This study explored multiple approaches to developing robust indicators of age-friendly neighbourhood environments with potential application to urban and rural communities in countries with a wide range of national-income levels. Using existing measures of community environments collected within a global epidemiological study, we successfully adapted these measures to define and describe a novel set of indicators aligned with the World Health Organization’s age-friendly cities framework . In the context of a longitudinal study for which recruitment began five years before the publication of the WHO framework, we were able to address six of the original eight WHO domains: 1) outdoor spaces and buildings; 2) transportation; 3) social participation; 4) civic participation and employment; 5) communication and information; and 6) community support and health services. However, we were unable to include indicators for two domains — 1) housing; and 2) respect and social inclusion — which are among those most commonly excluded from comparator audit tools as well, included solely in an attempt to apply the WHO criteria to informal settlements in Nairobi, Kenya .
The current effort builds upon the WHO’s pilot-testing process, which was carried out across locations representing a range of population densities, cultures, and demographic profiles in 2014–2015. That study reported an average of 24% fidelity between the standard indicators and available metrics across these sites, citing difficulties with data collection as the principal issue impeding the use of the indicators . Although the specific indicators available for use in PURE vary from those recommended by the WHO, they align with 75% of the broad domains included in the original WHO age-friendly cities framework. The geographic scope of the current project is also considerably larger, reflecting 496 communities located in 20 countries, as compared to the 15 communities across 11 countries included in the WHO’s pilot tests .
In addition, our findings regarding the impact of urbanness complement existing efforts to extend the WHO framework from urban regions to rural and remote areas, such as that carried out via a series of focus groups held in ten communities across Canada in 2007 . This project highlighted several features of greater importance to residents of rural and remote areas than urban ones, particularly driving safety, expanded public transportation, alternative channels for information provision, and the creation of a “one-stop shop” to provide healthcare and other support services in a single, accessible location . Although this phase of our study was not designed to assess the importance of individual indicators or broader domains to specific health outcomes, it demonstrates striking differences in access to supports for healthy ageing by level of urbanness, including in specific areas such as public-transit availability and access to government services highlighted by the Canadian effort .
In this way, our effort builds upon prior studies that have cited the importance of urbanness to healthy ageing without examining its relationship to specific indicators. For example, while developing the Neighbourhood Design Characteristics Checklist (NeDeCC) in England, Burton et al. reported that the urban–rural status of older adults’ residences had one of the strongest associations with well-being of all 25 included indicators, but the researchers were unable to examine variation in the other indicators by urbanness due to sample-size limitations . The creators of the Older People’s External Residential Assessment Tool (OPERAT) reported significantly higher scores in the domains of natural elements and incivilities and nuisance in the most-urban environments, but noted their small, non-random sample of 500 adults and their focus on Wales alone as important limitations . An effort to adapt the WHO framework to both urban and rural communities across China by integrating data from the China Health and Retirement Longitudinal Study (CHARLS) was explicitly designed to examine variation by urbanness and found that all indicators of age-friendliness were more common in urban areas across the six included domains (see Supplemental Fig. 1), similar to our own outcome . However, their extensive adaptation of individual indicators means these results are not directly comparable to ours. Looking beyond the healthy-ageing literature, our findings regarding urbanness and public-transit availability align with those reported in a study that applied an adapted version of the EPOCH 1 tool to assess community-level features associated with CVD risk factors in 2,074 urban and rural communities across Canada, which similarly found significantly lower availability of buses and trains in rural settings than in urban ones .
Research examining the relationship between country-level income and environmental indicators related to healthy ageing is rare. In fact, none of the comparator tools identified in our scoping review integrated data from more than one low- or middle-income country [14, 15, 23], except the efforts led by the WHO itself [12, 13]. However, neither of the two WHO efforts were designed to assess variation in the availability of indicators by country-level income, preventing any direct comparison. We identified substantial variations in the individual indicators in domains 1 and 2 by country-income class, suggesting that healthy-ageing indicators need to be adapted to specific resource levels and contextual settings. For example, using a single distance to train and bus stations to define public-transit availability in both high- and low-income countries overlooks the fact that residents of low-income countries are less likely to have access to a vehicle to travel to such a station , reducing the maximum distance that reflects practical accessibility.
