J Immigrant Minority Health DOI 10.1007/s10903-017-0625-1
ORIGINAL PAPER
Examining the Role of Income Inequality and Neighborhood Walkability on Obesity and Physical Activity among Low-Income Hispanic Adults Samuel D. Towne Jr.1 · Michael L. Lopez2 · Yajuan Li3 · Matthew Lee Smith1,4 · Judith L. Warren2 · Alexandra E. Evans5 · Marcia G. Ory6
© Springer Science+Business Media, LLC 2017
Abstract Obesity is a major public health issue affecting rising medical costs and contributing to morbidity and premature mortality. We aimed to identify factors that may play a role in obesity and physical activity at the individual and environmental/neighborhood levels. We analyzed data from an adult sample who were parents of students enrolled in a school-based health and wellness program. The sample was restricted to those who were Hispanic and whose children were on free/reduced lunch (n = 377). Dependent variables: body mass index (BMI); neighborhood walkability. Walk S core® was used to assess neighborhood walkability. Overall, 46% of participants were obese and 31% were overweight. The median age of respondents was 34 years, and the majority were female (88%) and married (59%). Participants who resided in a census tract with a higher relative income inequality (high, OR 2.54, 90% CI 1.154–5.601; moderate-high OR 2.527, 90% CI
1.324–4.821) and those who were unmarried (OR 1.807, 90% CI 1.119–2.917) were more likely to be obese versus normal weight. Overweight individuals that resided in areas that were walkable versus car-dependent averaged more days engaging in walking for at least 30-min (p <.05). Identifying individual and neighborhood factors associated with obesity can inform more targeted approaches to combat obesity at multiple ecological levels. The importance of understanding how neighborhood characteristics influence health-related and behavioral outcomes is further reinforced with the current findings. Identifying effective strategies to engage communities and organizations in creating, implementing, adopting, evaluating, and sustaining policy and/or environmental interventions will be needed to combat the obesity epidemic.
* Samuel D. Towne Jr.
[email protected]
2
Texas A&M AgriLife Extension, Texas A&M University, College Station, TX 77843, USA
Michael L. Lopez
[email protected]
3
Department of Agricultural Economics, Texas A&M University, College Station, TX 77843, USA
Yajuan Li
[email protected]
4
Department of Health Promotion and Behavior, College of Public Health, The University of Georgia, Athens, GA 30602, USA
5
Michael & Susan Dell Center for Healthy Living Division of Health Promotion and Behavioral Science School of Public Health, University of Texas, Austin, TX 78701, USA
6
Center for Population Health and Aging, Texas A&M School of Public Health, 1266 TAMU, College Station, TX 77843‑1266, USA
Matthew Lee Smith
[email protected] Judith L. Warren
[email protected] Alexandra E. Evans
[email protected] Marcia G. Ory
[email protected] 1
Texas A&M School of Public Health, 1266 TAMU, College Station, TX 77843‑1266, USA
13
Vol.:(0123456789)
Keywords Health and place · Hispanic · Minority · Low income · Physical activity · Income inequality
Background Obesity and low levels of physical activity are interrelated and serve as major public health concerns given the associated risks [1]. In the USA, the percentage of obese adults grew from 30% in 1999–2000 to 35% in 2011–2012 [2]. In addition, 69% of adults were classified as overweight [2]. The rate was higher among Hispanic individuals of Mexican descent, where the percentage of overweight males and females was 82 and 78%, respectively [2]. Furthermore, the percentage of the population who were obese was 35% (35% for males and 36% for females) for the general population [2]. Yet for Hispanic individuals of Mexican descent, the percentage of males and females that were obese was 40 and 46%, respectively [2]. Thus, Hispanic individuals face clear disparities in the rate of being overweight and obese when compared to the general population. The percentage of adults that met both aerobic activity and muscle-strengthening national guidelines in 2013 was 20% up from 15% in 1998 [2]. In contrast, the percentage of Hispanic individuals that met these guidelines in 2013 was 17% compared to 9% in 1998 [2]. Further, the percentage of individuals that met both aerobic activity and muscle strengthening guidelines decreased with decreasing income levels. For example, approximately 29% of those at 400% or more of the federal poverty level met aerobic activity and muscle-strengthening guidelines versus as little as 13% among those at 100% or below the federal poverty level [2]. In contrast, nearly 9% of Hispanic individuals at 100% or below the poverty level met these same guidelines [2]. Thus, Hispanic individuals continue to face disparities in meeting recommended physical activity guidelines when compared to the general population. Being overweight or obese and having related risk factors, including low levels of physical activity, are highly prevalent among Hispanic individuals versus the general population. As such, there is a critical need to identify geospatial areas (e.g. neighborhoods) with the greatest need of multi-faceted and multi-level interventions that target obesity and related risk factors among Hispanic individuals. Identifying place-based factors related to health disparities, namely neighborhood characteristics that adversely impact health and related outcomes is critical for targeting tailored area-specific interventions. The intersection of health and place, namely the role of the neighborhood, can provide a more complete picture when examining factors associated with individual-level health-related outcomes [3] thereby
