Obesity as a causal risk factor for depression: Systematic review and meta-analysis of Mendelian Randomization studies and implications for population mental health

8 studies included.
Of the 23 studies that were retrieved for detailed inspection for eligibility, 15 were excluded because of the following reasons (Fig. 1): Two studies reported associations separately for depression (Walter et al., 2015a) and phobic anxiety (Walter et al., 2015b) from the same Nurses' Health Study, so we excluded the study of anxiety (Walter et al., 2015b) so as not to include the same data twice, and because we were primarily interested in depression as the outcome. One study (Samaan et al., 2015) was excluded because it only reported associations between the BMI genetic risk score and depression (OR = 1.01 per obesity-associated allele of a 21-SNP risk score, 95%CI = 0.99, 1.02) but not the gene–instrumented association between BMI and depression. Two studies from the UK Biobank cohort study were excluded: The first (Millard et al., 2019) reported associations between BMI and neuroticism-type personality characteristics separately by each questionnaire item (e.g., being nervous, a worrier, tense, and ‘suffering from nerves'). High neuroticism is related to depression (Hakulinen et al., 2015), but the study was excluded because of the use of single personality items as the outcomes. The second study (Wootton et al., 2018) examined how BMI was related to five domains of life satisfaction (work, health, finances, friends, and family) and single-item scale of happiness. This study was also excluded because of the use of single-item questions. Two studies (Kivimäki et al., 2011b; Lawlor et al., 2011) used only single genes (FTO or MC4R) as instruments for BMI, which limits the validity of the instrument, and these two studies were therefore excluded.