 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
a Department of Nutrition, Schools of Public Health and Medicine, Carolina Population Center
c Spatial Analysis Unit, Carolina Population Center, University of North Carolina, Chapel Hill, North Carolina
b Division of Epidemiology and Community Health, University of Minnesota, Minneapolis, Minnesota
 |
ABSTRACT |
|---|
 TOP ABSTRACT SUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
OBJECTIVE. In a nationally representative cohort, we sought to assess the geographic and social distribution of PA facilities and how disparity in access might underlie population-level PA and overweight patterns.
DESIGN, SETTING, AND PARTICIPANTS. Residential locations of US adolescents in wave I (1994–1995) of the National Longitudinal Study of Adolescent Health (N = 20745) were geocoded, and a 8.05-km buffer around each residence was drawn (N = 42857 census-block groups [19% of US block groups]). PA facilities, measured by national databases and satellite data, were linked with Geographic Information Systems technology to each respondent. Logistic-regression analyses tested the relationship of PA-related facilities with block-group socioeconomic status (SES) (at the community level) and the subsequent association of facilities with overweight and PA (at the individual level), controlling for population density.
MAIN OUTCOME MEASURES. Outcome measures were overweight (BMI 
95th percentile of the Centers for Disease Control and Prevention/National Center for Health Statistics growth curves) and achievement of 5 bouts per week of moderate-vigorous PA.
RESULTS. Higher-SES block groups had a significantly greater relative odds of having 1 or more facilities. Low-SES and high-minority block groups were less likely to have facilities. Relative to zero facilities per block group, an increasing number of facilities was associated with decreased overweight and increased relative odds of achieving 5 bouts per week of moderate-vigorous PA.
CONCLUSIONS. Lower-SES and high-minority block groups had reduced access to facilities, which in turn was associated with decreased PA and increased overweight. Inequality in availability of PA facilities may contribute to ethnic and SES disparities in PA and overweight patterns.
Key Words: environment • health disparities • adolescence • environmental health • population-based studies
Abbreviations: PA—physical activity • SES—socioeconomic status • GIS—Geographic Information Systems • SIC—Standard Industrial Classification • MVPA—moderate-vigorous physical activity • MET—metabolic equivalent • CDC—Centers for Disease Control and Prevention • NCHS—National Center for Health Statistics • OR—odds ratio • CI—confidence interval
Understanding the causes of health disparities is critical for improving health and reducing social inequality. Inequality in obesity and its underlying factors, in particular physical activity (PA) and inactivity, contribute greatly to health disparity. Minority and groups of low education are at highest risk for obesity and most other major noncommunicable diseases.1,2
Considerable research indicates that socioeconomic status (SES) at the neighborhood level is related to obesity, PA, and other health-related behaviors.3–7 Moreover, limited research shows that access to community facilities is positively associated with PA levels.8–11 In fact, this area of research is being strongly encouraged by both the National Institutes of Health12 and the Robert-Wood Johnson Foundation.13 However, there are no population-level analyses on the relationship between SES and the distribution of recreational facilities. Furthermore, there are very few studies investigating the relationship between neighborhood facilities and PA and obesity patterns in large and diverse populations.
This study fills a critical gap by examining the potential role that the built environment might play in inequality of PA and obesity at the national level with a large, ethnically diverse sample of adolescents and exact measures of PA-related facilities. Understanding physical environment factors such as the possible inequitable distribution of such resources is important for public policy related to ameliorating health disparities.14
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Respondent-Location Data
All Add Health respondents had residential street addresses that were recorded at each interview. Residential street addresses were geocoded by a commercial geocoding services vendor, Geographic Data Technology (now TeleAtlas, Lebanon, New Hampshire). Respondent residences that could not be properly address-geocoded had their positions recorded by using global positioning system devices. Locations for wave I respondents were then assigned from the best available source in the following priority order: addresses geocoded with street segment match (N = 17119), addresses geocoded with global positioning system (N = 3242), addresses geocoded with zip code + 4 or zip code + 2 centroid match (N = 163), addresses geocoded with 5-digit zip-code centroid match (N = 88), and addresses geocoded by the respondent's school location (N = 133). For respondents with problematic wave I (1994–1995) location data, wave II (1996) location information was substituted if the respondent information indicated that they had not moved. Two adjustments were made to the resulting data layer. The 163 respondents with residential locations >75 miles from their schools were presumed to be in error and were reassigned to their school locations. During data quality review of the respondent-location data layer, the respondent locations based on Geographic Data Technology address geocoding in 1 community seemed to be systematically offset from their correct locations when compared with digital orthophotography for the area. These respondent locations were adjusted to match the geography shown in the orthophotography by using a standardized Geographic Information Systems (GIS) "rubber-sheeting" operation.15
An 8.05-km (5-mile) buffer was drawn around each respondent, on the basis of empirical evidence that this distance would likely capture relevant PA facilities.16,17 An 8.05-km buffer was built around each respondent (N = 20745), and then all were aggregated to create the full set of 42857 census-block groups used in this study. In this study, the buffers were used for the sole purpose of generating the large sample of census-block groups that are described below.
