Although the lumbar puncture is a cool procedure, and very satisfying when it goes as planned, it is time-consuming, possibly painful to the patient, requires discussion/consent that can take a fair amount of time, and requires CSF fluid analysis which can also take a lot of time. It also requires skill (and occasionally luck), which some have more than others. What this all boils down to is most emergency physicians hoping that there is a better, easier test than an LP when we are ruling-out subarachnoid hemorrhage (SAH) in our headache patients.
When I was a resident, I remember distinctly that there were widely varying practices among attendings. Some were fine with just getting a plain CT and stopping the work-up there; some wanted CT/CTA; others wanted the traditional CT followed by LP. I was given variable explanations for these practices, and was left wondering: who was right?
If you really want to listen to some good lectures about SAH in general and have this question (and many others) answered thoroughly, you owe it to yourself to listen to SMARTEM’s podcasts entitled “Subarachnoid Hemorrhage: A Rational Approach” from Dec 18, 2010 (1h 53min) and the newer “SAH: A Picture is Worth a Thousand LPs” from Nov 7, 2012 (2h 11min). If these seem too long for you, you can hop over to ERCast’s podcast entitled “The Subarachnoid Enigma” from May 8, 2015 (19 min).
Here’s the spoiler alert: CTA cannot replace LP. To be specific, we are talking about patients with a normal plain CT and a normal neurological examination. The problem with CTA is that it is very good at finding aneurysms > 3mm. Great, right? No! Because the issues are:
1. The CTA will NOT tell you if that aneurysm has bled at any point in time
2. Roughly 2.5% of people in this country have aneurysms, so it may be a completely incidental finding.
So who cares if you find an aneurysm incidentally, haven’t you still done the patient a huge favor? Not necessarily…You have now basically assured the patient that they will get an angiography (the real, invasive kind) and possibly a neurosurgical intervention. These are not benign procedures and have significant complications and fairly high complication rates.
So you need to do the LP. Still don’t believe me? Check out these resources:
-Life in the Fastlane has a nice general summary of headache assessment/management in the ED
Life in the Fastlane link
-There is a wonderful article in Academic Emergency Medicine called “What are the Unintended Consequences of Changing the Diagnostic Paradigm for Subarachnoid Hemorrhage After Brain Computed Tomography to Computed Tomographic Angiography in Place of Lumbar Puncture”. You can view the article in it’s entirety with this link: CTA in Place of LP article
I really encourage you to take a look at these things so that you can have an active discussion with your attendings while working-up these patients and so that you are knowledgeable about this very important topic.
Please share any questions/comments you may have!
I think there is far more to this and this is well beyond settled.
Unfortunately, CT-LP is advocated as the “standard of care;” A standard which developed in a time with less capability and less of a drive for high quality studies.
As much as a negative CT-LP may be the “Gold Standard,” for ruling out a SAH, the actual utility is much lower than we’d like to think. Only around 1/3 of LPs are truly bloodless and it only requires an RBC count of > 5 cells per microL to be “positive.” Quite opposite to some practice, decreasing RBC counts cannot be used to suggest traumatic tap. What about xanthocromia? Xanthocromia detection starts approximately 12 hours after the start of bleeding, leaving a window period where many won’t trust a negative CT, but also shouldn’t trust a negative LP. To make it worse, most labs do not use spectrophotometry to measure for xanthochromia, but instead just look at the sample by eye. The false positive and false negative rate of the LP are not actually known (especially in anemic patients)…unless we circle back and use angiography as our confirmatory test. Realistically, LP is just an additional screening test that we add to CT. It may have a strong negative predictive value, but the risk pool becomes so small at that point, that a dice roll might give similar results.
While some editorialists have brought it up, LP is not benign, and the morbidity does need to be considered. When using a 22 g needle, up to 40% will experience a post LP headache and 15% of those individuals will go on to have a blood patch. When you look at the factors that go into a post LP headache, residents are going to have the highest complication rate (inexperience and more pokes). Many institutions still use cutting needles for the LP, which makes the risk of this complication at the highest end.
