As for dogmas to be lysed, no falsity is more resilient than the one that every thunderclap headache (TCH) needs to be tapped to rule out subarachnoid hemorrhage.
Walking through the numbers, our 90,000 pt emergency room probably treats 2,000 headache patients yearly, with 5-10% being thunderclaps, i.e. about 100-200 patients – let’s assume 200. We know from old (Landtbloom 2002) and new (Sayers 2015) series that about 7-15%% of these will be real subarachnoids or about 30 bleeds a year in the worst case. If we tap all of them, we will necessarily induce 10-20% = 40 artificial bleeds, which – according to the best literature we have – is at least hard if not impossible to differentiate from real bleeds:
- the red blood cell (RBC) count need not decrease in artificial bleeds
- but it can in SAH
- Xanthochromia, Ferritin, erythro- and siderophages need time to develop
- there is no proper threshold for RBCs.
If you look at Sayers 2015 series of LPs in TCH, about 15% were uninterpretable and 13% were inconclusive, while 4% were clearly positive. This amounts to 32% LPs that lead to further workup or 64 patients a year for us. Of these 64 patients between 2% (if you use CTA) or up to 5% (if using DSA) harbour incidental harmless aneurysms, so about 2 patients will have one, which then is coiled or operated on.
On the other hand, at most 1 patient of our 40 real subarachnoids yearly will be CT negative (regardless of the time – see Sayers or Dubosh 2016, but at least in the first 6-24 hours). Unfortunately our LP workup for SAH in Germany (visual inspection for xanthochromia) only has about a third sensitivity. We don’t really know how many SAH do have enough red blood cells in their lumbal CSF and what a proper threshold might be. So it turns out, that we might actually miss the one CT negative patient after all.
In all this discussion I did not even consider other adverse events of lumbar puncture.
Here, then, is my take on the issue:
- TCH is way more complex than just ruling out SAHs. Think about the differential (see below) and do a proper history, physical exam and ensure follow-up, if possible with MRI (think of RCVS, SVT and other stuff). A rate of two thirds etiologically unresolved TCHs („primary“) is too much.
- Always keep in mind that Hunt and Hess I-II may actually have a better prognosis than H&H III-V.
- Discuss the probabilities with your department and your patient and be reasonable: LP is not the goldstandard test for subarachnoid!
- If you do an LP to rule out SAH, leave at least 12 hours time (from onset) for xanthrochromia to develop and yes: all this time the patient is in danger, but the probability that he has an SAH is extremely low.
- The two most important papers to discuss CT sensitivity are Sayers 2015 paper and the recent meta analysis by Dubosh in Stroke 2016.Still the old Perry series of 2011 is remarkable and Mark’s collection of 55 CT- „subarachnoids“ is impressive – study the 11 real patients (table 3).
- For TCH read up in Ducros 2013 BMJ article and then read Devenneys 2014 extended list.
- CSF analysis in the setting of TCH is best set out in the international consensus article of 2013 by Nagy.