Pediatric acute respiratory distress syndrome (ARDS) is a heterogeneous clinical syndrome of acute hypoxemic respiratory failure arising after a known insult, reflecting inflammatory lung injury and impaired gas exchange. Its pathophysiological features include diffuse alveolar damage driven by dysregulated inflammation, endothelial–epithelial injury (1, 2).

The Pediatric Acute Lung Injury Consensus Conference (PALICC) definition was established in 2015 and later updated by PALICC-2 in 2023, which refined pediatric-specific diagnostic criteria (2, 3). According to the PALICC-2 guideline, pediatric ARDS (PARDS) is defined in children under 18 years who develop acute hypoxemia and new chest-imaging opacities within 7 days of a known clinical insult, not fully explained by cardiac failure or fluid overload, and who meet oxygenation thresholds based on respiratory support. Severity is stratified according to oxygenation thresholds using the oxygenation index/oxygen saturation index and PaO₂/FiO₂ ratios across both invasive and noninvasive modes of support (2).

PARDS arises from a spectrum of direct pulmonary (e.g., pneumonia, aspiration) and indirect extrapulmonary (e.g., sepsis, systemic inflammation) causes (4, 5). Pneumonia has been reported as the leading etiology, accounting for 35%–70% of cases, followed by sepsis, contributing approximately 13.5%–43% (6–8).

Despite advances in respiratory support, PARDS remains associated with substantial mortality and morbidity. Reported mortality ranges from approximately 20% to 50% (5, 7–9). A systematic review and meta-analysis showed a pooled mortality of approximately 33.7% (10). Mortality increases stepwise with PARDS severity and with higher oxygenation index (11) Additional mortality risk factors include extrapulmonary PARDS, higher severity of illness, multiorgan failure, fluid overload, immunodeficiency, and higher ventilator pressures (5–10, 12, 13).

The interface between pediatric sepsis and PARDS remains incompletely characterized. The prevalence of PARDS among pediatric patients with sepsis remains poorly defined, with a few studies showing wide variability ranging from 9% to 83%, largely due to inconsistent definitions used for both PARDS and sepsis (14, 15). In addition, the Phoenix pediatric sepsis definition was recently published in 2024 (16). The aim of this study was to characterize PARDS in sepsis by comparing the epidemiology, clinical characteristics, outcomes, and mortality risk factors among children with sepsis with and without PARDS, and between pulmonary and extrapulmonary PARDS phenotypes.

In this retrospective analysis of a prospective cohort of children with sepsis, the prevalence of PARDS was substantial, affecting nearly 40% of Phoenix-defined sepsis cases. PARDS was associated with a five-fold increase in mortality. EP-PARDS demonstrated a predominance of bacterial infections and a markedly higher inflammatory burden; however, these differences did not translate into a statistically significant increase in mortality. Instead, mortality in sepsis-associated PARDS was associated with greater multiorgan failure and higher cumulative fluid balance.

The epidemiology of PARDS in pediatric sepsis remains poorly described, largely because earlier studies did not apply standardized definitions for either syndrome. Estimated prevalence ranges widely from 9% to 83% owing to the heterogeneous criteria used for both conditions (8, 14, 15). A single-center cohort from Indonesia showed a comparable prevalence of 33.9%; however, the study did not specify the sepsis definition applied, and mortality among the sepsis cohort reached 57.1%, exceeding the globally reported rate of 11%–40% (17). In contrast, the present study incorporated structured diagnostic frameworks, enhancing the accuracy and interpretability of PARDS epidemiology within rigorously defined pediatric sepsis and addressing the gap created by the absence of standardized criteria in earlier studies.

In our cohort, PARDS was associated with a five-fold increase in mortality among children with sepsis, with mortality rising in a stepwise manner across PARDS severity categories. This risk is consistent with the findings from multiple cohorts. In a multicenter study across four PICUs in Thailand, children with sepsis-related PARDS had significantly higher mortality than those with nonseptic PARDS (42.3% vs. 17.9%, p = 0.016) (8). A higher mortality rate was also reported in a single-center Indonesian cohort, in which 73.7% of children with sepsis-related PARDS died (17). Additionally, a multicenter analysis of the PACCMAN network revealed a mortality rate of 48.8% in EP-PARDS, a subgroup predominantly affected by sepsis (85.4%) (5). Altogether, these studies demonstrate that sepsis-associated PARDS consistently carries a substantially elevated risk of death across diverse healthcare settings.

