Admission SpO₂/FiO₂ ratio for mortality risk stratification in children with severe pneumonia: a multicenter retrospective cohort study

Abstract

Background

Severe pneumonia remains a major cause of pediatric intensive care unit (PICU) admission and death, particularly in settings where access to vaccination, timely care, and reliable oxygen systems is uneven. Conventional prognostic assessment combines clinical severity, organ dysfunction, and laboratory biomarkers, but many components are intermittent or invasive. This study aimed to evaluate whether the admission peripheral oxygen saturation to fraction of inspired oxygen (SpO₂/FiO₂; S/F) ratio provides additional, non-invasive prognostic information in children with severe pneumonia requiring respiratory support.

Methods

This multicenter retrospective cohort study included 218 children with severe pneumonia admitted to the PICUs of three teaching hospitals in China from December 2021 to November 2024. Admission clinical characteristics, respiratory support, oxygenation indices, and laboratory variables were extracted from electronic medical records. The primary outcome was in-hospital mortality. Associations with mortality were evaluated using group comparisons, correlation analysis, exploratory logistic regression, and receiver operating characteristic (ROC) analysis.

Results

Sixty-one children died, giving an in-hospital mortality rate of 28.0%. Non-survivors had lower admission S/F and SpO₂/FiO₂ (P/F) ratios and more frequently received invasive mechanical ventilation. Higher lactate, alanine aminotransferase (ALT), and blood urea nitrogen (BUN) levels were associated with increased mortality risk. In exploratory logistic regression, a higher S/F ratio was associated with lower mortality risk [odds ratio (OR) 0.988 per one-unit increase, 95% confidence interval (CI): 0.981–0.995; P=0.001]. The S/F ratio showed modest single-marker discrimination [area under the curve (AUC), 0.637].

Conclusions

The admission S/F ratio was inversely associated with in-hospital mortality in children with severe pneumonia requiring respiratory support. It may be used at PICU admission and after changes in oxygen therapy as a rapid adjunct to clinical assessment, severity scores, lactate, and organ dysfunction markers. Because discrimination was modest and respiratory-support modalities were heterogeneous, the S/F ratio should not be used alone or as a fixed treatment threshold without prospective validation.

Highlight box.

Key findings

• A lower admission SpO₂/FiO₂ (S/F) ratio was associated with in-hospital mortality in children with severe pneumonia requiring respiratory support; its single-marker discrimination was modest (area under the curve 0.637).

What is known and what is new?

• The S/F ratio is a non-invasive surrogate of oxygenation impairment, but pediatric pneumonia-specific prognostic evidence across non-invasive and invasive support is limited.

• This multicenter cohort links admission S/F with mortality while showing that lactate and organ dysfunction markers provide complementary information.

What is the implication, and what should change now?

• Clinicians may use the S/F ratio at pediatric intensive care unit admission and after oxygen-support changes to trigger integrated reassessment, not as an isolated mortality test or an unvalidated treatment cutoff.

Introduction

Pneumonia and other lower respiratory infections remain major causes of pediatric morbidity and mortality. The Global Burden of Disease 2023 analysis estimated that mortality from lower respiratory infections among children younger than 5 years decreased by 33.4% from 2010 to 2023, yet the aggregate mortality rate in sub-Saharan Africa remained furthest from the global target, and marked socioeconomic gradients persisted (1). The disproportionate burden in low- and middle-income countries (LMICs) reflects interacting child-level and environmental determinants, including undernutrition, prematurity, incomplete vaccination, chronic disease, household crowding, indoor and outdoor air pollution, delayed care-seeking, limited access to diagnostics and antibiotics, and unreliable oxygen delivery (2,3). The Child Health and Mortality Prevention Surveillance Network further showed that pneumonia contributed substantially to deaths among young children in sub-Saharan Africa and South Asia (4). Vaccination programs have reduced pneumonia-related hospitalization and mortality, but progress remains uneven across regions and health systems (5-7).

