What's Actually Going On
The internal carotid and vertebral arteries run up through your neck to feed the brain. In a dissection, the innermost lining of one of them tears, and blood is driven into the wall itself under pressure. That forms a bruise inside the vessel wall, which either narrows the channel from the inside or bulges outward.
The important part is what happens next. That bruise is sticky, so a clot forms at the tear, and a piece of it can break away and travel into the brain. The harm comes from travelling clot, not from the artery closing off. That is why the treatment is blood-thinning medication rather than anything hands-on, and it is why the stroke, when it happens, arrives days after the pain started.
Most of these happen spontaneously. An underlying weakness in the vessel wall, partly inherited, is thought to be why an ordinary event can trigger one in some people and nothing at all in everyone else.
How to Identify It
- Head or neck pain is the presenting symptom in about 80% of cases.
- In people who already got headaches, the pain was different in quality in about 3 out of 4.
- Median severity was 5 out of 10, so it is not necessarily severe.
- Sudden explosive onset happened in only 5.5%.
- Only 14.7% got any relief from over-the-counter medication.
- In tears of the vertebral artery, dizziness or vertigo is the most common symptom at 58%, ahead of headache (51%) and neck pain (46%).
- Fewer than half had an obvious injury, and only 7.9% had a known connective tissue condition, so a clean history does not clear you.
- Drooping eyelid, cranial nerve problems and pulsatile ringing in one ear are real signs, but they are late ones.
And here is what it is not. The single hardest part of spotting this is that it wears the clothes of an ordinary neck problem for the first few days. It can also arrive looking like a trapped nerve: in the reported cases of tears presenting as nerve pain, the C5 nerve root was the one most often affected, producing shoulder and arm pain with weakness lifting the arm out to the side.
For clinicians, the two figures worth knowing: the structured history framework currently in use Sn: 50% | Sp: 63% and the positional neck test it replaced Sn: 0-57% | PPV: 0%.
The Debate
The safety test that was measured, and then retired
Standard practice for decades
Before manipulating a neck, turn and extend it and watch for symptoms. A clear result meant it was safe to proceed.
vs
Hutting 2013; Côté 1996; CADRE 2017
Measured properly against blood flow, the test detected 0% of cases, with a range of 0 to 57% across the only four studies that exist. A search of 9,022 articles found no evidence supporting Doppler flow testing for this either.
The profession's own international body removed positional testing from its framework in 2020 and replaced it with a structured history. That was the right call, and it was made on the profession's own evidence.
Then the replacement was measured too
The framework, 2020 and 2023
An internationally agreed structure for thinking through vascular risk before treating a neck, endorsed across 22 countries.
vs
Go4Safe, 2021 and 2023
Checked against artery scans in 150 primary-care patients, the framework caught 50% of cases and flagged 54.7% of an ordinary caseload as needing caution. Two therapists assessing the same patient agreed only moderately.
The researchers who ran both studies are manual therapists testing their own field's framework, and they also published a case where a serious event followed gentle hands-on treatment, specifically to make the point that following the framework does not guarantee detection. The honest reading: there is no test that rules this out, the framework is the best available way to think rather than a safety net, and a profession that measures and publishes the limits of its own safety procedure is doing the thing you would want it to do.
Honest Limitations
The caution has a cost nobody has counted
The current framework flags more than half of ordinary neck pain patients as needing caution. Given how rare this condition is, the large majority of those are false alarms, and each one means someone does not get treatment that would have helped.
Nobody has calculated whether that trade is worth making. That is not an argument for ignoring the caution. It is an argument against believing it is free.
These are prompts, not tests
Every recognition feature on this page comes from studying people who definitely had a tear. That tells you what those patients reported.
It does not tell you how many people with ordinary neck pain report exactly the same things, which is why none of these features can be turned into a percentage. Use them to decide to get checked, not to decide you are fine.
Even the scans are imperfect
In the best trial available, independent expert review found that 52 of the 250 patients enrolled with a diagnosed tear did not actually have one.
This does not change the advice to get assessed, because the alternative is not getting assessed. It does mean that if your history strongly fits and a scan comes back unclear, that is worth a conversation rather than instant reassurance.
The Nuance
Can having your neck manipulated cause this? Genuinely unsettled, and the newest evidence points the other way to the oldest. Older studies comparing patients with the general population found a link. Newer analyses comparing them instead with people who saw any clinician for the same neck pain found no excess at all.
The most likely explanation for both sets of results is the same: a tear already in progress causes the neck pain that sends someone to get treated in the first place. So the people getting treated were already the people with a tear. A 2015 review looking for any study measuring how often this actually happens after manipulation found that none had ever been done.
None of that changes the practical advice, because it does not depend on the answer. Skipping a neck manipulation while something is being checked costs you nothing, and there is always another way to treat a sore neck.
What happens to people who get it? Most do well. Around two thirds make a good functional recovery. A repeat tear happens in about 4% of people, another stroke in roughly 2 to 2.4% in the first year, and the pain itself typically clears within about two weeks. The outcome is decided overwhelmingly by whether it is found before the stroke rather than after, which is the reason this page spends almost all its words on spotting it.
