Skip to main content

Cross-sectional and longitudinal associations between the built environment and walking: effect modification by socioeconomic status

Abstract

Background

Although socioeconomic status (SES) has been shown to modify associations between the neighborhood built environment and physical activity, contradictory results exist. Objectives of this cross-sectional and longitudinal analysis were to: 1) examine whether overall neighborhood walkability and specific built characteristics were associated with walking among adults at a single point in time and after they relocate neighborhoods, and 2) test for effect modification of these associations by SES.

Methods

We linked longitudinal data from 703 adults who relocated urban neighborhoods between two waves of Alberta’s Tomorrow Project (2008–2015) to neighborhood built environment data. We created a walkability index from measures of population counts, street connectivity, and destination diversity within 400 m of participants’ homes. In cross-sectional analyses, we used generalized linear models to estimate associations between built characteristics and minutes walked per week at baseline. For the longitudinal analyses, we used fixed-effects linear regression models to estimate associations between changes in built characteristics and minutes walked per week. We also assessed if indicators of SES (individual education or household income) modified both sets of associations.

Results

Most cross-sectional and longitudinal associations were small and statistically non-significant. Neighborhood population count (b = 0.03, 95% CI: 0.01, 0.07) and street connectivity (b = − 1.75, 95% CI: − 3.26, − 0.24) were cross-sectionally associated with walking duration among the overall sample. None of the longitudinal associations were statistically significant among the overall sample. There was limited evidence of effect modification by SES, however, we found negative cross-sectional associations between street connectivity and walking among adults with lower education and income, and a positive association between percent change in walkability and change in walking among lower educated adults.

Conclusions

Despite population count and street connectivity being associated with walking at baseline, changes in these built environment variables were not associated with changes in walking following residential relocation. Our findings also provide evidence, albeit weak, that changes in neighborhood walkability, resulting from residential relocation, might more strongly affect walking among low SES adults. Further longitudinal research is needed to examine built environment characteristics with walking for different purposes and to test for inequitable socioeconomic impacts.

Peer Review reports

Background

Regular physical activity can reduce the risk of chronic disease and premature mortality [1,2,3]. The World Health Organization’s Global Action Plan for Physical Activity and Health has a target of a 15% reduction in inactivity levels by 2030 and recommends several policy actions, including the creation of physical activity supportive environments [4]. Evidence, derived mainly from cross-sectional studies, suggests that numerous neighborhood built characteristics, such as street connectivity, land-use mix, residential density, and walkability (the combination of several built characteristics) are associated with physical activity [5,6,7]. However, cross-sectional studies are limited in their ability to establish temporality in the associations between built environment characteristics and walking [7, 8]. Furthermore, it is possible that participants self-select into neighborhoods with particular built characteristics based on their walking preferences leading to biased estimates in cross-sectional studies [9, 10]. Longitudinal studies, and specifically those examining the impacts of residential relocation, are able to measure physical activity before and after participants have been exposed to a new neighborhood built environment [11, 12]. Several residential relocation studies have found that increases in neighborhood walkability are associated with increases in walking duration among adults [13, 14]; however, the associations between specific built characteristics, such as street connectivity, land-use mix, and availability of public transportation, with walking are mixed [15, 16].

Despite evidence suggesting positive associations between socioeconomic status (SES) and physical activity [17], there is inconsistency regarding potential benefits of the built environment for physical activity among socioeconomically disadvantaged groups [18,19,20]. A systematic review of longitudinal studies investigating associations between the built environment and physical activity [19] found that most studies reported no evidence of effect modification by individual-level SES, except for one study, which found that new walking and cycling routes were used more by people with higher incomes, higher education levels, and by people who were currently employed [21]. In contrast, in their narrative review, Adkins et al. [20] reported that most studies in their review found weaker associations between the built environment and physical activity among socioeconomically disadvantaged groups compared to socioeconomically advantaged groups. Given the limited number of studies and mixed findings on differential impacts of the built environment on walking for different SES groups, more research is needed.

The objectives of this study were to: 1) examine whether or not overall neighborhood walkability and specific built characteristics were associated with walking among adults at a single point in time (cross-sectional) and after they relocate neighborhoods (longitudinal), and 2) test for effect modification of these associations by SES (education and household income).

Methods

Study and sample design

This analysis used data from Alberta’s Tomorrow Project (ATP), a prospective cohort study designed to investigate the etiology of cancer and other non-communicable diseases [22]. ATP recruited adults aged 35–69 years living in Alberta, Canada who had no history of cancer, other than non-melanoma skin cancer [22], between 2000 and 2008 using random digit dialing [23]. Of the 63,486 eligible and interested adults contacted, 31,072 consented to participate [23]. Participants completed a survey at the time of enrollment to measure demographic, personal and family health, and lifestyle information, and then follow-up surveys in 2008 and 2009–2015 for updating and collecting new information. The flow of sample recruitment and participation through ATP over time have been described elsewhere [22].

