Generating GPS activity spaces that shed light upon the mobility habits of older adults: a descriptive analysis

Mobility is defined as the “ability to move oneself (either independently or by using assistive devices or transportation) within environments that expand from one’s home to one’s neighborhood and regions beyond” [1]. Mobility is key to older adults leading active, healthy, independent lives [2] and is central to them conducting commercial, cultural, and social activities [3–5]. Older adults’ well-being and quality of life are also closely linked to their mobility [5–8]. The demographic shift toward an aging population is unprecedented in western society and demands novel solutions that evaluate and promote the mobility of older adults. These solutions may be embedded within transportation systems and planning [9–13].

It is essential to identify effective tools to describe older adult mobility so as to better understand the influence of neighborhood on their health and mobility [14]. “Life-space” is a frequently used measure of older adult mobility [15–17]. This self-reported measure of the extent of recent travel (using thresholds such as: within the home, into the local neighborhood, or beyond) was positively associated with diminished cognitive decline and Alzheimer’s Disease [18, 19], lower risk of both death [20] and becoming more frail [21]. Older age, being female, and having physical limitations have been associated with smaller life-spaces, as is having had a stroke, high depressive symptoms, and being obese [22, 23]. Higher education, better lower extremity function and muscle strength were associated larger life-spaces [22, 23]. Importantly, the ability to drive plays a key role in the mobility of older adults as captured using life-space measures [23–26].

Global Positioning Systems (GPS) technology may provide a means to calculate geographic range as a measure of mobility. GPS has been used to objectively characterize life-space [27–32] and also to detect outside physical activity [33]. Most research using GPS for older adult mobility comes from one study—the Senior Tracking (SenTra) project, based in Germany and Israel. SenTra used GPS points or metrics of out-of-home behaviors (e.g. number of visited places, time spent outside the home, and distance traveled from home) to assess mobility patterns in a sample of older adults who were cognitively impaired or had Alzheimer’s disease, and compared this clinical group with community-dwelling older adults [34–47]. It may not be possible to generalize outcomes from SenTra across geographically and culturally diverse settings, or with older adults who are independent and able to make their own travel decisions. Therefore, there is a need to investigate GPS applications in community-dwelling older adults in the North American context, given the limited knowledge base on measurement approaches to define the geographic patterns (i.e. shape) and extent (i.e. size) of older adult mobility in this population. Additionally, the potential to characterize older adults’ geographic extent or pattern was not explored in any of these studies, leaving questions about where older adults travel.

To represent daily mobility, neighborhood studies of physical activity have used GPS-based “activity spaces” as an individual-based measure of spatial behavior [48–52]. Activity spaces differ from the life-space measure, in that they focus on neighborhood (out of home) behavior only, rather than mobility both within and beyond the home. Additionally, with the exception of recent efforts to incorporate GPS [27–32], the majority of life-space studies are based on self-reported travel extents (within the home, into the local neighborhood, or beyond), not spatially-located travel data. Thus, the expansion of activity spaces to the investigation of older adult mobility will give additional insight into the community factors and resources that shape neighborhood activity. It is hypothesized that activity spaces may vary in size and shape across different populations, such as those with low incomes or different age groups [50, 53]. To date the utility of GPS to measure, describe, and compare mobility patterns in older adults has not been fully explored. Previous work proposed using a “mobility envelope”, the length of the outer perimeter of spatial excursions made by individuals, as an outcome measure for mobility studies in older adults [54]. By integrating various metrics of geographic extent from activity space studies, the mobility envelope concept could be advanced. For example, additional metrics evaluating different dimensions of individuals’ geographic scope may be useful for understanding mobility. In particular, in urban planning the shape or “compactness” of activity spaces is a metric of how circular a polygon is and is a concept thought to illustrate the capacity of neighborhoods to provide opportunities to “live, work, shop, and socialize at the local scale” [53]. Specifically, compactness has been shown to vary across different travel modes [55]. However, comparisons of compactness among different individuals (and within a particular type of activity space) may reveal important information on the role of driving in older adult mobility.

Therefore our objectives are twofold: to create and compare different types of geographic activity spaces for community dwelling older adults, so as to clarify the extent and pattern of their mobility; and then within each activity space approach, to assess individual sociodemographic and resource characteristics that are associated with larger (or smaller) or more (or less) compact activity spaces. We hypothesize that the size and shape of activity spaces will vary by the approach used to create them. Within each activity space type we hypothesize that participants who are younger, healthier, and with better access to transportation (driving or material resources for going outside) will have larger activity spaces. Furthermore, we hypothesize that those who drive less will have more compact activity spaces, and also that participants who live in more walkable neighborhoods will have more compact activity spaces, since amenities may be closer to home.

This research is one of the first studies to utilize GPS tracking to create activity spaces as a means to assess older adult mobility. We extend the literature by demonstrating that GPS tracking can be used to create three different types of activity spaces as a valuable tools to better understand the geographic mobility patterns of older adults. Not surprisingly, older adults deemed most mobile based on their age, the walkability of their neighborhoods, and whether or not they drove, had the largest activity spaces. The trends in activity space areas were not different, regardless of what approach was used to generate them. However, shape of activity spaces (measured as compactness) varied by approach.

