Background: The relation between body mass index (BMI) and mortality is not clear in the literature. An inverse relation between height and mortality has been suggested. We explore these relations in a very large cohort in Norway.
Methods: We studied two million men and women, age 20-74 years, who were measured during 1963-2000. These persons were followed for an average of 22.1 years. We used Cox proportional hazard models in the analyses. Also, the optimal BMI (the BMI at the time of measurement that was subsequently related to the lowest mortality) was estimated.
Results: Over the study period, 723,000 deaths were registered. The relative risk of death by BMI showed a J- or U-shaped curve, with the lowest rates of death at BMI between 22.5 and 25.0. In men, the optimal BMI increased from 21.6 when measured at age 20-29 to 24.0 when measured at age 70-74. In women, the optimal BMI was consistently higher, increasing from 22.2 to 25.7. Mortality decreased with increased height in men; in women, mortality decreased with height only up to heights of about 160-164 cm and then increased among the tallest women.
Conclusions: The relation between BMI and mortality was J- or U-shaped, with the "optimal" BMI varying by age and sex. Height was inversely related to mortality in men and in women up to a height of 165 cm.