Clinical risk factors and a predictive nomogram for mechanical ventilation in pediatric patients with infection-related plastic bronchitis

Abstract

Background

Plastic bronchitis (PB) in children may cause respiratory failure requiring mechanical ventilation (MV), but systematic risk assessments are lacking. This study aimed to identify clinical risk factors for MV in infection-related PB and to develop a predictive nomogram for individualized risk stratification.

Methods

In this retrospective cohort study, pediatric patients diagnosed with infection-related PB at our center between August 2019 and August 2025 were included. Univariate and multivariate logistic regression analyses were performed to identify independent risk factors for MV. A predictive model and nomogram were developed, with discrimination evaluated by area under the receiver operating characteristic curve (AUC) and calibration assessed using bootstrap resampling.

Results

A total of 103 pediatric patients were included in the study, comprising 78 with single-pathogen infections and 25 with mixed infections. Among them, 20 (19.4%) required MV. Multivariate analysis identified younger age (OR = 0.93, 95%CI: 0.88–0.97), elevated PaCO₂ (OR = 5.44, 95%CI: 2.10–21.1), and more extensive lobar involvement (OR = 6.53, 95%CI: 1.85–32.4) as independent risk factors. The model based on age and PaCO₂ achieved an AUC of 0.922; adding lobar involvement slightly increased the AUC to 0.954 without statistical significance (p = 0.118). Nomogram-predicted probabilities closely matched observed outcomes (mean absolute error = 0.023).

Conclusions

Age, PaCO₂, and lobar involvement were independent predictors of MV in pediatric infection-related PB. The model based on age and PaCO₂ demonstrated high discrimination and reliable calibration, and the derived nomogram serves as a practical tool for early risk assessment and individualized management.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-025-12242-1.

Plastic bronchitis (PB) is characterized by the formation of bronchial casts that can cause varying degrees of airway obstruction and respiratory dysfunction [1, 2]. In recent years, the incidence of pediatric PB has risen, largely in association with outbreaks of respiratory pathogens such as influenza virus and Mycoplasma pneumoniae [3–5]. While most children present with mild to moderate disease, a subset may progress rapidly to severe respiratory failure requiring mechanical ventilation (MV). In particular, infection-associated PB often exhibits fulminant progression. Early identification of high-risk patients is therefore crucial to enable timely intervention, optimize treatment strategies, and improve clinical outcomes.

Previous studies have primarily investigated the mechanisms underlying infection-associated PB and identified several poor prognostic factors [6–8]. However, systematic analyses of risk factors for MV in pediatric PB are limited, and validated predictive tools to guide clinical decision-making are lacking. This study aimed to develop a practical risk prediction tool for the early identification of pediatric patients at high risk for MV, thereby supporting timely intervention, guiding clinical management, and improving clinical outcomes.

Methods

Study design and population

This retrospective cohort study was a single-center study, approved by the Ethics Committee of Guangzhou Women and Children’s Medical Center (Approval No. 516A01-2024), with informed consent waived. Pediatric patients diagnosed with infection-associated PB at our center between August 2019 and August 2025 were screened for eligibility. The inclusion criteria were as follows: (1) age ≤ 18 years. (2) PB confirmed by bronchoscopy as infection-related [9]. All enrolled patients underwent bronchoscopy due to clinical or radiologic evidence suggestive of bronchial obstruction. According to our institutional protocol for critical care management, bronchoscopy was routinely performed in children with respiratory infections who presented with any of the following conditions: persistent atelectasis or imaging findings indicative of airway obstruction; severe wheezing or respiratory distress accompanied by mucus retention; or difficulty in airway secretion clearance. The procedure was performed under sedation and continuous monitoring by experienced pediatric pulmonologists. The diagnosis of PB was established based on characteristic bronchoscopic findings and the presence of bronchial casts. The diagnosis of infection was confirmed based on laboratory findings, including complete blood count, C-reactive protein (CRP), procalcitonin (PCT), and pathogen identification by polymerase chain reaction or blood culture. (3) availability of complete clinical data. Exclusion criteria were: (1) PB caused by non-infectious factors; and (2) the presence of severe underlying cardiac, pulmonary, immunological, oncological, or neurological diseases. MV was initiated in pediatric patients who met any of the following criteria: persistent severe hypoxemia (PaO₂ < 50 mmHg or SpO₂ < 85%, or PaO₂/FiO₂ < 200 mmHg despite FiO₂ ≥ 0.6), severe hypercapnia with acidosis (PaCO₂ >60 mmHg and pH < 7.25), recurrent apnea or impending respiratory arrest, impaired consciousness with loss of airway protective reflexes, or circulatory instability requiring controlled ventilation to maintain adequate oxygenation and acid–base balance. Patients were categorized into the MV group or the non-MV group according to whether invasive MV was required.