For this analysis, we focused on two time points using the surveys released in 2008 (referred to herein as ‘baseline’) and in 2009–2015 (referred to as ‘follow-up’) due to the consistency of the physical activity tool used in both surveys (described below). The eligibility criteria for this study included participants who completed the baseline and follow-up surveys and those who relocated (both to and from an urban area) between the two time points (based on a change in residential postal code). Those participants are hereafter referred to as “movers” for simplicity. Note that we retained “non-movers” (those who completed both the baseline and follow-up surveys and stayed in an urban area between the two time points) to test for differences between “non-movers” and “movers” before excluding them from the sample. Among the 9779 urban-dwelling participants who completed both baseline and follow-up surveys, participants were excluded if they did not move (n = 8952), had missing data (n = 123), or were immobile at baseline (n = 1; assessed through the mobility-related component of the EQ-5D [24]), resulting in a sample size for our analysis of 703 participants. The University of Calgary Conjoint Research Ethics Board approved this study (REB17–1466).

Measures

Exposure variables

The neighborhood built environment exposure measures for this analysis consisted of specific built environment characteristics (population counts, diversity of destinations, and street connectivity) and a composite walkability index estimated for 400-m Euclidean buffers around each participant’s residential 6-digit postal code at baseline and follow-up. In Canada, urban residential postal codes have a geographical area that typically ranges from one apartment building to one city block [25], and postal codes can be used as a valid proxy for household locations [26]. A 400-m buffer was chosen because it represents the approximate distance that can be travelled by walking in approximately 5 minutes [27]. Geographic Information Systems (GIS) were used to process built environment data for all urban postal codes in Alberta for each year between 2008 and 2015, and participants’ home locations were linked to objective built environment data with their postal code for baseline and follow-up based on the year they completed each survey.

Population counts were estimated using 2006, 2011, and 2016 Statistics Canada Census data. Population counts for non-census years were estimated based on averages imputed from the available census data. Diversity of destinations was estimated by summing the number of different destination types (classified by Standard Industry Codes obtained from DMTI Spatial’s CanMap Enhanced Points of Interest data). Examples of destination types included hardware stores, department stores, grocery stores, restaurants, banks, libraries, barbershops, museums, and schools. Street connectivity was estimated by the count of three and four-way street intersections within the buffer (derived from DMTI Spatial’s CanMap Streetfiles and Route Logistics dataset). For the street connectivity measure, we summed the number of three- and four-way street intersections together, but with three-way intersections weighted half as strongly as four-way intersections. These specific built characteristics have been used in a previous study with ATP data that found associations between the built environment and weight status [28].

The walkability index was created using the built characteristics described above (i.e., population counts, diversity of destinations, and connectivity). For the walkability index, the built environment characteristics were standardized relative to the maximum value of each characteristic observed between 2008 and 2015 across all urban postal codes in Alberta (making the scores relative to urban walkability values across the province). The maximum values found for each built characteristic were: 1) population count = 7956; 2) number of different destination types = 23; 3) number of four-way intersections = 51, and; 4) number of three-way intersections = 92. This percent of maximum possible (“POMP”) method [29, 30] results in a standardized score for each built characteristic that is a proportional score bound by 0 and 100%. The four standardized scores were summed together. Similar to prior research that down-weighted three-way intersections relative to four-way intersections [31], and consistent with our connectivity measure, the three-way intersections variable in our index was given a weight of 0.5 instead of one because three-way intersections have been shown to have a weaker association with walking compared to four-way intersections [32]. The resulting walkability index was bounded at 0, with a theoretical maximum of 350 (see Supplementary Material for more detailed information about the built environment measures).

After estimating walkability values at both time points, changes in the walkability values were operationalized in three different ways: 1) per 10-unit change; 2) percent change; and 3) per tertile of change (corresponding to decreased walkability, minimal change, and increased walkability).

To assess convergent validity, we estimated a Pearson correlation coefficient comparing the walkability index for 2012 with an external measure, the 2012 Walk Score®, a publicly available walkability index that reflects the level of access to nearby walkable amenities. Specifically, Walk Score® estimates neighborhood walkability based on proximity to amenities (e.g., grocery stores, restaurants, parks, libraries). Amenities within a five-minute walking distance are given the most points and a decay function is used to provide fewer points, up to a 30-minute walk. The Walk Score® algorithm has been shown to be a valid indicator of neighborhood walkability [33, 34] and is positively associated with walking [35], including among adults with low incomes [36, 37]. Our walkability index was positively correlated with Walk Score® (r = 0.69, p < 0.001; data not shown).

Walking

Both surveys measured walking during the past week using the International Physical Activity Questionnaire (IPAQ) short or long forms, which have been validated in multiple countries, including Canada [38]. The IPAQ asks participants to report on how many days they walked (for at least 10 minutes) during the previous week and how much time they usually spent walking on one of those days. Walking minutes per week at each time point was estimated following the IPAQ scoring system [38] and change scores were calculated to estimate changes in walking minutes per week from baseline to follow-up.

SES potential effect modifiers

The two potential effect modifiers, baseline education (≤ high school versus > high school) and baseline household income (low-income household versus not a low-income household) were examined. Participants were asked to report the highest level of education they finished, with nine response options ranging from “did not complete grade eight” to “completed university post-graduate degree”. For the income question, participants were asked to report their total household income before taxes, with 18 response options ranging from “less than $10,000 to “$250,000 or more”. The 2006 Statistics Canada Census data were used to determine the household low-income status cut point. Participants who reported a household income below the 2006 median income level for Alberta ($63,988) were classified as living in a household with a lower income [39]. Due to the categories used in collecting the data, $60,000 was used as the cut point. Two alternative measures of household income were also considered as potential effect modifiers for both the cross-sectional and longitudinal sets of analyses: 1) tertiles of baseline income, and 2) the Statistics Canada Market Basket Measure (a low-income line threshold, which reflects differences in living costs across regions and across household sizes in Canada). Tertiles of baseline income were created from using the midpoint of each income response category. We averaged the Market Basket Measure for all urban Alberta regions and then calculated separate cut points for each year and household size using the square root equivalency recommended by Statistics Canada [40].

Covariates

Data on age, sex, relationship status (married/common law, single/never married), IPAQ survey type (short or long-form), and years between surveys were available from self-administered questionnaires. Baseline age, sex, relationship status, the presence of children at home, season of survey completion, and years between surveys were included in the cross-sectional models. Changes in relationship status, presence of children at home, season of survey completion, as well as follow-up IPAQ survey type (short or long form) and years between surveys were included in the longitudinal models. We selected covariates based on their known associations with walking, their availability within the data, and/or whether they could potentially explain relocation to higher or lower levels of walkability (e.g., change in the presence of children in the home [41]).

Statistical analyses

Descriptive analyses compared movers and non-movers on baseline demographic characteristics, after which non-movers were excluded from the remaining analysis. We used independent samples t-tests and Pearson’s chi-square tests, as appropriate, to test for differences between non-movers and movers. Similarly, baseline demographic characteristics for participants from the three walkability-change groups (tertiles of walkability-change) were compared using one-way analysis of variance (ANOVA) and Pearson’s chi-square tests.

For the cross-sectional analyses (using baseline data), we used generalized linear models with a gamma distribution and an identity link function because residuals from ordinary least squares linear regression models were heteroskedastic and highly skewed. A gamma distribution requires all values to be greater than zero, so 1 minute was added to all participants’ baseline walking minutes-for the cross-sectional models. Akaike Information Criteria (AIC) values were used to compare model fits for a Gaussian (normal) distribution (with an identity link) and Gamma distributions (with identity, inverse, or log links). There were substantial reductions in AIC values between the Gaussian compared to the Gamma distributions, but minimal change between the three different linking functions. Thus, we used Gamma distributions with identity links to improve interpretability and comparability with the longitudinal analyses. We also assessed for non-linearity of the fully adjusted associations by including squared and cubed terms of the built environment predictors, and then testing if model fit was improved using likelihood ratio tests.

For the longitudinal analyses, we used fixed-effects linear regression models to estimate associations for within-person change in walkability and specific built characteristics (population counts, diversity of destinations, and street connectivity) with changes in walking. We created change scores by subtracting follow-up measurements of built environment characteristics from baseline measurements. Because fixed-effects models control for measured and unmeasured confounding by time-invariant covariates [15], only time-varying covariates were included in the models [42]. Linear regression models were used to estimate beta coefficients (b) and 95% Confidence Intervals (CI) for changes in built environment characteristics and walking minutes, while adjusting for time-varying covariates. We also assessed for non-linearity of the fully adjusted associations by including squared and cubed terms of the built environment predictors, and then testing if model fit was improved using partial F tests.

Effect modification of the associations between the built environment and walking was examined by including interaction (product) terms for education and household low-income status with the built environment measure separately in each model. Models with significant interaction terms were stratified by the SES indicator and adjusted for the time-varying covariates.

Sensitivity analyses were conducted to account for baseline walking in the longitudinal associations and to test the two alternative measures of household income as potential effect modifiers (in both the cross-sectional and longitudinal analyses). Listwise deletion was used for missing data. Analyses were conducted using R version 4.0.3, and the level of significance was set at P < 0.05 for all analyses.

Results

Comparison of characteristics between movers and non-movers

There were statistically significant differences (P < 0.05) between movers (n = 703) and non-movers (n = 8094) at baseline. Non-movers were older (mean 55 versus 53 years of age) and more likely to be married/common law (78% versus 69%) compared to movers (Supplementary Material).

Sample characteristics (movers only)

The majority of participants were women (64%), and at baseline most participants were married/common law (69%), with participant ages ranging from 37 to 77 years (mean = 53, SD (Standard Deviation) = 9 years; Table 1). Most participants completed at least some post-secondary education (79%), and most were from a household above the Alberta median household income cut point (71%). Most participants reported some walking at baseline (79%) and follow-up (84%), with a mean increase from 210 to 271 minutes/week at follow up (SD =381). The time between completing the surveys ranged from 0 to 7 years (mean = 2.9, SD = 1.4 years). Seventy percent of participants completed the follow-up survey in a different season than their baseline survey.

Table 1 Sociodemographic characteristics of the cohort (n = 703), Alberta’s Tomorrow Project, Canada, 2008–2015

Neighborhood built environments

Participants’ mean baseline neighborhood walkability value was 57.02 (SD = 37.69; Table 2). On average, participants moved to neighborhoods with − 0.25 points lower walkability (SD = 47.91). At baseline, participants’ mean connectivity scores were 15.60 (SD = 9.50) and mean change in connectivity was 1.06 (SD = 10.88). Participants lived in neighborhoods with a mean of 4.75 different destination types (SD = 6.03) at baseline and moved to areas with fewer destination types (mean = − 0.32, SD = 7.81) compared to their previous neighborhood. Mean population counts at baseline were 1054 (SD = 692), and participants moved to areas with lower population counts (mean = − 17, SD = 883), on average, relative to their baseline neighborhoods.

Table 2 Built environment characteristics of the cohort (n = 703), Alberta’s Tomorrow Project, Canada, 2008–2015

Cross-sectional associations between built environment characteristics and walking at baseline

Neighborhood street connectivity was negatively associated with minutes walked per week at baseline (b = − 1.75, 95% CI: − 3.26, − 0.24). In contrast, neighborhood population count was positively associated with minutes walked per week at baseline (b = 0.03, 95% CI: 0.01, 0.07). Neighborhood walkability and diversity of destinations were not significantly associated with baseline walking in the overall sample. There was no evidence of non-linear associations between the built environment variables and walking. However, there was evidence of confounding, with several of the estimated association changing substantially after covariate adjustment (crude and adjusted estimates shown in Table 3). Sensitivity analyses testing for effect modification by other household income measures (income tertiles and low-income line threshold) produced similar results (Supplementary Material).

Table 3 Cross-sectional associations between built characteristics and walking, Alberta’s Tomorrow Project, Canada (n = 703), 2008–2015

Only the statistical interaction terms for education and household income with street connectivity were statistically significant (Table 3). The associations between street connectivity and baseline walking were only significant among the lower SES groups (lower education: b = − 3.74, 95% CI: − 5.42, − 2.06; lower income: b = − 3.47, 95% CI: − 4.87, − 2.07). Statistical interactions for walkability and diversity of destinations with education and income approached statistical significance (Ps < 0.09). Although none of the stratified estimates were statistically significant, for both these sets of associations, the coefficient point estimates were negative amongst the lower SES groups and positive amongst the higher SES groups (Table 3).

Longitudinal associations between changes in built environment characteristics and walking following neighborhood relocation

There were no statistically significant associations between changes in the built environment characteristics and walking for the overall sample (Tables 4 and 5). There was also no evidence of non-linear associations between the built environment variables and walking. There was limited evidence of confounding, with few of the estimates changing substantially after covariate adjustment (crude and adjusted estimates shown in Tables 4 and 5).

Table 4 Longitudinal associations between walkability and walking, Alberta’s Tomorrow Project, Canada (n = 703), 2008–2015
Table 5 Longitudinal associations between specific built characteristics and walking, Alberta’s Tomorrow Project, Canada, 2008–2015

Among the longitudinal associations, the interaction terms between education and percent change in walkability (Table 4) and education and change in diversity of destinations (Table 5) were significantly associated with change in walking minutes. However, in stratified analyses, only the association between percent change in walkability and change in walking minutes among the low education participants was significant (b = 0.01, 95% CI: 0.00, 0.01)) (Table 4).

Sensitivity analyses adjusting for baseline walking and testing for effect modification by other household income measures (income tertiles and low-income line threshold) produced similar results (Supplementary Material).

Discussion

Our study examined cross-sectional and longitudinal associations between the neighborhood built environment and walking among adults and if individual-level SES modified these associations. Among the cross-sectional analyses, in the overall sample, street connectivity was negatively associated with minutes walked per week, and population count was positively associated with weekly minutes of walking. Walkability and diversity of destinations were not significantly associated with baseline walking for the overall sample. Our results for street connectivity, walkability, and diversity of destinations are inconsistent with prior evidence demonstrating positive associations between these built characteristics and walking [43, 44]. We detected evidence of effect modification of cross-sectional associations by SES only for street connectivity. Among the longitudinal associations, there were no statistically significant associations between changes in neighborhood walkability nor the specific built environment characteristics with walking duration for the overall sample. Evidence of effect modification by SES of longitudinal associations was only found for relative walkability-change and change in diversity of destinations by education.

Although our study found statistically significant associations for the overall sample in the cross-sectional analyses, we did not find the same associations for the longitudinal analyses. Our results are similar to those of Braun and colleagues [45] who undertook a similar cross-sectional and longitudinal analysis of the built environment, but with cardiometabolic risk factors. Their study also found mainly non-significant results, with different metabolic risk factors associated with the built environment at baseline (diastolic blood pressure) compared to across time (triglycerides and systolic blood pressure) [45].

Our lack of statistically significant longitudinal findings contrast with those of Hirsch and colleagues [13] who also used a residential relocation study design and fixed effects models, but found positive associations between changes in absolute walkability and walking. However, in contrast to our study that included an overall measure of walking, Hirsch and colleagues [13] examined walking for transportation and leisure as separate outcomes, and found that associations were only statistically significant for transportation walking. Indeed, our results align with the findings of another residential relocation study (that also used fixed effects models) and combined transportation and leisure walking into a single outcome and found no statistically significant associations between changes in walkability and changes in participation or duration of walking [45]. Most of our study participants completed the IPAQ-short version, rather than the IPAQ-long, which prevented separate assessments of transportation and leisure walking.

Our longitudinal findings regarding specific built characteristics mirror some of the mixed results that have been previously found [15, 16]. Changes in neighborhood population count, diversity of destinations, and street connectivity following residential relocation were not related to changes in walking. Although a UK residential relocation study found a positive association for residential density with daily steps [16], an Australian study found a non-statistically significant association between residential density and transportation walking [15]. These two studies also found positive associations between land use mix and walking [15, 16], whereas our study found no relationship, among the overall sample, between increases in diversity of destinations and changes in walking. These differences in findings for specific built characteristics may be attributable to the larger neighborhood buffer sized used in the previous studies (the UK study used a 1 km network buffer [16] and the Australian study used a 1.6 km network buffer [15]). Our neighborhood buffer size of 400 m excluded potential changes in diversity of destinations located outside the buffer that may have still been within walking distance.