Early risk stratification in severe pediatric pneumonia currently relies on respiratory distress, hypoxemia, altered mental status, hemodynamic instability, and evolving organ dysfunction, together with scores such as the Pediatric Risk of Mortality, Pediatric Index of Mortality, pediatric Sequential Organ Failure Assessment, Respiratory Index of Severity in Children, and modified Predisposition, Insult, Response, and Organ dysfunction scores. Common laboratory markers include lactate, procalcitonin (PCT), C-reactive protein (CRP), blood urea nitrogen (BUN), creatinine, transaminases, coagulation indices, and inflammatory or hematological ratios (8-11). These measures are clinically useful but may require blood sampling, multiple variables, or delayed laboratory processing. A continuously available oxygenation index could therefore add value by identifying deterioration between formal laboratory or scoring assessments.

The PaO₂/FiO₂ (P/F) ratio is widely used to assess oxygenation impairment in severe respiratory disease and acute respiratory distress syndrome (ARDS), but it requires arterial blood gas sampling and is less convenient for repeated pediatric assessment (12,13). The SpO₂/FiO₂ (S/F) ratio uses pulse oximetry and the administered fraction of inspired oxygen, providing a non-invasive and repeatable estimate of oxygenation impairment (14-18). The Second Pediatric Acute Lung Injury Consensus Conference recognized SpO₂-based oxygenation metrics for pediatric ARDS severity assessment, and pediatric studies have demonstrated clinically meaningful relationships between S/F- and PaO₂-based indices (19,20). Its value is not that it replaces clinical scores or organ dysfunction biomarkers, but that it provides an immediate respiratory signal that can be trended at the bedside and interpreted alongside them.

Evidence regarding the prognostic value of the admission S/F ratio specifically in children with severe pneumonia receiving a spectrum of non-invasive and invasive respiratory support remains limited. We therefore evaluated the association between the admission S/F ratio and in-hospital mortality and explored its role as a practical adjunct to P/F-based and multisystem risk stratification. We present this article in accordance with the STROBE reporting checklist (21) (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0577/rc).

Methods

Study design and participants

This multicenter retrospective cohort study was conducted in the pediatric intensive care units (PICUs) of three tertiary teaching hospitals in China: Children’s Hospital, Zhejiang University School of Medicine, the Children’s Hospital of Fudan University, and Guangzhou Women and Children’s Medical Center. Consecutive eligible children admitted from December 2021 to November 2024 were screened. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Ethics Committee of Children’s Hospital, Zhejiang University School of Medicine (No. 2021-IRB-302). All participating hospitals were informed of and agreed to the study. Individual consent for this retrospective analysis was waived.

During the study period, the participating centers had the following service profiles: the Children’s Hospital of Zhejiang University School of Medicine had approximately 2,900 licensed beds, 24 PICU beds, and managed approximately 5,000–8,000 pediatric pneumonia cases annually; the Children’s Hospital of Fudan University had approximately 1,000 licensed beds, 40 PICU beds, and managed approximately 4,000–5,000 pediatric pneumonia cases annually; and Guangzhou Women and Children’s Medical Center had approximately 1,400 licensed beds, 33 PICU beds, and managed approximately 4,000–7,000 pediatric pneumonia cases annually.

Children were eligible if they were older than 1 month and younger than 18 years, were admitted to a participating PICU with severe pneumonia, and received respiratory support [high-flow nasal cannula (HFNC), non-invasive ventilation (NIV), invasive mechanical ventilation, or extracorporeal membrane oxygenation (ECMO)]. Pneumonia required compatible acute respiratory symptoms and new pulmonary infiltrates on chest radiography or computed tomography. Severe pneumonia was operationalized by at least one documented objective feature: hypoxemia (room-air SpO₂ <90% or need for supplemental oxygen), severe respiratory distress, respiratory failure requiring HFNC/NIV/invasive ventilation/ECMO, shock or vasoactive support, altered consciousness, ARDS, sepsis, empyema, or pneumothorax. This definition was based on Chinese pediatric community-acquired pneumonia guidance and World Health Organization (WHO) severe-pneumonia criteria and distinguished the cohort from children with mild or moderate pneumonia managed without intensive respiratory support (22-24). Patients with incomplete key records or inability to complete relevant examinations or treatment were excluded. The study size was determined by the number of consecutive eligible patients with complete data during the study period.

Data collection

The primary outcome was in-hospital mortality, defined as death before hospital discharge; survivors were patients discharged alive. The principal predictors were the S/F and P/F ratios recorded at PICU admission. Demographic characteristics, comorbidities, respiratory support type, and admission laboratory parameters were extracted from electronic medical records, including white blood cells (WBCs), BUN, CRP, PCT, aspartate aminotransferase (AST), alanine aminotransferase (ALT), hemoglobin (Hb), pH, sodium, potassium, blood glucose, activated partial thromboplastin time (APTT), lactate, D-dimer, and serum creatinine.

