| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH AND PRACTICE |
Ana V. Diez Roux and Latetia Moore are with the Center for Social Epidemiology and Population Health, University of Michigan, Ann Arbor. Kelly R. Evenson and Aileen P. McGinn are with the Department of Epidemiology, University of North Carolina, Chapel Hill. Daniel G. Brown and Shannon Brines are with the School of Natural Resources and Environment, University of Michigan, Ann Arbor. David R. Jacobs, Jr, is with the Division of Epidemiology, University of Minnesota, Minneapolis.
Correspondence: Requests for reprints should be sent to Ana V. Diez Roux, MD, PhD, Department of Epidemiology, 1214 South University, 2nd Floor, Ann Arbor, MI 48103 (e-mail: adiezrou{at}umich.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSData on Recreational ResourcesRESULTSDISCUSSIONReferences |
|---|
Objectives. Using data from a large cohort of adults aged 45 to 84 years-old, we investigated whether availability of recreational resources is related to physical activity levels.
Methods. Data from a multiethnic sample of 2723 adult residents of New York City, NY; Baltimore, Md; and Forsyth County, NC, were linked to data on locations of recreational resources. We measured the availability (density) of resources within 0.5 (0.8 km), 1, 2, and 5 miles of each participant’s residence and used binomial regression to investigate associations of density with physical activity.
Results. After adjustment for potential confounders, individuals in the tertile of participants residing in areas with the highest density of resources were more likely to report physical activity during a typical week than were individuals in the lowest tertile. Associations between availability of recreational resources and physical activity levels were not present for the smallest area assessed (0.5 miles) but were present for areas ranging from 1 to 5 miles. These associations were slightly stronger among minority and low-income residents.
Conclusions. Availability of resources may be 1 of several environmental factors that influence individuals’ physical activity behaviors.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSData on Recreational ResourcesRESULTSDISCUSSIONReferences |
|---|
Past work on residential environments and physical activity has been hampered by limited data on the specific features of residential environments that may be relevant.6 A growing body of recent work has begun to measure features of the physical environment using survey data as well as objective measures of the location of recreational facilities.7 The presence of a positive association between objective availability of resources and physical activity would suggest that improving spatial access to resources is an appropriate strategy to increase population levels of activity.
Evidence on whether availability of physical activity resources is an important predictor of physical activity behavior is mixed.8–22 Although important, existing research has focused largely on simple measures, such as distance to selected facilities or number of facilities within a given area,8–10,13–19,22 that do not account for the resources offered at a particular location. In addition, these studies have viewed space as discrete areas rather than a continuous field. Questions remain regarding the relevant spatial scale and sensitivity of empirical results to different spatial scales.
Using data from a large, multiethnic cohort of adults aged 45 to 84 years-old, we investigated associations between objective measures of the availability of recreational resources and physical activity. We used geographic information system methods to quantify the density (per area and per population) of physical activity resources weighted by the number and types of activities available at each location. We hypothesized that greater availability of recreational resources would be associated with a greater probability of residents being physically active. Because individual characteristics may result in individuals being more or less dependent on local resources, we also investigated whether any associations observed differed according to individual-level income or race/ethnicity.