13
J Immigrant Minority Health
going beyond examining the role of only individual characteristics (e.g., race and ethnicity), but also incorporating key area-level information (e.g., neighborhood characteristics) for a more holistic perspective. Neighborhood Walkability The relationship between neighborhood walkability and socioeconomic status has been shown to influence healthrelated behaviors and conditions such as physical activity and overweight/obesity [4]. Further, the relationship between neighborhood walkability and physical activity has been explored in a sample of older adults [5] and among a large sample of those with atherosclerosis [6] indicating a positive association between neighborhood walkability (using Walk Score® [7] to assess walkability) and walking. Yet, limited information regarding walkability exists for U.S. adults, given narrowly focused past studies (e.g. focusing on single areas/cities) [4, 8, 9]. Therefore, having a clearer understanding of the relationship between physical activity and neighborhood walkability is crucial. This is especially important among the growing population of Hispanic individuals in the USA. In particular, minority individuals with low incomes face multiple health disparities that challenge good health. Therefore, more research is needed to identify predictors of health-related outcomes, especially surrounding obesity and physical activity. Income Inequality Income inequality has been associated with poor health outcomes [10], and the need to investigate income inequality at multiple levels (e.g. counties, neighborhoods) has been suggested as a critical need [10]. For example, neighborhood level income inequality has been shown to effectively predict BMI differences among non-institutionalized Black and White adults [11]. Therefore, more studies that estimate the relative relationship between income inequality and health-related outcomes among diverse populations are needed. The purpose of the current study was to identify neighborhood characteristics associated with physical activity and obesity among low income Hispanic individuals. Objective We aimed to identify the relationship between: (1): neighborhood income inequality and health-related outcomes, namely overweight and obesity among Hispanic adults; and (2) neighborhood walkability and health-related behaviors, namely physical activity among Hispanic adults. These
J Immigrant Minority Health
aims were associated with the following research questions: (1) Are Hispanic residents of neighborhoods with higher income inequality (relative to those living in neighborhoods with lower income inequality) more likely to be overweight or obese?; (2) Are Hispanic residents of neighborhoods with greater walkability (versus those living in neighborhoods with lower walkability) more likely to report higher levels of physical activity?
The theoretical frameworks used in the current study can inform research questions in a broad sense, but can also be useful for building statistical models (e.g., identifying what variables to include as important covariates). This can help to ensure research questions are based on a strong theoretical foundation supported through the relevant research literature.
Methods
We analyzed data from a sample of adults (n = 867) who were parents or legal guardians of children enrolled in a school-based garden and nutrition intervention study residing in one of seven cities throughout Texas [16]. All parents or legal guardians were those of 3rd grade children in Title I elementary schools. All schools primarily served students from low income families (at least 40% of students from low-income families) [17]. The sample was restricted to Hispanic adults whose children were on free or reduced lunch and had complete information on study variables collected at baseline (n = 377). We included parents whose children were on free or reduced lunch to serve as a proxy for low socioeconomic status [18]. Our final sample size was further reduced in multivariable analyses and reported in the results (see Tables 1, 2, 3, 4, 5, 6).
Theoretical Foundation & Social Determinants of Health The theoretical framework in the current study is based on the Social Ecological Theory [12] and the World Health Organization’s Framework for the Social Determinants of Health [13]. These frameworks support the rationale for social and spatial determinants of health, which are key to the current study. The World Health Organization’s (WHO) framework for the Social Determinants of Health suggests several key determinants of health play a role in one’s health or health-related outcomes [13]. This framework has been used in several studies to identify key determinants (e.g. race/ethnicity, income, education) of health and health-related outcomes [14, 15]. As such, it was used as a model in the current study to examine the influence of neighborhood characteristics on health (obesity) and health behaviors (physical activity) among a sample of low income Hispanic individuals (see Fig. 1). As seen in the figure, place-based factors can influence health and related outcomes indirectly by affecting neighborhood factors (e.g., income inequality, walkability) and more directly through other means not specifically addressed in the current study.
Study Area, Setting and Sample
Variables Outcome Variables BMI The first major dependent variable was body mass index (BMI). BMI was calculated based on self-reported height and weight using the following formula: weight (lb)/ [height (in)]2 × 703 [19].
