Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Apr 1.
Published in final edited form as: Am J Public Health. 2011 Nov 28;102(4):689–697. doi: 10.2105/AJPH.2011.300150

The Association Between Neighborhood Characteristics and Body Size and Physical Activity in the California Teachers Study Cohort

Theresa HM Keegan 1, Susan Hurley 2, Debbie Goldberg 3, David O Nelson 4, Peggy Reynolds 5, Leslie Bernstein 6, Pam L Horn-Ross 7, Scarlett L Gomez 8
PMCID: PMC3410673  NIHMSID: NIHMS388961  PMID: 21852626

Abstract

Objectives

We considered interactions between physical activity and body mass index (BMI) and neighborhood factors.

Methods

We used recursive partitioning to identify predictors of low recreational physical activity (<2.5 hours/week) and overweight and obesity (BMI≥25.0 kg/m2) among 118 315 women in the California Teachers Study. Neighborhood characteristics were based on 2000 US Census data and Reference US business listings.

Results

Low physical activity and being overweight or obese were associated with individual sociodemographic characteristics, including race/ethnicity and age. Among White women aged 36 to 75 years, living in neighborhoods with more household crowding was associated with a higher probability of low physical activity (54% vs 45% to 51%). In less crowded neighborhoods where more people worked outside the home, the existence of fewer neighborhood amenities was associated with a higher probability of low physical activity (51% vs 46%). Among non–African American middle-aged women, living in neighborhoods with a lower socioeconomic status was associated with a higher probability of being overweight or obese (46% to 59% vs 38% in high–socioeconomic status neighborhoods).

Conclusions

Associations between physical activity, overweight and obesity, and the built environment varied by sociodemographic characteristics in this educated population.

The prevalence of overweight and obesity has reached epidemic levels in the United States. The most recent national prevalence data report that 68% of US adults are overweight or obese.1 This prevalence poses a significant public health problem given that obesity is a risk factor for many chronic conditions and certain cancers.1–3 Because of the health benefits of physical activity in preventing and treating overweight and obesity, increasing physical activity has become a public health priority.4

With these high levels of overweight and obesity in the United States,1 researchers have begun to search for neighborhood factors, including the built environment, that may influence physical activity and overweight or obesity. The built environment comprises the physical attributes of a person’s surroundings, including the existence and condition of sidewalks and walking trails for walking and other types of recreation, the spatial configuration of street networks, the availability of health-promoting resources, and the number of walkable destinations. The research to date suggests that elements of the built environment influence physical activity and levels of overweight or obesity.5–7 Most studies, however, have focused on relatively specific geographic areas (e.g., a given city or county), have not assessed the influence of the built environment across different sociodemographic groups, and have not considered interactions between individual and neighborhood factors.

We aimed to determine the association between measures of the built environment and physical activity levels and body mass index (BMI, defined as weight in kg divided by height in m2) across geographically and sociodemographically diverse neighborhoods. We applied recursive partitioning, a tree-based classification method, to identify both independent associations and interactions between these variables in a large, established cohort of California women.

METHODS

The California Teachers Study (CTS) is a prospective study of 133479 women who were active or retired California teachers or administrators at the time the cohort was established in 1995 to 1996.8 Women joined the cohort by completing a self-administered questionnaire that assessed a wide range of lifestyle characteristics and exposures (including physical activity during the past 3 years) as well as personal and family medical history (including height and weight).

Geocoding

The current study was based on 118315 CTS participants who resided in California at the time of the baseline survey and for whom residential address could be geocoded to a census block group. Residential addresses at study entry were geocoded through a variety of efforts described previously.9 Briefly, batch geocoding was performed by Geographic Data Technology GDT (Geographic Data Technologies, Lebanon, NH). We then made extensive efforts to manually geocode addresses that could not be geocoded by the batch method. Together, these approaches resulted in assigning 95% of California baseline residences to a latitude and longitude and 98% to a US Census block group.

Outcome Data

Information on physical activity and BMI was obtained from responses to the baseline questionnaire. Women were asked about their physical activity levels over the previous 3 years, including the average number of hours per week and the number of months per year that they performed strenuous (e.g., swimming laps, aerobics, running, basketball, cycling on hills) and moderate (e.g., brisk walking, golf, volleyball, cycling on level streets) exercise activities or sports. The average hours per week and months per year of moderate and strenuous physical activity were summed and categorized into either meeting (≥2.5 hours per week, henceforth referred to as high), or not meeting (<2.5 hours per week, henceforth referred to as low), the physical activity recommendations of the American Cancer Society.3 Long-term metrics of these physical activity measures have been associated with cancer risk and mortality in this cohort.10–12 We excluded women from the physical activity analysis for whom physical activity data were missing (n=948).

Weight (lb) and height (ft, in) as self-reported at baseline were used to calculate BMI, which was then categorized as normal weight (<25.0 kg/m2) and overweight or obese (≥25.0 kg/m2).2 We excluded from the BMI analysis women with missing (n=5290) or extreme (<16.0 or >54.8 kg/m2; n=61) BMI values.

Sociodemographic Characteristics

Age at baseline was calculated from self-reported birth date and the date the questionnaire was completed. Self-reported race/ethnicity was categorized as non-Hispanic White (henceforth referred to as White), African American or Black, Hispanic, Asian or Pacific Islander, and other or not reported. All public school teachers and administrators have at least a college education; therefore, education as a measure of individual socioeconomic status (SES) is not useful within this cohort. However, individual SES for these women is partially determined by spousal income as well as other factors that are reflected in the diversity of neighborhoods in which they live.9 Thus, we used 2000 US Census data to characterize neighborhood urbanization and SES at the block group according to the address of residence at baseline.9 The degree of urbanization was categorized into 5 groups: urban, suburban, city, small town, and rural.9 This SES metric incorporates population measures of average neighborhood education, income, and occupation, as described elsewhere,9 and was categorized into quartiles based on the distribution of the block groups across the state.

