1. Introduction
The U.S. Department of Health and Human Services Physical Activity (PA) Guidelines for Americans recommend that children and adolescents participate in at least 60 min of moderate to vigorous PA (MVPA) daily [
1]. Time spent in MVPA is inversely associated with cardiometabolic risk [
2], overweight and obesity [
3], and depression [
4]. U.S. children and adolescents are not meeting PA recommendations in general [
5,
6,
7], with African American adolescent girls at the highest risk for inactivity [
8].
The built and natural environment (all aspects of surroundings, man-made and natural) may influence risk for obesity by creating a setting that supports or hinders activity [
9,
10]. The association between specific environmental characteristics and physical activity in childhood has been extensively studied [
11,
12,
13,
14]. For example, an increased number of PA facilities has been shown to be associated with more time spent in MVPA [
15]. Environments shown to support PA among children typically enable recreational activities, including parks, recreation fields, and green space (open land covered in grass or vegetation that could potentially be used for PA) [
16,
17,
18,
19]. The availability of green space, recreational space (e.g., recreational fields, basketball courts, etc., including those located on school grounds), parks, and facilities that support PA has been linked to increased PA [
20,
21,
22,
23,
24] among children and adolescents. Examining a combination of physical activity promoting aspects of the built and natural environment together, in a single construct, could clarify how multiple environment factors relate to individual-level health behaviors.
Geographic information systems (GISs) data and methods have been used to map aspects of the built and natural environment and to explore the relation between environment and PA using objective spatial measures [
22,
25,
26,
27,
28,
29]. Locations used for PA have been mapped to identify disparities in spatial accessibility. The limited PA locations in low-income communities put adolescents who live in low-income communities at increased risk for inactivity [
21,
30,
31]. The lack of safe spaces for outdoor PA, based on a physical environment inventory and caregiver and child perceptions further contribute to limited PA opportunities in low-income, urban neighborhoods [
32,
33]. GIS methods can be used to identify neighborhoods with characteristics that increase the risk for inactivity, enabling policy makers and stakeholders to identify areas to be targeted for additional PA locations.
The ecological model of active living conceptualizes how the neighborhood environment relates to individual behaviors, including participation in PA [
34]. The model posits that PA is influenced through four hierarchical domains (1—policy environment, 2—behavior settings, 3—perceived environment, and 4—intrapersonal). The absence of PA promoting opportunities in the higher-ranking domains (1–3), limits individuals’ ability to pursue PA in the intrapersonal domain (4). This study focuses on the second ranking domain (behavior settings), which includes access to the built environment (e.g., neighborhood, recreation, and schools) and natural environment (e.g., vegetated open space, parks, and green space) in low-income, urban communities, assessed in relation to objectively measured PA in predominantly African American adolescent girls.
Based on the ecological model of active living [
34], the purpose of this study is two-fold. Our first objective is methodological—we sought to generate a replicable neighborhood-level physical activity location availability score (PALAS) constructed from data variables associated with PA and apply this score to neighborhood statistical areas (NSAs, developed by the Baltimore City Planning Department based on 278 recognizable city neighborhoods) in Baltimore City [
35]. Second, we examined the relation between objectively measured PA among predominantly African American adolescent girls and the PALAS rating of their neighborhood environment (neighborhood PALAS) and the home neighborhood area (HNA, PALAS variables/subcomponents within 0.25 miles of home). We hypothesized that adolescents living in high PALAS neighborhoods (indication of greater availability of PA locations) would engage in more minutes per day of MVPA compared to adolescents living in low PALAS neighborhoods. We also hypothesized that adolescents with PA locations in their HNA, including government-owned PA locations, schools, recreation centers, or parks, would engage in more minutes MVPA/day than adolescents without nearby PA locations.
4. Discussion
To investigate the second ranking domain of the ecological model of active living, this study successfully generated a PALAS to map PA location availability in Baltimore City, MD by NSA and illustrated that neighborhoods varied in availability and variety of PA locations. The PALAS was then used to analyze if a relation existed between the PALAS and PA among predominately African American adolescent girls living in low-income, urban communities. Using individual variables and subcomponents of the PALAS, PA location availability around each girl’s home (HNA) was also examined in relation to PA. As a result, this study found that a higher neighborhood PALAS and the presence of a recreation center in the home neighborhood area were associated with more minutes per day in MVPA.
This study found that adolescents living in high PALAS neighborhoods engage in more minutes of MVPA per day compared to adolescents living in low PALAS neighborhoods, supporting our first hypothesis. This result supports the ecological model of active living, showing that the neighborhood environment is associated with individual behaviors. This finding also adds to existing literature showing a link between greater variety and availability of PA locations and PA among children and adolescents, specifically among African American adolescents living in low-income communities, an understudied group at risk for inactivity and obesity [
62]. Furthermore, this study’s focus on adolescent girls is important given the gender-based PA disparity that consistently shows girls as having lower activity levels and higher body fat percentages than boys [
63]. Our examination of relations between PA location access throughout the entire neighborhood (not just surrounding the home) and objectively measured PA is novel and has potential policy implications. The methods used to generate the PALAS map could be used by other jurisdictions to identify areas with limited access to PA locations. Demonstrating a link between PA and the PALAS among a population at increased risk for inactivity and obesity adds support for this method as a tool for stakeholders and policy makers. The information can be used for investment decisions for new PA locations or zoning regulations to support PA locations in neighborhoods with limited access.
