A MULTILEVEL STUDY OF EFFECTS OF SOCIOECONOMIC STATUS, INCOME INEQUALITY, AND THE BUILT ENVIRONMENT ON ADULT OBESITY IN CHINA BY LIBIN ZHANG DISSERTATION Submitted in partial fulfillment of the requirements for the degree of doctor of philosophy in Sociology in the Graduate College of the University of Illinois at Urbana-Champaign, 2012 Urbana, Illinois Doctoral Committee: Professor Tim Liao, Chair Associate Professor Margaret Kelley Assistant Professor Ilana Redstone Akresh Assistant Professor Kathy Baylis
294
Embed
a multilevel study of effects of socioeconomic status, income - Ideals
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Guizhou 0.5563 22.36 3.32 17.90 1,078 Note: In the following tables that report results from bivariate or multivariate analysis, the Gini coefficient is replaced by the Gini index (i.e. Gini coefficient 100) so that interpretation of the result can be based on each unit of change of the Gini index (i.e. each 0.01 change in the Gin icoefficient).
254
Table 10 Bivariate analysis of BMI vs. obesity outcome (N=9,586) BMI Obesity (Yes=1)
Coefficient SE Odds Ratio SE
Level-1 Variables: Demographic Control
Female 0.093 0.074 1.061 0.049 Age 0.015*** 0.002 1.011*** 0.002 Married 1.089*** 0.099 1.735*** 0.120
Level-2 Variables: Mean Income (in 1000 yuan) 0.030*** 0.005 1.020*** 0.003 Mean education (in years) 0.141*** 0.017 1.079*** 0.011 Urbanicity Index 0.014*** 0.002 1.009*** 0.001 Urban (ref.=Rural ) 0.525*** 0.080 1.319*** 0.065 Gini Coefficient *100 -0.085*** 0.005 0.956*** 0.003 Jiangsu -0.157 0.117 0.996 0.072 Guangxi -1.681*** 0.112 0.338*** 0.032
Notes: *Significance at 5% level; ** significance at 1% level; *** significance at 0.1% level
255
Table 11 Estimates from random intercept multilevel models of BMI (N=9,586)
Model 1 Model 2 Model 3 Model 4 Variables Coefficient SE Coefficient SE Coefficient SE Coefficient SE Fixed effects Intercept 18.332*** 0.339 18.193*** 0.377 18.367*** 0.443 18.025*** 0.629 Level-1 Variables: Demographic Control
Relative Income (RDI) -0.302 0.404 -0.268 0.518 Behavioral Control
Smoker Alcohol Consumption
Level-2 Variables: Community Characteristics
Mean Income (in 1000 yuan) -0.011 0.015
Mean Education (in years) 0.067 0.054 Urbanicity Index 0.000 0.006 Urban (ref.=Rural ) 0.106 0.215 Gini Coefficient*100
Province Indicator Jiangsu Guangxi
Cross-level Intearaction Gini*Low Income (Q2) Gini*Middle Income (Q3) Gini*High Income (Q4) Gini*Top Income (Q5) Gini*Middle Education Gini*High Education
Random-effects Parameters Random Intercept 1.056 (0.063) 0.979 (0.060) 0.978 (0.060) 0.971 (0.060) Level-1 Residual 3.417 (0.025) 3.403 (0.025) 3.403 (0.025) 3.403 (0.025) Model Fit Statistics (BIC) 51,175.360 51,193.960 51,202.570 51,236.030
256
Table 11 (Cont.)
