Article Text

Download PDFPDF

Preventing stroke in patients with heart failure: why are patients losing out?
  1. Martin R Cowie1,
  2. Rosita Zakeri1,2
  1. 1 National Heart and Lung Institute, Imperial College London, London, UK
  2. 2 Department of Cardiology, Royal Brompton and Harefield NHS Foundation Trust, London, UK
  1. Correspondence to Professor Martin R Cowie, National Heart and Lung Institute, Imperial College London, London SW3 6LY, UK; m.cowie{at}

Statistics from

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.

According to international guidelines, including those from the European Society of Cardiology (ESC), patients with atrial fibrillation and an additional risk factor for thromboembolic events, such as heart failure, should be considered for long-term oral anticoagulation.1 Risk assessment should include the use of the CHA2DS2-VASc score, which among other things gives a point for the presence of the signs or symptoms of heart failure (or objective evidence of reduced left ventricular function), as well as for being female or older age (1 point for age 65–74 years, 2 points for age 75 years or older).2 Thus, for the vast majority of patients with heart failure and atrial fibrillation, oral anticoagulation should be the default position, the only exception being those who have excessively high bleeding risk and/or those who (after full discussion) do not wish to take an anticoagulant.

In England, The National Institute for Health and Care Excellence (NICE) also recommends that the CHA2DS2-VASc stroke risk score should be used for all patients with atrial fibrillation, and that oral anticoagulation should be ‘offered’ (the strongest level of recommendation) to people with a CHA2DS2-VASc score of 2 or above, taking bleeding risk into account.3 As in the ESC guideline,1 the HAS-BLED bleeding risk score is recommended to provide a more objective estimate of bleeding risk—with emphasis on tackling those risk factors for bleeding that may be reversible (such as uncontrolled hypertension, concomitant use of aspirin or non-steroidal anti-inflammatory drugs or harmful alcohol consumption). Of note, several of the factors that are associated with increased bleeding risk are also strongly associated with an increased risk of thromboembolism, making clinical decision-making complex in some cases.

Such recommendations should mean that most people with heart failure are strongly encouraged to take oral anticoagulation. Sadly, this appears not to be the case. SWEDE-Heart has produced many insights from its ‘Big Data’ approach of linking data from several health and social datasets in one country. This gives us a rich insight into secular trends in cardiovascular health and care, and a better understanding of the associations between clinical, demographic and social factors, and healthcare usage, processes and outcome. In their Heart manuscript, Savarese et al,4 using data from the SWEDE-Heart registry prior to the licensing of the new oral anticoagulants, report that less than 60% of Swedish patients with coexisting heart failure and atrial fibrillation were on an oral anticoagulant. Unsurprisingly, non-use was associated with a worse stroke and mortality outcome than use. Non-use was more likely in subgroups at particularly high risk of stroke: older age, being female and those under non-specialist care, for example. In addition, those with more severe heart failure, a longer history of heart failure, paroxysmal rather than persistent atrial fibrillation and those who lived alone or had a lower socioeconomic background also appeared to be less likely to be anticoagulated.

The chance of being anticoagulated did rise as the CHA2DS2-VASc score rose, but at a less steep gradient than the chance dropped as the HAS-BLED score rose. In other words, physicians were more influenced by a rise in bleeding risk than they were by a rise in stroke risk. This example of omission bias is well known and is perhaps explained by physicians still being of the mindset that if a stroke occurs, it can be considered ‘natural’ whereas causing a bleed by anticoagulation is iatrogenic and therefore more likely to be perceived as their ‘fault’. Other work confirms this explanation, including a study of general practitioners in Canada comparing the effect on prescribing behaviour of having had a patient with atrial fibrillation with an adverse event related to bleeding on an anticoagulant compared with a patient with known atrial fibrillation who had a stroke when not on an anticoagulant.5

How relevant are the findings to other countries and healthcare systems? Work from the same time period in the UK showed that there was an (albeit reducing) older age and female gender bias against anticoagulation for newly identified patients with atrial fibrillation, but no analysis of whether the presence of heart failure modified these effects.6 Savarese et al 4 highlight work from other European countries, which paint a similar picture.

