Characteristics of the Built Environment Associated With Leisure-Time Physical Activity Among Adults in Bogotá, Colombia: A Multilevel Study

Transcription

1 Journal of Physical Activity and Health, 2010, 7(Suppl 2), S196-S Human Kinetics, Inc. Characteristics of the Built Environment Associated With Leisure-Time Physical Activity Among Adults in Bogotá, Colombia: A Multilevel Study Luis F. Gomez, Olga L. Sarmiento, Diana C. Parra, Thomas L. Schmid, Michael Pratt, Enrique Jacoby, Andrea Neiman, Robert Cervero, Janeth Mosquera, Candance Rutt, Mauricio Ardila, and José D. Pinzón Background: Even though there is increasing evidence that the built environment (BE) has an influence on leisure-time physical activity (LTPA), little is known about this relationship in developing countries. The objective of this study was to assess the associations between objective built environment characteristics and LTPA. Methods: A cross-sectional multilevel study was conducted in 27 neighborhoods in which 1315 adults aged years were surveyed. An adapted version of the IPAQ (long version) was used to assess LTPA. Objective BE characteristics were obtained using Geographic Information Systems. Associations were assessed using multilevel polytomous logistic regression. Results: Compared with inactive people, those who resided in neighborhoods with the highest tertile dedicated to parks (7.4% to 25.2%) were more likely to be regularly active (POR = 2.05, 95% CI = ; P = 0.021). Those who resided in neighborhoods with presence of TransMilenio stations (mass public transportation system) were more likely to be irregularly active (POR = 1.27, 95% CI = , P = 0.009) as compared with inactive people. Conclusions: These findings showed that park density and availability of TransMilenio stations at neighborhood level are positively associated with LTPA. Public health efforts to address physical inactivity should consider the potential influences of urban planning and mass public transportation systems on health. Keywords: urban health, active living, public parks The promotion of physical activity has been identified as an important strategy for facing the growing epidemic of chronic diseases. 1 According to the World Health Organization, the burden of chronic diseases is growing and 80% of worldwide mortality due to chronic diseases occurs in low and middle-income countries, generating premature deaths as well as significant social Gomez and Mosquera are with the División de salud, Fundacion FES Social, Bogota, Colombia. Gomez is also with the Facultad de medicina, Universidad Javeriana, Bogota, Colombia. Sarmiento is with the Dept of Social Medicine, Universidad de los Andes, Bogota, Colombia. Parra is with the Prevention Research Center in St. Louis and the George Warren Brown School of Social Work, Washington University in St. Louis. Schmid, Pratt, Neiman, and Rutt are with the Division of Nutrition and Physical Activity, Centers for Disease Control and Prevention, Atlanta, GA. Jacoby is with the Dept of Healthy Eating and Active Living, Non-communicable Disease Unit, Pan American/World Health Organization, Washington, D.C. Cervero is with the Dept of City and Regional Planning, University of California, Berkeley, Berkeley, CA. Ardila and Pinzón are with the Centro de Estudios Urbanos, Corporación de Universidades del Centro de la Ciudad, Bogota, Colombia. and economic burdens. 2 Colombia and its capital city of Bogotá are not exceptions, as chronic diseases such as coronary heart disease, stroke, and cancer are among the leading causes of mortality. 3,4 Despite the recognized health benefits of regular physical activity 5 7 the majority of the adult population in Bogota is inactive. 8 In fact, the last National Nutrition Survey from Colombia reported that 44.7% of the adult population in Bogotá met recommendations for physical activity and only 8.6% met these criteria during leisure time. 8 Prevalence of leisure time physical activity (LTPA) was even lower among women, people with low education levels, and those who resided in poor and disadvantaged neighborhoods. 8 Promotion of LTPA has special relevance for public health because of its well-established physical and mental health benefits. 9 LTPA has been strongly associated with perception of wellness and quality of life 10,11 and it may also contribute to social interactions that can bring about better community cohesion and increased social capital. 12,13 Ecological models emphasize the links between policy and environmental attributes with the promotion of physical activity. 14 In addition, studies conducted by S196