Data collection

Baseline demographic data (age, sex, and body weight), clinical manifestations (fever, cough, and pre-admission disease duration), laboratory findings (arterial blood gas analysis, complete blood count, inflammatory markers, organ function indices, and microbiological results), imaging characteristics (extent of lobar involvement, consolidation, pleural effusion, etc.), major therapeutic interventions (fiberoptic bronchoscopy, corticosteroids, intravenous immunoglobulin, and antiviral therapy), and outcomes (duration of MV, length of hospital stay, survival status, and sequelae including obliterative bronchiolitis, obliterative bronchitis, bronchiectasis, and asthma) were collected. Laboratory and radiographic parameters were selected based on the most abnormal values obtained within the first 24 h after admission, in order to better reflect the initial severity of illness and enhance the clinical relevance of the analysis. Imaging evaluations were independently performed by two experienced radiologists, and discrepancies were resolved through discussion until consensus was reached.

Statistical analysis

To identify factors associated with the need for MV, we systematically evaluated clinical, laboratory, and imaging parameters. Continuous variables were expressed as medians with interquartile ranges and compared between groups using the Mann–Whitney U test. Categorical variables were summarized as frequencies and percentages and compared using the χ² test or Fisher’s exact test. Variable selection was conducted in two steps: first, variables with p < 0.05 in univariate analysis were considered candidates; second, among highly correlated variables, only the most clinically representative indicators with VIF < 5 were retained for inclusion in the multivariate logistic regression model. The final set of clinical, laboratory, and imaging variables was entered into a multivariate logistic regression to identify independent risk factors for MV in children with infection-associated PB.

A predictive model was developed based on the results of multivariate logistic regression. Model discrimination was assessed using ROC curves and the area under the curve (AUC), with performance compared across models using the DeLong test. A nomogram was then constructed for the final model, and internal validation was performed using bootstrap resampling (B = 1000) to evaluate calibration, with mean absolute error (MAE), mean squared error (MSE), and the 90th percentile of absolute error calculated. To minimize bias from missing data, variables with more than 20% missing values were excluded, and the maximum missing rate among the included variables was 1.94%. Multiple imputation was performed using robust linear regression via the impute_rlm() function in the simputation package (R), based on clinically relevant covariates such as white blood cell count, neutrophil percentage, CRP, hospital stay days, and the number of lung lobes involved, to impute missing values in physiological and biochemical variables (e.g. pH, PO₂, PaCO₂, lactate, glucose, and monocyte percentage). All statistical analyses were conducted using R software (version 4.3.3), and a two-sided p < 0.05 was considered statistically significant.

Results

Pediatric patients characteristics

Between August 2019 and August 2025, 110 pediatric patients with infection-associated PB were screened. After excluding 3 with severe underlying pulmonary disease, 2 with nephrotic syndrome, and 2 with hematologic malignancies, 103 patients were included in the final analysis. Pathogen distribution (Supplementary Table 1) showed 78 single-pathogen infections (75.7%) and 25 mixed infections (24.3%), with Mycoplasma pneumoniae being the most frequent pathogen (69 patients, 67.0%). MV was required in 20 patients (19.4%), and one patient (1.0%) died during hospitalization.

