Wheezing in the City of Lights


The PARIS Trial – High-Flow Oxygen in Bronchiolitis

The timing of the publication of the PARIS (Paediatric Acute Respiratory Intervention Studies) 1 trial is perhaps fortuitous, as we brave few saddle up to venture forth once more unto the breech, dear friends, into the cough-wracked, wheeze-ridden whirlpool of snot that the ED paediatric area will soon become… for winter is coming. And the negative-sense single-stranded RNA virus that is RSV (human Respiratory Syncitial Virus) doesn’t care about your misery. Oh no. With its cell-membrane-piercing lipoprotein F & G complexes and its 2-8 day (but usually 4-6 day) incubation period, it will come for you. It can’t be bargained with. It can’t be reasoned with. It doesn’t feel pity, or remorse, or fear. And it absolutely will…not…stop…ever…until you are dead. Or, you know, at least your soul is. C’est la vie. You work in a mixed ED. Own your choices, people.

The PARIS trial, published somewhat appropriately on the bleak, foggy morning of March 22nd amidst snow flurries and a top of 2°C at the NEJMs Boston headquarters, is a pragmatic multi-centre randomised controlled trial, looking at infants presenting to hospital with clinically diagnosed bronchiolitis, who required supplemental oxygen. They randomised kids to receive either bog standard “low flow” oxygen to keep their SaO2 > 92% (or > 94% in centres where 92% was deemed a little too scary), or immediate treatment with “high-flow” oxygen at 2 L/kg/min. Their primary outcome was the need for “escalation of care”, which we’ll get to in due course, and their secondary outcomes included admission to ICU, hospital LOS (length of stay), ICU LOS, and duration of oxygen treatment.

The study involved 15 Australian and 2 New Zealand hospitals. Funding was sourced from six bodies including the NHMRC, and five non-profit institutional research foundations. While there was no industry funding, Fisher & Paykel provided the high-flow equipment for use in the trial. For budding health ethics & policy researchers out there, an interesting follow-up study might be there for the taking: F&P’s sales of high-flow units before and after the publication of the PARIS trial.

Some 20,795 infants were identified and screened between October 2013 and August 2016, with 18,578 excluded (11,081 had bronchiolitis but didn’t need oxygen, 7,341 had a different illness, 535 simply missed the boat logistically/administratively, 156 had bronchiolitis but were sick enough to be whisked away immediately to ICU before enrolment could occur, 44 had consent refused up-front, and another 166 had consent withdrawn at a later time), leaving 1,472 kids who made it through for analysis. With 733 patients in the SOT (standard oxygen therapy) group, and 739 in the HFOT (high-flow oxygen therapy) group, this was a reasonably large, well-powered study.

So… what did they find? Was it important? Will it change practice?? Will it be misinterpreted??? Will Fisher & Paykel sell more high-flow oxygen units next year than they otherwise would have???? The answers are, respectively: Things. Yes. Probably. Almost certainly. Even more certainly.

Breaking down the important bits:

HFOT is safe

Adverse outcomes were exceedingly rare in both groups. The pessimist (or pedantic statistician) might suggest that this results in the low frequency of events meaning that comparisons between the two groups are difficult, verging on meaningless. The optimist (or those willing to think a little more carefully) would suggest that this means that no matter what we do with oxygen treatment in these kids, it’s exceptionally safe. Each group had 3 episodes of apnoea, and 1 pneumothorax per group as well. No other badness ensued. The take-home point here is that HFOT does not seem to cause an increased risk of pneumothorax (or apparently any other pulmonary barotrauma), a reasonable concern based on the physiology and mechanics involved.

HFOT Reduces the Need for “Escalation of Care” :  12 % vs 23 %

Right. This is their primary outcome, and the ONLY positive finding in the trial. Some 12% of kids in the HFOT group required escalation, compared to 23% of the SOT group. The p-value is irrelevant but more than small enough to satisfy those who don’t quite understand how p-values should be used. This is a big difference. It represents an NNT of 9. Which is pretty good, given we get pretty excited about aspirin for AMI, etc. with its NNT ~ 50 (and tPA in stroke, with an NNT ~ ∞).

This is impressive stuff.

Or… is it?

Escalation of care was defined as “increased respiratory support and/or transfer to ICU”. For those in the SOT group, “escalation of care” was invariably escalation to HFOT. This was, in fact, recommended/suggested by the trial investigators.

