Changes in social participation between ages 65 and 70 and incident dementia: a Bayesian competing risk analysis

Social participation may protect against dementia, but few studies have examined whether cessation of social participation during the period surrounding retirement is associated with subsequent dementia risk accounting for the competing risk of death. This study examined whether changes in social participation between ages 65 and 70 are associated with incident dementia in community-dwelling Japanese older adults.

This prospective cohort study analysed 1,681 participants of the NISSIN Project (Nisshin City, Aichi Prefecture, Japan) enrolled at age 65 (1996-2005), re-examined at age 70, and followed for incident dementia through the Japanese National Long-Term Care Insurance scheme until 2023. Social participation was assessed using 10 items at ages 65 and 70. Participants were classified into four groups: Continuation, Cessation, Initiation, and Persistent nonparticipation. Bayesian cause-specific hazard models were used to estimate hazard ratios for incident dementia, treating death without dementia as a competing event. Sensitivity analyses addressed reverse causation and model assumptions.

During 22,116 person-years of follow-up (median 13.6 years), 497 participants developed dementia and 138 died without dementia. In the fully adjusted model, Cessation was associated with a higher hazard of dementia than Continuation (HR 1.27, 95% credible interval [CrI] 1.02 to 1.64; posterior probability of HR > 1, 0.98). This association remained after lag exclusion and adjustment for cognitive function at age 65. Neither Initiation nor Persistent nonparticipation was associated with dementia.

Cessation of social participation between ages 65 and 70 was associated with increased dementia incidence. Recent cessation of social activities may help identify older adults who could benefit from preventive support during the period surrounding retirement. Limitations include a single-suburb sample, which may limit generalisability, and identification of dementia through long-term care insurance certification rather than clinical diagnosis.