A brief note: I get to do the editing duty this week (Dr Andrew Weatherall that is) and I could not let it pass without a word of tribute to Dr John Hinds. I had only had the chance to learn from the good Dr Hinds via his online presence. It was a big presence.
As one who did not know him personally, I can only reflect that he demonstrated many of the best qualities of a passionate doctor and that his passing, far too soon, has revealed many of the best qualities of his colleagues.
Just in case you needed another reminder, you could watch him in action here, or read good words by @Eleytherius here, or sign a really worthwhile petition to deliver a vision for a better prehospital service for patients in NI here.
As to this week’s post, Dr Alan Garner has a post on looking for the right outcomes so we’re doing the right thing for our patients.
Can’t see the wood for the damn trees
As part of their intubation quality program many services now report their first look intubation rate. We have been doing so for a couple of years now. This looks like a really good thing to do. We know that more than one attempt at intubation is associated with greater incidence of serious adverse events in critically ill patients, and the more attempts the more likely those adverse events become (reference 1).
Therefore a strategy of aiming for first look success is probably a good idea, a strategy that my own service employs. So this should be a good thing to report as a quality measure too. Indeed why would you not? After all, the more attempts, the worse things get right?
Well wait a minute …
First let’s have a think about why we would report it. Is it telling us something that actually matters?
The outcomes that really matter are did they die or end up with hypoxic brain injury. The process issues that really matter are did they get hypoxic or have a cardiac arrest during the intubation process. There are other hard complications/process issues you can measure too like aspiration with unnecessary additional ventilator days, or even did you break their teeth.
First look intubation tells us none of these things. It does not tell us if the patient became hypoxic, aspirated or even arrested. Yes it is associated with lower incidence of these complications but it does not tell you if the complication actually occurred.
And what if emphasising first look intubation rate as a quality measure shifts the focus in the wrong direction? Could you risk making the risk of hypoxia higher?
Am I losing the plot here? Let’s go back to first principles.
The outcomes that really matter are death and hypoxic injury. I don’t think anyone is going to argue these should be avoided. Fortunately the incidence of these is pretty low so we tend to use surrogates for these things instead, things like the incidence of hypoxia or hypotension/bradycardia during intubation. These are pretty direct measures reflecting outcomes that matter.
First look intubation isn’t an outcome. It’s not even a surrogate for an outcome – it’s a surrogate for a surrogate of an outcome. My concern is that surrogates for an outcome, rather than the actual outcome can lead you way up the garden path. The MAST suit again comes to mind. The patient’s BP went up so it had to be a good thing surely. Of course when someone finally did a decent study on the outcome that really mattered, mortality, it was trending to worse not better.
Although there are no randomised controlled trials showing hypoxia to be bad for you, the circumstantial evidence is pretty overwhelming so I agree this is not quite like the MAST suit situation. However in using first look intubation as a quality measure we are now reporting a surrogate for a surrogate of the outcome that actually matters. I.e. we are reporting first look as it is associated with lower rates of hypoxia because lower rates of hypoxia are associated with lower rates of death and brain injury.
This is a risky game and recent audits of my own service show why. For the past year we have had a monitor that records the vital signs every 10 seconds and we download the data at mission end and attach it to the record. I have been going through these records to see what our rates of peri-intubation hypoxia actually are.
First thing I need to say is that our first look intubation rate so far this year is 100%. However we did have a couple of episodes of significant hypoxia.
My concern is that by reporting the first look rate, we draw attention to it and we send the message to our teams that this is the thing that we think matters. So better to press on a little bit longer even though the sats are falling to make sure I nail that tube first time!
Why are we reporting a surrogate for a surrogate? I have really accurate data from the monitor on the peri-intubation hypoxia rate, hypotension, bradycardia and arrest. Why report a surrogate for these things that might actually encourage our staff to focus on the surrogate and cause an episode of hypoxia, bradycardia, hypotension etc.
It remains important to emphasise optimising conditions for the first intubation attempt as that appears to have lower complication rates. However it is a means to an end. We should emphasise the outcomes (or at least the surrogates with only one degree of separation from that outcome) that matter. Why report a surrogate for a variable when you have the data to report the actual variable?
