The history in SAH is crucial. The onset of the headache must be specifically enquired about. The majority of haemorrhages occur with a severe headache, usually the worst of the patient's life, and are maximal within seconds.

Patients presenting in an acute confusional state may give no history of headache and careful witness accounts must be sought.

In patients with gradual onset, retro-orbital headache, always examine the pupils and eyelids carefully – a partial or complete III nerve palsy mandates that an expanding posterior communicating artery aneurysm is excluded. Discuss with neurosurgery, even if a plain CT scan is normal.

Some patients with SAH will present as physiologically unstable, with abnormal cardiac rhythms, such as atrial fibrillation, or with signs of myocardial ischaemia. It is important to separate the cardiac sequelae from the neurological presentation, as these events usually require no specific treatment, and antiplatelet or anticoagulant therapy may be disastrous.

Recent North American literature has suggested that in selected patients, a normal CT scan within 6h of symptom onset is sufficiently specific to exclude SAH without the requirement for lumbar puncture. This is not yet widely accepted, and usual UK neurosurgical practice would be to advise lumbar puncture in all patients in whom SAH is suspected, if CT is normal.

The obtaining, measurement, and interpretation of CSF bilirubin is prone to error and care must be taken. Samples should be obtained after 12h and analysed as soon as possible. The ‘three tube’ technique to assess for falling red blood cell counts as a method to distinguish traumatic tap from SAH is notoriously unreliable. If spectrophotometry is unavailable (e.g. overnight), lumbar puncture should be delayed until the biochemistry lab can perform the analysis. The presence of bilirubin indicates red cell breakdown products in the CSF. In a rapidly-analysed sample, this indicates blood in the CSF present prior to the lumbar puncture, that is, SAH. Spectrophotometry has replaced the visual inspection of CSF supernatant for xanthochromia.

A ‘traumatic tap’, where inadvertent venous contamination causes the CSF to be overly bloody, often results in an equivocal biochemistry analysis, as the large spectrophotometric peak produced by oxyhaemoglobin ‘masks’ any bilirubin peak which may be present. One option is to immediately repeat a lumbar puncture, if this can be done less than 6h since the first attempt; otherwise such patients should be discussed with neurosurgery.