Cohort-based income gradients in obesity among U.S. adults

Objectives No studies have focused on socioeconomic disparities in obesity within and between cohorts. Our objectives were to examine income gradients in obesity between birth-cohorts (inter-cohort variations) and within each birth-cohort (intra-cohort variations) by gender and race/ethnicity. Methods Our sample includes 56,820 white and black adults from pooled, cross-sectional National Health and Nutrition Examination Surveys (1971-2012). We fit a series of logistic hierarchical Age-Period-Cohort models to control for the effects of age and period, simultaneously. Predicted probabilities of obesity by poverty-to-income ratio were estimated and graphed for 5-year cohort groups from 1901-1990. We also stratified this relationship for four gender and racial/ethnic subgroups. Results Obesity disparities due to income were weaker for post-World War I and II generations, specifically the mid-1920s and the mid-1940s to 1950s cohorts, than for other cohorts. In contrast, we found greater income gradients in obesity among cohorts from the 1930s to mid-1940s and mid-1960s to 1970s. Moreover, obesity disparities due to income across cohorts vary markedly by gender and race/ethnicity. White women with higher income consistently exhibited a lower likelihood of obesity than those with lower income since early 1900s cohorts; whereas, black men with higher income exhibited higher risks of obesity than those with lower income in most cohorts. Conclusions Our findings suggest that strategies that address race and/or gender inequalities in obesity should be cognizant of significant historical factors that may be unique to cohorts. Period-based approaches that ignore life-course experiences captured in significant cohort-based experiences may limit the utility of policies and interventions.