https://orcid.org/0000-0003-2769-6522
Abstract
In recent decades, the prevalence of obesity and diabetes has risen substantially in North America and worldwide. To address these dual epidemics, researchers and policymakers alike have been searching for effective means to promote healthy lifestyles at a population level. As a consequence, there has been a proliferation of research examining how the “built” environment in which we live influences physical activity levels, by promoting active forms of transportation, such as walking and cycling, over passive ones, such as car use. Shifting the transportation choices of local residents may mean that more members of the population can participate in physical activity during their daily routine without structured exercise programs. Increasingly, this line of research has considered the downstream metabolic consequences of the environment in which we live, raising the possibility that “healthier” community designs could help mitigate the rise in obesity and diabetes prevalence. This review discusses the evidence examining the relationship between the built environment, physical activity, and obesity-related diseases. We also consider how other environmental factors may interact with the built environment to influence metabolic health, highlighting challenges in understanding causal relationships in this area of research.
On average, individuals who live in highly walkable neighborhoods have increased levels of physical activity and lower body weights than those living in less walkable areas.
In studies based on large populations, high neighborhood walkability was associated with a lower incidence of diabetes and hypertension.
In many studies, neighborhood-level differences in metabolic health were significant after accounting for age, sex, and the social characteristics of local residents; however, important inconsistencies were noted.
Other characteristics of neighborhood environments may offset or magnify the benefits of neighborhood walkability.
Lower concentrations of fast-food restaurants and air pollution and greater proximity to green space were associated with a reduced risk of obesity and diabetes.
Systematic differences in neighborhood design and resources may contribute to structural social and racial inequities in health.
Further research is needed to understand which policies (or combination thereof) can optimally improve metabolic health at a population level.
The Weight of Place
The role of place in shaping health has been a long-recognized concept that explains differences in disease burden across populations. In Air, Water and Place, Hippocrates theorized that physical features of the environment, such as the winds and waters, explained variations in the constitution and illnesses of citizens in different cities (1). The quality of air and water are just as vital to our health today as in Hippocrates’ time, as reflected in the number of attributable deaths to air pollution, crises surrounding lead in drinking water, and the COVID-19 pandemic (2, 3). However, human health is also shaped deeply by the physical environments we construct. In 1950, one third of the world’s population resided in urban areas, but this figure has grown substantially to more than half (55%) of people globally, with some areas (eg, North America, 82%) having substantially higher proportions of residents residing in urban areas (4). In the latter half of the 20th century, urban design practices in North America and many other regions have led to the development of car-oriented, suburban areas, characterized by low population density (urban sprawl); the separation of residential developments from retail, commercial, and business areas; and street designs that lack pedestrian infrastructure (5). With the increasing dominance of this mode of development came a recognition that these new built environments made walking and cycling for transportation impractical, which in turn may adversely affect health. There is an accumulating body of literature describing how built environments might contribute to the risk of cardiometabolic diseases and putative mechanisms by which it might exert these effects (6-14).
The aim of this review is to describe major concepts and recent findings on the relation between the built environment and metabolic health and disease, including conditions such as obesity, insulin resistance, type 2 diabetes mellitus, and hypertension. The intent was not to undertake a systematic review of research conducted in each of these fields; instead we direct interested readers to several excellent systematic reviews on these individual topics (15-,17). We first provide a brief discussion of the concept of the built environment. Then, we review recent work examining connections between different components of the built environment, physical activity, obesity, and diabetes. We also consider work in other fields of environmental epidemiology that have attempted to contextualize the built environment, along with other environmental correlates of metabolic health, such as the retail food environment, neighborhood socioeconomic status, air pollution, and green space. Finally, we conclude with a discussion of the major challenges facing research on the built environment and future opportunities in this field to understand the potential benefits of built environment interventions on metabolic health. Of note, we focus on research from North America, Australia, and Western Europe, and many important urban areas across Central and South America, Africa, and Asia are not captured in this review, due predominantly to limitations in data availability.
The Built Environment
The built environment consists of all physical structures in the environment that have been made or modified by humans (18, 19). Some of its elements are easily imagined, such as buildings, street design, and land uses (eg, residential, commercial, industrial areas), but many definitions also include other features, such as transportation infrastructure and parks. The built environment can be discussed on many scales, from the macro (eg, metropolitan areas, multicity regions), to meso (eg, neighborhoods), to micro (eg, block faces, benches, curb cuts, street crossings) and from both objective and subjective perspectives. As a general concept, it also relates to other popular terms such as urban sprawl. The primary focus of this review is on the attributes of the built environment that are believed to impact metabolic health by encouraging physical activity, with a major emphasis on neighborhood walkability. We will also discuss how other environmental attributes, such as neighborhood socioeconomic status, food environment, air pollution, and green space coexist and potentially interact with the built environment to affect metabolic health. Over the 20th century, a mode of development emerged favoring separation of land uses (eg, residential from nonresidential spaces), low-density housing, and curvilinear street design (5) (Fig. 1). Drivers of these changes included the growing affluence among the middle class, improvements in road infrastructure allowing individuals to commute to work more efficiently, and zoning policies aimed at reducing congestion in residential areas, thereby providing residents with larger lot sizes and more space for parking (5, 20). Many of these neighborhood design features favored or necessitated the use of automobiles for transportation and excluded walking as impractical or dangerous. In contrast, highly “walkable” or “activity-friendly” developments generally have higher densities of residents and services, a greater mix of land uses, and a more connected street network (Fig. 1). Over the past several decades, models have emerged in the transportation literature to capture the variables that determine a preference for driving over other modes of transportation, variously referred to as the “3-Ds,” “5-Ds,” or other iterations (21, 22). These D variables can include (1) density (of population, residences, jobs, etc.), (2) diversity (of land uses), (3) design (of street networks, streetscapes, and surrounding locations), (4) destination accessibility, and (5) distance to transit.
Overview of differences between built environments with high and low levels of walkability. Built environments that support functional physical activity, such as walking to work, to local transit, or to perform errands (active transportation) often have unique designs that contrast with less walkable, suburban developments. In this figure, 2 neighborhoods in the Greater Toronto Area, Canada, are shown along with the distances that can be traveled in a 10-minute walk. Walkable areas (left panel) typically have a grid-like street network with shorter block lengths, minimizing the distance that one needs to travel to reach nearby locations and increasing the overall distance that can be traveled by foot as walking routes are more efficient. In contrast, less walkable neighborhoods (right panel) typically have a curvilinear street design and fewer street intersections, making travel less efficient even when the straight line distance between 2 locations is small. Population and residential density is increased in highly walkable areas, favoring a greater concentration of jobs and services and a more compact urban design. These areas also have a greater mixture of land use types (eg, residential and commercial), which makes it easier to walk or cycle when completing short trips for daily activities.