Model 5 Model 6 Model 7 Model 8 Coefficient SE Coefficient SE Coefficient SE Coefficient SE Fixed effects Intercept 22.516*** 0.762 23.930*** 0.759 23.681*** 0.950 23.948*** 0.950 Level-1 Variables: Demographic Control
Random-effects Parameters Random Intercept 0.566 (0.051) 0.566 (0.051) 0.565 (0.051) 0.566 (0.051) Level-1 Residual 3.404 (0.025) 3.404 (0.025) 3.402 (0.025) 3.396 (0.025) Model Fit Statistics (BIC) 51173.050 51,110.050 51,157.590 51,143.450
Notes: Number of observation = 9586; Number of community = 218. Numbers are coefficients. Random-effects parameters have standard errors in parentheses. *Significance at 5% level; ** significance at 1% level; *** significance at 0.1% level
257
Table 12 Odds ratio estimates from random intercept multilevel logistic models of obesity (N=9,586)
Model 1 Model 2 Model 3 Model 4 Fixed effects
OR SE OR SE OR SE OR SE Level-1 Variables: Demographic Control
Relative Income (RDI) 0.607 0.158 0.611 0.212 Behavioral Control
Smoker Alcohol Consumption
Level-2 Variables: Community Characteristics
Mean Income (in 1000 yuan) 0.996 0.009 Mean education (in years) 1.015 0.033 Urbanicity Index 1.002 0.004 Urban (ref.=Rural ) 1.005 0.129 Gini Coefficient *100
Province Indicator Jiangsu Guangxi
Cross-level Intearactions Gini*Low Income (Q2) Gini*Middle Income (Q3) Gini*High Income (Q4)
Notes: Number of observation = 9586; Number of community = 218.
Numbers are odds ratios. ICC and Level 2 variances have standard errors in parentheses. *Significance at 5% level; ** significance at 1% level; *** significance at 0.1% level
261
Table 13 Dimensions of the built environment Dimension Definition Example of Measures Density & intensity
Amount of activity in a given area
Persons per acre, jobs per square mile Ratio of commercial floor space to land area
Land use mix Proximity of different land uses
Distance from house to nearest store Share of total land area for different uses Dissimilarity index
Street connectivity Directness and availability of alternative routes through the network
Intersections per square mile of area Ratio of straight line distance of network distance Average block length
Street scale Three-dimensional space along a street as bounded by buildings
Ratio of building heights to street widths Average distance from street to buildings
Aesthetic qualities Attractiveness and appeal of a place
Percent of ground in shade at noon Number of locations with graffiti per square mile
Regional structure Distribution of activities and transportation facilities across the regions
Rate of decline in density with distance from downtown Classification based on concentrations of activity and transportation network
Source : Handy, S. L., M. G. Boarnet, R. Ewing, and R. E. Killingsworth. 2002. "How the built
environment affects physical activity: views from urban planning." American Journal of Preventive Medicine 23:64-73.
262
Table 14 Categories of community context for understanding the obsogenic environment
Local Recreational and Sports Environment
Yes No Local Fast Food
Environment Yes Presence-Presence Presence-Absence No Absence-Presence Absence-Absence
Table 15 Descriptive statistics of obesity for four categories of community context
Local Environment Mean BMI SD N of obese Total N % of obese
Fast Food Restaurant -0.265 0.155 Sports and Rec. Facility
0.147 0.125
Absence-Absence (ref.)
Absence-Presence
0.266* 0.136
Presence-Absence
0.06 0.208 Presence-Presence
-0.313 0.218
Random-effects Parameters
Random Intercept 0.561 (0.051)
0.562 (0.051) 0.545 (0.051)
Level-1 Residual 3.404 (0.025)
3.404 (0.025)
3.404 (0.025)
Model Fit Statistics (BIC) 51,116.310 51117.850 51,128.410
Notes: Number of observation = 9586; Number of community = 218. Numbers are coefficients. Random-effects parameters have standard errors in parentheses. *Significance at 5% level; ** significance at 1% level; *** significance at 0.1% level
265
Table 20 Multivariate analysis of community contexts and obesity outcome (N=9,586)
Notes: Number of observation = 9586; Number of community = 218. Numbers are odds ratios. ICC and Level 2 variances have standard errors in parentheses. *Significance at 5% level; ** significance at 1% level; *** significance at 0.1% level
267
Table 21 Stratified analysis of community contexts and BMI outcome (N=9,586)
Overall Model Absence-Absence Absence-Presence Presence-Absence Presence-Presence
Fixed effects Coefficient SE Coefficient SE Coefficient SE Coefficient SE Coefficient SE Intercept 23.930*** 0.759 25.043*** 1.061 24.261*** 1.480 23.322*** 2.821 20.75*** 2.195 Level-1 Variables:
Notes: Number of observation = 9586; Number of community = 218. Numbers are coefficients. Random-effects parameters have standard errors in parentheses. In the Presence-Absence context, Guangxi Province is omitted due to collinearity. *Significance at 5% level; ** significance at 1% level; *** significance at 0.1% level
Overall Model Absence-Absence Absence-Presence Presence-Absence Presence-Presence
Fixed effects Coefficient SE Coefficient SE Coefficient SE Coefficient SE Coefficient SE Province Indicator Jiangsu 1.444*** 0.231 1.357*** 0.388 1.218** 0.384 2.297** 0.814 2.125*** 0.547
Notes: Number of observation = 9586; Number of community = 218. Numbers are odds ratios. Random-effects parameters have standard errors in parentheses. In the Presence-Absence context, Guangxi Province is omitted due to collinearity. *Significance at 5% level; ** significance at 1% level; *** significance at 0.1% level
271
Figure 1 Overweight and obesity prevalence by age: United States, 1971-2006
Source: CDC/NCHS, NHANES 2008. "U.S. Obesity Trends: Trends by State 1985–2008." vol.
