JC: Intubate or not intubate? That is the question…



Clinical scenario:

A multi-morbid, elderly patient with renal failure and recently diagnosed hyperkalaemia suffers a cardiac arrest in your busy resuscitation area before you can start appropriate treatment. You start CPR with a fully available team and give instructions to treat the reversible cause during chest compressions. One of your very keen senior trainees (who has just returned from her anaesthetic rotation) offers to intubate the patient during the above in order to secure the airway and avoid potential aspiration as she puts it. Everyone looks up at you as you seem rather hesitant to give the go ahead for this surely vital intervention…


Good quality published studies on tracheal intubation during adult in-hospital cardiac arrests are scarce. The European Resuscitation Council over the past decades have de-emphasised the importance for this and highlighted the importance of early defibrillation and good quality chest compressions (see here1). Emergency physicians know the multitude of potential risks inherent to endotracheal intubation during a cardiac arrest scenario: it might lead to interruptions in chest compressions, distract from treatment of identified reversible causes, delay defibrillation, result in inadequate ventilation/oxygenation etc.

Having dreamed of a career in anaesthesia in a previous life, I still remember how difficult it was for me to understand, acknowledge and observe the hands off technique at the beginning of my career. I was therefore pleased to find this paper published in JAMA last month2 which looked at establishing if early tracheal intubation during in-hopsital cardiac arrest was beneficial in terms of outcome.

Please do read the paper in its entirety using the link above. This blog post is an accompanying piece of work only and your conclusions should be made only after you have read the paper in full.

Objective(s) of the study:

This study attempted to determine whether tracheal intubation during in-hospital cardiac arrest was associated with improved survival rates at point of hospital discharge.

Study design:

This was a multicentre, retrospective, observational matched cohort study over a period of almost 14 years using a USA-based registry of in-hospital cardiac arrests. This is a very long period (kudos for the organisers and authors) but the retrospective nature of the data collection can be criticised for all its inherent flaws. Waiver of informed consent was granted for pragmatic reasons.


Tracheal or tracheostomy intubation during in-hospital cardiac arrest at any given minute (0 – 15 minutes) was defined as the intervention. Any patients with an invasive airway already in place during the time of the arrest were excluded.

Time to intubation was defined as interval from loss of pulse until tube placement and was expressed in whole minutes. It is not clear from the paper itself how this was confirmed (visualisation of tube going through vocal cords or tracheostomy cannula, ETCO2 wave or value…) but a supplement details this.


The primary outcome was survival to hospital discharge . An easy one I hear you say but maybe not that much if you think about it in more in depth.

The authors probably had the same thoughts when they defined the secondary outcomes as ROSC and favourable functional outcomes at discharge. A cerebral performance category score of 1 or 2 was considered a good outcome consistent with the Utstein guidelines. Read about the cerebral performance score here. Simply put: this is relevant as not all discharges from hospital are equal in terms of functionality, quality of life, burden to family/society etc. It is therefore important to look at this also, especially if you have such a huge cohort of patients studied over a period of over a decade.

The patients:

262,832 eligible patients were enrolled and 119, 022 excluded leaving a study population including 108,079 patients from 668 hospitals. Gargantuan numbers I thought.

The baseline characteristics differences between patients were deemed to be minimal.

The median age was 69 years and 69.9% of patients were intubated. The median time to intubation was 5 minutes (IQR 3-8 minutes). 336 patients (0.5%) received a tracheostomy.

A total of 24,256 patients (22.4%) survived to hospital discharge.

For complete breakdown and details, I would like to refer you (once again) to the original paper.

A few words about the stats (but a few only!):

The nature of the study needed an approach that would account for the fact that intubation might not occur if the ROSC or termination of resuscitative efforts occur before the intervention. For this this reason, the authors used a time-dependent propensity score matching. Simply put: patients who were being intubated at any minute were separately and sequentially matched with patients who were at risk of being intubated before and within the same minute.

Do not get too worried if you cannot grasp this immediately, You can read more about time-dependent propensity matching here.


In the unadjusted analysis, patients intubated within the first 15 minutes had lower survival compared to those not intubated: 17% vs 33.2% (RR = 0.58, 95% CI 0.57-0.59, p<0.001).

In the propensity matched cohort, survival was lower among the exposed group than among the unexposed group: 16.3% vs 19.4% respectively (RR = 0.84, 95% CI 0.81-0.87, p<0.001). Good functional outcome was also lower among the exposed group than the unexposed one: 10.6% vs 13.6% (RR = 0.78, 95% CI 0.75-0.81)

Sensitivity analyses did not alter these findings. In the matched group, survival was 19.5% vs 36.2% in the intubated versus non-intubated groups with RR = 0.54, 95% CI 0.52-0.56.

For further details, please refer to the original paper.


I have already mentioned the retrospective nature of the study above. Always be cautious of retrospective studies as they are seriously prone to bias. We have a podcast on this here 3 .

Confounders included the underlying cause of cardiac arrest, the skills of the intubator, the quality of chest compressions. The authors quite rightly mention these in their discussions. It is also worth mentioning that data on unsuccessful intubation attempts was not available from the registry.


The authors concluded that initiation of tracheal intubation within any given minute during the first fifteen minutes of resuscitation was associated with decreased survival to hospital discharge.

My thoughts:

I like: massive cohort, long study period, clearly defined and relevant outcomes, sound(ish) stats

I don’t like: hidden confounders, unknown skills of the intubator and medical team

There is a paucity of good quality published evidence in advanced life support interventions. Hopefully you are already using a hands off technique from intubation maximising early defibrillation and good quality CPR. This paper will probably add or change little to your practice.


What do YOU think?




Before you go please don’t forget to…

Lu B. Propensity score matching with time-dependent covariates. Biometrics. 2005;61(3):721-728. [PubMed]
Andersen LW, Granfeldt A, Callaway CW, et al. Association Between Tracheal Intubation During Adult In-Hospital Cardiac Arrest and Survival. JAMA. 2017;317(5):494. doi: 10.1001/jama.2016.20165
CAN 6 Retrospective and Prospective studies. St.Emlyn’s. http://www.stemlynspodcast.org/e/can-6-retrospective-and-prospective-studies/. Published 2016. Accessed March 16, 2017.


  1. Steve Gill (@ICMdoc)

    “Multi-morbid elderly patient with renal failure” … Surely the question is whether to resuscitate not whether to intubate?!

    The likelihood of discharge from hospital with any semblance of QOL is very small whatever piece of plastic you put in the mouth.

    1. Janos Baombe

      Thank for you comment and thoughts Steve.
      The scenario is merely an attempt to illustrate how an intervention like intubation can distract from treating reversible causes of cardiac arrest. Arguably, hyperkalaemia is one of the easy one to identify and treat during CPR…

  2. Pingback: JC: Intubate or not intubate? That is the question… – Global Intensive Care

  3. Pingback: ETT for in hospital cardiac arrest…. – Z5 Clinical Education Space

  4. francis loutsky

    Thanks Janos,
    I would be concerned that retrospective observational studies (regardless how large or fancy the propensity matching) are used to support any change (or bolstering ) of practice, as they are best used to help devise the definitive prospective interventional study question/hypothesis. There is a tendency to continue to interpret as causal, the impressive looking result, even though the association is just as likely to be confounding or bias (both big problems with this study design.) Our favoured narrative is supported, rather than the truth elucidated…


Thanks so much for following. Viva la #FOAMed

Translate »