BERKE Et Al (2007)_ Association of the Built Environment With Physical Activity and Obesity in Older Persons

 RESEARCH AND PRACTICE  Association of the Built Environment With Physical Activity and Obesity in Older Persons | Ethan M. Berke, MD, MPH, Thomas D. Koepsell, MD, MPH, Anne Vernez Moudon, Dr es Sc, Richard E. Hoskins, PhD, MPH, Eric B. Larson, MD, MPH In the United States, obesity has been called an epidemic: an increasing proportion of Americans are overweight or obese.1,2 Numerous studies have highlighted the large proportion of overweight and obese adults, and the number of older adults
of 7

Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  American Journal of Public Health|March 2007,Vol 97,No. 3486|Research and Practice|Peer Reviewed| Berke et al.  RESEARCH AND PRACTICE  Objective. We examined whether older persons who live in areas that are con-ducive to walking are more active or less obese than those living in areas wherewalking is more difficult. Methods. We used data from the Adult Changes in Thought cohort study fora cross-sectional analysis of 936 participants aged 65 to 97 years. The Walkableand Bikable Communities Project previously formulated a walkability score topredict the probability of walking in King County, Washington. Data from the co-hort study were linked to the walkability score at the participant level using a ge-ographic information system. Analyses tested for associations between walkabilityscore and activity and body mass index. Results. Higher walkability scores were associated with significantly more walk-ing for exercise across buffers (circular zones around each respondent’s home)of varying radii (for men, odds ratio [OR]=5.86; 95% confidence interval [CI]=1.01,34.17 to OR=9.14; CI=1.23, 68.11; for women, OR=1.63; CI=0.94, 2.83 to OR=1.77;CI=1.03, 3.04). A trend toward lower body mass index in men living in more walk-able neighborhoods did not reach statistical significance. Conclusions. Findings suggest that neighborhood characteristics are associatedwith the frequency of walking for physical activity in older people. Whether fre-quency of walking reduces obesity prevalence is less clear.( Am J Public Health. 2007;97:486–492. doi:10.2105/AJPH.2006.085837) gender-specific analysis of physical activity,citing differences in the perception of environ-ment, convenience to destinations, and automo- bile use. 18,19 Indeed, older women appear totake fewer trips per day than do older men,indicating that the tendency to travel by anymeans, including walking, varies by gender. 20 Little is known about the association be-tween the built environment and activity andobesity in older men and women. Recently,new measurement tools have made study of the relation between obesity, physical activity,and the built environment possible. 17,21,22 Ourstudy explored whether more walkable neigh- borhoods are associated with more activityand less obesity in older men and women. METHODS Participants GroupHealthCooperativeisaconsumer-governed,staff-modelhealthmaintenanceor-ganizationinWashingtonStatewithmorethan500000members.Studyparticipants Association of the Built Environment With Physical Activityand Obesity in Older Persons |Ethan M. Berke,MD,MPH,Thomas D. Koepsell,MD,MPH,Anne Vernez Moudon,Dr es Sc,Richard E. Hoskins,PhD,MPH,Eric B. Larson,MD,MPH weredrawnfromtheAdultChangesinThought(ACT)study,aprospective,longitu-dinalcohortstudyofolderpatientsaimedat detectingtheonsetofdementia.TheACTstudybeganin1994andinitiallyenrolledapproximately2500randomlyselected,cog-nitivelyintactparticipantsaged65yearsorolderwhowereGroupHealthpatientsinclinicsservingwesternKingCounty.Detailsofthesampleareavailableelsewhere. 15,23 Approximately2000participantswereusedinouranalysis,correspondingtothesamplesizeavailableatthe2002assessment.Theincludedparticipantswerecognitivelyintact asdefinedbyaCognitiveAbilitiesScreeningInstrumentscoreof86orhigher, 24 whichcorrespondedtoaMini-MentalStatusExami-nationscoreof25to26orhigher.Allpartic-ipantsresidedinKingCounty, Washington,wherethegeographicmodelusedinthisanalysiscouldbeapplied.Datawereex-tractedfromACTstudyassessmentsoccur-ringbetweenJanuary1,2001,andDecem- ber31,2003,becausethisperiodmost In the United States, obesity has been calledan epidemic: an increasing proportion of Americans are overweight or obese. 