obesity and the built environment
TRANSCRIPT
C
OK
ABaIbwmtrsscsstcmmtahdobeJ
Momgout(
KaWHa
PsE
A
S
RESEARCH
urrent Research
besity and the Built Environment
ATIE M. BOOTH, PhD; MEGAN M. PINKSTON, MA; WALKER S. CARLOS POSTON, PhD, MPHefaprv
ITlataFciUnobline(chpttbsc
NNarae(sdipv
mtcvf
BSTRACTiological, psychological, behavioral, and social factorsre unable to fully explain or curtail the obesity epidemic.n this article we review research on the influence of theuilt environment on obesity. Studies were evaluatedith regard to their methods of assessing the environ-ent and obesity, as well as to their effects. Methods used
o investigate the relationships between the built envi-onment and obesity were found to be dissimilar acrosstudies and varied from indirect to direct. Levels of as-essment between and within studies varied from entireounties down to the individual level. Despite this, obe-ity was linked with area of residence, resources, televi-ion, walkability, land use, sprawl, and level of depriva-ion, showing promise for research utilizing moreonsistent assessment methods. Recommendations wereade to use more direct methods of assessing the environ-ent, which would include specific targeting of institutions
hought to vary widely in relation to area characteristicsnd have a more influential effect on obesity-related be-aviors. Interventions should be developed from the in-ividual to the neighborhood level, specifically focusingn the effects of eliminating barriers and making neigh-orhood level improvements that would facilitate thelimination of obesogenic environments.Am Diet Assoc. 2005;105:S110-S117.
any investigators have attempted to explain theobesity epidemic, but no single theory has suffi-ciently explained all of the factors contributing to
verweight and obesity. For instance, although genesay increase susceptibility for obesity, no dominant
enes have been discovered whose presence is necessaryr sufficient to cause obesity (1). Despite the emphasis onnderstanding and modifying individual characteristicshat influence dietary and physical activity patterns2-4), little progress has been made in halting the obesity
. M. Booth is a postdoctoral fellow, M. M. Pinkston isclinical psychology–health emphasis PhD student, and. S. C. Poston is an associate professor with theealth Research Group, Mid America Heart Institutend University of Missouri-Kansas City.Address correspondence to: Walker S. Carlos Poston,
hD, MPH, 4825 Troost, Suite 124, University of Mis-ouri-Kansas City, Kansas City, MO 64110.-mail: [email protected] © 2005 by the American Dietetic
ssociation.0002-8223/05/10505-1016$30.00/0
idoi: 10.1016/j.jada.2005.02.045
110 Supplement to the Journal of the AMERICAN DIETETIC ASSOCIATION
pidemic. As a result, researchers also have begun toocus on the interaction between environmental factorsnd the development of overweight and obesity. The pur-ose of this article is to provide an overview of the currentesearch assessing the relationship between the built en-ironment and obesity.
NFLUENCE OF THE BUILT ENVIRONMENT ON OBESITYhe built environment includes urban design factors,
and use, and available public transportation for a region,s well as the available activity options for people withinhat space (5). The built environment can both facilitatend hinder physical activity and healthful eating (6,7).or example, areas with few recreational facilities, safetyoncerns, uneven and hilly terrain, or insufficient light-ng can hinder physical activity (8). Many areas in thenited States are designed specifically for vehicles witho concessions for pedestrians (6), and zoning restrictionsften lead to land use where specific distinctions existetween commercial and residential properties (ie, lowand use mix) (5). In contrast, areas with high connectiv-ty (ie, the directness or connectedness of travel in aeighborhood with multiple pedestrian access points),ncourage more walking and bicycling for transportation8). An increasingly high density of fast-food restaurants,onvenience stores, bars, food distribution programs withigh-fat foods, and concentrated media marketing, allromoting unhealthful food choices, hinder good nutri-ion (9). For example, poorer neighborhoods have threeimes fewer supermarkets than wealthier neighborhoodsut contain more fast-food restaurants and conveniencetores, limiting the availability of nearby healthy foodhoices (10).
eighborhood Influenceseighborhoods are commonly defined by census bound-ries (ie, block groups or tracts) and have been linked toesidents’ health outcomes (11). Census tracts includepproximately 4,000 people, and boundaries are delin-ated to include a comparatively homogenous population12). Census data are aggregated to represent the expo-ure to neighborhood environments that may indepen-ently affect human behavior, unique from measures ofndividual attributes (eg, individual income) (11). Thus,hysical environments can influence the health of indi-iduals beyond that of individual health risk factors (13).For example, safer neighborhoods, which include aixture of houses, commercial, retail, and recreation des-
inations, often result in more physical activity and socialapital and less overweight and obesity (8,14). Along witharious available neighborhood destinations, pedestrianacilities and public transportation help facilitate walk-
ng and bicycling for transportation (8).© 2005 by the American Dietetic Association
udtbgamrpa
istlmlpno
RImws(mkmnmr
tattfFt(ut
tdiatceowonipcatFtowvtpoSt
aaawcnshoclimaaw
F .
