Severe respiratory failure is a leading cause of pediatric ICU admissions for pneumonia. Early identification of children at risk could improve survival and optimize resource allocation

Mei Huang¹, Li Wei², Xia Liu³

Abstract

Background:
Severe respiratory failure is a leading cause of pediatric ICU admissions for pneumonia. Early identification of children at risk could improve survival and optimize resource allocation.

Objective:
To evaluate the accuracy of an early warning algorithm combining clinical signs and laboratory markers for predicting severe respiratory failure in hospitalized children with pneumonia.

Methods:
A prospective multicenter cohort study was conducted in five tertiary pediatric hospitals in China from January 2022 to December 2023. A total of 1,024 children aged 1 month to 12 years with radiologically confirmed pneumonia were enrolled. Data on respiratory rate, oxygen saturation, work of breathing, CRP, and lactate were collected at admission. Severe respiratory failure was defined as the need for invasive mechanical ventilation within 48 hours of admission. Model performance was evaluated using ROC curves.

Results:
Severe respiratory failure occurred in 142 children (13.9%). The algorithm demonstrated an AUC of 0.92 (95% CI: 0.90–0.94), with sensitivity of 88.0% and specificity of 85.2%. Independent predictors included SpO₂ < 90% (OR = 4.6, p < 0.001), respiratory rate above age-specific thresholds (OR = 3.8, p < 0.001), and lactate > 2.0 mmol/L (OR = 2.9, p < 0.01). Early identification reduced median time to escalation of care by 3.5 hours (p < 0.001).

Conclusion:
An integrated early warning algorithm combining clinical and laboratory parameters can accurately predict severe respiratory failure in pediatric pneumonia. Incorporation into hospital triage systems could reduce ICU delays and improve outcomes.