In the absence of an a priori theory on how the distance one must travel to access recreational resources affects one’s use of those resources, we explored areas ("windows") of different sizes (0.5 miles [0.8 km], 1 mile, 2 miles, and 5 miles) around each person’s home. Our a priori assumption was that, within a window of a given size, resources closer to one’s residence would have more effect on physical activity than those further from one’s residence. Therefore, kernel densities (which assign more weight to resources closer to one’s residence than those closer to the boundary of the window) were our primary measure of resource availability. However, in addition, we examined whether results differed when simple densities (which assume equal effects of all resources located within the window) were used. Although the presence of a resource may affect physical activity regardless of the number of people who reside in the area, the presence of a larger population also implies more competition for the resources available. We therefore examined whether our results differed when density of resources was calculated in terms of population or simply in terms of area.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSData on Recreational ResourcesRESULTSDISCUSSIONReferences |
|---|
A semiquantitative questionnaire adapted from the Cross-Cultural Activity Participation Study24 was used to collect data on physical activity. In our analyses, we focused on the types of activities that we hypothesized would be linked to density of recreational resources: (1) team sports (e.g., softball, volleyball, basketball, soccer), (2) dual sports (e.g., tennis, racquetball, paddleball), (3) individual activities (e.g., golf, bowling, yoga, tai-chi), and (4) moderate- or heavy-effort conditioning activities (e.g., aerobics, bicycling, running, jogging, rowing, swimming, judo, karate). Participants who reported that they engaged in any of these 4 types of activities during a typical week were categorized as physically active. In some analyses, information on total number of minutes per week spent engaging in these activities was also used.
Data on covariates (age, gender, race/ ethnicity, income, and perceived neighborhood violence) were obtained from the baseline MESA examination as well. Race and ethnicity were classified using questions from the US census. Family income was grouped into 3 categories (less than $20 000, $20 000–$49 999, $50 000 or more). Participants rated the extent to which they perceived that violence was a problem in their neighborhood. Participants who reported that they exercised at least once a week were also asked about the frequency with which they exercised within 1 mile of their home. Home addresses for all participants were geocoded.
|
Data on Recreational Resources |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSData on Recreational ResourcesRESULTSDISCUSSIONReferences |
|---|
The types of resources available at each facility or park were recorded, and 48 different types of resources were identified. The distribution of these resources by broad types is shown in Table 1
. Information on counts (of tracks, roller and ice skating rinks, skate parks, pools, tennis courts, racquetball or squash courts, general sports fields, and baseball, cricket, and football fields) was obtained for some types of resources. By summing the total number of resources (weighted by the count when appropriate), we obtained variables for each point location or park that represented the total number of resources available at that location. In the case of parks, we assumed that any resources were evenly distributed in space over the park.
|
For comparison purposes, we also estimated simple densities by dividing the count of resources by the area of the window without assigning more weight to resources located closer to participants’ residences. To obtain resource density estimates adjusted for population density, we divided the resource density value by a population density value for the same window.26 We estimated population density using block population data and an approach identical to that used for recreational densities. These population-adjusted densities can be interpreted as number of resources per 100 000 population.
When the area for which density was estimated for a given participant fell partly outside the geographic areas for which data were collected, we applied a correction factor according to which we assumed that densities for the unobserved area were identical to those in the observed area. The percentages of participants for whom 80% of the density window was in the study area were 96%, 92%, and 77% for the 1-mile, 2-mile, and 5-mile windows, respectively. We estimated densities separately for team and dual sports, conditioning, and individual activities and then summed them to obtain an overall activity density measure. We also estimated densities separately for facilities with and without use fees.
Statistical Analyses
A total of 2742 MESA neighborhood study participants lived within the geographic areas for which physical activity resource data were collected. Of these individuals, 19 were excluded because they had missing physical activity data, leaving 2723 participants for the analysis. Population-adjusted densities were divided into 3 groups based on tertiles derived from the full sample. We used binomial regression to calculate relative prevalences of physical activity by density categories adjusted according to age, gender, race/ethnicity, and individual-level income.27 We used stratified analyses and included interaction terms in models to examine effect measure modification (multiplicative interaction).
We assessed associations between density and the amount of activity reported by physically active participants using linear regression analyses in which logged weekly minutes of reported activity was the outcome. We repeated selected analyses using kernel densities unadjusted for population density. We used site-specific tertiles (instead of full-sample tertiles) in these analyses because of the large differences in densities observed across sites when population density was not taken into account. Also, as mentioned, we repeated selected analyses using simple densities. All P values reported were derived from 2-tailed tests.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSData on Recreational ResourcesRESULTSDISCUSSIONReferences |
|---|
). The most common resources available at all 3 sites were team sports resources (27%–39% of all activities across sites), followed by aerobics/ cardiovascular equipment/weight training activities (17%–21%).
