Project:
An Age-Period-Cohort Analysis of the Rise in the Prevalence of the U.S. Population Overweight and/or Obese (PO&O)

Year: 2004

Investigator: Daponte, Beth Osborne, and Andrew Cook

Institution: Yale University

Project Contact:
Beth Osborne Daponte
Institution for Social and Policy Studies
Yale University
P.O. Box 208209
New Haven, CT 06520-8209
Phone: 203-432-6141
E-mail: beth.daponte@yale.edu

This research examines trends in the propensity of obesity and overweight (PO&O) in the United States from 1976 to 2001. An age-period-cohort analysis examined:

1. The life course trajectory of weight gain
2. The years in which Americans displayed the greatest degree of weight gain
3. Whether trends differ by birth cohort.

The analyses use data from the National Health Interview Survey (NHIS), a nationally representative annual health survey conducted by the National Center for Health Statistics. The NHIS data are adjusted for the misreporting of weight and height. The analyses use demographic methods, specifically age-period-cohort analysis, and ordered logit models.

The ordered logit models show that the probability of being obese versus overweight, and overweight versus normal weight has increased in every 5-year period, after controlling for a host of factors. Between 1976 and 2001, the probability of the adult population being obese has tripled. The probabilities of obesity for women are now nearing those of men. Of the race/sex groups studied, Black and Hispanic females possess the highest probabilities of being obese, with Black women having a higher probability. Hispanic men have the highest probability of being obese. When the data are disaggregated by year, the results show the youngest age groups are increasing their likelihood of being obese at the fastest rates.

Findings indicate that for every age group and for every birth cohort, current group average Body Mass Indexes (BMIs) are larger than the previous period’s BMI. Altogether, the adult population in the United States has increased its BMI every 5-year period between 1976 and 2001. Further, the growth in PO&O seems to be accelerating over time. Taking into account the characteristics of the population does not mitigate the effect of period on Americans’ increased weight gain. One limitation of the data is that the sample contained adults only. It is not known how many individuals entered the sample already overweight or obese. Because every birth cohort and age group showed an increase in BMI, explanations that focus on specific subpopulations were not supported by the data. Initially, it was hypothesized that a particular time period, age group, or cohort could be isolated as an indicator of BMI increase. If that had been the case, one could target policy interventions and design appropriate policy interventions to mitigate the situation. The findings demonstrate that this was not the case and that period effects dominate: the PO&O increases with every time period, without exception.

This study provides some important insights into the age-period-cohort effects of PO&O. Future research might concentrate on the aspects of American culture that result in eating behavior that yields PO&O, how the American lifestyle might be adapted to expend more energy, and how public policies might arrest the PO&O trend in the United States.