Census-Block–Group Locations
The 8.05-km circular buffers for each respondent were combined to form an aggregated buffer data layer. A GIS polygon-on-polygon overlay combined the aggregate buffers and the census-block–group boundaries. All census-block groups in which a respondent residence was located, plus all census-block groups fully subsumed within the aggregate buffers, were included in the sample (N = 42857 census-block groups) except block groups that represented military/merchant ships. A GIS point-in-polygon overlay was performed to select all the PA facilities and resources located within these census-block groups (Fig 1).
|
PA Facilities and Resources
A commercially purchased set of digitized business records recorded in a proprietary 4-digit extension to the 4-digit Standard Industrial Classification codes (SIC) were used. These detailed 8-digit SIC codes are identical to those used by the Census Bureau. Comprehensive retrospective data for time period of interest were used. The standard set of data used for addresses and linkage to facilities from commercial databases provides 6-digit SIC codes, which do not provide depth in categorization of facilities or accurate retrospective data (ie, most databases include only facilities that are currently in business).
A comprehensive list of 169 of the 8-digit SIC codes for PA facilities and resources was built. YMCA/YWCA facilities do not have unique SIC identifiers, so a textual query was developed to select them based on the facility name field in the database. The SIC code list, YMCA/YWCA query, and zip codes encompassing the respondent 8.05-km buffers were sent to the commercial vendor. All records of interest matching the criteria and falling within the zip-code areas specified were returned along with facility names and street addresses contemporaneous to the wave I calendar year 1995. Facility locations were geocoded by using ESRI StreetMap 2000 1.1 (Environmental Systems Research Institute, Inc, Redlands, CA) (which uses street and address data from Geographic Data Technology), with an address-geocoding match rate of 85.9%. An additional 8.2% of the facilities were located by using latitude and longitude coordinates that were provided by the commercial vendor, yielding a total of 94.1% of the 71286 records returned from the commercial vendor that were entered into the recreational-facilities GIS data layer.
Measures of recreational facilities were derived from the larger list of SIC resources. The SIC codes were summarized into a single measure of all facilities and subdivided into smaller categories of specific types of facilities, with some overlap between categories (eg, outdoor facilities and public facilities) (Table 1).
|
Moderate-Vigorous PA
The wave I questionnaire included a standard PA-behavior recall8,21 that is similar to other self-report questionnaires that have been used and validated in other large-scale epidemiologic studies.22–25 Information was elicited on participation in moderate-vigorous PA (MVPA) (5–8 metabolic equivalents (METs); skating and cycling, exercise, and active sports) in the previous week. One MET is defined as the energy expenditure associated with quiet sitting. The Add Health PA questions were worded as such, "During the past week, how many times did you..." followed by a listing of activities such as walking, basketball, softball, and so forth, allowing calculation of frequency (bouts) of specified activities per week by MET value. A binary variable was created to represent whether the respondent achieved 5 bouts of MVPA per week. Those with missing data were excluded.
Body Mass
Height and weight were self-reported in wave I during in-home surveys, and overweight status was defined as a BMI 95th percentile of age- and gender-specific cut points from the 2000 Centers for Disease Control and Prevention (CDC)/National Center for Health Statistics (NCHS) growth charts.26 Those with missing data were excluded.
