Treatment Pathways for Patients With Atrial Fibrillation

J. A. Hodgkinson; C. J. Taylor; F. D. R. Hobbs


Int J Clin Pract. 2012;66(1):44-52. 

In This Article


Summary of Main Findings

We found no relationship between maintenance or discontinuation of anticoagulants and stroke risk as measured using CHADS2 scores in this cohort of patients with AF, disproving our hypothesis. Although there is more evidence of treatment maintenance than treatment change, especially in the first year after diagnosis, the amount of therapeutic change is still noteworthy and appears higher than in some other studies.[19,29] Prescription patterns for AF therapy, including from 2002 onwards, suggest that too few high-risk patients are receiving best practice treatment, and particularly of concern is that, some of these patients are being transferred away from best practice treatment over time.

There is someone in every category of treatment change, even in the first year after diagnosis, i.e. every possible change and direction of change has been undergone by at least one person, and while there are clearly patterns by which some therapeutic routes are more frequent than others, no one treatment pathway dominates to the exclusion of others. This could imply a lack of certainty about what is best practice, and also what is best for some people/groups (especially in the context of a relative lack of change to treatment in general), or just reiterate the importance of trying different approaches, especially where patient preferences, difficulties in diagnosis and concern over securing the appropriate balance between risks and benefits need to be taken into account.

Comparison With Existing Literature

The lack of relationship between treatment pathways involving anticoagulant use and CHADS2 scores reiterates the findings of much previous research regarding the underuse of warfarin. A smaller study of AF patients breaking down the constituent strata of CHADS2 score (data available for 2381 patients) found the only pairwise comparison that was significantly different for prescription of antithrombotic drugs was between score 0 and 1,[30] and an Italian study found no correspondence between risk stratification and anticoagulant treatment.[31] Another Italian study not only found that risk stratification only partially influences antithrombotic management but also that underuse and discontinuation of anticoagulant treatment is common in incident AF patients.[32]

The results of our sensitivity analyses did not echo those of a French study[33] that identified 1992–1994 as the period of most significant increase in the use of anticoagulants (14.6% in 1985–1987 compared with 40.3% in 1992–1994, p < 0.05). The lack of relationship between CHADS2 score and use of anticoagulants even when patients with an index date prior to 2002 were excluded was surprising, but even post-2002, there has been considerable development in the understanding of best practice for AF [the first derivation of CHADS2 was only published in 2001,[11] and previous research suggests that best practice has frequently not been applied even when relatively widely known.

The discontinuation of anticoagulants amongst a proportion of apparently suitable patients found in some earlier studies was replicated. A study of warfarin use in new-onset AF and its change over time in 1005 adults in the United States and Canada[34] initially found that 65% of patients were prescribed warfarin, but only 44% were taking it at 30 months. Even in 'ideal' candidates for warfarin, the rate of warfarin prescription decreased from 70% at baseline to 50% at 30 months.

Although some patients moved to combinations that still included anticoagulants, as much as one-third of treatment time of all those starting on a therapeutic approach involving anticoagulants featured no use of anticoagulants (either separately or in combination with other treatment types) over the 5-year period. In this light, our findings suggest a very different conclusion to that taken by the authors of the GEFAUR study in Spain that anticoagulation was underused partly precisely because of a reluctance to change current antiplatelet therapy,[19] as we also identified change away from anticoagulation.

There was more treatment change for both patients started on rate and on rhythm control than found in the AFFIRM substudy,[29] but little evidence those on rhythm control are more likely to transfer to another therapy than those on rate control. This could be interpreted as partial evidence of the relative effectiveness of this treatment approach.

The relatively high proportion of treatment change – and the finding that there is someone in every category of treatment change – could be interpreted not as a lack of certainty about what is best practice, but rather to complement the finding from some previous studies that treatment change produces better outcomes than no treatment change.[29]

Strengths and Limitations

Our study benefited from the generic strengths of the GPRD – a large general practice research database, containing a wide extent of clinical data that previous validation studies have confirmed the validity and data completeness of.[35] As a result of the large number of cases included, our study provides reliable estimates of treatment pathways – with large enough numbers in the database for the last 10 years to reasonably demonstrate current practice – and we were able to examine time trends in the association between time since start of diagnosis and treatment approach.

