As a mathematician I am quite puzzled by the weird way that medicine understands itself as a science. But there are some basic techniques that can recover some of the scientific spirit in everyday life and one of them is probabilistic reasoning, of which the core is Bayes theorem. Now we had a session once that covered the intricacies of odds vs. probabilities in using Bayes theorem, but perhaps it is more in order to discuss the theorem itself, so that is what we did on Tuesday, using the case of D-dimer for sinus thrombosis, where a sensitivity of “only” 94% is used to discredit it’s use as a rule-out criterion both in the guidelines and in reviews by many german writers.

So consider any headache (as we now know, it usually is non-specific) in the ER and estimate the possibility of it being an SVT. Then use the sensitivity and specificity of the test to come up with post-test probability after D-dimer is negative and positive.



We concentrate on the non-classical neurocritical applications of transcranial b-picture sonography, i.e., everything apart from investigating the vessels for their own sake. Starting with a female patient with ICH due to sinus thrombosis, we discuss the neuroanatomy of the mesencephalic and diencephalic slices and consider

  • quantification of midline shift
  • screening for temporal or deep ICH
  • monitoring of ICP (using the pulsatility)


  • The Widder (in its recent new edition)
  • This article on using transcranial midline shift assessment in deep ICH


Imagine three scenarios:

  1. A happy minor stroke patient who is brought by the EMTs with nasal prongs and 4l/min O2, 8hrs after the beginning
  2. A major stroke patient, still intubated after endovascular recanalization of his right MCA embolic occlusion, ventilated with the “standard” FiO2 of 0,35
  3. A patient with a stroke and COPD, now exacerbated because of aspiration bronchitis, oxygen running at 5l/min, being drowsy

In all of these patients, oxygen therapy does or might play a role. So what is the evidence?

  1. Routine oxygen supplementation in (minor) stroke might be harmful – cf Rønning 1999 (interestingly, the current AHA stroke guidelines still recommend it for EMT personnel)
  2. While oxygen therapy might be beneficial in non-recanalized strokes (no proper evidence, though), after recanalization it might actually be harmful
  3. In COPD exacerbations (stroke or not), high flow oxygen therapy kills – as opposed to titrated oxagen therapy (with a goal of 88-90% SaO2)

So what is it with oxygen – is it good or bad? I think, that oxygen is a therapy that you have to carefully indicate and not shower over all your patients. The role of oxygen as the most important molecule for survival has been vastly overstated – in stroke, at least, oxygen is only one of the problems of the ischemic tissue. Remember that people dying with prolonged hypoxemia don’t have pathological changes in their brain. Also you can survive a pO2 of 7,5 without neurological damage.

In many medical emergencies, the use of oxygen is either not evidence based or proven to be harmful, as this article points out (this is in contrast to wound infections, where oxygen therapy might be quite helpful). For instance, in most states of reperfusion (such as after cardiac arrest, MI, or tPA/endovascular therapy), oxygen therapy should probably avoided, if possible. If recanalization is not feasible, improving oxygen supply should be improved if possible, but

oxygen supply = oxygen content x blood flow

Remember that local blood flow might be reduced by increased pO2. Also, if you really want to improve oxygen content, defined as

oxygen content = Hb-bound-O2 + O2-in-solution
= some-factor x Hb x SaO2 + 0,023 x pO2

so it should make sense to increase Hb just as much as it does to increase the SaO2 by 4%.

So what about your stroke unit patient, whose SaO2 runs below our standard range of 93%-100%? Where does the threshold of 93% come from? We don’t know. We arbitrarily chose one value to recognize impending gas exchange problems, so that they can be managed early, not because the stroke will become worse with 92 or 91%. If a nurse calls for that kind of desaturation alarm, you should go to your patient and investigate and repair the cause of O2 gas exchange problem, rather than just supplement O2. In fact, when you do supplement oxygen, it will take much longer to realize that the problem has become worse, because the time to alarm is prolonged.

So these are the core messages:

  • Oxygen is a therapy with benefits and harms – try to avoid it after recanalization
  • Desaturation indicates a problem to be solved rather than automatic oxygen therapy

Reference. Look at this great article for intensivists.

Cerebral hyponatremia: salt wasting syndrome vs SIADH

The story of neurological hyponatremia is probably as interesting as the management is complicated. While early reports (where side effects of therapy cannot have played a role) clearly depicted salt wasting, the discovery of SIADH left the salt wasting syndrome forgotten until rediscovered in the 80ies, not the least by Wijdicks, who then saw it everywhere.

