Using relative BMI distributions to explore differences within and between gender, race/ethnic, and education subgroups, United States, 1998-2006

More than one-third of U.S. adults were obese in 2005-2006, including 33.3% of men and 35.3% of women. Overall the body mass index (BMI) distribution has shifted to the right, with a greater increase in the prevalence of those who are extremely obese. However, the processes affecting obesity disparities among groups are not well understood. While the BMI distribution contains a wealth of information, traditional analyses rely on capturing differences in averages between groups. This study uses a novel statistical approach, relative distribution methods, to explore inequalities and disparities in obesity prevalence both between and among gender and race/ ethnic subgroups. The relative distribution is effectively a transformation of two distributions into a single distribution that facilitates comparison. It is the set of percentile ranks that cases from one distribution would have if they were placed in another distribution. Data from 1998-2006 from the National Health and Nutrition Examination Survey (NHANES) are used to compare BMI distributions across time both within and among gender, race/ethnic, and education subgroups. Among men, preliminary results suggest a polarized BMI distribution between race/ethnic minorities compared to Whites. Among women, the BMI distribution is shifted to the right for race/ethnic minorities compared to Whites but distributions are similar in shape. Further analyses will decompose shape and location effects while adjusting for age and education covariates. A greater understanding of these differences can help to identify causal mechanisms explaining disproportionate obesity burden in populations, and to more effectively target health interventions to those at greatest risk.