Given their correlation with measured height and weight and the acceptance of self-report measures in epidemiologic studies, self-reported height and weight were used.27
Statistical Analysis
This is a descriptive ecological study investigating the association between block-group–level sociodemographic factors and availability of PA and recreational facilities. In addition, the association between community PA and recreational facilities and individual-level PA and overweight was assessed. Statistical analyses were conducted by using Stata 8.2.28 Three sets of logistic-regression models were run at 2 distinct levels (national and individual): (1) population-level models that tested the relative odds of having 1 of various types of recreational facilities by census-level education status, controlling for the proportion of the census-level population of nonwhite ethnicities (N = 42187; 651 were missing block-group education and minority data, and 19 were missing block-group education data); (2) interactive population-level models (minority census-level population x census-level education status) that tested the relative odds of having 1 recreational facility per block group at combined levels of census-level education and minority population (N = 42187); and (3) individual-level analyses that assessed the association between number of facilities within an individuals' residential block group and relative odds of overweight (BMI 95th percentile of the CDC/NCHS growth curves; N = 17950) and high MVPA (5 bouts of MVPA per week; N = 18413). To retain comparability of our research with other disparity-related research, this analysis maintains the census-block–group definition of neighborhood for the individual-level analyses. Individual-level models assessed only the census-block group in which the individual resided. No analyses were made at the buffer level in this study. All models controlled for population density within the block group.
The widely accepted series of Stata survey procedures were used to correct SEs for multiple stages of cluster sample design in models predicting the likelihood of overweight and PA.
|
RESULTS |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
5 bouts of MVPA per week increased with each additional PA facility per block group. Having just 1 PA facility was associated with an increased relative odds of engaging in 5 bouts of MVPA per week by 3% relative to having no such facilities (OR: 1.03; 95% CI: 1.01–1.06; P = .009). Individuals who lived in census-block groups with 7 PA facilities were 32% less likely to be overweight and 26% more likely to be highly active than those who lived in block groups with no PA facilities.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
5 bouts of MVPA per week and a decreased relative odds of overweight. In addition, there is a significant beneficial effect linked with each additional PA facility. Of particular relevance is the fact that, in addition to all facilities showing inequitable distribution by sociodemographic characteristics, categories of facilities that were expected to be distributed equitably (eg, public facilities, youth organizations, parks, YMCAs, and schools) were actually distributed inequitably. Ethnic minorities and those of lower education are at the highest risk for lack of PA and recreational facilities. These findings, coupled with the association between the availability of these resources at the block-group level and the increased likelihood of engaging in PA and the decreased relative odds of overweight status, suggest that inequality in the built environment might underlie important ethnic and sociodemographic health disparities.
Three recent review articles conclude that environmental factors, measured either objectively or perceptively, are consistently related to PA.29–31 Objective and perceived access to facilities and opportunities to exercise are consistent predictors of PA in smaller regional studies.11,32–34 One study in a small mid-Western US city shows some association between distribution of facilities and sociodemographic factors.35 A larger study in Atlanta, GA,7 and another using national data at the county and metropolitan level7 show an association between the built environment and BMI and PA. Future work needs to determine which specific types of environmental changes are likely to impact PA and obesity, which at present is unknown.30
The association between SES and PA, independent of race/ethnicity, has been shown in a variety of populations.11,36 Although some have suggested that lower-SES communities are located in denser commercial space and therefore have access to more facilities,11,37 our research at the national level suggests that this access to commercial space may not translate into access to a variety of recreational facilities across diverse settings. The vast geographic coverage of these national data provide a rare opportunity to examine these relationships at a national level that had not yet been examined.
This cross-sectional ecological study shows crude associations between census-block–level sociodemographics and availability of PA and recreational facilities and advance the association between these facilities and individual-level behaviors. These findings have important policy and intervention implications and emphasize the connection between the built environment and health outcomes.
Although this study investigates the availability of PA and recreational facilities across census-block groups, availability is just one dimension that should be addressed. Affordability, quality, and accessibility, as well as availability, are important. Nonetheless, this study shows that availability alone is associated with a significant increase in bouts of PA and decreased overweight. This is a cross-sectional study, and as such it is impossible to demonstrate that recreational facilities have been distributed inequitably over time. Other factors in the built environment might equally impact obesity and obesity-related behaviors. Some work suggests a similar relationship between inequitable distribution of grocery stores by SES and race/ethnicity,38,39 and these factors are not considered in the present study. Our future work will delve deeper into measures of the built environment that were not captured in this study (eg, neighborhood walkability, land use, community design) at the individual, as opposed to the ecological, level.
