Aspects on Oxygenation in Preterm Infants before, Immediately after Birth, and Beyond

Oxygen is the only drug on the World Health Organization list of 479 essential medicines that has no alternative [1]; i.e., no other drug can be substituted for oxygen. Oxygen is critical for life but too little (hypoxia) or too much oxygen (hyperoxia) can be harmful, lead to serious morbidity, or even be fatal. Whilst the effects of hypoxia are well known in all humans, including newborn infants, hyperoxia is also indelibly linked to harm, including cerebral vasoconstriction [2], major organ injury [3], and in preclinical studies, neurodevelopmental impairment and death [4] via mechanisms such as generation of reactive oxygen species that cause oxidative stress and damage to lipids, proteins, RNA, and DNA [5, 6]. Oxygen control is especially important in premature infants who have immature lungs that require supplemental oxygen to prevent hypoxia but also inadequate antioxidant defenses raising susceptibility to oxygen stress, injury, and toxicity [7, 8]. It was not until relatively recently that the appreciation of the delicate balance between hypoxia and hyperoxia shifted to focus on the preterm infant [9, 10].

The oxygen needs of the preterm infant, especially for those born at margins of viability, are complex and continue to be unclear and confounded by crucial knowledge gaps. For example, a hypoxemic in utero environment is necessary for normal lung development [11] but severe postnatal hypoxia, especially in the first few minutes of life, may increase risk of death and serious morbidities [12]. Oxygen needs may change again days to weeks after birth, particularly if there is evolving bronchopulmonary dysplasia (BPD) where oxygen is both a mainstay of treatment, and perpetrator in pathogenesis [13].

This narrative review will consider historical and emerging evidence informing current recommendations for oxygen therapy for the preterm infant. We will address aspects of oxygen of the fetus prior to birth, but also oxygen needs during the infant’s first few minutes of life, during admission in the neonatal intensive care unit (NICU) and after discharge home. We examine the challenges of applying concepts that are generated from research to clinical care, often within an NICU or home environment that has other competing factors for best practice. Finally, we highlight considerations for ongoing and emerging research which can be conducted to address critical knowledge gaps for ensuring safe and best treatment of the preterm infant with oxygen.

Lack of appropriate oxygen levels during intrauterine life can precipitate serious harm. For example, profound maternal hypoxemia is indelibly linked to fetal neurological injury. The sensitive subventricular and pallidum zones, for example, rapidly undergo apoptosis [14], structural and functional changes [15] during hypoxemia. In utero inflammation may also increase susceptibility to hyperoxia after birth [16]. Animal studies suggest that some level of inflammation may be protective and promote organ maturation [17] but inflammation can also precipitate organ injury particularly within the placenta [18], brain [19], and lungs [20].

A stable hypoxic environment (PaO₂ 20–30 mm Hg) is important for normal lung development during the pseudoglandular and canalicular stages. This process is driven by a group of transcription factors including the hypoxia inducible factors (HIFs) which control >100 genes governing the regulation of cellular metabolism, proliferation, angiogenesis, extracellular matrix formation, and survival [21]. The human HIF system is mature by 8 weeks gestation [22] and high levels of HIF mRNA and protein are detected in the fetal lungs [22, 23], where HIF-1α protein expression in all cell types is induced by hypoxia [24]. HIF-2α protein, on the other hand, is expressed only in vascular endothelium and type II pneumocytes [25].

Genetic studies illustrate the importance of the HIF system. Homozygous deletions of HIF-1α and HIF-1β are fatal in mice by embryonic day 10, with severe cardiovascular malformations being the most common cause of death [26]. HIF-2α gene deletion results in death of 50% of fetuses while survivors have abnormally developed lungs with reduced surfactant production, postnatal respiratory distress, and early death [27].

At birth, infants need to adapt to breathing air, which, even at 21% oxygen, is relatively hyperoxic for the preterm infant compared to the intra-uterine milieu. During the saccular stage of pulmonary development (26–36 weeks gestation), respiratory epithelium thins out, lung acini grow and the capillary network forms, allowing for effective gas exchange [28]. At this stage, type II epithelial cells become more mature and surfactant synthesis and homeostasis mechanisms mature [29].