Overall, 45% of participants were younger than 65 years, and 46% were men (Table 2). The racial/ethnic composition of the sample was roughly similar to that of the geographic area from which each sample was drawn. Moderate- and heavy-effort conditioning activities were the most common activities reported (34.2% of participants overall). Other individual activities (13.1% of participants) and team or dual sports activities (4.3% of participants) were less common. These patterns were consistent across the 3 sites, with the exception of a lower prevalence of individual activities in New York. Of the study participants who reported being physically active outside of their home, a majority (64%) reported that they exercised within 1 mile of their home all or at least half of the time.
|
Table 3 shows physical activity prevalence ratios (PRs) by sociodemographic characteristics and tertiles of population-adjusted densities. Participants in the tertile with the highest density of resources were significantly more likely to report engaging in physical activities during a typical week than those in the lowest tertile (PR = 1.14; 95% confidence interval [CI] = 1.03, 1.26). In addition, when analyses were restricted to individuals who reported that they exercised regularly, density of recreational resources within a 1-mile radius was positively associated with the probability of exercising within 1 mile of one’s residence all or most of the time (Table 3).
|
). With the exception of the 5-mile window, similar results were obtained when resource densities were not adjusted for population density (in comparisons of the lowest and highest site-specific resource tertiles, prevalence ratios were 1.15 [95% CI = 1.04, 1.27] for 1 mile, 1.15 [95% CI = 1.04, 1.27] for 2 miles, and 1.06 [95% CI = 0.96, 1.18] for 5 miles).
|
), but this heterogeneity of associations was statistically significant (P<.1) only for race/ethnicity. Among those reporting activity, only 5-mile densities were positively associated with weekly minutes of activity (adjusted percentage differences in minutes for the top and middle tertiles vs the bottom tertile were 29% [95% CI=–2%, 71%] and 13% [95% CI=–4%, 33%], respectively).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSData on Recreational ResourcesRESULTSDISCUSSIONReferences |
|---|
Past work relating objective measures of the availability of recreational resources to physical activity has produced mixed results. In one of the first studies linking spatial accessibility of resources to physical activity, Sallis et al. documented a positive association between density of pay-per-use neighborhood facilities (defined in windows ranging in size from 1 to 5 km) and frequency of exercise among neighborhood residents.9 Other studies have reported an inverse association between distance to resources such as bikeways or parks and use of these resources or physical activity levels.8,10,13,21 Giles-Corti and Donovan11 found only weak and nonsignificant associations between access to natural and built facilities and physical activity among 1803 adults in Perth, Aus-tralia. Other studies have also documented null or weak associations between objective measures of availability of recreational resources and physical activity.14,16,18,19
Our study was unique in that we studied the relationship between a quantitative measure of the availability of specific types of resources (i.e., not simply the presence of facilities) and the probability that people reported physical activity. The use of a quantitative, objective measure (as opposed to participant reports) eliminated the possibility of same-source bias (i.e., physically active participants being more likely to report resources in their local area).
Associations appeared to be stronger among Hispanic and non-Hispanic Black participants than among non-Hispanic White participants, suggesting that certain groups may be more affected by local environmental features. Similar to earlier work by Sallis,9 there was some evidence in our data that associations between density and physical activity were stronger for pay-per-use activities (or were present only for these activities); however, as a consequence of the limited cost data available, it is difficult to draw conclusions from this finding. The presence of a fee may simply be a proxy for facility quality and attractiveness.
Previous work examining the spatial accessibility of recreational resources has involved the use of measures such as presence of a facility within a given area or distance measure (e.g., distance to nearest park).8–10,13–16,18,19,22 Giles-Corti and Donovan11 used a measure based on the gravity model28; that is, they estimated distances from each residential location to all facilities within a bounded study area. In contrast, we estimated densities for windows of different sizes. Our approach did not create artificial boundaries implying barriers to movement from one area to the next. However, it did require establishing a window size beyond which the density was assumed to be 0 (just as the gravity model requires defining a boundary beyond which the distance decay parameter is 0).
The kernel density method also assigns within-window weights that decline following a bivariate normal distribution. A priori, we think that this is reasonable given that if space affects resource use, we would expect resources located close to one’s residence to have a greater effect on one’s physical activity than those located closer to the boundary of the window. Use of the kernel density method was recently proposed in studying the spatial accessibility of primary care services.26 Our results did not vary substantially when kernel densities or simple densities were used, because of the very high correlations between the measures (all correlations were above 0.95). A limitation of our approach is that we did not account for transportation options and we used Euclidian distances rather than network distances.
There is scant information on how distance affects use of different types of recreational resources. Ideally, one would develop specific hypotheses about the size of the area that is relevant (which may vary for different types of resources) and collect data to test these hypotheses. In the absence of such a priori theory, studies must necessarily be exploratory. Our focus on the 1-mile window was consistent with the fact that the majority of physically active individuals in our sample reported using resources within 1 mile of their home; however, we also tested the sensitivity of our results to window size.
Giles-Corti and Donovan11 allowed for the possibility that the relationship of distance to facility use differs for different types of facilities by including different distance decay parameters for different facility types in their analysis of Australian data. Unfortunately, we are not aware of any data on how distance affects use of the different types of facilities we investigated in the US context. As additional information becomes available, it will be possible to refine our method by estimating densities for windows of different sizes on the basis of knowledge regarding how far people are willing to travel to use different resources.
Associations between density of resources and physical activity levels were present for windows ranging from 1 to 5 miles. Our data did not allow us to investigate larger windows. The presence of associations even for relatively large windows is compatible with the notion that, in a car-oriented culture, resources spread over relatively large areas may be relevant to behaviors. A limitation of our analyses is that they were based exclusively on home residences and did not include information on availability of resources around work locations. Also, we had limited cost data and no data on quality or attractiveness of facilities.
Another complex issue is whether density of resources should be adjusted for population density. There is little information on how population density affects use of resources. The sites we studied differed substantially in population density and density of resources. We think it is unlikely that the much larger unadjusted densities in New York reflect better access; they may simply reflect the much greater concentration of people and things in space. For this reason, we chose to investigate population-adjusted densities using an approach similar to one recently proposed for the study of accessibility to medical services.26
We examined the sensitivity of our results to population adjustment and found that results were similar for all windows except the 5-mile window. In the case of that window, no associations were observed when densities were not adjusted for population. Unadjusted and population-adjusted estimates of resource densities were less correlated for the larger than for the smaller windows. Our results suggest that population adjustment may be necessary to detect effects of availability on physical activity, but more work is needed to understand whether or under what circumstances population density affects use of resources.
Spatial accessibility of physical activity resources appears to be a positive, albeit weak, predictor of activity levels. Improved theory and data on the ways in which distance affects use of different types of resources will allow testing of much more specific hypotheses regarding the relationship between density of resources and physical activity. Intervention studies are also needed to determine whether increased availability increases levels of physical activity. Other characteristics, including quality and attractiveness of resources as well as features of the built environment that facilitate the use of public spaces for walking and exercise, may be as relevant as or more relevant than pure spatial accessibility of resources.
|
Acknowledgments |
|---|
We thank the other investigators, the staff, and the participants of the Multi-Ethnic Study of Atherosclerosis for their valuable contributions.
Human Participant Protection
The Multi-Ethnic Study of Atherosclerosis was approved by institutional review boards at the study sites. The analyses reported here were also approved by the University of Michigan institutional review board. All participants provided written informed consent.
|
Footnotes |
|---|
Contributors
A. V. Diez Roux drafted the article and conducted analyses. K. R. Evenson supervised data collection and assisted with the writing of the article. A. P. McGinn assisted with data collection and critically reviewed the article. D. G. Brown provided expertise on accessibility measures and critically reviewed the article. L. Moore assisted with data analyses and critically reviewed the article. S. Brines created geographic information system measures and critically reviewed the article. D. R. Jacobs, Jr, provided expertise on measurement of physical activity and critically reviewed the article.