2 Built Environment and PA in Bogotá S197 transportation and urban planners have made important contributions to the knowledge of which attributes of the built environment are associated with active transportation (ie, walking and cycling). In the 3D model developed by Cervero and Kockelman, characteristics of the built environment are grouped in 3 dimensions: density, diversity, and design. 15 According to this model, people who reside in high-density neighborhoods have more opportunities to access destinations that encourage nonmotorized means of transportation. Diversity is related to mix land use and people who reside in neighborhoods with high diversity are more likely to walk or bike for transportation. Attributes of design include connectivity, density of road networks, and presence of parks and trees, among other characteristics. 15 To date, most research has explored the association between these 3 attributes of design with active transportation. 14 The factors that promote and motivate LTPA may differ from those that promote or discourage utilitarian physical activity. However, they may also be influenced by different combinations of personal and environmental factors. 15,16 Objective indicators obtained using geographic information systems (GIS), including proximity to parks, have been positively associated with meeting recommendations for LTPA. 17,18 In addition, a study conducted by Ewing et al, which used a sprawl index calculated from indicators of land use and street networks identified a significant correlation between this index and minutes walked during leisure time. 19 Little is known about the association between objective environmental characteristics and LTPA in the Latin American region. A better understanding of these connections in the context of the city of Bogotá will provide valuable guidance for future efforts to promote physical activity in the city and others urban settings in Latin America. In consequence, this study examines the associations between objective built environment characteristics and LTPA among adults residing in the urban area of Bogotá. Study Setting Methods Bogotá, the capital city of Colombia, has a population of approximately 7 million 20 and is located on a plateau at 2600 m above sea level. In the city, there have been a number of efforts to create policies aimed at changing social norms with the intention of increasing the mobility of citizens and recover public space. 21 For more than a decade, one of the main goals for the city is to recognize the rights of pedestrians, giving them priority over motor vehicles. 21 The city has implemented a number of urban changes that have included the construction of bicycle paths, recovery of public spaces, creation and improvement of public parks, and enhancement of existing recreational programs. 21 For example, from 2001 to 2003 Bogotá increased the availability of green area per inhabitant from 2.5 to 4.12 m 2. 21,22 The CicloRutas project is a network of approximately 300 km of bicycle paths. This network is in part connected with one of the public transportation systems of the city known as TransMilenio and with various parks of the city. 23 TransMilenio is a rapid mass transportation system of buses that operate in exclusive lanes and have fixed stations approximately every 500 m. 21 Another initiative that has given recognition to the city of Bogotá is the Ciclovía program in which 121 km of the main avenues of the city are closed to motor vehicles on Sundays and holidays from 6 AM to 2 PM and opened solely for pedestrians and cyclists. 21,23 Sample Areas and Study Population A multistage, cross-sectional, multilevel study was carried out during 2005 in the urban area of Bogotá among 30 neighborhoods selected as primary sampling units (mean area, 447,204 m 2 ; SD, 275,262 m 2 ; min = 51,476 m 2 ; max = 1,110,379 m 2 ). To ensure representativeness of community design, neighborhoods were selected after a prior stratification by socioeconomic status (SES) (low = 2, middle = 3 4, middle-high = 4), slope of the terrain (average slope 10% vs >10%), proximity to TransMilenio stations ( 500 ms vs >500 ms), and public park provisions ( 6% of total land devoted to parks vs >6%). The SES index was determined using the classification from the Bogotá Planning Department based on physical characteristics of the household and surrounding areas (ie, conditions and accessibility of the roads, presence of sidewalks, and construction materials of the house). After sampling stratification, 2 cells did not have neighborhoods. Once the neighborhoods were selected, 5 blocks were randomly selected in each neighborhood and 10 houses were randomly selected within each block. One adult aged years who had at least 1 year of residence was selected per household. Due to small sample size in 3 neighborhoods of high SES as a result of low response rates (0% 10%), only 27 neighborhoods with 1315 participants were finally included in the study representing a response rate of 66%. Characteristics of each neighborhood are included in Table 1. All the protocols and questionnaires were reviewed and approved the IRB of Universidad de los Andes in Bogota. All the participants were asked to provide informed consent before the survey. Outcome Variables A culturally adapted version of the long form of the International Physical Activity Questionnaire (IPAQ) was used to assess overall levels of physical activity by domain including leisure time physical activity. 24 Based on previous experience in administering the IPAQ in

