The classic axiom in emergency medicine is to follow a negative CT with a lumbar puncture, our gold-standard bedside test for definitively ruling out SAH. Unfortunately, our gold standard is fraught with difficulty.
Nontraumatic headache is the presenting symptom in about two percent of ED patients, and approximately one percent of these patients have a subarachnoid hemorrhage (SAH). (CJEM 2002;4:333.) Some 60 to 80 percent of SAHs are caused by aneurysmal bleeds, which result in have high morbidity and mortality. Delay in diagnosis and treatment increases risk of rebleeding, disability, and death.
Noncontrast CT overall is reported to be between 90-100% sensitive for diagnosing SAH, but sensitivity is dependent on timing and the amount of hemorrhage. Unfortunately, CT is less reliable when bleeds are small or patient presentation is delayed. Current ACEP guidelines (Level B recommendation) reinforce this axiom by stating that in “patients presenting to an ED with sudden-onset, severe headache, and a negative non-contrast head CT scan result, lumbar puncture should be performed to rule out subarachnoid hemorrhage.” (Ann Emerg Med 2008;52:407.)
Not so fast. Lumbar puncture is classically considered a low-risk procedure, but complications range from mild and frequent (pain, post-LP headache) to rare and severe (epidural abscess/hematoma). LPs are also time-intensive procedures, but perhaps the biggest risk lies in the possibility of getting a false-positive result. Up to 15 percent of LPs are traumatic, and no method can reliably distinguish a benign traumatic tap from a life-threatening SAH. Decreasing RBC counts in serial tubes is an unreliable determinant and visual analysis of xanthochromia, which most US hospitals use and which is only 50% sensitive.
LP is helpful when negative, but it can confound the diagnosis when abnormal, leading to unnecessary admission and more invasive testing. Considering the incidence of asymptomatic intracerebral aneurysm in the general population is about two percent (Stroke 1998;29:251), blood from a false-positive LP might be incorrectly attributed to an asymptomatic, incidental aneurysm discovered on more invasive testing, which could ultimately result in the neurosurgical repair of an aneurysm that never bled in the first place.
Credible evidence in two recent studies reveals the need to question an LP after a negative CT in selected headache patients. A retrospective study at a neurosurgical referral center reported that current generation CT was 100% sensitive for ruling out SAH in unaltered patients when CT was completed within six hours of headache onset. (Stroke 2012;43:2115.) Another study prospectively looked at more than 3,000 neurologically intact adult ED patients who reported headache that peaked in intensity within one hour of onset. This study reported CT sensitivity of 100% for detection of SAH if performed within six hours of ictus. (BMJ 2011; 343:d4277.)
A recent editorial by the lead author of the 2008 ACEP Clinical Policy on headache supported this new imaging rule and suggested that physician practice and the ACEP policy he helped formulate need to change. (Stroke 2012;43:2031.) Now, a new study in the July issue of the Annals of Emergency Medicine questions the wisdom of attempting to revise one of our most entrenched axioms.
Nontraumatic Subarachnoid Hemorrhage in the Setting of Negative Cranial Tomography Results: External Validation of a Clinical and Imaging Prediction Rule
Mark DG, Hung YY, et al
Ann Emerg Med
This was a retrospective, matched, case-control study of patients presenting to Kaiser Permanente EDs in Northern California between 2000 and 2011. The study sought to validate the six-hour imaging rule by Perry et al (BMJ 2011;343:d4277) and a potential clinical decision rule that excludes SAH after a negative CT in patients over 40, neck stiffness or pain, LOC, or headache onset during exertion. (BMJ 2010;341:c5204.) Fifty-five cases of SAH after a negative CT were identified.
The clinical decision rule identified 97.1 percent of patients with subarachnoid hemorrhage in this group, but analysis of the imaging rule revealed that CT scan completed within six hours of headache missed the diagnosis of SAH in 11 of 55 cases (20%).
Some important points deserve mention before establishing any conclusions. This was a retrospective chart review, and the clinical data are less reliable. And when a SAH case in this study had a pertinent data point missing from the record, it was substituted with a likely value (a process called imputation) so the case could be retained for analysis rather than deleting it entirely. Retrospective analysis also opens up the possibility that historical features or clinical findings documented in the chart were influenced by imaging or LP data that had already been resulted. Think about how many incomplete charts you have at the end of a shift or that are not completed until a few days later.
Less than 70 percent of SAH cases in this study diagnosed via LP after a negative CT had an identifiable source (aneurysm, AVM) that resulted in neurosurgical intervention. Even the authors acknowledge that some of these repaired lesions might have been asymptomatic and incidental. Also of significance, this study used local facility radiologists without training in neuroradiology to interpret CTs, unlike the specialty radiologists upon which the 2012 Stroke and 2011 BMJ studies relied. These limitations aside, this study at worst identified only 11 additional possible cases of SAH missed by the six-hour CT rule over 11 years throughout an entire local hospital network of EDs. Considering the previously discussed risks of further workup, missing one patient in their network per year is probably an acceptable risk.
Perhaps it is time we begin treating SAH like other potentially catastrophic life-threatening diseases we see in the ED. Once reasonable noninvasive testing puts ACS and PE patients in a low-risk group (troponin, EKG, PERC rule), we counsel them, give them return precautions, and send them home, a strategy that requires accepting a small amount of risk. Data now suggest that patients with a negative CT performed within six hours of headache onset that was read by a neuroradiologist is sufficient to put them patients into an extremely low-risk group. In 2013, subjecting these patients to an imperfect “gold-standard” LP is no longer in their best interest.
Click and Connect! Access the links in EMN by reading this issue on our website or in our iPad app, both available on www.EM-News.com.
* Read the full text of the Annals of Emergency Medicine article, “Nontraumatic Subarachnoid Hemorrhage in the Setting of Negative Cranial Tomography Results: External Validation of a Clinical and Imaging Prediction Rule,” for free at http://bit.ly/15YOGb1.
* Read all of Dr. Lovato's past columns at http://bit.ly/JournalScan.
* Comments about this article? Write to EMN at firstname.lastname@example.org.