However, the use of the GPRD is not wholly unproblematic. The diagnoses of AF were not verified – the data quality markers for the GPRD do not include assessments of the accuracy of diagnoses or the appropriateness of Read terms selected. Although we had no information on the diagnostic criteria used for AF diagnosis (such as electrocardiograms), a previous GPRD study did report a high level of validity in the recording of AF by GPs.[28] In addition, the onset of AF symptoms may not always be similar to the date of first recording of them in the GPRD, because of delays, either in patients visiting their GP or in diagnosis of AF. More generic limitations of the data source are that there is a lag time before data are available from the GPRD, the practices in the GPRD are volunteers and so may not be typical of general practices across England and Wales, and that there may be some under-recording of some minor comorbidities used, for example, to identify patients with contraindication for anticoagulation.

Further limitations of our study are that those patients with a contra-indication, for example to warfarin, were not excluded, and it is likely that more of those with higher CHADS2 scores would be contra-indicated. CHADS2 scores were applied retrospectively, and certainly would not have been available for therapeutic decision-making at the time treatment decisions were made. Nevertheless, information about the individual risk factors that comprise CHADS2 would have been available. In practice, CHADS2 scores may have changed over time, but the scoring for patients here reflects their state in the 6 months before and after diagnosis. If a practitioner responsible for prescribing for a patient had calculated a patient's CHADS2 score, this may have led to a different rating or alternatively to different drug regimen recommendations.

There may be difficulties in interpreting the meaning of some of the findings, as we did not aim to explore treatment pathways for specific drugs, because the number of combinations would be immense. Rather, we put together certain treatments as rhythm control (flecainide, sotalol and amiodarone), rate control (beta blockers and calcium channel blockers), anticoagulants (warfarin) and antiplatelets (aspirin or clopidogrel). This means that the results may have been confused, for example, by patients being prescribed a beta blocker for hypertension. While it would be desirable to confirm what beta blockers, particularly sotalol, were taken for, as it could explain why, for instance, the 'anticoagulant + rate + rhythm' category was so common, we cannot do this from the results generated.

Treatment received was collated at points in time, rather than continuously through the 5-year period. While it would be useful to know how many months patients were exposed to drugs, we cannot know this, as our percentage estimates of time spent on particular therapy combinations relates to information about treatment extracted at certain points in time, rather than continuously through the 5-year period.

The fact that we were unable to conduct sensitivity analyses stratifying by age is also a weakness of the study, and it is recommended that future research consider the impact of age on treatment pathways for AF.

There may also be possible deficiencies in the prescribing data, if, for example, patients buy their aspirin over the counter or have their anticoagulant monitoring and prescribing carried out wholly in secondary care. This figure is unlikely to be large, however. Finally, health system-related factors such as physician volume, the geographical location of the practice, and specific type of referrals may have influenced treatment patterns, and these are not controlled for.

Implications for Future Research and Practice

Prescription patterns for AF therapy, including as recently as 2002, suggest that too few high risk patients are receiving best practice treatment. The lack of relationship between treatment pathways involving use of anticoagulants and CHADS2 scores is striking, and suggests that this should be a cause for concern. It will be essential to revisit an analysis such as this to consider more recent treatment patterns, as our sensitivity analyses were unable to discriminate any indication of improvement in the provision of appropriate therapies over time. The National Institute of Health and Clinical Excellence (NICE) published guidelines for the management of AF in June 2006, which may have a potential impact on antithrombotic, rate and rhythm prescribing patterns by clinicians.[10] This study could then provide baseline data for a comparison with results in 2–3 years time. Sensitivity analyses stratifying treatment pathways by age would also be instructive. As it stands, although, the results of this study provides further grist to the argument that more education is needed about the most appropriate management of AF. If one limitation of the study is that practitioner involvement in calculating a patient's CHADS2 score may have led to a different set of findings, then this also highlights the importance of GPs conducting such assessments.

Research Ethics

The GPRD Group has ethical approval from a Multi-centre Research Ethics Committee (MREC) for all purely observational research using GPRD data, namely, studies that do not include patient involvement, as here. No individual patients are identifiable through this research.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: