Early Identification of Children at Risk for Complicated Parapneumonic Pleural Effusion: Development of a Prediction...

Parapneumonic pleural effusion (PPE) is a frequent complication of pediatric pneumonia. Some cases progress to complicated PPE (cPPE), which is associated with greater morbidity, longer treatment, and a higher likelihood of need for pleural drainage. Early identification of children at risk of progression remains difficult at presentation, and pediatric evidence on early predictors is scarce and inconsistent.

We conducted a retrospective observational study including children aged <18 years admitted with PPE between 2015 and 2025. cPPE was defined through standard laboratory and/or imaging criteria. Twenty clinical, laboratory, and radiological candidate predictors available at initial evaluation were assessed. After univariate screening, variables with statistical association and clinical relevance were entered into a multivariate logistic regression model. After model assessment, a simplified score was derived from the original model for bedside use.

A total of 122 children were included, 77 (63%) with cPPE and 45 (37%) with uncomplicated PPE. In univariate analyses, younger age, dyspnea, chest wall retractions, reduced lung sounds, C-reactive protein, lower hemoglobin, pleural effusion thickness, and mediastinal shift were associated with cPPE. In the final multivariate model, five variables remained independently associated: age (OR 0.73, 95% CI 0.63-0.84), retractions (OR 3.23, 95% CI 1.16-8.95), C-reactive protein ≥176.5 mg/L (OR 2.83, 95% CI 1.02-7.82), pleural effusion thickness (OR 2.16, 95% CI 1.27-3.70), and mediastinal shift (OR 10.75, 95% CI 2.24-51.64). The model showed good discriminatory performance, with an AUC of 0.883 (95% CI 0.817-0.948). The optimal probability threshold was 0.49, yielding a sensitivity of 92.2% and a specificity of 71.1%. A simplified additive clinical score derived from the model demonstrated similar performance, allowing stratification into four risk groups with progressively increasing observed rates of cPPE.

Early risk stratification of pediatric PPE is feasible using a small set of readily available clinical, laboratory, and radiological variables. The proposed model and the adapted clinical score showed good discriminatory ability and may help identify children at increased risk of progression to cPPE who could benefit from closer monitoring and early optimized management. External validation is needed before routine clinical implementation.