Background Physical activity (PA) is cross-sectionally associated with lower prevalence of depression and other indicators of psychological well-being. Few longitudinal studies examine the possibility of bi-directional associations between activity and psychological health.
Using data from the 1958 British Birth cohort (maximum N = 14,072) over four decades of life we examined (i) cross-sectional associations to establish whether psychological ill-health is concurrent with lack of PA at several life-stages, and (ii) longitudinal associations to elucidate the direction of association, i.e. (1) psychological health and prior PA, and (2) PA and prior psychological health.
Methods Psychological health was assessed using validated, age appropriate instruments from questionnaires to teachers (11y, 16y) and participants (23y, 33y, 42y, 50y). Measures were converted to (i) standard deviation (SD) scores and (ii) categorised to represent “severe distress” (top 10% at each age). PA reported by mothers (11y) and participants (16y, 23y, 33y, 42y, 50y), was classified at all ages into four categories. Covariates from different life-stages included: body mass index, socioeconomic position, education and physical limiting illness.
Results Using linear regressions, we found that individuals with poorer psychological health were less physically active, but patterns of association varied by age. In adolescence, there was a dose-response relationship: e.g. at 16y, per category higher activity, psychological health was lower by 0.08 (0.07, 0.10) SDs. In adulthood, there was a threshold effect: e.g. at 50y, those active 1+/wk had a 0.20 (0.15, 0.24) lower SD score than those active <1/wk. Associations persisted after adjustment, although attenuated. We assessed directionality using random effects models to (1) psychological health (16-50y) with time-varying PA as an explanatory variable, and (2) PA (16-50y) with time-varying psychological health as an explanatory variable. Preliminary analysis suggests that relationships are bi-directional. The odds ratio of severe distress was 0.81 (95% CI 0.75, 0.89) for those physically active at the previous sweep compared to the inactive. However, there was a suggestion that the relationship weakened with age. Prior psychological health predicted PA similarly across ages. Between 16-50y, the odds of being physically active if severely distressed previously was 0.67 (0.62, 0.73). Results were unchanged after adjustment.
Conclusion PA and psychological ill-health are concurrent over decades of life. The relationship appears to be bi-directional, although weakening with age for PA to subsequent psychological ill-health. Findings suggest that PA throughout adolescence and young adulthood may alleviate subsequent psychological ill-health.