Objectives: To assess the relationship between body mass index (BMI; kg/m2) and mortality in a large nationally representative sample of US adults over age 70 years.
Design: Prospective longitudinal cohort study, the Longitudinal Study of Aging (LSOA). Subjects were all those 7260 black and white people (2769 men, 4491 women) initially interviewed in 1984 for whom height and weight were available. These subjects were followed through to 1990.
Measurements: Measurements included self-reported height and weight, date of death if subjects died, sex, age, race, measures of socio-economic status, number of living first degree relatives, and responses to questions asking whether the subject had retired due to poor health, had difficulty eating, worried about their health, and felt their health was worse than during the prior year. Smoking status was not assessed.
Results: When analyzed via Cox proportional hazard regression, the relationship between BMI and mortality, represented by means of hazard ratio, was clearly U-shaped for both men and women. The base of the curves was fairly wide suggesting that a broad range of BMIs are well tolerated by older adults. The minimum mortality (estimated from the fitted proportional hazard models) occurred at a BMI of approximately 31.7 for women and 28.8 for men. The results were essentially unchanged, if analyses were weighted, if various disease states were controlled for, and if apparently unhealthy subjects were excluded.
Conclusions: The finding of the relatively high BMI (27-30 for men, 30-35 for women) associated with minimum hazard in persons older than seventy years supports some previously documented findings and opposes others and, if confirmed in future research, has implications for public health and clinical recommendations.