Immediately after birth, an increased fraction of oxygen is required to meet higher energy demands [30]. However, at this stage, exposure to higher levels of oxygen, even room air, interferes with crucial developmental processes and deranges lung maturation, alveolarization, and surfactant synthesis and production [31]. This has the potential to precipitate a vicious cycle of harm, where more supplemental oxygen and ventilatory support are needed to overcome impaired gas exchange from damaged pulmonary anatomy and physiology. In experimental preterm lamb models of severe RDS, levels of HIF-1α, HIF-2α and their downstream regulation target, vascular endothelial growth factor (VEGF) mRNA, an important factor for normal formation of vascular beds during embryonic development [32] are decreased. This possibly contributes to continuing and chronic abnormalities in lung growth, and possibly, to an increased susceptibility to development of chronic lung disease of prematurity or bronchopulmonary dysplasia if the infant survives [33]. Downregulation of HIF-1α, and VEGF are also implicated in regression or cessation of retinal vascularization causing retinal hypoxia [34]. Later, this leads to pathological retinal neovascularization and proliferation seen in retinopathy of prematurity [34].

Historically, the practice of liberally administering oxygen during delivery room care of newborns [35] was driven by the rationale that hypoxia was the underlying cause of cardiorespiratory depression. The hypoxia concept was further reinforced in 1862, when William Little associated birth asphyxia with cerebral palsy [35]. For decades, oxygen was used for newborn birth resuscitation despite no evidence of either harm or benefit [35]. There was further uptake of this therapy into standard practice when oxygen was noted to drive improvement in the standardized newborn resuscitation score, the Apgar score, primarily by increasing the score for color or “pinkness” [36].

In the early 1990’s, Ola Saugstad and Siddarth Ramji conducted the first delivery room trials. Based on evidence from animal studies showing increased oxidative stress and injury with pure oxygen exposure after asphyxia [37, 38] the Resair studies randomized depressed mature infants (≥35 weeks gestation) to respiratory support with either air (21% oxygen) or pure (100% oxygen) oxygen based on their day of birth (quasi-randomization) [39, 40]. Infants were given either air or 100% oxygen at initial resuscitation. Those in the air group were switched over to 100% oxygen if there was no improvement after 90 s. Oxygen saturations were not monitored because that was not standard practice at that time. Oxygen concentrations were also not adjusted during resuscitation – infants received either air of 100% oxygen.

The Resair studies and others that followed showed for the first time that: (1) newborn infants could be resuscitated with air instead of oxygen without major difference in death [40] and neurodevelopmental morbidity [41], and (2) using 100% oxygen even for a few minutes increased the risk of oxidative stress and major organ injury that could last for weeks. These studies formed the basis for several meta-analyses which collectively showed in >1,300 infants, that initiating resuscitation of asphyxiated term or near term infants with air decreased early death (typical risk ratio, RR: 0.71 (0.54, 0.94), typical RD: −0.05 (−0.08, −0.01), NNT: 20 (12, 100) [42‒44] and did not worsen neurodevelopmental in survivors until 2 years of age [45].

This novel finding challenged decades of practice and created a paradigm shift in resuscitation guidance. From 2005, consensus on science and treatment recommendations stated for the first time that either air (21% oxygen) or 100% oxygen could be used to initiate respiratory support of hypoxic/asphyxiated term/near term infants if oxygen was not available [46]. In some countries like Australia, guidelines stated that air should be used for all infants regardless of gestation, due to concerns for oxidative stress and injury with the use of 100% oxygen [47].

Prior to the widespread availability of pulse oximetry devices, the effect of delivery room resuscitation procedures on oxygen levels of sick newborn infants during the first minutes of life was unknown [48]. Resuscitation measures were guided by clinical signs (e.g., color, heart rate, appearance, respiratory efforts).

Due to the findings of the Resair [39, 40] and subsequent works [44] demonstrating clinical impact and biochemical and organ injury from exposure to even a few minutes of oxygen [3], there was a rapid loss of equipoise for using high levels of oxygen in the delivery room [49]. Therapeutic drift toward using lower levels of oxygen was further reinforced by observational studies demonstrating that healthy mature infants took up to 7–8 min to reach SpO₂ levels of at least 90% [50, 51] that administration of 100% oxygen rapidly increased SpO₂ to supraphysiological levels and was associated with oxidative stress, organ injury and even reduced cerebral blood flow [2, 52].

As technology improved in the early 2000’s [53], the routine use of pulse oximetry became standard in the delivery room as an accessible and non-invasive method of measuring response to oxygen therapy [54]. Even though many centres caring for preterm infants were routinely using SpO₂ monitoring since as early as the 80’s [48], it was not until 2010 that consensus statements suggested that fractional inspired oxygen (FiO₂) should be adjusted to target SpO₂ levels derived from healthy, full-term, spontaneously breathing infants [55] during the first 10 min of life. Despite this, whether these titration strategies were achievable in practice and whether oxygen titration benefited preterm infants, were unknown.