Neighborhoods with low socioeconomic status (SES)sually have fewer physical activity resources than me-ium- to high-SES neighborhoods, leading to more inac-ivity of neighborhood residents (3,6). In low-SES neigh-orhoods, many incivilities (eg, physical decay, litter,raffiti) and unsafe conditions are common, creating un-ppealing and even dangerous neighborhood environ-ents (15). However, high levels of walking behavior are
eported in low-SES neighborhoods, likely due to highopulation density, walking to work, and a greater reli-nce on public transportation (3).Residents in neighborhoods with more available phys-
cal activity resources, including sidewalks and safetreets, report higher activity levels (2). The proximity ofhese resources is important because people are moreikely to use nearby resources (8). Making neighborhoods
ore walkable (ie, designed for pedestrians with mixedand use) might help increase physical activity. However,revious walkability studies (14) have used self-reportedeighborhood environment variables instead of direct,bjective measurements.
ESEARCH ON OBESITY AND THE BUILT ENVIRONMENTn the emerging field of investigating obesogenic environ-ents, a range of assessment methods have been used,ith few studies using similar methods. Methods for as-
essing the built environment include direct assessmentseg, in-person audits by trained observers), intermediateeasures (eg, use of telephone book yellow pages or mar-
eting databases to identify institutions), and indirecteasures (eg, aggregation of census data to approximateeighborhood SES). Figure 1 illustrates the continuum ofethods used to assess environmental factors in the cur-
ent research.Studies reviewed in this article investigated the rela-
ionship between characteristics of the built environmentnd obesity. The studies were evaluated according toheir attempts to define the obesogenic environmenthrough aspects of community design, the prevalence ofood stores, and neighborhood and material deprivation.igure 2 outlines the purposes of the studies, categorizeshem by the type of environmental assessment methodie, indirect, intermediate, direct) used to define individ-al and environmental variables, and also summarizes
igure 1. Continuum of methods for measuring the built environment
he limitations of the studies. b
May 2005 ●
The primary question facing researchers investigatinghe built environment and obesity is whether communityesign factors might prevent individuals from engagingn physical activity (16) while encouraging them to selectnd eat more energy-dense and low–nutrient-value foods,hus contributing to the obesity epidemic. Giles-Corti andolleagues (7) included both indirect and intermediatenvironmental measures in their study and found thatverweight yet healthy and working Australian adultsere more likely to live near highways. In addition, both
verweight and obese adults were more likely to live ineighborhoods that lacked adequate sidewalks and prox-
mal places for physical activity. In fact, participants withoor access to recreational facilities had a 68% greaterhance of being obese. Interestingly, residents withoutccess to a motor vehicle were twice as likely to be obesehan residents who always had access to a motor vehicle.inally, participants who watched 3 or more hours ofelevision per day almost doubled their odds of beingverweight and were likely to be obese when comparedith people who watched less than 3 hours of daily tele-ision. Although the self-reported sedentary activity ofelevision watching and the lack of proximal places forhysical activity were associated with greater levels ofverweight and obesity, this study did not find that theES of healthful and working neighborhoods was relatedo overweight and obesity (7).
The density and land use mix (ie, types of zoning) ofreas also has been found to impact obesity risk. Saelensnd colleagues (8) first identified residents of high-walk-ble (single- and multiple-family residences) and low-alkable (single-family residences) neighborhoods with
omparable SES using census data. They comparedeighborhood residents based on physical activity mea-urements, weight status, and self-reported neighbor-ood perceptions. Residents then directly recorded theirwn physical activity using accelerometers. Results indi-ated that residents from high-walkability neighborhoodsived in neighborhoods more conducive to physical activ-ty (ie, higher residential density and street connectivity,
ore diverse and accessible land use, better aesthetics,nd pedestrian safety) than did residents from low-walk-bility neighborhoods. Accordingly, residents of low-alkability neighborhoods tended to report higher mean
ody mass indexes (BMIs) and have higher rates of over-Supplement to the Journal of the AMERICAN DIETETIC ASSOCIATION S111
Article Purpose Outcomes
Environmental Assessment Methods
Findings LimitationsType of Measurement Category
Burdette andWhitaker (18)
Examined the relationshipbetween overweight inpreschool children andthree factors: theproximity of thechildren’s residences toplaygrounds, to fast-food restaurants, andthe safety of theneighborhoods
● Height, real height obtained● Weight, real weight obtained● BMIa● Demographics, WICb database
● Prevalence of playgrounds,County Health Departmentdatabase
● Identification of fast-foodrestaurants, yellow pagesfrom the phone book andInternet
● Crime data from the Cincinnatipolice department
● GISc used to analyze thespatial relationships
Intermediate ● No relationship betweenoverweight or nonoverweightlow-income children anddistance to playgrounds orfast-food restaurants and levelof safety.
