A stroke without a cause identified is a stroke waiting to repeat. Finding the cause is not an outpatient afterthought: the two findings that change treatment most — a narrowed neck artery and an irregular heart rhythm — are time-critical, and both are missed when the search is half-done. This chapter lists what should be looked for, in what order, and when a wider search is justified.
1. The five mechanisms
Ischaemic strokes are classified by mechanism because the prevention differs for each:
| Mechanism | What it is | What prevents the next one |
|---|---|---|
| Large-artery disease | A cholesterol plaque in the carotid or another big artery sheds clot | Surgery on the carotid within days; intensive lipid lowering; antiplatelet |
| Cardioembolic | A clot forms in the heart, usually in atrial fibrillation, and travels | Anticoagulation, not antiplatelets |
| Small-vessel disease | The tiny deep arteries damaged by blood pressure and diabetes | Blood pressure, glucose, antiplatelet |
| Other determined cause | Dissection, vasculitis, a clotting disorder, sickle cell, genetic disease | Specific to the cause |
| Undetermined (cryptogenic) | Nothing found, or two competing causes | Keep looking, particularly for hidden atrial fibrillation and a hole in the heart |
About a quarter to a third of ischaemic strokes end up in the last row after a standard workup. The purpose of the rest of this chapter is to make that row smaller.
2. The standard workup, and its time limits
| Test | Looking for | When |
|---|---|---|
| Brain imaging: CT on arrival, MRI where the CT is unclear or the stroke is small or posterior | Type, territory, size and age of the lesion; whether it is one stroke or several (several territories points to the heart) | Arrival; MRI within days |
| Carotid imaging (ultrasound, CT or MR angiography) | A narrowing of 50–99% on the side of the stroke | Within 24 hours (UK), because surgery must follow within days |
| 12-lead ECG, then continuous monitoring | Atrial fibrillation, present or intermittent | ECG on arrival; at least 24 hours of monitoring (UK), extended if nothing is found |
| Echocardiogram | A clot in the heart, a valve problem, a weak ventricle, an enlarged left atrium, a patent foramen ovale | Days; a bubble study or transoesophageal echo in younger people and unexplained strokes |
| Bloods | Full blood count, glucose and HbA1c, a full lipid profile, kidney and liver function, clotting, inflammatory markers; lipoprotein(a) once in a lifetime | First 24 hours; lipids are valid if taken within 24 hours of onset |
| Blood pressure over days | Hypertension, which is the commonest cause of the commonest mechanism | Throughout; a home monitor from discharge |
3. The carotid: the clock inside the cause
A symptomatic carotid narrowing of 70–99% carries a high risk of a further stroke in the weeks after the first, and surgery to clear it (endarterectomy) is one of the most effective operations in medicine when it is done quickly: in the NASCET trial, the two-year stroke rate fell from about 26% to 9%, roughly one stroke prevented for every six operations. The benefit falls steeply with delay, and is largely gone after twelve weeks. The UK guideline asks for surgery within seven days of symptoms for a 50–99% narrowing; the US guideline says within two weeks. Stenting is the alternative where surgery is unsuitable; in older patients it carries more procedural strokes. A narrowing below 50%, or one on the wrong side, is treated with the medicines in the next chapter, not an operation.
4. Atrial fibrillation: the one most often missed
Atrial fibrillation causes around a fifth of ischaemic strokes, and the treatment that prevents the next one — anticoagulation — is different from and better than the antiplatelet given for everything else. The problem is that AF comes and goes, and a single ECG catches a minority of intermittent cases.
- A 12-lead ECG on arrival finds AF in roughly one in twelve stroke patients who did not know they had it.
- Continuous monitoring on the ward for 24–72 hours finds more.
- A wearable monitor for a few weeks finds more again.
- An implantable loop recorder, a matchstick-sized device under the skin of the chest, finds the most: in CRYSTAL-AF, in people whose stroke had no cause found, AF was detected in 8.9% within six months and 12.4% within a year, against 1.4% and 2.0% with standard care, and about 30% by three years. STROKE-AF found the same in people whose stroke had been attributed to other causes.
UK guidance: at least 24 hours of monitoring for everyone; prolonged monitoring (weeks, or an implanted recorder) where the stroke is unexplained and anticoagulation would be given if AF were found. US guidance says the same and is more explicit that long-term monitoring is reasonable after a cryptogenic stroke. If your stroke had no cause found and nobody has offered monitoring beyond the ward, ask.
A caution the trials taught. Finding a few minutes of AF on a long recording is not the same as finding the cause. Two trials (NAVIGATE ESUS and RE-SPECT ESUS) gave anticoagulants to everyone with an unexplained stroke on the theory that hidden AF was common, and found no benefit and more bleeding. Anticoagulation is for AF that has actually been found.
