Prone positioning didn't clearly cut ventilation escalation in infant bronchiolitis
The PROPOSITIS trial put babies on high-flow oxygen face-down or on their backs. Escalation of care was 15% versus 21% — a difference that did not reach significance, and the authors call the study inconclusive.
| Group | Value (%) |
|---|---|
| Prone position | 15 |
| Supine position | 20.8 |
Turning infants with severe bronchiolitis onto their fronts, rather than leaving them on their backs, did not significantly reduce the need to escalate their breathing support, according to the PROPOSITIS randomised trial [s1]. The result is not a clean negative — the difference leaned in favour of the prone position but was not statistically significant, and the researchers say explicitly that the study was not definitive [s1]. It is a careful trial reaching an honestly uncertain answer to a practical bedside question.
The question
Bronchiolitis, a viral lower-airway infection, is the commonest reason young infants are admitted to hospital in winter, and the sickest are supported with high-flow nasal cannula (HFNC) oxygen. Lying prone is known to improve the mechanics of breathing and oxygen levels in the short term, which raises an obvious question: does that translate into fewer babies needing to be stepped up to more aggressive ventilation? Short-term physiology often fails to convert into a benefit patients actually experience, so the question needed a trial rather than an assumption [s1].
What the trial tested
PROPOSITIS was a multicentre, randomised, open-label trial run in 15 paediatric intermediate-care or intensive-care units in France, and had been registered before enrolment began [s2]. It enrolled infants aged six months or younger, admitted within the previous 24 hours with moderate to severe acute bronchiolitis and requiring HFNC support, between January 2021 and November 2023 [s1]. It was led by an academic hospital group, the Hospices Civils de Lyon, with no drug or device company standing to gain from the result — positioning costs nothing and has no commercial sponsor [s2]. Infants were randomly assigned to the prone position (221) or the supine position (230); those in the prone group were kept face-down for 24 hours or longer during the first 48 hours, and all infants received standardised HFNC support at 2 litres per kilogram per minute [s1]. The primary outcome was the need to escalate care to non-invasive or invasive ventilation within the first 72 hours, judged against prespecified criteria [s1]. Because positioning cannot be concealed, the trial was open-label — clinicians knew which position each infant was in — though the design mitigated this with prespecified escalation criteria and an independent adjudication committee for treatment failure [s1].
What it found
Of 451 infants randomised, 446 were included in the primary analysis — median age 41 days (interquartile range 19 to 72), 54% male [s1]. Escalation of care occurred in 80 infants (17.9%) overall: 33 of 220 (15.0%) in the prone group and 47 of 226 (20.8%) in the supine group [s1]. The adjusted odds ratio was 0.66, but its 95% confidence interval ran from 0.40 to 1.07 and crossed 1 (P = .09), so the difference did not reach statistical significance [s1]. The secondary outcomes did not differ significantly between the groups either [s1].
On safety, serious adverse events were rare and similar: 2 of 180 infants (1.1%) in the prone group and 2 of 264 (0.8%) in the supine group [s1]. Prone positioning of infants carries a particular sensitivity given long-standing safe-sleep advice to place healthy babies on their backs, but this trial concerns closely monitored infants in intensive care, a different setting from unsupervised sleep at home [s1].
How to read it
The authors' own framing is the right one: prone positioning did not significantly reduce escalation of care, and the wide confidence interval means the study was not definitive [s1]. An odds ratio of 0.66 is the kind of effect that would matter clinically if real, but with an interval reaching past 1 the trial cannot confirm it — it was underpowered for the effect size that actually turned up, which is a common fate when event rates come in lower than planned [s1]. The honest reading is neither "prone works" nor "prone is useless," but "not proven, and worth a larger trial" [s1].
The main limits follow from the design: an open-label positioning study cannot blind caregivers, and the trial was conducted in one country's intensive-care system in infants already sick enough to need HFNC, so it does not speak to milder cases or to routine ward care [s1].
What to watch
Whether a larger, adequately powered trial is mounted to settle the question the confidence interval left open, and whether the suggestive 15%-versus-21% gap holds up or fades [s1]. This article describes research and is not medical advice; the positioning and respiratory support of a sick infant are decisions for treating clinicians.
Sources
Sources
More on
A single-dose RSV antibody held up against five monthly shots in high-risk babies
The SMART trial tested clesrovimab against palivizumab in premature and heart- and lung-affected infants. It was built to show safety, not to prove one prevents more RSV than the other.
Skipping routine stomach checks was safe for ventilated children and fed them better
In GASTRIC-PICU, 4,700 ventilated children were randomised to routine gastric residual checks or none. Dropping the checks was non-inferior for survival and ventilator-free days, and improved nutrition.
Intravenous arginine did not shorten sickle cell pain crises in a phase 3 trial
The NIH-funded STArT trial in children and young adults was halted for futility: time to crisis resolution was 60.8 hours with arginine versus 65.8 with placebo, well within the range of chance.
Sodium bicarbonate did not improve survival after in-hospital cardiac arrest
A double-blind Danish trial gave the drug or placebo during resuscitation at 21 hospitals. Restored circulation occurred in 39% versus 37% — no meaningful difference — and bicarbonate raised blood alkalosis and sodium.