An ecologic analysis of socioeconomic and environmental determinants of overweight and obesity prevalence among US adolescents using Bayesian small area estimation with 2003 and 2007 National Survey of Children's Health Público

Abstract

Objective: Childhood obesity continues to be a growing public health concern with 12.5 million (17%) school-age children and teens affected. We aim to predict county-level overweight and obesity prevalence among U.S. adolescents with Bayesian small area estimation. Using data from the 2003 and 2007 National Survey of Children's Health, we will explore the association of adolescent overweight and obesity prevalence with sociodemographic and physical environment characteristics.

Methods: We used a Bayesian hierarchical Poisson model based on the Markov chain Monte Carlo (MCMC) approach to estimate the county-level prevalence of childhood obesity nationwide. Negative binomial Poisson regression models will be used to perform an ecologic analysis of social and environmental determinants of health on adolescent overweight and obesity prevalence. Both the small area estimation and the county-level contextual analysis were performed for three excess weight prevalence outcomes: overweight including obesity, overweight alone, and obesity alone.

Results: Using small area estimation, 30.7% of US adolescents are overweight or obese. The South region and East South Central Census division have the highest prevalence. The geographic disparities differ when examining overweight and obesity separately. An increased risk of excess weight among adolescents is associated with certain county-level indicators of poor adult health, poor socioeconomic environment such as child poverty rate and high unemployment rates, home and school food assistance and racial composition.

Conclusions: The model-based childhood obesity prevalence estimates displayed a geographic pattern similar to that of adult obesity prevalence. Higher prevalence was predominantly seen in the southeast and in counties with indicators of poor social and physical environment and increased food assistance. The prevalence varies across and within states and contextually by county suggesting an opportunity for intervention at the local, state and national level. These results suggest that childhood obesity is a complex and multi-faceted issue and may be influenced by the combined effects of the social, economic and physical environments.

Table of Contents

TABLE OF CONTENTS

1. List of Tables...i
2. List of Figures...ii
3. List of Abbreviations...iii
4. Chapter 1: Background/Literature Review...1
5. Chapter 2.1: Manuscript 1 - Small area Bayesian model-based estimation of overweight and obesity prevalence among US adolescents (10-17 years) by county...19
6. Chapter 2.2: Manuscript 2 - Socioeconomic and environmental correlates of county-level (small area) overweight and obesity prevalence among US adolescent...33
7. Tables...46
8. Figures...57
9. Chapter 3: Summary, Public Health Implications, Possible Future Directions...65
10. Appendix...77
11. References...88

LIST OF TABLES

Table 1. Basic descriptive statistics of state-level, public use and county-level, restricted use 2003 and 2007 NSCH data (n, % [unweighted]) for US children aged 10-17 years...46
Table 2. Basic descriptive statistics for model-based estimates for overweight and obesity prevalence among 10-17 year old US children...47
Table 3. Basic descriptive statistics of county-level contextual variables...48
Table 4. Model-based prevalence estimates and basic statistics for county-level covariates for counties that are below the national overweight and obesity means (posterior mean and 95% credible intervals exclude the national mean using 2003 and 2007 NSCH)...49
Table 5. Model-based prevalence estimates and basic statistics for county-level covariates for counties that are above the national overweight and obesity means (posterior mean and 95% credible intervals exclude the national mean using 2003 and 2007 NSCH)...51
Table 6. Model fit comparison assessing for best fit when considering all non-collinear covariates, geography (rurality and Census-based region) and population (normal and log transformed total population and population density...53
Table 7. Negative binomial regression model using standardized covariates (following tests of correlation and collinearity) for Overweight including obesity prevalence among US adolescents...54
Table 8. Negative binomial regression model using standardized covariates (following tests of correlation and collinearity) for Overweight alone prevalence among US adolescents...55
Table 9. Negative binomial regression model using standardized covariates (following tests of correlation and collinearity) for Obesity alone prevalence among US adolescents...56

LIST OF FIGURES

Figure 1. Model-based predicted estimates of Overweight including obesity prevalence among US children (10-17 years) for US counties, using 2003 and 2007 NSCH data...57
Figure 2. Model-based predicted estimates of Overweight alone prevalence among US children (10-17 years) for US counties, using 2003 and 2007 NSCH data...57
Figure 3. Model-based predicted estimates of Obesity alone prevalence among US children (10-17 years) for US counties, using 2003 and 2007 NSCH data...58
Figure 4. Model-based predicted estimates of Overweight including obesity prevalence among US adolescents (10-17 years) for US counties below and above the national mean (posterior mean and 95% credible interval excludes the national mean...58
Figure 5. Model-based predicted estimates of Overweight alone prevalence among US adolescents (10-17 years) for US counties below and above the national mean (posterior mean and 95% credible interval excludes the national mean...59
Figure 6. Model-based predicted estimates of Obesity alone prevalence among US adolescents (10-17 years) for US counties below and above the national mean (posterior mean and 95% credible interval excludes the national mean...59
Figure 7. Model-based predicted estimates of Overweight including obesity prevalence among US adolescents (10-17 years) for US counties in the lowest and highest decile of the mean...60
Figure 8. Model-based predicted estimates of Overweight alone prevalence among US adolescents (10-17 years) for US counties in the lowest and highest decile of the mean...60
Figure 9. Model-based predicted estimates of Obesity alone prevalence among US adolescents (10-17 years) for US counties in the lowest and highest decile of the mean...61
Figure 10. Sensitivity results for model-based Overweight including obesity prevalence estimates among US children (10-17 years) for all US counties...62
Figure 11. Sensitivity results for model-based Overweight alone prevalence estimates among US children (10-17 years) for all US counties...63
Figure 12. Sensitivity results for model-based Obesity alone prevalence estimates among US children (10-17 years) for all US counties...64

About this Master's Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Rollins School of Public Health
Department	Epidemiology
Degree	MPH
Submission	Master's Thesis
Language	English
Research Field	Health Sciences, Public Health
Palavra-chave	small area estimation obesogenic environment childhood obesity county-level prevalence
Committee Chair / Thesis Advisor	Kramer, Michael R, Emory University
Partnering Agencies	Emory University schools, faculty or affiliated programs CDC

Última modificação

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	An ecologic analysis of socioeconomic and environmental determinants of overweight and obesity prevalence among US adolescents using Bayesian small area estimation with 2003 and 2007 National Survey of Children's Health ()	2018-08-28 14:52:49 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions