Social
Heterogeneous association between social isolation and all-cause mortality among Japanese older adults Frances Rom Lunar* Frances Rom Lunar Naoki Kondo Yukiko Honda Atsushi Nakagomi Toshiaki Komura Kosuke Inoue Koichiro Shiba
Background: Social isolation has been linked to higher mortality; however, it remains unknown how this association varies across individuals and which social groups are vulnerable.
Methods: We used data from a nationwide cohort study of nearly 20,000 Japanese older adults with a 9·4-year follow-up period (Japan Gerontological Evaluation Study, 2013–2022), adjusting for 60 pre-baseline covariates. We linked the cohort data to a national insurance database to ascertain 4,299 all-cause deaths. Using a causal machine-learning method, we estimated the heterogeneous association between social isolation and mortality.
Results: Social isolation was more prevalent among those with lower income and educational attainment and associated on average with 65-day shorter survival time (restricted mean survival time [RMST] difference; 95% CI: -106, -24·2). There was evidence of heterogeneity such that social isolation was associated with even shorter survival days in some sub-groups (e.g., -214 days among the bottom quintile of the estimated RMST differences [95% CI: -332, -97·0]). These sub-groups tended to report lower educational attainment and pre-existing health problems. The adverse association of social isolation was particularly strong when lower educational attainment was combined with higher income, especially among women. Our simulation indicated that decreasing the prevalence of social isolation would reduce education- and income-based disparities in survival time.
Implications: Social isolation may be more harmful among those from socially disadvantaged backgrounds, and a combination of some characteristics can synergistically amplify its adverse effects on mortality. Population-level interventions reducing social isolation would not only improve health on average but also mitigate existing health disparities.
