Risk factors for mortality in children with hypoxemia in resource-constrained settings: a secondary analysis of Global Paediatric Acute Critical Illness Point Prevalence Study (PARITY)

BACKGROUND: Hypoxemia, a mortality predictor and hallmark of pediatric acute respiratory distress syndrome (PARDS), is disproportionately common in resource-constrained settings (RCS). The burden of PARDS in RCS is likely substantial considering the high prevalence of known clinical triggers (e.g., sepsis, pneumonia, trauma), but it is challenging to diagnose due to limited diagnostic resources. We aimed to: (1) describe respiratory care resource availability in RCS hospitals and test whether availability was associated with mortality; (2) determine the proportion of children who presented to RCS hospitals with hypoxemia and their associated outcomes; and (3) test whether, in children with hypoxemia, having a PARDS trigger was associated with mortality. METHODS: We developed and applied operational definitions for five tiered respiratory care resource bundles. Through a secondary analysis of Global Paediatric Acute Critical Illness Point Prevalence Study (PARITY) data, we performed descriptive statistics, hypothesis testing (i.e., chi-square and Wilcoxon rank-sum tests), and logistic regression analyses. RESULTS: Among the entire Global PARITY cohort (n = 7538), 763 (10.1%) were admitted with hypoxemia. Seventy percent (n = 531) were treated at a site with the intermediate or less respiratory care resource bundle available. Mortality was 6.8% (n = 52) and inversely associated with respiratory resource availability. The odds of mortality were higher for patients treated at sites with the intermediate bundle or less compared to those with the advanced or expert bundle available (adjusted odds ratio [OR] 18, 95% confidence interval [CI] 4.1-83). Fifty-six percent (n = 430) had a PARDS trigger, most commonly pneumonia (n = 256), bronchiolitis (n = 116), and sepsis (n = 58). There was no association between the presence of a PARDS trigger and mortality. Ninety-four percent of patients with a PARDS trigger (n = 405/430) had insufficient data available for a PARDS-related diagnosis according to the Second Pediatric Acute Lung Injury Consensus Conference (PALICC-2) guidelines. CONCLUSIONS: Children with hypoxemia treated at hospitals with respiratory care resource constraints in countries with lower socio-demographic index (SDI) had significantly higher mortality. These findings highlight the importance of ongoing work to improve resource availability, strengthen health systems, and support pediatric healthcare providers in identifying PARDS in order to help clinicians risk stratify children, focus resources, and tailor management to optimize outcomes.