3 S198 Gomez et al Table 1 List of the 27 Neighborhoods of the Study Population with the Selected Characteristics Used in the Sampling Design Neighborhood SES a terrain b Slope of Proximity to TransMilenio c % area dedicated to public parks d % 0 3.6% % 0 8.2% % % % 0 7.5% % 0 2.1% 6 3 2% % % 0 6.7% 8 2 2% % % 0 3.9% % 1 9.6% % 0 8.5% % % % 0 2.1% % 0 7.6% % 0 7.2% % 1 5% % 0 3.4% % 0 7.3% % % % 0 0.1% % 0 2.2% % 0 5.2% % 1 1.5% % 0 0.8% % 0 0.7% % 0 0.5% % 0 3.1% a SES = socio-economic status. b Average slope of the neighborhood was calculated in different topographic triangles in the terrain levels and an average of these values was determined. c Calculated as the existence of TransMilenio within the area of neighborhood. d Park area/land area 100. Colombia and from results of cognitive interviews, 25 changes in wording and the order of questions were made to reduce over reporting of activity that has been found in other IPAQ studies. In addition, the duration of activities in each of the days was also reported, which allowed the calculation of a daily average of physical activity. To validate the modified version of the long IPAQ, a subsample of 41 persons wore accelerometers (Uniaxial Computer Science and Application, Inc. s accelerometers, CSA-model-7164) for at least 5 days. We obtained a Spearman correlation coefficient of 0.42 (P =.006) between the scoring in metabolic equivalents (METS) calculated from the IPAQ and the accelerometers. The test-retest reliability of IPAQ was calculated among 147 adults and a Spearman correlation of 0.69 (P <.001) was obtained. Methods used in this validation analysis were consistent with the procedures followed by Craig et al 26 and the measurement properties are comparable with other validated questionnaires. 27 For this analysis, 3 LTPA categories were defined: regularly active (those who reported engaging in at least 30 minutes of LTPA per day for at least 5 days within the

4 Built Environment and PA in Bogotá S199 last 7 days), irregular active (those who reported at least 10 minutes of LTPA in the last 7 days, but did not meet the criteria to be regularly active), and inactive (those who reported less than 10 minutes of LTPA in the last 7 days). Characteristics of the Built and Natural Environment Measured by Geographical Information Systems (GIS) Measures of the built environment for this study were developed with data from the Cadastre Department of Bogota using Arc-View software (ArcInfo, version 9, Redlands, CA, Environmental Systems). Characteristics of the built environment were grouped in the 3 dimensions described in the model developed by Cervero 15 : density, diversity, and design. Using empirical evidence from a previous study conducted in Bogotá, 28 the following neighborhood measures were selected: housing density, land-use mix (index ranging from 0, which indicates 1 single land use, to 1 which indicates maximum heterogeneous land use), park density (percentage of land covered by parks), and the presence of Ciclovía routes, TransMilenio stations, and bicycle paths. In addition, the slope of the terrain as a natural environment attribute was included. Table 2 describes the operational definition of each variable and their distribution within the study neighborhoods. We conducted analyses at both the block and the neighborhood levels and found similar associations between LTPA and built environment attributes with the exception of park density which was only significant at the neighborhood level. In this manuscript we focus on the results from the neighborhood level analysis. Individual Characteristics Covariates included gender, age groups (18 35 yrs, yrs, and years), and level of education (secondary or less versus more than secondary). Statistical Analysis Objective environmental characteristics were based on the tertiles of the following indicators: housing density, land use-mix and park density. New binary variables for Ciclovía, TransMilenio and bicycle paths (present, not present) were created. Because the outcome variables of this study had 3 categories, a multilevel polytomous logistic regression was conducted using HLM6. 29 This analysis assumed a random intercept form, and regression coefficients were taken as fixed. 30 Results were presented as odds ratios with 95% confidence intervals. Environmental attributes were included in the final model when the P value in the bivariate analysis was less than SES was not Table 2 Environmental Measures Obtained by Geographic Information Systems (GIS) in the Selected 27 Neighborhoods Variable Definition Mean or % SD* VC** Min Max Housing density Number of housing units/total number of properties Land-use mix 1 ((Σi(pi)(lnpi))/lnk) where p = proportion of total land uses, i = category of land use, ln = natural logarithm, and k = number of land-use categories Park density Park area/land area Length of Ciclovía a Total length in meters in neighborhood Completeness of bicycle paths Bicycle path km/street network km TransMilenio stations b 0 1 or more 24 3 Slope Average slope of the neighborhood was calculated in different topographic triangles in the terrain levels and an average of these values was determined * SD = standard deviation. ** VC = variation coefficient. a Ciclovía= Program in which 121 km of the main avenues of the city are closed to motor vehicles on Sundays and holidays and opened solely for pedestrians and cyclists. b Total number of TransMilenio stations in neighborhoods.