CT-ango is not risk free. Contrast reactions do occur, although the real numbers indicate about 0.5% of any reaction and 0.08% of a moderate reaction. Additional radiation exposure is frequently brought up. The calculations on additional cancers from CT scanning is questionable at best. The calculations rely on a extrapolation of the stochastic effects of radiation. The data sets used included exposures of 10s and even 100s of times higher. While it may be “best data,” is still isn’t very good. A CT head with and without contrast (ie two scans), provides a radiation dose of 4 mSv. By contrast, a CT Abdomen/Pelvis is about 10 mSv. Frankly, the stochastic affects of radiation exposure below 100 mSv is tough to predict (there is even a study that suggest low dose radiation exposure is beneficial – a la hormesis).
The CT-angio effects are only prevented if you actually get a negative-negative study. For the rest, the LP delays care and adds a painful procedure with potential complications.
From the perspective of “what do you do.” I don’t think there is a clear answer. CT-LP is what is done and for good reasons. Technology gives a new modality that provides an excellent, similar, but slightly different answer. I’m also not sure if part of the problem is that the question isn’t clear. If you ask 100 emergency physicians what they are ruling out on a patient with a sudden onset headache, I think you’ll be surprised at the answer. I think you’ll find a remarkable number don’t actually want to rule out sub-arachnoid hemorrhage…they really want to rule out an aneurysm.
Hey Sam, thanks for the comments. I totally see what you’re saying. But the point that people need to understand is that identifying an aneurysm on CTA does not equal SAH. I agree that a negative CT + negative CTA is pretty darn good, no doubt about that. But it isn’t 100% (close enough though, I agree) and more worrisome is finding the aneurysm that has nothing to do with the headache. You still have done very little to prove SAH just by finding the aneurysm. I agree that in patients in whom you are unable to obtain an LP (ex. elevated INR, pt adamantly refuses, etc), there certainly may be a role for CTA. I just want to emphasize that CT/CTA should not be the first line diagnostic algorithm for the average “r/o SAH” pt. What do you think?
Hmm, I have yet to get through all the literature you have kindly provided, but LITFL is citing “Negative CT followed by negative CTA will give post-test probability of excluding SAH of 99.43%. But note up to 2% population may have an ‘incidental’ small aneurysm.”
Please correct me if I am wrong, but it sounds like the argument is that while CT/CTA is pretty good at picking up SAH, the concern is for the incidentalomic aneurysm. I don’t think its fair that we can’t use a reasonable test to essential rule-out life threatening pathology, because of fear of neurosurgery inappropriately managing the patient. There are patients who won’t tolerate the LP and there are nights it is too busy to perform the LP in a timely manner. The CT/CTA protocol is a valuable tool in our arsenal, and I certainly don’t think that it should be out the door because of the evidence presented.
If my patient with sudden onset headache does have an aneurysm, I think I would be pretty content with bringing them in for neurosurgery eval, neuro checks, etc. There are sentinel bleeds with risk of rebleed, and there are chest pain patients who come in with way weaker stories than “acute headache + newly identified aneurysm”. We can’t be 100% specific in our diagnosis in the ED. Of course, there is an art to this and testing should be applied proportionately to risk.
The LP is certainly the gold standard and has broader diagnostic utility. There was mention of CTA being negative because of vasospasm after bleed
“This is because in some patients with aneurysmal SAH, the aneurysm is not visualized on the first study due to parent vessel spasm or thrombosis or a major change in size due to the rupture; this occurred in 14% (4 of 28 cases) in one study. So some CT-negative SAH patients with a negative CTA will not be identified without an LP. This underscores the need to know if a SAH has taken place (not just the presence of an aneurysm)”
The AEM article you provided has a nice list of other pathologies the LP may be useful for “Some diagnoses can only be made or suggested by LP, including various forms of meningitis, spontaneous intracranial hypotension, pseudotumor cerebri, and cerebral venous sinus thrombosis (CVST).”