Longitudinal evidence, beginning with the Multi-Ethnic Study of Atherosclerosis (MESA) analysis of Hirsch and colleagues (2014), links favorable changes in urban form — more nearby destinations, retail zoning, street connectivity, and transit access — to increases in adults' walking for transportation. What remains unresolved is the causal dose–response question those authors posed explicitly: how much of which specific environmental change produces how much additional transport walking, in whom, and net of residential self-selection. Later work has consolidated the direction of the association, supplied quasi-causal evidence for composite walkability via a genetically informed twin design, and estimated associational distance-decay thresholds by destination type, but no study has delivered causally identified threshold or dose–response estimates for changes in specific built-environment features in adults, nor settled demographic heterogeneity. The problem is to produce and defend such estimates, including their functional form and thresholds, their scale specification, their variation by demographic characteristics, and their self-selection robustness.

Background

Physical inactivity is a leading modifiable risk factor for cardiovascular disease and other chronic conditions, and walking is the most common form of physical activity among adults. Epidemiologists therefore study how the built environment — the human-made surroundings such as land-use patterns, street networks, and transit systems in which people live — shapes walking behavior. A standard distinction separates transport walking (walking as a means of getting to destinations, e.g., shops or transit stops) from leisure or recreational walking (walking for exercise or pleasure), because the two behaviors respond to different environmental features. Walkability is an umbrella term for the degree to which an environment supports walking, typically quantified with geographic information systems (GIS) as measures such as population density, the share of land zoned for retail, the density of walkable destinations, street connectivity (how directly the street grid links origins to destinations), and distance to transit. These features are usually summarized within a buffer — a circle or street-network polygon drawn around a residence at some chosen radius, a choice that itself affects the measured exposure. In 2014, Hirsch, Moore, Clarke, Rodriguez, Evenson, Brines, Zagorski, and Diez Roux published a longitudinal analysis in the Multi-Ethnic Study of Atherosclerosis (MESA), a diverse cohort of older adults in six US cities followed from 2000 onward with time-varying, GIS-derived built-environment measures around participants' homes. They found that both baseline levels of and increases in destination density, retail zoning, and street connectivity were associated with greater increases in transport walking over time, while built-environment change was not associated with leisure walking. Because the observational design cannot exclude confounding and because people who prefer to walk may choose walkable neighborhoods — the problem of residential self-selection — the authors could not attach causal effect sizes to these associations. They closed by stating that questions remain about how much or what types of changes are necessary to increase physical activity, urging future work to identify potential thresholds or physical forms of these features that optimize health behavior change, to resolve the geographic scale (buffer size and shape, home versus whole activity-space exposure) at which each feature matters, and to examine effect modification by demographic characteristics. Subsequent systematic reviews of longitudinal studies and natural experiments (research designs exploiting real-world environmental changes, such as new infrastructure, analyzed with before-and-after or difference-in-differences comparisons) have since consolidated the overall direction of the environment–walking association, and quasi-experimental designs including fixed-effects analyses (which compare each person with themselves over time, thereby controlling for all time-invariant personal characteristics) and genetically informed twin analyses (which compare identical and fraternal twins to control for genetic and shared-family confounding) have begun to address self-selection, but the quantitative core of the 2014 question — the causal dose–response and threshold structure of specific environmental changes for transport walking in adults — remains open.

Problem Statement

Derived from the future-work question stated by Hirsch et al. 2014 (doi:10.1093/aje/kwu218): identify, causally, the dose-response relationship between specific changes in the built environment and subsequent changes in transport walking among adults. Concretely, all four of the following components are required: using changes (not merely baseline levels) in objectively measured features — density of walkable destinations by type, share of land zoned for retail, street connectivity, and access to transit — estimate, with an identification strategy that addresses residential self-selection and time-invariant individual confounding, how much additional transport walking (minutes per week or trips per week, and distinguishing transport from leisure walking) is produced by a given increment of each environmental change; characterize the functional form of this relationship, including whether threshold or saturating behavior exists (that is, how much change is needed before walking increases, and where further change yields no further increase); specify the geographic scale (buffer size and shape, or activity-space exposure) at which each feature's effect operates; and quantify how the dose-response varies by demographic characteristics, at minimum age, sex, race/ethnicity, and socioeconomic position. The causal interpretation and generalizability of the estimates depend on the population and urban context studied and on the exposure and outcome measurement protocols, relative to the source question of how much and what types of built-environment change are necessary to increase transport walking.