5 S200 Gomez et al included in the model, as it was found to be significantly correlated with education level. All the models were adjusted for gender, age group, education level, slope of the terrain and by the environmental attributes finally included. Results Table 3 shows the sociodemographic characteristic of the study population and the distribution of LTPA. The mean age was 36 years (SD = 13.5), 43% of the sample were between years old. Sixty-five percent of participants were women and 76% had more than secondary. Slightly more than 47% of the respondents engaged in any leisure activity, and 9.2% were regularly active during leisure time. The average time of residence in the neighborhood for the total sample was 14.5 years. Table 4 includes results from the multilevel polytomous logistic regression models for being irregularly and regularly active in leisure time. After adjustment for covariates, the model showed that compared with inactive people, those who resided in neighborhoods with a park density between 7.4% and 25.2% were more likely to be regularly active (OR = 2.05, 95% CI = ). The same positive association persisted for being irregularly active but it was not significant at the 0.05 alpha level (OR = 1.33, 95% CI = ). As compared with inactive people, those who resided in neighborhoods with presence of TransMilenio stations were more likely to be irregularly active (OR = 1.27, 95% CI = ). Finally, residing in a neighborhood with a slope of the terrain of 4% or more, was negatively associated with being regularly active during leisure time (OR = 0.37, 95% CI = ). Discussion This study found that park density and access to Trans- Milenio were associated LTPA. In addition, living in a neighborhood with a slope of the terrain of 4% or more was negatively associated with being regularly active. The positive association between availability of parks and LTPA has been documented in several studies and enhances the importance of public parks in the promotion of active living in developing urban settings. 17,18 The association between the presence of Trans- Milenio stations and LTPA has been previously documented. 28 This finding may be explained in part by the urban interventions that have occurred alongside the construction of TransMilenio, which include the enhancement of pedestrian infrastructure such as sidewalks, cross walks and pedestrian bridges. To better interpret the results from this study it is important to understand Bogotá s cultural, economic, and urban characteristics. Bogotá is a city with high levels of housing density and land-use mix with relatively low Table 3 Sociodemographic Characteristics of the Study Population and Distributions of any Leisure-Time Physical Activity (LTPA) and Meeting Recommendations Through LTPA Among 1315 Adults Aged Years Characteristics Total sample (% or mean) (n = 1315) Any LTPA (% or mean) (n = 619) Regular active in LTPA (% or mean) (n = 121) Total 100% 47.1% 9.2% Mean age (yrs) 36 (SD = 13.5) 37.9 (SD = 14.2) 41.2 (SD = 14.6) Age groups (yrs) * % 52.7% 8.7% % 41.0% 7.9% % 45.5% 12.0% Sex ** ** Male 35.3% 58.8% 14.4% Female 64.7% 40.7% 6.4% Education level ** ** Basic level (complete or incomplete) 24.0% 36.8% 5.7% Secondary (complete or in complete) 47.2% 46.4% 8.1% More than secondary 28.8% 56.7% 14.0% Years of residence in the neighborhood 14.5 (SD = 11.7) 14.2 (SD = 11.4) 14.4 (SD = 10.4) * P <.01. ** P <.001.

