Suicides, homicides, accidents, and other external causes of death among blacks and whites in the Southern Community Cohort Study.

Abstract

Prior studies of risk factors associated with external causes of death have been limited in the number of covariates investigated and external causes examined. Herein, associations between numerous demographic, lifestyle, and health-related factors and the major causes of external mortality, such as suicide, homicide, and accident, were assessed prospectively among 73,422 black and white participants in the Southern Community Cohort Study (SCCS). Hazard ratios (HR) and 95% confidence intervals (CI) were calculated in multivariate regression analyses using the Cox proportional hazards model. Men compared with women (HR = 2.32; 95% CI: 1.87-2.89), current smokers (HR = 1.74; 95% CI: 1.40-2.17), and unemployed/never employed participants at the time of enrollment (HR = 1.67; 95% CI 1.38-2.02) had increased risk of dying from all external causes, with similarly elevated HRs for suicide, homicide, and accidental death among both blacks and whites. Blacks compared with whites had lower risk of accidental death (HR = 0.46; 95% CI: 0.38-0.57) and suicide (HR = 0.55; 95% CI: 0.31-0.99). Blacks and whites in the SCCS had comparable risks of homicide death (HR = 1.05; 95% CI: 0.63-1.76); however, whites in the SCCS had unusually high homicide rates compared with all whites who were resident in the 12 SCCS states, while black SCCS participants had homicide rates similar to those of all blacks residing in the SCCS states. Depression was the strongest risk factor for suicide, while being married was protective against death from homicide in both races. Being overweight/obese at enrollment was associated with reduced risks in all external causes of death, and the number of comorbid conditions was a risk factor for iatrogenic deaths. Most risk factors identified in earlier studies of external causes of death were confirmed in the SCCS cohort, in spite of the low SES of SCCS participants. Results from other epidemiologic cohorts are needed to confirm the novel findings identified in this study.