A Review of Oxygen Use During Chest Compressions in Newborns—A Meta-Analysis of Animal Data

Background: International consensus statements for resuscitation of newborn infants recommend provision of 100% oxygen once chest compressions are required. However, 100% oxygen exacerbates reperfusion injury and reduces cerebral perfusion in newborn babies. Objective: We aimed to establish whether resuscitation with air during chest compression is feasible and safe in newborn infants compared with 100% oxygen. Methods: Systematic search of PubMed, Google Scholar and CINAHL for articles examining variable oxygen concentrations during chest compressions in term newborns. Results: Overall, no human studies but eight animal studies (n = 323 animals) comparing various oxygen concentrations during chest compression were identified. The pooled analysis showed no difference in mortality rates for animals resuscitated with air vs. 100% oxygen (risk ratio 1.04 [0.35, 3.08], I2 = 0%, p = 0.94). ROSC was also similar between groups with a mean difference of −3.8 [−29.7–22] s, I2 = 0%, p = 0.77. No difference in oxygen damage or adverse events were identified between groups. Conclusions: Air had similar time to ROSC and mortality as 100% oxygen during neonatal chest compression. A large randomized controlled clinical trial comparing air vs. 100% oxygen during neonatal chest compression is warranted.


INTRODUCTION
Approximately, 3% of infants require respiratory assistance at birth and 0.1% require chest compressions (CC) (1,2). During respiratory support of term and near term newly born infants air (21% oxygen) should be given as trials and meta-analyses reported a significant reduction in mortality in infants resuscitated with air (relative risk 0.71 [95% CI 0.54 to 0.94], risk difference −0.05 [−0.08 to −0.01]) (3,4), which is also reflected in the neonatal resuscitation guidelines (1,2). Oxygen (O 2 ) use in the delivery room is associated with potential adverse effects; hyperoxia slows cerebral blood flow, brief periods of 100% O 2 causes long-term reductions in cerebral blood flow. High concentrations of O 2 lead to generation of oxygen free radicals, which have a role in reperfusion/reoxygenation injury after asphyxia especially to oxyregulator tissues such as myocardium (5)(6)(7)(8). Thus, air might be a more appropriate gas than 100% O 2 .
If starting with air has been unsuccessful, the current resuscitation guidelines suggest to titrate oxygen and increase oxygen to 100% once chest compressions are started (1,2). However, there is lack of supporting evidence of the beneficial effects of either titrating oxygen or using 100% O 2 during cardiopulmonary resuscitation (CPR). The aim of oxygen use during resuscitation is to reactivate mitochondrial activity and energy provision and prevent tissue damage from oxygen deprivation during asphyxia while avoiding adverse effects of oxidative stress on the respiratory system and cerebral circulation; and tissue damage from oxygen free radicals (9).
If air is equally effective as 100% O 2 in newborn infants requiring CC, the use of air instead of 100% O 2 could reduce morbidity and mortality in asphyxiated infants. The aim of the meta-analysis was to compare the efficacy of air compared to 100% O 2 during chest compression in the resuscitation in newborn infants immediately after birth. Further aims included assessment of oxidative stress and inflammatory markers using air vs. 100% O 2 .

METHODS
We searched PubMed, Google Scholar, and CINAHL using the following search terms (last searched on June 6, 2018, Appendix 1): "infant, " "newborn, " "resuscitation, " "chest compression, " "oxygen, " and "delivery room." Publications were assessed based on title, abstract, and methods. Studies were included if they compared different oxygen concentrations during CC. Studies were excluded when no CC were performed, or if they did not define the infants as newborn. The initial search was aimed to only identify human trials. However, no human trials were identified and therefore the search was expanded to include animal studies. A manual search through the references of the obtained articles was additional performed.

Study Selection
Two reviewers (CGH and GMS) independently reviewed citations for selection. Studies were included in the review if they met the following criteria: randomized controlled trial; comparing use air vs. 100% O 2 during neonatal CPR; and presented the outcomes of either death or ROSC. Our primary outcome measure was mortality during neonatal CPR. Secondary outcomes included time to ROSC, oxygenation, and indicators Abbreviations of organ injury or damage. Full articles for potentially relevant studies were retrieved and independently assessed for their eligibility using a standardized data collection form. We also aimed to identify and if available include multiple publications describing the same study. Authors were contacted for data on return of spontaneous circulation (ROSC) as they were only reported as median (IQR) in the respective articles. No language restrictions were applied. Discrepancies regarding inclusion were resolved with another member of the review team (ALS).

Data Extraction
Data were recorded using a standardized data collection form to record study design and methodological characteristics, patient characteristics, interventions, and outcomes thereof, including their RR (95% CI). Data extraction was independently performed by two investigators (GMS, CGH) and discrepancies were resolved in consultation with another member of the review team (ALS).

