Aging
Body mass index at different late-life stages and risk of all-cause mortality in a diverse population with exceptional longevity: LifeAfter90 Study Claire Meunier* Claire Meunier Meunier Meunier Meunier Meunier Meunier Meunier University of California, Davis
The association between late-life obesity and all-cause mortality varies across older age. Among the younger-old populations, obesity is associated with a higher mortality risk. However, studies among oldest-old have reported null or protective effects, with little known about diverse populations. We examined the association of BMI measured at different stages of late-life and all-cause mortality in 849 participants who lived until 90 years in the LifeAfter90 study. Early late-life BMI (mean age 79.9; range 73.8-95.1) was defined as the first clinically measured BMI obtained at age 65 from electronic health records beginning in 2005. Late late-life BMI (mean age 91.7; range 89.2-104.6) was defined as the BMI measure closest to study baseline. Time to mortality, which was ascertained from the EHR, was calculated as years since baseline. Cox proportional hazards models estimated the association between BMI and mortality adjusting for demographic variables. BMI distributions differed by timing, early late-life: 30% normal, 46% overweight, 25% obese; late late-life: 48% normal, 39% overweight, 13% obese. During six years of follow-up, 53% of participants died. In models accounting for demographics, age at BMI, and baseline age, obesity in early late-life was associated with a higher hazard of mortality after age 90 compared with normal BMI (HR(95% CI):1.45(1.12,1.88)). Obesity measured in late late-life was not associated with mortality, though point estimates suggest a lower risk of mortality (HR:0.87(0.64,1.18)). For both late-life time points, overweight BMI was not associated with mortality (Figure 1). In this diverse cohort of the oldest-old, obesity in early late-life was associated with increased mortality risk, whereas obesity in late late-life was not. The reversal of point estimates later in late life may reflect age specific effects of obesity, selective survival, or reverse causality due to illness-related weight loss prior to death.
