Are you in the 700 club? Most emergency physicians are. If you trained in emergency medicine you learned that the work-up of a potentially dangerous headache isn’t finished until the CSF is back. In this month’s SMART EM (www.smartem.org) we flesh out our educational session from ACEP 2012, a talk which asked how many LPs it takes to find one aneurysmal subarachnoid hemorrhage—and what the fallout is.
Is it worth 700 LPs to find one aneurysmal subarachnoid hemorrhage?
Are you in the 700 club? Most emergency physicians are.
If you trained in emergency medicine you learned that the work-up of a potentially dangerous headache isn’t finished until the CSF is back. In this month’s SMART EM (www.smartem.org) we flesh out our educational session from ACEP 2012, a talk which asked how many LPs it takes to find one aneurysmal subarachnoid hemorrhage—and what the fallout is.
The literature on this topic is both fresh and remarkably good. The studies are recent, large and prospective, with excellent follow-up. And the results are straightforward. It turns out that following classic dogma means performing roughly 700 lumbar punctures for every one aneurysmal subarachnoid hemorrhage detected. Thus, if you trained in emergency medicine, you’re probably in the 700 club.
And why are we in the club? Faith. For years we have been true believers that performing an LP after a negative CT increases a patient’s safety. But the numbers raise an important question: if we’re poking hundreds of backs for just one to benefit, how much safety is lost? LPs, after all, cause infections, pain, headaches, and on rare occasion serious morbidity. They also lead to false positives, which typically mean a second LP, or perhaps advanced imaging with radiation, and contrast exposure. Sometimes patients are admitted after a false positive LP and have cerebral angiography, a procedure with definite risks. And once in a blue moon someone undergoes risky neurosurgery due to a false positive, because imaging can’t discern which aneurysms were causing a headache and which are incidental.
There’s more. The 700 calculation presumes that patients undergoing LP have acute onset, first time, mostly worst-of-life headaches. In relevant studies roughly 8-10% of such patients have a subarachnoid hemorrhage, making this a high risk group. But that’s not most headache patients. In patients with half the risk, the ratio of LPs to aneurysmal SAH jumps from 700 to 1400. And for those with lower risk headaches the number is in the thousands, which helps to explain why finding an aneurysmal SAH by LP is a Ghostbusters moment.
Now, I love doing LPs. It is a deft hand that swiftly and smoothly threads the needle, coaxing fluid. But 700 is a lot. And thousands of LPs is pretty much LP hell—at both ends of the needle. After all, in addition to being painful and having side effects, up to a third of LPs are falsely positive.
After seeing these numbers we decided to calculate a ‘Test Threshold’ – a formula that crunches the life-saving benefits of finding a true positive LP, and the life-threatening harms of LP for everyone else (we left out minor harms like headache, pain, and infection). The answer surprised us: the threshold is roughly 1%. In other words, performing LPs on more than about 100 people for each true positive LP leads to more life-threatening harm than life-saving good.
This is a revelation. According to our best calculations, unless a patient has high risk headache features (syncope, stiff neck, etc.) the most beneficial approach to diagnosing SAH is a CT scan—with no LP—because the scan typically gets you well under the 1% mark.
It’s a tough pill to swallow, so don’t take our word for it. Listen to the full audio at www.smartem.org, then contemplate, and consider. Then tell us what you think. To us it seems like the 700 club, and the dogma of classic teaching, are more religion than science.
David H. Newman, MD is Author of Hippocrates’ Shadow: Secrets From The House Of Medicine
12 Comments
Now, if we can just get our “expert” colleagues to quit testifying that LP after CT is the standard of care, we can start doing the right thing for patients based on the evidence. Until then, unnecessary testing and procedures will continue.
I always felt that the yield of LP after negative head CT was low, especially if CT is performed within a few hours after symptom onset. In an ideal world, we would be able to translate our experience to a diagnostic algorithm that would define a standard of care. And if that standard were met, there would by definition be no grounds for malpractice. I hate to sound cynical, but malpractice fear is what is driving the “700 Club”, and without some form of legal protection, overuse of LP’s is likely to continue- ironically and tragically to the detriment of our patients.
Can’t wait for the commercial “Have you or your loved one had a poor yield spinal tap? You might be entitled to compensation. Call 1 800 BAD TEST”
As I understand it we are talking about the ACUTE headache patient with the newest generation CT scanners. If you are using an older generation CT scanner or your patient’s HA is > 6-12 hours old, these numbers don’t hold water. A table on the smartem.org site summarizing the data, including type of CT scanner, duration of symptoms, type of study, and mechanism of follow up would be awesome.
I welcome this data, as I can now turn to my patients and say, “one out of 700 LP’s in this situation will reveal a SAH – would you like to proceed or decline” instead of the old “your CT is negative, but the text books and certain experts say an LP is needed now”.
While there is no 100% safe way to practice medicine or avoid litigation, documenting the discussion and allowing a patient to make an informed refusal may help your attorney defend you later.
I love this, but can’t yet decide whether this is actionable info or just a different religion that we like better.
My patients do not get dogmatically dictated LPs now.
To follow up on Dr. Kaminski’s comments, I’ve been doing something similar for the last three to four years. As more data comes out, the number keeps getting smaller, btw. I go over the potential risks of LP (mainly post LP headache), then explain that the risks of missing something serious — like a subarachnoid — range from 1/100 to a 1/1,000. In my experience, about 95% of patients opt out. I document their refusal, and no, I don’t make them sign out AMA.
For patients who have perfectly normal exams, who are generally healthy, and who are being evaluated only because they stated they had the sudden onset of the worst headache in their lives, the only acute concern is SAH. That being the case, why not save the radiation and dollars and just do the far less expensive (and still apparently more accurate, even if only minimally so) LP?
I wrote about this in 2006 and was astounded by the vehemence of some of the reaction!
Diagnosis of subarachnoid haemorrhage following a negative computed tomography for acute headache: a Bayesian analysis. Coats TJ, Loffhagen R. Eur J Emerg Med. 2006 Apr;13(2):80-3.
A good emergency physician HAS to be a good Bayesian – but there are a lot of people out there who have a very strong belief that suggesting that an LP is essential. I love seeing this discussion becoming more mainstream. We also need a large database so that we can better risk stratify patients.
My only concern is those E.D. physicians who are of the original religion consisting of “mandatory L.P’s” in every E.D. for every headache-related CT. Those docs will contend there is no clinical judgement allowed when they testify against you in court as the standard of care.
Agree with much that is said, generally the main things I have picked up is the odd viral meningitis, never picked up xanthochromia.
I think if you CT early in the presentation then an LP is less essential, if the patient presents at 5-7 days, then the CT does get let sensitive and hence an LP may be more likely to have some benefit (or maybe not!)
I am a big fan of a quick CTA of head in delayed presentation to look for aneurusmal source. But I have picked up an aseptic meningitis or too that was a surprise as noted by another poster here.