Evolution of obesity prevalence in France. An age-period-cohort analysis

Abstract

We investigated the impact of birth cohort on the changes in obesity prevalence after taking into account age and survey period. We analyzed data from 4 national surveys conducted in 1997, 2000, 2003, and 2006. For each survey, self-reported data on weight and height were recorded on mailed questionnaires sent to a sample of 20,000 households, representative of the French population. We modelled the prevalence of obesity using logistic regression with age, cohort and period as explanatory variables. As these variables are linearly dependent, only nonlinear effects can be estimated uniquely and interpreted, after imposing specific chosen constraints in the models. There was a substantial depa..