Evidence for urban design and public health policy and practice: Space syntax metrics and neighborhood walking
Gavin R. McCormack a,b,c,*, Mohammad Javad Koohsari d,e,f, Liam Turley a, Tomoki Nakaya g, Ai Shibata h, Kaori Ishii f, Akitomo Yasunaga i, Koichiro Oka d
Abstract
Most walkability indices do not capture the topological structure of urban forms. Space syntax models these topological relationships. We estimated associations between the space syntax measure of street integration and walkability (SSW) and neighborhood-specific leisure (LW) and transportation (TW) walking among 4422 Canadian adults. Street integration and SSW were found to be positively associated with TW and LW participation in a usual week. A one-unit increase in SSW was associated with a 6-min increase in usual weekly minutes of TW. Street integration and SSW were also positively associated with TW participation in the last week. Higher street integration and walkability conceptualized using space syntax support neighborhood walking.
Keywords:
Built environment Physical activity
Street configuration
Pedestrian
Urban form
Street pattern
1. Introduction
Neighborhood built features support or even impede walking (Salvo et al., 2018; Farkas et al., 2019; Saelens and Handy, 2008). Street connectivity, land use diversity, residential density, and overall walkability support physical activity (Salvo et al., 2018; Farkas et al., 2019; Saelens and Handy, 2008; Hajna et al., 2015). Often, walkability indices include relative or absolute counts of built features within a neighborhood (Hall and Ram, 2018; Maghelal and Capp, 2011), however these indices do not always reflect the configuration and topological structure of urban forms. Topological structure of urban forms refers to how open spaces (i. e., streets and parks) relate to each other to form the overall open space network. In the topological view, the number of turns required to reach a destination within a network is used, instead of simple metric distance. Space syntax provides a different approach, focusing on the topological aspect of urban forms for estimating supportiveness of the built environment for walking. Space syntax takes into account the topology of street layouts, in particular, the configuration between open spaces and built-up areas (Hillier and Hanson, 1984). To obtain the space syntax measures, “axial lines” are estimated which represent lines of sight. The configuration of these axial lines is used to estimate the “street integration” – a key space syntax measure (Hillier, 2009). In relation to the measurement of the 3D’s (density, diversity, and design) of active urban design, the space syntax concept corresponds primarily with estimating diversity and design (Koohsari et al., 2016a). Application of space syntax in physical activity research remains novel, and among the few existing studies, several have found associations between integration and other space syntax metrics and walking (Koohsari et al., 2016a, 2016b, 2017b; Baran et al., 2008; Wineman et al., 2014).
Space syntax theory is intuitive, reflecting human movement decision-making, and includes metrics, such as street integration, that can inform urban design policy (Cutumisu and Spence, 2009; Koohsari et al., 2014; Karimi, 2012). Nevertheless, similar to most previous built environment-physical activity studies (McCormack and Shiell, 2011), studies estimating associations between space syntax metrics and walking have not controlled for residential self-selection that can bias estimates. Furthermore, studies have linked space syntax metrics with pedestrian behavior captured via travel surveys (Baran et al., 2008) and general walking captured using questionnaire (Koohsari et al., 2016a, 2016b, 2017b; Wineman et al., 2014), but none have used neighborhood-specific measures of walking. Considering these two factors when estimating associations between the built environment and physical activity is important for informing causal inferences. Notably, there are no studies, which have examined associations between space syntax metrics and walking in Canadian adults (Farkas et al., 2019). Our application of space syntax in the current study responds to calls for more policy-relevant research in the built environment-physical activity field (Raine et al., 2012; Giles-Corti et al., 2015). Specifically, we estimated the associations between space syntax metrics and neighborhood transportation (TW) and leisure walking (LW) in adults, adjusting for residential self-selection, neighborhood tenure, and sociodemographic characteristics.