Statistical Analysis
The principal summary measures were RR (95% CI) for dichotomous outcomes. Heterogeneity was explored using a chisquare test, and the quantity of heterogeneity was measured using the I 2 statistic. We summarized RR estimates using randomeffects models. Analyses were performed in RevMan version 5.3 (Cochrane Collaboration, 2014). All p-values are 2-tailed.

RESULTS
We did not identify any clinical human study on the subject. Therefore, all results come from controlled animal models. Six (75%) studies used a swine model (10)(11)(12)(13)(14)(15), and two (25%) an ovine model (16,17). Included studies assessed mortality (Figure 1), ROSC and/or circulatory recovery, oxygenation, severity of oxygen injury, and adverse events caused by oxygen. Table 1 presents all identified animal studies, intervention groups, and primary outcome, and conclusion ( Table 1). Table 2 presents the pH, pCO 2 , base excess, and lactate prior the start of resuscitation ( Table 2). Studies resuscitated animals with air or 100% O 2 during CPR. All piglet studies used a post-transitional model (18), while the lamb studies used a fetal-to-neonatal transitional model (18). Dannevig et al. (10,11) assessed brain and lung inflammation/injury in asphyxiated newborn piglets resuscitated with 21 vs. 100% O 2 , different compression to ventilation ratios (C:V) and variable durations of initial positive pressure ventilation. Alsaleem   O 2 while an additional study compared 3:1 C:V vs. continuous CC with asynchronous ventilation (CCaV) in addition to air vs. 100% O 2 and assessed time to ROSC and oxidative stress (14,15).

Mortality
For the pooled analysis, the study by Solevåg et al. (15) was excluded because the study by Dannevig et al. (10) included the same piglets reported by Solevåg et al. (15). A total of 323 animals were included with an overall mortality of 17(5%). All animals in the ovine studies survived and therefore no pooled analysis was possible. For the outcome of mortality, the pooled analysis of the porcine studies showed no difference in mortality for piglets resuscitated with air vs. 100% oxygen (odds ratio 0.84 [0.26, 2.72], p = 0.77, I 2 = 0%; Figure 1).

Return of Spontaneous Circulation and Circulatory Recovery
Overall, the individual studies and the pooled analysis showed no difference in the time to ROSC with air vs. 100% O 2 with a mean difference of  (12,13) randomized piglets to air for the duration of resuscitation, 100% O 2 for 3 min followed by air, or 100% O 2 for 30 min followed by air after apnea induced cardiac arrest. Time until ROSC (heart rate >150 bpm) were similar among groups with median (IQR) times of 67 (60-76) s , 88 (76-126) s, and 68 (56-81) s in the air, 100% for three min group, or 100% for 30 min group, respectively.
Alsaleem et al. (17) induced cardiac arrest in fetal lambs by umbilical cord occlusion. The lambs were delivered and resuscitated with air then randomized to continue to receive air at the onset of CC or the O 2 to be increased to 100%. There was no difference in mean (±SD) time to ROSC 211 (±145) s and 306 (±270) s in the air and 100% O 2 group, respectively. Perez-de-Sa et al. (16), induced cardiac arrest in fetal lambs and randomized them to either air, 100% for the first 3

Oxygenation
All studies confirmed the significant arterial hyperoxemia in the group of animals resuscitated with 100% oxygen. Using cerebral near-infrared spectroscopy, studies measured speed of recovery of brain oxygenation by how fast cerebral regional oxygen saturation reached 30% and when brain tissue oxygenation (PbtO 2) had increased 0.1 kPa from its nadir. Studies in near-term lambs reported higher PbtO 2 with 100% O 2 , and significantly increased partial pressure of arterial oxygen levels immediately after ROSC compared to lambs who remained in air (165 ± 145 vs. 41 ± 16 mmHg), (p = 0.046) (17). However, throughout CC, no significant difference between blood oxygen saturation, partial pressure of arterial oxygen, arterial oxygen content, or O 2 delivery to the brain between groups was observed (17). Perez-de-Sa et al. reported brain tissue oxygenation as measured by the partial pressure of oxygen in extracellular fluid (PbtO 2 ) reaching a maximum of 56 kPa (420 mmHg) in the group initially ventilated with 100% O 2 for 30 min, a level only previously seen in hyperbaric conditions (16

Indicators of Organ Injury or Damage
There were two studies that also examined the tissue oxidative stress or damage when 100% O 2 vs. air was used in the resuscitation after asphyxia induced cardiac arrest in newborn piglets. Dannevig et al. (10,11) found no difference in brain or lung inflammatory markers (e.g., lactate, lactate/pyruvate, or IL-1) in the groups with 100% O 2 or air. Dannevig et al. also examined damage associated with different C:V ratios and  duration of initial PPV (10,11), however these results were excluded as they were not within the scope of this review. Damage caused by oxidative stress was measured in the piglet study assessing 100% O 2 and air with 3:1 C:V or CCaV (14). The oxidized glutathione to glutathione ratio, a marker for oxidative stress, was significantly higher with 100% O 2 than air, 0.14 (0.11-0.22) and 0.10 (0.08-0.11) (p = 0.005), respectively.

