Computed Tomography Angiography of the Head is a reasonable next test following a Negative Non-contrast Head Computed Tomography in the Emergency Department work up of Subarachnoid Hemorrhage

Although subarachnoid hemorrhage (SAH) is rare, identifying a sentinel bleed is one of the most beneficial things an Emergency Physician (EP) can accomplish. For this discussion, we assume that a negative non-contrast computed tomography (CT) scan of the head is inadequate to rule out a sentinel SAH, and must be followed by another test; traditionally- a lumbar puncture (LP). We argue that following a non-diagnostic CT, CT angiography (CTA) is an acceptable alternative in patients with a reasonable prior probability of SAH.

Although the sensitivity of LP for SAH approaches 100%, LP has other suboptimal characteristics. It is invasive, painful, and often causes considerable patient anxiety, such that some patients will leave against medical advice if given no other choice. Not infrequently LP causes a significant postdural puncture headache, the incidence of which is estimated to range from 10–40%[1]. This complication can require medical therapy, epidural blood patch and even admission. (The literature is unfortunately not helpful regarding the frequency of any of these.) LP results are often inconclusive, mostly because of traumatic taps (in 15–20%) – and neither arbitrary cutoffs for red blood cell count, change in counts between tubes, nor evaluation of xanthochromia can resolve this problem[2]. Lastly, LP can be technically difficult, depending on patient body habitus, and is time-consuming for the EP, which can interfere with the care of other acutely ill patients.

Substituting CTA for LP has several advantages. It is painless and non-invasive, and can be performed very rapidly; from the EP standpoint, it requires little time. Diagnostically, it can identify other important causes of worrisome headache (e.g. venous sinus thrombosis, ischemic stroke, and arteriovenous malformation), even if SAH is not present, albeit while forgoing CSF analysis [3]. A recently published mathematical probability model suggests that this strategy would have a sensitivity of 99.5% for aneurysmal SAH and is eminently reasonable for most such patients[4].

There are important disadvantages of CTA. First, the reported prevalence of incidental cerebral aneurysms in the general population is between 0.5% and 6%, depending on study design, and is likely to be at least 2% [5].The vast majority of these will never cause a clinical problem. Thus, approximately one in 50 patients being evaluated for possible SAH, in whom SAH is not present, will have an incidental aneurysm identified. It is not currently possible to distinguish these from a truly dangerous aneurysm that is causing the index headache. This could lead to harm from unnecessary neurosurgical intervention. Fortunately, this problem can be mitigated – simply by following any CTA that shows an aneurysm, but no visible bleeding, with an LP!

The second major concern with CTA is increased radiation. The effective radiation dose of this test is approximately 2 mSv for adults[6]. To put this into context, this represents less than one third the radiation received during a CT scan of the abdomen (8 mSv) and even less than natural annual background radiation (3 mSv) in the United States[7]. Directly and precisely estimating increased cancer risk from acute exposures of less than 10 mSv is very difficult to achieve but is certainly not zero[7]. Nonetheless, extrapolation from epidemiologic data suggest that the lifetime attributable risk of death from cancer associated with a single CT head is less than 0.01% for individuals over 25 years old[8]. While obviously important, this risk is quite small, with a number needed to harm of 10,000, and even less in older adults – who comprise the majority of those worked up for SAH.

Thirdly, CTA exposes the patient to the risk of acute allergic reaction and contrast-induced nephropathy. However, the risk of severe allergic reaction is extremely low, ranging from 0.001% to 0.02% and has decreased over time since the advent of nonionic, low-osmolality contrast formulations[9]. Recent reports estimating the incidence of contrast-induced nephropathy from CTA of the head suggest that the risk of clinically significant, patient-oriented adverse outcomes is negligible[10].

In sum, we don’t argue that CTA is superior to LP as a next test after a negative non-contrast CT head, but rather that it is an acceptable alternative diagnostic strategy in adults who are at risk for SAH. We don’t mean to suggest that routinely adopting an approach involving CTA is better than adopting one involving LP … but neither would we argue the opposite. Rather, we propose that a third approach, centered on shared decision-making, is the one that should be adopted: after informing our patient about each of these perfectly reasonable approaches, we help her choose, based on her own values and preferences.