MANUSCRIPT CITATION

Sotiropoulos JX, Oei JL, Schmölzer GM, Libesman S, Hunter KE, Williams JG, Webster AC, Vento M, Kapadia V, Rabi Y, Dekker J, Vermeulen MJ, Sundaram V, Kumar P, Kaban RK, Rohsiswatmo R, Saugstad OD, Seidler AL. Initial Oxygen Concentration for the Resuscitation of Infants Born at Less Than 32 Weeks’ Gestation: A Systematic Review and Individual Participant Data Network Meta-Analysis. JAMA Pediatr. 2024 Aug 1;178(8):774-783. doi: 10.1001/jamapediatrics.2024.1848. PMID: 38913382; PMCID: PMC11197034.

REVIEWED BY

Dr Kiran Shrivastava
Title: Consultant, Department of Neonatology
Institution: MRR Children’s Hospital, Thane, Mumbai
Email: drkiran.neo@gmail.com

Dr Abdul Razak
Title: Neonatologist and Doctoral Student
Institution: Department of Paediatrics, Monash University, Melbourne, Australia
Monash Newborn, Monash Children’s Hospital, Melbourne, Australia
The Ritchie Centre, Hudson Institute of Medical Research, Melbourne, Australia
Department of Paediatrics, Joan Kirner Women’s and Children’s Hospital, Melbourne, Australia
Email: abdul.razak@monash.edu

TYPE OF INVESTIGATION

Systematic review and Individual Participant Data (IPD) and Network Meta-analysis (NMA)

QUESTION

In Preterm infants born at less than 32 weeks’ gestation (P) is low (≤0.3), intermediate (0.5-0.65), or high (≥0.90) initial oxygen (I) for delivery resuscitation best to prevent:
• Primary outcome: All-cause mortality before hospital discharge (latest available mortality data until discharge, or as defined by trial).
• Secondary outcomes:
o Severe (grade III or IV) intraventricular hemorrhage
o Bronchopulmonary dysplasia
o Retinopathy of prematurity (any)
o Saturation of peripheral oxygen (SpO2) in target range (80%-85%) at 5 minutes
o SpO2 at 5 minutes

METHODS

Design: Systematic review and IPD-NMA by inviting investigators to share IPD from eligible studies.

Participants: Infants (singleton and multiple births) born at less than 32 weeks’ gestation; regardless of timing of umbilical cord clamping.
Study Selection: Eligible studies were randomised clinical trials (RCTs) that were included if they compared at least 2 initial oxygen concentrations for delivery room resuscitation – low (≤0.3), intermediate (0.5-0.65), or high (≥0.90) fraction inspired oxygen (FiO2), and if they provided IPD.

o Interventions with low or intermediate FiO2 were only included if the delivered oxygen concentration was titrated according to the infants’ clinical status after birth.
o Static oxygen concentrations were excluded.
o For the high FiO2 group, interventions without titration (eg, static FiO2 1.0) were also included.

Data Sources: Medical databases {MEDLINE (Medical Literature Analysis and Retrieval System Online), Embase (Excerpta Medica Database), CENTRAL (Cochrane Central Register of Controlled Trials ), and CINAHL (Cumulated Index to Nursing and Allied Health Literature)} from 1980 onwards and clinical trial registries (ClinicalTrials.gov and WHO ICTRP (World Health Organisation International Clinical Trials Registry Platform) up to October 10, 2023.

Data Extraction and Synthesis: IPD were recoded into a common format and checked by 2 independent reviewers for completeness, trustworthiness (with a purpose-built IPD integrity tool), and Risk of Bias (Cochrane Risk of Bias tool, adapted for IPD).

The Confidence in Network Meta-Analysis tool was used to grade certainty of evidence.

Analysis: All analyses were intention to treat and prespecified in the statistical analysis plan, performed with R version 4.2.1 (R Foundation) and the multinma package.

All available IPD were synthesized in a single step, using generalized linear mixed models.

MAIN RESULTS

• 13 eligible studies were identified, with 12 providing IPD; and 11 studies (1003 infants) were included in primary analysis.
• 10 of 12 studies included FiO2 titration in all allocation arms.
• 10 studies were conducted in high-income countries; 2 were in middle-income countries.
• 8 RCTs (702 infants) compared high vs low initial FiO2, while 4 RCTs (353 infants) compared intermediate vs low initial FiO2.
• No studies compared intermediate vs high FiO2.

Primary:
– High initial FiO2 reduced the odds of death compared to low or intermediate FiO2 and the certainty of evidence was very low (Table 1). The effects of intermediate vs low FiO2 remained inconclusive.

Secondary:
• No significant differences were found between the 3 interventions for the outcomes, chronic lung disease (odds ratio (OR), 1.17; 95% Credible interval (CrI), 0.55-2.52), severe intraventricular hemorrhage (OR, 0.56; 95% CrI, 0.10-1.82) or retinopathy of prematurity (OR 1.17; 95% CrI, 0.58-2.20).
• High initial FiO2 led to higher SpO2 at 5 minutes compared to low FiO2, with more odds of reaching SpO2 ≥80% at 5 minutes (OR, 3.67; 95% CrI, 1.15-12.21)

Subgroup analyses:
• No evidence of differential treatment effects across gestational ages or sex.
• Prespecified subgroup analysis by country income classification, faster versus slower titration could not be explored.