Our results contribute to the inconsistent findings regarding differing impacts of the built environment on physical activity for people from different SES groups [19, 20, 46]. A systematic review of longitudinal studies investigating the associations between the built environment and physical activity found little support for effect modification by SES [19]. In a UK residential relocation study, Clary et al. [16] tested for effect modification by SES within their associations between changes in various built environment measures and accelerometer-derived physical activity measures. Their study found evidence of effect modification only for the association between increases in accessibility to public transit (a composite measure of proximity, number of stops, and frequency of service) with daily steps, such that there were no statistically significant associations among the two higher SES groups, but a marginally significant negative association for the low SES group (i.e., defined by social housing recipient status [16]). Our findings are similar to Clary at al [16] because most associations had no evidence of effect modification by SES and we had some preliminary evidence of negative associations among the low SES groups.

The study has several strengths, including the use of both a cross-sectional and a longitudinal design and the testing of modification by SES indicators. This study is relevant to requests for more policy-relevant research methods [47, 48] because we examined the impact of specific built environment characteristics, which may be more useful to policymakers compared to multi-component walkability measures. Furthermore, we estimated built environment characteristics using data from the same year that the participant completed each questionnaire (except for population count data, where the census data were weighted to approximate the years where the census was not collected). Several other residential relocation studies have used WalkScore®, which is a proprietary measure of walkability, and thus those studies only had the most current data available [13, 45].

The study was limited by relatively small subsamples of participants who could be classified as having low SES, which decreased the precision of our estimates and may have reduced our ability to detect statistically significant associations and interactions. The overall ATP cohort has relatively high SES [23] and is likely affected by a “healthy volunteer” bias that is common among prospective cohort studies [49, 50]. However, the ATP cohort has rates of chronic diseases (e.g., diabetes and hypertension) that are similar to the general Canadian population [51]. Our study was also limited by the self-report measures of walking that did not specify the location where walking occurred and did not differentiate by walking purpose. Our use of Euclidean buffers meant that the geographically defined neighborhoods were standardized across participants, however, this delineation also meant that while certain built environment features were present in the buffer they might not have been easily accessible via walking [52]. We did not have access to data regarding participants’ moving date or their reasons for moving, and as such, we were unable to consider exposure time in neighborhood or reasons for moving (e.g., changing occupations or familial responsibilities). Similarly, there was still potential for residential self-selection bias, even in the longitudinal analyses because some adults who were interested in increasing their walking may have chosen to move to a neighborhood with greater walkability. Finally, the study sample consisted of middle-aged adults (ATP recruitment inclusion criteria included adults aged 35–69 years only) who relocated neighborhoods; thus, the results may not be generalizable to younger adults or non-movers.

Given the limited number and mixed findings on differential impacts of the built environment for SES groups, there is a need for more longitudinal research examining equity impacts of changing the built environment. The mixed evidence so far [19, 20, 46], in addition to our own findings, precludes us from making definitive recommendations in regard to how and which urban planning policies should be modified to encourage more walking among low SES adults. Nevertheless, to improve our understanding and to generate evidence that might inform policy in the future, methodological improvements in future studies examining associations between the built environment and walking among low SES adults may be needed. For example, future studies would benefit from context-specific measures of walking and from separating leisure and transportation walking when investigating the impact of changes in the built environment on walking [53, 54]. It would also be useful for future studies to examine built characteristics that may be particularly relevant for adults with low SES (e.g., transit stops, safety, low-cost or free recreation centers). Such studies may involve natural experiments that involve examining the effects of modifying specific features of the built environment on walking among residents of low SES neighbourhoods. Regardless of our findings, other interventions may be needed to increase population levels of physical activity to avoid negatively impacting adults with lower SES. For instance, more fundamental or ‘upstream’ interventions (e.g., living wage policies) may be needed to address socioeconomic disparities in physical activity [55, 56].

Conclusions

Our findings suggest that an increase in neighborhood walkability, relative to the prior neighborhood, is associated with increases in walking after relocating neighborhoods. Despite street connectivity and population count being associated with walking in the cross-sectional analyses, these built environment variables were not associated with changes in walking following residential relocation. We found limited evidence that these associations differ by education and household income. Although the evidence is weak, our findings also suggest that changes in neighbourhood walkability, following residential relocation, might more strongly affect walking among low SES adults. Further longitudinal research is needed that examines specific built environment characteristics and walking for different purposes among disadvantaged populations.

Availability of data and materials

The data that support the findings of this study are available from Alberta’s Tomorrow Project, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are available, however, from the authors upon reasonable request and with permission of Alberta’s Tomorrow Project. More information can be obtained via https://myatpresearch.ca.

Abbreviations

SES:

Socioeconomic status

ATP:

Alberta’s Tomorrow Project

CI:

Confidence interval

GIS:

Geographic Information Systems

IPAQ:

International Physical Activity Questionnaire

ANOVA:

Analysis of variance

SD:

Standard deviation

References

  1. McTiernan A, Friedenreich CM, Katzmarzyk PT, Powell KE, Macko R, Buchner D, et al. Physical activity in cancer prevention and survival: a systematic review. Med Sci Sports Exerc. 2019;51(6):1252.