Fig. 1 Place-based factors related to health behaviors and related health outcomes Table 1 Distribution of major independent variables
Gini coefficient Walk score®
Min
Max
Lower quartile
Median
Upper quartile
Mean
Std.
0.31 0
0.52 77
0.39 14
0.41 38
0.42 53
0.41 35.20
0.05 21.89
13
13
BMI category
Employment
Marital Status
Education
Age
Sex
Male Female Mean +/−SD (Range) High school or less College or higher Married Unmarried, Divorced, or widowed Full-time employment Part-time employment Not employed or retired Normal weight Overweight Obese
45 332 34.19 +/− 6.74 (19–65) 255 122 224 153 169 59 149 86 117 174
n
Total n
67.64 32.36 59.42 40.58 44.83 15.65 39.52 22.81 31.03 46.15
11.94 88.06
%
Table 2 Sample distribution across selected characteristics and BMI (n = 377)
11 136 33.59 +/− 6.68 (19–65) 96 51 87 60 55 27 65 41 48 58
n
65.31 34.69 59.18 40.82 37.41 18.37 44.22 27.89 32.65 39.46
7.48 92.52
%
Low income inequality
Gini Coefficient Quartiles
13 80 34.40 +/− 5.97 (22–51) 74 19 61 32 50 11 32 26 26 41
n
79.57 20.43 65.59 34.41 53.76 11.83 34.41 27.96 27.96 44.09
13.98 86.02
%
Low-medium Income Inequality
12 71 34.42 +/− 7.16 (23–54) 51 32 49 34 33 13 37 12 24 47
n
61.45 38.55 59.04 40.96 39.76 15.66 44.58 14.46 28.92 56.63
14.46 85.54
%
Medium–high income inequality
9 45 35.04 +/− 7.45 (23–59) 34 20 27 27 31 8 15 7 19 28
n
High income inequality
62.96 37.04 50.00 50.00 57.41 14.81 27.78 12.96 35.19 51.85
16.67 83.33
%
J Immigrant Minority Health
J Immigrant Minority Health Table 3 Sample distribution across selected characteristics and walkability (n = 377) Total
Sex Age Education Marital Status Employment
Numbers of days walking at least 30 min
Number of days with any moderate physical activity
Male Female Mean +/−SD (Range) High school or less College or higher Married Unmarried, Divorced, or widowed Full-time employment Part-time employment Not employed or retired Mean +/−SD (Range) 0 days 1 day 2 days 3 days 4 days 5 days 6 days 7 days Mean +/−SD (Range)
Walkability walkable
Car-dependent (Most errands require a car)
Car-dependent (Almost all errands require a car)
n
%
n
%
n
%
n
%
45 332 34.19 +/− 6.74 (19–65) 255 122 224 153 169 59 149 3.06 +/−2.73 (0–7) 99 13 18 24 22 41 15 52 2.28 +/−2.24 (0–7) 118 27 64 52 17 32 8 29
11.94 88.06
13 96 34.40 +/−7.55 (24–59) 63 46 53 56 50 20 39 3.00 +/−2.80 (0–7) 32 2 7 3 5 15 4 15 2.54 +/−2.33 (0–7) 31 8 19 14 10 8 2 12
11.93 88.07
13 114 34.33 +/−6.04 (19–52) 93 34 82 45 58 18 51 3.13 +/−2.71 (0–7) 32 6 3 11 10 12 6 18 2.28 +/−2.34 (0–7) 38 13 19 15 2 10 4 11
10.24 89.76
15 99 34.06 +/−6.72 (22–65) 84 30 75 39 47 18 49 2.87 +/−2.68 (0–7) 29 5 7 9 6 9 4 14 1.88 +/−1.90 (0–7) 41 5 22 21 3 9 1 3
13.16 86.84
Walking The second major dependent variable was selfreported days participants walked at least 30 min in the past week (range from 0 to 7). Respondents were asked: ‘During the last 7 days, how many days did you walk for at least 10 min at a time?’ This was followed up with: ‘How much time did you usually spend walking on one of those days?’ Responses for these questions were combined to create a variable of the number days walking where participants reported at least an average of 30 min of walking. This measure of physical activity (i.e., walking) was then used to asses differences across walkability [20].
