*3.7. Covariates*

A very important issue in clinical trials is the impact of the covariates. In the studies about anorexia discussed in the present review, covariates such as age, BMI, leptin levels are usually recorded and reported, as part of the diagnosis, prognosis and follow-up criteria for the evaluation and management of the disease. Only a few studies included a detailed specific statistical analysis of the impact and association of covariates. Burdo et al. [49] further tested for several associations between nutrient levels and BMI, as well as eating-disorder symptoms, and they reported that vitamin B12 was negatively associated with BMI in AN. Traits of highly specific markers such as the epoxy-fatty acids and other oxylipins were considered together for ANOVA with other covariates such as age, BMI or anxiety [52,53]. Covariates such as BMI and psychiatric comorbidities were also considered [56], but only to evaluate their correlation with the metabolomics findings (individual fatty acids).
