Clinical Case 107: Subarach Sans Scan

This case is inspired by a Twitter conversation started by Dr Brent May ( @DocBrent ) – he is an Anaesthetist and motorsport doc in Australia.

Brent asked the simple question: “SAH Q – no CT available. What is the “usefulness” of RBC count on LP? Lab “unable to do xanthochromia”.  This is a real problem in many hospitals – especially the smaller places with no CT or specific lab crew.  So lets look at a case and try to get to the bottom of this one!

You are working in a remote clinic with a small Emergency area.  It is a slow day.  You are 400 km from the closest hospital which has a CT and laboratory services.


Jan is a 45 year old local woman.  She presents with a bad headache – this is not usual for her, she is not usually a “headachy” person. No PMHx of migraine or other headache syndrome.  It woke her from sleep this AM – about 4 hours ago. Maximal intensity from the onset.  She says it has not responded to simple analgesia and resting at home.  She has had no syncope or neurological symptoms.  The headache has remained relatively constant.  She is complaining of neck stiffness – though has normal ROM in her neck.  She has no fever, vomiting or photophobia.

On examination she has no neurological signs.  Her obs are all normal, BP = 140/85, afebrile.  You cannot reproduce any objective neck stiffness.

So – first question.  Based on that clinical info – what would you estimate Jan’s “probability” of having a subarachnoid haemorrhage ?  Have a Gestalt guess… 2%, 5%, 10%, 50 %???

How low would you want your “probability” to be to avoid a transfer and subsequent work-up?  i.e.. what is your lower test threshold for SAH? 1:100,  1: 1000, less?

Here is a copy of Perry & Steill’s landmark paper in the BMJ 2011 – check out tables 1 & 2.  Jan is pretty much a “average Jan” for the cohort of 3000 pts in their paper.

About 7.7  % ended up having a subarachnoid bleed – and another 2 – 3 % had another non-benign cause for their headache.  So 10% is a reasonable “pretest probability” for badness in Jan’s case.

Now a 10% risk of serious intracranial pathology.  That would mandate a work up in most places.  Unfortunately your hospital only has rudimentary lab gear – you can do an LP and get red cell counts, but not xanthochromia.   A CT would be helpful if you can do it in less than 6 hours – but even if you called for transfer now – it would be well past 6 hours by the time Jan made it into the doughnut 400 km away.

So here we are.  You can do an LP and get red cell counts on the CSF.  Xanthochromia is probably not helpful as it is early (less than 12 hours) and simple visual inspection for yellow change is inaccurate. [Best read by non-colour blind female lab techs apparently!  Can you ask for that on the forms?]  Most Australian labs use automated spectroscopy to analyse for xanthochromia – and this is usually only available in larger, tertiary labs.

The accepted gold standard diagnostic features of SAH are variable – but the Canadian group who do a heap of research use the following:

“Subarachnoid haemorrhage was defined by:-

– subarachnoid blood on unenhanced computed tomography of the head,  OR

– xanthochromia in the cerebrospinal fluid,  OR

– red blood cells (>5×106/l) in the final sample of cerebrospinal fluid,

PLUS  an aneurysm or arteriovenous malformation evident on cerebral angiography.”

So now the big question.  Is there any way that you can effectively “rule out” SAH in Jan without transferring her 400 km to the next hospital?

Does a low CSF red cell count allow us to “exclude” SAH?  Can we use any other tests to decrease the risk for Jan?

Tough yet common problem in many rural areas in Australia.  I would love to hear your thought and comments below.  Does it depend on the timing?  Can we watch and wait if the initial LP is clear?  Or do we need to get imaging and a review for xanthochromia on all the Jan’s we see in ED?

If you are interested in the diagnostic dilemma that is the “SAH work up” then I highly recommend listening to David Newman’s SMART EM episode on the topic from way back in 2010. (WARNING: it is nearly 2 hours of podcast!)

Let me know what you think



  1. I am a bit rusty, but I know these types of situation come up on a fairly regular basis, even in small hospitals in the USA (i.e. cath labs, etc.)

    Especially, in the USA, if the patient has no money or insurance. Things may change a bit with Obamacare, but doubt it.

    At any rate, I would mildly medicate, and observe with bed rest in a dark room. Prepare as best you can for transfer, alerts, etc.

    After all, what is a neurosurgeon or neurologist going to do, if you rush this patient to the nearest facililty? Would they give Thrombolytics in the facility? It has been a while for me, and I don’t know Australia’s standard of care, or here too, matter of fact.