● Lack of variation in theenvironmental exposurevariables
● Categorized exposures at aneighborhood level
● Limited mobility of the studypopulation; no idea how longthey had lived at address
Cubbin andcolleagues (12)
Examined the relationshipbetween neighborhoodmaterial deprivationand CVDd risk factorsas independent ofsocioeconomic status inminority participants
● Physical inactivity, NHANES IIIe● Type 2 diabetes, NHANES III● Smoking status, NHANES III● BMI● Systolic blood pressure–
NHANES III, blood tests● Cholesterol (non-HDL-C)f,
NHANES III, blood tests● Age, education, income,
NHANES III
● Townsend material deprivationindex derived from the 1990census� occupation status� car ownership� renter occupied housing� overcrowding
Indirect ● African-American women withhighest BMI in materiallydeprived neighborhoods
● Other CVD factors foundamong different ethnic groups
● Self-report● Cross-sectional● Neighborhood effects possibly
due to self-selection● No information of length of
residency
Ellaway andcolleagues (19)
Examined if neighborhoodis associated with bodysize and shape
● Height, real height obtained● Weight, real weight obtained● BMI● Waist circumference, real
circumference obtained● Waist-to-hip ratio● Social class, self-report of
occupation
● Neighborhood materialdeprivation index, self-reportof housing tenure, carownership, and income
Indirect ● Lower social class had higherBMIs
● Participants in the mostdeprived areas had higherBMIs, larger waists, andhigher waist-to-hip ratios thanparticipants from the non-deprived areas.
● Neighborhood of residenceassociated with BMI and otherphysical factors
● No measure of the builtenvironment
Ewing andcolleagues (16)
Examined the relationshipbetween urban sprawl,health, and health-related behaviors
● Physical Activity Outcomesincluding leisure time physicalactivity, BRFSSg
● Weight-related outcomes (BMI,obesity), BRFSS
● Morbidity outcomes(hypertension, diabetes,coronary heart disease),BRFSS
● Smart Growth America’smetropolitan sprawl index� residential density� land use mix� degree of centering� street accessibility
● County sprawl index (based onUS Census data and data fromthe Natural ResourcesInventory of the USDepartment of Agriculture)� low residential density� poor street accessibility
Indirect Individuals in sprawling counties:● Weighed more● Exercised less● Had hypertension
● Completed at a county levelrather than neighborhood level
● Cross-sectional, cannot suggestcausal relationship
● Characterization of physicalactivity
● Other sprawl indexes notmeasured
● Did not examine caloric intake● BRFSS is self-report
Frank andcolleagues (17)
Examined the relationshipbetween the builtenvironment and travelpatterns and comparedwith BMI, and obesity
SMARTRAQh study● Height, self-report● Weight, self-report● BMI● Demographics
SMARTRAQ study● Sociodemographic variables,
self-report● Minutes spent in car● km walked● Built environment with GIS
� county-level tax assessor’sdata
� digital aerial photography� street network map� census data
Intermediate
Mix ofintermediateand indirect
● Obesity associated with landuse mix as mediated byphysical activity and distancewalked
● Participation bias● Self-reported BMI● Limited range of urban form● No account of time in transit
use● Cross-sectional, cannot suggest
causal relationship
S112M
ay2005
Suppl1Volum
e105
Number
5
Article Purpose Outcomes
Environmental Assessment Methods
Findings LimitationsType of Measurement Category
Giles-Corti andcolleagues (7)
Examined the relationshipbetween environmentaland lifestyle factors andoverweight and obesity
● BMI, self-reported weight andheight
● Demographics, self-report● Lifestyle factors, self-report● Recreational physical activity,
self-report● Social environmental factors,
self-report● Perception of sidewalks,
self-report● Physical environmental
factors, GIS surveying ofneighborhoods
Intermediate
Indirect
● Quality of physicalenvironment related tooverweight and obesity
● Overweight participants livedon streets without sidewalks
● Overweight and obeseparticipants reported not livingwithin walking distance tostores
● Self-report● Possibility that those who are
sedentary to begin with mightnot notice areas to do physicalactivity
● Perceptions of the environmentwere not validated
● High standard of livingenvironment sampled from (ie,the working well)
Kinra andcolleagues (21)
Examined the relationshipbetween socioeconomicdeprivation andchildhood obesity
● Height, real height obtained● Weight, real weight obtained● BMI
● Townsend Material DeprivationScore derived from 1991census data:� unemployment� overcrowding� wealth� income
Indirect ● Children in more deprivedareas were 2.5 times moreobese than the rest of thepopulation of the UnitedKingdom
● Sample was limited to whites● Sample was from only state
schools● Townsend index used, rather
than asking individual levelquestions
Saelens andcolleagues (14)
