Accurate diagnosis of subarachnoid haemorrhage

The annual incidence of subarachnoid haemorrhage (SAH) is nine cases per 100,000 population with 85% of SAH caused by rupture of a cerebral aneurysm.^1,2 Case fatality in aneurysmal haemorrhage is around 40% with 30% of patients suffering a re-bleed within four weeks. ² Treatment is effective: with early diagnosis, appropriate medical management and aneurysmal obliteration, mortality and significant disability at one year can be reduced to 24%. ³

It is estimated that 8% to 12% of patients with a thunderclap headache (abrupt headache of severe onset) will have had a SAH. ⁴ Most cases of SAH are confirmed with a non-contrast computed tomography (CT) scan of brain; earlier imaging from headache onset improves the sensitivity for identifying SAH. In patients with a negative CT scan, a lumbar puncture is required to look for evidence of haemoglobin breakdown products in cerebrospinal fluid (CSF). ⁵ In the late 1980s and early 1990s, it was estimated that about 2% of patients with SAH have a normal CT scan of brain if performed within 12 h of headache onset. ⁶ There is, however, emerging evidence that a negative CT scan within 6 h of headache onset in patients who are neurologically intact and without neck pain, may be sufficiently sensitive to avoid the need for lumbar puncture to exclude SAH,^7,8 but this has not yet been adopted by guidelines. Lumbar puncture for the detection of CSF xanthochromia, therefore, continues to be required for CT-negative patients.

Delayed diagnosis of SAH remains common and frequently attracts medicolegal attention; 9% of the total neurological litigation has been attributed to misdiagnosis of SAH. ⁹ In the 2011 UK national confidential enquiry into patient outcome and death (NCEPOD) from aneurysmal SAH, 49 of 383 patients (13%) did not have a timely diagnosis of aneurysmal SAH; this was mostly due to delayed CT scanning. ¹⁰ Of the 12 patients in the cohort who did not have SAH identified radiologically, 10 underwent lumbar puncture which confirmed the diagnosis.

Spectrophotometric analysis of CSF for the presence of bilirubin is widely used in the UK and other European countries although visual inspection of the CSF for pigmentation remains common practice elsewhere. ¹¹ Other less commonly used approaches include the measurement of CSF ferritin and the direct measurement of CSF bilirubin. ¹¹ The UK National External Quality Assessment Scheme guidelines (revised in 2008) ⁵ propose the measurement of net oxyhaemoglobin absorbance (NOA) and net bilirubin absorbance (NBA) in a CSF sample taken at least 12 h after headache onset. There are, however, limitations to this test.

First, increased CSF bilirubin is not specific for SAH, ¹² and may occur in other conditions such as meningitis and spontaneous intracranial hypotension. Second, the diagnostic absorbance cut-offs proposed are somewhat arbitrary. The reference range for CSF bilirubin absorbance of <0.007 AU (initially proposed by Chalmers and Kiley ¹³ ) was derived from a control sample of only 16 patients. The 2008 UK guidelines recommend an NBA >0.007 AU as a clear indication for angiography. ⁵ This was based on review of a series of 740 CSF spectrophotometric scans from four centres on CT-negative patients (27 of whom had NBA >0.007 AU and proceeded to angiography). However, little clinical information was provided on the patient cohort, in particular the certainty with which a diagnosis of SAH had subsequently been confirmed or excluded. This raises the possibility of both spectrum and verification bias which may affect the diagnostic performance of the test. Although it is commonly suggested that bilirubin persists in the CSF for up to 2 weeks, this has only been demonstrated for patients with a CT brain scan showing SAH¹⁴ and there is anecdotal evidence that aneurysmal SAH with lumbar puncture at day 8 may not reach the NBA threshold of 0.007 AU. ¹⁵ In this edition of the Annals, Birch et al. ¹⁶ investigate whether CSF protein concentration can aid interpretation. In a series of 132 patients with NBA >0.007 AU and NOA >0.02 AU, those patients in whom a diagnosis of SAH was not confirmed had the highest CSF protein concentrations. Although Birch et al. could not identify a CSF protein concentration above which SAH could be reliably excluded, they remind us of the importance of clinical setting for the interpretation of CSF spectrophotometry.

As highlighted by the NCEPOD survey, ¹⁰ future improvements in the diagnosis of SAH will rest heavily on clinical skills in suspecting SAH at an earlier stage with timely CT scanning. CSF spectrophotometry will continue to play an important role, but further work is needed to define the diagnostic performance of spectrophotometry in CT-negative patients.