    Article  PubMed  PubMed Central  Google Scholar 

  2. Lear SA, Hu W, Rangarajan S, Gasevic D, Leong D, Iqbal R, et al. The effect of physical activity on mortality and cardiovascular disease in 130 000 people from 17 high-income, middle-income, and low-income countries: the PURE study. Lancet. 2017;390(10113):2643–54.

    Article  PubMed  Google Scholar 

  3. Warburton D, Bredin S. Health benefits of physical activity: a systematic review of current systematic reviews. Curr Opin Cardiol. 2017;32(5):541–56.

    Article  PubMed  Google Scholar 

  4. World Health Organization. Global action plan on physical activity 2018–2030: more active people for a healthier world: World Health Organization; 2019.

    Google Scholar 

  5. Hajna S, Ross NA, Brazeau A-S, Bélisle P, Joseph L, Dasgupta K. Associations between neighbourhood walkability and daily steps in adults: a systematic review and meta-analysis. BMC Public Health. 2015;15(1):768.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Barnett DW, Barnett A, Nathan A, Van Cauwenberg J, Cerin E. Built environmental correlates of older adults’ total physical activity and walking: a systematic review and meta-analysis. Int J Behav Nutr Phys Act. 2017;14(1):103.

    Article  PubMed  PubMed Central  Google Scholar 

  7. Ding D, Gebel K. Built environment, physical activity, and obesity: what have we learned from reviewing the literature? Health Place. 2012;18(1):100–5.

    Article  PubMed  Google Scholar 

  8. McCormack GR, Shiell A. 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. 2011;8(1):125.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Lamb KE, Thornton LE, King TL, Ball K, White SR, Bentley R, et al. Methods for accounting for neighbourhood self-selection in physical activity and dietary behaviour research: a systematic review. Int J Behav Nutr Phys Act. 2020;17:1–22.

    Article  Google Scholar 

  10. Frank LD, Kershaw SE, Chapman JE, Campbell M, Swinkels HM. The unmet demand for walkability: disparities between preferences and actual choices for residential environments in Toronto and Vancouver. Canadian journal of public health-revue Canadienne De. Sante Publique. 2015;106(1):ES12–20.

    Google Scholar 

  11. Kärmeniemi M, Lankila T, Ikäheimo T, Koivumaa-Honkanen H, Korpelainen R. The built environment as a determinant of physical activity: a systematic review of longitudinal studies and natural experiments. Ann Behav Med. 2018;52(3):239–51.

    Article  PubMed  Google Scholar 

  12. Ding D, Nguyen B, Learnihan V, Bauman AE, Davey R, Jalaludin B, et al. Moving to an active lifestyle? A systematic review of the effects of residential relocation on walking, physical activity and travel behaviour. Br J Sports Med. 2018;52(12):789–99.

    Article  PubMed  Google Scholar 

  13. Hirsch JA, Diez Roux AV, Moore KA, Evenson KR, Rodriguez DA. Change in walking and body mass index following residential relocation: the multi-ethnic study of atherosclerosis. Am J Public Health. 2014;104(3):e49–56.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Wasfi RA, Dasgupta K, Eluru N, Ross NA. Exposure to walkable neighbourhoods in urban areas increases utilitarian walking: longitudinal study of Canadians. J Transp Health. 2016;3(4):440–7.

    Article  Google Scholar 

  15. Knuiman MW, Christian HE, Divitini ML, Foster SA, Bull FC, Badland HM, et al. A longitudinal analysis of the influence of the neighborhood built environment on walking for transportation: the RESIDE study. Am J Epidemiol. 2014;180(5):453–61.

    Article  PubMed  Google Scholar 

  16. Clary C, Lewis D, Limb E, Nightingale CM, Ram B, Page AS, et al. Longitudinal impact of changes in the residential built environment on physical activity: findings from the ENABLE London cohort study. Int J Behav Nutr Phys Act. 2020;17(1):1–10.

    Article  Google Scholar 

  17. Pan SY, Cameron C, DesMeules M, Morrison H, Craig CL, Jiang X. Individual, social, environmental, and physical environmental correlates with physical activity among Canadians: a cross-sectional study. BMC Public Health. 2009;9(1):1–2.

    Article  Google Scholar 

  18. Christie CD, Consoli A, Ronksley PE, Vena JE, Friedenreich CM, McCormack GR. Associations between the built environment and physical activity among adults with low socio-economic status in Canada: a systematic review. Can J Public Health. 2021;112(1):152–65.

    Article  PubMed  Google Scholar 

  19. Smith M, Hosking J, Woodward A, Witten K, MacMillan A, Field A, et al. Systematic literature review of built environment effects on physical activity and active transport–an update and new findings on health equity. Int J Behav Nutr Phys Act. 2017;14(1):158.

    Article  PubMed  PubMed Central  Google Scholar 

  20. Adkins A, Makarewicz C, Scanze M, Ingram M, Luhr G. Contextualizing walkability: do relationships between built environments and walking vary by socioeconomic context? J Am Plan Assoc. 2017;83(3):296–314.

