Mortality, health services use, and health behavior in a cohort of well older adults

OBJECTIVES: To construct an evidence-based Good Health Behavior Score and examine the relationship between aggregate health behaviors, mortality, and health services utilization in the last year of life in a cohort of well older adults. DESIGN: A prospective cohort. SETTING: A large health maintenance organization. PARTICIPANTS: 1867 older enrollees who responded to a health promotion survey. MEASUREMENTS: A baseline self-administered questionnaire was used to ascertain health behaviors in 1987-1988, and vital status was determined 48 months later. A Good Health Behavior Score was calculated, and a Cox proportional hazards model was used to compare high, middle, and low score groups regarding risk of death. For those who died, differences in amount and type of health services utilization in relation to the summary score were compared for the year before death. RESULTS: During the 4 years of follow-up, the mortality rate for the mid-level score group was 50% less, and in the highest score group was 70% less, than in the lowest score group. Among decedents, no significant differences were found between high and low Good Health Behavior score groups for inpatient and outpatient utilization, pharmacy use, or total cost during the last year of life. CONCLUSION: An easily developed and simple health behavior score can predict short term mortality quite strongly. Medical care costs in the last year of life were similar in individuals with higher and lower health behavior scores.