Recently Geert Meyfroidt published an article in Intensive Care Medicine describing ten false beliefs in neurocritical care shown here:
It's a great article, but I think they could have been more aggressive about challenging neurocritical care dogmas (1). In response, here is a list of ten dubious beliefs that goes farther to challenge the status quo. To avoid being overly negative, this is a mix of classic dogmas and some reasonable ideas which may be outdated.
Dubious Belief #10: Patients with Myasthenia Gravis or Gillian-Barre Syndrome require frequent monitoring of pulmonary function tests. They should be intubated prophylactically based on vital capacity and/or negative inspiratory force (NIF).
Newer concept: Bedside pulmonary function tests are notoriously effort-dependent. The decision to intubate should be based on clinical judgement and respiratory failure, not arbitrary cutoff values of vital capacity/NIF.
Pulmonary function tests can be used to evaluate the diaphragmatic strength of patients with muscle weakness (e.g. Myasthenia Gravis or Guillain-Barre Syndrome). Unfortunately, these tests are extremely effort-dependent, causing them to fluctuate quite a bit. Furthermore, patients with bulbar muscle weakness may be unable to maintain a mouth seal required to perform the tests well. Pulmonary function tests may have some utility in triage (determining which patients require ICU admission) and in following patient trajectory. However, the decision to intubate should not be dictated by a single test result. Instead, this decision is a clinical judgement based on the patient’s overall trajectory and clinical examination (e.g. tachypnea, cough strength, oxygenation, work of breathing, use of accessory muscles, chest radiography). Repeating pulmonary function tests frequently can disrupt patient sleep, unintentionally promote respiratory fatigue, and provide spurious results due to variable patient effort.
Dubious Belief #9: 3% saline must be infused into a central vein.
Newer Concept: 3% saline is safe for peripheral administration through a well-functioning IV line.
Traditional pharmacy dogma is that any solution with osmolarity >900 mOsm must be given via central line to avoid thrombophlebitis or skin necrosis following extravasation. 3% saline has an osmolarity of 1026 mOsm, so many textbooks have recommended central administration. However, this is a gross oversimplification: the tendency of a solution to cause phlebitis or skin necrosis depends on numerous properties besides osmolarity (e.g. pH, chemical composition).
Clinical evidence doesn’t support the need for central infusion. There don’t appear to be any reports of 3% saline causing skin necrosis. Jones 2017 reported a series of 157 patients treated with peripheral 3% saline, with six extravasation events but no consequent skin necrosis. A recent comparative study showed that peripheral infusion of 3% saline at rates up to 100 ml/hr didn’t affect the rate of catheter phlebitis when compared to catheters used for routine therapies (Meng 2018). Other authors have also reported the use of peripheral 3% saline safely (Dillon 2018 and additional references curated by Scott Weingart here)(2).
Dubious Belief #8: Prophylactic infusion of 3% saline magically prevents badness from occurring, in any situation involving brain edema.
Newer Concept: Prophylactic infusion of 3% saline to patients at risk for worsening edema has numerous potential drawbacks, without offering any proven benefit.
Imagine a patient with a large ischemic stroke, with a risk of worsening edema over time. There are roughly two strategies for use of hypertonic saline:
Prophylactic hypertonic saline infusion strategy:
- Start hypertonic saline pre-emptively.
- Raise the sodium to an arbitrary level (e.g. 155 mEq/L).
- After the patient starts to improve, allow the sodium to drop back to normal.
- Follow the patient clinically.
- If critical deterioration occurs, provide boluses of hypertonic saline.
- Use hypertonic therapy as a temporary bridge to definitive treatment (e.g. decompressive craniectomy).
- After definitive therapy has been provided, withdraw hypertonic therapy.
There are numerous drawbacks to prophylactic infusion of hypertonic saline:
- Neurons rapidly adapt to increased tonicity. If the patient deteriorates further after the sodium has been raised, there is little room to increase the sodium further.
- After the patient has started improving, lowering the sodium to normal exposes them to a risk of rebound edema and deterioration. Sodium normalization usually occurs after the initial crisis has passed, at a time when practitioners’ attention to the patient has often slackened.
- Hypertonic infusions may cause volume overload, pulmonary edema, acute kidney injury, and hyperchloremic metabolic acidosis.
