Religion and mortality among the community-dwelling elderly.

OBJECTIVES: This study analyzed the prospective association between attending religious services and all-cause mortality to determine whether the association is explainable by 6 confounding factors: demographics, health status, physical functioning, health habits, social functioning and support, and psychological state. METHODS: The association between self-reported religious attendance and subsequent mortality over 5 years for 1931 older residents of Marin County, California, was examined by proportional hazards regression. Interaction terms of religion with social support were used to explore whether other forms of social support could substitute for religion and diminish its protective effect. RESULTS: Persons who attended religious services had lower mortality than those who did not (age- and sex-adjusted relative hazard RH = 0.64; 95% confidence interval CI = 0.52, 0.78). Multivariate adjustment reduced this relationship only slightly (RH = 0.76; 95% CI = 0.62, 0.94), primarily by including physical functioning and social support. Contrary to hypothesis, religious attendance tended to be slightly more protective for those with high social support. CONCLUSIONS: Lower mortality rates for those who attend religious services are only partly explained by the 6 possible confounders listed above. Psychodynamic and other explanations need further investigation.