1,2 Numer-ous studies have highlighted the large propor-tion of overweight and obese adults, and thenumber of older adults who are overweight or obese continues to rise. 3,4 Obesity has beenassociated with many health problems, includ-ing cardiovascular disease, diabetes, somecancers, depression, and arthritis. 2,5–10 Physicalactivity is believed to be an important deter-minant of health and body weight. Most Americans do not regularly engage in physicalactivity, 11 and efforts are being made nation-ally to increase the activity level of the popula-tion to prevent comorbid disease.Older people are at increased risk of de-cline in functional independence as they age.Of community-dwelling adults aged 75 yearsor older, 10% lose independence each year,as measured by activities of daily living. 12 Adecline in independence is associated with higher rates of hospitalization and mortality. 13 In addition to its inverse association with obe-sity, exercise is associated with a slowing infunctional decline 14 and dementia 15 and may help some older persons maintain functionalindependence. Older adults may choosewalking as a form of physical activity, bothfor recreation and as a means of transport forcompleting tasks of daily living.An older person’s activity level may be influ-enced by the built environment, which is de-fined by the Centers for Disease Control andPrevention as human-formed, developed, orstructured areas. 16 Neighborhood aesthetics,convenience to destinations, availability of pathsand sidewalks, and other environmental attrib-utes are believed to influence the walkability of a neighborhood. 17 As people age they mayspend a greater amount of time around their homes and have the opportunity to walk forexercise or for transportation; thus the study of the built environment in relation to activity andobesity is important. Some have called for  March 2007,Vol 97,No. 3|American Journal of Public Health Berke et al. |Peer Reviewed|Research and Practice|487  RESEARCH AND PRACTICE   TABLE 1—Predictors of Walkability in the Built Environment: Adult Changes in ThoughtStudy,King County,Washington,2001–2003 Environmental Characteristic (Threshold Value) a Odds Ratio b (95% Confidence Interval)Shorter distance to closest grocery store (<440 m)2.26 (1.12,4.56)*More dwelling units per acre of the parcel where the residence is located (>21.7)1.96 (1.15,3.35)*More grocery store,restaurant,or retail clusters in 1-km buffer (>1.8)1.70(1.11,2.60)*Fewer educational parcels in 1-km buffer (<5.1)1.55 (0.94,2.58)**Fewer grocery stores or markets within 1-km buffer (<3.7)1.50 (1.02,2.20)*Smaller size of closest office complex (<36659 sq m)1.28(1.08,1.53)*Longer distance to closest office/mixed-use complex (>544 m)1.27 (1.04,1.56)*Smaller size of block where residence is located (<23876 sq m)1.19 (0.99,1.43)** Source .Adapted from Moudon et al. 33 In press. Note .Pseudo- R 2 of full model=0.34 (using the Cox and Snell test). a  Threshold values of characteristics for active walking environments are derived from mean values for subjects walking>150min/wk vs not walking. b Odds of walking>150 min/wk vs not walking,using a straight-line measurement* P  <.05; ** P  <.1. closelycorrespondedtotheperiodinwhichgeographicdatawerecollected.Aftertheseselectioncriteriawereapplied,1967partici- pantswereeligible.Duringin-personvisitsconductedevery2 yearswithACTparticipants,informationwascollectedonactivityandobesity.Measured heightandweightwereusedtocalculatebodymassindex(BMI;weightinkilogramsdivided byheightinmeterssquared),acommonmea-sureofoverweightandobesity.Aself-reportof  physicalactivitywascollectedateachbiennialvisit.Awrittensurveyqueriedparticipantsonthenumberoftimesperweektheypartici- patedinvariousphysicalactivitiesforexercisethatlastedatleast15minutespersession.Thesurveyisdescribedinmoredetailinarecently publishedstudyoftherelationbetweenphysi-calactivityanddementia. 