Source Document: Perry JJ, et al. Is the combination of negative CT and negative LP sufficient to rule out subarachnoid hemorrhage? Ann Emerg Med. 2008; 51(6): 707-13 Edlow JA, et al. How good is a negative cranial computed tomographic scan result in excluding subarachnoid hemorrhage? Ann Emerg Med. 2000; 36(5): 507-16. High risk clinical characteristics for subarachnoid haemorrhage in patients with acute headache: prospective cohort study. BMJ. 2010.
Narrative: Headache is an extremely common problem in outpatient cohorts and a small proportion of patients who are neurologically normal will have brain hemorrhages caused by leaking aneurysms that require repair. Reliably predicting which individuals with new headaches will have a hemorrhage without over-treating and over-testing has been extremely difficult. This review examines the highest quality data available from the only existing large, high quality, multi-center emergency department dataset examining the evaluation of acute headache.1
For many emergency physicians in well-resourced settings a common question is whether or not to perform lumbar puncture after a CT scan of the head shows no hemorrhage. CT scan detects approximately 90% of subarachnoid hemorrhages, which means that 10% are missed. Patients often are uncomfortable with the idea of lumbar puncture and a discussion of potential harms, potential benefits, and risk thresholds for lumbar puncture is frequently necessary. These data should guide this conversation, and suggest that if the patient is believed to be at the same initial risk of brain hemorrhage as the patients from the study (roughly 6-7%) then the chance of hemorrhage is roughly 0.5%, or 1 in 200, after a CT scan is normal. This changes considerably depending upon the initial risk, therefore it is important to note that the patients in this dataset were high risk. Nearly 4 out of 5 were experiencing the worst headache of their life, nearly 1 in 3 were vomiting, and 5% had lost consciousness in association with the headache. If a patient is felt to have a very high risk of SAH, say 50%, on initial evaluation, then the risk of hemorrhage after a normal CT is reduced from 50% to 10% (not 0.5%). In those at low risk, for instance those in whom the physician feels the risk of hemorrhage is closer to 1% to begin with, the risk of hemorrhage becomes 0.1%, or 1 in 1000.
Interesting notes from this dataset include the features that seemed to be associated with higher risk. Some features that are classically considered to be seen with subarachnoid were indeed statistically more common among those headache subjects with subarachnoid hemorrhage than those without hemorrhage. These features included more rapid onset of headache, worst headache of life, onset with exertion, syncope, neck stiffness or pain, and vomiting. However, all of these features were very common among headaches that were not due to subarachnoid hemorrhage (except for syncope).
Caveats: These data are from a single large study of headache patients evaluated in ED's and will not apply to most headaches in which the prevalence of subarachnoid hemorrhage is considerably lower than 6.5%. In addition, enrolled patients are at higher risk than most headache patients evaluated in ED's because of the typical selection bias that comes with study enrollment (study subjects are usually higher risk than those who do not get enrolled in such studies). This comports with most emergency physician experience, which suggests that it is considerably less than 6.5% of patients that have a brain hemorrhage among those in whom hemorrhage is considered diagnostically.2 Therefore these risk estimates may be misleadingly high when applied to a general headache population in the ED.
Alternately, these patients were evaluated thoroughly by physicians for neurologic abnormalities and high-risk features, as part of the study. Therefore in the absence of this thorough physician evaluation it is not possible to know whether or not a specific patient is appropriately classified as a headache meeting the above criteria. In addition, some high risk features were exclusions from this study. For instance patients could not have a known history of aneurysm or hemorrhage, papilledema (swelling of the optic nerve) on examination, or a focal neurologic abnormality on examination.
Author: David Newman, MD
Published/Updated: September 21, 2010
If you have suggestions, requests, or questions about a particular NNT review, please send us a message and we’ll try to address it as soon as possible.