    But in my day, no thrombolytics were being given, and the few patients I did speak to about it, did not want to try it.

    john bennett md

  2. I think SAH diagnosis can be a dilemma but not in this situation. This patient needs to go to a bigger hospital. She could have had an ICH into an AVM / undiganosed tumour or secondary to HTN, and the LP might be negative in that case. And if she had a SAH, she needs a CTA to look for the source and literature shows that she might be better off in a bigger centre

    Careful with the CT within 6 hours as well. While our patient is “the average Jan”, the CT in our hospital is unlikely to be the “average CT ” in the BMJ study -- high resolution CT is not available in the great majority of Australian non-tertiary hospitals.

    As for red blood cell count, if xantochromia is not available, CSF needs to be collected in 3 containers. If the RBC decreases from tube 1 to 3 and is 0 in tube 3, SAH can be ruled out.

    • Great thoughts Ivy

      Thanks for the link to the paper showing high-volume centres do better. I imagine this is true of a lot of uncommon, critical diseases. In fact that is one of our main motivations for this blog -- to try and close the knowledge gap and hopefully deliver better initial care to these folk.

      How would it be different if Jan had a history of chronic daily headaches and this was her third presentation in a month with a headache?
      Tougher then?

      But I agree -- it is hard to get below the “safety threshold” on Jan without doing a CT and an LP. This will mandate a transfer either way.
      Also good point about CT technology -- we have a 16 slice CT and a radiologist 2000 km away -- usually reported by a trainee after hours -- not the same as the Perry cohort as described in their paper -- so apples and pineapples really!

  3. Ron Cassano says:

    Hi Casey
    Where I am now fits your scenario; I would estimate a rough 10% chance of a SAH in this otherwise well lady; so in order this is what I’d do:
    1/ Get retrieval organised first, as well as talk to neurological team at accepting hospital, given all obs are normal and GCS is 15
    2/ given the typical headache, I would treat as SAH, keeping BP in normal range probably around 145 systolic, and treat her headache symptomatically, with appropriate analgesics and anti emetics, and no NSAIDS or aspirin etc
    3/ I honestly don’t see the need for an urgent LP at all; your management won’t change, and you may cause very negative consequences
    4/ the difficulty comes if her headache abates significantly: I would still push for immediate transfer, given her history of this headache and no PH of migraine etc. As we all know this headache may simply be the precursor of the real thing, coming soon, so she definitely needs a CT, probably with contrast/angiography if non contrast is NAD; the big hurdle then is convincing the relatively junior neurosurg reg to accept her.
    How I might do that is probably not allowable in this blog but might include blackmail coercion and direct action, skills very necessary if you practice in remote Aus, to protect patients, as they would be looked after in the city.

  4. Patrick Linehan says:

    I think that you’re looking at this patient from the wrong perspective: you are limiting your diagnosis to yes/no subarachnoid haemorrhage and not considering the other diagnoses that patients with an acute severe headache could have. Admittedly SAH is the most common life-threatening diagnosis, but 90% of your patients won’t have it.

    In the 3132 patient cohort in the paper you quoted <> the final diagnoses were
    Benign headache and other benign causes 62%
    Migraine 20%
    Subarachnoid haemorrhage 8%
    Viral illness 3%
    Acute ischaemic stroke/TIA 2%
    Postcoital headache 2%
    Sinusitis 1%
    Syncope 1%
    Neck strain <1%
    Brain tumour with mass effect <1%
    Subdural haematoma <1%
    Intracerebral haemorrhage <1%
    Hypoglycaemia <1%
    Weakness—not determined <1%
    Bacterial meningitis <1%

    and we can probably add other rare diagnoses like cerebral sinus thrombosis, pituitary apoplexy, spontaneous intracranial hypotension to that list.

    The question to ask is "Would the results of an LP change my management of the patient going forward (e.g. avert a transfer, persuade me to tee up a CT angiogram or CT venogram if the plain CT is negative), and would the LP cause an adverse outcome to my patient if I do it before the transfer out for CT?"

    Suppose you do the LP. Let's assume you do best practice and use a tiny needle, that you can get cell count, protein, gram stain, cultures, visual inspection for xanthochromia, (and by the way, I can't find a study that compares visual and spectrophotometric measures of xanthochromia that measures any patient oriented outcomes rather than surrogate lab outcomes--please let me know what I am missing!) and measure an opening pressure.