Examined physical activityand weight status ofresidents comparedwith neighborhoodenvironmental survey
● Physical activity,accelerometers and self-report
● Height and weight, self-report● Demographics, self-report
● Neighborhoods determinedusing 1990 census data togain high- and low-walkabilityneighborhoods
● Neighborhood EnvironmentWalkability Scale, self-reportsurvey
Indirect
Intermediate
● Trend found that residents oflow walkability neighborhoodshad higher BMIs than thosewith high walkabilityneighborhoods
● Low walkability neighborhoodresidents walked significantlyless than those residents fromhigh walkability neighborhoods
● Unable to determine if one’sneighborhood can be definedas area of physical activity
● Low recruitment rate● Demographic differences
between neighborhoods● Did not validate environments● Self-report data
van Lenthe andMackenbach (20)
Examined the relationshipbetween neighborhooddeprivation andoverweight and lookedto see if thisassociation wasmodified by education,age, or sex
● BMI, self-report on a postalquestionnaire from the Dutchlarger study—Health andLiving Conditions of thePopulation of Eindhoven andsurroundings (GLOBE)
● Neighborhood deprivationbased on:� educational level� occupational level� employment status
Indirect ● As neighborhood deprivationincreased, prevalence ofoverweight increased
● Self-report data● Neighborhoods defined by
aggregates of samples
aBMI�body mass index.bWIC�Special Supplemental Nutrition Program for Women, Infants, and Children.cGIS�geographic information system.dCVD�cardiovascular disease.eNHANES III�Third National Health and Nutrition Examination Survey (1988-1994), self-report survey.fHDL-C�high-density lipoprotein cholesterol.gBRFSS�Behavioral Risk Factor Surveillance System (1998-2000), self-report survey.hSMARTRAQ�Strategies for Metro Atlanta’s Regional Transportation and Air Quality study, self-report survey.
Figure 2. Review of studies where the built environment was investigated in relationship to body mass index.
May
2005●
Supplement
tothe
Journalofthe
AMERICAN
DIETETICASSOCIATION
S113
wawpr
cArauttnqAelrtru0a
mrlnotSsrBt
isautfrsmwvlaisi
caaeesplB
huaNsbaahrargwdsg
mnmfthLldcdiwwetmrtblctsov
tnvoabaHIauwbAdia
S
eight than high-walkability neighborhood residents. Inddition, residents in the high-walkability neighborhoodalked significantly more (eg, a difference of 52 minuteser week of moderate to vigorous physical activity) thanesidents in the low-walkability neighborhood (14).
Frank and colleagues (17) investigated the impact ofommunity design and physical activity on obesity intlanta, GA using both indirect and intermediate envi-
onmental data sources. Neighborhoods were designateds connected or disconnected (ie, high or low walkability)sing land-use mix data from the county tax assessor andhe 2000 census within a Geographic Information Sys-ems (GIS) framework. Participant data within eacheighborhood were drawn from a transportation and airuality survey, which measured individual level factors.fter adjusting for the effects of age, income, and level ofducation, a significant relationship was found betweenand-use mix and the prevalence of obesity, although thiselationship was mediated by physical activity (ie, dis-ance walked during a 2-day time period). For instance,esearchers found that a single quartile increase in land-se mix was related to a 12.2% reduction (odds ratio�.878; 95% confidence interval�0.839-0.919) in the prob-bility of being obese.Ewing and colleagues (16) used only indirect environ-ental assessment methods when they investigated the
elationship between sprawl (ie, low housing density, lowand-use mix, no strong centers of activity, and poor con-ectivity) and physical activity level and prevalence ofbesity. Self-reports of behavioral and health-status ques-ions from the BRFSS (Behavioral Risk Factor Surveillanceystem) and the Smart Growth America’s metropolitanprawl index, adapted to the county level, indicated thatesidents of sprawling counties walked less, had higherMIs, and higher obesity and hypertension prevalence
han did residents of more compact counties.County-level analyses controlling for minutes walked
ndicated that sprawl seemed to have a linear relation-hip to BMI and obesity (ie, more sprawl�higher BMIsnd obesity rates) and an indirect relationship with min-tes walked. When the same outcomes were measured athe metropolitan level, no significant relationship wasound between obesity and the sprawl index. The authorseported that the county level analysis was more repre-entative of daily lifestyles of residents rather than theetropolitan level, which consists of multiple countiesith varying built environments. Although this studyiewed the environment at the county and metropolitanevels and examined its relationship to physical activitynd health, further research should examine the samenformation at the community or neighborhood level topecifically define the living and working environments ofndividuals.
