Clinical Case 129: A SMACC in the Head (ache)


I need a little help with writing my talk for SMACC in Dublin this June – it is all about diagnosis in the ED.

So I want to present a simple case to you here…  and I want you to answer it honestly, without referring to any papers – I am after your “gut instinct” you “Gestalt” for lack of a better English word!  So please have a read of the scenario below and tell me the answer to the simple question by completing the poll

 Karen is a 45 year old accounts manager. She is usually well aside from occasional asthma and long-standing, mild depression.  Takes salbutamol and citalopram.

She presented to her GP this morning with a history of a bi-frontal headache which came on over a period of 15 minutes whilst at work at her computer.  The headache increased rapidly in intensity to “9/10” pain.  It is the “worst headache she has ever had”.  She felt nauseous after it peaked and nearly vomited.  She has had no syncope or light headedness.  She described some neck stiffness to her GP, but here are no objective findings of meningism on examination.

Her Neuro exam is normal with no changes on fundoscopy.

Her BP is 143/82 and she is in sinus rhythm at 80/min.  ECG Normal.

Her GP was worried about a possible subarachnoid haemorrhage (SAH) and sent her into your ED by ambulance.

By the time you have seen her the headache has been present for 3 hours. It has improved a little with paracetamol and a dose of ketorolac en route…

So that is all the data you will get.. now onto he poll:

[Please try and be as instinctual as possible, answer as if you were in your ED and had to guesstimate the answers to these questions]


I promise that I will get back to you with some poll results and data once I have mined your minds!  See you in Dublin!

OK – after a week and about 250 responses to the questions above I am posting the results:

Q1:  What is the pre-test probability based on the history given above?

pretest SAH

InterestingPOst test risk results with a wide spread of risk assessments and a double Bell curve emerging.

Two peaks around 8%  and one at about 33%.

OK – so that is the patient’s risk, but how risk averse are the Docs?  What were the responses to the question: “What post-testing risk level would you cease further investigation?”

This is a bit of a clearer Bell curve based around the 1% level – although there are a handful of very risk tolerant folk out there who would accept a 1/5 risk – I think that this may represent misunderstanding / a poorly worded question?

So – that is what we found…

comments welcome



  1. Pik Mukherji says

    Plugged in numbers that weren’t pre-looled up. But many prev decisions around this topic inform those responses, even without firm numbers in mind.
    Point being: if I read case and WOULD work up, I’ll make sure my pretest is far higher than post-test I’d accept and vice versa.

    If you look at #s, if #1 is more than 10x higher than #2, the clinician wants a workup. How accurate his actual #s are is more relevant to his a priori knowledge of base rate or his lack of numeracy. (ie. Does he really translate 0.5% into 1/200, or does his brain just think “that’s a low one.”)

    • Thanks Pik
      Agree. How many ER docs think about prior before embarking on work up ?

      Does it help inform the content of the “work up”?
      Do we have enough data to do this accurately?

  2. Hey Casey,

    That’s a really interesting survey, thanks.

    I think the problem is when people start taking data, eg the Perry paper, and start using it as a ‘one size fits all ‘. For example, as you point out, if our gestalt pretest probability is really high, even a negative ct within 6hrs still won’t get you to a low enough risk and i would argue that a LP is still warranted. I think that this is an important point that doesn’t get discussed much. Are you aware of any structured calculator of pretest probability based on history/exam?

    Cheers, Adam.

    • Hi Adam
      I wrote this patient case based on the mean data set of pts in the Perry paper. So aimed for average of 3000+ headaches worked up.
      Their SAH rate was 7.7% with another ~1% having dangerous ICH etc
      Agree there is a wide range of pretest risk, guesswork etc
      As far sai know we don’t have a validated score yet – but Perry et al published their models from a derivation study – so maybe an Ottawa SAH rule on the horizon !

      • Hi Casey,

        That’s really interesting. Especially that that patient was about average for the Perry study and that the pretest probability for a patient like that was 7%. Those numbers are really useful for guiding decision making…and even shared decision making!

        Happy Easter to you and everyone at Broome.

        Cheers, Adam.

  3. Geoffrey Menzies says

    I hope i am wearing something nice when the ACA reporter chases me down the street saying ‘Why didn’t you do your job properly, doctor?. Why did you throw her out of the ED when she was still obviously very sick?’.
    So even the 95% need a robust safety net. And if she comes back a second time, the post-test is going to need to be something very very low. Phone a friend, share the load.
    And that approach is actually somehow mostly also good for the patient, fortunately.

Speak Your Mind