This issue has only ever come up once in the college exams, which is fascinating because one of the main aims of every intubation is the prevention of hypoxia. Careful and deliberate steps are taken to prevent it, and then scrambling desperate steps are taken to manage it.
This chapter is essentially an overly complicated answer to Question 19 from the first paper of 2017, where the candidates were invited to "describe the strategies for minimising hypoxaemia in the period immediately pre- and post-intubation". An ideal answer would probably be something like the brief summary paragraph which follows; the time-poor candidate would be able to safely stop there, as a detailed exploration of the author's long rambling digression is non-essential. Of the critically important references, a brief skim through the review article by EmCrit's Weingart and Levitan (2011) would be enough to answer Question 19. Specifically, one may take the unaltered content of their Figure 3 (p. 8) and use that as the answer. For a more detailed (and more recent) overview, the meta-analysis by Russotto et al (2017) is probably the best reading.
In brief summary:
Pre-intubation preparation Peri-intubation
- Head up 20-25° (especially valuable in the obese patients)
- 100% FiO2
- Deep breaths × 8
- Or, 3-4 minutes of breathing the oxygen-rich mixture
- The effect is enhanced by positive airway pressure
- There is no benefit in extending this period beyond 4 minutes
- Use NIV unless contraindicated
- PEEP 5-10 cm H2O
- Not to exceed 25 cm H2O
- Alternatively, use a PEEP valve on the bag-valve mask
Minimisation of oxygen use
- Use generous amounts of muscle relaxant
- The use of non-depolarising agents is preferred, as fasciculations can increase the total body oxygen demand
Anticipation of hypoxia
- Preparation of staff and equipment for rapid desaturation
- Continued application of CPAP during the apnoeic period (i.e. while waiting for optimal intubating conditions)
- Use of bag-valve mask to gently ventilate the patient, promoting flow of fresh oxygen into the FRC
Preparation for failure
- Extend invitation to ENT or senior anaesthetic staff to be present at the intubation
- Make surgical airway equipment easily available
And now, in exhausting detail:
Let's say you stop breathing room air. How long will it take for your oxygen saturation to drift into dangerously unserviceable levels, killing your brain? In 1961, 8 adult surgical patients "with no obvious respiratory disease" agreed to find out. They signed up with Heller and Watson who had planned to make them apnoeic in the operating theatre and measure their oxygen depletion by means of the recently invented Clark electrode. This "Polarographic study of arterial oxygenation during apnea in man" is a highly cited article, and worth stealing via Sci-Hub (as NEJM will not allow you free access to their legacy content). It is only two pages long, minus the table of data.
"The results obtained in this interesting study are of practical value in many respects", wrote the authors in an act of massive understatement. What they found was nothing short of revolutionary, given that contemporary practice was to "conduct anesthesia with 20 to 33 per cent oxygen in the inhaled atmosphere". From breathing room air, the apnoeic subjects rapidly became quite hypoxic. A small fragment of Heller and Watson's table can be seen to the left. The patients were were given a dose of some barbiturate and a muscle relaxant (specifically which - this is not mentioned); their blood oxygen content was then measured immediately by the polarographic electrode, and in 30-second increments thereafter. The patients started off with an average PaO2 of 94 mmHg, corresponding to a saturation of 99%; after 60 seconds the PaO2 was down to 60 mmHg on average (which corresponds to a saturation of around 90%, depending on individual p50 values.) Heller and Watson did not wait to find out how much worse it got after that, and rescued the patients.
The authors then experimented with apnoea preceded by three minutes of hyperventilation. They asked their healthy(ish) volunteers to breathe with a minute volume of around 10L, inhaling pure oxygen. What followed was a period of safe apnoea during which the body oxygen stores were slowly depleted. The patients started with a 0-minute PaO2 of around 400 mmHg; four minutes later it was still around 120 mmHg. One could safely go for a mid-intubation toilet break and then come back to the airway with that sort of safety margin.
This data offers the tangent-prone author a useful segue into a discussion of body oxygen stores and exploration of the mechanisms which deplete oxygen in the apnoeic patient.
My source for this is the excellent "The Physiology Viva" by Kerry Brandis, a succinct reference text for the ANZCA or CICM primary exam candidate. I specifically refer to Chapter 4, p. 130 ("Physiology of preoxygenation"). In summary:
|Site||21% FiO2||100% FiO2|
|Functional residual capacity of the lung||270 ml||1825 ml|
|Blood (haemoglobin and dissolved O2)||
Hb-bound: 805 ml
Dissolved: 15 ml
Total: 820 ml
Hb-bound: 810 ml
Dissolved: 100 ml
Total: 910 ml
|Tissues (mainly dissolved O2)||45 ml||50 ml|
|Myoglobin||200 ml||200 ml|
|Totals||1335 ml||2985 ml|
Brandis got this table from somewhere, and it is not clear where. However, as with all such fundamental topics in physiology, if one sets the date range in Google Scholar to "before 1970" one is usually able to find something interesting. In this case, one stumbles upon "Oxygen and carbon dioxide gas stores of the body" by Cherniack and Longobardo. On their page 199 one is able to find data which is similar to the Brandis table ("Magnitude of O2 stores") as well as interesting digressions on the volumes and mechanisms of oxygen storage in the seal, honeybee larvae and toadfish. For the record, seals are absolute champions of oxygen storage, adapting over millions of years to survive prolonged apnoea while diving. Presumably over millions of years the selection pressure imposed by incompetent anaesthetists will also lead to similar adaptation in man.