15 Ouranalysisusedthemeasureofwalkingforexercisefromthequestionnaire,withthequestion,“Duringthelastyear,howmanydaysperweekdidyouwalkforexerciseforatleast15minutesatatime?”Othermeasuresofactivity,suchasswimming,biking,andweightlifting,werenotusedinouranalysis,becausewefelttheywouldnotbesignificantlyinfluencedbyneigh- borhoodwalkability.TheACTstudyalsoprovidedseveralco-variatesthatmayconfoundtherelationbe-tweenneighborhoodwalkabilityandwalkingactivityandobesity,includingage,gender,educationlevel,income,livingalone,tobaccouse,andself-reportedinformationonarthri-tis.DepressionwasmeasuredusingtheCen-terforEpidemiologicalStudiesDepressionScale,a20-itemquestionnairevalidatedinanolderpopulation. 25,26 GroupHealthCoop-erativeprescriptionclaimsrecordswereusedtoassesschronicdiseaseburden.TheRxRisk score,derivedfrompharmaceuticaluseasanindicatorofdiseaseburden,wascalcu-latedforeachrespondent. 27 Geographic Data An earlier study, the Walkable andBikable Communities (WBC) project, pro-vided scores for neighborhood walkability.The WBC project, based in King County,Wash, and unrelated to the ACT study,identified components of the built environ-ment that contributed to increased walkingand biking. It used a behavioral model of environment and the travel-based principlesof srcin, destination, route, and area tostructure the environmental determinants of  physical activity. 28 Information from a tele- phone survey of 608 randomly selectedadult respondents in an 88-square-mile re-gion of urbanized King County was com- bined with objective tax parcel–level geo-graphic data from publicly availablesources. 29 The spatial sample frame con-sisted of medium- and high-density residen-tial areas of King County, with services(shops, schools, offices, etc.) close to homes.Details of the spatial sample frame con-struction are described elsewhere. 30 The27-minute survey included questions fromthe Behavioral Risk Factor SurveillanceSystem, International Physical ActivityQuestionnaire–Long, and National HealthInterview Survey, as well as additional ques-tions described by Brownson et al. 31 The WBC study then captured data onapproximately 200 directly observableneighborhood attributes with 900 relatedmeasures within 1-km and 3-km circularzones (buffers) around each respondent’s home and measured distance to destinationsup to 3km from a respondent’s home. Ob- jective measures of the built environment included land-use characteristics from the parcel-level assessor’s files, park informa-tion, streets and foot and bike trails, landslope, vehicular traffic, and public transit data. The WBC study estimated, by multino-mial logistic regression, the likelihood of walking more than 150 minutes per week,corresponding to the Centers for DiseaseControl and Prevention recommendationsfor sufficient physical activity, 32 versus not walking at all or walking moderately (<150minutes per week).We used variables from the survey in a2-step modeling process to create a basemodel. Those survey variables found statisti-cally significant or considered theoreticallyimportant were kept in the final models. Thevariables retained from the first step plus 200environmental variables were used for thesecond step. Two final models were created:1 for straight-line distances from the respon-dents’ homes (i.e., as the crow flies) and 1 fornetwork distances along existing transporta-tion routes (i.e., traveling along the streets). Of the 200 objective environmental variablesassessed, 8 were found to have a significant effect on walkability in the straight-line modeland were used to compute the walkabilityscores (Table1). Details of the methods usedto derive the walkability index are describedelsewhere. 21,22,34 Finally, we calculated walkability scores forthe entire surface of the spatial sample frame.This surface model was based on the finalstraight-line model. We controlled for surveyvariables and calculated walkability scores forthe significant environmental features of the  American Journal of Public Health|March 2007,Vol 97,No. 3488|Research and Practice|Peer Reviewed| Berke et al.  RESEARCH AND PRACTICE  Note.  