    Article  Google Scholar 

  21. Goodman A, Sahlqvist S, Ogilvie D, Consortium I. Who uses new walking and cycling infrastructure and how? Longitudinal results from the UK iConnect study. Prev Med. 2013;57(5):518–24.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Robson PJ, Solbak NM, Haig TR, Whelan HK, Vena JE, Akawung AK, et al. Design, methods and demographics from phase I of Alberta's tomorrow project cohort: a prospective cohort profile. CMAJ open. 2016;4(3):E515.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Ye M, Robson PJ, Eurich DT, Vena JE, Xu J-Y, Johnson JA. Cohort profile: Alberta’s tomorrow project. Int J Epidemiol. 2016;46(4):1097–8l.

    Article  Google Scholar 

  24. Rabin R, Charro F. EQ-SD: a measure of health status from the EuroQol group. Ann Med. 2001;33(5):337–43.

    Article  CAS  PubMed  Google Scholar 

  25. Statistics Canada. Postal code conversion file (PCCF), Reference Guide (2017).

    Google Scholar 

  26. Bow CJD, Waters NM, Faris PD, Seidel JE, Galbraith PD, Knudtson ML, et al. Accuracy of city postal code coordinates as a proxy for location of residence. Int J Health Geogr. 2004;3(1):5.

    Article  PubMed  PubMed Central  Google Scholar 

  27. Ellis G, Hunter R, Tully MA, Donnelly M, Kelleher L, Kee F. Connectivity and physical activity: using footpath networks to measure the walkability of built environments. Environ Plann B. 2016;43(1):130–51.

    Article  Google Scholar 

  28. Nichani V, Turley L, Vena JE, McCormack GR. Associations between the neighbourhood characteristics and body mass index, waist circumference, and waist-to-hip ratio: findings from Alberta's tomorrow project. Health & Place. 2020;64:102357.

    Article  Google Scholar 

  29. Moeller J. A word on standardization in longitudinal studies: don't. Front Psychol. 2015;6:1389.

    PubMed  PubMed Central  Google Scholar 

  30. Cohen P, Cohen J, Aiken LS, West SG. The problem of units and the circumstance for POMP. Multivar Behav Res. 1999;34(3):315–46.

    Article  Google Scholar 

  31. Iravani H, Rao V. The effects of new urbanism on public health. J Urban Des. 2020;25(2):218–35.

    Article  Google Scholar 

  32. Herrmann T, Gleckner W, Wasfi RA, Thierry B, Kestens Y, Ross NA. A pan-Canadian measure of active living environments using open data. Health Rep. 2019;30(5):16–26.

    Google Scholar 

  33. Duncan DT, Aldstadt J, Whalen J, Melly SJ, Gortmaker SL. Validation of walk score® for estimating neighborhood walkability: an analysis of four US metropolitan areas. Int J Environ Res Public Health. 2011;8(11):4160–79.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Carr LJ, Dunsiger SI, Marcus BH. Validation of walk score for estimating access to walkable amenities. Br J Sports Med. 2011;45(14):1144–8.

    Article  PubMed  Google Scholar 

  35. Thielman J, Rosella L, Copes R, Lebenbaum M, Manson H. Neighborhood walkability: differential associations with self-reported transport walking and leisure-time physical activity in Canadian towns and cities of all sizes. Prev Med. 2015;77:174–80.

    Article  PubMed  Google Scholar 

  36. Winters M, Barnes R, Venners S, Ste-Marie N, McKay H, Sims-Gould J, et al. Older adults' outdoor walking and the built environment: does income matter? BMC Public Health. 2015;15(1).

  37. Chudyk AM, McKay HA, Winters M, Sims-Gould J, Ashe MC. Neighborhood walkability, physical activity, and walking for transportation: a cross-sectional study of older adults living on low income.(report). BMC Geriatr. 2017;17(1).

  38. Craig CL, Marshall AL, Sjöström M, Bauman AE, Booth ML, Ainsworth BE, et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003;35(8):1381–95.

    Article  PubMed  Google Scholar 

  39. Statistics Canada. Census trends for Canada, provinces and territories (table). Census Trends. 2006 Census; 2007.

    Google Scholar 

  40. Statistics Canada. Table 11-10-0066-01: Market Basket Measure (MBM) thresholds for the reference family by Market Basket Measure region, component and base year.

  41. Clark WA, Huang Y. The life course and residential mobility in British housing markets. Environ Plan A. 2003;35(2):323–39.

    Article  Google Scholar 

  42. Allison PD. Fixed effects regression models: SAGE publications; 2009.

    Book  Google Scholar 

  43. Saelens EB, Handy LS. Built environment correlates of walking: a review. Med Sci Sports Exerc. 2008;40(7 Suppl 1):S550–S66.

    Article  PubMed  PubMed Central  Google Scholar 

  44. Farkas B, Wagner DJ, Nettel-Aguirre A, Friedenreich C, McCormack GR. Evidence synthesis-a systematized literature review on the associations between neighbourhood built characteristics and walking among Canadian adults. Health Promot Chronic Dis Prev Can. 2019;39(1):1–14.

    Article  PubMed  Google Scholar 

  45. Braun LM, Rodriguez DA, Song Y, Meyer KA, Lewis CE, Reis JP, et al. Changes in walking, body mass index, and cardiometabolic risk factors following residential relocation: longitudinal results from the CARDIA study. J Transp Health. 2016;3(4):426–39.

