Abstract
Background:
Recent years have seen increased interest in the role of neighborhood factors in chronic diseases such as cancers. Less is known about the role of neighborhood factors beyond individual demographics such as age or education. It is particularly important to examine neighborhood effects on health among African American men and women, considering the disproportionate impact of cancer on this group. This study evaluated the unique contribution of neighborhood characteristics (e.g., racial/ethnic diversity, income) beyond individual demographics, to cancer control behaviors in African American men and women.
Methods:
Individual-level data were drawn from a national survey (N=2,222). Participants’ home addresses were geocoded and merged with neighborhood data from the American Community Survey. Multi-level regressions examined the unique contribution of neighborhood characteristics beyond individual demographics, to a variety of cancer risk, prevention, and screening behaviors.
Results:
Neighborhood racial/ethnic diversity, median income, and percentage of home ownership made modest significant contributions beyond individual factors, in particular to smoking status where these factors were associated with lower likelihood of smoking (ps < .05). Men living in neighborhoods with older residents, and greater income and home ownership were significantly more likely to report prostate specific antigen testing (ps < .05). Regional analyses suggested different neighborhood factors were associated with smoking status depending on the region.
Conclusion:
Findings provide a more nuanced understanding of the interplay of social determinants of health and neighborhood social environment among African American men and women, with implications for cancer control interventions to eliminate cancer disparities.
Keywords: Neighborhood, Cancer, African American, Disparities, Multilevel, Health Behaviors
1. Introduction
A fundamental principle of health geography is that where a person lives or their neighborhood, affects individual health. There is considerable literature on the impact of neighborhood factors and a wide range of cancer-related outcomes from prevention and early detection[1–3] to survivorship[4,5]. Studies have more recently begun to examine multilevel models of the role of individual and neighborhood factors in cancer-related outcomes[5,6,7]. However additional research is needed to clarify whether neighborhood factors are simply a context to the lives of people or whether they significantly contribute to health and wellbeing, above and beyond the individual-level factors such as a person’s age or education. This is particularly the case when considering cancer control in diverse populations in the context of cancer disparities.
Recent writings further exemplify the importance of neighborhood factors, analyzed using geospatial techniques, in cancer control and population science research[8–10]. Krieger noted that there is “profound and startling lack of empirical research on the impact of residential segregation” on the cancer continuum from both a racial and economic perspective[9]. Because of historic and contemporary forces that perpetuate racial residential segregation, African American men and women in particular are more likely to live in disadvantaged neighborhoods with concentrated poverty[11,12]. The confluence of neighborhood poverty and racial segregation gives rise to distinctive physical and social neighborhood factors that are associated with poor health outcomes[13]. Several authors argued that to fully evaluate neighborhood effects on health, the complex and multilevel associations among individual socioeconomic status, neighborhood socioeconomic status, neighborhood social capital, segregation, and discrimination[13–17] must be investigated.
Although previous research suggests that neighborhoods can play a role in outcomes along the cancer continuum18, more investigation is required to understand the role of neighborhood characteristics in cancer control behaviors, a primary driver of cancer disparities19,20, as well as the relative contribution of neighborhood in relation to demographics such as age or education. The current study tests a multilevel model examining the extent to which indicators of neighborhood social environment (age, racial/ethnic diversity, income) contribute above and beyond key individual-level demographics, to a variety of cancer risk, prevention, and screening behaviors among African American men and women.
We hypothesize that neighborhood characteristics will make significant additional contributions to cancer control behaviors beyond the statistical contributions of individual demographics, such that African American men and women living in neighborhoods where the median age of residents is older, with greater racial/ethnic diversity and median income, and greater rates of home ownership will report more cancer control behaviors. Secondary analyses examined the neighborhood models for smoking status, the outcome most consistently associated with neighborhood factors, comparing/contrasting three US geographic regions. Study findings will enhance our understanding of cancer etiology and can inform multilevel cancer prevention and control interventions aiming to reduce persistent cancer disparities impacting this group.
