Read the complete EBM Focus and earn CME.
Reference: BMJ 2015 Feb 18;350:h568 (level 2 [mid-level] evidence)
Aneurysmal subarachnoid hemorrhages occur in 21,000-33,000 persons in the United States each year, with an incidence of 10.5 per 100,000 person-years worldwide (N Engl J Med 2006 Jan 26;354(4):387). A classic feature of a subarachnoid hemorrhage is the presence of a sudden onset severe headache (thunderclap headache) (Am Fam Physician 2013 May 15;87(10):682). While the usual approach is to perform a non-contrast computed tomographic (CT) scan of the brain in patients presenting to the emergency department with a thunderclap headache (Stroke 2012 Jun;43(6):1711), the sensitivity of CT for diagnosis of subarachnoid hemorrhage decreases if the scan is performed more than 6 hours after headache onset (BMJ 2011 Jul 18;343:d4277, Stroke 2012 Aug;43(8):2115). In patients with a normal CT scan, but who are still considered to be at significant risk of subarachnoid hemorrhage, a lumbar puncture should be performed to assess for elevated opening pressure, xanthochromia, and red blood cells in cerebrospinal fluid (CSF) (N Engl J Med 2006 Jan 26;354(4):387). Unfortunately, it can be difficult to differentiate blood in the CSF from bleeding due to a traumatic tap, and further diagnostic procedures may be required to unequivocally rule out a subarachnoid hemorrhage. A recent diagnostic cohort study evaluated the cerebrospinal fluid of 1,739 alert patients presenting to the emergency department with acute non-traumatic headache to determine the optimal cutoff for distinguishing a traumatic lumbar puncture from red blood cells in the CSF caused by a subarachnoid hemorrhage.
Abnormal CSF samples were defined as red blood cell counts > 1x106 cells/L in the final (fourth) tube or xanthochromia in any tube. CSF samples with increased white blood cell counts, but normal (< 1x106 cells/L) red blood cell counts in the final tube and no xanthochromia were considered normal. Patients were diagnosed with an aneurysmal subarachnoid hemorrhage if they had blood in the subarachnoid space on non-contrast brain CT, CSF with xanthochromia, or red blood cells in the final CSF tube plus an aneurysm on cerebral angiography. The patient also must have required neurovascular intervention or died to be included as a true positive. Non-aneurysmal subarachnoid hemorrhages were not included in this definition. Six hundred forty one patients (36.9%) had abnormal results after lumbar puncture and 15 patients (0.9%) were diagnosed with a subarachnoid hemorrhage. Of these 15 patients, 7 were diagnosed by the presence of xanthochromia and 8 patients had an abnormally high red blood cell count on the final CSF tube (range 9,750-600,000x106 cells/L). All 15 patients had an aneurysm on cerebral angiography. The combined definition of high subarachnoid hemorrhage risk, including the presence of xanthochromia in any CSF tube or a red blood cell count in final CSF tube ≥ 2,000x106 cells/L, had 100% sensitivity and 91.2% specificity for the diagnosis of subarachnoid hemorrhage. The negative predictive value was 100% with a high positive likelihood ratio of 11.4 and negative likelihood ratio of 0.
The results of this study suggest that together, the absence of xanthochromia in any CSF tube and a red blood cell count of ≤ 2,000x106 cells/L in the final CSF tube may rule out a subarachnoid hemorrhage in patients presenting to the emergency department with an acute non-traumatic headache. This rule may distinguish a true subarachnoid hemorrhage from a traumatic lumbar tap and prevent the need for further patient assessment. However, these results are tempered by the fact that the prevalence of subarachnoid hemorrhage was very low in this population and the results of this study have not yet been validated in an independent population. Further studies are required to validate this rule before it can be trusted to help diagnose or rule out a subarachnoid hemorrhage in the emergency department.
For more information, see the Headache and Subarachnoid hemorrhage topics in DynaMed.