Development and internal validation of a nomogram for predicting flexible bronchoscopy utilization in children with lobar pneumonia caused by Mycoplasma pneumoniae: a single-center retrospective cohort study.

Abstract

Background: The recent resurgence of Mycoplasma pneumoniae (MP) pneumonia has increased the number of children with lobar pneumonia who may receive flexible bronchoscopy (FB), yet the decision to perform FB remains subjective, and no quantitative tool exists for this specific high-risk subgroup. This study aimed to develop and internally validate a nomogram to predict FB utilization in children with MP-related lobar pneumonia using routine admission indicators to support real-time bedside decision-making.

Methods: This single-center retrospective study included 345 children (aged 1 month-18 years) with serology-confirmed MP-related lobar pneumonia (January 2022-March 2025); those with congenital heart disease, immunodeficiency, or plastic bronchitis were excluded. Patients were divided into FB (n=171) and non-FB (n=174) groups based on whether FB was performed. Laboratory predictors were restricted to values obtained within 72 hours of admission and before the FB decision. FB was recommended if symptoms or signs persisted after ≥72 hours of standard therapy. Multivariate logistic regression was used to identify predictors, and a nomogram was constructed and internally validated by 10-fold cross-validation. Model performance was assessed by area under the curve (AUC), calibration, and decision curve analysis (DCA).

Results: Viral coinfection [odds ratio (OR) =23.271, 95% confidence interval (CI): 9.299-64.900, P<0.001] was the strongest independent risk factor for FB utilization, while bacterial coinfection (OR =0.215, 95% CI: 0.088-0.507, P<0.001) and male sex (OR =0.146, 95% CI: 0.061-0.313, P<0.001) were independent protective factors. The nomogram incorporating these six variables demonstrated moderate discrimination (apparent AUC =0.741, 95% CI: 0.689-0.792; cross-validated AUC =0.705) and good calibration (Hosmer-Lemeshow P=0.56). At the optimal cutoff value of 0.56, the model had a sensitivity of 51.5% and a specificity of 85.1%. DCA demonstrated net clinical benefit over a "treat-all" strategy within a threshold probability range of 0.1-0.3.

Conclusions: The nomogram showed modest ability to predict FB use in this single-center cohort, but its clinical utility for identifying children with lobar pneumonia caused by MP who truly require FB remains uncertain.