Defining away sepsis

Why do we need new definitions for sepsis? We want the term to denote a syndrome with serious consequences, something that needs the full force of an ICU to recover.

No simple infections. So not a young girl with viral feverish URI (who happens to have leucocytes < 4). Neither a simple pneumonia. Maybe the myriad of negative sepsis studies stem from the fact that they included too many simple infections. What used to constitute severe sepsis, the endangered endorgans should become requisite in the definition of sepsis.

High sensitivity. What about the 90 yo dehydrated nursing home resident with positive nitrite stick and altered mental status? He might recover with some fluids and antibiotics, but then he also might not and deteriorate, so that he needs treatment response monitoring – he could be septic, but he also could just be cured in 2 hours.

I see the recent initiative to improve our definition of sepsis (dubbed Sepsis 3.0 by the FOAMED community) in this light. While the new definitions sound a bit esoteric, they are actually very practical, focussing on two situations:

  • Non-ICU: The normal ward, private practice or ER, where we need to recognize those patients with suspected infection that might deteriorate and need close monitoring, while ressources are not that good, that every patient can have an arterial blood or extensive lab works. Sepsis 3.0 recommends to use the qSOFA, also known as BAT score, where 2 out of 3 criteria suffice.
    • Blood pressure < 100
    • Altered mental status
    • Tachypnea 22 breaths/min
  • In ICU, we should screen with the daily labs we already have, using the SOFA score (routinely computed in electronic charts, also called CAR-LOG = coagulation-arterial pressure-renal-liver-oxygenation-GCS) and use a 2 point change as a threshold for sepsis, if infection is suspected. Septic shock is defined by need for noradrenaline to keep the MAP above 65 with remaining lactate geq 2 mmol/l, provided that enough fluids have been given.

Who is suspected? The beauty of the old and new definition of sepsis that “suspected infection” is not further specified. You might be warned by fever, CRP or leucocytosis, but also by a good story for an infection (postoperative patient, smelly urine). While some folks find this too vague, I think we always know when to suspect infection.

Think of sepsis when organs bail out. The article reminds us to think of sepsis (and thus suspect infection), whenever any organ failure occurs.





Hepatic encephalopathy in chronic and acute liver failure

The pathophysiology of hepatic encephalopathy is complex and not really well understood. We use the case of a neuro consult for an obtunded  hepatic failure patient to discuss the various (mostly hypothetical) components

  • Ammonia toxicity, leading to
  • intracerebral glutamine production in astrocytes, disrupting their function
  • Disruption of blood brain barrier
  • Increased cerebral blood flow (NO plays a role)
  • Inflammation (prostaglandins play a role)
  • Other neurotoxic agents (neurosteroids play a role)

As for triggers, we develop a differential for the usual reasons for worsening hepatic encephalopathy in cirrhosis, with infection/fever, dehydration, constipation, GI bleeding and electrolyte disturbances highest on the list.

The symptoms of HE are hard to understand, since most of the pathology takes place in the astrocyte rather than the neuron, but still you can easily list and grade them. It is important to stress that most of the symptoms and signs are not specific (not even asterixis!) and even a bright yellow patient is entitled to a proper differential diagnosis (think of, among others, subdurals, ICH, stroke, status, elyte disturbances, Wernicke’s, encephalitis) as well as some straightforward tests (CT/MRI, EEG, labs, maybe CSF).

With respect to therapy in the chronic case, we talk about the standard (lactulose or lactitole – some evidence, antibiotics – even less, probiotics – worse, diet – some hints) and the experimental approaches (indomethacin, sildenafil, …).

The final twist then is to understand the difference between two possible clinical scenarios:

  • Worsening (or first ever signs) of HE in a cirrhotic patient
  • Acute liver failure.

In the case of the latter, the astrocytes have no time to adapt to the bombardment by pathophysiological missiles, so hyperperfusion and rapidly increasing ICP lead to a fulminant clinical picture that requires immediate action:

  • Transfer to a transplant center (if you don’t happen to be in one)
  • ICP monitoring and control if grade > III
  • experimental bridging regimens (MARS, …) until liver transplant


  1. A greek review of the pathophysiology and standard treatment in the slow case 2011
  2. Raghavan and Marik’s great review of the critical care management of acute liver failure from 2006 Neurocritical care
  3. An update from 2013 on the latter topic

Monitoring midline shift

While the indication of decompressive craniotomy has become much harder now that everyone has been shown to profit, the prognostication of herniation in probably malignant MCA stroke has always been difficult.