To reduce information bias, oxygenation and laboratory variables were based on the earliest values documented at PICU admission using a standardized extraction framework across centers. Missing key variables were handled by complete-case analysis, and no statistical imputation was performed.

Microbiological investigations—including blood culture, sputum or tracheal-aspirate culture, respiratory pathogen polymerase chain reaction, and other pathogen-specific tests—were performed according to clinical indications. Because sampling strategies and testing platforms were not standardized across centers and complete pathogen results were not available in the harmonized dataset, microbiological etiology was not included as a covariate in the present analysis.

Statistical analysis

Continuous variables were summarized as mean ± standard deviation or median with interquartile range, and categorical variables as n (%). Between-group comparisons used independent-samples t-tests or Mann-Whitney U tests for continuous variables and Chi-squared or Fisher exact tests for categorical variables, as appropriate. Point-biserial Pearson correlation, exploratory univariable logistic regression, and receiver operating characteristic (ROC) curve analysis were used to evaluate associations with mortality and single-marker discrimination. Given the modest area under the curve (AUC), heterogeneous respiratory-support modalities, and absence of external validation, a data-derived Youden cutoff was not promoted as a clinical treatment threshold. Two-sided P values <0.05 were considered statistically significant.

Results

Patient characteristics

A total of 218 children with severe pneumonia who received respiratory support in the PICU met the eligibility criteria and had complete key data for analysis. Among them, 157 survived and 61 died, corresponding to an in-hospital mortality rate of 28.0%. The median age was 6.0 years, and 135 patients (61.9%) were male. Ninety-five patients (43.6%) had no known comorbidity. The most frequent underlying conditions were cerebral palsy or neuromuscular disease [45/218 (20.6%)], hematological disease, immune deficiency, or malignancy [33/218 (15.1%)], bronchopulmonary dysplasia [26/218 (11.9%)], cardiovascular disease [8/218 (3.7%)], endocrine or metabolic disorders [3/218 (1.4%)], and other comorbidities [8/218 (3.7%)].

Administrative records from the three participating centers identified 13,478 pediatric pneumonia admissions during the study period. The 218 severe pneumonia cases admitted to PICUs and included in the analytic cohort therefore represented 1.6% of all pediatric pneumonia admissions.

Clinical presentations

Clinical characteristics are summarized in Table 1. Non-survivors more frequently received invasive mechanical ventilation than survivors (90.2% vs. 77.1%), whereas HFNC or NIV was more common among survivors (22.9% vs. 9.8%). Both admission S/F and P/F ratios were lower in non-survivors, indicating more severe oxygenation impairment at PICU admission. Vasoactive-drug use was also more frequent among non-survivors. Other demographic and baseline clinical variables showed no consistent statistically significant differences between outcome groups.

Table 1. Clinical characteristics of the total cohort and by outcome groups.

Variable	Total patients (N=218)	Survival group (N=157)	Fatal group (N=61)	P value
Age ≥6 years	128 (58.7)	87 (55.4)	41 (67.2)	0.11
Weight, kg	16.94±11.99	16.03±11.11	19.28±13.82	0.11
Male sex	135 (61.9)	92 (58.6)	43 (70.5)	0.10
Body temperature, ℃	37.32±1.06	37.38±0.94	37.15±1.32	0.22
Heart rate, bpm	140.61±27.48	141.12±27.47	139.31±27.69	0.67
Respiratory rate, breaths/min	35.57±10.36	35.50±10.22	35.75±10.82	0.88
PaCO2, mmHg	43.44±14.87	43.56±15.12	43.12±14.35	0.85
PaO2/FiO2	239.85±77.85	248.51±78.48	217.57±72.15	0.01*
SpO2/FiO2	190.12±40.72	196.52±36.85	173.66±45.64	0.001*
Ventilatory support
HFNC or NIV	42 (19.3)	36 (22.9)	6 (9.8)	0.045*
Invasive ventilation	176 (80.7)	121 (77.1)	55 (90.2)	0.045*
Vasoactive drugs	75 (34.4)	40 (25.5)	35 (57.4)	<0.001*
Comorbidities
Cerebral palsy/neuromuscular disease	45 (20.6)	28 (17.8)	17 (27.9)	0.15
Cardiovascular disease	8 (3.7)	7 (4.5)	1 (1.6)	0.55
Endocrine or metabolic disorders	3 (1.4)	2 (1.3)	1 (1.6)	1.00
Bronchopulmonary dysplasia	26 (11.9)	21 (13.4)	5 (8.2)	0.41
Hematological disease/immune deficiency/malignancy	33 (15.1)	23 (14.6)	10 (16.4)	0.91
Other comorbidities	8 (3.7)	4 (2.5)	4 (6.6)	0.31
No known comorbidity	95 (43.6)	72 (45.9)	23 (37.7)	0.35
Total hospital stay, days	23.14±21.07	23.50±19.86	22.22±24.06	0.71
PICU stay, days	15.25±15.57	13.88±11.94	18.77±22.10	0.11

