A Comfortable Miss Rate? Who Needs an LP after CT?

Clinical Scenario: A middle-aged man with a history of hypertension, diabetes mellitus, obesity, and peripheral vascular disease presents to the ED after an episode of syncope about 30 minutes ago. He is now completely alert and oriented and complains of a severe headache. Your initial workup, including basic labs, EKG, troponin, and non-contrast head CT, is unremarkable, and you prepare to admit him for observation and an inpatient syncope workup. The hospitalist service calls back to request a lumbar puncture to rule out subarachnoid hemorrhage before the patient comes up to the floor. Should you get the LP?

Clinical Question: In patients with a non-diagnostic non-contrast head CT, is a lumbar puncture necessary to completely rule out subarachnoid hemorrhage?

Literature Review:  In medical school, we all learn that head CT alone is not sufficient to rule out a subarachnoid hemorrhage (SAH) in a patient with a sufficiently suspicious history- you also need a lumbar puncture (LP), to evaluate for blood in the CSF or xanthochromia. On standardized tests, no patient with risk factors and a sudden-onset headache gets to escape the LP needle… but is this the right way to go about things in clinical practice?

Most experts and clinical guidelines continue to recommend LP after negative head CT in patients at high risk of SAH. In the 2012 guidelines for the diagnosis and treatment of SAH, the American Heart Association and American Stroke Association recommend that “head CT, if nondiagnostic, should be followed by lumbar puncture (Class I, Level of Evidence: B).”[1] This recommendation is based mainly on older studies demonstrating decline in the sensitivity of head CT over the course of days.

However, newer studies using modern multi-detector CT scanners may have identified a subclass of patients in whom an LP is not required to rule out SAH.  Perry and colleagues performed a multi-center prospective cohort study to assess the sensitivity of modern third-generation CT in ED patients being evaluated for SAH [2]. Patients presenting to 11 Canadian tertiary care referral centers between November 2000 and December 2009 with suspected SAH were prospectively enrolled. Alert (GCS=15) patients over 15 years of age presenting with non-traumatic acute headache or syncope associated with headache were included in the study. Exclusion criteria included the presence of focal neurologic deficits or papilledema, known history of CNS abnormality (such as neoplasm, aneurysm, or shunt), recurrent headaches, and transfer from another center with an established diagnosis of SAH. The gold standard for diagnosis of SAH was subarachnoid blood on non-contrast head CT, any visually identified xanthochromia on CSF analysis, or RBCs in the final tube of CSF collected AND aneurysm identified on CT angiography. A major weakness of the study was that not all patients enrolled had both a head CT and a lumbar puncture. In an attempt to correct for this weakness, all patients who did not have a definitive diagnosis based on neuroimaging OR a negative LP were followed for six months to ascertain their outcomes. By the conclusion of the study, 3,132 patients had been enrolled; of these, 240 had confirmed SAH. For all comers, the sensitivity of head CT was 92.9% (95% CI 89.0%-95.5%) and the negative predictive value was 99.4% (99.1%-99.6%). However, for patients who were scanned within six hours of headache onset, the sensitivity of head CT was 100% (97.0%-100%), and the negative predictive value was 100% (99.5%-100%). Likelihood ratios were not reported; however, using data available in the paper, they were calculated as a negative likelihood ratio of 0.07 (0.05-0.11) for all comers, and an impressive 0.00 (0.00-0.03) for patients scanned within six hours of headache onset. 

The results of this study were later replicated by Backes and colleagues [4]. In this retrospective single-center cohort study, patients presenting to the ED with a history suspicious for SAH between 2005 and 2012 were enrolled. Patients with clinical suspicion of a non-traumatic SAH and a normal level of consciousness (GCS=15) were included. Exclusion criteria included unknown time of symptom onset, focal neurologic deficits on presentation, referral from another hospital with a confirmed diagnosis of SAH, and LP in the month before presentation. At the study site, all patients with suspicion of SAH undergo non-contrast head CT, and all patients with a nondiagnostic head CT undergo LP with CSF analysis at least 12 hours after symptom onset; patient databases were reviewed to generate a study population of 250 patients who met criteria.  In all comers, head CT had a sensitivity of 95.4% (89.5%-98.5%), negative predictive value of 96.6% (92.2%-98.9%), and negative likelihood ratio of 0.05 (0.02-1.11). In patients scanned within 6 hours of symptom onset, sensitivity was 98.5% (92.1%-100%), negative predictive value 98.6% (92.3%-100%), and negative likelihood ratio 0.02 (0.00-0.10). In fact, only one patient with a non-diagnostic head CT had any findings on LP; this was a patient with atypical symptoms who was subsequently found to have a bleeding cervical AVM. The authors conclude that in patients with typical symptoms who present and are scanned within six hours of headache onset, there is no need for an LP after non-diagnostic head CT to rule out SAH. Weaknesses of this study included its small sample size and retrospective design.

There are, of course, many patients who still warrant an LP after non-diagnostic head CT. Patients with an altered level of consciousness or focal neurologic deficits were excluded from the above studies and require more intensive diagnostics. These findings are not generalizable to patients with an unknown time of symptom onset, significant anemia, pediatric patients, or patients who present to community centers that lack 24/7 coverage by experienced neuroradiologists—note that both studies were performed at academic tertiary referral centers. Some experts also raise the possibility that stopping the ED workup after a non-diagnostic head CT might miss minor “sentinel” bleeds [5], citing a 1987 study showing that head CT missed “sentinel” bleeds in 55% of patients, while LP, when performed, was positive in all patients later diagnosed with SAH[6]. However, this study was performed in 1987, prior to the introduction of modern third-generation CT scanners, and any attempt at replication would likely show improved testing characteristics for CT alone.

Clinical Takehome : In alert adult patients with a suspected non-traumatic SAH and no focal neurologic deficits who are scanned within 6 hours of symptom onset, a non-diagnostic head CT is sufficient to exclude SAH in patients with a low to moderate pre-test probability of SAH.

Submitted by Kevin Baumgartner, PGY-1
Faculty Reviewed by Brian Cohn

Additional related #FOAMed resources:
LP for subarachnoid hemorrhage: The 700 Club
SGEM #134: on what British docs say about LP    

References: 
1. Connolly et al. “Guidelines for the management of aneurysmal subarachnoid hemorrhage: a guide for healthcare professionals from the American Heart Association/American Stroke Association.” Stroke 2012 Jun; 43(6): 1711-1737 
2. Perry et al. “Sensitivity of computed tomography performed within six hours of onset of headache for diagnosis of subarachnoid haemorrhage: prospective cohort study.” BMJ 2011; 343 
3. Alan Schwartz. “Diagnostic Test Calculator.” Department of Medical Education, University of Illinois at Chicago. [http://araw.mede.uic.edu/cgi-bin/testcalc.pl] 
4. Backes et al. “Time-dependent characteristics of head computed tomography in patients suspected of nontraumatic subarachnoid hemorrhage.” Stroke 2012 Aug; 43(8):2115-9 
5. Singer RJ, Ogilvy CS, Rodorf G. “Clinical manifestations and diagnosis of aneurysmal subarachnoid hemorrhage.” UpToDate. Literature review complete through September 2015; article last updated September 2013. 
6. Leblanc R. “The minor leak preceding subarachnoid hemorrhage.” J Neurosurg 1987; 66(1):35