Blending and monitoring oxygen levels require substantial investment in equipment such as oxygen blenders and pulse oximeters and in education [56, 57]. In a survey of 695 clinicians from 21 countries in 2020 to 2021, Sotiropoulos et al. [58] found that 28% and 34% of clinicians from high and low-middle income countries, respectively, did not have oxygen blenders in the delivery room even 11 years after initial recommendations. Indeed, only 45% of clinicians titrated FiO₂ to target predetermined SpO₂ levels.

A series of RCTs were conducted to examine whether using lower oxygen levels (ranging from 21% to 30%) to initiate resuscitation of preterm infants could decrease risk of death and other morbidities, compared to using higher (60%–100%) levels of initial oxygen levels. The majority of these studies [59‒68], except for three [69‒71], were developed prior to acquisition of knowledge of “normal” healthy infant early SpO₂ [51] and publication of guidelines for FiO₂ blending and SpO₂ targeting [72]. These studies also used different initial FiO₂ levels and changed FiO₂ to achieve a wide range of SpO₂ targets. Individually, none of these studies were sufficiently powered to answer the primary question of whether using lower (FiO₂ ≤0.3) levels of oxygen improved survival without neurodevelopmental disability in infants below 32 weeks gestation, when compared to higher (FiO₂ ≥0.6) initial oxygen levels. The Targeted Oxygen for the Resuscitation of Preterm Infants and their Developmental Outcomes (To2rpido) study remains the largest (n = 290) study but even this was ended at 15% recruitment due to loss of equipoise from clinicians using high levels of oxygen for newborn resuscitation [69].

Standard pairwise meta-analyses of these studies have yielded inconclusive results [73]. The NETMOTION network meta-analysis of individual participant data examined whether low (21–30%) intermediate (approximately 60%) or high (90–100%) initial FiO₂ improved morality in preterm infants <32 weeks’ gestation. In analysis of 12 RCTs with total 1,055 participants, high initial FiO₂ possibly reduced mortality (OR: 0.45 95% credible interval 0.23–0.86) compared to low initial FiO₂ [74, 75]. However, heterogeneity in titration strategies and risk of bias due to blinding downgraded the certainty of evidence to low. Several ongoing studies will be synthesized in a prospective meta-analysis, which may add further insights [76].

Data obtained from the 8 RCTs provided information on the evolution of SpO₂ levels in preterm infants <32 weeks’ gestation treated with different levels of oxygen during the first few minutes of life. These studies found that only 23% (159 out of 768) participants met study SpO₂ targets despite strict trial conditions, illustrating the difficulties in meeting recommendations from international guidance [77]. Pooled analyses showed that those who were initially resuscitated with lower fractions of inspired oxygen (FiO₂ ≤30%) were less likely to reach the minimum recommended SpO₂ of 80% by 5 min [77]. No infant died during resuscitation but those who did not reach an SpO₂ of 80% by 5 min were more likely to have lower heart rates (mean difference −8.37, 95% CI: −15.73 to −1.01, p < 0.05) [77], which was in turn, associated with increased risk of severe intraventricular hemorrhage (IVH; OR: 2.04, 95% CI: 1.01–4.11, p < 0.05) and death (OR: 4.57, 95% CI: 1.62–13.98) [77]. Even though further study is needed to determine best SpO₂ targets and whether reaching these targets are feasible for clinicians in routine practice, existing information suggests that oxygen levels are important even in the first few minutes of life.

The need to maintain appropriate oxygen levels in the preterm infant does not stop even after the infant is admitted into the NICU, where evolving lung physiology and other competing factors such as infection may interfere with oxygen therapy. The importance of the need to maintain oxygenation within a safe margin was shown by the NeOProM meta-analysis which prospectively synthesized individual patient data from 5 RCTs that aimed to compare outcomes of infants randomized to be nursed in lower (85–89%) or higher (91–95%) SpO₂ levels. NeOProM examined oximetry data from 4,965 infants below 28 weeks gestation and found no significant difference regardless of whether the infants were nursed in lower or higher SpO₂ target ranges for the primary outcome of death or major disability at 18–24 months corrected age.