Compared with the non-MV group (n = 83), MV patients were significantly younger, lighter, and had a shorter pre-admission disease course (all p < 0.05), while sex distribution was similar. Fever and cough were nearly universal, and peak temperature did not differ between groups. On admission, MV patients exhibited more pronounced gas-exchange and acid–base abnormalities, with significantly lower pH and higher PaCO₂ (both p < 0.001), whereas PaO₂ was similar between groups. Most inflammatory markers did not differ, but the MV group had significantly higher leukocyte and neutrophil counts (both p < 0.01), with similar lymphocyte and platelet levels. Imaging revealed more extensive disease in MV patients, with multilobar involvement more frequent (p = 0.006). Consolidation, pleural effusion, ground-glass opacity, pneumothorax, atelectasis, and mediastinal emphysema occurred at similar rates (Table 1). Notably, some patients exhibited rapid radiographic progression, with initially mild findings rapidly advancing to extensive pulmonary involvement accompanied by marked clinical deterioration (Fig. 1).

Table 1.

Baseline characteristics of patients with and without mechanical ventilation

Variables	Overall(n = 103)	Non-MV group(n = 83)	MV group(n = 20)	p
Demographics
Female, n (%)	52 (50.5)	45 (54.2)	7 (35.0)	0.196
Age(months)	68.00 [45.00, 87.50]	80.00 [55.50, 91.00]	35.50 [20.00, 50.25]	< 0.001
Weight (kg)	18.50 [14.80, 22.45]	19.40 [16.00, 23.25]	14.00 [11.50, 17.12]	< 0.001
Clinical symptoms
Days before admission(day)	6.00 [4.00, 8.00]	6.00 [5.00, 8.00]	4.00 [3.00, 6.25]	0.013
Fever, n (%)	99 (96.1)	79 (95.2)	20 (100.0)	0.721
Highest temperature (°C)	39.60 [39.00, 40.00]	39.60 [39.00, 40.00]	39.75 [38.82, 40.00]	0.623
Cough, n (%)	103 (100.0)	83 (100.0)	20 (100.0)	NA
Laboratory findings
pH	7.40 [7.37, 7.43]	7.41 [7.38, 7.44]	7.33 [7.24, 7.38]	< 0.001
PaO₂ (kPa)	14.92 [12.70, 16.70]	15.00 [13.20, 16.70]	13.85 [9.93, 16.75]	0.361
PaCO₂ (kPa)	4.90 [4.50, 5.70]	4.80 [4.50, 5.30]	6.40 [4.85, 8.20]	< 0.001
Lactate abnormal, n (%)	25 (24.3)	23 (27.7)	2 (10.0)	0.171
White blood cell (×10⁹/L)	7.70 [6.15, 12.20]	7.30 [5.93, 10.61]	13.30 [8.45, 17.64]	0.001
Neutrophils (×10⁹/L)	5.13 [3.57, 9.64]	4.84 [3.38, 6.81]	11.31 [5.76, 15.45]	< 0.001
Lymphocytes (×10⁹/L)	1.79 [1.13, 2.30]	1.85 [1.28, 2.30]	1.63 [0.81, 2.28]	0.189
Platelets (×10⁹/L)	258.00 [215.00, 336.00]	253.00 [214.50, 324.50]	322.50 [240.25, 421.25]	0.101
C-reactive protein (mg/L)	23.60 [11.57, 39.21]	23.60 [11.48, 44.55]	21.97 [12.53, 27.61]	0.708
ALT abnormal, n (%)	15 (14.6)	13 (15.7)	2 (10.0)	0.771
Creatinine abnormal, n (%)	6 (5.8)	3 (3.6)	3 (15.0)	0.156
Fibrinogen abnormal, n (%)	47 (45.6)	41 (49.4)	6 (30.0)	0.189
Imaging findings
Lung lobes involved, n (%)				0.006
1	32 (31.1)	31 (37.3)	1 (5.0)
2	51 (49.5)	40 (48.2)	11 (55.0)
3	18 (17.5)	11 (13.3)	7 (35.0)
4	1 (1.0)	1 (1.2)	0 (0.0)
5	1 (1.0)	0 (0.0)	1 (5.0)
consolidation, n (%)	102 (99.0)	82 (98.8)	20 (100.0)	1
Atelectasis, n (%)	36 (35.0)	25 (30.1)	11 (55.0)	0.067
Ground-glass opacity, n (%)	4 (3.9)	4 (4.8)	0 (0.0)	0.721
Pleural effusion, n (%)	41 (39.8)	32 (38.6)	9 (45.0)	0.784
Pneumothorax, n (%)	5 (4.9)	4 (4.8)	1 (5.0)	1
Mediastinal emphysema, n (%)	6 (5.8)	3 (3.6)	3 (15.0)	0.156

