National guidelines for analysis of cerebrospinal fluid for bilirubin in suspected subarachnoid haemorrhage.

It is crucially important to detect subarachnoid haemorrhage (SAH) in all patients in whom it has occurred in order to select patients for angiography and preventative surgery. A computed tomography (CT) scan is positive in up to 98% of patients with SAH presenting within 12h but is positive in only 50% of patients presenting within 1 week. Cerebrospinal fluid (CSF) bilirubin spectrophotometry can be used to determine the need for angiography in those few CT-negative patients in whom clinical suspicion of a SAH remains high; it may remain positive for up to 2 weeks after the event. The lumbar puncture (LP) should only be performed >12h after the onset of presenting symptoms. Whenever possible, collect four sequential CSF specimens. Always ensure that the last CSF sample taken is sent for bilirubin analysis. Protect the CSF from light and avoid vacuum tube transport systems if possible. Always use spectrophotometry in preference to visual inspection. All CSF specimens are precious and should be analysed no matter how they were transported, where necessary with appropriate notice of the caveats regarding oxyhaemoglobin. Centrifuge the specimen at >2000 rpm for 5 min as soon as possible after receipt in the laboratory and in any case within 1 h of collection. Store the supernatant at 4 degrees C in the dark until analysis. An increase in CSF bilirubin is the key finding which supports the occurrence of SAH, but it is not specific for this. In most positive cases bilirubin will occur with oxyhaemoglobin. Oxyhaemoglobin occurring on its own is difficult to interpret and may be increased as a result of in vitro haemolysis of red cells introduced into the CSF during lumbar puncture. This process is exacerbated by vacuum tube transport systems. Results should be interpreted in the light of other investigations (e.g. if scan shows bilirubin then measure serum bilirubin and CSF oxyhaemoglobin and protein) and other confounding variables such as the time elapsed from presentation to LP.