2. Methods
2.1. Study and sample design
Details of the study methods are available elsewhere (McCormack et al., 2012; Jack and McCormack, 2014; McCormack et al., 2014; McCormack et al., 2010). Briefly, using random digit dialing of residential land-line telephone numbers, two independent cross-sectional samples of Calgary (Canada) residents 18 years of age were recruited for interviews during August–October 2007 (summer/early autumn; n ¼ 2199, response rate ¼ 33.6%) and January–April 2008 (winter/early spring; n ¼ 2223, response rate ¼ 36.7%). Of participants who completed the telephone-interview and agreed to follow-up (n ¼ 3602), 2006 completed a postal survey. The telephone and postal surveys captured, among other information, physical activity, sociodemographic characteristics, reasons for residential self-selection, and residential postal code. The University of Calgary Conjoint Health Research Ethics Board approved this study (REB# 20798).
2.2. Data collection
2.2.1. Physical activity
In the telephone survey, the Neighborhood Physical Activity Questionnaire (NPAQ) (Giles-Corti et al., 2006) captured participation (<10 min/week “none” versus 10-min/week “any”) and minutes of transportation and leisure walking (TW and LW, respectively) undertaken inside the neighborhood (10–15 min walk from home) during a usual week. The NPAQ walking items are reliable among Canadian adults (McCormack et al., 2009).
In the postal survey, a modified version of the International Physical Activity Questionnaire (“Neighborhood-specific IPAQ” or NIPAQ) (Frehlich et al., 2018) captured participation (<10 min/week “none” versus 10-min/week “any”) and minutes of neighborhood TW and LW in the last 7-days. The original NIPAQ, which does not define a specific neighborhood boundary for respondents, is reliable and valid in Canadian adults (Frehlich et al., 2018, Frehlich et al., 2018). For consistency with the NPAQ, in our study the NIPAQ items captured walking undertaken within a 10–15 min of home. Given the differences in the psychometric properties of usual and past week self-report physical activity questionnaires (Doma et al., 2017), we included both NPAQ (usual week) and NIPAQ (last 7-days) captured walking in our analysis.
2.2.2. Residential self-selection
Of the original 19 telephone survey items that captured participant’s reasons for residing in their neighborhood (response options: not at all important, somewhat important, and very important), 16 loaded on one of four residential self-selection scales (access to places for physical activity, access to services, sense of community, and ease of driving) and all had acceptable reliability (Jack and McCormack, 2014; McCormack et al., 2012).
2.2.3. Built environment
Participant’s 6-digit residential postal codes were geocoded (Google Maps Platform, 2018). A street integration score was calculated for each street segment considering all the other street segments within a 1.6 km distance from its center. A street with a higher integration score requires fewer turns to be reached compared with lower street integration scores. Using Axwoman (Jiang, 2012) and DepthMap (Turner, 2004), we calculated street integration from street centerline data (DMTI Spatial Inc, 2008). We linked postal codes to dissemination areas to estimate gross population density for each participant’s neighborhood (using the 2006 Canadian Census). Informed by previous research (Koohsari et al., 2016a), we standardized (z-score) gross population density and street integration and summed them to create a composite Space syntax walkability (SSW) index. Notably, SSW is positively correlated with other measures of walkability (Koohsari et al., 2016a), and street integration is positively associated with street intersections and availability of local destinations (Koohsari et al., 2016b) and Walk Score® (Koohsari et al., 2017a).
2.2.4. Sociodemographic characteristics
The telephone interview captured the participant’s gender, age, the highest level of education completed, home ownership status, number of children <18 years of age at home, and length of neighborhood residence (tenure). 2.3. Statistical analysis We estimated descriptive statistics (mean standard deviation; frequencies) for the sample. Generalized Linear Models (GLM; log distribution with binary link function) estimated odds ratios (OR) and 95 percent confidence intervals (95CI) for associations of street integration and SSW with achieving: 1) any (10 min/week) neighborhood TW, and; 2) any neighborhood LW. Among participants reporting 10 min/ week, another type of GLM (gamma distribution with identity link function) estimated unstandardized beta coefficients (b) and 95CIs for associations of street integration and SSW with weekly minutes of neighborhood TW and LW. We estimated separate models for NPAQ and NIPAQ walking outcomes and models adjusted for residential self- selection, neighborhood tenure, and sociodemographic characteristics, and season. Statistical significance was achieved at p < .05. We used SPSS (version 24) for the analysis.
