Response

We thank Kate Shipman, Satheesh Ramalingam, Charlotte Dawson and Zhainab Yasear for their comments and recognise there is ongoing debate regarding the additional benefit of the traditional CT-lumbar puncture (LP) algorithm vs CT alone in excluding subarachnoid haemorrhage.¹

We agree that CT pick up of subarachnoid haemorrhage has improved significantly. It is however important to emphasise that studies reporting near 100% sensitivity and specificity were with CT performed within 6 hours of onset. In one recent UK study in routine practice, only ∼10% of patients were imaged within this timeframe.² Sensitivity falls with increasing delay to presentation and the importance of considering lumbar puncture correspondingly increases. Furthermore, detection of subarachnoid haemorrhage on CT imaging remains operator dependent. In routine practice, scans are generally not reported by an experienced neuroradiologist as in the majority of the published studies but rather by a trainee general radiologist, often out of hours.

We would therefore strongly caution against false reassurance from a negative CT report in a patient with a suggestive clinical history, particularly with a delayed presentation. It is for this reason that major international guidelines^3,4 continue to recommend CT/LP in cases with high clinical suspicion. However, we entirely agree that pre-test probability should always be carefully considered in evaluating the need for lumbar puncture after CT. Ease of access to CT has resulted in increasing numbers of patients being scanned without adequate phenotyping of the presenting headache. Subsequent rote application of guidelines in patients in whom SAH was in any case clinically unlikely pre-CT unfortunately results in too many unnecessary lumbar punctures being performed.