Conservative versus liberal oxygenation targets in PICU
Every year in the UK, about 14,000 children needing intubation and ventilation are taken care of in children’s intensive care units, and they often receive supplemental oxygen as part of their treatment. Despite this, data regarding optimum SpO2 targets within PICU are scarce.
Recent studies, in both adults and children, highlight the potential harms of exposure to a high fraction of inspired oxygen (FiO2). In 2017, however, a retrospective analysis found that more than a quarter (26%) of saturation recordings were in the 99-100% range.
There are existing guidelines supporting lower saturation targets in ARDS (88-92%) and bronchiolitis (>90%). The Oxy-PICU study is the first RCT looking at optimum saturation targets within the PICU population.
The study
This is the first study looking at optimum saturation targets within the paediatric intensive care population and recruited a large number of participants, despite being conducted during the COVID-19 pandemic.
It was a pragmatic, multicentre, open, parallel-group, randomised clinical control trial conducted across 15 different Paediatric Intensive Care Units (PICUs) in England and Scotland and their associated transport services.
This study highlights how collaborative efforts and research registries can lead to large pragmatic trials that can potentially change practice.
Who were the Patients?
The centres recruited 2040 critically ill children, between 38 weeks corrected gestational age and 16 years old, from 15 different PICUs in England and Scotland.
The patients were all admitted to the PICU as an emergency and received invasive mechanical ventilation with supplemental oxygen at the time of randomisation. Patients with a brain injury, known/suspected uncorrected congenital cardiac disease and pulmonary hypertension were excluded.
What was the Intervention?
Patients were randomly allocated to either conservative (SpO2 88-92%) or liberal (SpO2 >94%) oxygenation targets in a 1:1 ratio.
The conservative target was based on recent pARDS guidelines.
To minimize between-group differences, the randomisation sequence was based on age (<12 months or ≥2 months), site, primary reason for admission (respiratory or other) and severity of abnormality of gas exchange (defined as an S/F ratio <221 with PEEP ≥5 or other).
The protocol specified hourly FiO2 and mean airway pressure measurements with adjustments towards the target saturations.
What were the Outcomes?
Primary outcome
The primary outcome was the duration of organ support at 30 days or death.
The need for organ support was determined from the Paediatric Critical Care Minimum Dataset, as collected for the PICANet registry. Participants were given a score between 1 and 30 based on the number of calendar days of organ support received. Death on or before day 30 was assigned a score of 31.
The team computed a probabilistic index with a 95% confidence interval. A value above 0.5 on this index suggests that there is a greater than 50% likelihood that the conservative treatment approach is more effective compared to the liberal treatment strategy.
Secondary outcomes
There were several secondary outcome measures:-
– Mortality at PICU discharge and 30 daysÂ
– Time to liberation from invasive mechanical ventilationÂ
– Duration of organ support
– Functional status at PICU discharge (measured by POPC and PCPC scales)Â
– Length of PICU stay and acute hospital stay.Â
– Incremental cost at 30 daysÂ
· Serious adverse events (e.g., severe lactic acidosis, cardiac ischaemia, AKI, seizure, event related to oxygenation target)
What were the results?
Oxygenation:
In the conservative group, the median of the mean SpO2 values was 94% (IQR 93–96) versus 97% (96–98) in the liberal oxygenation group.
The median of the mean FiO2 values was 0·27 (0·24–0·33) for the conservative group versus 0·35 (0·30–0·42) for the liberal group.
Patients in the conservative group spent a greater proportion of their total ICU time on invasive mechanical ventilation with a SpO2 of 88–92%.
Primary outcome
Duration of organ support or death at 30 days was lower in the conservative group (median 5 days (IQR 3-9)) compared to the liberal group (median 5 days (IQR 3-10).
The probabilistic index was 0.53 (95% CI 0.50 to 0.55, p value=0.04), suggesting a statistically significant benefit for the conservative group compared to the liberal oxygenation group.