Overall there is no prospective evidence supporting the use of prophylactic hypertonic infusions (unlike bolus therapy, which is supported by better evidence). There are numerous drawbacks to prophylactic hypertonic infusions, with most modern reviews recommending against this strategy (Shah 2016,Diringer 2016).
Dubious Belief #7: Mechanical cooling devices are the essential intervention for temperature control after anoxic brain injury.
Newer Concept: Pharmacologic temperature control plays an essential role to prevent shivering, avoid the need for sedatives/paralytics, and wean patients off mechanical cooling devices.
Mechanical cooling strategies are the most dramatic (most commonly an external adaptive cooling device such as the Arctic Sun, which costs ~$900 per patient)(3). Most discussions of therapeutic temperature management contain an extensive discussion of different mechanical devices, while paying little attention to antipyretic medications.
However, pharmacological temperature control is an enormously valuable component of therapeutic temperature management. A reasonable prophylactic regimen is acetaminophen 1 gram scheduled q6hr, buspirone 30 mg q8-12hr, and prednisone ~60 mg/day with gradual taper (yes, steroid is recommended for post-arrest patients by 2018 ESICM/SCCM guidelines)(4). An aggressive antipyretic regimen achieves two purposes:
- Prevention of shivering. Shivering may cause hemodynamic instability. Even more problematic, treatment of shivering often involves the use of opioids, paralytics, and long-acting sedatives, which can prolong the duration of ventilation and delay neuroprognostication.
- Prevention of rebound fever. After mechanical cooling is discontinued, patients still have brain injury leaving them at risk of harm from fever. A robust antipyretic regimen may help prevent rebound fever for several days beyond discontinuation of mechanical cooling.
Mechanical cooling devices are undoubtedly essential. However, in my opinion the goal should ideally be to prevent fever pharmacologically. The mechanical cooling device should be considered a second line of defense against fever, rather than the primary mode of thermoregulation.
Dubious Belief #6: It is impossible to assess fever in a patient who is being managed with therapeutic temperature management.
Newer Concept: Water bath temperature may provide a window of insight into the patient’s thermoregulation for patients maintained at 36C. Very low water bath temperature could be regarded as a fever-surrogate.
It is widely believed that fever cannot be assessed in a patient undergoing therapeutic temperature management. This might not be entirely true. Patients undergoing therapeutic temperature management at 36C are essentially being forced to be normothermic. The temperature of the water bath of an external cooling device is a measurement of the amount of “work” required to keep the patient at this temperature. For example, if the water bath temperature is above 36C, the patient’s temperature is falling and the bath is warming them up (a poor prognostic sign). If the water bath temperature is unusually low (e.g. it can drop to 4C), this indicates that the patient is “trying” to spike a fever. Very low water bath temperature might be a surrogate for fever, especially if this occurs despite the use of antipyretic medications (see #7 above). Low water bath temperature is also clinically important when discontinuing mechanical cooling, because it predicts that the patient will immediately spike a fever after disconnection (a potentially dangerous event, which could make a case for continued treatment with mechanical cooling and intensification of medical antipyretic therapies)(5).
Dubious Belief #5: Prognosis regarding anoxic brain injury cannot be assessed and shouldn’t be discussed with the family before 72 hours.
Newer Concept: Prognosis regarding anoxic brain injury can often be assessed and should be discussed with family within 24-48 hours in an exploratory fashion.
A popular dogma is that nothing can be said about prognostication following anoxic brain injury until the magical 72-hour timepoint. This isn’t true currently. In the past, prognostication was delayed by the use of therapeutic hypothermia to 33C. This typically involved the administration of paralytics, opioids, and sometimes benzodiazepines, all of which would prohibit a reliable clinical or electroencephalographic evaluation. Rewarming patients and allowing these drugs to wear off took a long time. Currently, most patients can be maintained at 36C without the use of any long-acting sedatives (ideally with absolute avoidance of opioids, benzodiazepines, paralytics). Propofol is often used for sedation, which can be discontinued briefly to facilitate reliable neurologic examination and EEG readings. This allows for more timely, accurate neuroprognostication while simultaneously facilitating early liberation from the ventilator (6).
