Last week emcrit proposed that we have lost our nerve: that we fail to act when action is called for, that we tend to commit errors of omission, rather than errors of commission, and so we should toughen up, give ourselves permission to act, provide maximally aggressive care everywhere. A chorus responded with the obvious objection that we are doing way, way too much and that we are causing massive amounts of harm with overtesting, overdiagnosing, and overtreating.
I review hundreds of high-risk cases per month, and in patients who have bad outcomes, we do often conclude that the bad outcome could have been prevented by doing more. It really does seem as though there is a reluctance by non-surgeons to perform dangerous procedures, and that this reluctance often results in harm.
The easiest way to explain this discrepancy is that maximally aggressive care everywhere might be a reasonable paradigm if everywhere for you is a resuscitation bay (cue Casey). But of course M.A.C.E., if interpreted as when in doubt, do more is an unreasonable paradigm even in the resus bay–we don’t want to do more, we want to do the right amount. Part of knowing the right amount is knowing medicine, and part of it is knowing yourself. Most of us tend toward overtesting and overtreating, but under-resuscitating (most of us but not all of us–we’ve all run into clinicians whose threshold to do dangerous procedures is too low, they are way scarier). This is why there is a lot of wisdom to contradictory rules of thumb like don’t just do something, stand there and if you’re not sure whether to intubate, intubate.
When you spend a lot of time reviewing cases, you start to think about the patient in front of you from the perspective of the person reviewing the chart next week; this mindset has spawned a useful cognitive strategy I call the preferred error. If you make the right choice, or the patient does well, fantastic. What matters is when you’re wrong, or there’s harm. So consider the consequences of being wrong on both sides of the decision, and determine which course of action fails better.
Looking at if you’re not sure whether to intubate, intubate from the future backward, its rationale becomes obvious. If you intubate a patient who doesn’t need to be intubated, how much harm is done? Some harm. And there’s the slight chance that you will fail to intubate and cause harm by trying, but this is quite unlikely. Now consider, what if you don’t intubate, but it turns out the patient needed to be intubated–how much harm is done? Potentially a lot of harm when that patients later requires a crash intubation, which is much more dangerous.
The preferred error considers how much harm if you’re wrong, but you must also consider how likely you are to be wrong, and factor that in. The hyperadrenergic patient–hypertensive, tachycardic, hyperthermic, agitated–has a differential filled with dangerous conditions, and it may take some time to sort out. You’re contemplating alcohol withdrawal, you think there’s a 30% chance that this is alcohol withdrawal. Should you treat for alcohol withdrawal? Consider what would happen if you treat with benzodiazepines, and he didn’t have alcohol withdrawal: not harmless, but probably minimal harm. And how likely is it that he doesn’t have alcohol withdrawal? About 70%, so 70% chance you might cause minimal harm. But what if you don’t treat with benzos, and the patient does turn out to have alcohol withdrawal? Very bad, lots of harm, the patient will get a lot sicker, might seize, might need a crash intubation. So 30% chance of causing a lot of harm if you don’t treat. Treat. Lots of tough decisions become easier when you consider the preferred error.
Lastly, when you’re still not sure what to do, you can hedge, and hedging means prepare. In the severe asthmatic you’re nervous about but you think you can turn around, but you’re not sure, and you really don’t want to intubate, but someone told you if you’re not sure whether to intubate, intubate, the way to manage that risk is get everything ready to intubate. Fully prepare, cognitively and materially, to intubate, as you throw every asthma therapy you’ve got at the patient. Preparation gives you the chance to be right, while minimizing the harm if you’re wrong. Preparation is the respect we pay to risk.
The following is adapted from my contribution to this discussion of how best to manage unfasted patients who require deep sedation to facilitate a painful procedure. I am responding mostly to Nicholas Chrimes, a thoughtful Australian anesthesiologist behind The Vortex Approach to airway management. Because aspiration leading to clinically relevant morbidity is a rare event and is often not straightforward to identify, we do not know which patients are most at risk, which sedation procedures confer greater risk, or how to reduce that risk (by, for example, fasting). Standard of care is therefore based on tradition and opinion which I believe is largely misguided and contrary to the interests of patients and providers. Though the focus of the discussion was on whether patients not known to have an empty stomach are better off receiving spontaneously breathing procedural sedation or RSI/endotracheal intubation, I was most struck by what felt to me a lack of appreciation of the harms of fasting. The harms of fasting are not adequately represented in this debate, so I think the topic is worth elaborating.