2. Material and methods
2.1. Data sources
2.1.1. Neighborhood-level data
We used U.S. Census tracts as the neighborhood unit of analysis. Typically encompassing between 2,500 to 8,000 people, a census tract is smaller than a city but larger than a block group or census block[21]. The census tract is roughly equivalent to a neighborhood established by the Bureau of Census for analyzing stable population characteristics and is typically integrated with large national surveys at the individual level[22–25]. While we had considered using census block groups as the neighborhood level of analysis, the distribution of study participants per block group was highly skewed with a number of block groups with zero participants. Therefore, census tract rather than block group was chosen as the neighborhood unit of analysis.
We used the following neighborhood characteristics from the U.S. 2010 American Community Survey (ACS) at the tract level. Median Age reports the median age, in years, of residents (21 years and above) living in a census tract. Racial/Ethnic Diversity is a tract-level racial/ethnic diversity index that represents the likelihood that two persons, chosen at random from the same census tract, belong to different racial/ethnic groups. The index ranges from 0 to 100 with higher scores indicating greater diversity. Median Income reports the median household income in dollars for the past 12-months in a given census tract. Home Ownership reports the percentage of owner-occupied housing units out of the total housing units in a census tract, further adding to the nuance of neighborhood socioeconomic status.
2.1.2. Cancer control behavioral data
Data from the National Cancer Institute-funded Religion and Health in African Americans (RHIAA) study[26] are utilized because the study focused on a national probability sample of African American men and women and collected detailed survey data on a variety of cancer control behaviors. The study included 2,370 African American participants who completed a telephone interview. Sampling and recruitment methods are detailed elsewhere [27,28]. Eligibility criteria included speaking English, self-identifying as African American, age 21 years or older and no history of cancer due to survey items on cancer prevention and screening behaviors. Professional interviewers read participants an informed consent script and obtained verbal assent. Interviews were roughly one hour and included cancer risk (e.g., heavy alcohol use, smoking), prevention (e.g., fruit/vegetable consumption, physical activity) and screening (e.g., breast, prostate, colorectal) behaviors. A demographic module provided data on individual-level demographics described in 2.1.3. The University of Maryland Institutional Review Board approved the RHIAA study as well as the current work.
Behavioral Risk Factor Surveillance System (BRFSS) items were used to assess alcohol and tobacco use[29]. The alcohol consumption item asked whether participants had any alcohol use in the previous 30 days. For those answering yes, a subsequent item assessed heavy alcohol use as the number of times in the previous 30 days with 4 (for men) or 5 (for women) or more drinks on one occasion. The tobacco use item asked if participants had smoked cigarettes 100 or more cigarettes in their lifetime. The items had adequate test-retest reliability in a previous sample of African American men and women [30].
An adapted National Cancer Institute’s 5-A-Day Survey was used to evaluate average daily fruit and vegetable consumption[31,32], with seven items assessing fruit consumption and 5 items targeting vegetables. The brief version of the International Physical Activity Questionnaire[33] was used to measure physical activity, where items evaluate number of minutes in the previous week participants engaged in vigorous and moderate activity and walking.
Participants reported on select age-and sex-appropriate cancer screening behaviors consistent with screening recommendations at the time of data collection and using items based on the BRFSS. Women age 40 and older were asked whether they had completed a mammogram within the previous year. Men were asked whether they had completed a prostate specific antigen test in the past year.
2.1.3. Individual-level data
Individual age, sex, and education were identified as individual-level factors of focus based on their ubiquity in social epidemiological research on demographic characteristics associated with health-related outcomes. These indicators were self-reported by RHIAA study participants. Income was not utilized due to extent of missing data.
