Background & aims
The relationship between vitamin D and common mental disorders (CMDs) remains unclear. We aimed to determine if behaviours affecting vitamin D concentrations differ between individuals with or without CMDs and evaluate, cross-sectionally and prospectively, the extent to which the association between 25(OH)D and CMDs are explained by these behaviours.

Methods
Data are from the 1958 British birth cohort (n = 7401). Behaviours were ascertained by questionnaire at age 45 years. CMDs (depression, anxiety, panic, phobia) were assessed using the Clinical Interview Schedule-Revised at 45 years and depression using Mental Health Inventory-5 at 50 years.

Results
Participants with CMDs at 45 years differed from others on some but not all vitamin D related behaviours. There were inverse, cross-sectional associations at 45 years of 25(OH)D with depression and panic, which persisted after adjustment for vitamin D related behaviours (OR = 0.57, 95% CI: 0.40,0.81 and OR = 0.33, 95% CI: 0.40,0.81, respectively). Association between 25(OH)D and subsequent (50 years) risk of depression was non-linear (p = 0.01), with lower risk for participants with 25(OH)D between 50 and 85 nmol/l compared with those with lower or higher concentrations.

Conclusion
This study provides support for an association of low 25(OH)D concentrations with current and subsequent risk of depression in mid-adulthood.