The more walkable neighborhood has a denser street network and better connectivity of streets than does the less walkable neighborhood.Although the less walkable neighborhood appearsto have more retail destinations,it is beyond the distance a respondent would be expected to walk. FIGURE1—Example analysis of participants in more walkable (a) and less walkable (b) neighborhoods. a b respondents’ home locations and for addi-tional points on a 1-km grid within the spatialframe. To obtain values for the continuoussurface, we used a radial basis function to in-terpolate the values of areas between the points, thereby creating a smoothed-surfacemodel.We used a geographic information system(ArcView 9.0, ESRI, Redlands, Calif) togeocode each ACT participant’s address tothe associated tax assessor’s parcel. If the ad-dress could not be geocoded with parceldata, King County street file data were usedto geocode the address. Circular buffers of 100, 500, and 1000 m were created aroundeach point (Figure1). Buffer sizes were rep-resentative of distances usually traveled onfoot, with smaller buffers representing dis-tances that may be more commonly traveled by older people. The 1000-m buffer corre-sponded to other analyses of behavior andthe built environment. 17,22 The model com- puted walkability scores on a scale of 0 to100 for each subject within the area of each buffer. These walkability data were thenmerged with the respondent data from ACTfor analysis. Statistical Analysis Participants were, a priori, stratified into 4gender-specific groups: those who lived at the same address 2 years prior to their clini-cal assessment (men and women) and thosewho moved to a new address in the 2 yearssince their last assessment (men andwomen). Only participants living in the same home for at least 2 years were included inthe analysis of BMI (n=740), because we hypothesized that the effect of the built envi-ronment on a change in BMI might takelonger than 2 years to detect. All 4 groupswere included in the analysis of self-reportedwalking (n=936), because adaptation of this behavior would be expected in less than 2 years. We chose to stratify participants ongender because previous research showeddifferent patterns of walking for activity be-tween men and women. 18–20,35 We used t  tests and χ 2 analyses to analyzedifferences between men and women and be-tween those living at the same address and adifferent address 2 years prior to the study.Multiple logistic regression was used to de-termine the associations between neighbor- hood walkability and self-reported walkingand BMI. Regression analyses controlled forCenter for Epidemiological Studies Depres-sion Scale score, income, education, tobaccouse, living alone, self-report of arthritis, age,and RxRisk as a measure of chronic disease burden. Statistical analyses were performedwith Stata version 9 (Stata Corp, CollegeStation, Tex).  March 2007,Vol 97,No. 3|American Journal of Public Health Berke et al. |Peer Reviewed|Research and Practice|489  RESEARCH AND PRACTICE   TABLE 2—Participant Characteristics: Adult Changes in Thought Study,King County, Washington,2001–2003  Total Women Men,(N=936)(n=601; 64.2%)(n=335; 35.8%) Age,y,* mean (SD)78.5 (6.1)78.9 (6.1)77.8 (6.0)65–7427.224.631.975–8454.454.953.485–9418.120.014.6 ≥ 950.30.50CES-D score,** mean (SD)5.8 (6.5)6.4 (6.9)4.7 (5.4)RxRisk,$,** mean (SD)4142.1 (2307.9)3924.3 (1422.7)4532.6 (2223.3)Income >$30000,** %49.337.669.1Education,>12 y,%69.768.871.2Lives alone,** %45.555.128.3Uses tobacco,**%10.44.720.3Suffers from arthritis,%,kg/m 2 ,mean (SD)27.0 (5.0)27.0 (5.7)27.1 (3.6)Overweight or obese,** %63.259.370.3Walks any amount for exercise,%48.446.150.9Living in same home at least 2 years, a %79.177.981.2 Note .CES-D = Center for Epidemiological Studies Depression Scale; RxRisk = chronic disease burden: BMI = body mass index. a Persons living at the same address for at least 2 years were younger ( P  =.007) and had higher RxRisk scores ( P  <.001) thandid those living at a different address 2 years prior to the study.