Some services like our own are now reporting 100% first look intubation rates, but no one is yet reporting 0% peri-intubation hypoxia rates. Aim for first look intubation as that appears to be a smart strategy, but tell your people it is the hypoxia that matters by making that the centre of attention in your reporting.
What do we mean by hypoxic?
Another thing I have been forced to look at is the definition of peri-intubation hypoxia. I had intended to use the definition of hypoxia used in many of the studies on this subject:
“Desaturation was defined as either a decrease in SpO2 to below 90% during the procedure or within the first 3 minutes after the procedure, or as a decrease of more than 10% if the original SpO2 was less than 90%.” (reference 2, see also 3-5)
I excitedly opened the data file of our first patient that we had intubated when we got our shiny new monitor a year ago to see what had happened. It was easy to identify the timing of intubation from the capnography data as we routinely pre-oxygenate our patients with a BVM device with the capnography attached. The sats pre-induction were a steady 90%, for 2 readings they were 89% (20 seconds) and then climbed to 98% when ventilation was commenced. So according to this definition we had a desaturation!
I don’t think anyone would claim a fall in SpO2 of 1% is clinically significant. It is also less than the error of the measurement quoted by the manufacturer of the oximetry system. This set of circumstances is not going to occur that often but it does not make sense to classify this case as a desaturation. We have therefore modified our definition to:
“Desaturation is defined as either a decrease in SpO2 to below 90% (minimum change at least 3%) during the procedure or within the first 3 minutes after the procedure, or as a decrease of more than 10% from the pre-intubation baseline if the original SpO2 was less than 90%.”
So what should we be reporting?
Thomas reported that each subsequent attempt at intubation was associated with an increased risk of hypoxia, aspiration, bradycardia, cardiac arrest etc. If we have the data on these variables then why not report them directly instead of reporting the surrogate for them. For hypoxia I would suggest our slightly modified definition above.
As for other variables why not use the definitions from Thomas’ paper?
|Bradycardia||HR <40 if >20% decrease from baseline|
|Tachycardia||HR >100 if >20% increase from baseline|
|Hypotension||SBP <90 mm Hg (MAP <60 mm Hg) if >20% decrease from baseline|
|Hypertension||SBP >160 if >20% increase from baseline|
|Regurgitation||Gastric contents which required suction removal during laryngoscopy in a previously clear airway|
|Aspiration||Visualization of newly regurgitated gastric contents below glottis or suction removal of contents via the ETT|
|Cardiac arrest||Asystole, bradycardia, or dysrhythmia w/non-measurable MAP & CPR during or after w/in intubation (5 min)|
For the physiological definitions Thomas includes percentage change from baseline like we do with the hypoxia definition. This acknowledges that these are critically ill patients and often have deranged physiology before we start. These definitions can therefore be used in the real world in which we operate. If we all adopted these definitions we could meaningfully compare ourselves with Thomas’ original paper and with each other.
And as for us…
We are seriously thinking about ditching the reporting of first look intubation rate. It is not telling us what really matters – and we can’t get better than our current 100% rate anyway. Despite this we are having occasional episodes of hypoxia and other complications, and it is possible that the rate of these complications are being exacerbated by emphasising first look.
We are therefore looking at moving to the much more comprehensive set of indicators used by Thomas (along with our modified hypoxia definition). This will demonstrate to our team members the factors that we think really matter, because we measure them and report them externally.
You could argue that the only way to achieve 0% hypoxia is to accept that we are not going to have a 100% first look intubation rate. I for one would gladly give up our 100% first look rate if in doing so we achieved 0% hypoxia. I don’t yet know if this is achievable but I have some ideas. Those who walk the quality & patient safety road with me know that we might never arrive, but that should not deter us from the journey.
- Anders Rostrup Nakstad MD, Hans-Julius Heimdal MD, Terje Strand MD, Mårten Sandberg MD, PhD. Incidence of desaturation during prehospital rapid sequence intubation in a physician-based helicopter emergency service. American Journal of Emergency Medicine (2011) 29, 639–644
- Reid C, Chan L, Tweeddale M. The who, where, and what of rapid sequence intubation: prospective observational study of emergency RSI outside the operating theatre. Emerg Med J 2004;21:296-301.
- Omert L, Yeaney W, Mizikowski S, et al. Role of the emergency medicine physician in airway management of the trauma patient. J Trauma 2001;51:1065-8.