2.2. Geocoding and study area
Geocoding is the process of converting addresses into geographic coordinates used to place markers on a map. Of the 2,370 participants, addresses of 45 (1.9%) people could not be geocoded due to incorrect or incomplete addresses. After geocoding, we merged datasets from the two data sources: individual demographics and cancer control behaviors from the RHIAA survey and neighborhood level data from the 2010 ACS. For presentation and visualization purposes, we used the 9 established U.S. Census divisions to characterize the data: New England, Middle Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific[21]. Due to small sample size and the possibility of unstable results, three divisions (New England, Pacific, and Mountain) were not included in study analyses, resulting in a final analytic sample of N=2,222 in the study area of 6 census divisions.
2.3. Statistical methods
First, descriptive analyses (e.g., means/standard deviations, percentages) were conducted on the neighborhood characteristics and cancer control behavior variables to contextualize the multilevel model findings (see Table 1). Second, we conducted parametric (for continuous variables) and non-parametric (for binary variables) correlations to estimate bivariate associations between neighborhood-level factors and cancer control behaviors (see Table 2). Finally, we used hierarchical logistic (for binary dependent variables) and linear (for continuous dependent variables) regression, entering individual-level demographics (e.g., participant age, sex, education) in step one followed by each of the neighborhood-level variables separately in step 2 (see Tables 3–4). This hierarchical, multi-level approach was used to identify the unique role of each neighborhood factor in the outcome variable above and beyond individual-level demographics. Due to the small increment of measurement, the neighborhood income variable was recoded into 5 levels for the regression models. A secondary exploratory analysis used US Census Divisions to create a Northeast, Midwest, and Southeast region and the models for smoking status were examined within these three regions.
Table 1.
Descriptive analysis of neighborhood characteristics and cancer control behaviors among African American study participants, 2008–2010
| Value | N | |
|---|---|---|
| Neighborhood characteristics | ||
| Median resident age (years) | 36.10 | 2221 |
| Racial/ethnic diversity index; Mean (SD) | 33.60 (21.92) | 2222 |
| Median resident income | $32,417.00 | 2221 |
| Percent home ownership; Mean (SD) | 45.94 (19.54) | 2222 |
| Cancer risk behaviors | Mean (SD) or % yes | N |
| Used alcohol in past 30 days | 41.1% | 2,218 |
| Had 4/5+ alcoholic beverages/day in past 30 daysa | 16.6% | 2,169 |
| Smoked > 100 cigarettes in lifetime | 44.1% | 2,217 |
| Cancer prevention behaviors | ||
| Fruit servings / day | 2.43 (1.36) | 2,222 |
| Vegetable servings / day | 2.16 (0.96) | 2,222 |
| Walking minutes / week | 48.40 (53.45) | 2,123 |
| Moderate physical activity minutes / week | 40.15 (53.47) | 2,134 |
| Vigorous physical activity minutes / week | 46.72 (57.45) | 2,141 |
| Cancer screening behaviors | ||
| Mammogram in the past yearb | 76.2% | 1,068 |
| Prostate specific antigen test in the past yearc | 68.1% | 429 |
Note. Sample sizes vary due to sub-sample and skip patterns.
heavy drinking defined as 4+/day for women, 5+/day for men
Among women age 40 and older
Among men age 45 and older
Table 2.
Pearson’s correlations of neighborhood factors and cancer control behaviors among African American study participants, 2008–2010
| Median Resident Age | Racial/Ethnic Diversity | Median Resident Income | Home Ownership | |
|---|---|---|---|---|
| Cancer risk behaviors | ||||
| Alcohol Use in past 30 days | −.04* | −.02 | .04 | −.02 |
| # days with 4/5+ alcoholic beverages/day in past 30 daysa | −.02 | −.04 | .04 | −.04* |
| Smoked >= 100 cigarettes | −.04 | −.06* | −.07** | −.08** |
| Cancer prevention behaviors | ||||
| Fruit servings per day | .04 | −.01 | .03 | .02 |
| Vegetable servings per day | .05* | −.03 | .05* | .03 |
| Walking minutes / week | −.01 | .02 | .01 | .02 |
| Moderate physical activity minutes / week | −.01 | −.01 | −.01 | −.02 |
| Vigorous physical activity minutes / week | .01 | −.01 | −.01 | .01 |
| Cancer screening behaviors | ||||
| Mammogram in past year | .07* | .05 | .06* | .03 |
| PSA test in past year | .11* | .07 | .10* | .12* |
= p < .01;
= p < .05
heavy drinking defined as 4+/day for women, 5+/day for men
PSA = prostate specific antigen
Table 3.