Values are n (%) or mean standard deviation. *, P<0.05. HFNC, high-flow nasal cannula; NIV, non-invasive ventilation; PICU, pediatric intensive care unit.

Laboratory findings

Admission laboratory findings are shown in Table 2. Compared with survivors, non-survivors had higher BUN, lactic acid, sodium, and ALT levels (all P<0.05). Serum creatinine, PCT, AST, and D-dimer were also numerically higher in non-survivors, whereas Hb, neutrophil count, and CRP were lower, although these differences were not statistically significant.

Table 2. Laboratory examination parameters by outcome groups.

Parameter	Survival group (N=157)	Fatal group (N=61)	P value
WBCs, 109/L	11.36±7.78	12.04±8.44	0.59
Neutrophils, %	79.88±19.42	72.33±19.76	0.50
Hb, g/dL	10.59±1.83	11.07±2.22	0.06
Platelet count, 109/L	286.03±174.81	229.82±129.92	0.06
BUN, mmol/L	4.70±4.17	6.65±10.93	0.04*
Serum creatinine, μmol/L	31.34±22.65	53.95±125.35	0.20
Total bilirubin, μmol/L	13.97±29.65	12.85±16.79	0.19
pH	7.37±0.11	7.38±0.10	0.64
Lactic acid, mmol/L	1.82±2.68	2.13±1.91	0.04*
Na+, mmol/L	136.02±4.91	137.50±5.16	0.04*
K+, mmol/L	3.87±0.63	3.73±0.66	0.16
Blood glucose, mmol/L	7.71±2.66	8.28±3.57	0.58
CRP, mg/L	57.53±70.06	44.46±58.45	0.28
PCT, ng/mL	8.31±24.10	10.55±22.02	0.37
AST, U/L	92.83469.12	240.34±630.63	0.22
ALT, U/L	44.71±162.39	101.57±236.38	0.02*
APTT, s	34.36±14.73	35.39±12.13	0.32
D-dimer, mg/L	4.49±17.45	5.14±9.10	0.23

Values are mean standard deviation. *, P<0.05. ALT, alanine aminotransferase; APTT, activated partial thromboplastin time; AST, aspartate aminotransferase; BUN, blood urea nitrogen; CRP, C-reactive protein; Hb, hemoglobin; PCT, procalcitonin; WBCs, white blood cells.

Regression and ROC analyses

Correlation analysis demonstrated inverse associations between mortality and both oxygenation indices: S/F ratio (r=−0.203, P=0.001) and P/F ratio (r=−0.143, P=0.02). ALT (r=0.131, P=0.03), lactate (r=0.124, P=0.04), BUN (r=0.124, P=0.04), and sodium (r=0.119, P=0.04) were positively associated with mortality (Figure 1; Table 3).

Figure 1.

Correlation coefficients between candidate prognostic variables and in-hospital mortality. Points to the left of zero indicate inverse associations, whereas points to the right of zero indicate positive associations. Values shown beside each point represent point-biserial Pearson correlation coefficients and corresponding P values. ALT, alanine aminotransferase; BUN, blood urea nitrogen; P/F, PaO₂/FiO₂ ratio; S/F, SpO₂/FiO₂ ratio.

Table 3. Correlation analysis between selected variables and mortality.