All in all, the results of both the FAMD and MTMM analyses and the availability of individual indicators across the diverse set of communities included in this study support calls in the existing literature to abandon uniformity in favour of complexity. In fact, the WHO’s guide to using the core indicators of the AFC framework states the guidelines are “something to be adapted, as necessary and appropriate, to build an indicator set that is most meaningful and relevant in the local context” , and a number of the studies identified in our narrative review described such adaptation to lower-income countries  and rural communities [14, 16]. Parallel efforts have generated indicators using local data rather than the WHO AFC framework, including the Multidimensional Assessment System of the Built Environment (MASBE), which was refined using case studies in Mexico and Spain , and the Age-friendly Urban Index (AFUI) in Ireland, which used confirmatory factor analysis to identify three domains and calculate a single score . In addition, numerous projects have narrowed in on specific aspects of age-friendliness, such as accessibility and protection from harmful exposures . These include the Mobility Over Varied Environments Scale (MOVES) tool, developed using data from a population-based survey of older Canadians , and the Senior Walking Environmental Assessment Tool (SWEAT), which measures features related to physical activity . However, because the WHO framework is so widely applied — the WHO Global Network for Age-friendly Cities and Communities comprised 1,114 sites home to more than 262 million individuals in late 2021  — adapting and applying the full set of the WHO’s AFC indicators across the broadest possible range of settings remains critically important .
Looking forward, this project provides the foundation for applying these indicators to multiple domains of healthy ageing among adults aged 50 and older within PURE’s unique study cohort. Critically, PURE’s longitudinal design will advance the exploration of complex causal pathways that link exposures recorded between 2010 and 2015 to outcome data captured in follow-up surveys completed through 2021. Three major epidemiologic studies are planned, each building on the prior effort. The first will examine social isolation based on a scale previously developed for PURE analyses and comprising marital status, social support, and group membership ; the second will look at three distinct measures of mental health (stress, depression, and suicide); and the third and final study will evaluate incident CVD and CVD mortality. In addition, future funding will be sought to repeat the EPOCH assessments using the same tools, expanded to include all of the domains recommended by the WHO. This process will allow us to document changes in community-level healthy-ageing indicators over time and then relate these shifts to changes in risk factors and measures of social, psychological, and physical functioning.
Strengths and limitations
The timespan of EPOCH data collection represents one of this study’s major limitations: because we collected environmental data for each community over a relatively short timeframe, the results presented here may not reflect the current age-friendliness of PURE study communities. However, a new round of EPOCH data collection is planned to update each community’s rankings and to examine changes over time. Conversely, the fact exposure data pre-date outcome data is a key strength of the planned epidemiologic studies. The planned data-collection effort could also help overcome another limitation of the current study: the fact that the tools were not developed specifically with age-friendliness in mind. This precluded our ability to assess whether public spaces, buildings, or public-transit vehicles are accessible to older adults with mobility, vision, or hearing limitations, a factor that is highlighted throughout the WHO criteria .
The EPOCH data collection that informs the current analysis was also limited in its geographic scope, with the bulk of the variables based on a systematic social observation conducted throughout a one-kilometre walk. Because the precise latitude and longitude of the centre of each PURE community has been recorded, however, one potential method of overcoming these limitations is the integration of similar indicators from satellite or other georeferenced data, which have become significantly more widely available over the past decade for the areas under study.
In addition, although participant observations inform several indicators, not all study participants were older adults. Earlier efforts that have integrated subjective assessments among members of this age group have demonstrated the utility of such an approach, particularly to rank  or weight [22, 34] objective indicators or for the assessment of indicators for which objective data may be lacking, such as accessibility to buildings by wheelchair users . However, a number of similar environmental audit tools failed to include any input from members of this age group [14, 21], making our approach an advancement over these others.
Perhaps the biggest limitation of this study is the fact that our indicator selection was confined to a single set of existing definitions applied universally across PURE’s diverse communities. Our analyses demonstrate substantial variation by both community-level urbanness and country-level income, indicating that the ideal construction of healthy-ageing indicator variables should take these moderating factors into account. For example, although EPOCH 1 defines access to a range of resources (such as government buildings and train stations) via a 20-km distance from the community centre, this single linear distance may equate to widely varying travel times in urban vs. rural locations; older adults’ sense of perceived accessibility is also likely to differ based on geographic and cultural factors .
Finally, although we were able to address the bulk of the broad domains identified by the WHO’s healthy-ageing indicators framework, the precise indicator definitions differed significantly. This last aspect of our study design will limit our ability to speak specifically to the relationship between the WHO’s age-friendly criteria and the health outcomes captured in PURE, but this concern is offset by the broad geographic scope, large sample size, and diversity of the PURE cohort.