The notion that, during PSA, aspiration causes clinically important harm with an appreciable frequency, or that this frequency can be reduced by fasting, is contrary to the outcomes and opinions reported in these registries and reviews:
- Smally 2011
- Thorpe 2010
- Molina 2010
- Roback 2004
- Agrawal 2003
- Ghaffar 2002
- Treston 2004
- Bell 2007
- Babl 2005
- McKee 2008
There are over 20,000 ED PSA cases reported in the literature, and, to my knowledge, two (2) reported clinically consequential aspiration events. In one case, the patient was NPO for 6 hours prior to the procedure, in the second, NPO for 24 hours. There is also theoretical evidence that fasting, which increases the volume and acidity of gastric secretions, makes aspiration events more dangerous. ACEP’s 2014 clinical policy offers this level B recommendation: Do not delay procedural sedation in adults or pediatrics in the ED based on fasting time. Preprocedural fasting for any duration has not demonstrated a reduction in the risk of emesis or aspiration when administering procedural sedation and analgesia.
The suggestion that we might expose a non-fasted patient with a shoulder dislocation to the risks of RSI so that an endotracheal tube can be placed for the three minutes it takes perform a propofol-facilitated joint relocation, for the purposes of reducing aspiration risk, a risk that has been demonstrated to be trivially small and not reduced by fasting, is, in my view, a dangerous, almost irresponsible suggestion. Attempting to view this from your perspective, I will stipulate that anesthesiologists are better than emergency physicians at routine intubation, and the situation described is closer to routine intubation than a usual emergency department intubation.
I recognize that the anesthesia guideline I quoted earlier concluded with a strong recommendation for fasting and did not mean to imply otherwise; my point was that the authors of that guideline explicitly acknowledge that their recommendation is not evidence-based. The physiologic arguments you confidently assert as support of a fasting recommendation seem to me academic when compared to the overwhelming evidence that performing PSA on non-fasted patients is safe, especially when viewed in light of the harms of fasting.
The harms of fasting are very important and are a source of considerable morbidity; that you trivialize them is a powerful testament to the occasional chasms between EM and anesthesia. I think as groups we agree on most issues that come up in the areas where our expertise overlaps, but there is no greater exception to this than how we view fasting and aspiration risk.
Every time a patient in the emergency department who requires a procedure is fasted, there is harm. The procedure they need usually involves a painful condition; by delaying the procedure, you prolong the length of time the patient is in pain (and hungry and, more distressingly, thirsty). The lesion that needs to be addressed progresses–the fracture swells, the dislocation stiffens, the heart more content to be in atrial fibrillation–the procedure therefore becomes harder or less effective. The patient remains in a bed in the ED, taking up geographic and nursing resources that could be diverted to others.
There are more insidious harms. Because of the fasting culture in medicine and the unfortunate policies that have arisen around them, in many emergency departments, the default nursing position is that patients are NPO until specifically authorized to eat by a provider. In practice this means most patients in the department are deprived of food and drink on the chance that someone is going to get upset because the patient has eaten. It gets even worse: radiologists have jumped on this train and now, in some centers, will not accept patients for IV-contrast CT unless they have been fasted for 4 or 6 hours. This on the chance that the patient will have an allergic reaction to the contrast, then somehow as a result vomit and aspirate. So we wait hours to get pulmonary embolism diagnosed, for a baseless, senseless, defensive policy that directly harms patients. I have tried to reverse this culture of fasting where I work by using a catchy slogan: everyone eats.
I just got off an airplane. As we begun our descent, the flight attendant passed by my aisle, looked sternly at the passenger seated next to me who was reading a Kindle, and said, sir, please turn off all electronic devices for landing, in the usual tone that suggests the safety of other passengers was imperiled by his operating a Kindle. We looked at each other and shook our heads.
Rapid sequence intubation, the simultaneous administration of a paralytic and induction agent immediately followed by laryngoscopy, provides the optimal view of the glottis and prevents emesis. RSI is the best strategy for most patients who require intubation, but not all.