Within the built environment literature, “walkability” has become a popular concept that reflects the extent to which a neighborhood encourages and supports walking as a means of transportation. This definition owes a great deal to previous concepts such as the “D variables” referenced in the transportation literature, which are typically reflected in these measures. While a number of walkability indices have been developed, they share common variables capturing features such as population density, land use, and street connectivity. Furthermore, walkability indices typically combine a variety of these variables into a single measure (23, 24). This combination is important, as changes in any single built environment attribute may be insufficient on its own to meaningfully impact population health (25).
The Built Environment and Metabolic Health
Physical Activity
Neighborhood designs that make it easier to engage in active forms of transportation, including walking and cycling, are hypothesized to increase physical activity. Studying 14 cities across 10 countries, the International Physical Activity and Environment Network study found that residents living in the most activity-friendly neighborhoods (on the basis of walkability, transit options, and park access) accrued up to 89 additional minutes per week of moderate to vigorous physical activity compared to the least activity-friendly neighborhoods studied (26). Thielman et al reported similar findings using data from the Canadian Health Measures Survey; those living in the most versus least walkable neighborhoods performed an average of 82.6 minutes more per week of moderate to vigorous physical activity (27). Based on these studies, residents that live in neighborhoods that are more conducive to walking appear to be more likely to achieve physical activity targets (ie, 150 minutes per week of moderate to vigorous physical activity), as recommended by the US Department of Health and Human Services and other agencies’ guidelines (28, 29).
These studies offer evidence for a strong association between urban design and physical activity, based on observations across different world regions and assessed using objective measurement techniques to capture physical activity. However, there is the potential for bias caused by self-selection, in that individuals who choose to live in walkable areas may be healthier and more active than those who choose to live in more car-dependent, suburban areas that are less conducive for walking. However, the opposite may also be true—individuals residing in the downtown neighborhoods of cities may have worse underlying health than those in suburban settings for reasons that relate to structural differences in the social and racial characteristics of urban populations and to their exposure to other adverse environmental factors. Recently, a number of studies have examined the health benefits of moving from one type of neighborhood to another, which enables researchers to compare the risk of disease in the same person under different conditions. By using analyses such as fixed-effects models, studies can focus on changes that occur within individuals over time, removing the impact of risk factors that contribute to differences across populations but that remain relatively stable over time in a given individual (eg, socioeconomic status or individual preferences) or that remain static (eg, sex at birth, ethnicity, genetic susceptibility). However, these studies can have other important sources of bias that are difficult to control for, such as the reasons why people move from one neighborhood to another (eg, changes of employment or family structure), which may in themselves be potent confounders. Nevertheless, this type of study design forms the basis of a “natural” experiment. Examples include studies that follow individuals who move between neighborhoods with differing urban designs or those who stay in the same neighborhood but are exposed to changes in policies or neighborhood infrastructure, such as new transit lines or bicycle paths. Further examples of each are described in the following discussion.
The Residential Environment Project (RESIDE) study conducted by Giles-Corti et al identified individuals who intended to move to new developments around Perth, Australia, and followed them for changes in physical activity (30, 31). The study showed significant increases in transportation-related walking among those whose access to walkable shops and services increased, compared to those whose access declined or remained the same at 1 year (30) and 7 years of follow-up (31). This study was unique in its careful capture of the reasons for relocation, prospective assessment of neighborhood exposures, and long follow-up window. Another key set of studies arose from the Multi-Ethnic Study of Atherosclerosis (MESA) Study in the United States, which evaluated participants who moved during follow-up. Hirsch et al found that for every 10-point increase in the Walk Score of their new neighborhood (out of a scale of 100), MESA participants spent 16 additional minutes per week engaged in transport-related physical activity and were 11% more likely to meet the goals outlined in the US Surgeon General’s “Every Body Walk!” campaign (32). Similarly, using data from Canada’s National Population Health Survey, Wasfi et al found that participants who moved from a less walkable neighborhood to a highly walkable one engaged in an additional hour or more of utilitarian walking per week compared to those moving to a similar neighborhood type as the one they left (33). In contrast, the same was not true among participants enrolled in the Coronary Artery Risk Development in Young Adults (CARDIA) study who were from Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California (34).
From recent systematic reviews of the literature, the majority of studies found that high neighborhood walkability was associated with a significantly increased rate of physical activity compared to low neighborhood walkability (16, 35-38), including studies using self-report (16, 36) and objective measures of physical activity such as accelerometry (35). While most studies have examined this phenomenon in younger and middle-aged populations, meta-analyses that focused specifically on older adults also support an association between walkability and physical activity levels within this population (37, 38). Despite this, authors of each of the previously noted systematic reviews found substantial variability in study findings. For instance, in a review of natural experiments, including longitudinal analyses of “movers” from one neighborhood to another, Kärmeniemi et al reported a statistically significant association between changes in walkability and physical activity in 2 of 3 studies that focused on transportation-related activity, but in only 1 in 5 studies that examined overall levels of physical activity (16). However, the majority (4 out of 6) reported positive associations if they focused specifically on walkable access to shops, parks, and public transportation, suggesting that having places to walk to is the biggest determinant of pedestrian activity.
The variability in study findings may be due to a number a factors, including heterogeneity in the regions and populations studied, as well as in the tools used to measure walkability, the outcomes tested, and the study design itself. Furthermore, since fixed-effects models focus on changes within individuals over time, these analyses may lack power. Indeed, in the RESIDE study, researchers noted that the vast majority of participants moved to a neighborhood that was similar to or had fewer walkable shops and services than the one they had left (30, 31). This was also true among participants in the CARDIA study, which did not detect an association between neighborhood walkability and physical activity. For most members of their sample, walkability did not appreciably change during follow-up, and those in whom it did experienced only small changes in mean walkability scores, likely limiting their ability to detect a significant association. Finally, issues with follow-up are of major concern for longitudinal studies. In the previously discussed CARDIA analysis (34), 26% of the eligible sample were excluded due to missing data, while 28% of included participants were lost to follow-up. Collectively, these issues can contribute to selection bias and other limitations, including a reduction in statistical power.
In summary, important advances have been made in understanding the role of the built environment in promoting physical activity in recent decades. Associations reported in the literature suggest that the magnitude of differences in physical activity across neighborhood types is clinically meaningful, translating in some cases to more than 50% of weekly recommended minutes of aerobic activity. Moreover, these findings were evident across a range of study designs and methodologies and appear to be generalizable to different geographic regions. However, given the lack of interventional studies in this area, important questions remain regarding the specific components of the built environment that need to be modified (and to what extent) to increase physical activity at a population level, particularly among those at high risk of developing diabetes. Nevertheless, the evidence to date suggests that more walkable built environments have the potential to improve an individual’s participation in physical activity.
Obesity
While there have been a large number of studies examining the association between the built environment and physical activity, far fewer studies have considered clinical outcomes such as diabetes, hypertension, or cardiovascular disease. Among clinical outcomes, weight-related endpoints, such as body mass index (BMI) and obesity, have been the most commonly studied. The early literature in this field was predominated by smaller, cross-sectional studies. Hence, we have focused here on larger studies that were able to follow a representative sample of the population or a recruited cohort over time and systematic reviews of the literature more broadly.