* P  =.01,comparing men and women; ** P  <.001,comparing men and women. RESULTS Of the 1967 potentially eligible respon-dents in the data set, 1770 participants weresuccessfully geocoded with tax parcel dataor King County street file data (90%). Of those participants, 936 (53%) were livingwithin the spatial sample frame of the WBCsurface model. Only the latter were studied.The remainder of the participants eitherlived outside the spatial sample frame (n=637) or were unsuccessfully geocoded be-cause of missing or incorrect address infor-mation (n=197).Participants ranged in age from 65 to 97 years, with a median age of 78 years. BMIranged from 14.2 to 65.4, with a medianBMI of 26.3. As a group, approximately 63%of participants had BMIs in the overweight (25.0–29.9kg/m 2  ) or obese (30.0kg/m 2 ormore) range, and approximately half reportedno walking for exercise (Table2). About 1 in5 participants had moved to a new home inthe 2 years prior to the assessment. A higher proportion of men than women reported in-comes of $30000 or more and tobacco use.Women were older and had a lower RxRiskscore, indicating less chronic disease burden, but had higher Center for EpidemiologicalStudies Depression Scale scores and moreoften lived alone.A statistically significant association wasdetected between neighborhood walkabilityand any self-reported weekly walking sessionsin men and women living at a different ad-dress in the 2 years prior to assessment, re-gardless of buffer size, and in women but not in men living at the same address for 2 yearsor more. Odds ratios of any self-reportedwalking for the difference between the 75th percentile and 25th percentile neighborhoodwalkability scores (interquartile range) are re- ported in Table3.There was no significant association be-tween higher neighborhood walkability andthe proportion of participants in the over-weight or obese range, although in most comparisons the association was in the hy- pothesized direction. Participants living in adifferent home 2 years prior to assessment didnot exhibit an association between BMI andneighborhood walkability (data not shown). DISCUSSION Our study suggests that the built environ-ment, as described by a neighborhood walka- bility score, is associated with increased walk-ing for exercise in men and women. Modelsof walkability that take into account types of and distance to destinations and residentialdensity may be a useful predictor of physicalactivity in older adults. The association wasseen at several buffer sizes representing po-tential distances traveled by older people. If this finding is confirmed by other studies, theassociation between neighborhood walkabilityand physical activity may be adapted for use by community planners, health care provid-ers, and older people. Planners could chooseto design neighborhoods that are more walka- ble, with both transportation and recreationdestinations. Health care providers could tai-lor specific activity recommendations, takinginto account where the patient lives. Olderadults may use information on neighborhoodwalkability as they select a new residence orcommunity after retirement.Wefoundnostatisticallysignificantassoci-ationbetweenthebuiltenvironmentandobesityinthosewhohaveremainedinthesamehomefor2yearsormore.Itispossiblethatahypothesizedlagtimeof2yearswasinsufficienttodetectanassociation.Otherre-searchershaveusedseveraldifferentmetricstofindvaryingstrengthsofassociationbe-tweenthebuiltenvironmentandweight, 36–41 indicatingthatadditionalstudyofthisrela-tioniswarranted.Other studies have found similar associa-tions between physical activity and walkabil-ity, isolating net residential density, street connectivity, and land-use mix as significant measures. 42–44 The WBC walkability modelused in this study encompasses these vari-ables and provides precise measurements aswell as additional information about the typeof land-use mix that optimizes walking. Themodel showed that proximity to grocerystores, smaller block sizes, and higher residen-tial density at the level of the respondent’s parcel was associated with more walkingwithin the neighborhood. Clusters of destina-tions, such as grocery stores, restaurants, andretail, also increased the odds of walking suffi-ciently to meet Centers for Disease Control
Related Documents
View more
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks