Social health factors and dementia risk – assessing potential pathways in a population-based cohort study

Background Epidemiological studies suggest an association between social health factors and dementia risk, but potential pathways between these factors remain largely unexplored. Therefore, the current study assesses the relationship between social isolation, social activity engagement, and dementia risk, including potential effect mediation by level of loneliness. Method We used data from 7520 participants of the English Longitudinal Study of Ageing (ELSA), assessed from baseline (Wave 2, 2004/2005) until Wave 9 (2018/2019) for incident dementia. Level of social isolation was defined as the frequency of contact with children, other family and friends in addition to the membership in clubs and organizations. Social activities were inferred by the self-reported frequency of engagement in activities with a strong social component (e.g. volunteering). Both variables were dichotomized, representing low vs. high social isolation or social activity engagement. Level of loneliness was measured using the short form of the UCLA loneliness scale. Depression diagnoses were assessed through self-report. Separate multivariable Cox proportional hazard regression models were fitted, with social isolation and social activities as predictors while controlling for age, sex, education, wealth, cardiovascular risk factors (model 1) and, subsequently, for depression and level of loneliness (model 2). Result After a median follow-up period of 11.8 years (IQR = 5.9 years-13.9 years), 495 people (67.1 cases per 10,000 person-years; 95% CI 61.5-73.3) developed dementia. Higher social isolation (HR = 1.30; 95% CI 1.03-1.64) and more frequent social activity engagement (HR = 0.74; 95% CI 0.60-0.92) were related to dementia risk in model 1. Social isolation still predicted dementia risk in model 2 (HR = 1.30; 95% CI 1.02-1.66), while social activity engagement did not (HR = 0.83; 95% CI 0.65-1.07). Both depression and loneliness significantly predicted dementia risk in the model including social isolation, while, for the model including social activity engagement, only depression was associated with dementia risk. The change in HR between model 1 and 2 for social activity engagement was mainly due to adding level of loneliness rather than depression to the model. Conclusion Lower social isolation and more frequent social activity engagement are associated with lower dementia risk. Part of the association might be explained by level of loneliness and depression.