Logistic regressions for role of individual and neighborhood factors in cancer control behaviors among African American study participants, 2008–2010
| Average Neighborhood Resident Age | |||||||
|---|---|---|---|---|---|---|---|
| Alcohol use in past 30 days | Independent variable | OR | 95% CI | Smoked >= 100 cigarettes | Independent variable | OR | 95% CI |
| Step 1*** | Age*** | 0.98 | 0.97–0.98 | Step 1*** | Age*** | 1.02 | 1.01–1.03 |
| Sex*** | 1.78 | 1.48–2.12 | Sex*** | 1.75 | 1.46–2.09 | ||
| Education*** | 1.16 | 1.07–1.26 | Education*** | 0.79 | 0.73–0.86 | ||
| Step 2^ | Age*** | 0.98 | 0.97–0.98 | Step 2^ | Age*** | 1.02 | 1.01–1.03 |
| Sex*** | 1.78 | 1.49–2.13 | Sex*** | 1.76 | 1.47–2.10 | ||
| Education*** | 1.17 | 1.08–1.28 | Education*** | 0.80 | 0.73–0.87 | ||
| N=2,206 | Neighborhood age^ | 0.99 | 0.97–1.00 | N=2,205 | Neighborhood age^ | 0.98 | 0.97–1.00 |
| Mammogram in past year | Independent variable | OR | 95% CI | PSA test in last year | Independent variable | OR | 95% CI |
| Step 1 | Age | 1.01 | 0.99–1.02 | Step 1^ | Age* | 1.03 | 1.00–1.05 |
| Sex | -- | -- | Sex | -- | -- | ||
| Education | 1.08 | 0.94–1.24 | Education | 0.99 | 0.81–1.20 | ||
| Step 2 | Age | 1.01 | 0.99–1.02 | Step 2* | Age* | 1.03 | 1.00–1.05 |
| Sex | -- | -- | Sex | -- | -- | ||
| Education | 1.06 | 0.92–1.21 | Education | 0.97 | 0.80–1.18 | ||
| N=1,061 | Neighborhood age | 1.03 | 1.00–1.05 | N=425 | Neighborhood age* | 1.04 | 1.00–1.08 |
| Neighborhood Racial/Ethnic Diversity | |||||||
| Alcohol use in past 30 days | Independent variable | OR | 95% CI | Smoked >= 100 cigarettes | Independent variable | OR | 95% CI |
| Step 1*** | Age*** | 0.98 | 0.97–0.98 | Step 1*** | Age*** | 1.02 | 1.01–1.03 |
| Sex*** | 1.77 | 1.48–2.11 | Sex*** | 1.75 | 1.46–2.08 | ||
| Education*** | 1.16 | 1.06–1.26 | Education*** | 0.79 | 0.72–0.86 | ||
| Step 2 | Age*** | 0.98 | 0.97–0.98 | Step 2* | Age*** | 1.02 | 1.01–1.02 |
| Sex*** | 1.76 | 1.47–2.10 | Sex*** | 1.74 | 1.45–2.07 | ||
| Education*** | 1.16 | 1.06–1.26 | Education*** | 0.79 | 0.72–0.86 | ||
| N=2,207 | Neighborhood diversity | 1.00 | 0.99–1.00 | N=2,206 | Neighborhood diversity* | 1.00 | 0.99–1.00 |
| Mammogram in past year | Independent variable | OR | 95% CI | PSA test in last year | Independent variable | OR | 95% CI |
| Step 1 | Age | 1.01 | 0.99–1.02 | Step 1^ | Age* | 1.03 | 1.00–1.05 |
| Sex | -- | -- | Sex | -- | -- | ||
| Education | 1.08 | 0.94–1.23 | Education | 0.99 | 0.81–1.20 | ||
| Step 2 | Age | 1.01 | 0.99–1.02 | Step 2 | Age* | 1.03 | 1.01–1.05 |
| Sex | -- | -- | Sex | -- | -- | ||
| Education | 1.08 | 0.94–1.23 | Education | 0.98 | 0.81–1.20 | ||
| N=1,062 | Neighborhood diversity | 1.01 | 1.00–1.01 | N=425 | Neighborhood diversity | 1.01 | 1.00–1.02 |
| Neighborhood Income | |||||||
| Alcohol use in past 30 days | Independent variable | OR | 95% CI | Smoked >= 100 cigarettes | Independent variable | OR | 95% CI |
| Step 1*** | Age*** | 0.98 | 0.97–0.98 | Step 1*** | Age*** | 1.02 | 1.01–1.03 |
| Sex*** | 1.77 | 1.48–2.11 | Sex*** | 1.75 | 1.46–2.08 | ||
| Education*** | 1.16 | 1.06–1.26 | Education*** | 0.79 | 0.72–0.86 | ||
| Step 2 | Age*** | 0.98 | 0.97–0.98 | Step 2* | Age*** | 1.02 | 1.01–1.03 |
| Sex*** | 1.77 | 1.48–2.11 | Sex*** | 1.75 | 1.47–2.10 | ||
| Education** | 1.15 | 1.06–1.26 | Education*** | 0.81 | 0.74–0.88 | ||
| N=2,207 | Neighborhood income | 1.01 | 0.95–1.08 | N=2,206 | Neighborhood income* | 0.93 | 0.88–0.99 |
| Mammogram in past year | Independent variable | OR | 95% CI | PSA test in last year | Independent variable | OR | 95% CI |
| Step 1 | Age | 1.01 | 0.99–1.02 | Step 1^ | Age* | 1.03 | 1.00–1.05 |
| Sex | -- | -- | Sex | -- | -- | ||
| Education | 1.08 | 0.94–1.23 | Education | 0.99 | 0.81–1.20 | ||
| Step 2^ | Age | 1.01 | 0.99–1.02 | Step 2** | Age | 1.03 | 1.01–1.06 |
| Sex | -- | -- | Sex | -- | -- | ||
| Education | 1.05 | 0.91–1.20 | Education | 0.93 | 0.76–1.13 | ||
| N=1,062 | Neighborhood income^ | 1.09 | 0.99–1.22 | N=425 | Neighborhood income** | 1.25 | 1.08–1.45 |
| Neighborhood Home Ownership | |||||||
| Alcohol use in past 30 days | Independent variable | OR | 95% CI | Smoked >= 100 cigarettes | Independent variable | OR | 95% CI |
| Step 1*** | Age*** | 0.98 | 0.97–0.98 | Step 1*** | Age*** | 1.02 | 1.01–1.03 |
| Sex*** | 1.77 | 1.48–2.11 | Sex*** | 1.75 | 1.46–2.08 | ||
| Education*** | 1.16 | 1.06–1.26 | Education*** | 0.79 | 0.72–0.86 | ||
| Step 2 | Age*** | 0.98 | 0.97–0.98 | Step 2** | Age*** | 1.02 | 1.01–1.03 |
| Sex*** | 1.77 | 1.48–2.11 | Sex*** | 1.75 | 1.46–2.09 | ||
| Education*** | 1.17 | 1.07–1.27 | Education*** | 0.80 | 0.74–0.88 | ||
| N=2,207 | Home ownership | 1.00 | 0.99–1.00 | N=2,206 | Home ownership** | 0.99 | 0.99–0.99 |
| Mammogram in past year | Independent variable | OR | 95% CI | PSA test in last year | Independent variable | OR | 95% CI |
| Step 1 | Age | 1.01 | 0.99–1.02 | Step 1^ | Age* | 1.04 | 1.00–1.05 |
| Sex | -- | -- | Sex | -- | -- | ||
| Education | 1.08 | 0.94–1.23 | Education | 0.81 | 0.81–1.20 | ||
| Step 2 | Age | 1.01 | 0.99–1.02 | Step 2** | Age* | 1.03 | 1.01–1.05 |
| Sex | -- | -- | Sex | -- | -- | ||
| Education | 1.06 | 0.93–1.22 | Education | 0.95 | 0.78–1.16 | ||
| N=1,062 | Home ownership | 1.00 | 1.00–1.01 | N=425 | Home ownership** | 1.02 | 1.00–1.03 |
Note:
= p < .10;
= p < .05;
= p < .01;
= p < .001
OR = odds ratio; 95% CI = confidence interval
Table 4.