Table 1: Summary of Primary Outcome (Mortality)

CONCLUSION:

Authors concluded that “high initial FiO2 may be associated with reduced mortality in preterm infants born at less than 32 weeks’ gestation compared to low initial FiO2 (low certainty). High initial FiO2 is possibly associated with reduced mortality compared to intermediate initial FiO2 (very low certainty) but more evidence is required.”

COMMENTARY:

The Individual Participant Data and Network Meta-analysis (IPD-NMA) examined 11 of 13 eligible studies, including data from 1003 babies, to address a critical question in neonatal care: What is the optimal initial oxygen concentration for resuscitating preterm infants born at less than 32 weeks’ gestation?

 

The authors categorized initial oxygen concentrations into three groups: low (21-30%), intermediate (50-65%), and high (90-100%). They found that high initial oxygen concentrations were associated with a reduction in death compared to low and intermediate concentrations. Furthermore, they observed no differences in secondary outcomes, except for higher oxygen saturation within 5 minutes in the high oxygen group compared to the low oxygen group.1

 

These findings stand in contrast to previously conducted meta-analyses, which showed no significant effect of initial oxygen concentration on mortality. These earlier analyses typically compared low versus high oxygen concentrations based on an arbitrary distinction classifying infants exposed to intermediate oxygen as part of the high oxygen group. This categorization likely diluted the apparent lower effect of high oxygen on mortality, ultimately leading to conclusions of no significant difference.2-3 In contrast, the IPD-NMA categorizes oxygen exposure more precisely, revealing a significant reduction in mortality with high oxygen compared to both low and intermediate exposures.

 

The striking impact of brief differences in initial oxygen exposure on such a critical outcome might prompt immediate consideration of this approach in clinical practice. However, it is essential to carefully weigh the certainty of this evidence, which remains low to very low. This lower certainty is driven by several factors, most notably statistical heterogeneity. In NMA, prediction intervals serve as a measure of heterogeneity and are based on posterior distribution of analysis. They tell us that, with 95% certainty, the effect of a new trial would fall within this range. The authors point out that the lower mortality seen with high oxygen levels becomes nonsignificant as the prediction intervals cross the null, suggesting significant heterogeneity and reducing our confidence in the result.

 

This statistical heterogeneity may, at least in part, be influenced by differences in how oxygen was titrated in the included trials (clinical heterogeneity). Most trials employed slow titration, while others used fast titration, and one trial even increased oxygen concentration from 0.21 to 1.0 rapidly. Such variations in oxygen delivery methods likely contributed to the observed heterogeneity.1 Additionally, the trials enrolled infants over a span of the last two decades and involved populations from different countries, further contributing to the variability. Beyond heterogeneity, other factors, such as the lack of blinding in many trials and the relatively low numbers of participants leading to imprecise estimates, also diminish confidence in the findings and raise questions about their overall reliability.

 

While this type of IPD-NMA provides more credible evidence compared to traditional pairwise meta-analyses, the above-mentioned concerns urge caution among neonatal care providers. Should we adopt this evidence from a meta-analysis of heterogenous trials examining 1003 infants, or should we wait for more homogenous clinical trials with larger populations, such as the TORPIDO (Targeted Oxygen saturation in Respiratory care of premature Infants at Delivery: effects on Outcome) trial (1,470 participants) and the HILO (higher versus lower oxygen) trial (1,200 participants)?4,5 This is just the beginning of our journey to determine the best approach to oxygen use in preterm resuscitation. Ongoing studies like OPTISTART (Optimization of Saturation Targets And Resuscitation) may further illuminate how to optimize oxygen therapy after birth for these vulnerable infants.6

REFERENCES:

1) Sotiropoulos JX, Oei JL, Schmölzer GM, et al. Initial Oxygen Concentration for the Resuscitation of Infants Born at Less Than 32 Weeks’ Gestation: A Systematic Review and Individual Participant Data Network Meta-Analysis. JAMA Pediatr. 2024;178(8):774-783. doi:10.1001/jamapediatrics.2024.1848

2) Welsford M, Nishiyama C, Shortt C, et al. Initial Oxygen Use for Preterm Newborn Resuscitation: A Systematic Review With Meta-analysis. Pediatrics. 2019;143(1):e20181828.

3) Saugstad OD, Aune D, Aguar M, Kapadia V, Finer N, Vento M. Systematic review and meta-analysis of optimal initial fraction of oxygen levels in the delivery room at ≤32 weeks. Acta Paediatr. 2014;103(7):744-751. doi:10.1111/apa.12656Brown JV, Moe-Byrne T, Harden M, McGuire W. Lower versus higher oxygen concentration for delivery room stabilisation of preterm neonates: systematic review. PLoS One. 2012;7(12):e52033

4) Ju-Lee Oei; Lower or higher initial oxygen concentration with Targeted Oxygen saturation in Respiratory care of premature Infants at Delivery: effects on Outcome (TORPIDO 30/60); Australian New Zealand Clinical Trials Registry; ACTRN1261800087926; Updated 30/08/2022. Accessed 2024-09-05. Available from https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=367049

5) Schmölzer GM, Asztalos EV, Beltempo M, et al. Does the use of higher versus lower oxygen concentration improve neurodevelopmental outcomes at 18-24 months in very low birthweight infants?. Trials. 2024;25(1):237. Published 2024 Apr 4. doi:10.1186/s13063-024-08080-2

6) Vishal Kapadia, MD. Optimization of Saturation Targets And Resuscitation (OptiSTART): Multicenter Randomized Controlled Trial. ClinicalTrials.gov (US). NCT05849077. Updated 2024-03-01. Accessed 2024-09-05. https://clinicaltrials.gov/study/NCT05849077#study-overview