difference to the observed RRs (online supplementary table 1).

Discussion

Paroxysmal AF is becoming more commonly recorded as a diagnosis in UK primary care. Patients with paroxysmal AF are less likely to be treated with anticoagulants than patients with other AF: in 2000, patients with paroxysmal AF who were eligible for treatment were almost half as likely to receive anticoagulants as patients with other AF, and while this has been steadily improving over the last 15 years, patients with paroxysmal AF continue to be around 20% less likely to receive anticoagulants than patients with other AF. This holds true after adjusting the data for age, sex and other potential confounders. The absolute anticoagulant treatment gap for paroxysmal AF remained relatively constant between 2002 and 2012 at around 20%; in recent years, this gap has narrowed but remains around 13%.

These results are consistent with findings from studies carried out in the USA among hospital outpatients with AF, in which patients with paroxysmal AF were found to be around 20% less likely to receive oral anticoagulants than patients with persistent/permanent AF after adjusting for potential confounders.26 27 Similar results were also found in a study of patients with AF admitted to hospital in Greece,28 and in the Euro Heart Survey on AF, an observational study of patients with AF attending hospital in the ESC member countries (including the UK).29 No comparable studies using primary care data in the UK are known to the authors.

ESC guideline-adherent antithrombotic management is associated with significantly better outcomes, including those related to mortality, thromboembolism, stroke and TIA.30 Current treatment practice does not appear to closely follow published treatment guidelines,31 which most likely results in elevated levels of preventable ischaemic stroke among patients with paroxysmal AF, even more so than in the wider AF population, leading to greater morbidity, mortality and overall cost to healthcare systems.32 33

Further research is needed to establish whether the difference in treatment between patients with paroxysmal AF and patients with other AF is the result of lower levels of treatment initiation by clinicians or whether patients with paroxysmal AF are more likely to stop their treatment and to explore reasons why this is the case.20 This will facilitate the development of methods to improve guideline adherence and/or uptake of anticoagulant prophylaxis, leading to improved outcomes for patients with paroxysmal AF.7

Strengths and limitations

This study uses a large dataset that is representative of the AF population in UK primary care and includes recent data up to 2015. The dataset comprises routinely collected data used by general practitioners to make clinical decisions.

It is possible that AF type could be misclassified, as some cases of paroxysmal AF may be recorded under a non-specific AF clinical code; however, this is more likely to dilute rather than inflate observed effect sizes. Treatment may be under-reported if patients are prescribed anticoagulants by their hospital; however, most anticoagulants are prescribed in primary care, and most secondary care prescriptions should be captured by anticoagulant/international normalized ratio (INR) monitoring clinical codes, so under-reporting is likely to be minimal.

Conclusion

Patients with paroxysmal AF are less likely to be treated with anticoagulants than patients with persistent or permanent AF. This difference is not explained by differences in patient demographics, stroke risk or contraindications. The question remains as to whether the difference in treatment between patients with paroxysmal AF and patients with other AF is the result of lower levels of treatment initiation by clinicians or whether patients with paroxysmal AF are more likely to stop their treatment, and to explore the reasons why.

Key messages

What is already known about this subject?

Patients with paroxysmal atrial fibrillation (AF) are at an elevated risk of stroke. Guidelines recommend anticoagulants for patients with paroxysmal AF.

What might this study add?

In 2000, patients with paroxysmal AF were half as likely to be prescribed anticoagulants as those with permanent AF. In 2015, patients with paroxysmal AF were 20% less likely to be prescribed anticoagulants as those with persistent/permanent AF.

How might this impact on clinical practice?

The treatment of paroxysmal AF in UK primary care.

References

View Abstract

Footnotes

  • Contributors AI had the original research idea. RR undertook data extraction. NJA designed and performed the analysis. NJA and AI wrote the first draft of the paper, which was revised in collaboration with TM, DF and RR.

  • Funding NJA and TM are funded by the NIHR CLAHRC West Midlands initiative. This paper presents independent research, and the views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.

  • Competing interests None declared.

  • Ethics approval The THIN data collection scheme and research carried out using THIN data were approved by the NHS South-East Multicentre Research Ethics Committee (MREC) in 2003; under the terms of this ethics approval, studies must undergo scientific review. Approval for this analysis was obtained from the Scientific Review Committee (for the use of THIN data) on 2 April 2015 (SRC reference number 15THIN021).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Correction notice This paper has been amended since it was published Online First. Owing to a scripting error, some of the publisher names in the references were replaced with ‘BMJ Publishing Group’. This only affected the full text version, not the PDF. We have since corrected these errors and the correct publishers have been inserted into the references.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.