Another aspect of community design is the role of ac-ess to areas promoting physical activity, such as walkingnd bicycling trails, parks, and playgrounds. Burdettend Whitaker (18) used both intermediate and indirectnvironmental assessments when they investigated theffects of community design for children. They hypothe-ized that overweight children would be less proximal tolaygrounds, closer to fast-food restaurants, and live iness-safe neighborhoods than nonoverweight children.
MI was determined from measured weights and i114 May 2005 Suppl 1 Volume 105 Number 5
eights, and intermediate measurement methods weresed to identify playgrounds (ie, database of playgroundddresses) and fast-food restaurants (ie, yellow pages).eighborhood safety was determined by police records of
erious crimes and emergency phone calls for each neigh-orhood. Children’s home addresses, the playgrounds,nd the fast-food restaurants were mapped using GIS,nd spatial distances were calculated. Not all childrenad access to either a playground or a fast-food restau-ant in their neighborhood. Contrary to findings in thedult literature, Burdette and Whitaker (18) found noelationship between overall neighborhood safety, play-round and fast-food restaurant proximity and over-eight status. Children with higher poverty levels alsoid not differ by weight status as related to neighborhoodafety or to the proximity of the neighborhoods to play-rounds and fast-food restaurants.Ellaway and colleagues (19) used only indirect environ-ental measures to examine the relationship between
eighborhood material deprivation and resident bodyass and obesity risk, independent of other individual
actors (ie, demographics, social factors). Face-to-face in-erviews were conducted in four different SES neighbor-oods, and participants were weighed and measured.ower SES participants had a higher overweight preva-
ence than higher SES participants. Similarly, the mosteprived areas (ie, lower income, less housing tenure, lessar ownership) had twice the proportion of obese resi-ents as compared with the more-affluent areas. Further,ndependent of other factors, neighborhood of residenceas associated with BMI, waist circumference, andaist-to-hip ratio, indicating that neighbohood can influ-
nce body size and shape. Similarly, a study by van Len-he and Mackenbach (20) also used indirect measure-ents with data from a self-report questionnaire of
esidents in 84 different neighborhoods (ie, administra-ive units) in the Netherlands. Increasing levels of neigh-orhood deprivation (ie, educational level, occupationalevel, and employment status) were associated with in-reasing mean BMIs and overweight prevalence, al-hough neighborhood deprivation had a stronger relation-hip for women and older individuals who wereverweight when compared with men and younger indi-iduals.Cubbin and colleagues (12), using indirect environmen-
al assessments, investigated the relationship betweeneighborhood material deprivation (ie, Townsend depri-ation index measured by using unemployment, no carwnership, renter-occupied housing, and overcrowding)nd the frequency of cardiovascular disease (CVD) riskehaviors (ie, physical inactivity, higher BMI) among USdults (ages 25 to 64) who completed the Third Nationalealth and Nutrition Examination Survey (NHANES
II). Both household interviews and on-site medical ex-minations were used in NHANES III to collect individ-al-level outcome data. The individual data were linkedith census neighborhood data to examine differencesetween ethnicities, while controlling for individual SES.one-unit increase in the neighborhood deprivation in-
ex was associated with an 18% increase in physicalnactivity for white men, a 12% increase in physical in-ctivity for white women, and a 10% increase in physical
nactivity for African-American and Mexican-American
mcaBdbivcf
cutfwyhwtawtKa
AAleoMa
MMc(telSabpsr
MSttr(hddamci
dsSwm
idomntnlcbht
mttmposphprlabaa
ooaAtetfaa
MOTsbitrvtillsb
en. After controlling for individual education and in-ome, African-American women were at a disproportion-tely higher risk for CVD, including presence of higherMIs, when living in neighborhoods of more materialeprivation. Overall, residing in a deprived area or neigh-orhood was associated with a higher probability of hav-ng an adverse CVD risk profile. Although the risk profilearied by ethnic group and sex, neighborhood deprivationonsistently exerted an independent effect on CVD riskactors, even after adjusting for individual SES (12).