Built Environment Measures

We derived measures of neighborhood housing and commuting characteristics from 2000 US Census data. Neighborhood density was characterized at the block group level by several measures: population density (number of people/m2), household crowding (percentage of occupied households with 1 or more persons/room), percentage of total housing units that are not single family dwellings, and percentage of total housing units that are housing structures with more than 10 units. Population-based commuting characteristics were summarized in the following manner: means of transportation (% of individuals in the block group who traveled to work via car or motorcycle, public transportation, walking or bicycle, other means, or worked at home) and typical travel time to work (% of individuals in the block group who traveled <30, 30–44, 45–59, or ≥60 min/d to work).

We obtained information on street connectivity, which was calculated at the US Census tract level, from the Rand Center for Population Health and Health Disparities.13 In addition to median area and median length of street blocks, the α and γ ratios were used to characterize the directness of links and density of connections in neighborhoods such that high values of these ratios (i.e., close to 1.0) indicated a pedestrian-friendly area with many short links and intersections and few dead ends.13 These measures have been shown to be predictive of health behaviors and outcomes in previous studies.5,14–17

We derived information on neighborhood amenities from Reference USA business listing data for 2002.18 Businesses of interest for the study were downloaded from the complete database, and their addresses were geocoded to a latitude and longitude by using ArcView GIS software (Esri, Redlands, CA). The number of businesses within a 400- or 1600-meter buffer around a participant’s residence was determined as amenities because these distances have been associated with physical activity and BMI.14,19–21 We selected and categorized businesses on the basis of a priori hypotheses regarding the influence such amenities might have on neighborhood walkability and BMI14,22,23 as follows: social capital (museums, historical sites, and similar institutions; theaters; and performing arts and spectator sports); recreation facilities (golf courses and country clubs, skiing facilities, marinas, fitness and recreational sports centers, bowling centers, and other amusement and recreation industries); beer, wine, and liquor stores; retail stores; gasoline stations; convenience stores; restaurants and eating places; specialty food stores; bars; parking lots and garages; and supermarkets and other grocery stores. Additionally, composite measures were created of desirable amenities (the number of social capital amenities and recreation facilities), undesirable amenities (the number of beer, wine, and liquor stores; gasoline stations; and convenience stores), and total destinations. Total destinations has been shown to be a reliable indicator of walkability.24

Statistical Analysis

We used recursive partitioning,25–27 conducted by using the RPART routine in the R statistical software program,26 to identify mutually exclusive subgroups of participants that varied with regard to the outcomes of interest (BMI and physical activity). Recursive partitioning is a nonparametric regression method that uses the current data set to find the decision tree with the lowest average misclassification rate for classifying future observations. The terminal nodes of the tree partition the participants into subgroups according to a set of explanatory variables. Starting with the original data set, recursive partitioning splits each node by examining each variable and selecting 1 binary split across the members in that node based on a variable that maximizes the purity in the outcome. This process is repeated until further partitioning is not possible. Because the final tree over-fits the data, 10-fold cross-validation is used to prune the tree.27 Finally, the “one-standard error” heuristic is used to find the simplest tree with a cross-validated error estimate no more than one standard error larger than the best tree.27 Recursive partitioning can detect multiway interactions and handle highly correlated variables; however, this method is less well-known and is not as powerful as parametric methods when the form of the underlying model is parametric and correctly specified.

We performed separate analyses for dichotomized outcomes of BMI and physical activity. We submitted 49 explanatory variables into the recursive partitioning procedure, including age group, race/ethnicity, neighborhood urbanization, SES, population density, 3 measures of housing density, 4 street connectivity measures, 9 measures of population commuting characteristics, and 22 specific and 6 composite neighborhood amenities within both 400 and 1600 meters of the residence. Because the results were similar for the neighborhood amenities variables regardless of buffer size, we only presented the results that assessed the 1600-meter buffer.

We used logistic regression models for low physical activity and overweight or obesity to estimate the adjusted odds ratios for categorical variables that represented each subgroup (terminal node) from the recursive partitioning analyses. Odds ratios were adjusted for age and BMI in the analyses of physical activity or age and physical activity in the analyses of BMI.

RESULTS

Most of the women included in these analyses were aged 46 to 75 years, were White, and had a normal BMI (Table 1). Nearly half of the participants engaged in moderate or strenuous physical activity for at least 2.5 hours per week, on average, in the 3 years before joining the cohort. Most participants lived in neighborhoods that were suburban, in the upper 2 quartiles of SES, had little or no household crowding, and were composed predominantly of single family homes. The participants’ neighborhoods tended to have relatively low average street connectivity (α ratio=0.15; γ ratio=0.43) and a high average percentage of residents commuting to work by car or motorcycle. The most common neighborhood services within 1600 meters of the participants’ homes were restaurants and eating establishments (mean number=32.6), followed by retail stores (mean number=14.7), social capital facilities (mean number=5.2), and supermarkets and grocery stores (mean number=4.9).

TABLE 1.

Demographic and Neighborhood Characteristics at Recruitment Among Study Participants: California Teachers Study, 1995–1996