A few studies have examined the relationship between the built environment and obesity in a large population-based sample, including a study by Creatore et al involving 15 cities in Ontario, Canada (Fig. 2) (7). In this time-series analysis, the population-weighted prevalence of overweight and obesity was substantially lower among young and middle-aged adults who were living in neighborhoods that were highly walkability compared to those living in low walkability areas, even after accounting for difference in age, sex, ethnicity, and area income (absolute difference in top vs bottom quintile: −10%, 43% vs 53%). The adjusted rates of overweight/obesity in this representative sample of the population (n = 32 767) remained stable in highly walkable areas over the 12-year period between 2001 and 2012, while rates significantly increased over time in less walkable areas. These findings are supported by a longitudinal analysis of the National Population Health Survey (n = 2935) by Wasfi et al, which explored the effects of walkability on self-reported changes in body weight among adults aged 18 to 55 years who resided in an urban area between 1994 and 2006 (9). Their analysis found that age-related increases in BMI were substantially diminished among men who moved from a neighborhood with low walkability to a neighborhood with high walkability during the follow-up period compared to those who consistently lived in a setting with low walkability. Similarly, men who moved from a neighborhood with high walkability to neighborhood with low walkability on follow-up experienced a steeper rise in BMI over time than those living in highly walkable neighborhoods during the entire period. Of note, this pattern was not apparent among women (9).
Relationships between neighborhood walkability, obesity, and diabetes in Ontario, Canada. Adjusted prevalence of overweight/obesity and incidence of diabetes (95% CI) across neighborhoods in 15 municipalities in southern Ontario. A consistent pattern is seen whereby the most walkable neighborhoods (quintile 5) had a lower prevalence of overweight and obesity across all time points from 2001 to 2012, with no statistically significant changes over time. In contrast, less walkable neighborhoods (quintiles 1-4) had small but significant increases in the prevalence of overweight and obesity over time. Comparably, individuals in the most walkable neighborhoods had a lower incidence of diabetes and demonstrated a significant decrease in incidence over time, whereas less walkable neighborhoods (walkability quintiles 1-4) demonstrated an initial rise and fall in diabetes incidence, leading to no significant difference between rates in 2012 compared to 2001. Overweight and obesity rates were calculated using data from participants in the Canadian Community Health Survey between 2001 and 2011/2012. Diabetes incidence was calculated using the Ontario Diabetes Database, which captures diabetes cases using records from hospitalizations and physician service claims. Values were adjusted for age, sex, income, and ethnicity. Source: This figure originally appeared in Creatore et al (7) and is used under license by the copyright holder.
Results were less consistent in several American studies based on recruited cohorts. For instance, individuals enrolled in MESA (n = 701) who moved to a more densely populated neighborhood experienced a modest reduction in BMI, despite significant increases in transportation-related physical activity (mean change: −0.06 kg/m2 per 10-point increase in Walk Score) (32). Conversely, participants in the CARDIA study (n = 1079) who moved to a more walkable neighborhood did not experience a change in BMI or waist circumference in either random effects and fixed effects models (34).
There have been several systematic reviews on this topic. Early reviews, such as those by Feng et al (39)., Durand et al (40), and Mackenbach et al (41) reported inconsistent findings, likely due to heterogeneity in study design and limited power to detect differences across groups. Most early studies (~90%) were cross-sectional in design, and few examined composite measures of walkability (19). A more recent review by Chandrabose et al published in 2019 included only longitudinal studies (n = 13) and concluded that individuals residing in neighborhoods with more favorable characteristics (high neighborhood walkability, reduced urban sprawl, and/or increased access to recreational facilities) had a lower mean BMI, waist circumference, and body weight and lower prevalence of obesity (17). Many of the limitations raised in the literature on physical activity also apply to studies focused on obesity. In addition, in several studies, a lack of variation in built environment exposures (eg, insufficient numbers of people living in or moving to high walkability neighborhoods) has likely made it more challenging to identify an association between built environment features and health outcomes.
Overall, there appears to be a growing consensus regarding the benefits of environments that support physical activity for obesity prevention. In general, studies that analyzed data from representative, population-based samples have shown more consistent associations than traditional cohort studies, which may recruit participants who are healthier than those in the general population (42). Moreover, population-based studies tend to be larger and therefore are better powered to identify small, but important, differences across populations, and the detection of where people move to may be more complete in these analyses. On the contrary, cohort studies tend to collect more detailed exposure and clinical data on their samples, including measured height and weight and thus can study changes in clinical markers of disease development, as discussed in the following section. Lastly, studies that found a beneficial relationship between high neighborhood walkability and body weight are well aligned with evidence that active commuting patterns (walking or cycling) are associated with lower rates of obesity, while routine car use is associated with an increase in likelihood (43-45). Thus, there are compelling reasons to consider the potential benefits of the built environment on diabetes development.
Prediabetes and Insulin Resistance
Relatively few studies have explored the relationship between the built environment and early markers of diabetes risk. Using administrative data from Ontario, Canada, Fazli et al studied 1.1 million adults living in urban centers who had normal glucose values on routine laboratory testing and no prior diagnosis of diabetes. After 8 years of follow-up, the incidence of prediabetes was 20% higher among immigrants living in the least vs most walkable areas (with similar differences noted in the general population); however, this association varied according to the world region from which they emigrated (46). Among Southeast Asian immigrants, prediabetes incidence was 32% higher in low vs high walkability settings, while no significant association was seen among immigrants from South Asia with respect to walkability and prediabetes. Susceptibility to the effects of the built environment may vary across ethnic groups due to differences in cultural norms, attitudes, and preferences for engaging in transportation-related physical activity (eg, walking) vs other forms of physical activity (eg, gym or home-based exercises). It may also reflect differences in the social characteristics of neighborhoods, such as social cohesion, which may be greater in suburban areas that are ethno-cultural enclaves (communities of individuals who share the same cultural identity), despite being less walkable.
Other cohort studies from Australia and the United States have reported that middle-aged residents from walkable neighborhoods (or those with characteristics tied to walkability) had a smaller rise in hemoglobin A1c (HbA1c) (47) or fasting plasma glucose (FPG) levels (48, 49) over time compared to those living in less walkable areas. While these results support the influence of walkability on glycemic control over time, the differences between groups were small and may not be clinically meaningful (ie, likely to result in an important difference in clinical outcomes between groups). In the CARDIA (n = 1037) (34) and MESA (n = 583) (50) cohorts, participants who moved to a less walkable neighborhood were no more likely to develop insulin resistance as measured by homeostatic model assessment for insulin resistance (34) or to experience a rise in FPG (50). Many of these studies excluded people with diabetes from their analysis but did not exclude those with prediabetes or other indicators of insulin resistance at baseline. An exception was the North West Adelaide Health Study, which observed a 12% decrease in the likelihood of developing either prediabetes or diabetes [HbA1c ≥ 5.7% (38.8 mmol/mol) or FPG ≥ 5.6 mmol/L (100 mg/dL)] for every 1 SD increase in walkability over a 3.5-year period (51). There was a sizeable number of individuals with prediabetes or diabetes at baseline (45% of their cohort) who were excluded from this study. In their meta-analysis, Chandrabose et al pooled findings for individuals with and without diabetes (49) but adjusted for the use of antihyperglycemic agents. While they found an inverse relationship between walkability and glycemia, it is unclear whether the results were driven by better glucose levels in those with established diabetes living in highly walkable settings or whether walkability is a protective factor against early stages of diabetes development. As before, many of these cohort studies had comparatively small samples in their longitudinal models, resulting in limited power, and may have been affected by selection bias caused by differences in the characteristics of populations residing in urban and suburban areas. For example, in some cities, heavily urbanized areas have higher concentrations of poverty, racialized populations, or other marginalized groups, while in other cities, these areas may in fact be wealthier. Accounting for these factors is critical for understanding reported associations and to identify populations that are most susceptible to environmental effects.
