Preterm Patent Ductus Arteriosus Trials Have Missed the Right Patients
Management of patent ductus arteriosus in preterm infants has long sat in an uncomfortable place between physiology and proof. Randomised trials using pharmacotherapy, most often ibuprofen as a cyclooxygenase inhibitor, have not delivered the expected clinical gains when infants were enrolled through standard echocardiographic criteria such as ductal diameter and shunt direction alone. In those studies, closing the duct was not enough to improve outcomes, and some treatment strategies carried evidence of harm.
The central problem is not that ductal shunting is irrelevant. It is that a duct is not a phenotype, and not every visible shunt carries the same haemodynamic burden.
Shunt Volume, Not Just Duct Size, May Define the High-Risk Infant
A more selective approach focuses on moderate-high volume PDA shunts and on infants whose echocardiographic profile suggests meaningful circulatory compromise. The biological logic is straightforward. A large left-to-right shunt can steal systemic blood flow, increase pulmonary overcirculation, and worsen respiratory instability, but the magnitude of injury depends on how much blood actually crosses the duct and how the infant tolerates that load. Two infants may share a similar duct diameter while having very different shunt burdens and very different trajectories at the bedside.
This matters because the current selection method can misclassify both low-risk and high-risk babies. Low-risk infants may receive drug exposure without any realistic chance of benefit, while high-risk infants may accumulate the adverse effects of therapy and ongoing haemodynamic stress if treatment fails to reduce the shunt. The failure mode is built into the design: the wrong infant receives the wrong intervention at the wrong time.
Why Earlier Randomised Evidence Has Been So Difficult to Translate
Earlier randomised controlled trials do offer an important lesson, even if that lesson is uncomfortable. Their neutral or harmful results do not necessarily mean that all active treatment for PDA is futile. They may instead reflect imprecise phenotyping, suboptimal timing, and the use of interventions that were not potent enough for the haemodynamic problem being treated. If the trial population contains many infants whose ductus would have closed spontaneously, then any drug effect is diluted. If the sickest infants are enrolled after systemic compromise is already established, closing the duct may arrive too late to reverse downstream injury.
The review argues that the unanswered question is narrower than the broader clinical debate has often suggested: does targeted management of high-volume shunts in the highest-risk preterm infants alter meaningful outcomes when the diagnosis is based on comprehensive echocardiographic adjudication rather than simple anatomic thresholds?
Trial Design Will Have to Be More Clinical and Less Convenience-Based
Future studies will need to be built around phenotype, not convenience. That means defining the study population using haemodynamic profiles that identify infants at genuinely high risk from ductal shunting. It also means deciding whether pharmacological treatment is the right tool or whether non-pharmacological strategies deserve equal attention. The article makes clear that both pathways remain underexplored and that neither can be dismissed on the basis of the existing trial literature alone.
Choice of intervention is only part of the problem. Timing is equally important, because the same therapy can have different effects depending on whether it is delivered before pulmonary vascular congestion, systemic hypoperfusion, or ventilatory escalation has become entrenched. In a preterm infant, a persistent high-volume shunt can amplify respiratory support needs and destabilise organ perfusion; if treatment is delayed until that cascade has advanced, the trial may measure irretrievable injury rather than prevention.
One sentence captures the challenge plainly: a drug cannot rescue a physiology it no longer controls.
Outcomes Must Matter to Families and Bedside Teams
The review also pushes trialists toward outcome measures that are clinically meaningful rather than merely convenient for data capture. Surrogate endpoints can track ductal closure, but closure alone does not answer the question that matters most to families and neonatal teams: did the infant do better in ways that count at the bedside? Relevant outcomes must reflect survival, morbidity, and the burden of care in a way that aligns with what parents and clinicians actually experience.
That is where collaboration becomes essential. High-quality practice-changing research will require haemodynamic specialists, epidemiologists, and trial methodologists to work from the same playbook. Haemodynamic expertise is needed to phenotype the infant and determine whether the duct is truly driving instability. Trial methodologists are needed to prevent design flaws that blur treatment effects. Epidemiologists are needed to ensure that selection criteria and outcome definitions support causal interpretation rather than noise.
Observational Signals and Randomised Proof Do Not Yet Align
Observational studies continue to suggest that some infants with significant ductal shunting may be different from the broader preterm population, but observational data cannot answer the causal question on their own. Confounding by indication is especially strong in neonatal intensive care, where the sickest infants are often the most likely to be treated and the most likely to have poor outcomes regardless of therapy. Randomised evidence remains the only credible path to separating treatment effect from disease severity.
The review therefore positions the field at a practical crossroads. Existing trials have not established that active treatment improves outcomes for the average preterm infant selected by standard echocardiographic criteria. Yet the door remains open for a narrower and more biologically coherent strategy focused on infants with high-shunt volume PDA and demonstrable haemodynamic risk.
The Next Question Is Not Whether to Treat, but Whom to Study
The article’s most important contribution is its insistence on precision. The field does not need another broad trial that repeats the limitations of earlier work. It needs a design that asks whether reducing a truly significant shunt in the right infant, with the right intervention and the right timing, can change outcomes that matter.
Until then, the evidence base will remain split between a mechanistic rationale for treatment and a clinical trial record that has not yet matched that rationale. The gap is not theoretical. It is measured in ventilator days, haemodynamic instability, drug exposure, and the unresolved question of which preterm infants actually stand to benefit from active ductal intervention.