6 Table 4 Multilevel Polytomous Logistic Regression Analysis for Being Irregular and Regular Active Versus Inactive in Leisure Time, Associated With Selected Built Environment Attributes Among 1315 Adults Aged Years Unadjusted models Irregular active Regular active Characteristics POR 95% CI P POR 95% CI P Slope of land Less than 4% (referent) 1 1 4% or more 0.71 ( ) ( ) Housing density 45.8 or less (referent) ( ) ( ) ( ) ( ) Existence of bike paths Yes 1.15 ( ) ( ) Land-use mix 0.48 or less (referent) ( ) ( ) ( ) ( ) Park density 4.3 or less (referent) ( ) ( ) ( ) ( ) Existence of Ciclovías a Yes 1.25 ( ) ( ) Existence of TransMilenio stations b Yes 1.39 ( ) ( ) Adjusted models Irregular active Regular active Characteristics POR CI 95% P POR CI 95% P Slope of land Less than 4% (referent) 1 1 4% or more 0.76 ( ) ( ) Existence of bike paths Yes 1.01 ( ) ( ) Park density 4.3 or less (referent) ( ) ( ) ( ) ( ) Existence of TransMilenio stations Yes 1.27 ( ) ( ) a Ciclovía = program in which 121 km of the main avenues of the city are closed to motor vehicles on Sundays and holidays and opened solely for pedestrians and cyclists. b TransMilenio = transport system which operates in exclusive lanes and with fixed stations located every 500 meters. Note. Participants in level 1 variables = Number of neighborhoods included in level 2 variables = 27. S201

7 S202 Gomez et al variability in both of these characteristics. This could contribute to explain the lack of association found with LTPA. Several limitations of this study should be noted. Some of the findings suggest that the study may have had a reduced power to obtain precise confidence intervals because of the modest sample size and the low response rate among high SES neighborhoods. The cross-sectional design of this study does not allow determining a causal relationship between characteristics of the built environment and physical activity. This study cannot rule out self-selection for example, residents who want to be active may select neighborhoods that provide easy access to recreational opportunities, however, in the US similar studies have reported that self-selection is only a partial factor in explaining differential levels of activities. 31 Taking into account that the majority of Bogota s citizens are from low or middle-low SES, and that the mean time of residency was 14 years, the decision of where to live may be more often based on economic factors rather than environmental attributes. Finally, we did not find associations between mix land-use and housing density. This could be explained by the following factors: a) density and mix land use already have very high levels with low variability; b) density was measured using housing density instead of population density due to the fact that the last census was conducted in 1993 and projections of population growth for the time of the survey were not accurate; as a result, population density could be even higher; c) mix land use was calculated using information from the Cadastro Department, however, there is a large number of informal businesses that are not being captured by this measurement (ie, street vendors); thus, land-use mix could be even higher. The survey instrument used in this study, which relied on self-reported information, did not allow us to determine which activities were carried out within or outside the neighborhood, which may be a concern in tying environmental attributes to LTPA. In addition, the boundaries of the neighborhood were determined by the city s administrative considerations and did not necessarily coincide with the perception of neighborhood boundaries that the study participants had. To better understand the links between attributes of the built environment and LTPA in the context of Latin American cities, new studies and methodological approaches should be undertaken to address these possible limitations. Although the provision of public parks may affect LTPA, factors such as the design of these parks, types of use, and community appropriation are potentially important. Researchers should consider including these variables in future studies. Results of this study also highlight the need to expand and refine the measurement of physical activity to include the possibility of identifying how environmental factors outside the neighborhood may also have an influence on patterns of physical activity. Despite the wide confidence intervals and the high random errors for some of the associations identified in this study, these findings have important relevance for Bogota and for other cities from developing countries, considering that small changes at the population level can result in substantial public health benefits. 