5. The hole in the heart
A patent foramen ovale (PFO), a flap between the two upper chambers of the heart that stays open from before birth, is present in about a quarter of adults and harmless in nearly all of them. In a younger person with a stroke that has no other explanation, it can be the route by which a clot from a leg vein reaches the brain. Three trials in 2017 settled what to do: closing the hole with a device, in people under 60 with an otherwise unexplained stroke and a large shunt or an atrial septal aneurysm, cut recurrent stroke substantially — in CLOSE from 6% to 0% over five years, in REDUCE from 5.4% to 1.4% — at the cost of new atrial fibrillation in about one in twenty, usually transient. The UK guideline: consider closure within six months in people under 60 with a PFO and a right-to-left shunt and no other cause. The US guideline: the same, with a formal scoring system (PASCAL) to judge how likely the PFO is to be the cause. Closure in people over 60, or where another cause exists, is not supported.
6. Younger people, and strokes that do not fit
Around a quarter of strokes occur under 65 and a meaningful number under 50. The standard workup is the same; what changes is how far the search goes when it comes back empty.
| Possibility | Clue | Test | What changes |
|---|---|---|---|
| Arterial dissection (a tear in the wall of a neck artery) | Neck pain or headache before the stroke; recent neck manipulation, trauma, violent coughing; a drooping eyelid with a small pupil | CT or MR angiography of the neck with fat-suppressed sequences | Antiplatelet or anticoagulant for three to six months; trials (CADISS, TREAT-CAD) found neither clearly better |
| Antiphospholipid syndrome | Previous clots in veins or arteries, recurrent miscarriage, a low platelet count, lupus | Lupus anticoagulant, anticardiolipin and anti-β2-glycoprotein antibodies, repeated at 12 weeks | Warfarin rather than a DOAC, long term |
| Vasculitis (inflamed arteries) | Fever, weight loss, rash, joint pain, headache, raised inflammatory markers; multiple small strokes | ESR and CRP, autoimmune screen, vessel-wall MRI, sometimes biopsy | Immunosuppression |
| Sickle cell disease | Known diagnosis; African or Caribbean heritage | Haemoglobin electrophoresis if not known | Transfusion programme |
| Cancer | Weight loss, strokes in several territories, a very high D-dimer | Guided by symptoms | Treatment of the cancer; anticoagulation |
| CADASIL (a genetic small-vessel disease) | Migraine with aura, strokes and early dementia in the family; a characteristic MRI pattern in the temporal lobes | NOTCH3 gene test | Aggressive vascular risk control; no antiplatelet advantage proven; family counselling. The UK guideline gained a CADASIL section in 2023 |
| Fabry disease | Unexplained stroke under 55 with kidney disease, heart thickening, burning pain in hands and feet, or a family pattern | Alpha-galactosidase A enzyme level (men) and GLA gene test (women) | Enzyme replacement therapy; found in around 1% of cryptogenic young strokes |
| Inherited thrombophilia (factor V Leiden and similar) | Rarely the cause of an arterial stroke on its own | Only if a venous route (PFO, venous sinus thrombosis) is plausible | Rarely anything; testing is not routine |
| Drugs | Cocaine, amphetamines, anabolic steroids | History and urine screen | Stopping them |
Every blood, genetic and heart test in this chapter, in tiers with what each result changes, is listed in the every-test chapter.
7. Haemorrhage: finding out why the artery burst
Most bleeds deep in the brain in older people are from long-standing high blood pressure, and the cause-finding is the blood pressure record. Bleeds nearer the surface (lobar), bleeds in younger people, and bleeds without hypertension need CT or MR angiography within 48 hours to look for an aneurysm, an arteriovenous malformation or a tumour, and often a delayed MRI at around three months, once the blood has cleared, to look again. Repeated lobar bleeds in an older person suggest cerebral amyloid angiopathy, which changes the decision about restarting blood thinners; the UK guideline gained a section on it in 2023.
8. When the search comes back empty
"Cryptogenic" should be the end of a process, not a label applied after a normal CT and one ECG. Before accepting it, the following should all have happened: MRI of the brain, imaging of the neck and brain arteries, an echocardiogram with a bubble study in anyone under 60, at least 24 hours of rhythm monitoring with a plan for prolonged monitoring, the full blood panel including lipoprotein(a) and HbA1c, and in younger people the table above. If it has, the treatment is the standard secondary prevention in the next chapter, and the rhythm monitoring continues.
9. Where UK and US guidance differ
- Carotid surgery timing. UK within seven days; US within two weeks. The evidence favours the UK number.
- Prolonged rhythm monitoring. Both recommend it for unexplained stroke; the US guideline states the loop-recorder option more explicitly and US practice uses it more.
- PFO. Both close it in the under-60s with no other cause; the US adds the PASCAL classification. The same trials underlie both.
- Genetic causes. The UK 2023 guideline has dedicated sections on CADASIL and amyloid angiopathy; the US guideline covers them within the young-stroke text. Neither recommends routine genetic testing without a clinical or family clue.