*2.5. Statistical Analysis*

Group differences in cognitive performance were tested for by multivariate analysis of variance (MANOVA), including BMI as a covariate and then by univariate ANOVA models (also including BMI as a covariate). The effects of medication use and comorbid depression and anxiety disorders were subjected to stratified analyses in the FM group only, using MANOVA models comparing patients using and not using analgesics, anxiolytics, opioids, and antidepressants, as well as patients with and without depressive and anxiety disorders (with BMI as a covariate). The effect sizes are indicated by adjusted eta squared (η**<sup>2</sup> <sup>p</sup>**) values.

Associations between clinical variables and cognitive performance were only analyzed in the FM group. Firstly, to reduce the number of correlations performed (and thus limit type I error), we performed a multiple correlation analysis (the correlation coefficient (R) indicates the existence of an association, but not its direction (positive or negative), between the predictor variables (anxiety [STAI], depression [BDI], fatigue [FSS], insomnia [COS], and the four clinical pain variables [MPQ]) and each cognitive domain (selective attention and sustained attention [d2\_TR, d2\_CON, d2\_TOT of the d2 Attention test], divided attention [Condition 4 of the TMT], visuospatial memory [copying and memory conditions of the ROFC], verbal memory [RI\_AT, RL\_CP, RL\_LP of the TAVEC], and information processing speed [Conditions 2 and 3 of the TMT, and Conditions 1–4 of the 5DT]). Secondly, multiple regression analyses using the stepwise method were conducted, with BMI and the clinical variables as predictors and the cognitive parameters as dependent variables. The adjusted *R*<sup>2</sup> was used to evaluate the changes in predictions associated with each new block.