Accepted for publication February 28, 2006.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSData on Recreational ResourcesRESULTSDISCUSSIONReferences |
|---|
2. Booth SL, Sallis JF, Ritenbaugh C, et al. Environmental and societal factors affect food choice and physical activity: rationale, influences, and leverage points. Nutr Rev.2001;59:S21–S39.[Web of Science][Medline]
3. French SA, Story M, Jeffery RW. Environmental influences on eating and physical activity. Annu Rev Public Health.2001;22:309–335.[CrossRef][Web of Science][Medline]
4. Diez Roux AV, Merkin SS, Arnett D, et al. Neighborhood of residence and incidence of coronary heart disease. N Engl J Med.2001;345:99–106.
5. Sundquist K, Winkleby M, Ahlen H, Johansson SE. Neighborhood socioeconomic environment and incidence of coronary heart disease: a follow-up study of 25,319 women and men in Sweden. Am J Epidemiol.2004;159:655–662.
6. Diez Roux AV. Residential environments and cardiovascular risk. J Urban Health.2003;80:569–589.[Web of Science][Medline]
7. Humpel N, Owen N, Leslie E. Environmental factors associated with adults’ participation in physical activity: a review. Am J Prev Med.2002;22:188–199.[CrossRef][Web of Science][Medline]
8. Duncan M, Mummery K. Psychosocial and environmental factors associated with physical activity among city dwellers in regional Queensland. Prev Med.2005;40:363–372.[CrossRef][Web of Science][Medline]
9. Sallis JF, Hovell MF, Hofstetter CR, et al. Distance between homes and exercise facilities related to frequency of exercise among San Diego residents. Public Health Rep.1990;105:179–185.
10. Wilson DK, Kirtland KA, Ainsworth BE, Addy CL. Socioeconomic status and perceptions of access and safety for physical activity. Ann Behav Med.2004;28: 20–28.[CrossRef][Web of Science][Medline]
11. Giles-Corti B, Donovan RJ. The relative influence of individual, social and physical environment determinants of physical activity. Soc Sci Med.2002;54: 1793–1812.[CrossRef][Web of Science][Medline]
12. Giles-Corti B, Donovan RJ. Socioeconomic status differences in recreational physical activity levels and real and perceived access to a supportive physical environment. Prev Med.2002;35:601–611.[CrossRef][Web of Science][Medline]
13. Troped PJ, Saunders RP, Pate RR, Reininger B, Ureda JR, Thompson SJ. Associations between self-reported and objective physical environmental factors and use of a community rail-trail. Prev Med.2001;32: 191–200.[CrossRef][Web of Science][Medline]
14. Evenson KHA, Huston S. Evaluating change in physical activity with the building of a multi-use trail. Am J Prev Med.2005;28:177–185.[CrossRef][Web of Science][Medline]
15. Reed JAB, Wilson D, Mixon G, Cook A. Awareness and use of community walking trails. Prev Med.2004;39:903–908.[CrossRef][Web of Science][Medline]
16. Hoehner CM, Brennan Ramirez LK, Elliott M, Handy S, Brownson R. Perceived and objective environmental measures and physical activity among urban adults. Am J Prev Med.2005;28:105–116.[CrossRef][Web of Science][Medline]
17. Sharpe PGM, Hutto B, Ainsworth B. Association of environmental factors to meeting physical activity recommendations in two South Carolina counties. Am J Health Promotion.2004;18:251–257.[Web of Science][Medline]
18. Rutt CCK. Examining the relationships among built environment, physical activity, and body mass index in El Paso, TX. Prev Med.2005;40:831–841.[CrossRef][Web of Science][Medline]
19. King WBS, Brach J, Simkin-Silverman L, Soska T, Kriska A. Objective measures of neighborhood environment and physical activity in older women. Am J Prev Med.2005;28:461–469.[CrossRef][Web of Science][Medline]
20. Giles-Corti B, Donovan RJ. Relative influences of individual, social environmental, and physical environmental correlates of walking. Am J Public Health.2003; 93:1583–1589.