Age, y, no. (%)
 ≤35 13 919 (11.8)
 36–45 22 155 (18.7)
 46–75 72 468 (61.2)
 ≥76 9773 (8.3)
Race/Ethnicity, no. (%)
 White (non-Hispanic) 102 135 (86.3)
 Hispanic 5089 (4.3)
 Asian/Pacific Islander 4321 (3.7)
 African American 3179 (2.7)
 Other or not reported 3591 (3.0)
BMI, no. (%)
 Normal, < 25.0 69 103 (58.4)
 Overweight, 25.0–29.9 28 140 (23.8)
 Obese, ≥30.0 15 782 (13.3)
 Unknown 5290 (4.5)
Recent moderate or strenuous physical activity, no. (%)
 Low activity, < 2.5 h/wk 58 973 (49.8)
 High activity, ≥2.5 h/wk 58 394 (49.4)
 Unknown 948 (0.8)
Area of residence, no. (%)
 Rural 14 151 (12.0)
 Small town 4075 (3.4)
 City 21 783 (18.4)
 Suburban 65 593 (55.4)
 Urban 12 262 (10.4)
 Unknown 451 (0.4)
Neighborhood SES quartile,a no. (%)
 1st, low 4821 (4.1)
 2nd 19 799 (16.7)
 3rd 38 615 (32.5)
 4th, high 54 591 (46.1)
 Unknown 489 (0.4)
Population density quartile,a no. (%)
 0.00–0.86 people/1000 m2 29 589 (25.0)
 0.87–1.91 people/1000 m2 29 569 (25.0)
 1.92–2.98 people/1000 m2 29 581 (25.0)
 2.99–66.76 people/1000 m2 29 575 (25.0)
 Unknown 1 (< 0.1)
Housing density,a %, mean ±SD
 Crowding: occupied households with ≥1 person/room 6.9 ± 8.8
 Nonsingle family housing units 22.6 ± 25.2
 Housing structures with ≥10 units 10.0 ± 16.2
Commuting characteristics,a %, mean ±SD
 Car/motorcycle 89.0 ± 8.3
 Public transportation 3.0 ± 5.2
 Walking or bicycle 2.5 ± 4.1
 Other means 0.6 ± 1.1
 Working at home 4.9 ± 3.8
 < 30 min 62.7 ± 14.3
 30–44 min 19.5 ± 8.6
 45–59 min 8.1 ± 5.4
 ≥60 min 9.7 ± 6.8
Street connectivity,b mean ± SD
 γ: ratio of streets/possible intersections 0.43 ± 0.05
 α: ratio of loops/possible loops 0.15 ± 0.07
 Median area of street blocks, sq ft 480 878 ± 1 221 385
 Median length of street blocks, ft 2843 ± 1564
No. of specific neighborhood amenities within 1600 m from address, mean ± SD
 Social capital 5.2 ± 9.2
 Beer, wine, and liquor stores 2.5 ± 3.3
 Recreation facilities 3.9 ± 4.0
 Retail stores 14.7 ± 18.0
 Gasoline stations 2.5 ± 2.5
 Convenience stores 1.3 ± 1.6
 Restaurants and eating places 32.6 ± 46.0
 Specialty food stores 3.5 ± 5.9
 Bars 1.9 ± 5.4
 Parking lots and garages < 1 ± 0.3
 Supermarkets and other grocery stores 4.9 ± 8.6
 Desirable amenitiesc 9.1 ± 12.2
 Undesirable amenitiesd 6.3 ± 6.1
 Total destinations 73.1 ± 95.2
Open in a new tab

Note. BMI = body mass index, defined as weight in kg divided by height in m2; SES = socioeconomic status. The sample size was n = 118 315.

a

Based on 2000 US Census block group data.

b

Based on 2000 US Census tract data.

c

Desirable amenities included social capital (museums, historical sites, and similar institutions; theaters; associations; performing arts; spectator sports) and recreation (golf courses and country clubs, skiing facilities, marinas, fitness and recreational sports centers, bowling centers, and other amusement and recreation industries).

d

Undesirable amenities included beer, wine, and liquor stores; gasoline stations; and convenience stores.

The built environment characteristics of the participants’ residences varied by neighborhood SES (Table 2). Compared with the lower SES neighborhoods, the highest SES neighborhoods were more likely to be suburban and to have lower population density, less household crowding, and a higher proportion of single-family homes. Block sizes were somewhat smaller in high-SES neighborhoods, and the percentage of the population that worked at home was lower in low-SES neighborhoods. Proximity to neighborhood amenities modestly differed by neighborhood SES.

TABLE 2.

Built Environment Characteristics by Quartile of Neighborhood Socioeconomic Status (SES): California Teachers Study, 1995–1996

Characteristic SES
Pa
Low, 1st to 3rd Quartiles (n = 63 235) High, 4th Quartile (n = 54 591)
Area of residence, no. (%) < .001
 Rural 11 952 (18.9) 2194 (4.0)
 Small Town 3779 (6.0) 296 (0.5)
 City 14 086 (22.3) 7686 (14.1)
 Suburban 25 824 (40.8) 39 747 (72.8)
 Urban 7594 (12.0) 4668 (8.6)
Population density quartiles,b no. (%) < .001
 0.00000–0.00086 people/m2 16 906 (26.7) 12 423 (22.8)
 0.00087–0.00191 people/m2 12 090 (19.1) 17 383 (31.8)
 0.00192–0.00298 people/m2 15481 (24.5) 14 034 (25.7)
 0.00299–0.06676 people/m2 18 758 (29.7) 10 751 (19.7)
Housing density,b %, mean ± SD
 Crowding: occupied households with ≥ 1 person/room 10.3 (10.4) 3.0 (3.5) < .001
 Nonsingle family housing units 27.2 (25.6) 17.3 (23.6) < .001
 Housing structures with ≥ 10 units 10.9 (16.7) 8.9 (16.3) < .001
Street connectivity,c mean ± SD < .001
 γ: ratio of streets/possible intersections 0.44 (0.05) 0.42 (0.05) < .001
 α: ratio of loops/possible loops 0.16 (0.07) 0.13 (0.07) < .001
 Median area of street blocks, sq ft 577 427 (1 539 695) 366 702 (667 541) < .001
 Median length of street blocks, ft 2944.4 (1913.4) 2720.6 (991.9) < .001
Commuting characteristics,a %, mean ± SD < .001
 Car/motorcycle 89.6 (8.2) 88.4 (8.4) < .001
 Public transportation 3.0 (5.2) 2.9 (5.0) .193
 Walking or bicycle 2.9 (4.4) 2.1 (3.8) < .001
 Other means 0.7 (1.2) 0.5 (0.9) < .001
 Working at home 3.9 (3.4) 6.0 (4.0) < .001
 < 30 min 63.9 (14.4) 61.4 (14.0) < .001
 30–44 min 18.6 (8.7) 20.6 (8.4) < .001
 45–59 min 7.6 (5.3) 8.6 (5.5) < .001
 ≥60 min 9.9 (7.3) 9.5 (6.3) < .001
No. of specific neighborhood amenities within 1600 m from address
 Social capital 4.8 (8.4) 5.8 (9.9) < .001
 Beer, wine, and liquor stores 2.7 (3.5) 2.2 (3.1) < .001
 Recreation facilities 3.6 (3.8) 4.2 (4.3) < .001
 Retail stores 14.3 (17.6) 15.2 (18.4) < .001
 Gasoline stations 2.7 (2.6) 2.3 (2.4) < .001
 Convenience stores 1.6 (1.7) 1.0 (1.3) < .001
 Restaurants and eating places 33.0 (46.3) 32.1 (45.5) < .001
 Specialty food stores 3.7 (6.2) 3.3 (5.6) < .001
 Bars 2.1 (5.6) 1.6 (5.2) < .001
 Parking lots and garages < 0.1 (0.2) 0.1 (0.3) < .001
 Supermarkets and other grocery stores 5.8 (9.3) 4.0 (7.5) < .001
 Desirable amenitiesd 8.4 (11.4) 10.0 (13.0) < .001
 Undesirable amenitiese 7.1 (6.4) 5.5 (5.6) < .001
 Total destinations 74.4 (96.2) 71.8 (94.2) < .001
Open in a new tab