Walking and cycling trips are often extend beyond traditional buffer sizes (1 mile, ½ mile) used to represent neighborhoods where older adults live [56]. We demonstrated that GPS may enable a more precise way to operationalize neighborhoods than residential buffers or administrative units [51, 57–62] as it better captures the locations individuals actually visit rather than a presumed neighborhood boundary. GPS has recently become a more popular option for measuring neighborhood exposure and context in health studies [33, 50, 63–67]. As the analytic approaches to GPS data develop, this technology may become a powerful tool to accurately and precisely describe people’s interactions with geographic space, including older adult mobility. It is possible that different types of activity spaces may be more appropriate depending on the application or the research question being addressed. Since the MCP-generated activity space is bounded by the outermost GPS points, it captures the envelope of the extreme extent of travel and thus may include large geographic areas that are not visited by, or important to, an individual [68]. In contrast, SDE-generated spaces may be more useful if one wishes to assess the direction and general shape of a person’s travel area, without introducing potential error introduced by using geographically distant points. In addition, SDE approaches may indicate the frequency that an individual visits a geographic area as more points are generated in that location - this ‘pulls’ the SDE-generated activity space toward that more often visited geographic area. However, since SDE-generated activity spaces are by definition ellipse shaped, they may also capture a substantial area that an individual may not have visited. Finally, we recommend using DPA to generate an activity space if a research question is about the destinations participants pass in daily travel, as it relies solely on buffering the routes actually traveled.

The consistency of activity space area across the three different approaches aligns with a previous study that showed associations across MCP, SDE, and line-based buffers [52]. However, no previous studies assessed the shape (compactness) of GPS-derived activity spaces. In our study, patterns of compactness differed depending on the approach used to generate activity space. This signals the need to choose an approach to generating an activity space that is relevant to the research question being asked. Of interest, compactness generated using the SDE and MCP approach was higher when walkability was lower, whereas for DPA-generated compactness was higher in neighborhoods with higher Walk Score. This finding may illustrate that participants in neighborhoods with lower walkability are clustering their trips to a nearby retail area. Thus, in neighborhoods with lower walkability the area of all three activity space types becomes larger, and the compactness of the SDE and MCP is higher. However, in this same scenario, DPA-generated activity spaces become more elongated, potentially with multiple trips heading in the same direction, which results in a less compact DPA-generated activity space. Our findings using DPA are consistent with work indicating that small and compact activity spaces increase the likelihood of walking and cycling [55].

A number of factors surfaced as important to older adults’ mobility. First, it is intuitive and supported by previous studies that an older adult’s ability to drive will influence the size of their activity space [23–26], and that the activity space of older adult will be smaller compared with a younger person, on average [22]. Sex did not surface as a differentiating factor in our study, in contrast with previous reports that women had smaller activity spaces than did men [22, 23]. However, as only one-third of our sample were men it is possible we lacked the statistical power to test this association. Second, when an older person is no longer able to drive, physical support to maintain mobility within the neighborhood becomes increasingly important. Given its apparent role in our study around encouraging older adults to travel within their neighborhoods, physical support may represent an effective intervention in future studies. Third, our results support previous findings – that neighborhood attributes (i.e. higher street connectivity, proximity to destinations, and traffic conditions, and parks) are associated with increased mobility among older adults [65, 69–71]. Although seemingly counterintuitive at first glance, our finding that those living in higher walkability areas had smaller activity spaces could reflect closer proximity and access to amenities, and the use of different modes of transport. That is, highly walkable neighborhoods are more likely to have destinations that older people deem important, within easier walking access. As walking trips are often shorter than driving trips, this would reduce distance traveled to these destinations. However, this study cannot disentangle the competing and related concepts of walkability and car usage. On one hand, car usage was associated with larger activity spaces, potentially indicating greater mobility, yet walkability was associated with smaller spaces, potentially indicating that the local neighborhood environment is sufficient to fulfill daily activities and amenities. Future work creating activity spaces by mode or considering the distance traveled or frequency of trips within an activity space may help tease apart these complex and complementary elements of older adult mobility. All of these results are novel and in our view, are worth pursuing in future trials that evaluate different groups of older adults (we recruited older adults with low incomes) who reside across diverse built environment settings.

The life-space literature indicates that greater mobility is related to a wide range of favorable health outcomes [18–21]. This study indicates that larger activity spaces are associated with a number of different resources such as younger age, access to a vehicle, or physical support for going outside. One possibility is that the geographic mobility of individuals, as measured either in life-space or activity space, is in fact a proxy for personal resources. Additional work, examining changes in activity spaces as older adults transition through life changes (e.g. retirement, loss of a spouse, move to a more walkable neighborhood, driving cessation) may help to tease apart the complex connections between older adult resources, geographic mobility, and health outcomes.

There are many different ways to represent geographic activity spaces where individuals travel to and spend their time. However, outcomes and interpretations may vary based on the approach used to generate an activity space. It is important to use an approach tailored to the needs of a specific research question and outcomes. Some factors we identified as important to geographic mobility of older adults may be used to inform interventions and to design policies that support older adults living and engaging independently with their community. Specifically, this work highlights the role of neighborhood walkability, driving patterns, and physical support to go outside as important factors in determining the size of older adult activity spaces. Identifying an approach that best captures the activity space of older adults may be useful for future work aimed at isolating features of the neighborhood environment that support older people ‘aging in place’ or informing interventions and policies that support older adults living independently in the community.