Patients presenting with sudden onset of headache, must have a subarachnoid haemorrhage(SAH) ruled out. If missed, the potential consequences of a rebleed can be devastating. The warning leak or sentinel headache is a window of opportunity to make the diagnosis. The approach so far, is to perform a non-contrast CT head. If this is normal, then a lumbar puncture must be performed and tested for xanthochromia, or  red cell numbers counted. Neither of these techniques is totally accurate.

 What if we could rule out SAH with a CT?

That’s right! Rule out, go home! Can the new 3rd generation CT scanners(>250 detectors) do this?

Perry, Stiell et al, from Ottowa, BMJ 2011;343:d4277propose that a normal CT brain with a third generation scanner, when performed within 6 hours of headache onset, rules out SAH. Let’s look at this study. Click the link above and read it for yourselves.

It was a prospective multicentre cohort study. Patients selected had a non-traumatic acute headache and GCS of 15. SAH was ruled in if the patient had: subarachnoid blood, xanthochromia in the CSF, >5×106 red cells in the final tube AND an aneurysm identified on CTA or MRA.

The clinician was permitted to organise investigations as they normally would.

3132 patients were recruited, 240 patients had a SAH, 911 patients had a normal CT and no red blood cells in the CSF.

The conclusions of the study were:

Overall performance of the CT

sensitivity of 92.9%, specificity of 100%

negative predictive value of 99.4%, positive predictive value of 100%

Performance of CT when done <6 hours from headache onset

sensitivity 100%, specificity 100%

negative predictive value 100%, positive predictive value 100%

The trend here is very promising. However the study needs some work. Looking more closely, not all patients had a lumbar puncture following a normal CT. This was at the discretion of the clinician. These patients(n=1931) had to be followed up at 6 months, to ensure no deaths. 50 of these patients were lost to followup. Of these 50, 13 had had a CT scan within 6 hours. If any of these had a subsequent SAH it certainly would affect those performance outcomes.

My conclusion, is that although this study is incredibly promising, it won’t necessarily change my practice just yet. What we currently use is not totally accurate. Cerebrospinal fluid when assessed for xanthochromia by visual inspection is unreliable and when assessed by spectrophotometry, can result in false positives. Perhaps this new approach is no worst than these. The current limitation is availability of  third generation scanners. I’m not sure that they exist in institutions apart from the major tertiary centres at present.

Great potential and we need another study and some validation.

What do you think?

Peter Kas