21. Giles-Corti B, Broomhall M, Knuiman M, et al. Increasing walking: how important is distance to, attractiveness, and size of public open space? Am J Prev Med.2005;28:169–176.[Medline]
22. Troped PJ, Saunders RP, Pate RR, Reininger B, Addy CL. Correlates of recreational and transportation physical activity among adults in a New England community. Prev Med.2003;37:304–310.[CrossRef][Web of Science][Medline]
23. Bild D, Bluemke D, Burke G, et al. The Multi-Ethnic Study of Atherosclerosis (MESA): objectives and design. Am J Epidemiol.2002;156:871–881.
24. LaMonte MJ, Durstine JL, Addy CL, Irwin ML, Ainsworth BE. Physical activity, physical fitness, and Framingham 10-year risk score: the Cross-Cultural Activity Participation Study. J Cardiopulm Rehabil.2001;21:63–70.[CrossRef][Medline]
25. Gatrell A, Bailey T, Diggle P, Rowlingson B. Spatial point pattern analysis and its application in geographical epidemiology. Trans Inst Br Geogr.1996;21: 256–274.[CrossRef]
26. Guagliardo MF. Spatial accessibility of primary care: concepts, methods and challenges. Int J Health Geogr.2004;3:3.[CrossRef][Medline]
27. Robbins AS, Chao SY, Fonseca VP. What’s the relative risk? A method to directly estimate risk ratios in cohort studies of common outcomes. Ann Epidemiol.2002;12:452–454.[CrossRef][Web of Science][Medline]
28. Hansen WG. How accessibility shapes land use. J Am Inst Planners.1959;15:73–76.
This article has been cited by other articles:
|
|
 |
|
  C. K. Chow, K. Lock, K. Teo, S. Subramanian, M. McKee, and S. Yusuf Environmental and societal influences acting on cardiovascular risk factors and disease at a population level: a review Int. J. Epidemiol., March 4, 2009; (2009) dyn258v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. G. Bertoni, M. C. Whitt-Glover, H. Chung, K. Y. Le, R. G. Barr, M. Mahesh, N. S. Jenny, G. L. Burke, and D. R. Jacobs The Association Between Physical Activity and Subclinical Atherosclerosis: The Multi-Ethnic Study of Atherosclerosis Am. J. Epidemiol., February 15, 2009; 169(4): 444 - 454. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J Pearce, K Mason, R Hiscock, and P Day A national study of neighbourhood access to gambling opportunities and individual gambling behaviour J Epidemiol Community Health, October 1, 2008; 62(10): 862 - 868. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. E. Foraker, K. M. Rose, A. P. McGinn, C. M. Suchindran, D. C. Goff Jr, E. A. Whitsel, J. L. Wood, and W. D. Rosamond Neighborhood Income, Health Insurance, and Prehospital Delay for Myocardial Infarction: The Atherosclerosis Risk in Communities Study Arch Intern Med, September 22, 2008; 168(17): 1874 - 1879. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Clarke, J. A. Ailshire, M. Bader, J. D. Morenoff, and J. S. House Mobility Disability and the Urban Built Environment Am. J. Epidemiol., September 1, 2008; 168(5): 506 - 513. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. H. Auchincloss and A. V. Diez Roux A New Tool for Epidemiology: The Usefulness of Dynamic-Agent Models in Understanding Place Effects on Health Am. J. Epidemiol., July 1, 2008; 168(1): 1 - 8. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. R. Greenberg Contemporary Environmental and Occupational Health Issues: More Breadth and Depth Am J Public Health, March 1, 2007; 97(3): 395 - 397. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