Note. SES = socioeconomic status.

a

T-tests and χ2 statistics were used, where appropriate, to test for differences by SES categories.

b

Based on 2000 US Census block group data.

c

Based on 2000 US Census tract data.

d

Desirable amenities included social capital (museums, historical sites, and similar institutions; theaters; associations; performing arts; spectator sports) and recreation (golf courses and country clubs, skiing facilities, marinas, fitness and recreational sports centers, bowling centers, and other amusement and recreation industries).

e

Undesirable amenities included beer, wine, and liquor stores; gasoline stations; and convenience stores.

Predictors of Physical Activity

The recursive partitioning analysis of physical activity identified 7 subgroups of women with variable probabilities of low physical activity (Table 3). Women aged 76 years or older (node 1) and Hispanic, African American, and Asian/Pacific Islander women younger than 76 years (node 2) had the highest probabilities of low physical activity (66% and 60%, respectively), whereas White women and those of other or not reported race/ethnicity aged 35 years or younger (node 3) were the most active, with only a 40% probability of low physical activity. Among the majority of women, those who were aged 36 to 75 years and of White or other or not reported race/ethnicity, physical activity level varied by aspects of the built environment. Among these women, those who lived in neighborhoods with more household crowding (node 4) were slightly more likely to have low physical activity levels (54%) than were women in neighborhoods with less household crowding (nodes 5, 6, and 7 who had a probability of low physical activity ranging from 45% to 51%). In less crowded neighborhoods that also had a higher percentage of people who worked at home, women with more desirable amenities (node 7) were 5% less likely to have low levels of physical activity (probability=46%) than were women who lived in neighborhoods with fewer of these types of destinations (node 6; probability=51%).

TABLE 3.

Probability of Low Physical Activity and Logistic Regression Modeling of the Risk Subgroups Identified From Recursive Partitioning: California Teachers Study, 1995–1996

Age, Years Race/Ethnicity Crowdinga Work at Homeb Desirable Amenitiesc Terminal Node No. Probability of Low Physical Activity (95% CI) OR AORd
≥76 1 9491 0.66 (0.65, 0.67) 2.99 2.95
< 76 Hispanic, African American, and API 2 12 235 0.60 (0.59, 0.61) 2.29 2.16
≤35 White and other 3 11 494 0.40 (0.39, 0.40) 1.00 1.00
36–75 White and other ≥8% 4 19 344 0.54 (0.53, 0.55) 1.80 1.65
< 8% ≥3.9% 5 40 277 0.45 (0.45, 0.45) 1.25 1.21
< 3.9% < 11.5 6 19 666 0.51 (0.51, 0.52) 1.61 1.51
≥11.5 7 4860 0.46 (0.45, 0.47) 1.31 1.25
Open in a new tab

Note. AOR = adjusted odds ratio; API = Asian and Pacific Islander; CI = confidence interval; OR = odds ratio. Low physical activity is defined as moderate or strenuous physical activity of < 2.5 h/wk.

a

Percentage of housing units with 1 or more occupants per room.

b

Percentage who worked at home.

c

Desirable amenities included social capital (museums, historical sites, and similar institutions; theaters; associations; performing arts; and spectator sports) and recreation (golf courses and country clubs, skiing facilities, marinas, fitness and recreational sports centers, bowling centers, and other amusement and recreation industries).

d

Adjusted for age and body mass index (continuous).

Even after adjustment for age and BMI, older women (≥76 years) and women of Hispanic, African American, and Asian/Pacific Islander race/ethnicity who were younger than 76 years were at least 2 times more likely to engage in low levels of physical activity than were young (≤35 years) White women (Table 3). Compared with young White women, White women aged 36 to 75 years who lived in neighborhoods with more household crowding or in neighborhoods with less crowding but fewer desirable amenities were at least 1.5-fold more likely to be less physically active. Further adjustment of these models for neighborhood SES did not change these results (data not shown).

Predictors of Overweight and Obesity

Our recursive partitioning analyses for BMI identified 5 subgroups of women with variable probabilities of being overweight or obese (Table 4). Overall, the lowest probability of being overweight (33%) was observed among the youngest (≤45 years) and oldest (≥76 years) women (node 1). Only among middle-aged (46–75 years) women were neighborhood SES and race/ethnicity associated with various probabilities of being overweight. In high-SES neighborhoods, African American women (node 3) had a substantially higher probability (58%) of being overweight than did women of all other race/ethnicities (node 2; probability=38%). In lower SES neighborhoods, African American and Hispanic women and women of other or not reported race/ethnicity (node 5) had higher probabilities of being overweight (59%) than did White and Asian/Pacific Islander women (node 4; probability=46%). After adjustment for age and physical activity, African American women in all SES neighborhoods and Hispanic women in the lower 3 quartiles of neighborhood SES had more than a 2.4-fold increased odds of being overweight compared with younger or older women.

TABLE 4.

