Abstract 1466: Obesity is Independently Associated With Long-term Risk of Cardiovascular Mortality and Underweight With Risk of Non-cardiovascular Mortality in Framingham Offspring Cohort
Background: Adiposity increases the risk of cardiovascular disease (CVD) mortality and morbidity, and its effect is mediated through other risk factors. Conversely, a U-shaped association has been observed between body mass index (BMI) and all-cause mortality.
Methods: We sought to investigate the incremental effect of baseline BMI on the long-term risk of CVD and non-CVD mortality adjusting for standard CVD risk factors (sex, age, systolic blood pressure and treatment, total and HDL cholesterol, smoking and diabetes) in 4609 Framingham Offspring Study participants (2370 women) aged 20 – 60 and free of CVD and cancer at initial exam in the early 1970s. On follow-up (median 33, max 36 years), there were 258 CVD (74 in women) and 770 non-CVD (306 in women) deaths (53% of latter from cancer). We used Fine and Gray competing risk model and restricted cubic splines to delineate the impact of BMI (modeled as continuous and categorical [underweight 30 kg/m2] variable) on the risk of CVD and non-CVD mortality.
Results: BMI was linearly (non-linearity p-value=0.97) and independently associated with CVD mortality (hazard ratio 1.22 per 5 unit increase, 95% CI: 1.05, 1.41, p-value=0.01). The association with non-CVD mortality was non-linear (non-linearity p-value=0.01): underweight status compared to normal BMI significantly increased the risk 1.44-fold (95% CI: 1.04, 1.98, p-value=0.03) but obese and overweight status did not. Compared to other BMI groups, the underweight had a higher proportion of smokers (53% vs. 45%, p-value=0.006) and a similar proportion of non-CVD deaths due to cancer (50% vs. 54%, p-value=0.64.) The relationship between BMI and CVD and non-CVD mortality was similar in women and men as well as in baseline smokers and non-smokers and interactions of BMI with sex and smoking status were not significant.
Conclusions: Our results suggest a potential explanation for the U-shape association of baseline BMI and all cause mortality adjusting for other CVD risk factors: with increasing BMI, we observe an increase in risk of CVD mortality, whereas in underweight individuals, non-CVD mortality risk is increased. Additional studies are warranted to confirm these observations.