The study, published in Frontiers in Stroke, examined the death registry of a single hospital between December 2020 and April 2024. Of 5,452 deaths registered during that period, 173 had intracranial haemorrhage listed as a cause of death on the Medical Certificate of Cause of Death, a document completed by the treating clinician and independently verified by a Medical Examiner under the United Kingdom’s statutory death certification system. Seventy-four of those patients had atrial fibrillation, 52 were taking oral anticoagulation, and after applying strict exclusion criteria, 46 patients formed the final cohort.

What the researchers found upends a common clinical assumption. The majority of the fatal bleeds, 78.3 percent, were spontaneous rather than traumatic, meaning they erupted without any preceding head injury. Only 11.7 percent were clearly linked to trauma. Among the spontaneous cases, the bleeding pattern on computed tomography scans was dominated by intraparenchymal haemorrhage with ventricular extension, blood bursting into the brain tissue and spilling into the fluid-filled chambers deep within it, rather than the subdural haematomas that typically follow falls in the elderly.

The radiological findings were the study’s most striking contribution. A radiologist and a stroke physician independently reviewed the neuroimaging of each patient to determine the likely underlying cause of bleeding. In 28 of the 36 spontaneous cases, or 77.8 percent, the imaging showed the classic fingerprints of hypertensive vasculopathy: the chronic, cumulative damage that persistently elevated blood pressure inflicts on small penetrating arteries deep in the brain. These vessels, weakened by years of mechanical stress, lipohyalinosis and fibrinoid necrosis, eventually rupture under the sustained pressure load. Cerebral amyloid angiopathy, the deposition of amyloid protein in vessel walls that is the other major cause of spontaneous lobar haemorrhage, accounted for only 11.1 percent of cases.

Hypertension was almost universal in the cohort. Overall, 84.8 percent of the patients had a recorded diagnosis of high blood pressure, with 86.1 percent in the spontaneous group and 80 percent in the traumatic group. The two groups were otherwise remarkably similar: median age was around 84 years in both, Clinical Frailty Scale scores hovered near 5, indicating mild frailty, and there was no evidence that the patients were frequent fallers or had significant mobility problems. Anticoagulation profiles were also comparable, with apixaban the most commonly used agent at 47.8 percent, followed by rivaroxaban, edoxaban and warfarin, and only a small minority taking concurrent antiplatelet drugs.

The findings carry particular weight because of a long-standing clinical dilemma. Doctors have traditionally worried that elderly patients with atrial fibrillation who are prone to falls face an unacceptable risk of fatal head bleeding if anticoagulated. Earlier work estimated that documented high falls risk raises intracranial haemorrhage risk nearly twofold, and modelling studies suggested that a patient would need to fall roughly 45 to 458 times a year before the bleeding risk outweighed the stroke-prevention benefit of anticoagulation. Yet this study’s cohort, with mild frailty and few recorded fall-related admissions in the year before death, died overwhelmingly of bleeds that had nothing to do with trauma.

The authors also applied SPARCtool, a decision-support instrument that calculates net clinical benefit by weighing a patient’s annual thromboembolic risk against their annual major bleeding risk under their specific anticoagulant. The results were sobering: 69.6 percent of the cohort, 32 of 46 patients, were classified as inappropriately anticoagulated, meaning their bleeding risk exceeded their thromboembolic risk, and the dominant modifiable factor driving that calculation was hypertension. Median annual bleeding risk was 2.8 percent in both groups, while the median net clinical benefit, expressed as strokes prevented per major bleed caused, was 3.1 in the traumatic group versus 2.4 in the spontaneous group.

These observations align with the 2024 European Society of Cardiology guidelines for atrial fibrillation, which made a deliberate philosophical shift. The updated guideline removed female sex from the stroke risk score, producing the CHADS-VA scheme, and moved away from bleeding risk scores such as HAS-BLED as reasons to withhold anticoagulation. Instead, it emphasises aggressive management of modifiable risk factors, with a strict blood pressure target range at its core. The rationale is grounded in evidence that poor blood pressure control is independently associated with haematoma expansion, ventricular extension and worse outcomes after intracerebral haemorrhage, regardless of anticoagulant use, and that a mere 5 mmHg reduction in systolic pressure lowers the risk of major cardiovascular events by 9 percent.

The study has clear limitations that the authors acknowledge. It is a single-centre, retrospective analysis of fatal cases only, so it cannot speak to patients who survived their bleeds, and the modest sample size limits statistical power, with a post hoc estimate suggesting the cohort was adequate for descriptive purposes but underpowered to detect small between-group differences. The researchers also could not assess the duration, type or adherence of anticoagulant therapy, nor reconstruct the precise temporal relationship between blood pressure control and the fatal event. The authors themselves caution that the statistically significant association between haemorrhage type and aetiology is partly circular, since aetiological categories are defined by whether the bleed was spontaneous or traumatic.

Even with those caveats, the message is difficult to ignore. Hypertension is not merely a co-traveller of atrial fibrillation; it is a substrate for the arrhythmia itself and, when left uncontrolled, a direct engine of the very bleeding complications that make clinicians hesitate to anticoagulate. If falls and trauma are being overestimated as drivers of fatal intracranial haemorrhage in this population, then the calculus of anticoagulation decisions changes: rather than withholding life-saving stroke prevention from frail elderly patients, clinicians may achieve far more by rigorously controlling blood pressure before and during anticoagulant therapy. In a condition affecting tens of millions worldwide, that reframing could save a substantial number of lives.

Subject of Research: Association between poorly controlled hypertension, evidenced by hypertensive vasculopathy on neuroimaging, and fatal spontaneous intracranial haemorrhage in anticoagulated patients with atrial fibrillation

Article Title: Association between poor blood pressure control and intracranial haemorrhage-related mortality in patients with atrial fibrillation receiving oral anticoagulation: a retrospective observational cohort study

Article References: Kim, J. U., Cahilog, Z., Kadani, Z., Olajoyegbe, T., Fernandes, D., Ail, D., Oojageer, K., Da Silva, T., Rahman, S., & Kabunga, P. (2026). Association between poor blood pressure control and intracranial haemorrhage-related mortality in patients with atrial fibrillation receiving oral anticoagulation: a retrospective observational cohort study. Frontiers in Stroke, 5, Article 1840317. https://doi.org/10.3389/fstro.2026.1840317

Image Credits: AI Generated

DOI: 10.3389/fstro.2026.1840317

Keywords: atrial fibrillation, hypertension, intracranial haemorrhage, oral anticoagulation, hypertensive vasculopathy, stroke, blood pressure control, ESC guidelines, Association, between, poor, blood