Probability of Being Overweight or Obese and Logistic Regression Modeling of the Risk Subgroups Identified From Recursive Partitioning: California Teachers Study, 1995–1996

Age, Years Neighborhood SES Race/Ethnicity Terminal Node No. Probability of Overweight or Obese (95% CI) OR AORa
≤45 or ≥76 1 43 495 0.33 (0.32, 0.33) 1.00 1.00
46–75 Highest quartile African American 3 772 0.58 (0.54, 0.61) 2.80 2.47
White, Hispanic API, and other 2 34 737 0.38 (0.37, 0.38) 1.25 1.18
Lower 3 quartiles White and API 4 30 549 0.46 (0.46, 0.47) 1.79 1.64
Hispanic African American, and other 5 3411 0.59 (0.58, 0.61) 3.03 2.68
Open in a new tab

Note. AOR = adjusted odds ratio; API = Asian and Pacific Islander; BMI = body mass index; OR = odds ratio; SES = socioeconomic status. Overweight or obese defined as BMI ≥25 kg/m2.

a

Adjusted for age and physical activity.

DISCUSSION

In this population of highly educated women living throughout the geographically diverse state of California, we found that a few elements of the built environment, including household crowding, commuting characteristics, and proximity to desirable neighborhood amenities, were associated with meeting physical activity recommendations, but only among White women. We did not find any specific measures of the built environment to be associated with overweight or obesity. Physical activity was also associated with age and race/ethnicity, and overweight or obesity was related to age, race/ethnicity, and neighborhood SES. Our analysis is unique in its application of recursive partitioning to identify interactive effects among individual and neighborhood predictors of physical activity and overweight and obesity. Most previous studies of these associations did not fully explore the potential interactions between sociodemographic variables and characteristics of the built environment, although characteristics of the built environment are strongly related to neighborhood racial/ethnic composition and SES.6,7,28–32

Previous studies have suggested that physical activity, primarily in the form of walking, is related to the built environment, with increasing activity levels associated with greater residential density, land-use mix, street connectivity, and access to recreational areas.5,15,16,19,33–35 The associations between our measure of recreational moderate or strenuous physical activity and the built environment were limited to middle-aged White women and were not as strong as findings from previous studies of walking, likely because of the differing physical activity measure. Among middle-aged White women, however, our finding of greater physical activity among those who lived in neighborhoods with more desirable amenities is consistent with several previous studies that considered more global measures of physical activity. These previous studies found greater levels of physical activity among people living near recreational resources such as parks, schools, and sports facilities.28,36–38 Household crowding and neighborhood commuting characteristics were also related to physical activity in our population. We found that women who lived in neighborhoods with more household crowding were less likely to meet physical activity recommendations. Our findings suggest that neighborhood attributes that facilitate recreational physical activity may differ from neighborhood attributes that favor walking for transportation.5 Furthermore, because the distributions of our household crowding and commuting variables varied strongly by neighborhood SES and because physical inactivity is associated with lower SES,15,39,40 these associations could also have been due to some residual SES-related effect, particularly individual-level SES.

Our recursive partitioning results indicate that the relationship between neighborhood SES and overweight or obesity is not uniform across racial/ethnic groups. Specifically, the probability of being overweight or obese was highest for African American women, regardless of neighborhood SES, whereas for women of other race/ethnicities, living in lower SES neighborhoods was associated with a higher probability of being overweight or obese. Although the lack of association between overweight and obesity and the built environment measures in our study differs from previous studies that found a lower prevalence of overweight and obesity associated with increased neighborhood walkability6,7,17,19,33–35,41,42 and to a lesser degree with neighborhood accessibility to food establishments that offer healthy food options,7 research on the built environment and body size and physical activity suggests that these associations may vary by sociodemographic characteristics. It has been hypothesized that people living in low-SES neighborhoods may be more influenced by the built environment because of transportation constraints or limited opportunities for mobility7,43; however, there are few consistent data to support this notion,44 perhaps because of the mediating influence of race/ethnicity, which is often not considered. In a population-based study in Atlanta, Georgia, elements of the built environment were related to walking behavior and obesity among White men but were not as evident among African Americans and women.45 In the same Atlanta study population, built environment features considered to enhance walkability were associated with lower rates of overweight among White, highly educated men, but relationships were opposite among women, non-White men, and less-educated men.15 Similarly, a recent analysis of national prevalence data reported that low neighborhood SES was associated with higher BMI among Hispanics and non-Hispanic Whites, but not among African Americans.46 By contrast, a recent study of predominantly low-income, Mexican-born Hispanics living in Texas revealed that higher SES was associated with greater BMI47; others have reported a similar positive association among Hispanics living in predominantly immigrant neighborhoods.46

The associations found in this population of educated professional women, although internally valid, may not be generalizable to other populations. Nearly half of CTS respondents reported meeting the American Cancer Society guidelines for physical activity, compared with only about one third of non-Hispanic White respondents to the 2005 National Health and Nutrition Examination Survey,40 and the prevalence of overweight and obesity among our study participants was considerably lower (37%) than were national (69%)48 or Californian (47%)8 prevalence data. However, the relationships between the sociodemographic variables and physical activity and overweight and obesity found in our study are consistent with those found in other studies. In the present study, older women, regardless of race/ethnicity, and younger African American, Asian/Pacific Islander, and Hispanic women had the highest probabilities of not meeting physical activity recommendations, which is consistent with well-documented lower levels of physical activity among minority women and older women.49,50 Similarly, our findings for overweight and obesity are consistent with national prevalence data indicating that the rates of overweight and obesity increase with age1,51 and are higher among non-White populations and in lower SES areas.6,50,52–54

Similar to most studies on this topic, this analysis was limited to cross-sectional data; therefore, the effects of the built environment cannot be distinguished from associations resulting from the self-selection of individuals to live in neighborhoods with certain attributes. We had no subjective measurements of the built environment (e.g., how safe it feels, how one uses their environment), which are likely to affect behavior as well.14 Contrary to most previous studies, which generally focused on relatively small (and usually urban) geographic areas, our study included women from a broad geographic area with a diversity of social and built environments. We were able to consider a large number of established and objectively measured elements of the built environment that were not subject to recall bias. Finally, our study sample provided a unique opportunity to evaluate neighborhood SES among a cohort of women who were relatively homogeneous with respect to education, which is one measure of individual SES.