Variable	Correlation coefficient	P value
S/F	−0.203	0.001*
P/F	−0.143	0.02*
ALT	0.131	0.03*
Lactic acid	0.124	0.04*
BUN	0.124	0.04*
Na+	0.119	0.04*

Point-biserial Pearson correlations are reported. *, P<0.05. ALT, alanine aminotransferase; BUN, blood urea nitrogen; P/F, PaO₂/FiO₂ ratio; S/F, SpO₂/FiO₂ ratio.

In exploratory logistic regression analyses, a higher S/F ratio was associated with lower mortality risk [odds ratio (OR) 0.988 per one-unit increase, 95% confidence interval (CI): 0.981−0.995; P=0.001], and a higher P/F ratio showed a similar inverse association (OR 0.995, 95% CI: 0.991−0.999; P=0.02). Higher ALT, lactate, BUN, and sodium were associated with increased mortality risk (Table 4). ROC analysis showed modest discrimination for admission S/F (AUC 0.637; Figure 2). Subgroup ROC or interaction analyses by respiratory-support modality were not performed because the HFNC/NIV subgroup included only 42 patients and 6 deaths, which would yield unstable estimates.

Table 4. Exploratory logistic regression and discriminatory performance of mortality-associated variables.

Variable	OR	95% CI	AUC	Logistic regression P	Pearson r	Pearson P
Lactic acid	1.12	1.01–1.24	0.589	0.04	0.124	0.04
P/F	0.995	0.991–0.999	0.614	0.02	−0.143	0.02
S/F	0.988	0.981–0.995	0.637	0.001	−0.203	0.001
ALT	1.003	1.000–1.005	0.588	0.03	0.131	0.03
BUN	1.050	1.002–1.100	0.575	0.04	0.124	0.04
Sodium	1.076	1.002–1.157	0.570	0.045	0.119	0.04

ORs are expressed per one-unit increase in continuous variables. ALT, alanine aminotransferase; AUC, area under the curve; BUN, blood urea nitrogen; CI, confidence interval; OR, odds ratio; P/F, PaO₂/FiO₂ ratio; S/F, SpO₂/FiO₂ ratio.

Figure 2.

Receiver operating characteristic curve of the admission S/F ratio for discrimination of in-hospital mortality. The S/F ratio yielded an AUC of 0.637; the diagonal dashed line represents the no-discrimination reference line. No data-derived cutoff is presented because the analysis was exploratory and lacked external validation. AUC, area under the curve; ROC, receiver operating characteristic; S/F, SpO₂/FiO₂ ratio.

Discussion

Key findings

In this multicenter cohort of children with severe pneumonia requiring respiratory support, lower admission S/F and P/F ratios were associated with in-hospital mortality. The S/F ratio provided an immediately obtainable respiratory signal, whereas lactate, ALT, BUN, and sodium reflected complementary hemodynamic or organ dysfunction. The AUC of 0.637 confirms that S/F alone is not a mortality test; its clinical role is as one element of structured early reassessment.

Comparison with similar research

Previous studies have demonstrated close relationships between S/F and P/F ratios in acute hypoxemic respiratory failure, pneumonia, acute lung injury, and pediatric ARDS (14-20,25). Adult critical-care studies have also associated lower S/F-based measures or greater time spent at low S/F values with deterioration and mortality (26-30), while pediatric studies support SpO₂-based respiratory indices for severity assessment in intensive-care populations (20,31,32). The present study extends this evidence to severe pediatric pneumonia across HFNC/NIV and invasive ventilation, while also showing the limitations of applying a single admission value across heterogeneous support modalities.

Explanations of findings

The clinical association between a low S/F ratio and mortality is biologically plausible because worsening ventilation-perfusion mismatch, shunt, alveolar filling, or diffuse inflammatory lung injury lowers oxygen saturation relative to the delivered FiO₂. However, severe pneumonia-related death rarely reflects hypoxemia alone. Elevated lactate may indicate tissue hypoperfusion, increased work of breathing, shock, or impaired oxygen utilization; ALT may reflect hypoxic or sepsis-associated hepatic injury; and BUN may reflect renal hypoperfusion, dehydration, catabolic stress, or evolving kidney dysfunction. These findings explain why S/F should be interpreted with circulatory status, neurologic findings, urine output, lactate, renal and hepatic markers, and formal severity scores rather than in isolation.