Start out with any patient on your neurocritical ward, say a subarachnoid hemorrhage patient, who develops hyponatremia. Before you think, grab all data you can get, such as

  • history (fluid balance, speed of development, drugs, …)
  • examination (hypovolemia vs. eu- or hypervolemia)
  • osmolality, Na, K, Cl, uric acid, urea/BUN, creatinine, TSH, (cortisol level if in doubt)
  • urine osmo, Na, K, Cl, uric acid, urea/BUN, creatinine

Then open your favorite textbook, review article or any proper neurocritical care reference (such as those below) to arrive at a categorical diagnosis. Most of the time, it will be diuretic-induced, but in your Neuro patient it could also be neurological hyponatremia, meaning

  • we don’t know where it comes from,
  • it has to do with the patient’s disease (rather than our treatment) and
  • it looks like SIADH, but could also be CSWS.

The problem is that both diagnoses might be unstable over time (what looks like SIADH today, can waste salt tomorrow) and that the volume status is the arbiter between them, while clinical assessment of volume status is difficult at best – we all know how bad CVP is for deciding the volume status and all other physical signs fare even worse. And how do you judge the volume status of your septic right heart failed hypalbuminic swollen post-SAH patient?

In these difficult cases  I recommend  to just give some saline, say 500-1000 cc. Then give the patient some time (say 1 hour) and remeasure sodium – if it is not worse, your strategy was fine. Otherwise try the management of SIADH (which includes vaptans nowadays). By the way – fludrocortisone can be helpful in either case.

Of course, you have to remember the 0,5 mmol/h maximum rule – if Na rises more quickly, counter that with D5.

With this strategy I am not as tough as Sterns his in his block-diuresis-and-fill-up-the-sodium-slowly approach, as recently studied in this series (which has not been tried in Neuro patients).


A localization conundrum

Every respectable textbook mentions this, but it is still a wonder to see it (as we did last Saturday).

A 52 yo female is brought in by the Notarzt (EMS doc) with dense hemiplegia on the right. It seems that the patient just managed to call the EMS, before she collapsed; the Notarzt found her with paralysis of the right arm and leg, progressing to dense hemiplegia over the 20 minutes it took to reach the ER. Now the patient is restless, throws her head back and forth, right and left, crying that she is in pain, her arm is hurting and her leg is. She does not respond to any question, is in grave distress. After an opiate she becomes somnolent and whispers that the problem started 10 mins before she called the EMS (about 40 mins ago now) and then stops, obviously at peace with the world.

On examination she is somnolent, GCS 10 with intact cranial nerves. Her right arm is MRC 0, as is her right leg. She is also analgesic on the right arm and leg, though not on the trunk. The face is intact. On the left she has some degree of weakness at her leg, but develops enough strength to turn in her bed. The reflexes are pretty much down on the right, normal at the left arm, brisk at the left lower extremity, with some ankle clonus and Babinski’s on the left.

This is it. You can make a diagnosis out of that. There is no differential. Any ideas?

Hemicrania continua

Since on our ward we are dealing with strokes mainly, there is a tendency to forget basic neurologic knowledge, in particular in emergency settings. This happened recently in the following case: a 43 yo storage worker comes at 2 am into the ER, complaining about 4 months of unilateral headache, non side-shifting, with ugly exacerbations from his usual 4 to 8-9/10. Upon examination you discover a Horner syndrome and some tearing and noserunning. Interestingly he describes unilateral photophobia.

This is not a dissection, neither an SAH. As with all headaches history is the way to get to the truth, and maybe some treatment. Here are some important questions to ask:

  • Does the headache shift side?
  • Is it diurnally fixed?
  • Worse on lying than on standing?
  • Better with rest?
  • Loccally triggerable?
  • Visual disturbance?
  • Nausea/vomiting?
  • Photo-/phono-/osmo-/kinetophobia? Unilateral?
  • What triggers are there?
  • What drugs help? 

In all the trigeminoautonomic cephalagias, you have to get a picture (i.e. MRI)  eventually, but this should not preclude from treatment. Try oxygen 100%, triptans and indomethacin. Our case does have hemicrania continua, which by definition should react to this unique drug – but this is recently doubted.

References: in addition to the above article try the series of Peter Goadsby’s lectures on headache on YouTube.