    Article  PubMed  PubMed Central  Google Scholar 

  46. McCormack GR, Shiell A, Doyle-Baker PK, Friedenreich CM, Sandalack BA. Subpopulation differences in the association between neighborhood urban form and neighborhood-based physical activity. Health Place. 2014;28:109–15.

    Article  PubMed  Google Scholar 

  47. Sallis JF, Bull F, Burdett R, Frank LD, Griffiths P, Giles-Corti B, et al. Use of science to guide city planning policy and practice: how to achieve healthy and sustainable future cities. Lancet. 2016;388(10062):2936–47.

    Article  PubMed  Google Scholar 

  48. Giles-Corti B, Sallis JF, Sugiyama T, Frank LD, Lowe M, Owen N. Translating active living research into policy and practice: one important pathway to chronic disease prevention. J Public Health Policy. 2015;36(2):231–43.

    Article  PubMed  Google Scholar 

  49. Fry A, Littlejohns TJ, Sudlow C, Doherty N, Adamska L, Sprosen T, et al. Comparison of sociodemographic and health-related characteristics of UK biobank participants with those of the general population. Am J Epidemiol. 2017;186(9):1026–34.

    Article  PubMed  PubMed Central  Google Scholar 

  50. Andreeva VA, Salanave B, Castetbon K, Deschamps V, Vernay M, Kesse-Guyot E, et al. Comparison of the sociodemographic characteristics of the large NutriNet-Santé e-cohort with French census data: the issue of volunteer bias revisited. J Epidemiol Community Health. 2015;69(9):893–8.

    Article  PubMed  Google Scholar 

  51. Dummer TJ, Awadalla P, Boileau C, Craig C, Fortier I, Goel V, et al. The Canadian Partnership for Tomorrow Project: a pan-Canadian platform for research on chronic disease prevention. CMAJ. 2018;190(23):E710–E7.

    Article  PubMed  PubMed Central  Google Scholar 

  52. Brownson RC, Hoehner CM, Day K, Forsyth A, Sallis JF. Measuring the built environment for physical activity: state of the science. Am J Prev Med. 2009;36(4):S99–S123. e12.

    Article  PubMed  PubMed Central  Google Scholar 

  53. Frehlich L, Friedenreich C, Nettel-Aguirre A, Schipperijn J, McCormack GR. Using accelerometer/GPS data to validate a neighborhood-adapted version of the international physical activity questionnaire (IPAQ). Journal for the Measurement of Physical Behaviour. 2018;1(4):181–90.

    Article  Google Scholar 

  54. Giles-Corti B, Timperio A, Bull F, Pikora T. Understanding physical activity environmental correlates: increased specificity for ecological models. Exerc Sport Sci Rev. 2005;33(4):175–81.

    Article  PubMed  Google Scholar 

  55. Schulz A, Northridge ME. Social determinants of health: implications for environmental health promotion. Health Educ Behav. 2004;31(4):455–71.

    Article  PubMed  Google Scholar 

  56. McLaren L, McIntyre L, Kirkpatrick S. Rose's population strategy of prevention need not increase social inequalities in health. Int J Epidemiol. 2009;39(2):372–7.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

The authors thank Anita Blackstaffe for assisting with the preparation of the data.

Funding

This work was supported by the Canadian Institutes of Health Research [FDN-154331]. Alberta’s Tomorrow Project is only possible because of the commitment of its research participants, its staff, and its funders: Alberta Health, Alberta Cancer Foundation, Canadian Partnership Against Cancer and Health Canada, and substantial in-kind funding from Alberta Health Services. The views expressed herein represent the views of the authors and not of Alberta’s Tomorrow Project or any of its funders. CDC was supported by a Cumming School of Medicine Graduate Scholarship, an Izaak Walton Killam Doctoral Scholarship, and a Vanier Canada Graduate Scholarship.

Author information

Authors and Affiliations

Authors

Contributions

GM and CC conceived the study. CF and JV were involved in the original ATP study design and data collection. CC analyzed the data and drafted the initial manuscript. LT conducted the GIS analysis. All authors interpreted the results, contributed to manuscript drafts, and approved the final manuscript.

Corresponding author

Correspondence to Chelsea D. Christie.

Ethics declarations

Ethics approval and consent to participate

The University of Calgary Conjoint Research Ethics Board approved this study (REB17–1466). Informed consent was obtained from all participants, and the methods were performed in accordance with the relevant guidelines and regulations.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Table S1.

Sociodemographic Characteristics of Movers (the Study Cohort) and Non-Movers (Excluded) at Baseline. Table S2. Longitudinal associations between walkability and walking, including baseline walking as a covariate, Alberta’s Tomorrow Project, Canada (n = 703), 2008–2015. Table S3. Longitudinal associations between specific built characteristics and walking, including baseline walking as a covariate, Alberta’s Tomorrow Project, Canada, 2008–2015. Table S4. Cross-sectional associations between built characteristics and walking, with additional income effect modifiers. Table S5. Longitudinal associations between walkability and walking, with additional income effect modifiers. Table S6. Longitudinal associations between built characteristics and walking, with additional income effect modifiers.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Christie, C.D., Friedenreich, C.M., Vena, J.E. et al. Cross-sectional and longitudinal associations between the built environment and walking: effect modification by socioeconomic status. BMC Public Health 22, 1233 (2022). https://doi.org/10.1186/s12889-022-13611-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12889-022-13611-0

Keywords