Partial pressure of oxygen on admission linked to mortality in pediatric ICU patients

A new study showed a link between partial pressure of oxygen at admission and mortality in pediatric ICU patients and other critically ill subgroups, researchers reported at the Society of Critical Care Medicine Congress.

“We know that hyperoxia can lead to increased formation of reactive oxygen species, which can cause cell damage, cell death and inflammation,” Caroline Holton, MD, third-year pediatric ICU fellow at Children’s Mercy Hospital in Kansas City, Missouri, said during the presentation. “Multiple adult studies have suggested an association between hyperoxia and mortality. Pediatric studies are slightly more limited due to smaller sample sizes and just looking at general pediatric ICU populations.”

The retrospective, multicenter, observational cohort study included 13,071 patient encounters from 2015 to 2019 of noncardiac pediatric patients with partial pressure of oxygen (PaO₂) recorded during the first hour of ICU admission. Researchers utilized data from the Virtual Pediatric Systems database of more than 135 participating hospitals and Pediatric Index of Mortality 3 scores to estimate illness severity at admission. Researchers also identified six specific diagnosis subgroups of patients with trauma, head trauma, sepsis, renal failure, hemorrhagic shock and post-cardiac arrest.

The primary outcome was ICU mortality.

Overall mortality was 13.5%, which Holton said represented a sicker patient cohort than typical ICU populations. Hyperoxia in this cohort was rare on admission, Holton said, with a PaO₂ greater than 200 mm Hg in 25.5% of patients, a PaO₂ greater than 300 mm Hg in 7.8% of patients and a PaO₂ greater than 500 mm Hg in 1.3% of patients.

Results showed a U-shaped mortality curve, “with the highest mortality being seen in patients who have a low PaO₂ and those who have a very high PaO₂,” Holton said. When the researchers looked at standardized mortality ratios (SMR), the highest SMR was seen in patients who were hypoxic and hyperoxic at admission, with a more subtle U-shape, Holton said.

In the other critically ill patient subgroups, overall mortality was 25.8% among the 2,702 patients with trauma, 28.86% among 1,850 patients with head trauma, 22% among 1,218 patients with sepsis, 36.12% among 969 patients with renal failure and 39.44% among 322 patients with hemorrhagic shock. “Again, there [was a] very clear U-shape to the data,” Holton said, with higher mortality in the patients with low and high PaO₂ at admission.

Mortality was 63.07% among 1,500 post-arrest patients. For these patients, there did not appear to be a difference in mortality based on PaO₂ at admission, Holton said. This result persisted when the researchers looked at SMR in post-arrest patients.

“To me, this suggests that the pathophysiology of certain disease states may modify the association with hyperoxia,” Holton said.

The researchers noted limitations of the current study, including its observational design and lack of admission PaO₂ for all PICU patients.