Objective: To examine the predictive value of psychosocial factors as risk factors for all-cause mortality.
Design: A community-based longitudinal cohort study: The Honolulu Heart Program.
Setting: Population-based study conducted in Oahu, Hawaii.
Participants: Three thousand four hundred and ninety-seven men age 71 to 93 were examined and followed prospectively for all-cause mortality for an average of 6 years.
Measurements: Psychosocial data were obtained using the Lubben Social Networks Scale (LSNS). The LSNS consists of 10 items-family relationships (three items), relationships with friends (three items), and interdependent social supports and living arrangements (four items). We divided the LSNS score into quartiles for comparison, with the first quartile representing the lowest social support and the fourth quartile representing the highest social support.
Results: A significant dose-response relationship was noted with LSNS score and total mortality: 33.8% in the first quartile died over the follow-up period, 23.4% in the second, 18% in the third, and 15.7% in the fourth (P < .001). Six-year age-adjusted mortality rates were 66.2, 45.7, 37.8, and 33.7 per 1,000 person years in the first, second, third, and fourth, respectively (P < .001). Using age-adjusted Cox proportional hazards models, with the first quartile of LSNS as the reference group, relative risk for mortality was 0.69 (95% confidence interval (CI) = 0.58-0.82), 0.57 (95% CI = 0.47-0.70), and 0.52 (95% CI = 0.43-0.64) in the second, third, and fourth quartiles, respectively. Cox models were repeated, controlling for age and smoking status, and low LSNS scores remained significantly associated with higher mortality (P = .0001).
Conclusions: Our findings suggest that social networks were significantly independently associated with 6-year all-cause mortality in this cohort of older Japanese-American men. Social interventions in old age may reduce early mortality.