Clinical score for early escalation in pediatric A2063G Mycoplasma pneumoniae pneumonia: a retrospective cohort study

Abstract

Background: Macrolide-resistant Mycoplasma pneumoniae (MRMP), primarily driven by the 23 S rRNA A2063G mutation, is increasingly prevalent among East Asian children, diminishing azithromycin efficacy. Although some patients benefit from its anti-inflammatory properties, delayed escalation in non-responders can prolong fever and increase complications. Given the age-related risks of tetracyclines and fluoroquinolones, determining which children truly require second-line therapy remains a clinical challenge. Methods: We retrospectively reviewed 112 children with MRMP carrying the 23 S rRNA A2063G mutation. Patients were categorized into an azithromycin group (n = 66) and a second-line therapy group (n = 46). Between-group comparisons were performed using the χ² test, independent-sample t test, or Mann-Whitney U test. Independent predictors of escalation were identified via multivariable logistic regression, and model performance was assessed using receiver operating characteristic (ROC) analysis. Results: Compared with the second-line group, the azithromycin group had longer fever duration (median 7.00 vs. 5.00 days, P = 0.003) and slightly higher peak temperatures (39.20 °C vs. 39.00 °C, P = 0.016). In contrast, escalated patients exhibited significantly higher procalcitonin (PCT) levels (1.23 vs. 0.30 ng/mL, P < 0.001), greater chest CT total severity scores (TSS) (13.50 vs. 4.00, P < 0.001), and more frequent Streptococcus pneumoniae co-infection (65.22% vs. 39.39%, P = 0.012). Logistic regression identified elevated PCT, higher TSS, and ≥ 2 co-pathogens as independent predictors of escalation, while therapeutic bronchoscopy was protective; age was included as a covariate. The model demonstrated excellent discrimination (AUC 0.938, 95% CI 0.89–0.99; sensitivity 100%, specificity 78.8%). A five-item bedside score (cutoff ≥ 3.6) retained high accuracy (AUC 0.926; sensitivity 95.7%, specificity 86.4%). Conclusions: A simple clinical scoring model incorporating PCT, TSS, co-pathogen burden, bronchoscopy status, and age demonstrated good predictive accuracy in identifying children with A2063G-positive MRMP pneumonia who may benefit from early escalation to second-line therapy. Its use in clinical practice may support timely intervention for high-risk patients while minimizing unnecessary antibiotic escalation. Further prospective validation in multicenter cohorts is needed to confirm its generalizability and clinical utility.