These findings suggest that a wide range of US PA and recreational facilities may be distributed inequitably by ethnicity and SES, with additional association with individual-level behaviors. It is imperative that intervention efforts address the disparities in access to facilities and the relationship between the built environment and obesity and PA. Increasing the availability of PA and recreational facilities in underserved communities may be a profitable strategy for increasing PA and decreasing overweight at a population level.
|
ACKNOWLEDGMENTS |
|---|
We thank Ms Frances Dancy for helpful administrative assistance. This research uses data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris, from the National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. Special acknowledgment is due to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design.
|
FOOTNOTES |
|---|
Address correspondence to Penny Gordon-Larsen, PhD, University of North Carolina, Carolina Population Center, University Square, 123 W Franklin St, Chapel Hill, NC 27516-3997. E-mail: gordon_larsen{at}unc.edu
Those who are interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 W Franklin St, Chapel Hill, NC 27516-2524 (www.cpc.unc.edu/addhealth).
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M.-C. Yeh, A. Viladrich, N. Bruning, and C. Roye Determinants of Latina Obesity in the United States: The Role of Selective Acculturation J Transcult Nurs, January 1, 2009; 20(1): 105 - 115. [Abstract] [PDF]
|
|
|
|
 |
|
 J. Veitch, J. Salmon, and K. Ball Children's active free play in local neighborhoods: a behavioral mapping study Health Educ. Res., October 1, 2008; 23(5): 870 - 879. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. K. Kumanyika, E. Obarzanek, N. Stettler, R. Bell, A. E. Field, S. P. Fortmann, B. A. Franklin, M. W. Gillman, C. E. Lewis, W. C. Poston II, et al. Population-Based Prevention of Obesity: The Need for Comprehensive Promotion of Healthful Eating, Physical Activity, and Energy Balance: A Scientific Statement From American Heart Association Council on Epidemiology and Prevention, Interdisciplinary Committee for Prevention (Formerly the Expert Panel on Population and Prevention Science) Circulation, July 22, 2008; 118(4): 428 - 464. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K E MacLeod, G C Gee, P Crawford, and M C Wang Neighbourhood environment as a predictor of television watching among girls J Epidemiol Community Health, April 1, 2008; 62(4): 288 - 292. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Friel, M. Chopra, and D. Satcher Unequal weight: equity oriented policy responses to the global obesity epidemic BMJ, December 15, 2007; 335(7632): 1241 - 1243. [Full Text] [PDF]
|
|
|
|
 |
|
 B. A. Spear, S. E. Barlow, C. Ervin, D. S. Ludwig, B. E. Saelens, K. E. Schetzina, and E. M. Taveras Recommendations for Treatment of Child and Adolescent Overweight and Obesity Pediatrics, December 1, 2007; 120(Supplement_4): S254 - S288. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. M. Wolfe Return to Work After Gastric Bypass in Medicaid-Funded Morbidly Obese Patients Invited Critique Arch Surg, October 1, 2007; 142(10): 941 - 941. [Full Text] [PDF]
|
|
|
|
 |
|
 Q. Zhang and Y. Wang Using concentration index to study changes in socio-economic inequality of overweight among US adolescents between 1971 and 2002 Int. J. Epidemiol., August 1, 2007; 36(4): 916 - 925. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Papas, A. J. Alberg, R. Ewing, K. J. Helzlsouer, T. L. Gary, and A. C. Klassen The Built Environment and Obesity Epidemiol. Rev., May 28, 2007; (2007) mxm009v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Wang and M. A. Beydoun The Obesity Epidemic in the United States--Gender, Age, Socioeconomic, Racial/Ethnic, and Geographic Characteristics: A Systematic Review and Meta-Regression Analysis Epidemiol. Rev., May 17, 2007; (2007) mxm007v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Prevalence of Fruit and Vegetable Consumption and Physical Activity by Race/Ethnicity--United States, 2005 JAMA, May 16, 2007; 297(19): 2072 - 2074. [Full Text] [PDF]
|
|
|
|
 |
|
 P. M. Minihan, S. N. Fitch, and A. Must What Does the Epidemic of Childhood Obesity Mean for Children with Special Health Care Needs? J. Law Med. Ethics, March 1, 2007; 35(1): 61 - 77. [PDF]
|
|
|
|
 |
|
 N. H. Brodersen, A. Steptoe, D. R Boniface, J. Wardle, and M. Hillsdon Trends in physical activity and sedentary behaviour in adolescence: ethnic and socioeconomic differences * COMMENTARY Br. J. Sports Med., March 1, 2007; 41(3): 140 - 144. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2006 American Academy of Pediatrics. All rights reserved. |
||
|
||