3. Results
3.1. Sample characteristics
A total of 4033 participants provided complete telephone survey data (NPAQ plus covariates) of which 1874 postal survey participants also provided NIPAQ data. The telephone survey sample consisted mostly of women (59.7%), university educated (41.9%), those with no dependents (62.7%), and homeowners (81.5%), with a mean age and neighborhood tenure of 47.1 15.6 years and 11.2 11.3 years, respectively. The subset of postal survey participants also consisted mostly of women (62.2%), university educated (44.7%), those with no dependents (66%), and homeowners (86.3%), and had a mean age and neighborhood tenure of 50.8 15.3 years and 12.9 12.0 years, respectively. Mean street integration was 191.14 65.79 (minimum ¼
17.63 and maximum ¼ 365.52) and mean SSW was 0.01 2.50 (minimum ¼ 6.0 and maximum ¼ 21.5). Estimated from the NPAQ, 74.9% participated in LW and 59.1% in TW inside their neighborhood in a usual week (Table 1). Estimated from NIPAQ, 58.8% participated in LW and 40.2% in TW inside their neighborhood in the last week (Table 1). Agreement between NPAQ and NIPAQ TW (kappa ¼ 0.40, p < .001; overall agreement ¼ 68.9%) and LW (kappa¼0.38, p<.001; overall agreement¼71.4% was moderate), respectively. Among participators, NPAQ and NIPAQ mean neighborhood TW was 121.2146.0 min/wk and 106.6117.4 min/wk, respectively (Spearman’s rank correlation ¼ 0.54, p < .001). NPAQ and NIPAQ mean neighborhood LW was 186.3 177.6 min/wk and 166.7 159.7 min/wk, respectively (Spearman’s rank correlation ¼ 0.60, p < .001).
3.2. Street integration and neighborhood walking
NPAQ. Adjusting for all covariates, street integration was positively associated with usual weekly participation in neighborhood LW (OR 1.00; 95CI 1.00, 1.01, p ¼ .011) (Table 1). Adjusting for all covariates, street integration was also positively associated with usual weekly participation (OR 1.01; 95CI 1.00, 1.01, p < .001) and duration of neighborhood TW (b 0.22; 95CI 0.15, 0.28, p < .001). NIPAQ. Adjusting for all covariates, street integration was not significantly associated with participation (p ¼ .625) nor duration (p ¼ .434) of neighborhood LW in the last week (Table 1). However, street integration was positively associated with participation in neighborhood TW in the last week (OR 1.01; 95CI 1.00, 1.01, p < .001).
3.3. SSW and neighborhood walking
NPAQ. Adjusting for all covariates, SSW was positively associated with usual weekly participation in neighborhood LW (OR 1.04; 95CI 1.01, 1.07, p ¼ .024) (Table 1). Adjusting for all covariates, SSW was also positively associated with usual weekly participation (OR 1.14; 95CI 1.11, 1.18, p < .001) and duration of neighborhood TW (b 6.82; 95CI 5.01, 8.64, p < .001). NIPAQ. Adjusting for all covariates, SSW was not significantly associated with participation (p ¼ .800) nor duration (p ¼ .176) of neighborhood LW in the last week (Table 1). However, SSW was positively associated with participation in neighborhood TW in the last week (OR 1.15; 95CI 1.10, 1.21, p < .001).
4. Discussion
We found higher street integration and SSW to be associated with an increased likelihood of participating in neighborhood TW in the last week, and in neighborhood LW and TW in a usual week. The magnitude of the estimated associations was stronger for neighborhood TW regardless of recall format, reflecting previous evidence showing consistent associations between the built environment and TW (Saelens and Handy, 2008; Farkas et al., 2019). Moreover, for each 1-unit increase in SSW, neighborhood TW in a usual week increased by 6.8 min, while for street integration, a 1-unit increase was associated with a 0.22 min increase in usual neighborhood TW. Notably, these associations were present even after adjusting for residential self-selection, neighborhood tenure, and sociodemographic characteristics. We found no other associations between walking duration and space syntax metrics.
