In Wednesday's post about the Colorado Compendium, Graham mentioned a new 2010 BMJ article on the high-risk signs suggestive of subarachnoid hemorrhage by the gurus in clinical prediction rules in Canada.
We excessively work-up patients for a subarachnoid hemorrhage with a nonspecific headache and no neurologic deficitis. This is because it's difficult to predict who is high, medium, and low risk for such a bleed. So we throw a wider net so that we don't miss such a devastating diagnosis. This usually means a CT and LP for many patients with a headache.
In this 5-year multicenter study, the investigators identified clinical decision rules to help identify the higher-risk groups for a subarachnoid hemorrhage. They derived 3 models, based on recursive partitioning. Each has a negative predictive value of 100%.
Before thinking about seeing if your headache patient has any of these high-risk features, pay special attention to see if s/he would have met the inclusion and exclusion criteria of this study.
Inclusion criteria:
- Neurologically intact adults (age ≥ 16 years) with a non-traumatic headache peaking within an hour.
- History of ≥3 recurrent HA’s of same character/intensity
- Referred from another hospital with confirmed SAH
- Returned for reassessment of same HA which was already evaluated for SAH
- Papilledema
- New focal neurologic deficits
- Previous dx of cerebral aneurysm or SAH
- Previous dx of brain neoplasm
- Known hydrocephalus
- Age ≥ 40 years
- Witnessed loss of consciousness
- Neck pain or stiffness
- Onset of HA with exertion
- Arrival by ambulance
- Vomiting
- DBP ≥ 100 mmHg or SBP ≥ 160 mmHg
Feel free to download this card and print on a 4'' x 6'' index card.
Perry JJ, et al. High risk clinical characteristics for subarachnoid haemorrhage in patients with acute headache: prospective cohort study. BMJ. 2010; 341:c5204. PMID: 21030443
Read article for free at BMJ Online.
Thanks Michelle! Great article. Also looks like, by the data, your risk after a negative CT is approximately 1% to have a positive LP. We think that's a pretty good statistic for patients who are wondering if they should get the LP.
ReplyDeleteDavid Newman and Ashley Shreves have also contacted the authors of the study to try to determine the CT sensitivity based on hours of onset, for that whole "CT is less sensitive after 12 hours" thing.
It would indeed be wonderful to safely eliminate LP from the diagnosis of a SAH. It seems so archaic. Nice find on the article, Graham.
ReplyDeleteI for one really hope the data behind CTA keeps getting stronger. No aneurysm hopefully = no aneurysmal bleed. One can argue "what about non-aneurysmal bleeding?" From what I've read on that, non-aneurysmal bleeding is venous, and the outcomes with venous SAH are excellent (basically, no morbidity/mortality)
ReplyDeleteThanks that point, Corey. Yes, I was always taught that the #1, #2, and #3 cause of non-traumatic SAH by far are aneurysmal bleed, aneurysmal bleed, and aneurysmal bleed. I think if this data is validated, LP will be a thing of the past for r/o SAH.
ReplyDeleteI am a little dissapointed by this study, they could have done better. Why did they choose "peaking within an hour", this dilutes the sample way too much. 30 minutes would have been enough I guess.
ReplyDeleteTwo of the 3 models have "arrived by ambulance" in them, which makes them rather useless. A city ER with lots of non-ambulance patients wont be able to use these consistently. A good example for fishing within the logistic regression model.
Leaves us model 1, according to which a 38 year old patient, with sudden thunderclap headache at rest, peaking within 30 seconds, no LOC, no neck pain etc. would be ruled out ???
Am I too picky or are they really some major flaws in this study?
Agreed that the "arrived by ambulance" is unhelpful since there is so much practice-variation across sites. Like you, I get a little sense of the "fishing" expedition. I'm waiting for the validated study results. In the meantime, I use the derivation study with extreme caution.
ReplyDeleteThe authors point out the limitations that you mention. Specifically about ambulances: "We derived three potential clinical decision rules rather than choosing just the best performing rule, mainly on the basis that arrival by ambulance might not be useful without validation in regions with different cultural or business models for ambulance services."
I don't think you are too picky, but I think there's something interesting to come out of the validation efforts. I'm guessing that once they get enough patients enrolled in study, Model 1 won't have a NPV and Sensitivity of 100%. It'll just be another tool that'll help us risk stratify after we get a negative head CT.
So, the clinician gestalt will trump the decision rule in your example, as in the case of any decision rule that is available.
Michelle-
ReplyDeleteA problem with trying to skip the LP in SAH work up is that you may find an aneurysm but no radiological evidence of hemorrhage. The LP can independently confirm the presence of hemorrhage, necessitating invasive intervention.
A small aneurysm in a patient with a headache but no evidence of hemorrhage may be best left alone. Until imaging can reliably detect the presence or absence of both aneursym AND hemorrhage the LP is here to stay.
Chris
Thanks for this.
ReplyDeleteFor PERC/Well's, how come we accept <2% return visit for confirmed PE; But not >99% sensitivity in CT/CTA without LP?
(Sorry, if I'm mixing up the numbers)
@Jay: Because SAH has a much higher mortality, and arguably missing a SAH sentinel bleed is worse, because their 2nd bleed leads to an even higher mortality rate. (But see this month's SMARTEM podcast for some interesting data!)
ReplyDelete@Chris: Yes, that's exactly what I'm hoping -- that a Head CT (noncontrast and angio) will eventually be sensitive enough to detect aneurysms with and without an associated acute bleed.
ReplyDelete