Clinical implications and actions needed

In routine practice, the S/F ratio may be calculated at emergency/PICU triage once the oxygen-delivery setting is documented, repeated after initial stabilization, and reassessed after any escalation or de-escalation of respiratory support. A persistently low or falling S/F ratio should prompt confirmation of signal quality and FiO₂, focused examination for increased work of breathing and fatigue, review of airway clearance and lung complications, consideration of arterial blood gas measurement, and evaluation of whether respiratory support is adequate. It should be combined with age-appropriate early warning or PICU severity scores and organ dysfunction markers such as lactate, BUN/creatinine, transaminases, coagulation indices, blood pressure, vasoactive requirement, and urine output. The S/F ratio should not delay arterial sampling, intubation, or other escalation when clinical deterioration is evident.

We did not propose an optimal S/F cutoff despite the reviewer’s clinically important suggestion. A Youden index derived from the same cohort would maximize in-sample sensitivity and specificity but, with a modest AUC, only 61 deaths, and mixed HFNC/NIV and invasive ventilation, could be unstable and easily misinterpreted as a treatment threshold. A clinically actionable cutoff should be defined in a prospective cohort using standardized SpO₂ acquisition, FiO₂ estimation, respiratory-support strata, calibration analysis, and external validation. Until then, serial trends and the broader clinical context are more defensible than a single unvalidated threshold.

Severe pneumonia in children without known comorbidity

The finding that 43.6% of patients had no known underlying disease is clinically important. Previously healthy children may still develop severe pneumonia because of age-related vulnerability, high pathogen burden or virulence, viral-bacterial coinfection, delayed presentation, rapidly progressive inflammatory lung injury, or unrecognized host susceptibility. Respiratory syncytial virus, influenza, adenovirus, Streptococcus pneumoniae, Staphylococcus aureus, and severe Mycoplasma pneumoniae infection can cause respiratory failure even without chronic disease. Because standardized microbiological and pre-hospital course data were unavailable, these mechanisms remain hypotheses rather than patient-level explanations in this cohort.

Strengths and limitations

Strengths include the multicenter design, inclusion of both non-invasive and invasive respiratory support, and simultaneous assessment of oxygenation and organ dysfunction markers. Several limitations should be considered. First, the retrospective design prevents causal inference and may have introduced selection, measurement, and information bias. Clinical documentation, timing of testing, and escalation thresholds may have differed across centers. Second, FiO₂ estimation during HFNC and NIV is less precise than during controlled invasive ventilation, and pulse oximetry has an upper-range plateau effect. Third, respiratory-support-specific prognostic performance was not evaluated; the small HFNC/NIV subgroup and confounding by indication would make such analyses unstable. Fourth, only admission values were analyzed, so dynamic S/F trajectories and treatment response were not assessed. Fifth, complete-case analysis may have introduced selection bias. Sixth, microbiological etiology was not available in the harmonized dataset. Although the aggregate denominator of pediatric pneumonia admissions was obtained from administrative records, individual-level characteristics of children with non-severe pneumonia were unavailable, precluding direct comparison with the PICU cohort. Finally, all sites were tertiary PICUs, limiting generalizability to general wards, emergency departments, children managed without respiratory support, and health systems with different escalation thresholds.

Conclusions

The admission S/F ratio was inversely associated with in-hospital mortality among children with severe pneumonia requiring respiratory support in tertiary PICUs. It can be used at PICU admission and after changes in oxygen therapy as a rapid, non-invasive adjunct to bedside examination, P/F assessment, pediatric severity or organ dysfunction scores, lactate, renal and hepatic markers, and hemodynamic status. A low or decreasing S/F ratio should trigger integrated reassessment of oxygen delivery, respiratory-support adequacy, and evolving multisystem dysfunction, but should not be used alone or as an unvalidated treatment cutoff. Prospective, modality-stratified studies are needed to define reproducible thresholds and determine whether serial S/F-guided assessment improves clinical outcomes.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Medical Ethics Committee of Children’s Hospital, Zhejiang University School of Medicine (No. 2021-IRB-302). All participating hospitals were informed of and agreed to the study. Individual consent for this retrospective analysis was waived.

Data Sharing Statement

Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-0577/dss

tp-15-08-303-dss.pdf ^{(66.4KB, pdf)}

DOI: 10.21037/tp-2026-0577