Abbreviations: MV, mechanical ventilation; ALT, alanine aminotransferase; PaO₂, partial pressure of oxygen; PaCO₂, partial pressure of carbon dioxide. Notes: Continuous variables are expressed as median (IQR); categorical variables are expressed as number (percentage)

Fig. 1.

A 6-year-old boy presented with an 8-day history of fever and cough, progressing to wheezing. Initial chest radiography suggested bronchitis (A), but within one day repeat imaging showed rapid progression to bilateral infiltrates and a small left upper lung pneumothorax (B). Emergency fiberoptic bronchoscopy revealed diffuse mucosal congestion and edema with adherent yellowish-white plugs, predominantly in the right bronchi. Representative images illustrate the openings of the main bronchi (C) and the right middle and lower lobe bronchi (D). Influenza A virus was identified as the pathogen

Treatment and outcomes of pediatric patients

Significant differences were observed between MV and non-MV groups regarding treatment and clinical course. MV patients more frequently received immunoglobulin (90.0% vs. 25.3%, p < 0.001) and antiviral therapy (75.0% vs. 20.5%, p < 0.001), while glucocorticoid and antibiotic use were comparable. The interval from fever onset to bronchoscopy was shorter in MV children (p < 0.001), and they required more bronchoscopies overall (p = 0.019). In terms of outcomes, MV patients had markedly longer hospital stays (median 17.5 vs. 9.0 days, p < 0.001) and underwent invasive ventilation for a median of 7.0 days. Fever duration was similar between groups. Complication rates, including obliterative bronchiolitis, obliterative bronchitis, bronchiectasis, and asthma, were low and did not differ significantly (Table 2).

Table 2.

Treatments and outcomes of patients with and without mechanical ventilation

Variables	Overall(n = 103)	Non-MV group(n = 83)	MV group(n = 20)	p
Treatment
Glucocorticoid use, n (%)	53 (51.5)	39 (47.0)	14 (70.0)	0.11
Immunoglobulin use, n (%)	39 (37.9)	21 (25.3)	18 (90.0)	< 0.001
Antiviral therapy, n (%)	32 (31.1)	17 (20.5)	15 (75.0)	< 0.001
Antibiotic therapy, n (%)	102 (99.0)	82 (98.8)	20 (100.0)	1
Interval from fever to bronchoscopy (days)	7.00 [6.00, 11.00]	8.00 [6.00, 11.00]	3.00 [1.00, 7.50]	< 0.001
Number of bronchoscopies, n (%)				0.019
1	65 (63.1)	57 (68.7)	8 (40.0)
2	30 (29.1)	22 (26.5)	8 (40.0)
3	8 (7.8)	4 (4.8)	4 (20.0)
Clinical course
Long fever duration (days)	8.00 [6.00, 11.00]	9.00 [6.50, 11.00]	7.00 [5.75, 10.25]	0.149
Invasive ventilation days	0.00 [0.00, 0.00]	0.00 [0.00, 0.00]	7.00 [6.00, 8.00]	< 0.001
Hospital stay (days)	10.00 [7.50, 14.00]	9.00 [7.00, 11.50]	17.50 [12.00, 20.25]	< 0.001
Post-discharge sequelae
Obliterative bronchiolitis, n (%)	1 (1.0)	0 (0.0)	1 (5.0)	0.437
Obliterative bronchitis, n (%)	5 (4.9)	4 (4.8)	1 (5.0)	1
Bronchiectasis, n (%)	1 (1.0)	1 (1.2)	0 (0.0)	1
Bronchial asthma, n (%)	0 (0.0)	0 (0.0)	0 (0.0)	NA

