You are called to the paediatric ward to review a 17-month-old child who was admitted 2 days ago with viral-induced wheeze. An hour ago, they were running around the playroom with a snack.
They are now napping, and oxygen saturation on room air has dropped to 88-89%.
Can they go home?
Challenging the status quo
For children admitted to hospital with acute respiratory distress, the SpO₂ thresholds used to start and stop oxygen therapy typically sit somewhere between 90% and 94% but there is limited evidence behind these numbers. Some studies support different targets in specific settings, such as 88–92% in critically unwell, mechanically ventilated children (the Oxy-PICU trial).
Respiratory presentations such as bronchiolitis, lower respiratory tract infection (LRTI) and viral wheeze are among the most common reasons children are admitted to hospital, so it is worth challenging the status quo and asking: what is a safe SpO₂ threshold?
And if that threshold can be safely lowered, could children go home sooner? Shorter admissions would save money, free up beds and staff, reduce treatment burden for families, and ease the winter pressures felt on general paediatric wards every year.
Louman, S., van Stralen, K.J., Koppelman, G.H., Vaessen-Verberne, A.A., Bekhof, J., Bosmans, J.E., Brackel, C.L., Kamps, A.W., de Kleer, I.M., Mulder, C.F. and Balemans, W., 2026. Oxygen saturation thresholds in children with acute respiratory distress (OxyKids): a multicentre, open, parallel-group, randomised clinical trial. The Lancet Respiratory Medicine.
The study
OxyKids is the first randomised controlled trial (RCT) to compare an oxygen saturation threshold of 88% with the conventional 92% in a general paediatric population presenting with acute respiratory distress. The aim was to determine whether lowering the SpO₂ threshold for intervention to 88% could safely reduce the time taken to meet discharge criteria.
This pragmatic, multicentre, open, parallel-group RCT took place across ten general and teaching hospitals in the Netherlands, none of which had a PICU.
Who were the patients?
A total of 566 patients were randomised between September 2023 and December 2024.
Inclusion criteria:
- Previously healthy children aged 6 weeks to 12 years
- Hospitalised on a general paediatric ward with respiratory distress due to bronchiolitis, LRTI or acute viral-induced wheeze
- Requiring oxygen according to standard care at the 92% threshold
Exclusion criteria:
- Children aged 6-12 years with acute asthma (commonly recommended SpO₂ targets in asthma care are ≥94%)
- Known pre-existing cardiopulmonary, immunological, neurological or haematological conditions
- Born at <32 weeks’ gestation
Intervention
Children were randomly allocated (1:1) to an SpO₂ threshold of either 88% (intervention) or 92% (control). Allocation was not masked, and oxygen therapy was mandated if the SpO₂ fell below the allocated threshold. To reflect real-world clinical practice, the data were analysed intention-to-treat, so all randomised patients were included regardless of protocol adherence.
Medical staff could still start oxygen above the allocated SpO₂ level based on symptoms of respiratory distress, and once oxygen was started there was no upper SpO₂ limit. All other treatments followed local protocols.
Primary outcome
Time from admission to the general paediatric ward to meeting all prespecified discharge criteria.
Secondary outcomes
There were multiple secondary outcomes, including:
- Effectiveness: length of stay and total duration of oxygen therapy
- Safety: observations at cessation of oxygen, time to resolution of illness, unscheduled healthcare visits or readmissions within 28 days of discharge, and PICU transfers
- Parental anxiety: assessed via electronic questionnaire at discharge and at 7 and 28 days after discharge
- Serious adverse events: recorded up to 28 days after discharge
What were the results?
Primary outcome
The main finding was a clinically meaningful reduction in the time taken to be fit for discharge in the 88% (intervention) group compared with the 92% (control) group.
Patients in the 88% group met the pre-set discharge criteria a median of 16.8 hours earlier than the 92% group (27.6 h (IQR 15.1-52.7) vs 46.6 h (IQR 24.2-85.1)), exceeding the 12-hour difference deemed clinically meaningful. (The 16.8-hour figure is the estimated median difference across the whole distribution, which is why it doesn’t match a simple subtraction of the two medians.)
The adjusted geometric mean ratio (GMR) of 0.64 suggests that the typical time to fitness for discharge in the 88% group was around two-thirds of that in the 92% group. Sensitivity analyses adjusting for other factors gave consistent results: the 88% group were ready for discharge earlier.