The criteria for triggering escalation of care were defined as follows, and you had to have >2 of these to pull the trigger:

  • HR unchanged or increased (ie failed to decrease)
  • RR unchanged or increased (ditto)
  • O2 requirement in HFOT FiO>0.4 …or… O2 > 2 LPM in SOT group
  • Local early-warning tool (PEWS for TCH) triggered a review
  • Clinician felt a disturbance in The Force (ie concern for any reason)

CAVEAT EMPTOR – Beware of composite endpoints…and this is one of them!

What’s composite about it? Well, the investigators have taken the liberty of lumping in “This kid required* moving from SOT to HFOT” with “This kid needed to go to ICU”. From a clinical perspective, these are two VERY different things, even allowing for a generous definition (see the above list) of “required”.

Why does it matter in the context of the reported outcomes in this trial? Well, it matters because the primary (composite) outcome showed a massive difference, and is therefore what will be reported, and remembered by clinicians, when perhaps…dare I suggest…a perusal of the secondary outcomes (which, conveniently in this case, allows disentanglement of the composite primary figures) may be more elucidating, and vastly more clinically relevant.

HFOT Does Not Improve Clinically Relevant Outcomes

The secondary outcomes measured in the PARIS trial were:

  • Admission to ICU
  • Hospital LOS
  • Duration of oxygen Rx

As suggested above, these are things we really (or at least should!) care about. You may or may not be surprised to discover that there was no statistically significant difference in any of these outcomes, between the two groups.

The part of that which really matters and should be emblazoned in glowing neon, is that the rate of ICU admission was no different. So the difference reported in the primary outcome, which you will recall functionally consists of:  Escalation = (“moved from SOT to HFOT” + “admitted to ICU”) is due, in its entirety, to the number of kids who triggered the escalation criteria and moved from SOT to HFOT. More specifically:

  • 167 / 733 (23%) in SOT were escalated as per criteria
  • 167 / 167 (100%) of these were put on HFOT
  • 102 / 167 (61%) of them improved (representing 14% of the initial SOT group overall)
  • 65 / 167 (39%) did not improve, and went to ICU (representing 9% of the SOT group)

But wait… Considering the above in context:

  • 9% of the SOT group ended up in ICU
  • 12% of the HFOT group ended up in ICU

Yep. More kids in the got-immediate-HFOT group went to ICU. And before you hurl rotten fruit and lengthy diatribes about selection bias in my direction, these kids were randomised to the SOT & HFOT groups. If they looked sick enough that someone wanted to start HFOT immediately without randomisation, they were not included in the study.

Summary  (All Staff)

This was a well-designed, large, multi-centre study sufficiently powered to find fairly small but real differences in outcomes in infants presenting to hospital with bronchiolitis who required oxygen. The primary outcome showed a large difference between the groups, but this was a composite outcome, with the entire difference between groups caused by escalation from SOT to HFOT. The analysed and reported secondary outcomes, which were far more clinically relevant, showed no statistical difference between groups, and indeed even showed a potential trend towards harm in the HFOT group (12% vs 9% required ICU admission).

Even granting clinical relevance to the primary outcome, we are left with an NNT of 9, which whilst numerically impressive, in the real world essentially translates to having to set up 9 patients on high-flow oxygen immediately, in order to avoid having to put 1 child on it later if they are not improving on standard low-flow oxygen. This represents a considerable imposition on resources including equipment, maintenance, and nursing staff.

Were we to take the primary outcome at face value, it is unclear that changing practice to commence HFOT in all cases of bronchiolitis with an oxygen requirement, would produce any clinically relevant benefit for our patients, and would impose a large logistic burden on the department.

Summary of the Summary   (ED Registrars still working on their ADHD)

  • High-flow oxygen Rx is safe
  • High-flow oxygen Rx makes no difference to clinically important outcomes
    • Admission to ICU
    • ICU LOS
    • Hospital LOS
    • Duration of oxygen Rx
  • Putting 9 kids on HFOT right away prevents you having to put 1 kid on it later if they don’t pick up with simple low-flow oxygen
  • It’s probably not worth it

Summary of the Summary of the Summary   (ED Consultants with advanced ADHD)

  • Use standard low-flow oxygen first
  • Move to high-flow if you’re not winning
  • Keep in mind that doing so will not change the outcome

PARIS Trial @ NEJM – http://www.nejm.org/doi/full/10.1056/NEJMoa1714855

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s