67.64 32.36 59.42 40.58 44.83 15.65 39.52
34.86 4.58 6.34 8.45 7.75 14.44 5.28 18.31
34.01 7.78 18.44 14.99 4.9 9.22 2.31 8.36
57.80 42.20 48.62 51.38 45.87 18.35 35.78
38.55 2.41 8.43 3.61 6.02 18.07 4.82 18.07
29.81 7.69 18.27 13.46 9.62 7.69 1.92 11.54
73.23 26.77 64.57 35.43 45.67 14.17 40.16
32.65 6.12 3.06 11.22 10.2 12.24 6.12 18.37
33.93 11.61 16.96 13.39 1.79 8.93 3.57 9.82
73.68 26.32 65.79 34.21 41.23 15.79 42.98
34.94 6.02 8.43 10.84 7.23 10.84 4.82 16.87
39.05 4.76 20.95 20 2.86 8.57 0.95 2.86
Independent Variables To ascertain whether income inequality influenced individuals’ BMI category, we used the Gini Coefficient [21]. Gini Coefficients can range from 0 to 1, with 0 indicating perfect equality and 1 indicating perfect inequality [22]. Income inequality measured through the Gini Coefficient was measured at the Census Tract [11] and split into quartiles. The Gini coefficient has been used as a measure of income inequality by the US Census Bureau [23] and the World Bank [22].
13
J Immigrant Minority Health
Table 4 Adjusted analyses predicting overweight or obese BMI versus normal BMI category BMI Category
Income inequality
Age Sex Marital status Employment Education
Low-moderate versus low Moderate-high versus Low High versus low Low-moderate versus low Moderate-high versus Low High versus low [Continuous variable] [Continuous variable] Male versus female Male versus female Unmarried, divorced, or widowed versus Married Unmarried, divorced, or widowed versus Married Unemployed or retired versus employed Unemployed or retired versus employed High school or less versus college or higher High school or less versus college or higher
Overweight versus normal weight Overweight versus normal weight Overweight versus normal weight Obese versus normal weight Obese versus normal weight Obese versus normal weight Overweight versus normal weight Obese versus normal weight Overweight versus normal weight Obese versus normal weight Overweight versus normal weight Obese versus normal weight Overweight versus normal weight Obese versus normal weight Overweight versus normal weight Obese versus normal weight
Odds ratio
0.860 1.844 2.234 1.132 2.527b 2.542a 1.017 1.043 1.606 1.066 0.908 1.807a 0.789 1.069 1.078 0.662
90% Confidence Interval Lower
Upper
0.470 0.924 0.977 0.644 1.324 1.154 0.977 1.005 0.735 0.483 0.537 1.119 0.471 0.662 0.626 0.403
1.574 3.677 5.112 1.991 4.821 5.601 1.058 1.082 3.508 2.351 1.532 2.917 1.322 1.727 1.856 1.087
a
alpha = 0.10
b
alpha = 0.05
In addition, we included Walk S core® to assess differences in walking patterns ranging from 0 to 100. Walk Score® is an objective measure that assesses walkability across the planet. This online database was used to match individual participants’ residence (using residential addresses) to calculate an individualized Walk Score® for each participant using Walk Score® software that incorporates distance to destination, etc. in its calculation. More information about the methodology for Walk S core® is available at: https://www.walkscore.com/methodology. shtml. We separated Walk Score® into three categories: (1) areas that were walkable (Walk S core® 50–100); (2) areas that were moderately car-dependent where most errands required a car (Walk Score® 25–49); and (3) areas that were highly car-dependent where almost all errands required a car (Walk Score® 0–24) [7, 24]. We also included the number of days participants reported any moderate physical activity. This was used for descriptive analyses only. Respondents were asked: ‘During the last 7 days, on how many days did you do moderate physical activities like carrying light loads, bicycling at a regular pace, or doubles tennis? Do not including walking.’ This was treated as a simple count variable (range from 0 to 7) and was introduced to describe participants’ physical activity levels.
13
Marital status was identified as married versus unmarried (i.e., single, divorced, or widowed). This was included as a measure to better contextualize family composition. Education was included as a key social determinant of health, coded as high school or less versus college or higher [13]. Employment status (unemployed or retired, employed) was also included given its ties to health (e.g. BMI) [25]. In addition, sex (male, female) and age (continuous) were also used to describe the sample. Statistical Analyses Bivariate analyses were used to describe the distribution of our sample across select characteristics. Based on the sample size used in the current study, analyses were restricted to comparisons where adequate cell sizes were able to be maintained. Multinomial logistic regression was used to model the likelihood of being overweight or obese versus normal weight. Alpha of 0.10 and 0.05 are reported for multinomial logistic regression analyses with 90% confidence intervals. In a stratified-group analyses we used Poisson regression to model the number of days in which participants reported walking at least 30 min on average. This was stratified by BMI group (normal weight group, overweight group, obese group). SAS 9.4 was used for all analyses [26].