Poor prognosis is likely in patients with frank edema and loss of grey-white matter differentiation on CT scan or unequivocal early-onset status myoclonus (Rossetti 2016). Emerging evidence shows that EEG patterns at 24-48 hours may be strongly prognostic of poor outcome, as well (e.g. isoelectric, low-voltage, or burst-suppression patterns) (Muhlhofer 2018). Absence of pupillary reflexes or motor response to pain at 24-48 hours is likewise worrisome. If early indicators suggest a poor prognosis, a preliminary discussion with the family after 24-48 hours may help them prepare for this eventuality. Waiting at least 72 hours before withdrawal is generally advisable, but the family should at least be prepared for a poor outcome throughout this period.
Dubious Belief #4: A patient with Glascow Coma Scale (GCS) below 8 must be intubated. (GCS below eight, intubate!)
Newer Concept: Glascow coma scale doesn’t reliably assess the patient’s ability to protect their airway. Patients with a very low GCS may not require intubation.
The Glasgow Coma Scale is a blunt instrument which is primarily useful for harassing medical students. Several studies show that poisoned patients with low GCS may do fine without intubation, even with GCS as low as 3 (Duncan 2009; El-Sarnagawy 2016; Donald 2009). Why? Airway protection depends on swallowing and coughing reflexes which are mediated by the brainstem (Nishino 2012). Such reflexes may still be intact despite the presence of depressed consciousness from global dysfunction of the cerebral cortex. Overall, the decision to intubate is complex, requiring an integration of the overall clinical scenario, the patient’s trajectory, and their ability to handle secretions. When in doubt, intensive monitoring with serial re-evaluation may be the most reasonable approach.
Dubious Belief #3: The gag reflex is an essential component of the neurologic exam.
Newer Concept: The gag reflex is frequently misleading and poorly evidence-based.
Roughly 30% of normal people lack a gag reflex, making this is a nonspecific test. The test is operator-dependent, with variations based on where stimulation is applied and how strongly. In patients with altered mental status, gagging may stimulate vomiting with a consequent aspiration risk. Finally, presence of a gag reflex is unreliable for determining whether the patient is at risk for aspiration (protection from aspiration requires an intact swallow reflex, an entirely different beast). Overall, this is an unreliable test that isn’t supported by evidence and should be abandoned.
Dubious Belief #2: Convulsive status epilepticus should be treated with an anti-epileptic infusion only after failing to respond to three conventional anti-epileptic agents.
Newer Concept: Convulsive status epilepticus is a medical emergency which should be lysed in under 20-30 minutes. Prompt control of status epilepticus will often require intubation and anti-epileptic infusion after failing benzodiazepine.
The traditional approach to status epilepticus is only to intubate and start an antiepileptic infusion after the patient has failed to respond to three conventional anti-epileptic agents (e.g. benzodiazepine, phenytoin, and phenobarbital). This strategy was codified within guidelines in 1993 that have been faithfully passed down through the medical literature for over two decades. Nearly identical versions of this strategy can be seen in current critical care textbooks (e.g. Irwin & Rippe 2018).
The problem is that generalized convulsive status epilepticus may cause severe complications if allowed to continue for over 20-30 minutes (e.g. aspiration, hyperthermia, neuronal death, hyperkalemia, rhabdomyolysis, myocardial infarction). Thus, there simply isn’t enough time to try three different antiepileptic agents. In fact, it’s often logistically impossible to administer two anti-epileptic agents within thirty minutes (e.g. lorazepam and levetiracetam). An accelerated algorithm for seizure termination is shown below and explored further here.
Newer Concept: The benefit of systemic thrombolysis is debatable. Given that thrombolysis increases short-term mortality, the risk/benefit ratio is dubious.
Evidence regarding stroke thrombolysis suffers from considerable optimism bias: attention is showered on positive studies, whereas negative or neutral studies get ignored. Indeed, most studies have been neutral or were stopped early due to harm (table above). Positive studies are often fragile, potentially reflecting random chance (NINDS-II has a fragility index of 3, ECASS-III has a fragility index of 1). NINDS-II is the landmark positive study, but its results are less impressive if baseline imbalances between different groups are accounted for (Hoffman 2009). The significance of this mixed collection of studies remains hotly debated: is thrombolysis beneficial or harmful?