Linear regressions for role of individual and neighborhood factors in cancer control behaviors among African American study participants, 2008–2010
| # days with 4/5+ alcoholic beverages/day in past 30 days | Independent variable | β | 95% CI | Fruit servings/day | Independent variable | β | 95% CI |
|---|---|---|---|---|---|---|---|
| Step 1*** | Age** | −.06 | −.02 – −.00 | Step 1*** | Age*** | .14 | .01 – .02 |
| Sex*** | −.08 | −.65 – −.18 | Sex** | .05 | .04 – .27 | ||
| Education^ | −.04 | −.21 – .01 | Education*** | .10 | .08 – .19 | ||
| Step 2 | Age** | −.01 | −.02 – −.00 | Step 2 | Age*** | .14 | .01 – .02 |
| Sex*** | −.42 | −.65 – −.18 | Sex** | .06 | .04 – .27 | ||
| Education^ | −.10 | −.21 – .01 | Education*** | .10 | .08 – .19 | ||
| N=2,154 | Neighborhood age | −.01 | −.03 – .02 | N=2,206 | Neighborhood age | .02 | −.01 – .01 |
| Vegetable servings/day | Independent variable | β | 95% CI | Walking minutes / week | Independent variable | β | 95% CI |
| Step 1*** | Age*** | .14 | .01 – .01 | Step 1*** | Age*** | −.12 | −.60 – −.29 |
| Sex* | .05 | .01 – .17 | Sex* | −.04 | −9.40 – −.02 | ||
| Education*** | .17 | .11 – .19 | Education | .01 | −2.05 – 2.29 | ||
| Step 2 | Age*** | .14 | .01 – .01 | Step 2 | Age*** | −.12 | −.60 – −.29 |
| Sex* | .05 | .01 – .17 | Sex^ | −.04 | −9.38 – .01 | ||
| Education*** | .16 | .11 – .19 | Education | .01 | −2.01 – 2.27 | ||
| N=2,206 | Neighborhood age | .03 | −.01 – .01 | N=2,108 | Neighborhood age | .01 | −.36 – .49 |
| Moderate activity minutes/week | Independent variable | β | 95% CI | Vigorous activity minutes/week | Independent variable | β | 95% CI |
| Step 1*** | Age*** | −.08 | −.42 – −.12 | Step 1*** | Age*** | −.13 | −.65 – −.33 |
| Sex*** | −.15 | −.20.66 – −11.35 | Sex*** | −.19 | −27.66 – −17.82 | ||
| Education | −.02 | −3.12 – 1.19 | Education | −.02 | −3.24 – 1.31 | ||
| Step 2 | Age*** | −.08 | −.42 – −.12 | Step 2 | Age*** | −.13 | −.65 – −.33 |
| Sex*** | −.15 | −20.69 – −11.36 | Sex*** | −.19 | −27.64 – −17.79 | ||
| Education | −.02 | −3.11 – 1.24 | Education | −.02 | −3.31 – 1.28 | ||
| N=2,119 | Neighborhood age | −.01 | −.47 – .37 | N=2,125 | Neighborhood age | .01 | −.37 – .52 |
| # days with 4/5+ alcoholic beverages/day in past 30 days | Independent variable | β | 95% CI | Fruit servings/day | Independent variable | β | 95% CI |
| Step 1*** | Age** | −.06 | −.02 – −.01 | Step 1*** | Age*** | .14 | .01 – .02 |
| Sex*** | −.07 | −.64 – −.17 | Sex*** | .05 | .04 – .27 | ||