The unadjusted odds ratio was 0.85 (95% CI 0.73 – 0.99). The adjusted odds ratio was 0.84 (95% CI 0.72-0.99). In the conservative group, the likelihood of needing prolonged organ support or the risk of death was 16% lower compared to the liberal group.
Secondary outcomes
The conservative oxygenation group had a statistically significant reduction in healthcare costs of £2143/$3166 compared to the liberal oxygenation group.
The time from randomisation to extubation was 3.5 hours sooner in the conservative oxygenation group – 69.6 hours vs 73.2 (hazard ratio 1.11, p-value 0.03)
There was no significant difference in mortality between groups.
There was no statistically significant difference in the duration of organ support between groups.
There was no significant difference in the length of PICU or hospital stay between groups.
What were the limitations?
As this was a pragmatic trial, many treatment decisions were left to the clinical team, including ventilation parameters, use of recruitment manoeuvre, sedation, and readiness for extubation. This increases the risk of bias.
Recruitment took place during the COVID-19 pandemic. Participants may not have reflected the usual PICU patient population. If Oxy-PICU was repeated in today’s PICU, the effect size may be more apparent.
72% of eligible patients were randomised. 5570 met inclusion criteria, with 2753 meeting one or more exclusion criteria. That left an eligible population of 2817, of whom 2040 were randomised.
The organ support criteria were predefined by the registry used to collect data for PICAnet. There was no sub-group analysis based on the type of support given. This definition may differ from what individuals working in emergency medicine may think of as organ support. Patients were given a score of 1 if they received any organ support even after discharge from PICU. For example, a blood transfusion on the ward counted as one day of ‘support’.
There was also no subgroup analysis of patients with saturations within the conservative group’s target range.
More patients had protocol deviations in the conservative oxygenation group (41.4% vs 16.6%). Most involved going above the designated oxygen target range. The authors provided several possible explanations, including physiotherapy, pneumothorax, acute deterioration, and lack of trial awareness. However, a process evaluation was not conducted to investigate the obstacles that prevented patients from achieving the conservative oxygen saturation goal.
Supplemental data provided by authors showed that there is still an overall difference in cumulative dose of oxygen by intervention group.
CASP checklist – How good was the paper?
Does the study address a clearly focused issue?
Yes.
Was the assignment of participants to interventions randomised?
Yes.
Patients were randomly allocated to either conservative (SpO2 88-92%) or liberal (SpO2 >94%) oxygenation targets. in a 1:1 ratio.
Were all participants who entered the study accounted for at its conclusion?
Yes.
Was it a blinded trial?
No.
This was an unblinded trial.
Were the study groups similar at the start?
Yes.
Was each study group they treated equally)?
Many decisions were left to the clinical team, such as ventilator parameters, recruitment manoeuvre, sedation, and readiness for extubation, which could increase bias.Â
Were the effects of intervention reported comprehensively?
Yes.
Was the precision of the estimate of the intervention or treatment effect reported?
Yes.
They reported the confidence Intervals.
Do the benefits of the intervention outweigh the harms and costs?
Yes.
There was a statistically significant reduction in healthcare costs up to 30 days of £2143/$3166 in the conservative oxygenation group.
Was the exposure accurately measured to minimise bias?
Patients were randomised in a 1:1 ratio to either conservative or liberal oxygenation targets. This was an open study due to the practicalities involved with the intervention.
Clinical decisions around adjustment of ventilator settings, weaning and timing of extubation were left to the clinical teams. This could introduce ascertainment and selection bias.
Was the outcome accurately measured to minimise bias?
The primary outcome was a composite of death and days of organ support at 30 days. This clear, objective measure can be accurately recorded, minimising the risk of subjective bias.Â
Objective measures like SpO2 levels for oxygen saturation provide a standardised way to assess outcomes, further minimising bias.Â
However, the trial’s open nature made blinding impossible, which could introduce observer bias.Â
Have the authors identified all important confounding factors?