Similar results have been demonstrated using adoles-ent and child populations. Kinra and colleagues (21)sed indirect environmental methods to study the rela-ionship between neighborhood deprivation (ie, usingour census variables) and the measured heights andeights of 20,973 children between the ages of 5 and 14ears in the United Kingdom. Children’s weights andeights were directly assessed, but the built environmentas estimated using a census index of material depriva-
ion (ie, unemployment, overcrowding, owner occupation,nd car ownership). Results demonstrated that childrenho lived in more deprived areas had rates of obesity 2.5
imes more than the national rate of obesity in the Unitedingdom, showing a linear association between obesitynd neighborhood material deprivation.
PPLICATIONS AND LIMITATIONS OF THE CURRENT LITERATUREs evidenced by the reviewed studies, promising data
inks neighborhood of residence and obesity risk. How-ver, a variety of methods have been used to assess thebesity-related outcomes and the built environment.ore consistent methods still need to be developed and
pplied in the field.
easurement of Obesity Prevalenceany studies used self-reported height and weight to
alculate BMI to determine obesity prevalence7,14,16,17,20). This is problematic because people tendo underreport their weight, leading to inaccurate BMIstimates and likely underestimation of obesity preva-ence and risk, particularly among lower SES groups (20).ubsequently, this results in an underestimation of thectual extent of overweight and obesity. Obviously, it isetter to directly measure the height and weight of studyarticipants (18,19), although body composition and bodyhape measurements also should be incorporated intoesearch studies when practical (19).
easurement of the Built Environmenttudies investigating the built environment and its rela-ionship with obesity typically did not directly measurehe environment. Instead, indirect measures of the envi-onment were used to represent it, such as census data10,11,17,21), GIS data (eg, road network distance, steepill barrier, grid of city blocks) (7,17), and street networkata (17). Although these methods can approximate con-itions of the built environment, they may not be asccurate as direct measurements because database infor-ation often is dated and might not correctly reflect
onditions at the time of the study. Other studies using
ndirect methods have created indexes such as material cMay 2005 ●
eprivation, neighborhood deprivation, and neighborhoodafety (6,18,20,21) to distinguish between neighborhoodES levels, thereby reflecting the conditions of the peopleho live in the neighborhoods and not the built environ-ent itself.Intermediate measures of the built environment have
ncluded self-reported perceptions of neighborhood resi-ents (7,13,14,20). As Kirtland and colleagues (22) pointut, this is problematic because only fair to low agree-ent has been demonstrated between self-reports of
eighborhood and community environments and objec-ive environmental audits. Neighborhood residents mayot be able to correctly perceive distances (ie, which items
ie within neighborhood boundaries) and may have a per-eption bias that leads them to judge their environmentsased on their own expectations and lifestyles. Neighbor-ood residents also might have a different definition thanhe researchers of what makes up their neighborhood.
Other intermediate measurements of the built environ-ent that have been used are regional land-use data from
ax assessors and aerial photography (17). Althoughhese measurements can approximate the built environ-ent, tax data are self-reported by individuals and aerial
hotography cannot show actual uses of buildings. Vari-us databases (eg, departments of environmental health,tate departments of agriculture, telephone book, yellowages online, police Web sites, school district lists) alsoave been used to track specific entities (eg, places whereeople can buy food, public playgrounds, fast-food restau-ants) that are available within certain areas (10,18). Theimitation of these studies, however, is that they did notudit the actual site of the entities reported within theuilt environment; therefore, they made assumptionsbout availability within the environment without actu-lly verifying the accuracy of these data sources.Direct measurement through environmental audits has
nly been used in one study with obesity as the primaryutcome (7). Measurements included the type of streetnd the presence of sidewalks for each study participant.lthough these measures specifically verified what was in
he physical environment, they were not sophisticatednough to adequately capture enough characteristics ofhe built environment to account for all environmentalactors that have influenced obesity, such as the typesnd frequency of different institutions available in thereas.