Mortality in PARDS is driven by an intensified systemic inflammatory response that contributes to both severe respiratory failure and multiorgan injury. PARDS is initiated by neutrophil-mediated alveolar epithelial–endothelial injury, a pulmonary injury (18). This pulmonary injury amplifies the dysregulated immune state of sepsis, worsening cytokine-mediated inflammation, endothelial activation, and generation of reactive oxygen species (4, 5). These mechanisms collectively lead to and worsen multiple organ dysfunction, as demonstrated in our cohort by higher PELOD-2 scores among children with PARDS [median 8 (IQR 5.25–11) vs. 5 (IQR 3–7)]. Similar findings were observed in the PACCMAN multicenter analysis, where extrapulmonary PARDS, predominantly caused by sepsis, was associated with a markedly greater burden of multiorgan failure (5).

In our cohort, EP-PARDS demonstrated a trend toward higher mortality compared with P-PARDS, although this difference did not reach statistical significance. These two phenotypes differed substantially in both etiology and inflammatory burden: EP-PARDS was predominantly associated with bacterial infections, whereas pulmonary PARDS was more frequently viral in origin. Children with EP-PARDS also exhibited higher inflammatory marker levels, consistent with a more intense systemic inflammatory burden and lower platelet counts suggestive of increased platelet consumption. These clinical patterns align with evidence that extrapulmonary lung injury is driven by systemic endothelial injury from circulating inflammatory mediators, whereas P-PARDS reflects primary alveolar epithelial injury, which is more commonly triggered by viral or localized pulmonary pathogens (4).

Furthermore, greater cumulative fluid balance was independently associated with increased mortality in children with PARDS. Similar patterns have been observed in a multicenter study in which children with sepsis-related PARDS had nearly double the degree of fluid overload at 72 h compared with those with nonseptic PARDS [11.5% (5.3–14.6) vs. 6.4% (3.5–9.9)], and this group also exhibited substantially higher mortality (42.3% vs. 17.9%) (8). The detrimental impact of fluid accumulation has also been demonstrated in a large single-center prospective cohort of 290 children with acute lung injury, where each 10 mL/kg/day increase in fluid balance was associated with higher mortality and fewer ventilator-free days. Together, these findings reinforce fluid overload as a key and potentially modifiable determinant of outcomes in sepsis-associated PARDS.

This study has several limitations. First, although sepsis was prospectively identified through a structured registry, PARDS classification relied on retrospective chart review, which may have introduced misclassification despite strict adherence to the PALICC criteria. Second, a key limitation is the absence of detailed ventilator parameters, including tidal volume, PEEP levels, and driving pressure, which prevented assessment of lung-protective ventilation practices. Third, as this was a single-center study conducted at a tertiary referral hospital, our findings may be influenced by local case mix and referral patterns, potentially limiting generalizability. Another important limitation relates to the use of the Phoenix pediatric sepsis criteria. The Phoenix framework was developed primarily to standardize sepsis identification and severity stratification at the population level, rather than as a definitive diagnostic or prognostic tool for individual patients. The aim of this study was not to validate the Phoenix criteria, but to apply this consensus definition to define a sepsis cohort and examine how PALICC-2–defined PARDS interacts with sepsis severity and organ dysfunction phenotypes. Accordingly, our findings should be interpreted as population and phenotype level associations and viewed as hypothesis generating, highlighting the need for future multicenter validation. Finally, residual confounding is possible despite multivariable adjustment, particularly regarding timing of diuresis and renal replacement therapy. Despite these limitations, this study has notable strengths, including the application of the contemporary Phoenix sepsis definition and PALICC-2 PARDS criteria, enabling precise phenotyping that has been lacking in earlier literature. Additionally, the sample size, particularly for a rigorously defined sepsis-associated PARDS cohort is substantially larger than those in previous pediatric studies, allowing for more robust stratified analyses and improved reliability of effect estimates. Collectively, these strengths enhance the validity of our findings and provide a clearer understanding of the epidemiology and outcomes of PARDS in children with sepsis.

In conclusion, PARDS emerged as a prevalent complication linked to substantially increased risk of adverse outcomes. It was associated with greater illness severity, increased organ support requirements, and a five-fold increase in mortality, with risk escalating across PARDS severity categories. EP-PARDS showed features of heightened systemic inflammation, and fluid overload was identified as an important and potentially modifiable contributor to poor outcomes. By applying contemporary sepsis and PALICC-2 PARDS definitions, this study provided a clearer epidemiological profile of sepsis-associated PARDS. Future multicenter studies are needed to validate these findings and to inform early identification and management strategies that may improve outcomes in this high-risk population.