On episode three of the Ottawa EM Podcast, Dr. Rajiv Thavanathan (R5) interviews Dr. Michael Hale (R5), on the ED presentation and diagnosis of subarachnoid hemorrhage in the context of the thunderclap headache. (Click here to access Podcast Main menu)

Subarachnoid Hemorrhage (SAH)

Why does the diagnosis matter?

  • 30-day mortality of almost 50%.
  • Of those who survive a SAH diagnosis, 30-50% are left with significant disabilities.
  • While the most common cause of SAH is trauma, this post will specifically focus on non-traumatic SAH, most often due to an aneurysm. 
  • Note while headache account for 2% of all ED visits, SAH accounts for only 1% of this percentile. 
  • Most ER doctors will see less than 50 SAH total during an entire career.

Predictive Signs & Symptoms 

  • Thunderclap headache
  • Worst headache of life
  • Neck stiffness
  • Vomiting
  • Loss of consciousness (LoC)

Note: each of these features alone are more likely to be caused by another diagnosis.

In fact, thunderclap headache, although common in SAH, isn’t overlying predictive of SAH. That is to say, most patients presenting with a describe thunderclap headache will not go on to be diagnosed with a SAH, speaking to the rarity of the diagnosis.

While a combination of the above symptoms increases your pre-test probability, the  combined likelihood still rarely exceeds 25%.

Thunderclap Headache Definitions

  • The inclusion criteria of the Ottawa SAH rule states peaking intensity within 1-hour. This was done to increase the sensitivity of the rule, that is, not to miss any cases.
  • However, the vast majority of SAH peak either instantly, or within a few minutes, very rarely extending beyond minutes.
  • The original textbook definition of thunderclap was “a headache of moderate-to-severe intensity that peaks within 60-seconds”.
  • The story that concerns neurosurgeons “an instantly peaking headache, snap of the fingers fast, comparable to being hit in the head with a baseball bat”.

Non-Contrast CT (NCCT) Head

In 2011, the initial SAH rule-derivation study showed 100% sensitivity to diagnosis SAH if the scan was performed within 6-hours.


In 2020, Dr. Perry’s SAH Validation paper published in Stroke, they found a 95.5% sensitivity. And in-fact the true sensitivity may lie somewhere in-between these numbers, as from the most recent study, of the 5-missed SAH cases of 111 SAH diagnosis:

  • 2 were false positives caused by traumatic lumbar puncture, and the aneurysms identified on CTA were deemed to be incidental by neurosurgery (Note: we know 2-3% of the population will have incidental aneurysms).
  • 1 case missed by radiology.
  • 2 cases were true misses, 1 being a rare cause of SAH (a dural venous fistula), 1 being a patient with sickle-cell anemia and a hemoglobin of only 63 g/L.

*Clinical Pearl: A hemoglobin < 100 g/L makes blood less hyperdense on CT-scan, and therefore much easier to miss. Therefore, clinicians should proceed with caution in applying this rule to the profoundly anemic patient.

For Dr. Michael Hale, if a patient presents to the ED under the 6-hour window, a NCCT will suffice unless the patient is:

  1. Anemic with a hemoglobin of < 100 g/L
  2. Classic story of an instantly peaking headache with neck stiffness, vomiting, where your initial clinical suspicion remains incredibly high.

CT Angiogram (CTA) vs Lumbar Puncture (LP)

  • The literature supports CTA and LP to be effective at ruling out SAH with a similar degree of clinical certainty.
  • For a LP, this does depend on your positive criteria: whether xanthochromia vs. # of RBCs vs. combination of the two.
  • (From Dr. Perry’s study: xanthochromia + 2000 x 106 RBCs as cutoff, combined with a negative NCCT, the miss-rate of SAH is well under 1 in 1000).


LP: Pain, bleeding, infection, post-LP headache, time, and can be technically challenging.

CTA: Radiation (double the radiation-dose, albeit still less than that involved a CT Abdo-Pelvis), contrast-allergy, and the risk of detecting an incidental aneurysm (2-3% of the population, the vast majority of which would never cause any problems).

Does time matter for CTA sensitivity as it does for NCCT?

Despite both being CT technology, the two tests are looking for different things. For NCCT, we’re looking for blood in the subarachnoid space, for CTA we’re looking for an aneurysm.

The factors that decrease the sensitivity of each are different:

  • For NCCT, sensitivity is decreased by the amount of blood, the time from onset to time of scan (as blood diffuses away from the area, is broken down and becomes more iso-dense) and anemia.
  • For CTA, sensitivity is decreased by aneurysm location (i.e., near the skull-base, where there is significant artifact), size of the aneurysm, and contrast-timing and technique.
  • Ultimately, what decreases the sensitivity of NCCT notably the timing is very different than what decreases the sensitivity of CTA.

Is there an added diagnostic yield to doing CTA over LP?

  • Yes, when you’re considering alternative diagnosis.
  • i.e., cervical artery dissection, cerebral venous sinus thrombosis (CVST), and reversible cerebral vasoconstriction syndrome (RCVS) can all present with thunderclap headache.

RCVS (Reversible Cerebral Vasoconstrictive Syndrome)

  • First-coined in 2007. A relatively new diagnosis, and more a reflection of recategorization of diagnosis (i.e. the post-coital headache, the exertional thunderclap headache, all likely fall under the category of RCVS).
  • Reversible, segmental vasospasm in the Circle of Willis that’s identified on CTA imaging.
  • These patients often present with recurrent thunderclap headaches i.e. 4 over a 1-month period.
  • This should be on your differential diagnosis in cases of those patients presenting with recurrent thunderclap headache.
  • Although the vast majority of RCVS patients have complete resolution in 1-3 months, a notable percentage of these patients will develop negative sequalae. SAH in 30%, seizures in 15%, and can also lead to stroke and posterior reversible encephalopathy syndrome (PRES).
  • Clinically we can’t predict which RCVS patients will go on to have a benign course, vs. severe complications, therefore neurology often admits these patients for monitoring and initiation or either PO or IV calcium-channel blockers (CCB) to prevent vasospastic complications.


30-year-old patient presenting with thunderclap headache that started 5-hours ago

  • 6-hour NCCT cutoff is sufficient. If normal, stop-there.
  • However, if you’re really considering an alternative diagnosis, i.e. this patient has significant neck pain concerning for cervical artery dissection, or in the immediate post-partum period concerning for CVST, consider proceeding with a CTA even in the below 6-hour group. This isn’t necessarily to rule-out a SAH, but to expand on the differential diagnosis.
  • If the patient with a really concerning story and multiple risk-factors, likely prudent to proceed with LP or CTA even within 6-hours to further lower pre-test probability. *a minority of patients
  • If the hemoglobin is <100g/L, NCCT even within 6-hours is likely non-sufficient.
  • If this patient presented outside the window (i.e., thunderclap headache 8-hours ago), you will want to obtain either a LP or CTA. The decision in choosing between LP and CTA is based both on shared decision-making with the patient on risks vs. benefits, and your clinical suspicion of an alternative thunderclap diagnosis.
Michael Hale

Michael Hale

Dr. Michael Hale is a 4th year Emergency Medicine resident at the University of Ottawa who is completing a Masters in Business Administration.
James Gilbertson

James Gilbertson

Dr. James Gilbertson is a first year emergency medicine resident at the University of Ottawa. He is a Junior Editor with the Digital Scholarship and Knowledge Dissemination team for the EMOttawaBlog.
Rajiv Thavanathan

Rajiv Thavanathan

Dr. Thavanathan is an 4th year Emergency Medicine resident at the University of Ottawa, currently undertaking a fellowship in ultrasound.