J Immigrant Minority Health Table 5 Contrast estimates for Poisson regression predicting mean days of walking for at least 30 min among overweight individuals BMI: Overweight Group (n = 77)
Intercept Walkable (Walk s core® 50–100) Moderately car-dependent (Walk score® 25–49) Highly car-dependent (Walk score® 0–24) Age Male Female Unmarried, divorced, or widowed Married Unemployed or retired Employed (full or part-time) High school or less College or higher
Estimate
Wald 90% Confidence limits
p value
2.1892b 0.4435b 0.0377 0 −0.0347b −0.3562a 0 −0.2939a 0 −0.2964a 0 0.1038 0
1.5133 0.1435 −0.2774 0 −0.0516 −0.6939 0 −0.5649 0 −0.5519 0 −0.149 0
<0.0001 0.015 0.8439
Mean estimate (exponential)
Mean confidence intervals Lower
Upper
1.5005b 1.5581b 1.345a 1.4278a 1.3417a Adjusted mean 1.0102 0.6044 0.5667 0.5789 0.8753 0.549 0.9051 0.8740 0.5801
1.0731 1.0898 0.9919 0.9548 0.9715
2.0981 2.2278 1.8237 2.1352 1.8529
2.8652 0.7435 0.3528 0 −0.0178 −0.0184 0 −0.023 0 −0.0409 0 0.3565 0
0.0008 0.0828 0.0744 0.0564 0.4995
Contrast estimate results for significantly different variables Walkability
Walkable versus moderately car-dependent (Walk score® 25–49) Walkable versus Highly car-dependent (Walk score® 0–24) Employed versus Unemployed or retired Female versus male Married versus unmarried, divorced, or widowed Walkable (Walk s core® 50+) Moderately car-dependent (Walk score® 25–49) Highly car-dependent (Walk score® 0–24) Unemployed or retired Employed Male Female Married Unmarried, divorced, or widowed
p value
0.0177 0.015 0.0564 0.0828 0.0744
a
alpha = 0.10
b
alpha = 0.05
Results Descriptive Statistics Table 1 presents the distribution of the major independent variables within our study sample. The mean Gini Coefficient was 0.41 (standard deviation 0.05, range 0.31–0.52), which was lower than the relative Texas state average at 0.48 [23] indicating greater equality. In addition, the distribution of Walk Score® ranged from 0 to 77. The median Walk Score® was 38 with a mean of 35.2 (standard deviation 21.89) which were both classified as car-dependent. Only the
upper quartile of Walk Score® in the current sample were in the walkable range (at/above 50) at a Walk Score® of 53. Table 2 presents the overall distribution of our sample across selected characteristics by income inequality. Overall, 23% were in the normal weight category compared to 31% in the overweight group and 46% in the obese group. Individuals residing in areas with the low and low-moderate relative income inequality had rates of being overweight or obese (i.e., 33, 40, 28, and 44%, respectively). Individuals residing in areas with the medium–high and high relative income inequality had rates of being overweight or obese at 29, 57, 35, and 52%, respectively.
13
J Immigrant Minority Health
Table 6 Contrast estimates for Poisson regression predicting mean days of walking for at least 30 min among obese individuals BMI: over weight group (n = 117)
Intercept Walkable (Walk s core® 50–100) Moderately car-dependent (Walk score® 25–49) Highly car-dependent (Walk score® 0–24) Age Male Female Unmarried, divorced, or widowed Married Unemployed or retired Employed (full or part-time) High school or less College or higher
Estimate
Wald 90% Confidence Limits
0.8287b −0.1371 0.1766 0 0.0091 −0.0172 0 −0.1659 0 0.0286 0 −0.0325 0
0.3023 −0.3865 −0.0306 0 −0.0052 −0.2967 0 −0.3485 0 −0.1555 0 −0.2301 0
Mean estimate (exponential)
Mean Confidence intervals Lower
Upper
0.7307b 0.8719 Adjusted mean 0.921 1.2347 1.0581
0.5576 0.6477
0.9576 1.1736
1.355 0.1123 0.3839 0 0.0235 0.2623 0 0.0168 0 0.2126 0 0.1651 0
0.0096 0.3659 0.161 0.2956 0.9194 0.1352 0.7986 0.7867
Contrast estimate results for significantly different variables Walkability
Walkable versus moderately car-dependent (Walk score® 25–49) Walkable versus Highly car-dependent (Walk score® 0–24) Walkable (Walk s core® 50+) Moderately car-dependent (Walk score® 25–49) Highly car-dependent (Walk score® 0–24)
p value
0.023 0.3659
a
alpha = 0.10
b
alpha = 0.05
Table 3 presents the overall distribution of our sample across selected characteristics by walkability. On average, the number of days participants walked for at least 30 min in the previous week was 3 days. This was similar across different levels of walkability. The distribution of age and sex remained consistent across walkability scores. Nearly one-third of individuals failed to walk at least 30 min/day in a given week. Among those who did not walk at least 30 min per day in a given week, approximately 2 times as many participants were located in car-dependent areas versus areas that were walkable. The average number of days where participants engaged in any moderate physical activity was 2.3, ranging from an average of 2.54 among walkable areas to 1.88 among the most car-dependent areas. Nearly one-third of participants did not engage in any moderate physical activity in a given week. Adjusted Analyses Fully adjusted analyses included age, sex, marital status, employment status, educational attainment, and
13
