When the “Standard of Care” Fails the Trial: What PROSpect’s Negative Findings Reveal About Pediatric Critical Care’s Evidence Crisis

For years, intensivists in pediatric ICUs across three continents have flipped children onto their stomachs and dialed up high-frequency oscillatory ventilation, convinced by physiological logic, small observational studies, and institutional habit that they were helping. The PROSpect trial, published in The Lancet, just told them the evidence was never really there.

The trial enrolled children with moderate-to-severe pediatric acute respiratory distress syndrome (PARDS), randomized them across a two-by-two factorial design to prone versus supine positioning and high-frequency oscillatory ventilation (HFOV) versus conventional mechanical ventilation (CMV), and measured ventilator-free days at 28 days. Neither prone positioning nor HFOV improved the primary outcome. The trial’s conclusion is blunt: routine use of these strategies in this population is not recommended. That verdict lands in PICUs where both interventions have been standard operating procedure for a generation.

Three signals are now converging in pediatric critical care trials. Each one, read in isolation, looks like a one-off finding. Read together, they point to something the field has not yet named: a systematic overestimation of intervention benefit in critically ill children, built on physiological plausibility in the absence of adequately powered randomized evidence.

The Evidence That Was Never There

Start with prone positioning. The rationale has always been compelling: flipping a child improves ventilation-perfusion matching, recruits collapsed lung regions, and reduces ventilator-induced injury. Adult data supported it, particularly after landmark adult trials showed a mortality benefit with prone positioning. Pediatric practice followed the adult signal, as it often does when randomized pediatric data is sparse.

But the adult-to-pediatric extrapolation has a structural flaw. Children’s chest wall mechanics, lung compliance, and respiratory drive differ materially from adults. A 2025 meta-analysis of thirteen randomized controlled trials involving 1,529 pediatric patients found a risk ratio for mortality of 0.67 (95% CI 0.57–0.79) favoring prone positioning, a signal that looked persuasive on a forest plot. PROSpect, with its prospective factorial design and response-adaptive randomization, is the rigorous test that meta-analysis could not substitute for. The signal did not hold.

HFOV carries a parallel history. The physiological promise, delivering small tidal volumes at high frequency to limit barotrauma while maintaining mean airway pressure, made it a logical candidate for the sickest PARDS patients. But adult HFOV trials found no mortality benefit and in some cases suggested harm. Pediatric practice has continued using HFOV well past the point where adult evidence stopped supporting it, partly because critically ill children were assumed to respond differently, and partly because no adequately powered pediatric RCT existed to challenge that assumption. PROSpect filled that gap and found the same null result.

Why Factorial Design Made This Finding Possible

The structure of PROSpect matters as much as its conclusions. A conventional parallel-arm trial testing prone positioning alone, or HFOV alone, would have required two separate large-scale enrollments and still could not have captured interaction effects between the two interventions. The two-by-two factorial design tested both simultaneously, generating four treatment arms (supine plus CMV, supine plus HFOV, prone plus CMV, prone plus HFOV) within a single trial infrastructure.

The response-adaptive randomization layer added precision that fixed randomization cannot deliver. As interim data accumulated, the allocation ratios shifted toward arms showing more favorable signals, protecting against continued enrollment into clearly inferior strategies while preserving statistical power. This is not theoretical elegance: it represents a meaningful reduction in the number of critically ill children exposed to less effective interventions during the trial period itself.

That design architecture is the second signal. Pediatric critical care has historically relied on small single-center studies, post-hoc subgroup analyses of adult trials, and physiological surrogates as the basis for PICU protocols. The PROSpect investigators built something different: an international, multicenter, phase 3 adaptive platform capable of resolving a question that had been generating inconclusive noise for two decades. The trial didn’t just produce a negative result. It demonstrated that this level of methodological rigor is achievable in pediatric critical care, which raises an uncomfortable question about every protocol currently written on weaker evidence.

The Operational Reckoning for Sponsors and Sites

For sponsors designing trials in pediatric critical care, PROSpect reframes the baseline assumption problem. When standard-of-care arms in a pediatric RCT include interventions that have not themselves been validated in adequately powered trials, the control arm may carry its own signal confound. A trial comparing a novel adjunctive therapy to “best supportive care” that includes routine prone positioning just inherited uncertainty about whether that control arm is itself beneficial, neutral, or subtly harmful.

Protocol teams writing PARDS studies in 2027 will need to specify their ventilation strategy explicitly and justify it against PROSpect’s findings. Regulators reviewing pediatric respiratory INDs will ask the question even if sponsors don’t raise it first. The FDA’s existing guidance framework for pediatric study plans (PSPs) requires sponsors to address how pediatric physiology informs trial design. PROSpect does not create new regulatory obligations, but it creates new evidentiary obligations: a sponsor who defaults to prone positioning in a PARDS control arm now needs to explain why, against a published phase 3 RCT that found no ventilator-free day benefit.

For CROs and site networks running pediatric critical care studies, the operational implication is more immediate. PROSpect coordinated enrollment across international PICUs, which means the infrastructure exists to run adequately powered pediatric critical care trials at scale. The question is whether sponsors will use it. The PIVOTAL adaptive platform trial, run through ICNARC, offers a parallel model in pediatric intensive care: multi-center Bayesian adaptive design capable of evaluating multiple interventions simultaneously. Both platforms challenge the assumption that pediatric critical care trials must be small, slow, and underpowered by necessity.

Here is the counterintuitive read that most of the field will miss. The conventional interpretation of a negative trial is that it closes a door. PROSpect’s negative result for prone positioning and HFOV actually opens a harder question: if two of the most physiologically intuitive interventions in PARDS produce no ventilator-free day benefit in a rigorously conducted phase 3 trial, how many other PICU protocols rest on equally untested physiological assumptions? The answer, based on the state of the pediatric critical care literature, is: more than anyone has formally counted.

Within 18 months, expect regulatory reviewers and guideline committees to cite PROSpect as a template, not just a data point. Pediatric study plan submissions that include prone positioning or HFOV as standard-of-care comparators will face documentation pressure. More consequentially, the adaptive factorial design architecture PROSpect used will become harder to argue against in protocol negotiations. Sponsors who resist it on grounds of operational complexity will be asked to explain why a slower, smaller, single-arm trial serves pediatric patients better than the design that just resolved a 20-year evidence gap. The first sponsor to apply PROSpect’s factorial adaptive model to the next unresolved PICU intervention question will set the methodological floor, and every subsequent pediatric critical care IND gets measured against it.

References

1. The Lancet, “Prone positioning and high-frequency oscillatory ventilation in early paediatric acute respiratory distress syndrome (PROSpect): an international, two-by-two factorial, response-adaptive, phase 3, randomised controlled trial”
2. ClinicalTrials.gov, PROSpect Trial Registry (NCT03896763)
3. Frontiers in Pediatrics, Meta-analysis of prone positioning in pediatric ARDS (13 RCTs, 1,529 patients; mortality RR 0.67, 95% CI 0.57–0.79)
4. ICNARC, Paediatric Intensive Care Adaptive Platform Trial (PIVOTAL)