This study demonstrated that adolescent girls living in low-income, urban communities with recreation centers in their HNA (within 0.25 miles of the home) engaged in more minutes of MVPA per day than adolescents who did not have recreation centers in their HNA, supporting our second hypothesis, the ecological model of active living, and the current literature [
64,
65]. We also hypothesized that all government-owned locations within the HNA would be associated with more PA, which was found in the bivariate analysis; however, in adjusted models the relation was no longer present. Additionally, the subcategory of government-owned school PA locations was significant in the bivariate analysis, but not in the adjusted model. Contrary to our hypothesis, parks showed no relation with PA in either bivariate analysis or adjusted models.
There are several possible reasons why we did not find strong support for relations between adolescent girls’ MVPA and access to government-owned and school PA locations and parks in this sample, yet did find a relation with access to recreation centers. First, recreation centers are more likely to provide child-specific programming, which was not measured in this study. Additionally, information was not gathered on the presence of joint-use agreements supporting community access to school grounds, which may be a barrier to accessing PA opportunities on school locations. Third, data on park quality, safety and programming were not gathered, which may moderate relations between access and MVPA [
66,
67]. A final potential reason for the lack of support for a priori hypotheses could be the population studied. We based our hypotheses on existing literature, which has had limited inclusion of African American adolescent girls from low-income, urban communities [
62,
68,
69]. Findings from other studies may not generalize to this population. For example, our group examined if positive perceptions of the neighborhood environment surrounding schools or the aesthetic features of these neighborhoods (assessed via a driving audit) were related to PA among predominantly African American adolescent girls in low-income, urban communities [
70]. Contrary to what is typically observed in suburban neighborhoods, we found that having an overall lower positive perception of the neighborhood environment surrounding the school and a greater density of displeasing neighborhood aesthetics (graffiti, broken windows, and abandoned homes) were each related to higher levels of activity, compared to the inverse. Taken together, findings from this study support access to government-owned PA facilities (recreation centers, schools, and other government-owned PA locations). Recreation centers, in particular, should be developed, supported, and maintained to promote PA among adolescents. Investment in PA locations is critical to increase PA, which is needed to maintain a healthy weight [
71]. Additionally, the dynamics between PA access and PA of populations living in low-income, urban neighborhoods should be explored further.
This study had several strengths and limitations. One strength of this study includes the use of objectively measured PA, which eliminates self-reporting errors and bias. Another strength is the attempt to include PA locations across multiple jurisdictions and categories and create a database of PA locations for analysis that represented a complete range of possible locations used for PA. Private and publicly available PA locations were included. Every attempt was made to include a location only if it was open and operational. A limitation of the study is the inclusion of only predominantly African American adolescent girls in low-income, urban communities, which limits generalizability. The age range of 10–14 years old may also be considered a limitation as children’s PA outdoors may depend on the availability of a caregiver to provide supervision. Future studies utilizing a PALAS should include older children with increased independence and autonomy to interact with their environment. Additionally, the cross-sectional design prohibits establishing a causal relationship and prohibits the examination of the selection bias (when PA spaces are established first and healthier people with more economic capital relocate nearby, known as selection bias, or if the PA spaces were developed after healthier people with more economic capital were living there, known as causation) [
72,
73]. Physical accessibility, programming (fitness classes, soccer leagues, etc.), quality, or capacity of the PA locations were not included in this analysis, representing a limitation. These factors may inhibit or facilitate local residents’ use of PA facilities and in turn their PA. A limitation to our mapping procedure could be missing private PA locations undetectable by internet search, although we tried to include all private PA locations. Finally, the PALAS and associated map were developed based on data from 2010 to 2014 to align with activity data collected in 2009–2012. If others wish to use the Baltimore City PALAS map and shapefiles from this study for a current analysis, it will need to be updated; however, all methods are described herein.
Generating a replicable method for mapping PA location availability could be useful for future researchers and policy makers. The PALAS includes many structures that may have been in place for decades (including schools and parks), however other components of the PALAS such as private sites are more mobile. The PALAS could be used in other cities as an indicator of PA location availability and used over time to distinguish changes in infrastructure for PA. PALAS data in addition to policy and planning documents could be used to illustrate connections between the built environment and policies pertaining to parks and recreation, changes in demographics, and urban development [
74]. An analysis of the relation between the PALAS, PA locations in the HNA, and other populations (children and adults of all ages, genders, and incomes) should also be conducted. To evaluate the selection bias, how/when/where PA locations change over time (retrospective and prospective) should be analyzed in relation to PA. This could determine the casual impact of PA locations and health promotion among residents, or whether healthy people move into areas with greater PA location availability. Finally, future studies should consider incorporating programs and/or quality of PA locations, and the expansion of the definition and categorization of parks and green space to include only developed and accessible land. By mapping and analyzing the programming, quality, and/or accessibility at PA locations, stronger linkages with PA may be established.