Destinations, functional features, and aesthetics of residential areas were associated with more transportational walking among community-dwelling older persons. An increase of one functional feature per street within a 400 meter buffer surrounding one’s residence, an increase of one destination within 400–800 meter buffers, and an increase in aesthetics within 800–1200 meter buffers were associated with increases in transportational walking up to 2.89 minutes per two weeks.
Higher scores on aesthetics were associated with more time spent on transportational walking, which is in contrast to previous studies [15, 16]. This discrepancy may be due to the fact that within these studies aesthetics were measured differently, i.e. by less items or via self-reports. There is inconsistent evidence for associations between the area characteristics functional features and safety and walking [6, 17]. The inconsistent findings concerning the association between safety and transportational walking among older persons has been attributed to the complexity of measuring safety . Our measure included both traffic- and social safety indicators, and additional analyses showed that both sets of indicators were not associated with transportational walking in any of the buffers. The association between the presence of destinations and transportational walking was found for buffers up to 800 meters, but was absent in the 1200 and 1600 meter buffers. This finding is in line with studies reporting associations in buffers up to 1000 meters [17, 18].
Whereas other studies often use a predefined buffer size , our results revealed that associations between area characteristics and walking behaviour differed by buffer size. Nagel et al. found that associations between environmental factors and total walking time among older persons aged 65 years and older were similar across buffer sizes (400 and 800 meters) . We extended this finding, as we also included buffer sizes larger than 800 meters for which also significant associations were found.
A decrease in variation with increasing buffer size was found. This has most likely not biased our results, since we observed associations for aesthetics within 800 and 1200 meters (with lower variation than within 400 meters). It is expected that the municipality of Spijkenisse focussed most on the maintenance and improvement of the (close by) areas where most residents live. This would result in lower scores for area characteristics further away from the residences, and lower average scores when larger buffers are considered. It is unclear how it would affect our results when the mean scores on the area characteristics had remained stable or even increased with increasing buffer sizes, since it depends on the walking behaviour of the elderly and whether they would be able and willing to walk further distances.
A possible explanation for the finding that destinations and functional features were particularly important for transportational walking in small buffer sizes may be that older persons are generally less functionally fit than their younger peers. Thus, their functional capacities may limit their activity patterns to use destinations and functional features (e.g. sidewalks, benches) in the close vicinity of their residence. Area aesthetics were particularly important for larger buffer sizes. Elderly may only go for an extended walk when the environment is pleasant (aesthetically appealing) to walk through. Whereas other studies found that a buffer of 1600 meters was important for associations between the built environment and PA among elderly [19, 20], our study did not corroborate this observation. This may be due to a distance of 1600 meters being too far for older persons to walk regardless of the area characteristics or because there was too little variation within this buffer. The larger the area in which the environment is measured, the more likely that environments of individuals will become similar which may reduce the chance of finding associations with PA levels.
Frail persons lived closer to facilities and had more functional features in their residential area as compared to non-frail persons. This could be the result of a selection process, whereby frail persons decide to move closer to facilities. However, in additional analyses, no differences between frail and non-frail persons were found in prevalence of and reasons for moving to their current residence in the past 5 years. The average total time per walk for frail persons was lower as compared to non-frail persons, which may suggest that frail elderly were more bound to smaller areas around their residences as compared to non-frail elderly. Knowledge of the exact amount of PA that was practiced within specific buffers for both frail and non-frail elderly would allow for a more accurate estimation of associations between area characteristics and walking behaviour in each buffer. It is therefore suggested to take this into account in future research, e.g. by combining GPS and accelerometer measurements .
Recently, differences were found in walking distances between disabled and non-disabled elderly . Also, stronger associations were found between area characteristics and PA levels for disabled than non-disabled elderly . As frail persons are at increased risk to develop disabilities [11, 12], the role of environmental characteristics for PA may become more important with increasing health complaints.
A strength of this study concerns the personal geographical space units, i.e. the walking path based buffers around participants’ residences, instead of the often used, pre-defined geographical units, for instance based on zipcodes or neighbourhood boundaries. A personal geographical space unit provides more specific information on environmental characteristics to which persons are exposed as compared to a geographical unit. Furthermore, detailed qualitative and quantitative information about the residential areas of the elderly was collected by street audits. A limitation of this study was that area characteristics were collected up to 13 months after the first interviews took place. Thus, there is a possibility that the area characteristics may have changed meanwhile. To the extent that area characteristics determine walking, such changes in the environment may have resulted in an underestimation of the associations reported. The ISAR questionnaire was used to measure frailty, which overlaps in terms of measuring functional limitations and predicting the risk of adverse outcomes. Other studies used the Tilburg Frailty Index (TFI) which includes a broader set of indicators of frailty. It remains unknown however, of the TFI would have altered these associations .
As this study was conducted in a (middle-sized) city in The Netherlands, and the design of cities may differ across countries, it is unclear how these results also would apply for cities in other countries.
Our study has several implications. Firstly, for the appropriate linkage of environmental characteristics to walking (and other health behaviours), specific buffer sizes need to be used. It requires insight into the expected level of variation in the area, and it is important to realize that such variation may differ in different countries. We recommend to explore the variation of an characteristic prior to the analyses. Ultimately, such an approach may results in more consistent findings.
Secondly, living longer independently can be facilitated by a residential area that facilitates and inspires elderly to walk for daily activities. Neighbourhood improvements may increase levels of transportational walking among community-dwelling elderly. More research is needed to get more insight in the role of area characteristics for frail elderly.