To our knowledge, this is the first study examining the association between space syntax measures and neighborhood-based walking in adults and the first to link space syntax metrics with physical activity in the Canadian context (Farkas et al., 2019). Similar to others (Koohsari et al., 2016a, 2016b, 2017b; Baran et al., 2008; Wineman et al., 2014), we found our space syntax variables were associated with walking, however we did so using neighborhood-based measures of walking captured by the NIPAQ and NPAQ. Linking specific contexts (e.g., neighborhood characteristics) with specific behaviors (e.g., neighborhood walking) is considered a strength in terms of causal inference and implications for policy (Giles-Corti et al., 2005). Our findings suggest that population density and the layout and connectedness of streets are important neighborhood features that might influence walking. Notably, only the street network was used to estimate street integration, however incorporating sidewalk and formal and informal pathway networks might improve the ability of street integration and SSW to predict local walking. In these same data, we have found other measures of the built environment associated with neighborhood walking (McCormack et al., 2012; Jack and McCormack, 2014; McCormack, 2017). The consistency in findings across multiple studies that use the same data but different operational definitions of the built environment variables provides confidence that the built environment is important for supporting walking in the Calgary context. However, the findings of the current study are also novel given that it included an application of Space Syntax theory using spatial data that are readily available (street networks) and thus replicable across settings.
Associations between space syntax street integration and walkability, and weekly neighborhood leisure and transportation walking. NPAQ: Neighborhood Physical Activity Questionnaire (usual weekly walking); LW: leisure walking; SSW: Space Syntax Walkability Index; TW: transportation walking; NIPAQ: Neighborhood-modified International Physical Activity Questionnaire (last week walking); OR: odds ratio; 95CI: 95 percent confidence interval; b: unstandardized beta coefficient. OR (binary with log link function) and b (gamma with identity link function) estimated using generalized linear regression. All estimates adjusted for gender, age, education, home ownership, children <18 years of age at home, neighborhood tenure, residential self-selection, and season.
Despite adjusting for residential self-selection, neighborhood tenure and other important covariates, we cannot infer causality from this cross-sectional analysis. Self-reported physical activity can be biased (Sallis and Saelens, 2000; Durante and Ainsworth, 1996), however collecting walking data using two different types of recall (last week and usual week) from the same respondents 1–2 weeks apart adds confidence in terms of the consistency of our findings. Our definitions of the neighborhood used for the self-report physical activity items (10–15 min walk of home) and for the space syntax measures (1600-m radial distance) while overlapping, may not completely correspond spatially. Notably, Frehlich et al., 2018 found that NIPAQ-captured physical activity was more strongly associated with physical activity estimated from combined accelerometer and global position systems data that was collected within 400-m of home than for larger neighborhood boundaries.
As might be expected, the NIPAQ (last week recall), resulted in smaller estimated prevalence and minutes of both TW and LW compared with these estimates when captured by the NPAQ (usual week recall). While not focusing on context-specific behavior, Doma et al. (2017) found last week self-reported physical activity to have better convergent validity compared with usual week self-reported physical activity. Last week recalls likely more accurately capture recent physical activity, however they are more sensitive to week-to-week variations in physical activity that could result from sudden external factors (e.g., weather, illness, and injuries), while usual week recalls might reflect habitual physical activity levels and be less sensitive to these external factors (Doma et al., 2017). Despite differences in estimated prevalence and duration between NPAQ and NIPAQ data, odds ratios for the association between the built environment variables and TW participation were of similar magnitude and direction. Regardless of whether the NIPAQ or NPAQ is used, the association between LW and TW participation and the built environment is likely to be similar. Nevertheless, different findings were observed for the duration of TW and LW. People may more accurately report duration of neighborhood TW, as this activity is likely undertaken regularly and consistently (same trip route, same time of day, same amount of time etc.) from week-to-week. However, time spent participating in neighborhood LW may be less consistently undertaken in terms of selected routes, time of day, total amount of time, and week-to-week patterns, thus making it more difficult for the participant to recall duration with high accuracy regardless of whether last or usual week recall is used. Our findings serve as an important reminder to consider the wording and recall format of physical activity items when interpreting associations between the built environment and self-reported physical activity. Beyond the scope of this study, further psychometric assessment of self-report physical activity tools that capture different types of physical activity within different contexts is needed.
