Does substance use explain social differences in terms of depression? Findings from the Constances cohort

International audienceBackground: The role of alcohol, tobacco and cannabis use in social differences in terms of depression is poorly understood.Method: We have applied mediation and moderated-mediation models stratified by gender to a population-based sample (N = 37,192) of French men and women from the Constances cohort with baseline and follow-up measures of depressive states. We have examined whether socioeconomic status (SES, measured by education and income) differences in the prevalence of depressive states may be explained by both differences in prevalence of substance use according to SES (mediating effects) and differential effects of substance use on depressive state according to SES (moderating effects).Results: In the mediation models, substance use only explained 5.3% and 2.4% of the association between low education and depressive state in men and women respectively, and was not a significant mediator for income. Moderated mediation models showed robust moderation effects of education and income in both men and women. The association of tobacco use with depressive symptoms, which was the only substance for which a mediation effect remained and for which the moderation effect of SES was the strongest, was significantly higher in participants with low SES.Limitations: The partially cross-sectional nature of the data restricts the possibility of drawing causality with regards to associations between SES and substance use.Conclusion: Targeting substance use, particularly tobacco, can especially reduce depression risk in individuals of low SES