2010 http://www.win.niddk.nih.gov/statistics/index.htm. Accessed on April 20th, 2010.
272
Figure 2 Percentage of adult population with a BMI >= 30 in OECD member countries
273
Figure 3 The world population in urban areas (source: Hoffman 2001, adapted from World
Bank, 2000, World Development Indicators)
274
Figure 4 Global infant and young child overweight trends, 1990‒2015
(by World Bank income group)
Source: WHO, 2011. "Global status report on noncommunicable diseases 2010: Description of the global burden of NCDs, their risk factors and determinants."
http://www.who.int/nmh/publications/ncd_report2010/en/index.html. Accessed on 08/30/2011.
Source: Huang, Yukon and Xubei Luo. 2008. "Reshaping Economic Geography: The China Experience." in Reshaping Economic Geography in East Asia, edited by Y. Huang and A. M. Bocchi. NY: World Bank.
276
Figure 6 Trends in the prevalence of overweight and obesity in China, 1992–2002
Reprinted from: Wang, H., S. Du, F. Zhai, and B. M. Popkin. 2007. "Trends in the distribution of body mass index among Chinese adults, aged 20-45 years (1989-2000)." International Journal of
Obesity 31:272-8. Used with permission.
277
Figure 7 Distribution of average BMI in adults by regions in China
Reprinted from: Zhuo, Q., Z. Wang, J. Piao, G. Ma, F. Zhai, Y. He, and X. Yang. 2009. "Geographic Variation in the Prevalence of Overweight and Economic Status in Chinese
Adults." British Journal of Nutrition 102:413-8. Used with permission.
278
Figure 8 Map of surveyed CHNS provinces in 2006
279
Figure 9 CHNS response rates at the individual-level, 1989-2006
0
20
40
60
80
100
1985 1990 1995 2000 2005 2010
Ra
tes
Year
CHNS Response Rates, 1989-2006
Response rates (%) based on1989
Response rates (%) based onprevious year
280
Figure 10 Schematic of data combination and sample selection
CHNS employment file
CHNS ID file CHNS physical exercise file
CHNS diet file
CHNS marriage file
CHNS education file
First-cut CHNS sample CHNS community file CHNS household income file
Exclusion where age<18 and exclusion of pregnant women; Listwise deletion of observations with missing values
Second-cut CHNS sample
Final Sample for this study
CGSS income inequality data
281
Figure 11 Gini coefficients by province, CHNS 2006
- 0.10 0.20 0.30 0.40 0.50 0.60 0.70
Gini
Coefficient
Province
Gini Coefficient by Province
282
Figure 12 Prevalence of obesity by individual SES, CHNS 2006
283
Figure 13 Prevalence of obesity by province, CHNS 2006
0%
10%
20%
30%
40%
Province
Prevalance of Obesity
by Province
284
Figure 14 Provincial Gini coefficients and obesity prevalence
-
5.00
10.00
15.00
20.00
25.00
30.00
35.00
40.00
45.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Obesity
Prevalance
Gini Coefficient
285
Figure 15 Provincial Gini coefficients and mean BMI by province
21.5
22
22.5
23
23.5
24
24.5
25
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Mean BMI
Gini
286
Figure 16 An ecological model relating the built environment to physical activity, diet and body weight.
Reprinted from: Powell, L., S. Slater, F. Chaloupka. 2005. A Multi-Causal Model of Eating,
Physical Activity and Obesity. www.impacteen.org/. Used with permission.