In summary, the largest oxygen storage capacity in the human body is as gas in the FRC. This gaseous volume at sea level will be approximately 30ml/kg in the supine position. Of this, 21% will be oxygen. By replacing the FRC with pure oxygen, one may increase the total oxygen stoires of the body by 1.5L, of which almost all ends up in the lungs. The other storage compartments add very little. The arterial blood will rapidly become maximally saturated, venous saturation will be unaffected, tissue water can only store some tiny amount of dissolved oxygen (3ml/L) and myoglobin is maximally saturated at even very low oxygen tensions owing to its absurdly low p50 value (around 4 mmHg). The replacement of uselessly oxygen-poor atmospheric gas with pure wall oxygen is the process of "denitrogenation", i.e. allowing inert atmospheric nitrogen to be completely replaced with oxygen in the blood and the FRC.
Where does the oxygen go, and how quickly? Brandis goes on to answer that 80-90% of total oxygen use is in mitochondrial oxidative phosphorylation. At rest, i.e. after you have given the muscle relaxant, this rate will resemble the basal metabolic rate, consuming about 1.25 calories of metabolic fuel per minute - more if they are in some sort of hypermetabolic state. If one were for whatever reason compelled to list such states, the list might look like this:
Those pathological states aside, the normal paralysed 70kg human uses up about 250ml of oxygen per minute. Thus, one should have about 12 minutes of oxygen in one's entire body. This figure is purely theoretical - it is unrealistic to expect that the patient will remain perfectly well until every last molecule will be used up; desperate symptomatic hypoxia will develop well before the tank is half-full (i.e. around the time when one exhausts the FRC).
All discussion so far has revolved around healthy(ish) surgical patients who demonstrated severely impaired judgment by agreeing to a 1960s physiology experiment. In the modern ICU the situation is often quite quite different. Our patients are obese, agitated, hypoxic, peri-arrest, suffering airway injuries, and they often didn't politely agree to preoxygenate with a minute volume of 10L for three minutes. Barriers to normoxic apnoea in our population don't need to be listed, but ... if one were to list them, the list might look like this:
In a rapid sequence induction from a room air environment, the time between the dose of muscle relaxant and intubation is said to be 45-60 seconds.
The whole purpose of this series of strategies is to extend the period of good haemoglobin saturation between the time the patient stops breathing from all the drugs you gave and the time you get the tube in and give them fresh gas. This period of time can be brief and uneventful, or it can be a prolonged and painful struggle against totally unworkable anatomy. Ergo, anything you can do to extend that period is a positive step, as it buys more time to work out how to access the larynx.
There are a series of strategies which will now be elaborated upon; these are all derived from the Weingart and Levitan article (2011).
The typical prelude to emergency intubation in ICU (while everybody is scrambling for drugs and equipment) is the few minutes of preoxygenation. The critically ill patient, tachypnoeic and agitated, gasps and gurgles at the ICU senior registrar who grabs them like a facehugger and applies wall oxygen via a bag-valve-mask apparatus. But is this ritual useful?
The answer is "maybe". Thomas Mort et al studied a cohort of emergency intubations with 4 minutes of pre-RSI oxygenation (in 2005) and again using 8 minutes (in 2009). These were all patients who somehow failed non-invasive methods of oxygenation. The baseline oxygenation prior to BVM manoeuvres was a PaO2 of around 60-70 mmHg. After 4 minutes, Mort's cohort ended up with an average PaO2 of around 100 mmHg. Extending this pre-oxygenation period to 8 minutes was ...marginally effective (in fact, one quarter of the patients actually desaturated during the extra four minutes). 36% of the patients got absolutely no benefit from bag-mask preoxygenation. A small proportion (19%) increased their PaO2 by at least 50mmHg in the first 4 minutes.
In summary, one can expect critically ill patients to derive only a marginal benefit from prolonged attempts at preoxygenation, and extending these efforts beyond four minutes can actually jeopardise one's attempt.
Being supine is a stupid position for a person with dyspnoea. This is the reason the agitated hypoxic patient will try to sit up or get out of bed. They want to sit on the edge and "tripod" themselves to recruit all available muscle groups into the fight for air.
For intubation, being supine is a disadvantage because of the effect of this posture on the FRC. Supine posture favours atelectatic collapse of the lung bases, which decreases the FRC and makes it more difficult to pre-oxygenate the patient. The ideal position is about 20 degrees head up, particularly for the obese patients. In the patient who cannot bend at the wasit (eg. pelvic traction or spinal precautions) the reverse Trendeleburg position (tilting the whole bed 30° feet-down) is an accepted alternative.
In fact, the exact angle is probably less important than the tragus-to-sternum rule. One is able to make this happen by using a clever motorised bed, or by constructing the "airway ramp", a makeshift wedge of folded pillows and linen or towels which elevates the patient's head and shoulders until the tragus of the ear is in horisontal alignment with the suprasternal notch. This is a well-known trick, and everybody has some sort of diagram to describe it - the one shown here was stolen from Dr. Christine Whitten (The Airway Jedi) with no permission whatsoever.
The importance of positioning has been confirmed empirically, but only to the extent of proving that the FRC is improved by this posture. For instance, Lane et al (2005) used a 20° head-up tilt, and found that it bought about 100 extra seconds of normoxia (i.e. time taken to drop below 95% SpO2). However, this study was performed in a somewhat idealised setting. The patients were preoxygenated, anaesthetised, intubated, and then the hypoxia time was measured (with the well-sedated intubated patient lying either supine or 20° head-up, disconnected from the ventilator). Ramkumar et al (2011, also quoted by Weingart et al) reproduced the same findings using virtually the same technique, again measuring apnoeic desaturation in intubated anaesthetised patients. Both studies give good evidence for the improvement of FRC by posture, which confirms what we already know from basic physiology. Neither trial explored the unique challenges of the ICU environment in the West of Sydney. The patients were excluded if they were obese, suffering respiratory failure, or in need of a rapid sequence induction.
However the obese and morbidly obese population is exactly what Question 19 from the first paper of 2017 is asking about. What of the evidence and experience with this group? Did any study focus on them, or at least not exclude them? Well. Weingart and Levitan quote two studies, both from 2005. Subsequent years have been barren for this topic, and since 2011 nothing but pointlessly repetitive review articles has been published, which is why Dixon et al (2005) and Altermatt et al (2005) are cited by about 300 papers, each.
Altermatt and colleagues tested preoxygenation among forty morbidly obese (BMI > 35) patients, half of whom were preoxygenated supine and the other half in a sitting position. Again, the patient's tolerance of the posture was tested only after the intubation, which was for some reason performed in the decubitus position (what??). The head-up group did slightly better - the extra FRC bought an additional 50 seconds on average. Dixon et al (2005) looked at an even more portly group (mean BMI 44-47) and also found about 45 seconds difference between groups. This time the patients were intubated in a conventional position, and Dixon et al used an SpO2 of 92% as their cut-off, wheres Altermatt's group used 90%. Aside from this nit-picking, a head up position seems to benefit obese patients (and the people intubating them).
Using raw untreated oxygen to preoxygenate the patient is a great idea, but the very process of denitrogenation removes and important "splint" from the alveoli. The rapid uptake of oxygen by the pulmonary capillaries can lead to "denitrogenation atelectasis" (see a discussion of this in the 2007 paper by Duggan et al). The logical antidote to this would be the use of positive pressure ventilation.
Weingart et al specifically recommend the use of NIV to preoxygenate those patients who are shunting too much, and are unable to preoxygenate beyond a saturation in the low 90s. They quote Cristophe Baillard et al (2005) in support of this practice. Baillard's group randomised hypoxic patients to 3 minutes of CPAP with 100% FiO2; the controls used a bag-valve mask. With CPAP Baillard's group started at a higher SpO2 baseline (98% vs 93%) and desaturated to a higher mean value (93% vs. 81%). However, this author used a CPAP of 5 cm H2O. That hardly seems like enough positive pressure. Probably this has something to do with the fact that Baillard et al generally intubated slender French people (the mean weight for their population was 72kg, with height arond 170cm). Locally, a CPAP of 10-15 is used routinely to preoxygenate the victims of the obesity epidemic.
The alveolus, if it remains open during apnoea, will continue to exchange gases. These will be mainly oxygen and carbon dioxide. You are going to consume about 200ml of oxygen, and (with a normal respiratory quotient) produce about 200ml of CO2 of which most ends up being buffered in the bloodstream. The small volume of remaining CO2 which ends up being liberated into the alveolus is not enough to compensate for the larger volume of removed oxygen, and a small negative pressure is generated by this net loss of gas volume.
This net negative pressure is enough to gently drag fresh gas into the alveoli, albeit slowly. Provided all the airways stay open, this contributes enough oxygen to sustain metabolism for a surprisingly long period of time. For instance, Draper et al (1949) were able to oxygenate anaesthetised mongrel dogs for 45 minutes in this manner. Intensive care physicians reproduce the experiment each time they perform the ritual of clinical examination for brain death.
What relevance has this got for the critically ill patient undergoing rapid sequence induction? Well; provided one can maintain patency of the airway, one may be able to the oxygenate the patient near-indefinitely using merely this "mass transfer" of gas. There is probably some evidence in support of this practice. At the time of their writing, Russotto et al (2017) managed to find four trials with a total of 358 patients, of whom three used some sort of high-flow nasal cannula. All trials produced a higher value for the minimum oxygen saturation with apnoeic oxygenation, but none were able to demonstrate a trend towards decreased significant severe desaturation events. In other words, if you lose patency of the airway, it does not work at all; and if you have a patent airway then - conceivably - just about any sort of oxygen delivery device might be good enough.