ECHANISMS FOR HOW THE BUILT ENVIRONMENT INFLUENCESBESITY RISKhe reviewed studies typically demonstrated a cross-ectional association between indirect indexes of neigh-orhood context and obesity risk. However, it also ismportant to incorporate assessments of institutionshat may vary across environments and impact obesityisk. For example, food store density and location mayary in high vs low SES neighborhoods, contributing tohe availability of food options for individuals and help-ng explain the differences in obesity risk based onevel of neighborhood deprivation. Morland and col-eagues (10) used both indirect and intermediate mea-urement data sources to investigate the relationshipetween prevalence of food stores (eg, supermarkets,
orner stores, convenience stores) and restaurants andSupplement to the Journal of the AMERICAN DIETETIC ASSOCIATION S115
nefmisamwI
mhinstlhrlfnfntnftto
FTpeplplawmta
bgoFdprseaohttt
s
ibnpthnaasbah
crtahwbpafaa
WgI(HH
R
S
eighborhood SES, with a secondary analysis of differ-nces between racially segregated neighborhoods. In-ormation from the 1990 census (ie, indirect measure-ent) was used to approximate neighborhood SES and
ndividual level variables, and the addresses of foodtores were collected from local health departmentsnd state agriculture departments (ie, intermediateeasurement). Each type of food store or food serviceas classified according to the 1997 North America
ndustrial Classification System.Morland and colleagues (10) found three times asany supermarkets (eg, defined as having the mostealthful food options) compared with other food stores
n the wealthier neighborhoods, though more conve-ience stores, small grocery stores, and specialty foodtores were in the lower wealth neighborhoods. Fur-her, fast-food restaurants were more prevalent in theower- and medium-wealth neighborhoods than in theigher-wealth neighborhoods. For comparisons by race,esults indicated that more black residents lived inower SES neighborhoods than did white residents, andour times as many supermarkets were located in whiteeighborhoods than in black neighborhoods (10). Inact, the ratio of supermarkets for predominantly whiteeighborhoods was one per 3,816 residents, whereashe ratio of supermarkets for predominantly blackeighborhoods was only one per 23,582 residents. Un-ortunately, they did not collect data on obesity risk, sohey were unable to demonstrate a relationship be-ween the differences in the socioeconomic distributionf food sources and obesity risk and SES stratification.
UTURE DIRECTIONShe current review provides sufficient evidence to sup-ort the need for further research of the obesogenicnvironment. Implications for the interaction betweenublic health and community design have been estab-ished in this growing field of research. As definedreviously, the built environment includes the design,and use, and available public transportation of anrea, as well as the available activity options for peopleithin that space (5). Because these areas are definedore explicitly through research, interventions can be
ailored to encompass each aspect from a neighborhoodnd individual perspective.Before public health research results can be com-
ined with community planning efforts, more investi-ations using direct methods need to be completed inrder to identify effective changes for neighborhoods.uture research should strive to strictly define an in-ividual’s neighborhood based on both objective anderceived measures of the neighborhoods. By doing so,esearch can provide a strong foundation for under-tanding the interaction between individuals and theirnvironment. As noted in the limitations section of thisrticle, much of the current research has used indirectr intermediate methods for investigating neighbor-ood features. Although direct methods may be moreime-consuming and costly, they are necessary becausehey provide the most accurate and consistent descrip-ions of the neighborhood environment.
Concurrent with the research, resulting interventions
hould target factors at both the community level and the116 May 2005 Suppl 1 Volume 105 Number 5
ndividual level, with a focus on barriers to healthfulehaviors. From a neighborhood perspective, city plan-ers and public health officials must work together toromote agendas at a public policy level for changes inhe built environment to occur. At an individual level,ealth care professionals are encouraged to evaluate theeighborhood barriers that prevent clients from pursuingdequate physical activity and choosing healthful foods,nd thus leading to declines in health (6). Further, asuggested by Saelens and colleagues, the design of neigh-orhoods should focus on preventing material deprivationnd improving the walkability conditions of the neighbor-oods as a means for increasing physical activity (8).Results of the research presented in this review
learly demonstrate strong preliminary evidence of aelationship between built environment features andhe prevalence of obesity. Lower SES neighborhoodsre a primary concern, as residents in these areas mayave less access to recreational facilities or food storesith healthful, affordable options, although the neigh-orhoods themselves may be designed in ways thatromote physically active transportation rather thanutomobiles. This information is important in the ef-orts of researchers to impact public policy decisionsbout the built environment that affect communitiesnd health outcomes.
ork on this manuscript was partially supported byrants from the National Institutes of Health, Nationalnstitute of Diabetes and Digestive and Kidney DiseasesNIDDK–1RO1 DK064284-01) and the Agency forealthcare Research and Quality (AHRQ–3RO1S011282-02S1).
eferences1. Bouchard C. Genetics of obesity: An update on mo-
lecular markers. Int J Obes Relat Metab Disord.1995;19(suppl 3):S10-S13.