In summary, early studies suggest an association between neighborhood walkability and both markers of glycemic control and insulin resistance. However, few studies adequately controlled for socioeconomic status and race, which are important determinants of diabetes risk. These factors could cloud the underlying relationship between the built environment and metabolic health or may modify a population’s responsiveness to a particular environmental exposure. Further research is needed to understand the environmental determinants that will have the greatest impact for one population over another to optimize our community-level approaches to diabetes prevention.
Diabetes
A number of studies have examined the association between walkability and the development of diabetes. Among the earliest of these, Booth et al used administrative healthcare data to study the relationship between walkability on diabetes incidence among 1.6 million working-aged adults who were living in Toronto, Canada, in 2005 (6). The study demonstrated a 30% to 50% higher likelihood of developing diabetes among long-term residents and recent immigrants living in low vs highly walkable areas. In the broader population of young and middle-aged adults in southern Ontario (n = ~3 million), residents in the most walkable neighborhoods had a substantially lower incidence of diabetes than those living in less walkable areas (7), after adjusting for age, sex, neighborhood socioeconomic status, and ethnicity (Fig. 2). These findings were consistent with differences in body weight across neighborhoods, as discussed earlier, and appeared to be driven in large part by differences in transportation choices. Furthermore, there appeared to be a threshold effect, whereby the neighborhoods with the highest walkability experienced substantially lower rates of overweight, obesity, and diabetes incidence and far greater levels of walking, cycling, and public transit use than all other categories of walkability. To account for systematic differences in population characteristics and the presence of comorbidities, Booth et al undertook a propensity score-based analysis known as inverse probability of treatment weighting (52). After pooling estimates for 5 large Canadian urban areas, the study reported a 15% lower weighted incidence of diabetes over a 10-year period in areas with high vs low walkability. However, these findings were observed only in working-aged adults (<65) and were not significant among older individuals. In addition, pooled estimates were comparable across socioeconomic status and immigration groups among the younger population (52).
Several cohort studies performed in other countries have shown positive associations between walkability and the incidence of diabetes, including the North West Adelaide Health Study, which had a combined endpoint of diabetes and prediabetes development (51). In addition, Christine et al discovered a 20% lower incidence of diabetes among members of the MESA cohort who reported living in a supportive environment for physical activity compared to those who rated their neighborhood as unsupportive (53). However, some studies have reported neutral effects. In an African American cohort (Jackson Health Study) by Gebreab et al, access to neighborhood resources, such as gyms, swimming pools, and dance studios, was unrelated to diabetes incidence (54). Using a population-based sample of adults living in Copenhagen, Denmark, Sundquist et al found that walkability appeared to have a protective effect against the development of diabetes; however, their models were no longer significant after adjusting for individual-level income status. In addition, they analyzed data from sibling-pairs who moved to different neighborhoods in adulthood, as an attempt to control for shared familial environments; however, the results remained nonsignificant (55).
A recent meta-analysis published prior to 2017 by den Braver et al found a 21% reduction in the pooled estimate for diabetes risk in neighborhoods with high vs low walkability, with 5 out of 6 studies demonstrating a statistically significant relationship (15) (Fig. 3). However, only 3 of 6 studies were longitudinal and thus were able to look at the development of new diabetes cases (incidence). Furthermore, the overall findings were largely driven by bigger population studies from Canada, and thus these findings may not be fully generalizable.
Meta-analysis of associations between walkability and diabetes risk/prevalence. This meta-analysis of 6 studies demonstrates an overall association between residing in a more vs less walkable neighborhood and a lower risk/prevalence of diabetes mellitus (relative risk = 0.76, 95% CI: 0.72, 0.87). Studies included in the meta-analysis were graded moderate to strong using Mackenbach et al’s Quality Assessment Tool for Quantitative Studies. Source: This figure originally appeared in den Braver et al (15) and is used under a Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium provided appropriate credit is given to the original authors. https://doi.org/10.1186/s12916-017-0997-z
Taken together, there appears to be a protective relationship between neighborhood walkability and the likelihood of having diabetes, but studies from different countries (Canada, United States, Australia, and Denmark) demonstrated variable results, potentially due to differences in study design, population characteristics, or differences in the quality of the built environment. European cities tend to have greater infrastructure to support transit use, cycling, and pedestrian activities, and consequently, their populations are substantially less dependent on cars. Thus, low walkability neighborhoods in Copenhagen may represent a much more supportive environment than low walkability areas in North American cities. The socioeconomic and racial characteristics of people who live in different neighborhood types may also vary between countries, with the downtown core of many North American cities having higher levels of poverty. Finally, as we discussed in the previous sections, associations between neighborhood walkability and diabetes were strongest in population-based samples compared to recruited cohorts, which may be due to the fact that some cohorts had relatively small sample sizes and less variation in built environment exposures between populations or due to other sources of confounding that were not addressed in these studies.
Blood Pressure and Hypertension
Given the apparent benefits of neighborhood walkability on metabolic outcomes such as obesity and diabetes, several studies have explored whether these environments have favorable effects on blood pressure. In a population-based Canadian study, moving from an unwalkable to a highly walkable neighborhood was associated with a 54% lower likelihood of being diagnosed with hypertension over a 10-year follow-up period compared with those who moved from one unwalkable neighborhood to another (56). Importantly, this analysis used propensity-score matching to control for a variety of social- and health-related factors, including household income, race, diet, baseline physical activity, stress, and comorbidities. In addition, the authors did not observe any differences in the rate at which participants received preventive healthcare during follow-up. These findings have been supported by other large, population-based studies with sample sizes of up to 2.5 million, demonstrating that individuals residing in highly walkable neighborhoods have lower mean systolic and diastolic blood pressure (34, 57-59) and lower odds of being diagnosed with hypertension (56, 60-62). Similar findings were reported from a cross-sectional analysis of the RECORD cohort in France (59) and the UK Biobank (62), as well as longitudinal studies conducted in the United States, including CARDIA (34) and a Portland, Oregon–based cohort (58). Conversely, several analyses of the MESA study found null or counterintuitive findings. Cross-sectional and longitudinal analyses of participants aged 45 to 85 years in MESA found that those living in neighborhoods rated as being more supportive of physical activity were as likely to have hypertension as those who did not, and both mean systolic and diastolic blood pressure levels were similar between groups (50, 63). Cross-sectional (n = 5970) and longitudinal (n = 3145) analyses of the Western Australian Health and Wellbeing Surveillance System Survey and North West Adelaide Health Study in Australia also did not find an association between walkability and self-reported or objectively defined hypertension (51, 64).
