*2.5. Statistical Analysis*

Differences in baseline characteristics between the two groups were examined by *t*-test ANOVA for continuous variables (means ± standard deviation [SD]) or Chi-square for categorical data (percentages). Statistical analyses of the metabolites were performed on log-transformed data that were scaled to SD units to facilitate comparisons across metabolites. The effect of the MedDiet intervention between point 1 and point 2 in this analysis (6 months) was assessed using a linear mixed-effects model adjusted for age and sex. As many of the metabolites are biologically correlated, applying a multiple testing correction using all the 225 biomarkers would be too strict. Thus, account for multiple testing, Bonferroni correction was applied with significance level defined as 0.05/21 = 0.002, with 21 being the number of principal components that explained 99% of the variation in the NMR data.

Cross-sectional linear associations between the adherence to the mMDS score and metabolite concentrations at both time points for both the MedDiet intervention and control groups were obtained using a linear regression model adjusted for age and sex. The analyses were performed using ggforestplot R package (version 0.0.2) and the linear mixed-effects model was fitted with the nlme R package (version 3.1.-144).