2. Giles-Corti B, Donovan RJ. Relative influences of in-dividual, social environmental and physical environ-mental correlates of walking. Am J Public Health.2003;93:1583-1589.
3. Yen IH, Kaplan GA. Poverty area residence andchanges in physical activity level: Evidence from thealameda county study. Am J Public Health. 1998;88:1709-1712.
4. Young LR, Nestle M. Expanding portion sizes in theUS marketplace: Implications for nutrition counsel-ing. J Am Diet Assoc. 2003;103:231-234.
5. Handy SL, Boarnet MG, Ewing R, Killingsworth RE.How the built environment affects physical activity:Views from urban planning. Am J Prev Med. 2002;23:64-73.
6. Jackson RJ, Kochtitzky C. Creating a Health Envi-ronment: The Impact of the Built Environment onPublic Health. Washington, DC: Sprawl Watch Clear-inghouse Monograph Series; 2001.
7. Giles-Corti B, Macintyre S, Clarkson JP, Pikora T,Donovan RJ. Environmental and lifestyle factors as-sociated with overweight and obesity in Perth, Aus-tralia. Am J Health Promot. 2003;18:93-102.
8. Saelens BE, Sallis JF, Frank LD. Environmental cor-
relates of walking and cycling: Findings from the
1
1
1
1
1
1
1
1
1
1
2
2
2
transportation, urban design, and planning litera-tures. Ann Behav Med. 2003;25:80-91.
9. Hinkle A, Wu E. Communities of color issue briefingpaper: Addressing the obesity epidemic—Public poli-cies for healthy eating and physical activity environ-ments. California Adolescent Nutrition and FitnessProgram & California Pan Ethnic Health Net-work; 2003. Available at: http://www.canfit.org/pdf/CANFit-CPEHNbrief.pdf. Accessed April 14, 2005.
0. Morland K, Wing S, Diez-Roux A, Poole C. Neighbor-hood characteristics associated with the location offood stores and food service places. Am J Prev Med.2002;22:23-29.
1. Lee RE, Cubbin C. Neighborhood context and cardio-vascular health behaviors in youth. Am J PublicHealth. 2002;93:428-439.
2. Cubbin C, Hadden WC, Winkleby MA. Neighborhoodcontext and cardiovascular disease risk factors: thecontribution of material deprivation. Ethn Dis. 2001;11:687-700.
3. Feldman PJ, Steptoe A. How neighborhoods andphysical functioning are related: The roles of neigh-borhood socioeconomic status, perceived neighbor-hood strain, and individual health risk factors. AnnBehav Med. 2004;27:91-99.
4. Saelens BE, Sallis JF, Black JB, Chen D. Neighbor-hood-based differences in physical activity: An envi-ronment scale evaluation. Am J Public Health. 2003;93:1552-1558.
5. Ross CL, Microwsky J. Neighborhood disadvantage,
May 2005 ●
disorder, and health. J Health Soc Behav. 2001;43:258-276.
6. Ewing R, Schmid T, Killingsworth R, Zlot A, Rauden-bush S. Relationship between urban sprawl and phys-ical activity, obesity, and morbidity. Am J Health Pro-mot. 2003;18:47-57.
7. Frank LD, Andresen MA, & Schmid TL. Obesity re-lationships with community design, physical activity,and time spent in cars. Am J Prev Med. 2004;27:87-96.
8. Burdette HL, Whitaker RC. Neighborhood play-grounds, fast food restaurants, and crime: Relation-ships to overweight in low-income preschool children.Prev Med. 2003;38:57-63.
9. Ellaway A, Anderson A, Macintyre S. Does area ofresidence affect body size and shape? Int J Obes RelatMetab Disord. 1997;21:304-308.
0. van Lenthe FJ, Mackenbach JP. Neighbourhood de-privation and overweight: The GLOBE study. Int JObes Relat Metab Disord. 2002;26:234-240.
1. Kinra S, Nelder RP, Lewendon GJ. Deprivation andchildhood obesity: A cross sectional study of 20,973children in Plymouth, United Kingdom. J EpidemiolCommun Health. 2000;54:456-460.
2. Kirtland KA, Porter DW, Addy CL, Neet MJ, WilliamsJE, Sharpe PA, Neff LJ, Kimsey CD, Ainsworth BE.Environmental measures of physical activity supports:Perception versus reality. Am J Prev Med. 2003;24:323-
331.Supplement to the Journal of the AMERICAN DIETETIC ASSOCIATION S117