Definitive angiographic detection of subarachnoid haemorrhage compared with laboratory assessment of intracranial bleed in CT-negative patients

Abstract

Background: Between 2·5% and 5% of cases of subarachnoid haemorrhage (SAH) give negative results for computed tomography (CT) scanning. Recent UK guidelines make recommendations as to standardization of laboratory methodology for the detection of SAH in individuals with a negative CT scan. We have compared this approach with our routine assessment using first-order derivative spectrophotometry, a recently reported iterative computer prediction model and measurement of cerebrospinal fluid (CSF) ferritin concentration. All methods were judged against cerebral angiography as the definitive means of determining patient outcome. Methods: Scanning spectrophotometry of CSF is necessary for both the UK guidelines and our in-house method. Absorbance measurements at 360, 405, 414 and 455 nm are required for the computer model. CSF ferritin concentration was measured using a serum method on a DPC Immulite 2000. This gave a value of <12 µg/L for normal CSF. Ethical approval and informed patient consent to additional investigation were obtained. Results: The sensitivity and specificity for the UK guidelines method, derivative spectrophotometry (in-house), iterative computer model and CSF ferritin were, respectively: 0·8, 1·0; 0·9, 1·0; 0·9, 0·5; 1·0, 0·78. Conclusion: First-order derivative spectrophotometry had best concordance with angiography. Combination of this with CSF ferritin measurement may improve sensitivity, although the zero-order method recommended in the UK guidelines for the detection of an intracranial bleed is more amenable and has comparable performance.