employment in the regression model. Hispanic individuals with children on free or reduced lunch in neighborhoods with moderately high (OR 2.527, 90% CI 1.324–4.821, p <.05) and very high (OR 2.542, 90% CI, 1.154–5.601, p <.10) versus low income inequality were more likely to be obese than normal weight after controlling for all other terms in the model. Estimates for days walking at least 30 min are provided only for those in the overweight (n = 77) and obese (n = 117) groups given the weight-based focus of this paper and limited sample size among the normal weight (n = 55) group. Table 5 presents contrast estimates for Poisson regression predicting means days of walking for at least 30 min among overweight individuals (n = 77). The average number of days of walking for at least 30 min was 50% higher (p = 0.018) and 56% higher (p = 0.015) among Hispanic individuals with children on free or reduced lunch who were overweight and residing in neighborhoods that were walkable (Walk Score® 50–100) versus moderately car-dependent (Walk Score® 25–49) or highly car-dependent (Walk S core® 0–24), respectively.
J Immigrant Minority Health
Table 6 presents contrast estimates for Poisson regression predicting means days of walking for at least 30 min among obese individuals (n = 117). The average number of days of walking for at least 30 min was 36% lower (p = 0.023) among Hispanic individuals with children on free or reduced lunch who were residing in neighborhoods that were walkable (Walk S core® 50–100) versus moderately car-dependent (Walk Score® 25–49) after controlling for all other terms in the model.
Discussion This study sought to identify neighborhood factors, namely income inequality and walkability, related to health (BMI) and health-related behaviors (physical activity) among a sample of low-income Hispanic parents or legal guardians of 3rd grade students receiving free or discounted food services. Understanding the role that the environment plays in influencing health and health-related outcomes is a critical step to identifying targets for individuals and settings for inclusion in interventions, especially among vulnerable populations. Past research confirms the link between neighborhood characteristics and health and related outcomes. For example, neighborhood walkability using Walk Score® has been shown to be associated with walking among a sample of middle-aged and older adults in Texas, yet this study was somewhat limited in the relatively small number individuals from more diverse ethnic minority populations [5]. Further, research in Europe has also found associations with neighborhood walkability (measured with residential density and other characteristics) and engaging in moderate-to-vigorous physical activity [27]. Further, the neighborhood social environment has been shown to be associated with physical activity, yet this study was limited to African American and White males in a single city [28]. In addition, a relatively smaller study (n = 107) in California found associations between neighborhood walkability and both increased physical activity and lower obesity [29]. The current study helps to narrow the relative gap in research surrounding neighborhood environments and health and health-related outcomes among individuals from ethnic minority populations by concentrating on a Hispanic population with children in predominantly low-income schools. Understanding both geospatial and individual-level influencers on obesity is a critical need in addressing this costly and debilitating issue affecting millions both in the USA and globally. Globally, income inequality has been linked to BMI and obesity and risk factors for morbidity and mortality in large cross-country comparisons [30]. Findings from this study suggest that higher income inequality may be a strong predictor of obesity. To better understand and contextualize the underlying characteristics
associated with this relationship between income inequality and obesity, future studies should examine a more comprehensive set of objectively measured neighborhood- (e.g. land use, housing characteristics, residential density, rurality) and individual-level factors (e.g. health status, number of chronic conditions). Our findings are consistent with previous research indicating the importance of identifying strategies to address severe income inequality, especially for vulnerable populations (e.g. parents with low incomes from minority backgrounds) [11]. We found a strong link between living in a more walkable area and walking at least 30 min/day in a given week. This is consistent with previous research that has shown that among different populations, living in a more walkable area is associated with achieving more physical activity in the form of walking and general physical activity [31]. While this finding is intuitive, additional longitudinal studies are needed to build a solid research base of patterns and trends over time, especially among vulnerable populations. Identifying built, natural, and social factors of the environment that influence health-related outcomes is critical if we are to understand multifaceted and effective interventions and strategies that help make physical activity a natural and default part of regular routines.
