Atrial fibrillation after 70: stroke risk, anticoagulation and the fear of bleeding

Atrial fibrillation affects roughly 1 to 2 per cent of the general population and around 10 per cent of people over eighty. In AF the atria quiver rather than contract, blood stagnates — particularly in the left atrial appendage — and clot forms. If that clot embolises to the brain, the result is a stroke, and AF-related strokes are on average larger and more disabling than other ischaemic strokes.

Untreated, AF increases stroke risk roughly fivefold. Anticoagulation reduces that risk by around two-thirds. Those two numbers are the whole subject.

How it presents, and how often it does not

Some people feel palpitations, breathlessness, chest discomfort or fatigue. A substantial proportion feel nothing at all, and asymptomatic AF carries the same stroke risk as symptomatic AF. It is frequently found incidentally on a routine pulse check or an ECG done for another reason.

This is the case for opportunistic pulse palpation in older adults, which several guidelines recommend and which takes fifteen seconds. Consumer devices — smartwatches with irregular rhythm notification, single-lead ECG devices — now detect a great deal of AF, and the resulting questions about what to do with screen-detected, brief, asymptomatic episodes are genuinely unsettled. A notification from a watch is a reason to see a doctor, not a diagnosis.

Estimating stroke risk

The CHA₂DS₂-VASc score assigns points for congestive heart failure, hypertension, age 75 or over (two points), diabetes, prior stroke or TIA (two points), vascular disease, age 65 to 74, and female sex.

Note what age does here. Being 75 contributes two points on its own, and 65 to 74 contributes one. Which means essentially every person over 75 with AF crosses the threshold at which anticoagulation is recommended, before any other condition is counted.

The bleeding worry, examined

Clinicians and families both reason that an older person is frail, might fall, might bleed. The instinct is understandable and the arithmetic does not support it.

The frequently cited analysis on this point concluded that a person on warfarin would need to fall roughly 295 times a year for the risk of subdural haematoma to outweigh the stroke prevention benefit. The figure is a modelled estimate and should not be treated as precise, but the order of magnitude has held up in subsequent work: falls risk alone is not a good reason to withhold anticoagulation.

The BAFTA trial randomised over-75s in primary care to warfarin or aspirin and found warfarin substantially reduced stroke with no significant increase in major haemorrhage. That result mattered because it addressed exactly the population clinicians were most nervous about.

Bleeding risk scores such as HAS-BLED exist and are useful — but their intended purpose is to identify and correct modifiable risk factors (uncontrolled hypertension, excess alcohol, concurrent NSAIDs or aspirin, labile INR), not to select people to leave untreated.

Aspirin is not the compromise it is imagined to be

For years, aspirin was prescribed to patients felt to be too frail for warfarin. Current guidance is clear that this is not an acceptable substitute: aspirin is substantially less effective at preventing AF-related stroke and carries a bleeding risk that is not much lower. It is the worst of both positions and has largely been removed from guidelines.

Which anticoagulant

Direct oral anticoagulants — apixaban, rivaroxaban, edoxaban, dabigatran — are now first-line for most patients with non-valvular AF. Compared with warfarin they are at least as effective at preventing stroke, carry a lower risk of intracranial haemorrhage, and require no routine monitoring or dietary restriction. Apixaban in particular has a favourable profile in older and frail patients.

They are not universally appropriate. Warfarin remains necessary in mechanical heart valves and moderate-to-severe mitral stenosis. Dosing of DOACs must be adjusted for renal function, age and weight, and inappropriate under-dosing is a documented and common error that leaves patients with the bleeding risk and less of the protection.

Left atrial appendage occlusion is an option where anticoagulation is genuinely contraindicated.

Rate or rhythm

Separately from stroke prevention, there is the question of managing the arrhythmia itself. For years, trials suggested rate control — simply slowing the ventricular response with a beta blocker or similar — was equivalent to attempting to restore sinus rhythm.

That has shifted somewhat. The EAST-AFNET 4 trial found that early rhythm control, within a year of diagnosis, reduced cardiovascular outcomes. Catheter ablation has a growing role, particularly in symptomatic patients and in those with heart failure.

The crucial point, frequently misunderstood by patients: restoring normal rhythm does not remove the need for anticoagulation. Stroke risk relates to the underlying substrate and to the CHA₂DS₂-VASc score, not to whether the rhythm happens to be regular today.

What to ask

What is my CHA₂DS₂-VASc score? Am I on an anticoagulant, and if not, what is the specific reason? Is my dose correct for my kidney function and weight? And are there modifiable bleeding risks — the blood pressure, the alcohol, the ibuprofen I take for my knee — that we could deal with instead of reducing the protection?