The systematic review by Chandrabose et al reported a strong link between higher walkability and improved blood pressure outcomes; in fact, it was the most consistently observed association among cardiovascular risk factors, outside of weight-related outcomes and diabetes (17). The 2 studies in this review that did not demonstrate significant associations included the longitudinal analysis of data from MESA and CARDIA (as discussed earlier), which may have been hampered by their smaller sample sizes (34, 50). Appropriate outcome measurement is also important in hypertension, where self-reported diagnoses may suffer from both underreporting (eg, patients not being aware of their diagnosis) and overreporting (eg, a single, 1-time measure of blood pressure). To capture differences across populations, it may be necessary to compare large numbers of individuals across a range of metropolitan areas to generate sufficient variation in walkability exposures, as occurred in studies that identified significant associations (57, 61). Müller-Riemenschneider et al, for instance, commented on how Perth and other Australian cities have limited variation in walkability, which may have limited their ability to find differences in outcomes across groups (64).
Lipids
In contrast to other cardiometabolic outcomes, relatively few studies have included lipid measurements in their outcomes. Of those that have, the results have been mixed and generally not clinically significant due to their small effect sizes. For instance, using data from primary care practices located within southern Ontario cities (n = 44 448), Howell et al found lower high-density lipoprotein (HDL) cholesterol levels among individuals who resided in the least vs most walkable neighborhoods (−1.67 mg/dL), but no associations were noted with respect to total cholesterol (57). Loo et al found similar differences in HDL cholesterol levels among adults aged ≥40 years (n = 78 023) from primary care practices in Toronto (+2.0 mg/dL in highly walkable areas) but no differences in triglycerides or low-density lipoprotein (LDL) cholesterol levels were observed (65). These studies were cross-sectional, which limits the conclusions that can be made from these findings. However, other studies that followed a recruited cohort over time have generally not detected an association between walkability and changes in mean HDL cholesterol (34, 50), LDL cholesterol (34, 50), triglycerides (34, 49, 50), or aggregated measures of dyslipidemia (51). In a sensitivity analysis, Braun et al noted a paradoxical increase in triglycerides for individuals moving to more walkable neighborhoods, but the changes observed (increase in triglycerides of 1.01 mg/dL per 10-unit change in walkability) were unlikely to be clinically meaningful (50).
While high triglycerides and low HDL cholesterol levels are early markers of insulin resistance, mean values are a fairly insensitive marker of the metabolic health of a population more broadly. Since most members of a community have normal triglyceride and HDL cholesterol levels, small changes that are averaged across the population may obscure meaningful differences in the prevalence of these abnormalities in susceptible individuals. None of the previously discussed studies focused on populations who had dyslipidemia at baseline or other markers of insulin resistance. A systematic review of randomized controlled trials of walking interventions did not find a statistically significant relationship between walking and total, LDL, or HDL cholesterol concentrations, suggesting that these measures may be less amenable to modification by the kind of physical activity generated by neighborhood walkability (66). Moreover, factors such as lipid-lowering medications or diet may offset any inherent differences that exist between people living in different neighborhood types and were not accounted for in all cases. With respect to diet, no study examined this as a covariate in their analysis. In 1 case, the authors selected covariates based on hypothesized causal models, which may have accounted for possible “off-target” effects of the built environment on lipids mediated by changes in dietary behaviors (50). In other cases, this reflects an absence of dietary data in established databases (65). Further exploration of this issue is warranted before conclusions are made regarding associations between the built environment and lipid outcomes. In summary, while available evidence is limited, existing studies have not demonstrated a significant association between built environment exposures and markers of dyslipidemia.
Summary
Overall, a growing body of research suggests that denser, more walkable urban areas are associated with improved metabolic health compared to less walkable, car-oriented suburban areas. A large number of studies, including globally representative samples, have documented greater levels of physical activity among residents living in neighborhoods that have higher levels of walkability, with differences that are clinically meaningful. Furthermore, many studies have demonstrated links between walkability and BMI, obesity, prediabetes, diabetes, and hypertension. These findings support the notion that city policies that increase neighborhood walkability could have tangible benefits for the metabolic health of a population. Nevertheless, research in this area has been limited by a lack of prospective cohort studies that control for neighborhood self-selection, which would improve the ability to demonstrate a causal relationship between walkability and metabolic outcomes. Furthermore, moving toward more standardized frameworks for the measurement of built environment variables may aid in evidence synthesis and in explaining discrepant findings between studies as they arise.
The Built, Social, and Natural Environment
Although it is convenient to discuss the built environment’s impact on health in isolation, there is a complex interplay between environmental exposures. Neighborhood socioeconomic conditions, food environments, air pollution levels, and natural features such as green space may independently influence metabolic health and must be considered when isolating the effects of the built environment.
Neighborhood Socioeconomic Status
Neighborhood income is often lower in highly walkable downtown or “inner city” neighborhoods of large municipalities in the United States (67) and Canada (7, 68, 69), which may contribute to disparities across studies of the built environment. Further, neighborhood socioeconomic status has an important influence on the health of a population over and above the effects that household income places on individuals (70). The former represents the health impact of concentrated poverty or wealth through broader factors, such as access to neighborhood resources, social and cultural norms, social supports, and safety (70, 71). Neighborhood socioeconomic status is often measured using aggregated data on income collected from local residents, for example, from census estimates of the median household income level for a given area or region. Many epidemiologic studies have found an inverse relationship between neighborhood income and weight-related outcomes (72-74) or diabetes (75, 76). However, this area of research is also notable for a randomized trial that assessed whether area-level economic conditions have an impact on metabolic health. The Moving to Opportunity study was a randomized controlled trial conducted in partnership with the US Department of Housing and Urban Development that randomized 4498 families living in public housing to receive either (1) housing subsidies to allow them to move to an area with a low level of poverty (low-poverty vouchers), along with short-term counseling to assist with their housing search; (2) housing subsidies with no restrictions on where they could move nor any counseling (traditional vouchers); or (3) no additional assistance (77). Overall, mothers in families that were randomized to low-poverty vouchers had a lower prevalence of severe obesity (defined either as a BMI ≥ 35 or BMI ≥ 40 kg/m2) and lower mean HbA1c after 10 to 15 years of follow-up compared to no-assistance controls, whereas those receiving traditional vouchers did not differ with respect to metabolic health outcomes in comparison to controls. Although there are a number of limitations to this study, its findings, along with those of observational studies in this area, provide relatively strong evidence that neighborhood-level poverty is causally associated with obesity and diabetes risk.
