Development and validation of a dual-risk prediction model for pediatric acute osteoarticular infections caused by CA-MRSA

To identify independent risk factors for complicated disease course and long-term sequelae in pediatric acute osteoarticular infections (OAI) caused by community-acquired methicillin-resistant Staphylococcus aureus (CA-MRSA), and to develop visualized prediction models. Clinical data of 91 patients with OAI caused by CA-MRSA admitted to a tertiary pediatric center in China from April 2013 to April 2025 were retrospectively analyzed. Patients were categorized into simple (S-MRSA)/ complicated (C-MRSA) groups and favorable (F-MRSA)/unfavorable (U-MRSA) outcome groups. After excluding multicollinearity, multivariate Logistic Regression and Random Forest algorithm were used to identify independent predictors. Nomograms were constructed and validated using bootstrap sampling. The clinical utility of the model was comprehensively evaluated by decision curve analysis (DCA) and the area under the receiver operating characteristic curve (ROC). Among 91 patients (mean age 7.5 ± 4.3 years), 37.4% had a complicated course and 29.7% developed long-term sequelae. After adjustment for confounders, multivariate logistic regression analysis showed that total duration of fever (TDF) > 10.5d, CRP_Slope > -4.6 and Intermuscular Abscess were independent predictors of C-MRSA infection. The nomogram constructed based on this model showed good discrimination (AUC = 0.872), with excellent calibration (mean absolute error [MAE] = 0.036). DCA analysis verified that the model had significant net clinical benefit within the risk threshold range of 2%-92%. In a separate model model, time to normalization of ESR (TTEN) ≥ 1 month, CRP_Slope ≥ -3.18 and Physeal Involvement were independent predictors of U-MRSA infection (AUC = 0.918, MAE = 0.022). Both models constructed based on these key predictors showed high predictive performance, robust calibration, and significant net clinical benefit, underscoring their substantial clinical utility for risk stratification. A dual-risk prediction model integrating inflammatory dynamics and anatomical involvement demonstrates favorable predictive performance for complicated course and long-term sequelae in pediatric CA-MRSA infections, and may assist in early risk stratification and individualized management.