DISCUSSION
Current neonatal resuscitation guidelines recommend 100% O 2 during CPR, however the most effective oxygen concentration in newborn infants remains controversial (2,19,20). Oxygen has been used in neonatal resuscitation for over 200 years (21). Its use spread rapidly in response to reports of brain damage in infants who had survived birth asphyxia (22). The inclusion of skin color in the Apgar score further contributed to an increased use of oxygen in the delivery room. The use of 100% oxygen was accepted based on experts' opinion despite a lack of experimental evidence. However, over the last decades the use of 100% O 2 has been questioned as even a brief exposure to 100% O 2 may be detrimental and several studies reported that air is as effective as 100% O 2 (4,(23)(24)(25)(26)(27); Indeed, 21% O 2 resulted in a significant reduction in mortality (28/ (4,4). This has led to a change in the 2010 neonatal resuscitation guidelines to start respiratory support with air and oxygen delivery titrated according to target oxygen saturations in term and near-term newborn infants (28). While these studies have examined oxygen use during respiratory support of term and near-term newborn infants no human study has compared air vs. 100% O 2 during neonatal chest compression.
High concentrations of O 2 delivery during CPR generates O 2free radicals, which play major role in reperfusion/reoxygenation injury after asphyxia, especially to oxyregulatory tissues (e.g., myocardium) (28). During CPR 100% O 2 causes significant arterial hyperoxemia and increased partial pressure of arterial oxygen levels immediately after ROSC (165 ± 145 vs. 41 ± 16 mmHg, p = 0.046) compared to resuscitation with air (17). However, this did not result in higher oxygenation at the brain. Furthermore, tissue oxygenation, and tissue/organ damage were not significantly different between air and 100% O 2 . Indeed, only the study by Solevåg   glutathione ratio, a marker for oxidative stress, was significantly higher with 100% O 2 than air, 0.14 (0.11-0.22) and 0.10 (0.08-0.11) (p = 0.005), respectively (14). However, other studies did not report any difference in brain or lung inflammatory markers (e.g., lactate, lactate/pyruvate, or IL-1) in the groups with 100% O 2 or air (10,11). Based on these findings pure oxygen is not associated with damage to the nervous system or lungs any more than air but is a cause of oxidative stress. In addition, 100% O 2 exposure at birth has been associated with increased risk of neonatal mortality and childhood cancer (29,30).

LIMITATIONS
Limitations of the current review among others include (i) only data from animal studies were included, (ii) the data should not be directly extrapolated to clinical practice, (iii) different animal models (e.g., piglets and sheep), (iv) transitional model (lambs delivered via cesarean section) vs. post-delivery model (piglets 1-3 days old), (v) induction of cardiac arrest by asphyxia or potassium chloride, or (vi) all animals were intubated with a tightly sealed endotracheal tube (except the latest by Linner et al.) to prevent any endotracheal tube leak (10)(11)(12)(13)(14)(15)(16)(17)(18)31). These variations might have influenced the results; however, subgroup analysis would have been impossible given the small sample size of each study. In addition, results in preterm infants might differ due to their immature antioxidant defense system and increased likelihood to need resuscitation (32)(33)(34). Furthermore, not all studies were randomized and only the study by Solevåg et al. (14) adhered to the ARRIVE guidelines (35), which would have been a strength and could reduce potential bias.

FUTURE DIRECTIONS
Our results indicate that air during CC is safe and human trials are urgently needed. Human studies should compare air vs. 100% and examine effects of both oxygen concentrations to reduce exposure to hypoxia and hyperoxia (36). Alternatively, attempting to mimic the gradual rise in oxygen saturation of healthy term babies in the first 10 min after birth by titrating the concentration to the baby's saturation or using any intermediate options should be assessed in regards of benefits and harms (1,2). Furthermore, any human trial should include long-term neurodevelopmental follow-up (29,30).

CONCLUSION
No human studies were identified and the results obtained are from animal models. The data suggest that using air instead of 100% oxygen during neonatal chest compression had comparable outcomes including time to return of spontaneous circulating and mortality. Hyperoxia and oxidative stress were significantly higher with 100% O 2 . Human trials comparing air vs. 100% and/or oxygen titration as an alternative to air or 100% oxygen during neonatal chest compression are urgently needed.

FUNDING
We would like to thank the public for donating money to our funding agencies: CG-H is a recipient of the Office of the Provost and VP (Academic) Summer Student Award, Faculty of Medicine and Dentistry, University of Alberta. GS is a recipient of the Heart and Stroke Foundation/University of Alberta Professorship of Neonatal Resuscitation, a National New Investigator of the Heart and Stroke Foundation Canada and an Alberta New Investigator of the Heart and Stroke Foundation Alberta. This research has been facilitated by the Women and Children's Health Research Institute through the generous support of the Stollery Children's Hospital Foundation.