Logistic regression analysis of mechanical ventilation risk factors

To explore risk factors for MV in pediatric patients with infection-associated PB, variables that showed significant differences between the MV and non-MV groups (Table 1, p < 0.05) were included in univariate and multivariate logistic regression analyses. Given the strong correlation between age and weight (r = 0.834), only age was retained in the analysis. Due to the strong correlation between pH and PaCO₂ (r = -0.751), PaCO₂ was selected as representative variable; and because of the high correlation between total WBC count and neutrophils (r = -0.853), neutrophils were used as the representative variable. Univariate analysis indicated that younger age, shorter duration before admission, elevated PaCO₂, increased neutrophils, and greater lobar involvement were significantly associated with MV. Multivariate analysis further revealed that younger age (OR = 0.93, 95% CI: 0.88–0.97, p < 0.001), elevated PaCO₂ (OR = 5.44, 95% CI: 2.10–21.1, p < 0.001), and greater lobar involvement (OR = 6.53, 95% CI: 1.85–32.4, p = 0.003) were independent risk factors (Table 3).

Table 3.

Risk factors for mechanical ventilation in pediatric patients with infection-associated plastic bronchitis

Variables	univariate analysis		multivariate analysis
	OR (95% CI)	p -value	OR (95% CI)	p -value	VIF
Age (months)	0.95 (0.93 to 0.97)	< 0.001	0.93 (0.88 to 0.97)	< 0.001	1.11
Days before admission	0.83 (0.67 to 0.98)	0.023	0.88 (0.65 to 1.10)	0.31	1.06
PaCO₂ (kPa)	2.72 (1.67 to 5.07)	< 0.001	5.44 (2.10 to 21.1)	< 0.001	1.13
Neutrophils (×10⁹/L)	1.11 (1.03 to 1.23)	0.007	1.07 (0.91 to 1.19)	0.38	1.08
Lung lobes involved	3.06 (1.57 to 6.63)	< 0.001	6.53 (1.85 to 32.4)	0.003	1.14

Abbreviations: CI, confidence interval; OR, odds ratio; PaCO₂, partial pressure of carbon dioxide

Predictive models and calibration for mechanical ventilation

Based on multivariable logistic regression analysis, two models were developed to predict the risk of MV. Model 1, which included age and PaCO₂, yielded an AUC of 0.922. When the extent of lung lobe involvement was added, Model 2 achieved a slightly higher AUC of 0.954. However, DeLong’s test showed no statistically significant difference between the two models (p = 0.118, Fig. 2). This suggests that age and PaCO₂ alone provide substantial predictive performance, which also avoids the subjectivity and radiation exposure associated with imaging evaluation. A nomogram was constructed based on Model 1 to facilitate clinical risk prediction (Fig. 3). Calibration was assessed using bootstrap resampling (B = 1000), and the results (Fig. 4) demonstrated good agreement between predicted and observed probabilities: mean absolute error = 0.023, mean squared error = 0.00075, and the 90th percentile of absolute error = 0.044, indicating that the predicted probabilities were reliable and clinically applicable.

Fig. 2.

Receiver operating characteristic (ROC) curves comparing Model 1 (age + PCO₂) and Model 2 (age + PCO₂ + lung lobes involved)

Fig. 3.

Nomogram for predicting the probability of mechanical ventilation. The nomogram incorporates age and PaCO₂

Fig. 4.