Our findings suggest that the associations between physical activity and overweight and obesity and the built environment vary by sociodemographic characteristics and highlight the importance of considering these interactions in future research. We found that characteristics of the built environment were related to physical activity among White women. In addition, whereas neighborhood SES influenced the probability of being overweight or obese for most women, this was not the case for African American women. This is important given that the obesity epidemic in the United States disproportionately affects socioeconomically disadvantaged and minority populations. The relationship between neighborhood SES and overweight and obesity, despite relative homogeneity in education in this cohort, highlights the importance of SES and underscores the need for improving our understanding of the contextual factors associated with neighborhood SES.

Acknowledgments

This work was supported by National Cancer Institute funds (R03 CA128009 to T. H. M. K. and R01 CA77398) and the California Breast Cancer Research Fund (contract 97-10500). The collection of the cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the National Cancer Institute’s Surveillance, Epidemiology and End Results Program under contract N01-PC-35136 awarded to the Cancer Prevention Institute of California (formerly the Northern California Cancer Center), contract N01-PC-35139 awarded to the University of Southern California, and contract N02-PC-15105 awarded to the Public Health Institute; and the Centers for Disease Control and Prevention’s National Program of Cancer Registries, under agreement #U55/CCR921930-02 awarded to the Public Health Institute.

The authors thank Melissa Kealey for her contributions to this work and the California Teachers Study Steering Committee, who are responsible for the formation and maintenance of the cohort within which this study was conducted but who did not directly contribute to the current paper: Hoda Anton-Culver, Ellen T. Chang, Christina A. Clarke, Dennis Deapen, Katherine D. Henderson, James Lacey, Huiyan Ma, Susan L. Neuhausen, Rich Pinder, Frederick Schmaucher, Daniel O. Stram, Giske Ursin, Dee W. West, Sophia Wong, and Argyrios Ziogas.

Note. The funding sources did not contribute to the design or conduction on of the study nor the writing or submission of this article. The ideas and opinions expressed herein are those of the authors and endorsement by the State of California, Department of Health Services, the National Cancer Institute, and the Centers for Disease Control and Prevention or their contractors and subcontractors is not intended nor should be inferred.

Footnotes

Reprints can be ordered at http://www.ajph.org by clicking the “Reprints/Eprints” link.

Contributors

D. Goldberg led the analyses. S. Hurley, S. L. Gomez, D. O. Nelson, P. Reynolds, L. Bernstein, and P. L. Horn-Ross interpreted the data and assisted with drafting the article. T. H. M. Keegan originated the study and led the writing.

Human Participant Protection

Cancer Prevention Institute of California (CPIC) institutional review board approval was received for this study.

Contributor Information

Theresa H.M. Keegan, Cancer Prevention Institute of California, Fremont CA.

Susan Hurley, Cancer Prevention Institute of California, Fremont CA.

Debbie Goldberg, Cancer Prevention Institute of California, Fremont CA.

David O. Nelson, Cancer Prevention Institute of California, Fremont CA.

Peggy Reynolds, Cancer Prevention Institute of California, Fremont CA.

Leslie Bernstein, Division of Cancer Etiology, Department of Population Sciences, City of Hope National Medical Center, Duarte, CA.

Pam L. Horn-Ross, Cancer Prevention Institute of California, Fremont CA.

Scarlett L. Gomez, Cancer Prevention Institute of California, Fremont CA.