Limitations This study was cross-sectional thus limiting our ability to describe trends over time. While the area-level factors described in this study are strong objective measures of walkability and income inequality, readers should note the ecological fallacy given these area-level factors may not specifically influence all individuals within the sample or general population. As such, we suggest likely associations, but causality is not implied, especially given the cross-sectional nature of the study. Self-reported data on physical activity were not as accurate as objectively measured data [32]. However, in lieu of using accelerometers or other objective measures, researchers often utilize surveys to study larger samples with limited resources. With this in mind, we were careful not to imply too great an accuracy or reliability on the level of specification from self-reported measures of physical activity in terms of minutes of physical activity. As such, we limited physical activity outcomes to weekly days of walking versus relying on exact minutes of moderate-to-vigorous physical activity. Even so, the relatively large sample size and rich data on both self-reported individual-level characteristics coupled with strong objective measures adds strength to the methodological rigor of the current study. Further, the generalizability of the study findings are limited, given data are drawn from several hundred
13
residents of several cities throughout one US state. In addition, this database contained administrative data (e.g., address, demographics) that were used to characterize study participants and this data was reliant on selfreport. Thus, recall bias could have affected some of the information collected. That said, surveying based on self report has been done throughout the US for several years in large national survey studies [33]. In addition, while the Framework for Action on the Social Determinants of Health [13] and the Social Ecological Theory [12] were used as the theoretical foundation for the current study, not all components of these frameworks were not included. The implications of the current findings should be taken in light of these limitations.
Policy Implications Policy and decision makers can use the current study findings to identify particularly at-risk areas to target funding for interventions to reach Hispanic adults and parents in areas with higher income inequality. Policies that target disparities or gaps in income, particularly at the neighborhood level or larger area, may be able to complement the already existing interventions and media campaigns targeting at-risk individuals. In addition, community-level interventions [34] that bring together key stakeholders may be better informed and prepared to increase community capacity to reach out to neighborhoods with relatively high income inequality. Strategic funding or other resources for interventions that target increased capacity and communication among key community players may hold promise given community empowerment may create sustainable partnerships that can target obesity-related issues and risk factors. Environmental interventions that target the built, natural, and social environments to be more walkable for both utilitarian and recreation walking have also been shown to be associated with greater physical activity [35] and lower obesity rates [4] than less walkable areas. Thus, policies that work to support walkable or activity-friendly community designs may also hold promise in increasing physical activity among particularly at-risk communities. For example, policy efforts may be needed to create walkways, pathways, or greenspaces in insecure areas to facilitate walking and other forms of physical activity. Environmental assessments may also be useful for identifying connectivity and proximity of insecure areas to resources such as utilitarian destinations. Further research must include larger samples and longitudinal analyses with objective measures of physical activity and comparison groups where possible. Increasing methodological rigor
13
J Immigrant Minority Health
of large scale studies will take, among other things, commitment from policy makers and other key stakeholders, including funders, to carry-out this critical research. Doing so will shed greater insight on large-scale sustainable solutions to these critical public health issues. Compliance with Ethical Standards Conflict of interest None of the authors have any competing interests in the manuscript Ethical Approval Ethical Approval was granted by the Texas A&M University Institutional Review Board (IRB Number: IRB20110012D). Informed consent Informed consent was obtained from all participants. Participation in this study was voluntary and the study was carried out in accordance with all ethical standards. Funding This material is based on work that is supported by the National Institute of Food and Agriculture, U.S. Department of Agriculture, under award number 2011-68001-30138. Any opinions, findings, conclusions, or recommendations expressed in this presentation are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture.