Calibration curve of the prediction model based on age and PaCO₂. Calibration was assessed using bootstrap resampling (B = 1000)

Discussion

This retrospective study analyzed 103 pediatric patients with infection-associated PB to identify independent risk factors for MV and to develop a predictive model and nomogram based on regression analysis. The results demonstrated that age, PaCO₂ levels, and the extent of lobar involvement were independent risk factors for requiring MV. The model based on age and PaCO₂ alone exhibited high discriminative ability (AUC = 0.922), and the addition of lobar involvement slightly increased the AUC (0.954), though the difference was not statistically significant (DeLong test, p = 0.118). Bootstrap calibration further indicated that the predicted probabilities from the age–PaCO₂ model aligned with observed outcomes, confirming its strong potential for clinical application.

In this study, age was identified as an independent risk factor for MV in pediatric patients with PB, with younger age being associated with a higher likelihood of requiring MV. This association is closely related to the anatomical, physiological, and immunological characteristics of young children [10]. First, younger children have narrower airways, and airway resistance increases exponentially as the diameter decreases, rendering them less tolerant to mucus or cast obstruction and more susceptible to respiratory compromise and respiratory distress. Second, the pulmonary compliance and diaphragmatic function in infants and young children are not fully developed, resulting in limited respiratory reserve and increased susceptibility to fatigue and respiratory failure under acute respiratory stress. Additionally, younger children have immature immune function and reduced mucociliary clearance, making them more susceptible to severe infections and inflammatory responses, which can exacerbate airway obstruction and promote the progression of PB [11, 12].

The results of this study showed that elevated PaCO₂ was an independent predictor of MV (OR = 5.44). Hypercapnia reflected carbon dioxide retention, primarily caused by alveolar hypoventilation, airway obstruction, or respiratory muscle fatigue, and served as a direct indicator of respiratory failure [13]. Recent studies have suggested that elevated PaCO₂ was associated with impaired systemic organ perfusion [14, 15]. Compared with other parameters, PaCO₂, as an objective arterial blood gas measurement, is sensitive and timely, providing a direct assessment of respiratory dysfunction. Therefore, it serves as an important marker for identifying critically ill patients and determining the need for MV, enabling clinicians to guide early interventions and tailor individualized treatment strategies.

In addition, patients in the MV group exhibited more extensive lobar involvement, and multivariate analysis further confirmed it as an independent risk factor for MV (OR = 6.53). When multiple lobes are obstructed by casts or mucus plugs, effective alveolar ventilation decreases, lung compliance is reduced, and ventilation–perfusion mismatch worsens, all of which readily lead to hypoxemia. Previous studies also indicated that extensive lobar involvement was closely associated with more severe disease progression and a higher risk of respiratory failure in PB [16, 17]. However, imaging assessment has inherent limitations. First, its interpretation is often subjective and varied depending on image quality, the radiologist’s experience, and institutional differences. Second, although chest CT provides direct visualization of the extent of pulmonary involvement, it involves radiation exposure, which should be used with caution, particularly in pediatric patients [18–20]. In addition, some patients with PB exhibited rapid disease progression, and early imaging changes were sometimes minimal, while airway casts had already significantly impaired ventilation, indicating that relying solely on imaging could underestimate disease severity.

Previous studies have mainly focused on the mechanisms and risk factors associated with infection-related PB. Huang et al. [21] identified elevated neutrophil counts as an independent risk factor for Mycoplasma pneumoniae-associated PB. Zhao et al. [22] developed and validated nomogram models to predict the occurrence of PB in children with refractory Mycoplasma pneumoniae pneumonia, incorporating indicators such as body temperature, inflammatory cytokines, and imaging findings. Guo et al. [23] conducted a comprehensive analysis of children with PB and found that mucosal erythema, secretion color, and PCT levels were risk factors. Moreover, Lin et al. [24] demonstrated that elevated CRP and LDH levels serve as independent risk factors for pleural effusion in children with pneumonia-induced PB. While these studies have provided valuable insights into PB formation and early risk factors, none have specifically focused on predicting the need for MV in pediatric patients with PB. It should also be noted that the study period coincided with the COVID-19 pandemic. Recent studies have highlighted the impact of viral evolution, vaccination, and long COVID on pediatric respiratory infections and immune responses, which may indirectly influence the clinical presentation and severity of PB [25–28]. In addition, PB should be differentiated from other airway inflammatory diseases, such as asthma, which also involve airway remodeling and inflammatory cell infiltration [29]. In this study, we systematically evaluated clinical, laboratory, and imaging variables and applied a rigorous variable selection strategy to minimize the impact of multicollinearity. Based on these analyses, we not only identified age and PaCO₂as independent risk factors, but also as one of the first studies to develop and validate a predictive model for MV risk in pediatric PB, established a corresponding regression model and nomogram. Particularly, the inclusion of age allowed the model to capture the increased vulnerability of younger children to airway obstruction and respiratory compromise, reinforcing its clinical relevance. This model demonstrated good discrimination and calibration, allowing intuitive quantification of the probability of MV, which could facilitate clinical risk stratification, early intervention, and individualized management. Notably, in situations where imaging availability was limited or radiation exposure required careful consideration, the age and PaCO₂ based model demonstrated substantial potential for clinical application due to its simplicity, objectivity, and bedside feasibility.

