Yu, Yan
Working paper no. 2008-18
Abstract
The majority of US adults are overweight or obese. Contemporary cohorts have been gaining weight continuously well up to the oldest ages. The negative health consequences of excess body weight have been well-documented. An intriguing question is whether these consequences decline over the life course, and excess body fat is no longer a significant mortality risk factor in old age. Previous studies concluded with a decline over age in the excess mortality of the overweight or obese, and call for weight guidelines that adjust for age. However, most prior research was based on comparing cross-sectional age groups that belong to a multitude of birth cohorts. In addition to the pitfall of cohort distortions, all previous studies used time since baseline as analysis time, leading to an age overlap in the comparison groups that makes it impossible to determine the age to which the weight disparities in mortality pertain. This study adopts an age- and cohort-specific framework to examine the age effect on the weight-mortality relationship for five cohorts born in the first half of the 20th century. To make weight disparities age-specific, age rather than time since baseline is used as analysis time. The analysis finds no decline in the long-term mortality consequences of baseline weight over the age of mortality followup. Mortality differentials tend to be stronger among later cohorts. When cohort differences are ignored, weight disparities in mortality may assume a rising or declining age pattern, and even if the age pattern is not distorted, estimates of mortality differentials may be biased towards nil.