| Education* | −.04 | −.22 – −.01 | Education*** | .10 | .08 – .19 | ||
| Step 2^ | Age** | −.07 | −.02 – −.01 | Step 2 | Age*** | .14 | .01 – .02 |
| Sex*** | −.07 | −.63 – −.16 | Sex*** | .05 | .04 – .27 | ||
| Education* | −.04 | −.22 – −.01 | Education*** | .10 | .08 – .19 | ||
| N=2,158 | Neighborhood diversity^ | −.04 | −.01 – .01 | N=2,210 | Neighborhood diversity | −.01 | −.01 – .00 |
| Vegetable servings/day | Independent variable | β | 95% CI | Walking minutes / week | Independent variable | β | 95% CI |
| Step 1*** | Age*** | .14 | .01 – .01 | Step 1*** | Age*** | −.12 | −.60 – −.29 |
| Sex* | .05 | .01 – .17 | Sex^ | −.04 | −9.35 – .04 | ||
| Education*** | .17 | .12 – .19 | Education | .01 | −2.10 – 2.24 | ||
| Step 2 | Age*** | .14 | .01 – .01 | Step 2 | Age*** | −.12 | −.60 – −.28 |
| Sex* | .05 | .01 – .17 | Sex^ | −.04 | −9.39 – .01 | ||
| Education*** | .17 | .12 – .19 | Education | .01 | −2.11 – 2.23 | ||
| N=2,210 | Neighborhood diversity | −.02 | −.01 – .01 | N=2,112 | Neighborhood diversity | .01 | −.07 – .14 |
| Moderate activity minutes/week | Independent variable | β | 95% CI | Vigorous activity minutes/week | Independent variable | β | 95% CI |
| Step 1*** | Age*** | −.08 | −.42 – −.12 | Step 1*** | Age*** | −.13 | −.66 – −.33 |
| Sex*** | −.15 | −20.69 – −11.38 | Sex*** | −.19 | −27.57 – −17.72 | ||
| Education | −.02 | −3.09 – 1.22 | Education | −.02 | −3.35 – 1.21 | ||
| Step 2 | Age*** | −.08 | −.43 – −.12 | Step 2 | Age*** | −.13 | −.66 – −.33 |
| Sex*** | −.15 | −20.68 – −11.36 | Sex*** | −.19 | −27.56 – −17.71 | ||
| Education | −.02 | −3.09 – 1.22 | Education | −.02 | −3.35 – 1.22 | ||
| N=2,122 | Neighborhood diversity | −.01 | −.12 – .09 | N=2,129 | Neighborhood diversity | −.01 | −.12 – .10 |
| # days with 4/5+ alcoholic beverages/day in past 30 days | Independent variable | β | 95% CI | Fruit servings/day | Independent variable | β | 95% CI |
| Step 1*** | Age** | −.06 | −.02 – −.01 | Step 1*** | Age*** | .14 | .01 – .02 |
| Sex*** | −.07 | −.64 – −.17 | Sex*** | .05 | .04 – .27 | ||
| Education^ | −.04 | −.22 – −.01 | Education*** | .10 | .08 – .19 | ||
| Step 2** | Age** | −.06 | −.02 – −.01 | Step 2 | Age*** | .14 | .01 – .02 |
| Sex*** | −.07 | −.64 – −.17 | Sex*** | .05 | .04 – .27 | ||
| Education | −.03 | −.19 – −.04 | Education*** | .10 | .07 – .19 | ||
| N=2,158 | Neighborhood income* | −.06 | −.19 – −.02 | N=2,210 | Neighborhood income | .02 | −.03 – .06 |