Researchers have also demonstrated a significant interaction between built and social environments with respect to a variety of health outcomes, including physical activity levels (78-80), body weight (81, 82), and diabetes (6, 52, 55). Social characteristics could mitigate the benefits of walkability by creating uncomfortable or unsafe environments for local residents to walk (eg, crime, overpolicing) or could potentiate its effects (eg, greater social cohesion, local norms favoring physical activity) (78, 83-85). Broadly speaking, the literature in this area is quite varied, showing diminished (81, 86), neutral (52, 55, 79), or greater (6, 87) benefits of walkability in socially disadvantaged compared to socially advantaged areas. Given this heterogeneity, further work is needed to clarify these relationships so that mitigating factors can be addressed. Moreover, built environment policies often target economically privileged areas, which may have the unintended consequence of exacerbating existing health inequities.
The Food Environment
The retail food environment has been a frequent target of public health intervention. A range of policies have been used as an incentive to promote healthier eating patterns, from taxes on sugar-sweetened beverages to bans on opening new fast-food restaurants in certain regions (88-91). However, the relationships between food environments and metabolic outcomes is complex. Systematic reviews examining a link between the food environment and obesity have reported mixed findings or no evidence for an association (92, 93). Some studies have even reported paradoxical findings, such as increased obesity rates in areas that have a greater availability of grocery stores or a decreased likelihood of obesity in areas that have more fast-food outlets (92, 93). These unexpected findings might be partly explained by the interrelationship between neighborhood walkability and food retail. As might be expected, denser urban areas generally have a greater concentration of retail outlets than less dense areas and therefore have higher numbers of retail food establishments of all types, including unhealthy retailers (94-96). Further, in some cities, there has been an exodus of large supermarkets from city centers, with higher volumes or sizes of outlets (eg, big box stores) in suburban areas (97-99). In studies that adjusted for aspects of neighborhood walkability, paradoxical associations between healthy or unhealthy food retailers and BMI were greatly attenuated and expected associations started to emerge. For instance, in a study by Rundle et al, residents living in New York City neighborhoods that had a high density of supermarkets or grocery stores (number per km2) had a ~1 point lower BMI and 13% lower likelihood of obesity than those living in low-density areas, after accounting for sociodemographic characteristics and population density (100). Socioeconomic status is a major confounder in these relationships. A number of studies from the United States and the United Kingdome have suggested that lower income communities have relatively fewer green grocers and supermarkets or no access to these stores (so-called food deserts) and higher numbers of unhealthy food retailers, such as convenience stores and fast-food restaurants, compared with socioeconomically advantaged neighborhoods (101, 102). This may relate to the socioeconomic distribution of populations within cities and the built environment in which lower income groups reside. In the North American context, low-income areas are often located in the center of cities and therefore have more restaurants and food retailers of any type (including unhealthy retailers) than high income neighborhoods (96, 103).
There is growing evidence that proximity to fast-food outlets is directly related to fast-food consumption patterns and the likelihood of being overweight or obese (92, 93). Indicators that reflect how fast food fits into the overall food environment show the strongest results. In several large studies from the United States and Canada, the ratio of unhealthy to healthy food retailers in an area was strongly related to the rate of obesity in the population (104, 105). This measure, referred to as the Retail Food Environment Index, correlates well with metabolic health outcomes but creates a dilemma in neighborhoods that lack supermarkets or other healthy food retailers since they have 0 denominators and therefore are typically dropped from the analysis (106). In other studies, this was addressed in part by examining the relative market share or proportion of food retailers that are considered unhealthy (107-110). Using data from the New York City Community Health Survey (n = 48 482), Stark et al found that the percentage of all food retailers in a given neighborhood that were selling fast food was directly related to BMI in areas that had low levels of poverty (108). A study on >50 000 participants in the UK Biobank found that such neighborhoods were associated with a 51% higher odds of obesity compared to areas where fast-food outlets comprised a minority of food retailers (109). Using a similar approach, Polsky et al found that the relative proportion and the absolute volume (total number) of fast-food outlets in an area magnify each other’s effects. Among high-volume neighborhoods (3-5 outlets) in Toronto and surrounding cities, obesity and diabetes incidence rates were 2.5 and 1.7 fold higher, respectively, among young and middle-aged adults living in areas where more than half of all local restaurants sold fast food compared to those in areas where fewer than 10% of restaurants did, after adjusting for socioeconomic status, walkability, and access to grocery stores (100, 111). These areas, referred to by some as “fast-food swamps,” are often located in highly walkable areas, highlighting their potential to mitigate the benefits of neighborhood walkability.
This research has led to a more nuanced understanding of the complex association between the retail food environment and health. In sum, while conceptual links between the availability of healthy and unhealthy food retailers and metabolic health are intuitive, further work is needed to tease apart how these different environmental factors interact to affect dietary behaviors and downstream health outcomes.
Air Pollution
In a number of studies, air pollutants, such as particulate matter (PM), organic carbon, nitrogen oxides, CO, and SO2 (112), have been found to be risk factors for adverse metabolic outcomes. Several meta-analyses have identified direct relationships between air pollution and diabetes (113-115), blood pressure (116), and hypertension (117), while there is more limited evidence regarding potential associations with other risk factors, such as BMI (118) or lipids (119). Focusing on diabetes-related studies, a 2020 meta-analysis by Yang et al found that each 10 ug/m3 increase in PM2.5 exposure was associated with a 10% increase in diabetes incidence, with comparable results for PM10. While some studies have found associations between NO2 concentrations and the neighborhood-level prevalence of obesity and diabetes, no consistent associations were observed between NO2 and incident diabetes (115).
The built environment and air pollution are deeply connected since population density, road networks, and other aspects of urban design are major determinants of car use. A number of studies from the United States and Canada have found that denser, more walkable urban areas have increased levels of ambient and traffic-related air pollution (60, 69, 120, 121). This may appear counterintuitive, since individuals who reside in more walkable neighborhoods are more likely to use nonmotorized transportation and generate less traffic-related air pollution (7, 122). However, this phenomenon likely reflects the high concentration of pedestrian activity and vehicle traffic in the center of densely populated areas, where people access shops and services. In addition to higher levels of exposure to air pollution, walkable areas may also potentiate the effects of air pollution if local residents are more likely to engage in outdoor physical activity (including walking) in these environments. To date, there has been limited examination of this possibility. In a recent Canadian study by Howell et al, individuals living in city neighborhoods that were more walkable were 25% to 35% less likely to have been diagnosed with diabetes or hypertension, respectively, if these areas also had low levels of traffic-related air pollution. However, differences by walkability status were eliminated at higher levels of traffic-related air pollution (60) (Fig. 4). Conversely, using data from the Nurses’ Health Study, James et al found that the relationship between walkability and BMI was stronger in areas that had higher concentrations of traffic-related air pollution (123). Several studies have examined whether the benefits of physical activity on health are reduced by ambient air pollution (124-127) but have had mixed results. Thus, further research in this area is needed.
