Social support and mortality: Analysis of a national sample

While lack of social support is associated with increased odds of mortality in select populations, no studies have investigated this in a nationally representative sample. In addition, few studies have investigated associations between social support and specific causes of mortality. This study sought to address these issues. Our sample comes from the nationally weighted National Health and Nutrition Examination Survey III (1988-1994) and the National Death Index, and includes adult, non-institutionalized civilians ages 17 and older (N=15,289). Each person was assigned one of five modified Berkman social network index scores: 0 (lowest social support), 1 (low), 2 (moderate), 3 (high), and 4 (highest). Our outcome variables were all-cause and cause-specific mortality by the year 2000. There was evidence of a social support gradient, with those with lower social support having significantly higher odds of mortality for several mortality types: all-cause (lowest vs. highest social support odds ratio [OR] 2.76, 95% confidence interval [CI] 2.00-3.81; low vs. highest OR 1.64, CI 1.29-2.07), cardiac (lowest vs. highest OR 2.67, CI 1.51-4.73; low vs. highest OR 1.78, CI 1.16-2.75), and respiratory (lowest vs. highest OR 7.78, CI 2.86-21.18; intermediate ORs were smaller but all significant at the p<.05 level). These gradients remained for all three types of mortality when controlling for demographic variables, and remained for all-cause and respiratory mortality when controlling for demographic, socioeconomic, and health variables. There were no significant differences in odds of mortality by social support for cancer, neurodegenerative, accidents, or self-harm mortality.