A Multivariable Prediction Model and Bedside Score for Death or Bronchiolitis Obliterans in Pediatric Severe Adenovirus Pneumonia

Background: Severe adenovirus pneumonia (SAP) in children is associated with both in-hospital mortality and long-term respiratory sequelae, particularly bronchiolitis obliterans (BO). However, most previous studies have focused on single biomarkers or single clinical endpoints, limiting integrated prognostic assessment. This study aimed to develop and internally validate a multivariable prediction model for a composite adverse outcome of in-hospital death or BO in pediatric SAP.Methods: This single-center, retrospective, patient-level integrated analysis included 90 unique children with SAP after cross-study deduplication and eligibility screening of three partially overlapping source cohorts. The primary outcome was a composite of in-hospital death or BO. Candidate predictors were screened using least absolute shrinkage and selection operator regression with 10-fold cross-validation, followed by multivariable logistic regression and backward elimination based on the Akaike information criterion. Model performance was assessed by discrimination, calibration, decision curve analysis, and bootstrap internal validation. A simplified bedside risk score was subsequently derived from the final model.Results: The composite adverse outcome occurred in 21 of 90 children (23.3%), including 13 in-hospital deaths and 8 cases of BO. Five predictors were retained in the final model: an increase in the neutrophil-to-lymphocyte ratio from baseline to the treatment-intensive phase (p = 0.001), systemic immune-inflammation index of 1500 or greater during the treatment-intensive phase (p = 0.007), higher treatment-intensive-phase lactate dehydrogenase (p = 0.011), longer total fever duration (p = 0.039), and bacterial co-infection (p = 0.038). The apparent area under the receiver operating characteristic curve was 0.887, and the bootstrap-corrected value was 0.845. The derived bedside score stratified patients into low-, intermediate-, and high-risk groups, with observed event rates of 4.2%, 25.0%, and 85.7%, respectively.Conclusion: In this single-center retrospective integrated analysis, a multivariable model incorporating dynamic inflammatory change, fever burden, tissue injury, and bacterial co-infection showed promising internally validated discrimination for the composite outcome of in-hospital death or BO in children with SAP. The derived bedside score should be regarded as a hypothesis-generating tool for internal risk stratification, and external validation is required before clinical implementation.