This study had several limitations. First, it was a single-center retrospective study with a limited sample size, which might have introduced selection bias. Although potential confounding factors were adjusted for in the multivariate analysis, the findings still need to be validated in larger cohorts. Second, due to the limited sample size and missing data for some variables, not all potentially relevant risk factors, including D-dimer, procalcitonin, and pathogen type, were included. Third, the relatively small number of patients who required MV might have reduced the statistical power and increased the risk of model overfitting. In addition, external validation was not performed, which may limit the generalizability of the model to other populations or clinical settings. Nevertheless, the model constructed using only age and PaCO₂ already demonstrated high discriminative ability and accuracy. Additionally, although institutional guidelines for MV indications were in place, the final decision and timing of ventilation were determined by the attending physicians based on individual clinical conditions, which may have introduced some variability and represented an inherent limitation of this study. Future studies should involve multicenter, large-scale prospective cohorts and explore the integration of artificial intelligence and machine learning approaches to further improve the predictive performance and clinical utility of the model. Moreover, recent advances in non-invasive imaging modalities have shown promise for assessing pulmonary ventilation and structure without ionizing radiation. Hyperpolarized gas MRI techniques—using propane, butane, or ether gases—have demonstrated rapid, high-resolution lung ventilation imaging in preclinical models and could serve as valuable diagnostic tools in pediatric airway diseases [30–32]. Future research may explore their applicability for PB diagnosis and follow-up.

Conclusion

In summary, age, PaCO₂, and the extent of lobar involvement are independent risk factors for MV in pediatric patients with PB. The predictive model based on age and PaCO₂ demonstrates high discriminative ability and good calibration, and the corresponding nomogram facilitates clinical application. This model provides an important reference for risk stratification and optimization of treatment strategies in pediatric PB patients.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

PB

Plastic bronchitis

MV

Mechanical ventilation

PaCO₂

Partial pressure of carbon dioxide

PaO₂

Partial pressure of oxygen

WBC

White blood cell

CRP

C-reactive protein

PCT

Procalcitonin

ALT

Alanine aminotransferase

ROC

Receiver operating characteristic

AUC

Area under the curve

PICU

Pediatric intensive care unit

CI

Confidence interval

COVID

Coronavirus disease

Author contributions

Conceived and designed the study: LF, YYY, YLZ and HYP. Collected and analyzed data: HYP, ZXL, JPT, JHZ, CMZ, FYC, RD, JH, XXW. Wrote the manuscript: HYP, LF. All authors had reviewed and approved the final manuscript.

Funding

This study was funded by Guangzhou Health Science and Technology Project (20231A011031). Guangxi Zhuang Autonomous Region Health Commission Self-Funded Research Project (Z-B20241344). Guangdong Province Medical Science and Technology Research Fund: Clinical Study on the Improvement of Microcirculation Disorders in Children with Shock by Penequinol Hydrochloride (B2024274).

Data availability

The datasets used in this study are available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Guangzhou Medical University Affiliated Women’s and Children’s Medical Center (Approval No. 516A01-2024), with informed consent waived due to the retrospective study design and anonymization of patient data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.