References

  1. Hirsch JA, Moore KA, Clarke PJ, Rodriguez DA, Evenson KR, Brines SJ, Zagorski MA, Diez Roux AV. Changes in the Built Environment and Changes in the Amount of Walking Over Time: Longitudinal Results From the Multi-Ethnic Study of Atherosclerosis. American Journal of Epidemiology 2014;180(8):799-809. doi:10.1093/aje/kwu218. https://doi.org/10.1093/aje/kwu218
  2. Li J, Auchincloss AH, Hirsch JA, Melly SJ, Moore KA, Peterson A, Sánchez BN. Exploring the spatial scale effects of built environments on transport walking: Multi-Ethnic Study of Atherosclerosis. Health & Place 2022;73:102722. doi:10.1016/j.healthplace.2021.102722. https://doi.org/10.1016/j.healthplace.2021.102722
  3. 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. Annals of Behavioral Medicine 2018;52(3):239-251. doi:10.1093/abm/kax043. https://doi.org/10.1093/abm/kax043
  4. Bandara TN, Higgs C, Turrell G, De Livera A, Gunn L, Zapata-Diomedi B. Longitudinal effects of the built environment on transportation and recreational walking and differences by age and sex: A systematic review. Social Science & Medicine 2025;368:117811. doi:10.1016/j.socscimed.2025.117811. https://doi.org/10.1016/j.socscimed.2025.117811
  5. Duncan GE, Avery AR, Pilgrim MJD, Amram O, Mooney SJ, Rundle AG. Longitudinal Association Between Walkability and Physical Activity in Twins. American Journal of Preventive Medicine 2025;69(3):107950. doi:10.1016/j.amepre.2025.107950. https://doi.org/10.1016/j.amepre.2025.107950
  6. Timmermans EJ, Visser M, Wagtendonk AJ, Noordzij JM, Lakerveld J. Associations of changes in neighbourhood walkability with changes in walking activity in older adults: a fixed effects analysis. BMC Public Health 2021;21(1):1323. doi:10.1186/s12889-021-11368-6. https://doi.org/10.1186/s12889-021-11368-6
  7. Hogendorf M, Oude Groeniger J, Noordzij JM, Beenackers MA, van Lenthe FJ. Longitudinal effects of urban green space on walking and cycling: A fixed effects analysis. Health & Place 2020;61:102264. doi:10.1016/j.healthplace.2019.102264. https://doi.org/10.1016/j.healthplace.2019.102264
  8. He D, Lu Y, Xie B, Helbich M. Large-scale greenway intervention promotes walking behaviors: A natural experiment in China. Transportation Research Part D: Transport and Environment 2021;101:103095. doi:10.1016/j.trd.2021.103095. https://doi.org/10.1016/j.trd.2021.103095
  9. Marquet O, Hipp JA, Miralles-Guasch C. Neighborhood walkability and active ageing: A difference in differences assessment of active transportation over ten years. Journal of Transport & Health 2017;7:190-201. doi:10.1016/j.jth.2017.09.006. https://doi.org/10.1016/j.jth.2017.09.006
  10. Lu Y, Sun G, Gou Z, Liu Y, Zhang X. A dose-response effect between built environment characteristics and transport walking for youths. Journal of Transport & Health 2019;14:100616. doi:10.1016/j.jth.2019.100616. https://doi.org/10.1016/j.jth.2019.100616
  11. Gunn LD, King TL, Mavoa S, Lamb KE, Giles-Corti B, Kavanagh A. Identifying destination distances that support walking trips in local neighborhoods. Journal of Transport & Health 2017;5:133-141. doi:10.1016/j.jth.2016.08.009. https://doi.org/10.1016/j.jth.2016.08.009