*2.5. Statistical Analysis*

Statistical analysis was performed with Stata17 for Windows [31]. The assessment of the pre-post changes in the quantitative measures was carried out through repeated measures analysis of variance (repeated-ANOVA), while the McNemar test was used for paired nominal data.

Additionally, the differences post-pre values were generated for the scores registered in the CIES, and for the weight (kg) and the BMI (kg/m2) (values equal to zero in these new variables indicate absence of change; negative values indicate a decreasing pre-post change; and positive values indicate an increasing pre-post change). Next, ANOVA procedures compared the mean differences between the diagnostic subtypes, the groups of age, and the continent.

In this study, the statistical analyses were adjusted by the sociodemographic sex and age, due to the differences between the groups (defined by the diagnostic subtypes and the origin of the samples).The effect size was estimated with Cohen's-h coefficient for the differences between the proportions and with Cohen's-d for the differences between the means (null effect size was considered for |h| < 0.20 or |d| < 0.20, low-poor for |h| > 0.20 or |d| > 0.20, moderate-medium for |h| > 0.50 or |d| > 0.50 and large-high for |h| > 0.80 or |d| > 0.80) [32,33]. Finner's method controlled the increase in the Type-I error due the use of multiple significance tests [34].