The topic of yesterdays was the ultrasonographic depiction of third ventricle midline shift (mainly used in german centers where ultrasound wizards reside). But there are several steps involved before you can understand what you do there:

  • Depending on the location of the stroke either uncal, posterior transtentorial or anterior-posterior shift can occur – only half of them lead to lateral mass shift.
  • Clinical deterioration not only can come from lateral shift (as determined by the midline shift), but from rostrocaudal shift, compression of the ACA or PCA.
  • Lateral shift can occur at the level of the mesencephalon, the diencephalon or above, thus leading to divergence of CT determination of septal shift and ultrasound determination of third ventricle shift.

Still, if you can spare half of the CTs of an intubated and sedated patient, it should be worth it – so try ultrasound and gain experience.

Fluid responsiveness

Patients on non-intensive care wards (such as our stroke unit) are not as simple to understand as a patient with arterial and central line, with ultrasound and more high tech equipment around every corner.

So assume a lady with advanced heart failure, medium size stroke and a decent aspiration pneumonia, who is raised to your attention by a stroke nurse that is worried about the blood pressure of 78/28 mmHg with a heart rate of 109/min (sinus rhythm). What do we have by way of judging her fluid status? More precisely, is she going to benefit from a decent fluid bolus (say 500 ccs of NS or LR)?

From all evidence-based reviews on the history and physical, we know that a hemorrhagic patient might be identifiable by postural hypotension, tachycardia and postural dizziness. But this might be hard to check in a patient who is not able to sit or stand such as ours. Even more, both history and physical fare just as badly as flipping a coin (an AUC of the ROC curve of 56%), at least in the hands of an intensivist… Jugular venous pressure might be interesting for cardiological reasons, but the connection to right ventricular preload is difficult to judge as we don’t know the compliance of the system.

What remains are dynamic studies of responses to fluid challenges, such as

  • Passive leg raising: flip the legs up to 45° and the trunk down to 0°, effectively keeping the standard angle of legs to trunk of about 30-45°, pushing about 400 ccs of blood up to the heart.
  • Controlled respiration with volumes high above the usual aim of 6 ml/kg and with proper pressures (this yields vast changes in pressures, volumes and fluid compartmentalization around the heart and thorax)
  • Saline bolus of 500 ccs.

So what are the physiological variables to monitor during such a maneuver? In order of increasing complexity and high-tech-yness

  • Blood pressure (MAP), or even better pulse pressure (syst – diastol)
  • Heart rate (if the carotid sinus reflex is intact and Bezold-Jarisch/Bainbridge keep quiet)
  • Perfusion measurement of your oxygen saturation monitor (reflecting stroke volume in the periphery, which is worse than central stroke volume, but better than most measures)
  • etCO2, if you happen to monitor it, reflecting minute volume
  • PPV or SVV if measured by pulse contour analysis (PICCO = LIDCO, both better than Vigileo)
  • Inferior vena cava diameter and collapsibility index
  • Stroke volume (if measured by TEE or pulmonary catheter)
  • Femoral artery (better carotid artery, but this hasn’t been studied) mean systolic velocity
  • NICOM (thoracic impedance boosted)
  • … many more devices I have never used

So on any normal ward with – say – a simple vital sign monitor, us Neurologists can do best with ultrasound of the IVC and the carotid artery and the pulse pressure plus (perhaps) heart rate during passive leg raise. But this isn’t too bad.


Delicious delirium

The screaming patient on a very empty cabrio stroke unit – he stripped himself of all monitoring equipment, infusion. Obviously, his hemiplegia is also much better.

You can judge the importance of the topic from the fact, that this is the second time in a year we take it up. This is by request by one of our blog followers, so I want to use the opportunity to encourage anyone (both readers in particular :-)) to enlist suggestions for future talks.

The problem with delirium is that you should spend more time in recognizing the beginning and preventing it than you should thinking about therapy. Still, that doesn’t help at 2 am when your average 84 yo nursing home resident starts beating up your stroke nurse and you have to react. So we try to cover the basics in a rush as in this overview and then concentrate on hardcore psychopharmacology – how do you apply which substance in what patient?

I recommend distinguishing the following categories:

  • Blue alert – emergency self defense: a violent dangerous patient
  • Red alert – preventing major harm – an agitated hyperactive patient who could harm himself (or has already) by falling out of bed or pulling tubes
  • Yellow alert: infectious delirium – a restless patient who screams on your ward the whole night (this tends to turn other patient delirious)
  • Disturbed sleep rhythm with sundowning
  • Quite coherent, but hallucinating
  • Impeding delirium: slightly confused

As for references, there are recent reviews, even free and by a respectable Neurologist, but I can’t really recommend anything except uptodate. And of course the last blog entry on delirium.

