Guidance is needed on which CT-negative patients should proceed to LP for exclusion of SAH

Dear Sir

As described by Goyale et al., ¹ the role of lumbar puncture (LP) and cerebrospinal fluid (CSF) analysis for visual xanthochromia (and in the UK also for spectrophotometric detection of bilirubin) in cases of suspected subarachnoid haemorrhage (SAH) with non-diagnostic brain imaging has been a matter of conjecture over recent years. It is useful to know that there are some cases of aneurysmal SAH that are still diagnosed by CSF spectrophotometry, allowing vital treatment.

However, I had some difficulty putting your results into a clinical perspective from my view point in the Emergency Department. Retrospective study design has predictable weaknesses. You identify that 660 CSF specimens were submitted for xanthochromic analysis during the study period, and of these 28 (4.2%) were suggestive of a diagnosis of SAH. Viral meningitis, in particular, is a frequent differential diagnosis in this group and one might question the retrospective exclusion of 10 of 28 patients. Even if the exclusion criteria were appropriate, they should also have been applied to the remaining 632 requests so as to achieve an accurate denominator (i.e. how many patients undergoing LP were actually suspected of SAH?)

While the study importantly demonstrates the identification of four cases of SAH requiring interventional treatment that would otherwise have been missed, part of the aim was to assess the overall role of LP and CSF analysis in the diagnosis of SAH. Of the 18 patients with CSF results suggestive of SAH, 11 were subsequently shown to be true positives, but in six of these the LP was unnecessary since the diagnosis had already been made (6 patients) or could have been made (1 patient) on review of initial or subsequent imaging. The false positive rate was 7/18 (39%) for SAH. Though I understand that there is added value to discovering other LP diagnoses such as meningitis, these patients are usually subject to vascular imaging with the potential to discover incidental intracranial aneurysms and raise further difficult clinical questions.

The possibility of SAH correlates well with abnormal neurology and altered conscious level and most of the debate around the routine need to perform an LP after a non-diagnostic computed tomography (CT) scan of the brain has been in neurologically pristine patients. Were the six patients diagnosed with SAH prior to the availability of LP results going to proceed to further imaging investigations in any case on account of concerning clinical features?

As your work demonstrates, specialist neuroradiology interpretation is important to avoid unnecessary LPs, but the presented results do not allow a full estimation of the performance of non-contrast CT in the diagnosis of SAH in your population. For this, we would need to know the number of CTs performed for exclusion of SAH and how many were positive. While most guidance still states that all CT-negative patients should continue to LP, it is clear that this investigation is not performed in all eligible patients. ² If the 10 patients you describe are true exclusions (the caveat regarding the accuracy of the denominator made above notwithstanding), CT in your population would have a negative predictive value of 639/643 or 99.4%; so about one patient in every 200 with a truly negative CT would benefit.

Lastly, with regard to the need for a 24-h spectrophotometry service, were there any adverse outcomes in those patients who waited increased lengths of times for a CSF result? If not then perhaps an ambulatory pathway for clinically suitable patients with visibly clear CSF on LP may be an option for centres with no 24 h access to spectrophotometry.

I think the take-home message is that there are still some patients who benefit from LP and CSF analysis in this clinical situation but that further work is required to delineate the characteristics of this population to minimize the number of LPs being performed unnecessarily. ³