Construction of a risk prediction model for multiple bronchoalveolar lavages in children with necrotizing pneumonia based on LASSO regression
摘要
To explore the risk factors for multiple bronchoalveolar lavages in children with necrotizing pneumonia (NP) and construct a prediction model.
MethodsTo analyze the clinical data of patients with NP from January 2019 to June 2025, and divided into a Single- Bronchoalveolar lavage (BAL) group and Multiple-BAL group, analyze and construct the model.
ResultsThe single-BAL group had 139 patients, the multiple-BAL group 196. The incidence of sputum plug obstruction and bronchial erosion with necrosis was greater in the multiple-BAL group (P < 0.05). The platelet distribution width (PDW), lactate dehydrogenase (LDH) in the multiple-BAL group were greater (P < 0.05). Sputum plug obstruction, bronchial mucosal erosion and necrosis with softening, fibrinogen (FIB) < 2.84 g/L, PDW > 9.5%, and LDH > 576 U/L are risk factors for multiple alveolar lavages in pediatric patients with NP. The combined model had an AUC of 0.801, indicating good discrimination. The combined model showed good predictive performance after Bootstrap validation.
ConclusionsThe presence of sputum plug obstruction, bronchial mucosal erosion and necrosis with softening, FIB < 2.84 g/L, PDW > 9.5%, and LDH > 576 U/L are risk factors for multiple bronchoalveolar lavages in pediatric patients with NP.
ImpactIdentified 5 risk factors for multiple-BAL in pediatric NP, and built a quantifiable guiding model. Fills the gap of no multiple-BAL criteria in pediatric NP, integrating pathology and hematological indicators. Optimizes treatment, reduces burden, and facilitates clinical standardization.