Spss v 17.0

Data were reported as median with their interquartile range (IQR). Differences were investigated for significance with non-parametric tests, either with the Wilcoxon test when paired or the Mann-Whitney U test when independent. ROC analysis was performed for edema and LGE data comparing the acute phase to controls to set thresholds and evaluate their sensitivity and specificity. A diagnosis of myocarditis, based on a clinical suspicion of myocarditis represented by the presence of cardiac signs or symptoms or rise of cardiac biomarkers in the absence of other cardiac pathologies, and a CMR examination, positive according to the Lake Louise criteria, was used as ground truth for ROC analysis. Thresholds maximizing sensitivity and specificity (minimum total number of false positive and false negative cases) were chosen. Positive and negative likelihood ratios were calculated. The reproducibility of the diagnostic criterion was assessed through Bland-Altman analysis and reported as bias and its coefficient of repeatability (CoR). Concordance of the diagnostic criteria was assessed through Cohen’s κ.
Statistical analysis was performed using SPSS v.17.0 (IBM SPSS Inc., Chicago, IL, USA). Whenever a statistical test yielded a p-value < 0.05, the null hypothesis was rejected.