SSW and street integration are positively associated with participation in neighborhood TW and LW, with some albeit less supporting evidence for their associations with time spent on neighborhood TW. Space syntax includes practical metrics that can be used with easily available data (e.g., street networks) and which can be used to inform urban design policy and practice (Cutumisu and Spence, 2009; Koohsari et al., 2014; Karimi, 2012). Our findings highlight that these easy to estimate built environment metrics are associated with walking among adults residing in a Canadian urban setting.
References
Baran, P.K., Rodríguez, D.A., Khattak, A.J., 2008. Space syntax and walking in a new urbanist and suburban neighbourhoods. J. Urban Des. 13, 5–28.
Cutumisu, N., Spence, J.C., 2009. Exploring associations between urban environments and children’s physical activity: making the case for space syntax. J. Sci. Med. Sport 12, 537–538.
Dmti Spatial Inc, 2008. DMTI CanMap RouteLogistics. Markham, Ontario).
Doma, K., Speyer, R., Leicht, A.S., Cordier, R., 2017. Comparison of psychometric properties between usual-week and past-week self-reported physical activity questionnaires: a systematic review. Int. J. Behav. Nutr. Phys. Act. 14, 10.
Durante, R., Ainsworth, B.E., 1996. The recall of physical activity: using a cognitive model of the question-answering process. Med. Sci. Sport. Exerc. 28, 1282–1291.
Farkas, B., Wagner, D.J., Nettel-Aguirre, A., Friedenreich, C., McCormack, G.R., 2019. Evidence synthesis – a systematized literature review on the associations between neighbourhood built characteristics and walking among Canadian adults. Health Promot Chronic Dis Prev Can 39, 1–14.
Frehlich, L., Friedenreich, C., Nettel-Aguirre, A., McCormack, G.R., 2018. Test-retest reliability of a modified International Physical Activity Questionnaire (IPAQ) to capture neighbourhood physical activity. J. Human. Sport. Exer. 13, 174–187.
Frehlich, L., Friedenreich, C., Nettel-Aguirre, A., Schipperijn, J., McCormack, G.R., 2018. Using accelerometer/GPS data to validate a neighborhood-adapted version of the international physical activity questionnaire (IPAQ). J. Measur. Phys. Behav. 1,
181–190.
Giles-Corti, B., Sallis, J., Sugiyama, T., Frank, L., Lowe, M., Owen, N., 2015. Translating active living research into policy and practice: one important pathway to chronic disease prevention. J. Public Health Policy 36, 231–243.
Giles-Corti, B., Timperio, A., Bull, F., Pikora, T., 2005. Understanding physical activity environmental correlates: increased specificity for ecological models. Exerc. Sport Sci. Rev. 33, 175–181.
Giles-Corti, B., Timperio, A., Cutt, H., Pikora, T.J., Bull, F.C., Knuiman, M., Bulsara, M., Van Niel, K., Shilton, T., 2006. Development of a reliable measure of walking within and outside the local neighborhood: RESIDE’s Neighborhood Physical Activity Questionnaire. Prev. Med. 42, 455–459.
Google Maps Platform, 2018. Google Maps geocoding API. California, United States. https://developers.google.com/maps/documentation/geocoding. https://d evelopers.google.com/maps/documentation/geocoding. accessed June 2019. Hajna, S., Ross, N.A., Brazeau, A.-S., Belisle, P., Joseph, L., Dasgupta, K., 2015. Associations between neighbourhood walkability and daily steps in adults: a systematic review and meta-analysis. BMC Public Health 15, 768.
Hall, C.M., Ram, Y., 2018. Walk score® and its potential contribution to the study of active transport and walkability: a critical and systematic review. Transp. Res. D Transp. Environ. 61, 310–324.
Hillier, B., 2009. The city as a socio-technical system a spatial reformulation in the light of the levels problem and the parallel problem. In: Proceedings of Conference on Spatial Information Theory. Aber Wrac’h, France.
Hillier, B., Hanson, J., 1984. The Social Logic of Space. Cambridge University Press.
Jack, E., McCormack, G.R., 2014. The associations between objectively-determined and self-reported urban form characteristics and neighborhood-based walking in adults. Int. J. Behav. Nutr. Phys. Act. 11, 71.
