Radiographic and inflammatory factors associated with fever-based clinical improvement in children with macrolide-resistant Mycoplasma pneumoniae pneumonia

Purpose Clinical improvement during initial macrolide-based management varies among children with macrolide-resistant Mycoplasma pneumoniae (MRMP) pneumonia. We evaluated the associations between inflammatory burden, chest radiographic phenotypes, and clinical improvement. Methods We conducted a retrospective multicenter cohort study at nine university hospitals in South Korea from 2012 to 2024. Eligible patients had PCR-confirmed M. pneumoniae infection and sequencing-confirmed macrolide resistance. Fever-based clinical improvement was defined as defervescence within 5 days of macrolide initiation before any rescue therapy. Multivariable logistic regression analysis was used to evaluate factors independently associated with clinical improvement. Results Among the 169 identified MRMP pneumonia cases, 130 were included, and 47 (36.2%) showed clinical improvement. The proportion of cases that showed clinical improvement decreased across radiographic phenotypes, from 60.0% with peribronchial infiltration to 10.0% with dense lobar consolidation or diffuse haziness. Peak C-reactive protein (CRP) levels of ≥ 6 mg/dL were independently associated with lower odds of improvement (adjusted odds ratio [aOR], 0.25; 95% confidence interval [CI], 0.07–0.88), as was dense lobar consolidation or diffuse haziness compared with peribronchial infiltration (aOR, 0.08; 95% CI, 0.01–0.73). When stratified jointly by CRP level and radiographic involvement, the proportion of cases with clinical improvement was 53.7% among those with CRP < 6 mg/dL and non-lobar involvement and 12.9% among those with CRP ≥ 6 mg/dL and lobar or diffuse involvement. Conclusions In sequencing-confirmed MRMP pneumonia, fever-based clinical improvement varied with inflammatory burden and radiographic phenotype. These exploratory findings may help inform structured reassessment during macrolide-based treatment but require prospective external validation.