    First: harm. I assert that in a otherwise well patient with an acute headache, and a normal neuro exam, that there is no serious harm from an LP apart from a post-dural puncture headache. (NEJM article that I don't have at hand right now)

    Lets start with a normal LP as this will be the most common finding:
    The best guess for sensitivity of this type of LP analysis for SAH that I can find comes from <> which found a sensitivity of 93% for patients with a normal CT (this is probably the lower bound for patients who have not yet had a CT because the patients in this study would have been biased towards having small bleeds). Starting with a pretest probability of 10%, a negative test would give the patient a post-test probability of <1%. Would you feel comfortable forgoing the CT after a negative test? If you did you could save 90% of your transfers.

    Next, what if you found red cells and/or xanthochromia? This would be the next most likely outcome. The specificity in the study was 95%, so I would strongly urge the radiology resident to go ahead with a CT angiogram while the patient is in the scanner, rather than have some faffing about to get someone to do the LP at the receiving hospital.

    Then you get into the rarities:
    What if you found a lot of white cells and no blood? If the results were suggestive of a viral or bacterial meningitis would you still send the patient for a CT?
    An elevated opening pressure and no blood? Maybe try to arrange a CT venogram to check for a cerebral venous thrombosis?

    Still, I think the main result of an LP first strategy would be a marked decrease in patient transfers without any significant misses.

    • Thanks Patrick
      I used the Perry paper to get a data set for SAH pretest probability. Most Docs I ask overestimate the risk based on “jan’s Hx etc”.

      I agree -- there are many other possible diagnoses -- but the data suggests SAH is the big fish and the others are Dx on CT as “bycatch” or later. So I think most ED docs are doing CT/ LP to look for SAH in the first instance.

      agree that the LP is safe in a pt with normal consciousness and no neuro findings prior to CT -- did a post on this a while ago here:

      Thanks for your insights.

  5. “It woke her from sleep this AM – about 4 hours ago. Maximal intensity from the onset”

    Shouldn’t that be intensity unchanged since waking rather than maximal at onset? This may seem a pedantic point but we don’t actually know when the headache started -- just that she woke with pain and it hasn’t changed since. If you believe that CT is 100% sensitive in the first 6 hours after onset of SAH then I think this is an important distinction. If she presented to a hospital with a CT scanner then some people would rule out a SAH after a normal scan if they argue that 4 hours since waking is the same as 4 hours since onset. On the other hand, even if you argue that we don’t know the time of onset, a normal CT would have to be followed by an LP (in my opinion) if you want to exclude a SAH.

    Just to get even further off topic, the point you make about the CT being read by a trainee is important. Where I woke, after hours scans are reported by radiology reg’s. Despite this, if the scan is normal many people seem happy to say that a SAH has been excluded. I think this is at least in part due to the fact that many people(myself included) dislike doing LP’s.

  6. Hi Casey interesting discussion.
    We have just had a new CT installed 64 slice I’m told.
    This necessitated a shut down of CT for ten days during which time we had a couple of Jans visit.
    The probabilities are up for debate but it’s clear there is a moderate risk the patient has a significant finding.

    It’s important to discuss this with Jan in some manner as her response will often guide you more than the exact clinical situation.

    I find it hard to believe that 10% of atypical/worst ever headaches that present to my department have major findings. But then maybe that was with the old doughnut.


    • Thanks Andre

      The Perry papers did look at “worst headache of life” and traditionally it is a pretty strong risk factor for SAH. Unfortunately in their data sets they found about 80% of the patients in their big series (82 in one and 78.5%) did describe their headache as “worst ever” so it was not a great discriminator.
      Interesting -- if a patient who never really gets headaches arrives -- it would be their “worst ever” regardless of severity if they are in ED -- so is it really better than having a “particularly severe headache?

      Then as a GP -- we all know that “worst ever” is often clouded by other factors -- stress, depression, relationships etc -- so it is a tricky one to call.


  7. there is a CT and LP in this womans future. it is written

  8. Great post.

    Rather than try to paraphrase a whole bunch of stuff, or cut & paste effusively, might I be rude enough to leave a couple of links to some articles I recently penned for our local ED registrar crew? (We have a currently private departmental website that will probably end up shared with the world in the nearish future… but for now I’ve cut & pasted out to a public WordPress blog for wider access):

    First post = general Dx approach to SAH, and addresses CSF xanthochromia in some detail.

    Second post = addresses RBC count in CSF specifically in some detail.


  9. Hi again Casey,

    Probably worth pointing out that formal spectrophotometric analysis of CSF for xanthochromia is absolutely _not_ the standard, or typical practice, in Australian pathology labs, even in tertiary centres (including mine). It’s visual all the way, baby. Which is craptacular.


Speak Your Mind