The evolving research strategy seems to ignore the scientific method. According to the scientific method, if a study is dubious then it must be replicated. The importance of replication is emphasized by many “landmark” trials which were enthusiastically incorporated into mainstream treatment guidelines but later failed replication (e.g. tight glycemic control, activated protein C for sepsis). Given methodologic limitations of NINDS-II, this trial truly requires replication (Hoffman 2009). However, attempting to replicate NINDS-II would deeply challenge the status quo of stroke neurology. Consequently, investigators have shifted to a “pharma method” that involves repeatedly testing thrombolysis for new indications. The result of this process is an ever-expanding number of indications, with each indication supported by a single un-replicated study (0-3 hour window = NINDS-II; 3-4.5 hour window = ECASS-III; wake-up stroke = WAKE-UP). This isn't science – it's a house of cards.
Thrombolysis increases short-term mortality and perhaps long-term mortality too (figure above). This results from causing a fatal intracranial hemorrhage in ~3% of patients. As discussed recently, all-cause mortality is a difficult endpoint to ever affect with any intervention. For example, nearly all treatments used in critical care have no measurable effect on all-cause mortality. Zero. Thus, the fact that thrombolysis was able to worsen all-cause mortality is impressively bad (even if this only reaches statistical significance at a single timepoint). Furthermore, the performance of any therapy in everyday practice is usually worse than its performance in a structured clinical trial – so thrombolysis is probably more dangerous in real life.
It is impossible to justify an intervention proven to increase all-cause mortality, unless there is incontrovertible proof of some other extraordinary benefit that outweighs the excessive mortality. Such proof seems to be lacking. In the history of medicine, other treatments shown to increase all-cause mortality have eventually been banished from clinical practice. Thrombolysis for stroke is the only medical therapy which is commonly and intentionally used, despite evidence of increasing all-cause mortality (7).
Acknowledgement: Thanks to Dr. Gilman Allen for thoughtful comments on this post.
- In fairness, their article was constrained by a 1000-word limit, no illustrations, and pre-publication peer review with veto power over the article. At the PulmCrit blog, I’m limited only by the extent of my own insanity.
- To be fair, I do think that some caution is required. I wouldn’t infuse a high volume of 3% saline into a tenuous peripheral IV in a patient’s hand or wrist. However, I think if you have a well-functioning peripheral IV proximal to the wrist then it’s safe to infuse 3% saline through it.
- Costs vary, but most sources (example here) cite the cost of a set of gel-adhesive pads for the Arctic Sun to be about 900 US dollars.
- Steroid plays roughly two roles in the post-arrest patient. First, it is useful to prevent or mitigate hemodynamic instability resulting from post-arrest SIRS (a cytokine storm caused by reperfusion injury which resembles septic shock). Second, it is useful as an antipyretic agent as discussed above.
- I’m unable to find any evidence on this, perhaps in part because maintaining patients at 36C is a relatively new phenomenon. Furthermore, the significance of the water temperature depends on the antipyretic regimen being used (e.g. a 10C water bath in a patient on no antipyretics may not mean much, but for a patient on three antipyretics this is more worrisome regarding infection). So I don’t think you can take this to the bank. But it is a factor which is worth paying some attention to – especially when discontinuing the arctic sun.
- Dexmedetomidine may also be used as a short-acting sedative. In order to achieve hemodynamic stability it may be necessary to combine propofol with a vasopressor (e.g. propofol + norepinephrine, or propofol + phenylephrine) or combine dexmedetomidine with a low-dose epinephrine infusion. In my opinion such combinations are entirely acceptable and preferable to the use of longer-acting sedatives which threaten to obscure prognostication and delay awakening.
- I queried the twitter-sphere about this, and some clever folks pointed out common clinical practices which are proven to increase mortality (e.g. over-transfusion in gastrointestinal hemorrhage, use of excessively large tidal volumes, over-resuscitation with crystalloid, excessively aggressive glycemic control, excessive inotropes for heart failure). However, most of these represent titration errors regarding therapies which are potentially beneficial. Some therapies were pointed out which are increasingly unpopular (e.g. recruitment maneuvers). As a discrete intervention proven to increase mortality, I think thrombolysis for stroke is somewhat unique in mainstream modern medical therapeutics.