Jiang, B., 2012. Axwoman 6.0: an ArcGIS Extension for Urban Morphological Analysis. University of Gavle, Swede€
Karimi, K., 2012. A configurational approach to analytical urban design: “Space syntax” methodology. Urban Des. Int. 17, 297–318.
Koohsari, M.J., Kaczynski, A.T., McCormack, G.R., Sugiyama, T., 2014. Using space syntax to assess the built environment for physical activity: applications to research on parks and public open spaces. Leis. Sci. 36, 206–216.
Koohsari, M.J., Owen, N., Cerin, E., Giles-Corti, B., Sugiyama, T., 2016. Walkability and walking for transport: characterizing the built environment using space syntax. Int. J. Behav. Nutr. Phys. Act. 13, 121.
Koohsari, M.J., Owen, N., Cole, R., Mavoa, S., Oka, K., Hanibuchi, T., Sugiyama, T., 2017. Built environmental factors and adults’ travel behaviors: role of street layout and local destinations. Prev. Med. 96, 124–128.
Koohsari, M.J., Sugiyama, T., Mavoa, S., Villanueva, K., Badland, H., Giles-Corti, B., Owen, N., 2016. Street network measures and adults’ walking for transport: application of space syntax. Health Place 38, 89–95.
Koohsari, M.J., Sugiyama, T., Shibata, A., Ishii, K., Liao, Y., Hanibuchi, T., Owen, N., Oka, K., 2017. Associations of street layout with walking and sedentary behaviors in an urban and a rural area of Japan. Health Place 45, 64–69.
Maghelal, P., Capp, C.J., 2011. Walkability: a review of existing pedestrian indices. J. Urban. Reg. Inf. Sys. Assoc. 23, 5–19.
McCormack, G.R., Friedenreich, C., Shiell, A., Giles-Corti, B., Doyle-Baker, P.K., 2010. Sex and age-specific seasonal variations in physical activity among adults. J. Epidemiol. Community Health 64, 1010–1016.
McCormack, G., Shiell, A., 2011. In search of causality: a systematic review of the relationship between the built environment and physical activity among adults. Int. J. Behav. Nutr. Phys. Act. 8, 125.
McCormack, G., Shiell, A., Doyle-Baker, P., Friedenreich, C., Sandalack, B., 2014. Subpopulation differences in the association between neighborhood urban form and neighborhood-based physical activity. Health Place 28, 109–115.
McCormack, G.R., 2017. Neighbourhood built environment characteristics associated with different types of physical activity in Canadian adults. Health Promot Chronic Dis Prev Can 37, 175–185.
McCormack, G.R., Friedenreich, C., Sandalack, B.A., Giles-Corti, B., Doyle-Baker, P.K., Shiell, A., 2012. The relationship between cluster-analysis derived walkability and local recreational and transportation walking among Canadian adults. Health Place 18, 1079–1087.
McCormack, G., Shiell, A., Doyle-Baker, P., Friedenreich, C., Sandalack, B., Giles- Corti, B., 2009. Testing the reliability of neighborhood-specific measures of physical activity among Canadian adults. J. Phys. Act. Health 6, 367–373.
Raine, K., Muhajarine, N., Spence, J., Neary, N., Nykiforuk, C., 2012. Coming to consensus on policy to create supportive built environments and community design. Can. J. Public Health 103, S5–S8.
Saelens, B.E., Handy, S.L., 2008. Built environment correlates of walking: a review. Med. Sci. Sport. Exerc. 40.
Sallis, J.F., Saelens, B.E., 2000. Assessment of LW 6 physical activity by self-report: status, limitations, and future directions. Res. Q. Exerc. Sport 71, S1–S14.
Salvo, G., Lashewicz, B.M., Doyle-Baker, P.K., Mccormack, G.R., 2018. Neighbourhood built environment influences on physical activity among adults: a systematized review of qualitative evidence. Int. J. Environ. Res. Public Health 15.
Turner, A., 2004. Depthmap 4, a Researcher’s Handbook. Bartlett School of Graduate Studies, University College London, London, UK.
Wineman, J., Marans, R., Schulz, A., Van der westhuizen, D., Mentz, G., Max, P., 2014. Designing healthy neighborhoods: contributions of the built environment to physical activity in Detroit. J. Plan. Educ. Res. 34, 180–189.