32 To our knowledge this is the first study to explore the links between objective characteristics of the built environment and physical activity during leisure time in a city from Latin America. It should be recognized as a preliminary effort to understand this relationship in the context of a developing country. The multisectoral approach of this study, drawing upon the knowledge and experience of experts from public health, transportation, and urban design, may serve as a template for future research addressing the complex relationships between urban form and health. The findings of this study suggest that public health efforts to address physical inactivity and prevention of chronic disease in urban areas of developing countries should include consideration of the influences of urban planning and design and transportation systems on health. Acknowledgments This study was supported by a grant from the International Union for Health Promotion and Education and U.S. Centers for Disease Control and Prevention. The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention. References 1. World Health Organization. Fifty Seventh World Health Assembly. Global Strategy on Diet, Physical Activity and Health. Geneva: WHO; World Health Organization. Preventing Chronic Disease: A Vital Investment. Geneva: WHO; Gonzáles M, De la Hoz F. Mortalidad por enfermedades crónicas no transmisibles en Colombia, 1990 a Inf Quina Epidemiol Nac. 2002;7: Lobelo F, Pate R, Parra D, Duperly J, Pratt M. Burden of mortality associated to physical inactivity in Bogota, Colombia. Rev Salud Publica (Bogota). 2006;8(2, Suppl): Berlin JA, Colditz GA. A meta-analysis of physical activity in the prevention of coronary heart disease. Am J Epidemiol. 1990;132: Lee IM. Physical activity and cancer prevention data from epidemiologic studies. Med Sci Sports Exerc. 2003;35: Stewart KJ. Exercise training and the cardiovascular consequences of type 2 diabetes and hypertension: plausible mechanisms for improving cardiovascular health. JAMA. 2002;288: Instituto Colombiano de Bienestar Familiar. Encuesta Nacional de la Situación Nutricional en Colombia. Bogota: Instituto Colombiano de Bienestar Familiar; US Department of Health and Human Services. Physical Activity and Health: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human

8 Built Environment and PA in Bogotá S203 Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion; Schnohr P, Kristensen TS, Prescott E, Scharling H. Stress and life dissatisfaction are inversely associated with jogging and other types of physical activity in leisure time The Copenhagen City Heart Study. Scand J Med Sci Sports. 2005;15: Wendel-Vos GC, Schuit AJ, Tijhuis MA, Kromhout D. Leisure time physical activity and health-related quality of life: cross-sectional and longitudinal associations. Qual Life Res. 2004;13: Bedimo-Rung AL, Mowen AJ, Cohen DA. The significance of parks to physical activity and public health: a conceptual model. Am J Prev Med. 2005;28(2, Suppl 2): Lindstrom M, Hanson BS, Ostergren PO. Socioeconomic differences in leisure-time physical activity: the role of social participation and social capital in shaping health related behavior. Soc Sci Med. 2001;52: Sallis JF, Cervero R, Ascher W, Henderson KA, Kraft MK, Kerr J. An ecological approach to creating active living communities. Annu Rev Public Health. 2006;27: Cervero R, Kockelman K. Travel demand and the 3Ds: density, diversity, and design. Transp Res. 1997;2: Humpel N, Marshall AL, Leslie E, Bauman A, Owen N. Changes in neighborhood walking are related to changes in perceptions of environmental attributes. Ann Behav Med. 2004;27: Diez Roux AV, Evenson KR, McGinn AP, et al. Availability of recreational resources and physical activity in adults. Am J Public Health. 2007;97: Duncan M, Mummery K. Psychosocial and environmental factors associated with physical activity among city dwellers in regional Queensland. Prev Med. 2005;40: Ewing R, Schmid T, Killingsworth R, Zlot A, Raudenbush S. Relationship between urban sprawl and physical activity, obesity, and morbidity. Am J Health Promot. 2003;18: Departamento Administrativo Nacional de Estadística. Censo Available at: files/presultados.pdf 21. Parra D, Gómez LF, Pratt M, Samiento OL, Triche E, Mosquera J. Policy and environmental changes in Bogota and their relevance in promoting physical activity. Indoor Built Environ. 2007;16: Instituto Distrital de Recreación y Deporte. Artículo 242 Decreto No190. Instituto Distrital para la Recreación y el Deporte, División de Parques. Bogota: Instituto Distrital de Recreación y Deporte; Gomescáseres TM. Deporte, juego y paseo dominical: la recreación n en los espacios públicos urbanos, el caso de la Ciclovía a de Bogota. Monografia para optar por el título de Socióloga. (Sport, games and strolling around: recreation in the public urban spaces, the case of the Bogota Ciclovıa. Monograph submitted for the title of Sociologist). Bogota: Universidad Nacional de Colombia; International Physical Study. International Physical Activity Questionnaire, Available at: Willis G. Cognitive interviewing. A tool for improving questionnaire design. Thousand Oaks, London, New Delhi: SAGE Publications; Craig CL, Marshall AL, Sjostrom M, et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003;35: Sallis JF, Saelens BE. Assessment of physical activity by self-report: status, limitations, and future directions. Res Q Exerc Sport. 2000;71: Cervero R, Sarmiento OL, Jacoby E, Gómez LF, Neiman A. Influences of built environments on walking and cycling: Lessons from Bogotá. International Journal of Sustainable Transportation. 2009;3: Raudenbush SW, Bryk A, Congdon R. HLM: Hierarchical linear and nonlinear modeling (Version 5) [computer software]; Raudenbush SW, Bryk A. Hierarchical Linear Models. Applications and Data. Thousand Oaks, CA: Sage; Handy SL, Xinyu C, Patricia LM. Self-selection in the relationship between the built environment and walking. J Am Plann Assoc. 2006;72: Rose G. The Strategy of Preventive Medicine. Oxford: Oxford University Press; 1992.