References 1. Committee, P.A.G.A. Physical activity guidelines advisory committee report, 2008. Washington: US Department of Health and Human Services; 2008. pp. A1–14. 2. National Center for Health Statistics (US). Health, United States, 2014: with special feature on adults aged 55–64, Hyattsville, MD: National Center for Health Statistics (US); 2015. 3. Kearns RA. Place and health: towards a reformed medical geography. Profess Geograph. 1993;45(2):p. 139–47. 4. Sallis JF, et al. Neighborhood built environment and income: examining multiple health outcomes. Soc Sci Med. 2009;68(7):1285–93. 5. Towne SD Jr, et al. Using walk score™ and neighborhood perceptions to assess walking among middle-aged and older adults. J Commun Health. 2016;41:977–88. 6. Hirsch JA, et al. Walk score® and transit score® and walking in the multi-ethnic study of atherosclerosis. Am J Prev Med. 2013;45(2):p. 158–66. 7. Score, W., Walk score methodology. http://www. walkscore. com/professional/methodology.php. Last viewed January, 2011. 2012;10:3503. 8. Hosler AS, Gallant MP, Riley-Jacome M, Rajulu DT. Relationship between objectively measured walkability and exercise walking among adults with diabetes. J Environ Public Health. 2014. doi:10.1155/2014/542123. 9. Lochner KA, et al. Social capital and neighborhood mortality rates in Chicago. Soc Sci Med. 2003;56(8):1797–805. 10. Kawachi I, Kennedy BP. Income inequality and health: pathways and mechanisms. Health Serv Res. 1999;34(1 Pt 2):215. 11. Robert SA, Reither EN. A multilevel analysis of race, community disadvantage, and body mass index among adults in the US. Soc Sci Med. 2004;59(12):2421–34.
J Immigrant Minority Health 12. McLeroy KR, et al. An ecological perspective on health promotion programs. Health Educ Behav. 1988;15(4):351–77. 13. Solar O, Irwin A. A conceptual framework for action on the social determinants of health. World Health Organization. 2007. p. 1–77. 14. Marmot M, et al. Closing the gap in a generation: health equity through action on the social determinants of health. Lancet. 2008;372(9650):p. 1661–9. 15. Towne S. Differential effects of the great recession on minority populations. 2013. 16. Li Y, et al. Peer effects on childhood obesity from an intervention program. Health Behav Pol Rev. 2016;3(4):323–35. 17. U.S. Department of Education. Improving Basic Pro grams Operated by Local Educational Agencies (Title I, Part A). Available at: http://www2.ed.gov/programs/titleiparta/index.html. 2015; Available from: http://tea.texas.gov/ About_TEA/News_and_Multimedia/Press_Releases/2015/ TEA_identifies_more_than_400_Title_I_Reward_Schools/. 18. Hollar D, et al. Effect of a two-year obesity prevention intervention on percentile changes in body mass index and academic performance in low-income elementary school children. Am J Public Health. 2010;100(4):p. 646–53. 19. CDC. About adult BMI. Available at: https://www.cdc.gov/ healthyweight/assessing/bmi/adult_bmi/. [cited 2016]. 20. Van Cauwenberg J, et al. Environmental factors influencing older adults’ walking for transportation: a study using walk-along interviews. Int J Behav Nutr Phys Act. 2012;9(1):85. 21. Fan JX, Wen M, Kowaleski-Jones L. Tract-and county-level income inequality and individual risk of obesity in the United States. Soc Sci Res. 2016;55:75–82. 22. World Bank, Development Research Group. GINI index (World Bank estimate). 2015. 23. Noss A. Household income: 2013. Washington: US Census Bureau, Economics and Statistics Administration, US Department of Commerce, 2014.
24. Carr LJ, Dunsiger SI, Marcus BH. Validation of walk score for estimating access to walkable amenities. Br J Sports Med. 2010;bjsports69609. 25. Sarlio-Lähteenkorva S, Lahelma E. The association of body mass index with social and economic disadvantage in women and men. Int J Epidemiol. 1999;28(3):445–9. 26. Institute S, Base SAS 9.4 Procedures guide: statistical procedures. 2014: SAS Institute. 27. Van Dyck D, et al. Neighborhood SES and walkability are related to physical activity behavior in Belgian adults. Prev Med. 2010;50:S74–S79. 28. Thorpe Jr, RJ., et al. Race, social and environmental condi tions, and health behaviors in men. Fam Commun Health. 2015;38(4):297. 29. Saelens BE, et al. Neighborhood-based differences in physical activity: an environment scale evaluation. Am J Public Health. 2003;93(9):1552–8. 30. Kim D. et al. Is inequality at the heart of it? Cross-country associations of income inequality with cardiovascular diseases and risk factors. Social Sci Med 2008;66(8):1719–32. 31. Brown SC, et al. Walk Score®: associations with purposive walking in recent Cuban immigrants. Am J Prev Med. 2013;45(2):202–6. 32. Prince SA, et al. A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review. Int J Behav Nutr Phys Act. 2008;5(1):56. 33. BRFSS, Behavioral Risk Factor Surveillance System. Behavioral Risk Factor Surveillance System Questionnaire. 2012. 2013. 34. McLeroy KR, et al. Community-based interventions. Am J Public Health. 2003;93(4):529–33. 35. Medicine, T.R.B.a.I.o., Does the built environment influence physical activity? Examining the evidence. Special Report 282. 2005: Washington.
13