Editorial 

Reconsidering Oxygen at Birth for Preterm Infants: Lessons From TORPIDO 30/60

Alyssa R. Thomas, Elizabeth E. Foglia

JAMA Published Online: December 10, 2025

doi: 10.1001/jama.2025.23127

At birth, newborns must transition from the low-oxygen environment of the womb to postnatal normoxemia. Most preterm newborns require supplemental oxygen to support this process. Yet, active debate remains about the optimal starting fraction of inspired oxygen (Fio₂) for preterm delivery room resuscitation. Historically, high oxygen concentrations (Fio₂ of 1.0) were used routinely, guided by the belief that maximizing oxygen delivery was essential to support cardiorespiratory compromise. Over time, recognition of oxygen toxicity¹ and associations between high Fio₂ and increased mortality in more mature newborns^2-4 prompted a shift toward lower oxygen strategies for preterm newborns.⁵ In 2015, international guidelines for preterm infants recommended initiating respiratory support with lower Fio₂ (0.21-0.3), based largely on concerns that unnecessary oxygen exposure may cause harm without clear benefit for mortality, major morbidities, or neurodevelopment.^6,7 However, the recent NETMOTION collaborative individual participant data network meta-analysis, pooling 12 randomized trials including 1055 preterm newborns, challenged the start low paradigm, finding that higher initial Fio₂ (≥0.9) was associated with reduced mortality compared with lower (≤0.3) or intermediate (0.5-0.65) concentrations.⁸ These emerging data underscore that the optimal starting Fio₂ for preterm newborns remains undefined—a question with direct implications for millions of premature newborns worldwide each year.

In this issue of JAMA, Oei and colleagues⁹ report the TORPIDO 30/60 trial, the largest randomized clinical trial to date evaluating initial oxygen strategies immediately after birth for preterm newborns. Conducted across 31 sites in 6 countries, the trial analyzed outcomes for 1469 neonates born at 23 to 28 weeks’ gestation who were randomized to undergo resuscitation with initial Fio₂ of 0.6 (n = 728) or Fio₂ of 0.3 (n = 741). The authors should be commended for both the scale of enrollment and thoughtful use of consent waivers, developed with parent advocates, to enhance generalizability. The primary outcome, a composite of death or major brain injury at 36 weeks’ corrected gestational age, was similar between both groups (46.9% for Fio₂ of 0.6 vs 47.8% for Fio₂ of 0.3; relative risk, 0.98 [95% CI, 0.89-1.09]). Similarly, the individual components of the primary outcome did not differ between the groups.

How should we interpret these results? The NETMOTION meta-analysis prompted many to reexamine the role of initial Fio₂ to prevent mortality for preterm newborns, yet TORPIDO 30/60 did not demonstrate a survival benefit in the Fio₂ of 0.6 group. It is important to note that TORPIDO 30/60 specifically compared 2 relatively moderate oxygen strategies, thereby addressing a portion of the Fio₂ spectrum in which prior evidence suggested the least separation in outcomes. This context is apparent when viewed alongside the NETMOTION IPD meta-analysis, which reported the strongest mortality benefit for participants randomized to receive the highest initial Fio₂(≥0.9) compared with both the lowest (≤0.3) and intermediate (0.5-0.65) concentrations.

One possible interpretation of these collective findings is that initial Fio₂ during resuscitation may not be a major determinant of mortality for preterm newborns. The evidence in NETMOTION was rated as low or very-low certainty, in part due to heterogeneity across trials. If higher initial Fio₂ was associated with decreased mortality, one might reasonably expect a consistent dose-dependent pattern to emerge, with highest mortality rates among low initial Fio₂ strategies and interval improvement with intermediate and high initial Fio₂. However, point estimates for mortality by treatment groups in NETMOTION (14% for low Fio₂, 15% for intermediate, and 9% for high) did not demonstrate such a stepwise effect. Mortality rates in TORPIDO 30/60 were remarkably consistent: death by 36 weeks’ corrected gestational age occurred in 15.8% of patients in the Fio₂ of 0.3 group and 15.4% in the Fio₂ of 0.6 group. Alternatively, a threshold effect may exist, with survival benefits manifesting only at higher oxygen concentrations not tested in this trial. Whether even higher initial Fio₂ (such as 1.0) confers additional advantage without introducing new risks remains uncertain and requires further investigation. Importantly, TORPIDO 30/60 has not yet reported key oxygen-related morbidities, including bronchopulmonary dysplasia and retinopathy of prematurity, outcomes critical to understanding whether higher early oxygen exposure carries latent risks.

Although the primary outcome did not differ between treatment groups, the tertiary outcomes offer additional insight into short-term physiologic effects. Newborns randomized to receive the higher Fio₂ of 0.6 received chest compressions and epinephrine less frequently than those randomized to receive the lower Fio₂ (2% vs 5% for chest compressions and 1% vs 2% for epinephrine). However, these observed advantages were based on rare events and should therefore be interpreted with caution. The fragility index for these events illustrates this point: statistical significance for both outcomes would be lost with as few as 5 event changes for chest compressions and only 1 event change for epinephrine. Such statistical fragility¹⁰ highlights the need for replication before drawing firm conclusions. Additional evidence from the forthcoming HiLo trial (initial Fio₂ of 0.3 and 0.6, with planned enrollment of 1200 preterm newborns)¹¹ and future pooled analyses may provide greater precision.

The physiologic end points measured in TORPIDO 30/60 are also noteworthy. Newborns randomized to receive initial Fio₂ of 0.6 more frequently achieved oxygen saturation as measured by pulse oximetry (Spo₂) greater than or equal to 80% and heart rate greater than 100/minute at 5 minutes, targets previously associated with reduced mortality and severe intraventricular hemorrhage.^12-16 Observational analyses of multicenter cohorts demonstrated an association between 5-minute Spo₂ less than 80% and increased risk of death or severe intraventricular hemorrhage.¹⁴ Similarly, newborns with both prolonged bradycardia and Spo₂ less than 80% at 5 minutes experience increased odds of in-hospital mortality.¹⁶ Yet, the short-term physiologic improvements observed in TORPIDO 30/60 did not translate into reductions in death or major brain injury. This disconnect underscores the limitations of using early vital sign end points as surrogates for complex neonatal outcomes. The previous studies demonstrating an association between time-based vital signs and clinical outcomes were observational; these markers may reflect underlying illness severity more than modifiable physiology. Alternatively, perhaps transient improvements in oxygenation in the first minutes after birth do not protect against subsequent physiologic instability over days to weeks that culminates in brain injury or death. Either way, these results emphasize the need to identify time-critical physiologic targets that truly drive outcomes. Moving forward, progress should focus on prospectively testing varied physiologic targets during neonatal resuscitation, as planned in the ongoing OptiSTART trial,¹⁷ and developing more adaptive titration strategies to achieve those goals.

A key lesson from the TORPIDO 30/60 trial may be to not discount the improvements in postnatal physiology observed with the higher Fio₂ of 0.6 simply because clinically relevant outcomes were equivalent in the higher vs lower Fio₂ groups. Optimizing early stabilization may set the course even if it does not guarantee the desired destination. Meaningful progress will likely emerge through steady cumulative advances that bridge the delivery room and beyond. Within this continuum, delivery room management remains essential, not as an isolated event, but as the foundation on which future gains in survival and healthy development are built. The landmark work of Oei and colleagues in TORPIDO 30/60 serves as both culmination and catalyst, reflecting decades of inquiry to reshape understanding of oxygen in preterm transition and setting the stage for future investigation.