Skip to main content

One of the things I strongly dislike, is when a patient comes into the emergency department with a complaint of headache and it has features of a subarachnoid haemorrhage(SAH), and I just dont know if it is or not, so I err on the side of being careful and CT and LP.

When we read the studies, a SAH doesn’t have to be ‘the worst headache’ that patient has ever had, nor does it have to come on instantly. I used to ask, “is it really bad?” and “Did it come on…like…BANG,  being hit on the back of the head with a shovel?” I still ask these questions, but am less likely to rule out SAH. I will talk more about this at the EMCORE.

In fact, we miss about 5-6% of SAH in the emergency department (Vermeulen et al Stroke 2007;38:1216-21). This isn’t great, as about 50% of patients with Grade I SAH die and of the remaining patients, only 50% get back to their previous level of functioning.

So, first and worst, doesn’t count. What can we use? Linn et al(JNeurol Neurosurgand Psychaiatry 1998;65:791-93), tell us that if we are suspicious of a headache and it is new, and  has onset within 5 minutes and lasts longer than an hour, it should concern us and we should investigate the patient for a SAH.

Now remember, we aren’t talking about the patient with a depressed level of consciousness, or neurology. Those patients you’re going to scan. We are talking about the patient with headache as their only complaint.

Well here is  study in the BMJ(2010) by Perry et al. It was a prospective, multicentred, cohort study over 5 years. It enrolled 1999 patients with a non-traumatic headache. The study identified markers that could be used to determine a clinical rule with 100%sensitivity. I will say at the outset that this study had some significant limitations, one of which was that 1050 patients identified as eligible were not included in the study.

They came up with 3 clinical rules, with grouped signs and symptoms. These need to be validated. The total of all ‘red flags’, in a patient presenting with a rapid onset of headache were:

Any loss of consciousness, Stiff neckAge > 40, Came on during exertionPatient was brought to the ED by ambulance, VomitingHigh blood pressure > 160systolic, or >100 diastolic

Remember, none of these have been validated yet but interesting.

A few points on investigation

The approach, unfortunately,  is still Non-contrast CT brain, if normal, followed by lumbar puncture, if we are suspicious of a SAH, until a validated rule is found. The key is when to do the lumbar puncture. Some people wait for 12 hours to do an LP, primarily waiting for xanthochromia to develop, ie., the patient is there, waiting, for 12 hours. The reality is that, the sample is usually looked at under a light by the lab tech, who makes a visual judgement on whether the sample looks clear. Very few centres have spectrophotometers, and the reason is, that the band-width is set at such a level, so that no SAH’s are missed, that there is a significant false positive rate. The other concern is that there are a number of ultra-early re-bleeds that occur, yes, within the 12 hours.

Due to this, red cell analysis has been used. There needs to be adequate time (1-2 hours)for red cells to track down the canal, however, following this, red cells may help. The arguments about decreasing red cell count and our inability to judge traumatic taps is real, however, the red cell approach is receiving greater acceptance. Is it better than the guy in the basement, looking at the sample against a fluro light? I’m not sure.

The reality, is that in most cases, if you are concerned, you will still discuss this patient with the neurosurgery unit anyway. Now, if you think that will solve all problems, then you’re wrong. I don’t know how many variations of approaches I have encountered over the years. It depends on who I speak to, but they range form “Transfer for MRI”, to “Do a CT angiogram”, to “Send the patient home for outpatient MRI”, to “Can you pick up my laundry, I’m stuck in theatre and I’m a neurosurgeon and then I’ll tell you”, to I do’t know what.

The evidence is the evidence and I’m getting it all together. I’m presenting it at EMCORE and we’ll get a video of that lecture up soon afterwords.



  • mthomas says:

    Great article again Peter.

    Although not an expert, I feel that CT cerebral angiogram warrants consideration as an immediate investigation for suspicious headache (particularly if you plan to do a non-contrast CT brain anyway).

    Using the data from the quoted Perry study and some additional information, there is potentially a significant reason to consider it.

    Consider the following thought experiment:

    Perry found 130 SAH cases in a total of 1999 enrolled patients with headache. [This number is possibly an underestimate since they did not perform CT/LP on all patients.]

    From other data we know at least 70% of SAH cases are due to ruptured aneurism (conservative – some sources quote 85%). This would imply about 91 SAH cases with the “smoking gun” aneurism present.

    [I’m assuming that a ruptured aneurism would be visible on CTA, which might not necessarily be true.]

    From another study, the baseline prevalence of asymptomatic aneurisms is 0.4% in the community, so the remaining 1869 non-SAH patients in the Perry study would contain an estimated 7.5 patients with a silent aneurism, and 1861.5 without aneurism.

    Therefore in suspicious headache, an aneurism visible on CTA has a 91/98.5 chance of being associated with SAH (=92.4%).

    If there were no aneurism on CTA, the chance of the patient having a SAH is not zero, but 39/1861.5 (=2%). Which is substantially below the overall baseline SAH risk in the Perry cohort (approx 6%).

    So if this thought experiment was borne out by reality, CTAs could have a very significant utility in SAH diagnosis.

    A Positive Predictive Value of 92.4%, and Negative Predictive Value of 2% is a far greater spread than most other diagnostic tests in the whole of medicine!

    Doing CTA would avoid the 12 hour wait for xanthochromia study (plus the extra wait time for pathology to process the result).

    The only problem would be the radiation dose if CTAs were used more widely for headaches (instead of LP).

    The Perry clinical rules (if validated) might be a useful adjunct to “pre-screen” patients in order to reduce the number of unecessary scans.

    Seems like an excellent question for a research study!

    • Resus says:

      Great points and a good study topic. In fact, when I’m working in the very remote rural sector and I have a high clinical suspicion for a subarachnoid haemorrhage and we have a scanner, I will ask for the CT angiogram. I have still done an LP on the patient looking for RBC count. In every case, every case, the neurosurgical team on at the teaching hospital, have asked for a CTA. It helps them, it seems make the decision.

Leave a Reply