The randomisation process, which included factors like age, size, the primary reason for admission, and degree of abnormality of gas exchange, was designed to distribute potential confounding factors evenly between the groups.Â
However, there may be unidentified or unaccounted confounding factors in any clinical trial, especially in a diverse population.Â
Was the follow-up of the subjects complete and accurate?Â
The investigators aimed to follow up with subjects for 30 days post-randomization. This is a reasonable duration to assess the immediate outcomes of oxygen therapy in critically ill children.
Longer-term outcomes have not yet been published. They may impact oxygenation strategies going forward.
Less than 10% were withdrawn following randomisation. Numbers were similar between groups.
What were the results?
There was a statistically significant benefit in the duration of organ support or death up to 30 days in the conservative group when compared with the liberal oxygenation group. This was consistent across all measured components of the primary outcome.Â
The conservative oxygenation group had a significantly lower healthcare cost up to 30 days (£2143) than the liberal oxygenation group.Â
Do you believe the results?
Yes.
Can the results be applied to our local population?
Yes.
This was the largest randomised controlled trial in critically ill ventilated children ever conducted. Because most UK PICUs contributed patients, the results are applicable to the UK population.Â
The study was carried out during the height of the COVID-19 pandemic, which led to an atypical case mix, including fewer patients with respiratory infections. Perhaps if Oxy-PICU were to be repeated in today’s PICU, the effect size would be larger.
Two large patient cohorts were excluded – those with uncorrected congenital cardiac disease and those admitted with acute encephalopathy.
Do the results fit with the other evidence available?
Before the Oxy-PICU study, it was routine practice to maintain relatively high oxygen saturation levels, often targeting a SpO2 of 95-100% in critically ill children.Â
Previous research in both adult and PICU suggests that very high oxygen levels might contribute to oxidative stress and potential harm. There is ongoing debate around the optimal balance between avoiding hypoxia and preventing the potential harm of too much oxygen.Â
Studies in neonatal intensive care have shown that despite a significant reduction in severe retinopathy, more extreme preterm infants died with a conservative oxygenation target of 85 – 89%.
In adult critical care, there have been mixed results regarding the benefits of conservative versus liberal oxygen therapy. Some studies suggest no significant difference in outcomes between different oxygen targets, while others point to the potential benefits of more conservative oxygenation strategies.Â
The Oxy-PICU study is the only randomised controlled trial focusing on oxygenation targets in PICU. It suggests a potential benefit of lower oxygen targets (88-92% SpO2) over the more traditional targets (above 94% SpO2).
What did the authors conclude, and what can we take away from this study?
The Oxy-PICU study is the largest randomised controlled trial in the PICU ever. The authors conclude that among children and young people admitted to the PICU as an emergency who receive invasive mechanical ventilation, a conservative oxygenation target of SpO2 88–92% resulted in a significantly greater probability of a better outcome in terms of the duration of organ support or death at 30 days.
Patients in the conservative target range (SpO2 88-92%) were 6% more likely to have a better outcome.
However, as the authors also nicely stated, whether an intermediate SpO2 target (e.g., 92-94%) or strict adherence to the 88-92% range is yet to be determined.Â
Future research in this field might consider why there is a higher probability of an improved outcome in the conservative oxygenation group, exploring potential pathophysiologic mechanisms and identifying organ support factors that led to improved outcomes.
About PICSTAR
PICSTAR is a trainee-led research network open to all doctors, nurses and allied health trainees within Paediatric Intensive Care. We are the trainee arm of the Paediatric Critical Care Society – Study Group (PCCS-SG) and work with them on research, audit and service evaluation.
If you would like to join PICSTAR and get involved in projects, have ideas you would like to propose or get advice/mentorship via PCCS-SG, don’t hesitate to contact us at picstar.network@gmail.com. See their website for more: https://pccsociety.uk/research/picstar/
References
1. Ray, S., et al., Liberal oxygenation in paediatric intensive care: retrospective analysis of high-resolution SpO2 data. Intensive Care Medicine, 2017. 43(1): p. 146-147.