Fulminant myocarditis (FM) in children is a severe, rapidly progressive form of myocardial inflammation with a high risk of mortality and long-term cardiac sequelae. Early manifestations are frequently nonspecific, and the emergency department (ED) is often the first point of care. This single-center retrospective study aimed to identify early warning indicators available at ED presentation that are associated with FM, to support rapid recognition and escalation of care for children at high risk.
The study reviewed records of children aged 28 days to 16 years admitted through the ED of Shenzhen Children’s Hospital from 01/01/2015 to 31/12/2024 with a primary discharge diagnosis of acute myocarditis. After screening with ICD-10 codes and independent case review by pediatric emergency and cardiology specialists, 140 children met inclusion criteria: 86 with acute non-fulminant myocarditis (ANFM) and 54 with FM. The institutional Ethics Committee approved the protocol and waived informed consent for this retrospective analysis.
ANFM was defined according to pediatric-adapted criteria based on the 2013 European Society of Cardiology (ESC) position statement for clinically suspected myocarditis, requiring compatible clinical manifestations plus at least one diagnostic criterion from electrocardiography, elevated myocardial injury biomarkers (cTnI), or structural/functional imaging abnormalities. FM was defined using international guideline–aligned criteria as acute myocarditis with rapid hemodynamic deterioration within two weeks—manifesting as cardiogenic shock or life-threatening arrhythmia requiring vasoactive support or mechanical circulatory support. Cases with congenital heart disease, cardiomyopathy, coronary or valvular disease, hyperthyroidism, inherited metabolic disorders, non-myocarditis causes of cardiac dysfunction, symptom onset >30 days, or substantial prior treatment at another institution were excluded.
Variables were selected for clinical relevance and availability at the initial medical encounter. For direct ED presentations the earliest ED data were used; for transfers, the referring hospital's pre-transfer data were prioritized. Collected variables included demographics, duration of illness, categorized clinical manifestations (respiratory, gastrointestinal, cardiovascular, neurological), fever, hepatomegaly, laboratory tests (WBC, CRP, serum sodium and potassium, lactate, myocardial injury markers including cardiac troponin I [cTnI]), electrocardiography findings, chest radiography, and echocardiography when available. Missing data were limited (<5%) and imputed by median or mode as appropriate.
Univariable logistic regression identified candidate predictors (P < 0.10). To reduce multicollinearity and overfitting when considering multiple candidate variables, the study used least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation, performed both with and without left ventricular ejection fraction (LVEF) because echocardiography was not always completed at ED presentation. Variables selected by LASSO were entered into multivariable logistic regression constrained by events-per-variable considerations. Indicators with P < 0.05 in multivariable analysis were considered independently associated with FM. Discriminative performance was assessed using ROC curves and AUC, calibration via calibration curves and the Hosmer–Lemeshow test, and internal validation by bootstrap resampling (1,000 iterations) to obtain optimism-corrected AUC values.
Among children classified as FM, gastrointestinal symptoms were the most frequent presenting manifestation (88.9%). Compared with the ANFM group, children with FM presented with statistically lower serum sodium and higher lactate and myocardial injury marker levels (all P < 0.05). Specific counts were 54 FM and 86 ANFM patients in the cohort of 140.
After LASSO selection and multivariable logistic regression, five ED-available indicators were independently associated with FM:
These variables were all obtainable at or shortly after ED presentation and, in combination, may help identify children at higher risk of rapid deterioration from myocarditis.
The study evaluated discriminative ability using ROC curves and reported internal validation using bootstrap resampling to correct optimism in AUC estimates. Calibration was assessed by calibration curves and the Hosmer–Lemeshow test. Specific AUC and calibration statistics are reported in the full article; the authors used these assessments to support the predictive utility of the selected ED indicators.
The findings suggest that simple ED-available findings—particularly gastrointestinal symptoms, elevated lactate and cTnI, presence of third-degree atrioventricular block, and lower serum sodium—should raise suspicion for FM in children presenting with possible myocarditis. Timely recognition could prompt earlier hemodynamic monitoring, intensive care transfer, escalation to vasoactive or mechanical circulatory support, and specialist cardiology involvement.
This was a single-center retrospective study with diagnostic confirmation based primarily on clinical criteria; cardiac magnetic resonance and endomyocardial biopsy were not routinely performed. The sample size limited the number of variables in multivariable models, and findings require prospective multicenter validation before routine clinical implementation. The article reports these limitations and calls for larger, prospective studies to confirm and refine ED-based warning indicators.
In this cohort of 140 children with acute myocarditis, five ED-available indicators—gastrointestinal symptoms, elevated lactate, elevated cTnI, third-degree atrioventricular block, and lower serum sodium—were independently associated with fulminant myocarditis. These findings may assist emergency clinicians in early risk stratification, but prospective multicenter validation is needed as recommended by the authors.