There is now a blood test used after head injury, and it is frequently misunderstood in both directions. It is far more useful than most patients are told, and it does not do the thing most patients hope it does.
What the test actually measures
Two proteins are measured: GFAP (glial fibrillary acidic protein), released from astrocytes, and UCH-L1 (ubiquitin C-terminal hydrolase L1), released from neurons. Injured brain tissue leaks them into the blood, and the concentration reflects how much structural disruption occurred.
What it is good at: ruling out a bleed
In a prospective study across twelve European centres, adults presenting within 12 hours of head trauma had GFAP and UCH-L1 measured and a head CT as the reference standard. The combined assay reached a sensitivity of 95.5% and a negative predictive value of 97.3% for excluding clinically significant intracranial injury. Either biomarker alone performed worse than the pair.
A meta-analysis of 14 studies reached the same conclusion from the other direction: used as a rule-out at low thresholds, GFAP achieved 98% sensitivity with a 97% negative predictive value, and UCH-L1 99% sensitivity with a 99% negative predictive value, in adults with a Glasgow Coma Scale of 13 to 15.
The practical value is that a negative result can spare someone a CT scan and its radiation. That is a real benefit, and it is what the test was built for.
Milevoj Kopcinovic L, et al. Exclusion of intracranial lesions in mild traumatic brain injury using GFAP and UCH-L1: a European multicenter study. Eur J Emerg Med. 2025;32(5):351–358. doi:10.1097/MEJ.0000000000001234 · Karamian A, et al. Accuracy of GFAP and UCH-L1 in predicting brain abnormalities on CT scans after mild TBI: a systematic review and meta-analysis. Eur J Trauma Emerg Surg. 2025;51(1):68. doi:10.1007/s00068-024-02697-3
⚠️ What it is NOT: a blood test for concussion
This is the point that gets lost. Every one of those studies used an abnormal head CT as the reference standard. So the test answers the question “is there a visible structural lesion?” — not “did this person sustain a concussion?”
A normal biomarker result means the same thing a normal CT means: no bleed, no mass effect, nothing that needs a neurosurgeon tonight. It does not exclude a concussion, because concussion is a functional injury that does not produce a visible lesion in the first place. The logic is identical to the one on our diagnosis and testing page — a test that was never looking for your injury cannot rule it out.
If someone tells you a normal blood test proves you did not have a concussion, they have misread what the test does.
Timing: the window is real, and the two markers behave differently
These proteins are not static. In a multi-centre study sampling blood at enrollment and every six hours to 24 hours post-injury, GFAP rose about 3.7% per hour in patients who turned out to be CT-positive, while UCH-L1 and S100B did not rise significantly in that group. The authors concluded GFAP is the more robust marker at later post-injury times.
The practical consequences:
- The validated window is within 12 hours of injury. That is when the rule-out performance above was established.
- UCH-L1 is the earlier marker — it peaks early and falls away, so a late sample can be falsely reassuring.
- GFAP holds longer and keeps climbing where there is a structural lesion, so it is the more informative of the two once some hours have passed.
- Days later, neither is doing the job it was validated for. A biomarker drawn a week after the injury tells you very little.
So if you were injured last month and someone offers this test to settle whether you had a concussion, the answer it produces will not mean what it appears to mean.
Welch RD, et al. Modeling the Kinetics of Serum GFAP, UCH-L1 and S100B Concentrations in Patients with Traumatic Brain Injury. J Neurotrauma. 2017;34(11):1957–1971. doi:10.1089/neu.2016.4772
Age changes the performance — and this is under-communicated
In the European multicentre study, the biomarkers performed worst in adults over 65. Using standard cutoffs, both the individual markers and the combined assay lost accuracy in older patients. Re-optimising the cutoffs specifically for that age group recovered some of it — sensitivity 87.7%, negative predictive value 94.4% — but that is still meaningfully below the 95.5% and 97.3% achieved in the whole cohort.
Older adults are the group most likely to fall, most likely to be on anticoagulation, and most likely to be sent home from an emergency department on a normal result. They are also the group in whom this test is least reliable. If you are over 65, that is worth knowing about your own result.
How we use it
We are not an emergency department, and by the time most patients reach us the validated window has closed. Where a biomarker result exists from the acute presentation, it belongs in the record and we interpret it for what it is — evidence about bleeding, not about concussion. Where it does not exist, its absence proves nothing either.
What actually establishes the diagnosis at our stage is the clinical examination: vestibular and oculomotor testing, cognitive assessment, and structured symptom measures repeated over time.
Common questions
Is there a blood test that can tell me if I had a concussion?
Not in the way the question is usually meant. The available blood biomarkers, GFAP and UCH-L1, are validated to detect or exclude a structural lesion visible on CT. A normal result does not exclude a concussion. It is a very good test for bleeding and it is not a test for concussion.
How soon does the blood test need to be drawn?
Within 12 hours of the injury — that is the window in which its accuracy was established. UCH-L1 rises and falls early; GFAP holds longer and continues to climb where there is a structural lesion. A sample taken days later is outside what the test was validated for.
My blood test was normal but I still have symptoms. What does that mean?
It means you very likely did not have bleeding in or around the brain, which is genuinely good news. It says nothing about whether you have a concussion or post-concussive symptoms, which are functional problems that no blood marker currently rules out. See concussion without loss of consciousness.
Does the test work as well in older adults?
No. In the European multicentre study it performed worst in adults over 65 — sensitivity fell to 87.7% and negative predictive value to 94.4% even after cutoffs were re-optimised for that age group. Older adults are simultaneously the highest-risk group for falls and the group in whom the test is least reliable.