Older adults’ outdoor walking and the built environment: does income matter?

We extend the current literature by evaluating the association between neighbourhood walkability and physical activity in a population-based sample of older British Columbians. Specifically, our findings build on previous research linking walkability or access to destinations to outdoor walking, specifically in older adults [13‐15]. To our knowledge, ours is the first large study to use the Walk Score metric with a focus on the older adult populations. Given the devastating consequences of physical inactivity for older adults (loss of muscle and bone mass, increased obesity, increased prevalence of cancers, diabetes and heart disease [14] and a shortened lifespan [42]), it seems imperative to ensure that neighbourhoods are designed in ways that promote physical activity in this population.

Street Smart Walk Score was positively associated with the odds of achieving sufficient levels of physical activity specifically through outdoor walking. Three in five older adults in our sample reported walking outside for at least 150 min in the previous week, but respondents had over three times higher odds of achieving this if they lived in a ‘Walker’s paradise’ compared with a ‘Car dependent/very car dependent’ neighbourhood. Results did not vary significantly across levels of household income. One previous study of older adults (≥70 years) in Minnesota found no association between Walk Score and physical activity, potentially due to the small sample (n = 53) or limited environmental variability [43]. Four studies in adult populations (all ages) found positive associations between Walk Score and walking of similar magnitudes [29‐32]. Importantly, the Street Smart Walk Score is publicly available metric with consistent methodology across settings. As a result, our findings can be readily compared with those in other locations, a challenge when walkabilty metrics are study-specific.

Many studies have considered the purpose of walking – whether it is for transportation or recreation Typically, land use mix (destination diversity) and street pattern are more strongly associated with transportation-related walking [25], whereas proximity to parks has been independently associated with recreational walking [44, 45]. We were unable to assess this, as the CCHS-HA questions do not capture purpose. In older populations the distinction between transportation and recreation walking may be blurred, if daily walks are also errand trips. Local evidence shows older adults made fewer than 10 % of trips purely for exercise (travel diary data) [17], suggesting the majority of walking trips do have a utlitarian component. Street Smart Walk Score captures proximity to 13 types of destinations, one of which is proximity to parks. As such, the metric may have more relevance to recreational-related walking than have traditional land use mix and walkability indices. Irrespective of whether outdoor walking by CCHS-HA respondents in our study was for transportation or recreation, those living in more walkable neighbourhoods had far higher odds of partaking in sufficient outdoor walking to achieve health benefit.

Our results illustrate the potential for the built environment to support older adults to walk outside. This study provides further evidence that macroscale design features - connected neighbourhoods with commercial and recreational facilities – may better enable older adults to achieve health-enhancing physical activity levels through neighbourhood walking. Street Smart Walk Score does not explicitly capture detailed design level features, for example presence and condition of sidewalks and benches are not measured. We acknowledge the relevance of microscale built environment [13] and the concept of person-environment fit [11, 46] especially for older adults. Complementary studies have used street audits [16, 47, 48] and photovoice [49] to identify specific design features that help or hinder older adults in traversing their neighbourhoods. Taken together, studies of macroscale and microscale built environment features provide guidance on community design that supports active healthy aging at varying scales.

Our interest was in whether income levels moderated the effect of the neighbourhood environment on walking. Previous work suggests that the built environment has more influence on those with higher incomes, for example, Manaugh and colleagues [29] reported that Walk Score was more strongly associated with walking in wealthy households than lower income households. This may be due to restricted travel options faced by those with lower incomes, or less flexible time for recreational activities. In our study, household income was a not a significant moderator of the association between Walk Score and outdoor walking. This may be an important finding as the built environment has the potential to reduce health inequalities [50], if it serves to promote positive health behaviours, in not only high income by also low income populations. We used household rather than individual income, on the premise that household income is likely a better reflection of the resources available. Since 70 % of CCHS-HA respondents live alone, household income equates to individual income in most cases. A more pressing question may be whether income reflects the overall wealth. Statistics Canada typically asks about income “from all sources” [51], which may not capture wealth related to home ownership or retirement savings. Both of these are of particular relevance for older adults. Neighbourhood income levels may also play a role. For example, King and colleagues [19] recruited older adults who lived in four neighbourhoods types in Seattle and Baltimore: high and low walkable, and high and low income. Participants in high walkability, high income neighbourhoods were most likely to meet physical activity guidelines. We were unable to evaluate such interactions using the CCHS-HA.

The use of administrative data introduces some limitations. First, the CCHS-HA survey uses self-reported measures, subject to social desirability and recall error that may result in overestimates of physical activity. This would not likely vary across built environment settings, directing any bias toward the null. Recently objective measures of physical activity have been incorporated into a Canadian national survey [52] with a smaller sample size. As with any survey question related to income, we had substantial missing data for income (22.4 % of respondents, more likely to be female). We excluded these respondents from the effect modification analysis, which may have introduced bias if those who do not answer the income question are not evenly distributed across income levels. Studies have found that those who refuse are likely to be from higher occupational positions, whereas those who respond ‘don’t know’ are likely to be from lower occupational status [53, 54]. In our data, we found no significant differences for education and retirement status (other indicators of socio-economic status) between those with and without income data, and likewise no differences in Walk Score. Second, perceptions of the built environment were not captured in the CCHS-HA. The survey did not include questions that allow adjustment for self-selection (that people who like to walk may choose to live in more walkable neighbourhoods), although emerging evidence supports a role for the built environment independent of people’s preferences for walking [55‐57]. Third, the CCHS-HA had a question on pet ownership, but not specifically dog ownership; the literature suggests that it is dog ownership that is associated with walking [58, 59]. Fourth, survey data were collected in 2008–2009 and the Walk Score data were from 2013. This is of little concern given the slow nature of changes to the built environment. Finally, we conducted analysis of cross-sectional data, thus results should not be taken as causal, but rather they provide the backdrop to design and implement longitudinal studies [20].

Our study also had a number of strengths. We used a large, national survey of older adults (CCHS-HA) that provided a representative sample across a broad geographical area. Further, we used Street Smart Walk Score, a comparable, objective measure of walkability to allow our results to be placed into context. We were able to explore potentially moderating effects of income given the large sample size.