Secondary outcomes
The shorter time to meet discharge criteria in the 88% group translated into a shorter length of stay, by an estimated 17.6 hours (39.8 hours in the 88% group vs 60.8 hours in the 92% group, again an estimated difference rather than a simple subtraction of the two figures).
Fewer patients in the 88% group received supplemental oxygen at all, and those who did received it for less time.
There were no significant differences in emergency department visits or readmissions. For GP visits, the primary analysis showed no statistically significant difference within 28 days, although complete case analysis revealed numerically more GP visits in the 88% group.
Recovery was similar in both groups: symptom resolution, return to day care or school, return to normal health and parental anxiety did not differ.
On safety, 7% of patients in each group had one or more serious adverse events (readmission or PICU admission), and there were no deaths during the trial.
Limitations
Cost analysis: The paper quantifies the reduction in hospital admission time but does not calculate the real-world cost savings that would flow from it.
Burden on primary care: As the authors highlight, there were more GP visits in the 88% group, even though this was not statistically significant. Whether adopting an 88% threshold would genuinely shift work into primary care is unclear, but it suggests the need for closer monitoring, including of cost implications, if the threshold were introduced.
Neurocognitive sequelae: A significant concern for paediatricians when considering lower SpO₂ thresholds is the potential for adverse long-term neurocognitive effects. This study did not assess neurocognitive outcomes, but as the authors point out, there was no other indication of hypoxic brain injury during the trial, so the likelihood of harm seems low. Further research is needed to build an evidence base here.
Generalisability: This study included only previously healthy children, so the results may not be generalisable to those with pre-existing medical conditions. Importantly, children over 6 with asthma were excluded, so these parameters should not be applied to them. This highlights the need for further trials in these groups.
Oximetry bias: Of the 566 patients randomised, 480 were ethnically Dutch, so the results may not be generalisable to children with darker skin tones, particularly as pulse oximeters are known to overestimate SpO₂ in patients with darker skin.
Rare events: Finally, is this study big enough? Death in childhood is rare, so a cohort of this size cannot tell us whether changing SpO₂ thresholds would affect mortality.
The bottom line
For previously healthy children admitted with respiratory distress, an SpO₂ threshold of 88% reduced the time taken to be fit for discharge compared with 92%, with no evidence of harm.
Are we ready for it?
This trial has the potential to shift paediatric clinical practice, enabling clinicians to safely treat and discharge children with respiratory distress using an SpO₂ threshold of 88%. A lower threshold could mean earlier discharge, better use of resources and less unnecessary treatment for children.
The design of the Oxykids trial should make it easily applicable to clinical practice as the
study reflected real world scenarios by allowing for variations in local protocols and for
clinician discretion regarding oxygen delivery. It also highlights the importance of challenging
normal clinical practice and looking
But are we ready to disrupt the status quo? Can we resist the urge to treat these patients based on numbers alone? Are we willing to shift clinical practice without definitive evidence on the neurocognitive implications?
It is understandable that paediatricians may feel reluctant, especially as this represents a change to a long-standing “normal”. As with every big change, it will take collaboration and further data collection to ensure the shift can be implemented safely, and to monitor for rare complications.
So what about the 17-month-old on the ward?
Based on the OxyKids trial, intervening by starting oxygen to improve the monitor saturations could keep them in hospital for an extra 17 hours. Provided other clinical discharge criteria are met and you safety-net the parents, they can be safely discharged home.
References
Louman, S., van Stralen, K.J., Koppelman, G.H., Vaessen-Verberne, A.A.P.H., Bekhof, J., Bosmans, J.E., Brackel, C., Kamps, A.W.A., de Kleer, I., Balemans, W., Scheffer, M., Brouwer, M., van den Beukel, M., Andrinopoulou, E.R., Pijnenburg, M.W.H. and Boehmer, A.L.M., 2026. Oxygen saturation thresholds in children with acute respiratory distress (OxyKids): a multicentre, open, parallel-group, randomised clinical trial. The Lancet Respiratory Medicine.
Peters, M.J., Gould, D.W., Ray, S., Thomas, K., Chang, I., Orzol, M., O’Neill, L., Agbeko, R., Au, C., Draper, E. and Elliot-Major, L., 2024. Conservative versus liberal oxygenation targets in critically ill children (Oxy-PICU): a UK multicentre, open, parallel-group, randomised clinical trial. The Lancet, 403(10424), pp.355-364.
AI assistance disclosure: Gemini was used to tighten prose for this post. All clinical content,
references, and conclusions were written and verified by the author.
