When LP Is Not Necessary to Detect Subarachnoid Bleed

[Zanegray]

Did anybody see this?
Here is the actual reference:
Perry JJ et al. Sensitivity of computed tomography performed within six hours of onset of headache for diagnosis of subarachnoid haemorrhage: Prospective cohort study. BMJ 2011 Jul 18; 3434277.

Here is the capsule summary from Journal Watch:

Standard teaching is that lumbar puncture (LP) is essential in patients with suspected subarachnoid hemorrhage (SAH) despite normal head computed tomography (CT) scans. Researchers prospectively enrolled 3132 consecutive neurologically intact patients older than 15 who underwent head CT with third-generation multislice scanners to evaluate nontraumatic acute headache or headache with syncope at 11 tertiary emergency departments in Canada from 2000 to 2009. LP was performed at the discretion of the treating physician. Experienced radiologists who were blinded to the study interpreted all CT scans. SAH was defined by subarachnoid blood on CT, aneurysm on cerebral angiography, or xanthochromia in cerebrospinal fluid.
Mean headache peak pain severity was 8.7 on a 0–10 scale. LP was performed in 49% of patients after negative CT scans. Overall, 240 patients (7.7%) were diagnosed with SAH. The sensitivity of head CT for SAH was 92.9%, and the negative predictive value (NPV) was 99.4%. Emergency physicians identified all but three cases of SAH; all three patients were scanned >6 hours after headache onset. Among 953 patients who were scanned within 6 hours of symptom onset, head CT had 100% sensitivity and 100% NPV. Follow-up at 1 and 6 months did not identify any cases of missed SAH.
Comment: Because subarachnoid blood diffuses and hemolyzes within hours, CT might not be able to distinguish cerebrospinal fluid from blood as time passes. Patients with histories that raise concern for SAH should be prioritized to undergo CT within 6 hours of symptom onset. If CT is performed with a modern scanner and is interpreted as negative for SAH by an experienced radiologist, LP is unnecessary, unless it is being performed to detect other causes of headache.

Thoughts on this?

---

Members don't see this ad.

 

[xaelia]

I can't criticize the article itself too much, since it's similar to a lot of other literature out there. Their prevalence for SAH falls right in the middle of the pack of all studies - selection bias can get your SAH prevalence as low as 3% or as high as 16%. They only tap half their patients, so you don't honestly know if the other 51% of their enrolled cohort had a SAH - only that they didn't die from a recurrence during the follow-up period. This sort of follow-up is similar to other studies, but it's still a pseudo-gold standard.

That specific JWatch reviewer must hate LPs, because she's a proponent of using CTA to replace LP as well. Her summary statements are amazingly black and white without reservation as if a single, flawed article represents the new standard of care. But, to be fair, improvements in CT resolution are going to have the discriminatory power to detect smaller and smaller SAH. I don't think the sensitivity is ever going to be 100% considering you have to take into account the external validity of this study to your institution as well - I'd imagine practical sensitivity between 90-98% depending on local expertise and equipment. However, once you start getting into those upper 90s for sensitivity, you start making the post-test probability so low after a negative CT that you might start getting as many false positives as true positives off your LP.

We're getting closer to not LPing SAH early in the disease course as the standard of care, but this isn't yet definitive.

http://emlitofnote.blogspot.com/2011/08/slow-death-of-lumbar-puncture.html

 

[docB]

Why did they only tap half the patients? It seems like for any study of the sensitivity of something you'd want to do the study test then do the gold standard test on all subjects to see how many positives were missed.

 

[Arcan57]

There's some biologic plausibility to their data mining, but until there's a large RCT using the less than 6hr criteria (which will be a long time off considering it took 9 years to collect the patients for this study and fewer than a 1/3 were within 6 hrs of onset), I don't see my practice pattern changing. It does lead to some interesting questions if the CT is perfect within 6 hrs. For instance, do we then encourage people with headaches to rush to the ED ala CP and CVA sx? Does a stuttering course of HA time from initial onset or most recent onset? How arbitrary was the 6hr window?