Statistical analysis
Descriptive statistics were employed to determine numerical and categorical variables. Numerical variables were presented using measures of central tendency and dispersion (mean, median, standard deviation (SD) and interquartile range (IQR), as appropriate). Absolute and relative frequencies were presented for categorical variables. Overall interrater agreement, among three judges, was calculated using Fleiss’ kappa coefficient.[10] Cohen’s kappa coefficient with linear weighting[11] was used to measure pairwise judge concordance. For both kappa coefficient measures, all cases in which ADR were classified as unassessable by at least one judge were excluded. We considered 95% of confidence interval for all the analyses. The assessment of concordance was based on the quantitative scale proposed by Landis and Koch[12]: <0=poor, 0.00-0.20=slight, 0.21-0.40=fair, 0.41-0.60=moderated, 0.61-0.80=substantial, 0.81-0.99=almost perfect and 1.00=perfect.