References

  • 1.Flegal KM, Carroll MD, Ogden CL, Curtin LR. Prevalence and trends in obesity among US adults, 1999–2008. JAMA. 2010;303(3):235–241. doi: 10.1001/jama.2009.2014. [DOI] [PubMed] [Google Scholar]
  • 2.Bianchini F, Kaaks R, Vainio H. Overweight, obesity, and cancer risk. Lancet Oncol. 2002;3(9):565–574. doi: 10.1016/s1470-2045(02)00849-5. [DOI] [PubMed] [Google Scholar]
  • 3.Kushi LH, Byers T, Doyle C, et al. American Cancer Society Guidelines on Nutrition and Physical Activity for cancer prevention: reducing the risk of cancer with healthy food choices and physical activity. CA Cancer J Clin. 2006;56(5):254–281. doi: 10.3322/canjclin.56.5.254. quiz 313–314. [DOI] [PubMed] [Google Scholar]
  • 4.US Department of Health and Human Services. 2008 Physical Activity Guidelines for Americans. Washington, DC: US Department of Health and Human Services; 2008. [Accessed June 1, 2011]. Available at: http://www.health.gov/paguidelines/guidelines/default.aspx. [Google Scholar]
  • 5.Saelens BE, Handy SL. Built environment correlates of walking: a review. Med Sci Sports Exerc. 2008;40(7 suppl):S550–S566. doi: 10.1249/MSS.0b013e31817c67a4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Booth KM, Pinkston MM, Poston WS. Obesity and the built environment. J Am Diet Assoc. 2005;105 (5 suppl 1):S110–S117. doi: 10.1016/j.jada.2005.02.045. [DOI] [PubMed] [Google Scholar]
  • 7.Papas MA, Alberg AJ, Ewing R, Helzlsouer KJ, Gary TL, Klassen AC. The built environment and obesity. Epidemiol Rev. 2007;29:129–143. doi: 10.1093/epirev/mxm009. [DOI] [PubMed] [Google Scholar]
  • 8.Bernstein L, Allen M, Anton-Culver H, et al. High breast cancer incidence rates among California teachers: results from the California Teachers Study (United States) Cancer Causes Control. 2002;13(7):625–635. doi: 10.1023/a:1019552126105. [DOI] [PubMed] [Google Scholar]
  • 9.Reynolds P, Hurley S, Goldberg DE, et al. Regional variations in breast cancer among California teachers. Epidemiology. 2004;15(6):746–754. doi: 10.1097/01.ede.0000134863.45834.50. [DOI] [PubMed] [Google Scholar]
  • 10.Mai PL, Sullivan-Halley J, Ursin G, et al. Physical activity and colon cancer risk among women in the California Teachers Study. Cancer Epidemiol Biomarkers Prev. 2007;16(3):517–525. doi: 10.1158/1055-9965.EPI-06-0747. [DOI] [PubMed] [Google Scholar]
  • 11.Dallal CM, Sullivan-Halley J, Ross RK, et al. Long-term recreational physical activity and risk of invasive and in situ breast cancer: the California Teachers Study. Arch Intern Med. 2007;167(4):408–415. doi: 10.1001/archinte.167.4.408. [DOI] [PubMed] [Google Scholar]
  • 12.West-Wright CN, Henderson KD, Sullivan-Halley J, et al. Long-term and recent recreational physical activity and survival after breast cancer: the California Teachers Study. Cancer Epidemiol Biomarkers Prev. 2009;18(11):2851–2859. doi: 10.1158/1055-9965.EPI-09-0538. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Center for Population Health and Health Disparities Data Core. Pittsburgh, PA: Center for Population Health and Health Disparities, RAND Corp; [Google Scholar]
  • 14.Brownson RC, Hoehner CM, Day K, Forsyth A, Sallis JF. Measuring the built environment for physical activity: state of the science. Am J Prev Med. 2009;36(4 suppl):S99–S123. doi: 10.1016/j.amepre.2009.01.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Frank LD, Kerr J, Sallis JF, Miles R, Chapman J. A hierarchy of sociodemographic and environmental correlates of walking and obesity. Prev Med. 2008;47(2):172–178. doi: 10.1016/j.ypmed.2008.04.004. [DOI] [PubMed] [Google Scholar]
  • 16.Frank LD, Schmid TL, Sallis JF, Chapman J, Saelens BE. Linking objectively measured physical activity with objectively measured urban form: findings from SMARTRAQ. Am J Prev Med. 2005;28(2 suppl 2):117–125. doi: 10.1016/j.amepre.2004.11.001. [DOI] [PubMed] [Google Scholar]
  • 17.Li F, Harmer PA, Cardinal BJ, et al. Built environment, adiposity, and physical activity in adults aged 50–75. Am J Prev Med. 2008;35(1):38–46. doi: 10.1016/j.amepre.2008.03.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Reference USA. US Businesses Database. Omaha, NE: InfoUSA; 2003. [Google Scholar]
  • 19.Smith KR, Brown BB, Yamada I, Kowaleski-Jones L, Zick CD, Fan JX. Walkability and body mass index density, design, and new diversity measures. Am J Prev Med. 2008;35(3):237–244. doi: 10.1016/j.amepre.2008.05.028. [DOI] [PubMed] [Google Scholar]
  • 20.McGinn AP, Evenson KR, Herring AH, Huston SL, Rodriguez DA. Exploring associations between physical activity and perceived and objective measures of the built environment. J Urban Health. 2007;84(2):162–184. doi: 10.1007/s11524-006-9136-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Nagel CL, Carlson NE, Bosworth M, Michael YL. The relation between neighborhood built environment and walking activity among older adults. Am J Epidemiol. 2008;168(4):461–468. doi: 10.1093/aje/kwn158. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Lee C, Vernaz Moudon A. The 3Ds + R: Quantifying land use and urban form correlates of walking. Transp Res Part D Trans Environ. 2006;11(3):204–215. [Google Scholar]
  • 23.Yen IH, Kaplan GA. Neighborhood social environment and risk of death: multilevel evidence from the Alameda County Study. Am J Epidemiol. 1999;149(10):898–907. doi: 10.1093/oxfordjournals.aje.a009733. [DOI] [PubMed] [Google Scholar]
  • 24.Moudon AV, Lee C, Cheadle AD, et al. Attributes of environments supporting walking. Am J Health Promot. 2007;21(5):448–459. doi: 10.4278/0890-1171-21.5.448. [DOI] [PubMed] [Google Scholar]
  • 25.Nelson LM, Bloch DA, Longstretch WT, Jr, Shi H. Recursive partitioning for the identification of disease risk subgroups: a case-control study of subarachnoid hemorrhage. J Clin Epidemiol. 1998;51(3):199–209. doi: 10.1016/s0895-4356(97)00268-0. [DOI] [PubMed] [Google Scholar]
  • 26.Atkinson EJ, Therneau TM. An Introduction to Recursive Partitioning Using the RPART Routines. Rochester, MN: Mayo Foundation; 2000. [Google Scholar]
  • 27.Brieman L, Friedman JH, Olshen RA, Stone CJ. Classification and Regression Trees. Belmont, CA: Wadsworth; 1984. [Google Scholar]
  • 28.Gordon-Larsen P, Nelson MC, Page P, Popkin BM. Inequality in the built environment underlies key health disparities in physical activity and obesity. Pediatrics. 2006;117(2):417–424. doi: 10.1542/peds.2005-0058. [DOI] [PubMed] [Google Scholar]