When you’ve decided to intubate, first maximally preoxygenate. If the patient will not cooperate with your preoxygenation plan, even after you’ve asked nicely, that’s your cue to add cooperation in a vial, ketamine, and then oxygenate, before you push the paralytic and intubate. This is delayed sequence intubation. Otherwise, carry on with preoxygenation, and consider a couple of special situations.
The first is the patient who is about to arrest–obtunded, no blood pressure. Any induction agent will cause sympatholysis which, along with the transition to positive pressure ventilation, may precipitate arrest, so ideally we would avoid both while the patient is in the state of nearly arrested. Resuscitate aggressively with fluids, vasoactive drips and treatment of the underlying problem for as long as you can before intubating. If you must intubate the patient who is obtunded with no blood pressure, the safest way to do it is often without any drugs at all, while the patient continues to breathe. If you have to give meds, dose sedatives low and paralytics high.
The next special situation is the patient who has a severe oxygenation or ventilation deficit. The severe oxygenation deficit patient saturates less than 90% on 100% NIV; the severe ventilation deficit patient is compensating for a severe metabolic acidosis, e.g. DKA with pH 6.7. In patients with a severe oxygenation or ventilation deficit, even a brief period of apnea is very dangerous, and since paralysis is certain to cause apnea, it stands to reason that paralysis may not be the best approach. However, these patients are very ill, so conventional awake technique, which requires time and cooperation, will not work well. If only there were a drug that would immediately render the patient tolerant of laryngoscopy, while ventilation and airway reflexes are preserved.
Ketamine-supported intubation, KSI, is pushing an induction dose of ketamine over 20-30 seconds, then performing laryngoscopy. KSI is awake intubation with minimal or no local anesthesia, or, if you prefer, RSI without paralysis.
By omitting the paralytic, KSI carries a chance of two harms: suboptimal view of the glottis, and emesis/aspiration. I address these harms in detail in this discussion; the advantage in glottic exposure offered by paralysis is less significant in the era of video laryngoscopy, which almost always provides a great view of the cords, and the risk of emesis/aspiration is very small most of the time. In any case, these harms must be weighed against the harm of apnea for the patient in front of you. Others have described a similar strategy, augmenting ketamine with etomidate as necessary.
The last and most important special situation is high concern for difficult laryngoscopy. You assess all your endotracheal tube-requiring patients for difficult laryngoscopy, either intuitively or explicitly, and most of the time, you think, I got this, in which case, carry on with RSI like you always do. But if you think it is likely that laryngoscopy will fail, and the patient is presently benefiting from their own ventilatory efforts, abolishing those efforts with RSI may not be the best choice.
Even in scary laryngoscopy cases, RSI is probably still optimal if the patient is high risk to vomit (has been vomiting, upper GI bleed, bowel obstruction). In these scenarios, the protection against emesis afforded by paralysis is compelling, so proceed with RSI, but use a double setup, with your partner on standby, ready to cut the neck. Keep the head of the bed up and drop an NG tube in beforehand if you can.
The patient whose airway is a lawyer’s dream and isn’t a particular risk to vomit is ideally intubated without a paralytic, while continuing to breathe, awake intubation. Awake intubation has two pharmaceutical arms: local anesthesia, and systemic sedation. The more cooperative the patient and the less urgent the airway, the more you can rely on local anesthesia. So if you have time and cooperation, dose glycopyrrolate or atropine, then generously nebulize, topicalize, and atomize lidocaine, then you can slowly, carefully do your laryngoscopy, or flexible endoscopy, or whatever you want, as the patient is awake and breathing. In the OR, where patients and physicians are stable and cooperative, patients with concerning airways are intubated with minimal or no sedation at all, which affords an enormous margin of procedural safety. Patients being intubated in the ED are of course neither stable nor cooperative, but a similar degree of safety can be achieved using ketamine: the less time and less cooperation, the less lidocaine, the more ketamine.
In the extreme version of the emergency department awake intubation, give induction dose ketamine and go: KSI. Consider KSI for your severe oxygenation/ventilation deficit patients, but also when you are concerned that laryngoscopy is going to fail and the patient won’t cooperate with, or you don’t have time for, a more civilized, operating theater-type awake intubation. Have a paralytic ready in syringe, in case you want to convert to RSI at any point, and incorporate a double setup component to your approach, because your concerns about laryngoscopy might turn out to be well-founded.