Interaction between neighborhood walkability and traffic-related air pollution and the probability of diabetes. In Howell et al (60, 61), an antagonistic relationship between neighborhood walkability and traffic-related air pollution (indexed by a surrogate air pollutant, NO2) was identified. Individuals living in the most walkable neighborhoods (Q5) tended to have the lowest probability of having diabetes when levels of air pollution were low (5-20 ppb). However, at higher concentrations of traffic-related air pollution (eg, 30-40 ppb), the probability of having diabetes converged across neighborhood types (lowest: Q1) and no benefit to residing in a walkable neighborhood was observed. All probabilities presented were adjusted for age, sex, ethnicity, immigration history, neighborhood income, and number of medical comorbidities. Abbreviations: NO2, nitrogen dioxide. ppb, parts per billion; Q, quintile. Source: This figure originally appeared in Howell et al (60) and is used under the authors’ permission as the copyright holder.
Green Space
Neighborhood green spaces, including city parks, outdoor recreational areas, tree canopies, and nature trails, have generated substantial interest regarding their potential benefits for both physical and mental health (128). There are a number of putative mechanisms by which this might occur, including the protective effects of trees on the environment (eg, temperature regulation and shade, air pollution moderation), psychological well-being (eg, stress reduction, improved mood), social opportunities (social inclusiveness, networks and cohesion), and health behaviors (eg, increasing physical activity) (129-131). The combination of these factors may explain why proximity to green space (typically measured using satellite-based indices) has been linked to a variety of health outcomes including depression, dementia, cardiovascular disease, and all-cause mortality (132-137). In addition, several studies have identified inverse associations between the amount of greenspace near people’s homes and both BMI and the likelihood of being overweight or obese (138-141). Furthermore, there is some evidence to suggest that greenspace has positive benefits for downstream metabolic health outcomes. A number of cross-sectional studies have found favorable associations between greenspace exposures and the prevalence of diabetes (142-152), with a meta-analysis by den Braver et al reporting a significantly reduced odds of diabetes (9% lower) based on pooled estimates from 4 studies (15). Furthermore, Dalton et al identified a protective association between the amount of residential greenspace (woodlands, other natural areas and parks within 800 meters) and diabetes incidence among older adults enrolled in the European Prospective Investigation of Cancer cohort (Norfolk, United Kingdom). Diabetes incidence was 19% lower in the top vs bottom quartile of greenspace after adjusting for sociodemographic information, BMI, and family history of diabetes. Using data from the Sax Institute’s 45 and Up Study in Australia (n = 46 786), Astell-Burt and Feng found that a high degree of tree canopy (≥30% vs 0-9% of land area) within 1 mile of participant’s homes was associated with a 31% lower incidence of diabetes as well as a 17% and 22% lower incidence of hypertension and cardiovascular disease, respectively, after 6 years of follow-up, although total green space was associated with diabetes prevalence only (144). In another Australian study, living near public open spaces was associated with a lower incidence of combined diabetes/prediabetes, while the greenness of the space or its sports facilities were not (51).
Other factors in the built environment, such as walkability or air pollution, may contribute to these effects. In studies conducted in North American cities, communities that were more walkable had reduced levels of greenspace overall and decreased proximity to parks, albeit more opportunities for residents to walk to these amenities, if present (121, 123, 153, 154). Moreover, the combination of green space and walkable built environments may have additive effects on metabolic health and self-reported health (155, 156). For instance, using data from the UK Biobank, Sarkar found significant associations between greenspace and weight-related outcomes (BMI, waist circumference, whole body fat, and obesity), which was far greater in more urban areas (139). In contrast, in older women enrolled in the Nurses Health Study (aged ≥ 60 years), there were no interactions between neighborhood walkability and green space (123), and only walkability was related to BMI in multivariate models. In many studies, nonurban areas were included in the analyses, which may lead to paradoxical associations, since obesity levels tend to be higher in rural areas for reasons that relate to differences in the sociodemographic and occupational characteristics of populations, car-dependency, decreased access to healthcare, and social isolation.
Further research is needed to confirm and clarify the relationship between green space and metabolic health, although the evidence to date is highly supportive. Recent clinical trials suggest that people who were randomized to a “park” or “nature” prescription were more likely to continue to exercise after 6 months than people who received standard physical activity advice (157). These differences appeared to be mediated by the amount of time spent in nature being physically active. Further research is needed to understand the benefits of urban greenspace on physical activity and metabolic outcomes to inform healthy public policies. This includes recent interests in many jurisdictions throughout the world to create greener cities; with Barcelona, Melbourne, Seattle, Sydney, and Vancouver setting targets of 30% to 40% green cover by 2050. However, effective strategies to engage citizens from high-need communities to use natural environments and other resources for physical activity are needed.
Summary
It is clear that many environmental factors play a role in influencing metabolic health. Neighborhood exposures such as the level of air pollution, the availability of green spaces, the retail food environment, and social contexts have the potential to modulate a host of behavioral, psychological, and physiological responses that could alter an individual’s likelihood of developing metabolic disease. It is increasingly clear that these exposures may not act independently but may mutually reinforce or antagonize each other’s influence. While more evidence is needed to characterize the relationships between environmental exposures in different regions of the world, in the North American context, walkable built environments offer many benefits but have other, competing risk factors for metabolic disease, such as higher concentrations of traffic-related air pollution, a greater availability of fast food, reduced access to green space and trails, and a generally lower neighborhood socioeconomic status. This raises important questions for future research to identify which combination of factors may have the greatest protective influence against obesity and diabetes. By articulating what “healthier” neighborhoods might look like, it will help us to further refine our understanding of how urban environments affect our health and enable the successful implementation of healthy environmental policies.
Future Directions and Conclusions
Limitations of Current Research
Built environment research has evolved in recent years to encompass a broader range of health outcomes and employ novel data sources and innovative methods. Nevertheless, there are important limitations to the current literature that must be highlighted. These might broadly be categorized as challenges with study design, confounder assessment, and the measurement of built environment variables.
Some of the limitations in this field are similar to those faced by many maturing areas of epidemiologic research, namely an early reliance on cross-sectional designs that fundamentally limit researchers’ ability to make causal inferences. The preponderance of cross-sectional analyses was highlighted in many early reviews (39, 158) but is beginning to be addressed by a growing number of high-quality longitudinal studies (8, 9, 34, 52, 56). Still, further longitudinal analyses—in particular, of health outcomes such as diabetes and all-cause mortality—are needed. Additionally, given the maturity of this field of research, further efforts should be made to identify policy-relevant exposures and to evaluate both deliberate interventions and natural experiments. While more abstract concepts, such as neighborhood walkability, may be of primary interest in scientific research, concrete examples of how actionable policies and/or specific interventions affect population health may provide more impetus for policy changes and ease the translation of research findings into recommendations. As an example, the RESIDE Project in Perth, Australia, is evaluating local “Liveable Neighbourhoods Community Design Guidelines,” in which they are tracking individuals before and after moving to newly constructed neighborhoods that were identified by the local government to be either largely, partially, or noncompliant with the guidelines (159). These types of studies will provide valuable information on the effects of real-world policy decisions and the feasibility of implementing built environment interventions; including the alignment of stated planning goals and their execution (ie, how neighborhoods are actually built).