This did not mean that oxygen levels were unimportant. Nursing infants within lower SpO₂ target ranges was associated with a higher risk of death and necrotizing enterocolitis and a lower risk of retinopathy of prematurity requiring treatment (e.g., laser). However, limitations of the NeOProM studies should be recognized. Oxygen saturation probe calibration issues, amount of time spent in the intended target ranges, and differences across centers reduce the applicability of evidence to day-to-day practice. Nevertheless, the studies from which NeOProM was derived were the first indications that oxygen imbalance (whether too high or too low) in the NICU could have major consequences on preterm infants <28 weeks gestational age [78].

Maintaining desired oxygen saturations is a challenge in the NICU. Lim et al. [79] showed that in some cases, FiO₂ had to be adjusted >40 times a day to reach specific SpO₂ levels. This is an onerous task that is probably impossible to manage in NICUs with low staff to patient ratios [80]. One study found infants managed with clinician-controlled supplemental oxygen systems had many hyperoxic or hypoxic episodes, spending only 31% of time in target ranges. This effect, not surprisingly, was correlated to lower nurse to patient ratios [80].

In the last decade, devices with algorithms that provide automated oxygen control have been developed and these show great promise in improving adherence to recommended SpO₂ targets [81]. One recent systematic review and meta-analysis of thirteen RCTs comparing clinician-controlled versus automatic control algorithms showed automated oxygen control increased time spent within SpO₂ target range (90–95%), reduced time in hypoxic and hyperoxic ranges, and reduced burden on staff with decreased number of adjustments [82]. Further study is required to determine the impact of these algorithms on important longer term clinical outcomes including survival, bronchopulmonary dysplasia, pulmonary hypertension, retinopathy, and neurodevelopmental impairment.

Infants with bronchopulmonary dysplasia and other respiratory conditions may need longer term treatment with oxygen, even after hospital discharge. Home oxygen therapy (HOT) improves feeding and growth but whether HOT or SpO₂ monitoring at home benefit longer term outcomes, such as neurodevelopment is uncertain. DeMauro et al. [83] followed up 1,039 infants discharged home on supplemental oxygen and matched them to infants breathing room air. Those prescribed HOT had marginally better weight z scores (adjusted mean difference 0.11; 95% confidence interval (CI) 0.00–0.22), but were more likely to be rehospitalized for respiratory illnesses (adjusted relative risk 1.33; 95% CI: 1.16–1.53). Whether this was because HOT was more likely to be given to infants with worse lung disease and therefore more susceptible to later respiratory problems, is uncertain and needs to be clarified.

In the Remote Home Oximetry (RHO) study, Rhein et al. [84] randomized 196 infants with bronchopulmonary dysplasia (mean gestational age 26.9 weeks) who required HOT to either standard care (monthly clinic visits with in clinic weaning attempts of HOT) or RHO (weekly transmission of stored SpO₂ data to the study team by parents for advise on changing oxygen treatment). Infants in the RHO group had reduced duration of HOT (78.1 vs. 100.1 days) and were reported to have improved quality of life scores.

Currently, the ability to monitor and respond to oxygen treatment at home is limited by technology. Oximeters are wired machines that are limited by battery life and cause considerable inconvenience, distress, reduced sleep quality, and alarm fatigue for the infant’s caregivers [85]. Wireless technologies that have capacity to transmit SpO₂ data to central clinical stations, such as those used during the COVID-19 pandemic to provide safe out of hospital care for infected patients, may potentially improve capacity to oversee and ensure safe oxygen treatment for infants at home [86].

The harms and benefits of oxygen for the preterm infant must be managed meticulously. Additional evidence is urgently needed to address critical knowledge and practice gaps, especially during the first few minutes of life, when the infant’s physiology is changing at its most rapid pace. The use of novel technologies such as mobile oximeters, remote monitoring and devices equipped with automated oxygen-controlled systems may be of great potential to reduce the burden of over and under-oxygenation for the preterm infant, especially outside of a hospital setting.

The authors have no conflicts of interest to declare.

J.X.S. is supported by a National Health Medical Research Council Postgraduate Scholarship (Australia, GNT2031240). The funder had no role in manuscript conception, planning, writing and decision to submit for publication.

All authors contributed to conceptualization, investigation and writing – review and editing. J.X.S. and J.L.O. were responsible for writing – original draft. J.X.S. was responsible for project administration. J.L.O. and O.D.S. were responsible for supervision. All authors critically appraised and approved the final manuscript for submission. All authors meet ICMJE authorship requirements.