How might the shared decision-making between patients and their physicians be improved when it comes to the prevention of stroke related to atrial fibrillation, particularly when patients are likely to be older, already on multiple medications and perhaps rather overwhelmed by their health issues? The older fashioned approach of ‘the doctor knows best’ is increasingly seen as inappropriately paternalistic. It is vital for the physician (or other members of the multiprofessional team) to have a balanced narrative regarding the likely benefits and risks of any intervention (including drugs), and that this discussion is tailored to the biomedical status of the patient and also to their concerns. Instead of often asking ‘what is the matter with this patient?’ perhaps we should spend more time thinking ‘what matters to this patient?’ Failing to present our best estimate of the likely outcome from various courses of action (and in a way that a patient and their family can understand) is insupportable, but too often there is the temptation to take the easy route and perhaps lead the patient to the decision that is operationally easier for the healthcare system.

NICE has produced a shared decision-making tool for anticoagulation in atrial fibrillation,7 which seeks to help the patient understand the likely benefits and risks of anticoagulation in a more visual way, tailored to the estimated risk of stroke and bleeding (see figures 1 and 2 for a patient with CHA2DS2-VASc score 4 and HAS-BLED score 2). Patient groups can also provide information and support for people to help them come to an informed decision with which they are happy.

Figure 1

Atrial fibrillation (AF): anticoagulant options decision aid from the National Institute for Health and Care Excellence. Cates plots modified from Reference 7, with permission.

Figure 2

Atrial fibrillation (AF): anticoagulant options decision aid from the National Institute for Health and Care Excellence. Cates plots modified from Reference 7, with permission.

What should we do as clinicians (and as members of professional organisations) to ensure that patients with heart failure do not lose out on the potential benefits of anticoagulation if they have atrial fibrillation?

As international guidelines suggest, we should ensure we detect atrial fibrillation (whether persistent or paroxysmal), make an objective assessment of both the stroke and bleeding risk using an accepted tool, and then seek to have a discussion tailored to the individual patient so that a shared decision can be made on the best way forward for that individual. The options for patients have widened with the licensing of ‘newer’ oral anticoagulants. It is often forgotten that in the large randomised outcome trials of these agents, approximately one-half of the patients had a history of heart failure or left ventricular systolic dysfunction: those patients gained just as much benefit as those without heart failure.8 Audit of levels of anticoagulation in patients with heart failure and atrial fibrillation identified in either hospital or primary care, with benchmarking against colleagues, should enable extremes of practice to be identified and examined. Root-and-branch investigation of both severe bleeds and thromboembolic events should also enable fine-tuning of the care-pathway locally. Education of physicians, general practitioners and patients is essential. Policy-makers can also help ensure progress is made if treatment gaps or inequities in service are identified and brought to their attention.

Sadly, tens of thousands of people every year suffer a stroke (which can be fatal or associated with severe disability) that could have been prevented if the atrial fibrillation had been detected and anticoagulation offered. Patients with heart failure appear to be losing out more than most, perhaps because they tend to be older, are considered frailer and are already on multiple medications: action is needed now. Further publications from SWEDE-Heart will tell us whether the situation is improving in that country—but attention is needed on this issue globally.



  • Contributors Both MRC and RZ authored this editorial.

  • Competing interests MRC is a Non-Executive Director of the National Institute for Health and Care Excellence, but the opinions in this article are his own and do not necessarily reflect those of that organisation. He is also a member of the medical advisory boards of the Atrial Fibrillation Association and the Pumping Marvellous Foundation, two patient charities. RZ has no declaration of interests.

  • Provenance and peer review Commissioned; internally peer reviewed.

Linked Articles