  • 29.Ross CE, Mirowsky J. Neighborhood disadvantage, disorder, and health. J Health Soc Behav. 2001;42(3):258–276. [PubMed] [Google Scholar]
  • 30.Moore LV, Diez Roux AV. Associations of neighborhood characteristics with the location and type of food stores. Am J Public Health. 2006;96(2):325–331. doi: 10.2105/AJPH.2004.058040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Reidpath DD, Burns C, Garrard J, Mahoney M, Townsend M. An ecological study of the relationship between social and environmental determinants of obesity. Health Place. 2002;8(2):141–145. doi: 10.1016/s1353-8292(01)00028-4. [DOI] [PubMed] [Google Scholar]
  • 32.Diez Roux AV. Estimating neighborhood health effects: the challenges of causal inference in a complex world. Soc Sci Med. 2004;58(10):1953–1960. doi: 10.1016/S0277-9536(03)00414-3. [DOI] [PubMed] [Google Scholar]
  • 33.Saelens BE, Sallis JF, Black JB, Chen D. Neighborhood-based differences in physical activity: an environment scale evaluation. Am J Public Health. 2003;93(9):1552–1558. doi: 10.2105/ajph.93.9.1552. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Ewing R, Schmid T, Killingsworth R, Zlot A, Raudenbush S. Relationship between urban sprawl and physical activity, obesity, and morbidity. Am J Health Promot. 2003;18(1):47–57. doi: 10.4278/0890-1171-18.1.47. [DOI] [PubMed] [Google Scholar]
  • 35.Berke EM, Koepsell TD, Moudon AV, Hoskins RE, Larson EB. Association of the built environment with physical activity and obesity in older persons. Am J Public Health. 2007;97(3):486–492. doi: 10.2105/AJPH.2006.085837. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Diez Roux AV, Evenson KR, McGinn AP, et al. Availability of recreation resources and physical activity in adults. Am J Public Health. 2007;97(3):493–499. doi: 10.2105/AJPH.2006.087734. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Oakes JM, Forsyth A, Schmitz KH. The effects of neighborhood density and street connectivity on walking behavior: the Twin Cities walking study. Epidemiol Perspect Innov. 2007;4:16. doi: 10.1186/1742-5573-4-16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.King WC, Belle SH, Brach JS, Simkin-Silverman LR, Soska T, Kriska AM. Objective measures of neighborhood environment and physical activity in older women. Am J Prev Med. 2005;28(5):461–469. doi: 10.1016/j.amepre.2005.02.001. [DOI] [PubMed] [Google Scholar]
  • 39.Winkleby MA, Kraemer HC, Ahn DK, Varady AN. Ethnic and socioeconomic differences in cardiovascular disease risk factors: findings for women from the Third National Health and Nutrition Examination Survey, 1988–1994. JAMA. 1998;280(4):356–362. doi: 10.1001/jama.280.4.356. [DOI] [PubMed] [Google Scholar]
  • 40.US Department of Health and Human Services, Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System (BRFSS) US Physical Activity Statistics; 2007. 2007 State Demographic Data Comparison. [Google Scholar]
  • 41.Mobley LR, Root ED, Finkelstein EA, Khavjou O, Farris RP, Will JC. Environment, obesity, and cardiovascular disease risk in low-income women. Am J Prev Med. 2006;30(4):327–332. doi: 10.1016/j.amepre.2005.12.001. [DOI] [PubMed] [Google Scholar]
  • 42.Rundle A, Roux AV, Free LM, Miller D, Neckerman KM, Weiss CC. The urban built environment and obesity in New York City: a multilevel analysis. Am J Health Promot. 2007;21(4 suppl):326–334. doi: 10.4278/0890-1171-21.4s.326. [DOI] [PubMed] [Google Scholar]
  • 43.Black JL, Macinko J. Neighborhoods and obesity. Nutr Rev. 2008;66(1):2–20. doi: 10.1111/j.1753-4887.2007.00001.x. [DOI] [PubMed] [Google Scholar]
  • 44.Lovasi GS, Neckerman KM, Quinn JW, Weiss CC, Rundle A. Effect of individual or neighborhood disadvantage on the association between neighborhood walkability and body mass index. Am J Public Health. 2009;99(2):279–284. doi: 10.2105/AJPH.2008.138230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Frank LD, Andresen MA, Schmid TL. Obesity relationships with community design, physical activity, and time spent in cars. Am J Prev Med. 2004;27(2):87–96. doi: 10.1016/j.amepre.2004.04.011. [DOI] [PubMed] [Google Scholar]
  • 46.Do DP, Dubowitz T, Bird CE, Lurie N, Escarce JJ, Finch BK. Neighborhood context and ethnicity differences in body mass index: a multilevel analysis using the NHANES III survey (1988–1994) Econ Hum Biol. 2007;5(2):179–203. doi: 10.1016/j.ehb.2007.03.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Rutt CD, Coleman KJ. Examining the relationships among built environment, physical activity, and body mass index in El Paso, TX. Prev Med. 2005;40(6):831–841. doi: 10.1016/j.ypmed.2004.09.035. [DOI] [PubMed] [Google Scholar]
  • 48.Centers for Disease Control and Prevention. Healthy Weight, Overweight, and Obesity Among U.S. Adults. Hyattsville, MD: National Center for Health Statistics; 2003. [Google Scholar]
  • 49.US Department of Health and Human Services, Centers for Disease Control and Prevention. Early Release of Selected Estimates Based on Data From the January–June 2008 National Health Interview Survey. Hyattsville, MD: National Center for Health Statistics; 2008. Leisure Time Physical Activity. [Google Scholar]
  • 50.Centers for Disease Control and Prevention. Differences in prevalence of obesity among Black, White, and Hispanic adults–United States, 2006–2008. MMWR Morb Mortal Wkly Rep. 2009;58(27):740–744. [PubMed] [Google Scholar]
  • 51.Ogden CL, Carroll MD, Curtin LR, McDowell MA, Tabak CJ, Flegal KM. Prevalence of overweight and obesity in the United States, 1999–2004. JAMA. 2006;295(13):1549–1555. doi: 10.1001/jama.295.13.1549. [DOI] [PubMed] [Google Scholar]
  • 52.Grafova IB, Freedman VA, Kumar R, Rogowski J. Neighborhoods and obesity in later life. Am J Public Health. 2008;98(11):2065–2071. doi: 10.2105/AJPH.2007.127712. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Ellaway A, Anderson A, Macintyre S. Does area of residence affect body size and shape? Int J Obes Relat Metab Disord. 1997;21(4):304–308. doi: 10.1038/sj.ijo.0800405. [DOI] [PubMed] [Google Scholar]
  • 54.Cubbin C, Hadden WC, Winkleby MA. Neighborhood context and cardiovascular disease risk factors: the contribution of material deprivation. Ethn Dis. 2001;11(4):687–700. [PubMed] [Google Scholar]

RESOURCES