Neighbourhood and metropolitan-level determinants of adult overweight and obesity in the United States

Enrico Marcelli, Harvard University
David Cutler, Harvard University
S.V. Subramanian, Harvard School of Public Health
Ichiro Kawachi, Harvard School of Public Health

No study has estimated whether individual behaviour and area-level socioeconomic factors influence height-adjusted body weight in the United States simultaneously and over time using conventional or multi-level econometric models. We employ 1993-2002 Behavioural Risk Factor Surveillance Survey, 1990 Census, and other data to estimate how individual characteristics and behaviours and metropolitan-level factors influenced weight among non-elderly adults. Controlling for climate, several individual-level characteristics were positively (non-Latino black, Latino, low income, age) or negatively (married, employed, female, Asian, college graduate) associated with weight; and ethno-racial and economic segregation and the density of fast-food restaurants positively, and the price of fast-food negatively, influenced weight – even after controlling for diet, exercise, and smoking. These three behaviours explain more than 15% of the variation in weight, but were also influenced by several area-level factors. Thus, while individual characteristics and behaviours influenced body weight, area-level factors both directly (physiologically) and indirectly (psychologically) did also.