Built environment research often struggles with the ability to adequately control for important confounders. Given that this line of research is inherently observational, it is critical that studies identify, measure, and control for factors that confound our ability to assess whether built environment exposures are causally associated with better or worse health outcomes. A major limitation is the ability to control for individual self-selection (eg, choosing to live in a more favorable built environment) (26, 53, 60). Several approaches have been described to address these self-selection problems, including direct measurement of stated reasons for choosing a neighborhood or identifying an instrumental variable, a variable that is associated with the built environment exposure of interest and the health outcome but could only be caused by the exposure (160). These may be challenging to find without primary data collection, which limits the ability to reuse existing cohort and administrative data. In the absence of natural experiments, techniques such as these can address the lingering uncertainty regarding the causal interpretations of built environment findings.
Limitations also stem from how built environments are conceptualized and measured. In a 2010 review, Feng et al described over 51 built environment measures represented in the literature studied (39). While continuing to refine and explore new concepts is important to scientific development, the variety of measures analyzed has made it challenging to synthesize current evidence and compare findings across studies. As the field continues to mature, creating shared definitions and approaches may facilitate progress and reduce unnecessary duplication of research with similar underlying questions. A related issue stems from the technical approach to characterizing the built environment people are exposed to. Studies have relied heavily on “residential neighborhood” exposures, created using geographic information systems methodology, to describe the vicinity surrounding a participant’s home address (158). While this approach is convenient, it sidesteps the issue of how other, nonresidential exposures affect health behaviors and outcomes. These environments might be of interest in their own right (eg, workplace exposures) or may be contributors to an overall life-course exposure that is only partially assessed by capturing the home environment. Work by our group and others has identified how improved measurement of non-residential exposures to the built environment may affect the strength of association between the built environment and health behaviors (23, 110, 161). Even if questions surrounding the residential environment are of particular interest, omitting measurement of nonresidential exposures may bias empirical estimates (162). As GPS tracking becomes increasingly available through purpose-built devices or smart phone applications, researchers will have more access to temporally and spatially rich information. Greater adoption of these technologies will permit more accurate characterization of individuals’ exposures, reduce measurement error, and permit new types of questions to be addressed about environmental influences on health.
Opportunities
As built environment research has matured, new opportunities have arisen to expand the field’s scope, both in substance and scale. Research networks such as the Canadian Urban Environmental Health Research Consortium have created national data repositories that offer free, accessible, and standardized measures of environmental exposures (163). Furthermore a number of dashboards exist from which to download indicators from cities more broadly, including the US Centers for Disease Control and Prevention’s 500 Cities Project, the City Health Dashboard, and World Council of Cities Data (164-166). These organizations additionally serve as networks for research development, collaboration, and training. New approaches to characterizing the built environment are increasingly available through use of data such as Google Street View and Open Street Map (167-169). Google Street View data are useful to characterize street-level features of the built environment, such as traffic calming measures, neighborhood disorder, or building design variations, that would otherwise need to be collected through direct observation campaigns or surveys (167, 169). These methods may offer researchers access to more fine-grained information on the built environment and to hard-to-measure variables such as neighborhood aesthetic features, sidewalk disrepair, or street furniture. Use of machine learning techniques can aid researchers in processing these data so that features related to the “micro” environment in which people live can be studied more broadly and compared across populations living in different regions (170, 171).
Policy Implementation
Increasingly, local governments, governmental organizations, and private organizations have recognized the importance of the built environment on health. The World Health Organization, Centers for Disease Control and Prevention, the Public Health Agency of Canada, and the National Institute for Health and Care Excellence in the United Kingdom have highlighted the importance of the built environment in encouraging active lifestyles and improving population health (172-175). Other organizations, including the American Diabetes Association, American Heart Association, Diabetes Canada, and the Canadian Medical Association have also issued scientific or policy statements recognizing the role of the built environment as a determinant of physical activity and cardiometabolic health (176-179). While the direct implementation of built environment and health research into policy is still developing, some regional governments are beginning to include such provisions in local planning processes. In 2017, the region of Peel in Ontario, Canada, implemented a number of changes to its Master Plan, including requiring health assessments as part of their land development applications and requiring that planners incorporate an evidence-informed “Healthy Development Framework” into policies, as part of a push to encourage healthier urban design in development projects (180, 181). This example is 1 of numerous grassroots initiatives in the United States, Canada, and other countries; increasingly, many of these initiatives are receiving support through governmental (eg, US Environmental Protection Agency) and nongovernmental agencies (eg, Smart Growth America’s National Complete Streets Coalition), including tools, resources, workshops and assistance to develop complete street designs and revitalization projects (182, 183). Following policy changes over time will be critical to making a case for their adoption, more generally, and provides a unique opportunity for intervention research.
Conclusions
Over several decades, there has been a wealth of research on the built environment and metabolic health, with the overall evidence suggesting that walkable or activity-friendly environments provide more opportunities for physical activity and have the potential to decrease the burden of obesity and diabetes in the population. While there are important challenges facing this area of research, there is growing evidence from higher-quality longitudinal studies that support these conclusions. There have also been parallel advances in research on other environmental correlates of metabolic health and an appreciation for how environmental factors may jointly shape one’s likelihood of developing obesity and diabetes. As a result, there has been a tremendous shift toward uptake of this evidence into policy, which provides researchers with an opportunity to study natural experiments in this area. What is still unknown is the level of intervention needed to optimally shift behaviors in the population to meaningfully impact our diabetes prevention efforts. Furthermore, we need to understand what works, for whom, and in what context, since interventions that are effective in one population may be less so in another. Some communities have other, more critical barriers to lifestyle modification that need to be addressed to benefit from the physical environment in which they live. Built environment interventions may have a greater impact in communities that have the greatest need for lifestyle modification if they are part of a holistic approach to disease prevention that targets the structural barriers that perpetuate physical inactivity and unhealthy eating patterns. Doing so will require a commitment to invest in high-need communities and to engage them in discussions on what targets for intervention would be most meaningful to them. The COVID-19 pandemic has irreversibly changed our approach to population health and has demonstrated the extent to which experts from multiple sectors can work together to identify solutions to a problem that, for most, was previously unimaginable. As we reshift our attention back to chronic disease prevention, there are endless opportunities for intersectional collaboration to prevent obesity-related diseases such as diabetes. The built environment is now and will likely to remain an integral component of these efforts.
Financial Support
N.A.H. was supported by a Canadian Institutes of Health Research (CIHR) Vanier Canada Graduate Scholarship and a CIHR MD/PhD Studentship during the completion of this work. G.L.B. is supported by a Canada Research Chair in Policy Solutions for Diabetes Prevention and Management.
Disclosures
The authors do not have any conflicts of interest to disclose.
References
Howell NH, Tu JV, Moineddin R, et al. The probability and odds of diabetes and hypertension by levels of neighborhood walkability and traffic-related air pollution across 15 municipalities in Southern Ontario, Canada: a dataset derived from 2,496,458 community dwelling-adults. Data Brief. 2019;27:104439.