Management of raised intracranial pressure

We have talked about ICP is connected to tissue shift, how and why you would measure it and what the problem of raised ICP is (reduced perfusion, tissue shift). Today, we focused on the basic management aspects.

In my humble experience, you should distinguish between basic measures (for every patient) and advanced interventions that can only be applied in specific situations.

Basic measures

The rule is: keep everything normal that might increase ICP. To this end you have to measure quite a lot.

  • Monitor blood pressure, preferrably arterial, and keep blood pressure normal and avoid blood pressure excursions, because too high BP leads to vasoconstriction and ischemia and too low pressure might lead to vasodilation (increases ICP) and reverse steal phenomena.
  • Monitor temperature and avoid fever. Hypothermia is acceptable as long as bodily counter measures can be mastered (in particular: shivering).
  • Monitor pCO2 (preferrably by arterial blood gases) and try keep it in the lower half of normal (35-40 mmHg) – the rationale is as in blood pressure.
  • Monitor SaO2 (keep it in the normal range, in the nineties) and pO2 and avoid too low saturation (ischemia!) and too high pO2 (has bad effects, at least in traumatic brain injury).
  • Monitor pH (and thus HCO3) – rationale as in pCO2, but influenced by the kidney and a lot of the other stuff.
  • Monitor hydration and input/output, keep it neutral (not too dry, not too wet) – need a Foley for that.
  • Monitor electrolytes (Na, K, Cl, Mg, PO4) and correct them, in particular, when complications occur
  • Monitor and avoid raised central venous pressure: keep the head up and straight (e.g. 30° – no real evidence out there), correct central venous pressure, avoid abdominal pressure, prefer subclavian instead of jugular vein cannulation.
  • Monitor glucose and keep it in the normal range.
  • Avoid seizures. If in doubt, treat. Eg with phenytoin (might reduce ICP) or even better topiramate or zonisamid (both probably reduce ICP by reducing CSF production).
  • Avoid pain
  • Avoid stress: minimally invasive nursing, keep stress-by-relatives to a minimum, sedate if necessary.
  • Avoid obstipation: stool softeners
  • If working down this list requires sedation or intubation at any point, do so liberally.

Advanced ICP therapy is best described in 3 tiers of increasing invasiveness. All of these measures are controversial or at lest quite involved. For most of them the  physiologic effect is unclear or unpredictable.

  1. Osmotherapy (mannitol or hypertonic saline, if hypotensive) – need to monitor osmolarity – 1-3d effects at most (good for bridging the maximum swelling of a lesion)
  2. Hyperventilation – few hours effects at most (e.g. prep)
    Deep sedation (barbiturates or propofol or inhalative agents)
  3. Therapeutic hypothermia
    Bilateral craniectomy

Intracranial pressure

A 37 yo woman has extensive thrombosis (sinus sagittalis superior, rectus, transversus right), some lesions and bleeding, being somnolent and very hard to take a history. Suddenly she becomes agitated, complains about horrible headache, is restless, fidgety and simultaneously less vigilant, develops an abducens palsy.

We hypothesize that she has a plateau wave of increased intracranial pressure – so called A waves and thus discuss the theory behind these.

Here is the short version:

  • Raised ICP is (per se) pretty harmless – the brain works less, but is not really destroyed by raised pressure, UNLESS it’s perfusion decreases.
  • The symptoms of raised ICP (without reduced perfusion) are: headache, symptoms of papilledema (visual disturbance, field defects, …), somnolence, abducens palsies, the Cushing response.
  • It is not really clear where you should measure the ICP. The most reliable place is the ventricles (because the CSF equalizes the various pressures around the ventricles), yet it does not mirror the point of maximum pressure. The primary lesion (e.g., in traumatic brain injury) is not really helpful, because it is destroyed already. So we often chose the perilesional parenchyma, where pressure might counteract perfusion and thus damage tissue through ischemia. Finally, you can measure in the lumbar thecal sac – this can be equivalent to intraventricular pressure if no communication disturbance exists.
  • Plateau waves can arise from anywhere in the vicious circle     decreased CPP -> vasodilation -> ICP increase -> decreased CPP. Often, a decrease in arterial pressure causes them. They can lead to tissue ischemia.
  • You can treat plateau waves (or avoid them) by keeping the blood pressure up, reduce vasodilation (e.g. hyperventilation) or ICP (e.g. mannitol) or CSF (e.g. by carbonic anhydrase inhibitors or lumbar drainage). I am surprised as to how few people think of increasing blood pressure to break through them.

References. This is mostly folklore and can be found in many textbooks on neurointensive care.