| Vegetable servings/day | Independent variable | β | 95% CI | Walking minutes / week | Independent variable | β | 95% CI |
| Step 1*** | Age*** | .14 | .01 – .01 | Step 1*** | Age*** | −.12 | −.60 – −.29 |
| Sex* | .05 | .01 – .17 | Sex^ | −.04 | −9.35 – .04 | ||
| Education*** | .17 | .12 – .19 | Education | .01 | −2.10 – 2.24 | ||
| Step 2 | Age*** | .14 | .01 – .01 | Step 2 | Age*** | −.12 | −.60 – −.29 |
| Sex* | .05 | .01 – .17 | Sex^ | −.04 | −9.35 – .04 | ||
| Education*** | .16 | .11 – .19 | Education | −.01 | −2.33 – 2.16 | ||
| N=2,210 | Neighborhood income | .02 | −.02 – .04 | N=2,112 | Neighborhood income | .01 | −1.20 – 2.13 |
| Moderate activity minutes/week | Independent variable | β | 95% CI | Vigorous activity minutes/week | Independent variable | β | 95% CI |
| Step 1*** | Age*** | −.08 | −.42 – −.12 | Step 1*** | Age*** | −.13 | −.66 – −.33 |
| Sex*** | −.15 | −20.69 – −11.38 | Sex*** | −.19 | −27.57 – −17.72 | ||
| Education | −.02 | −3.09 – 1.22 | Education | −.02 | −3.35 – 1.21 | ||
| Step 2 | Age*** | −.08 | −.42 – −.12 | Step 2 | Age*** | −.13 | −.66 – −.33 |
| Sex*** | −.15 | −20.69 – −11.38 | Sex*** | −.19 | −27.56 – −17.71 | ||
| Education | −.02 | −3.06 – 1.38 | Education | −.02 | −3.56 – 1.14 | ||
| N=2,122 | Neighborhood income | −.01 | −1.95 – 1.35 | N=2,129 | Neighborhood income | −.01 | −1.30 – 2.19 |
| # days with 4/5+ alcoholic beverages/day in past 30 days | Independent variable | β | 95% CI | Fruit servings/day | Independent variable | β | 95% CI |
| Step 1*** | Age** | −.06 | −.02 – −.01 | Step 1*** | Age*** | .14 | .01 – .02 |
| Sex*** | −.07 | −.64 – −.17 | Sex*** | .05 | .04 – .27 | ||
| Education* | −.04 | −.22 – −.01 | Education*** | .10 | .08 – .19 | ||
| Step 2^ | Age** | −.06 | −.02 – −.01 | Step 2 | Age*** | .14 | .01 – .02 |
| Sex** | −.07 | −.64 – −.16 | Sex*** | .05 | .04 – .27 | ||
| Education^ | −.04 | −.21 – .02 | Education*** | .10 | .08 – .19 | ||
| N=2,158 | Home ownership^ | −.04 | −.01 – .01 | N=2,210 | Home ownership | −.01 | −.01 – .01 |
| Vegetable servings/day | Independent variable | β | 95% CI | Walking minutes / week | Independent variable | β | 95% CI |
| Step 1*** | Age*** | .14 | .01 – .01 | Step 1*** | Age*** | −.12 | −.60 – −.29 |
| Sex* | .05 | .01 – .17 | Sex^ | −.04 | −9.35 – .04 | ||
| Education | .17 | .12 – .19 | Education | .01 | −2.10 – 2.24 | ||
| Step 2 | Age*** | .14 | .01 – .01 | Step 2 | Age*** | −.12 | −.60 – −.29 |
| Sex* | .05 | .01 – .17 | Sex^ | −.04 | −9.37 – .02 | ||
| Education*** | .17 | .12 – .19 | Education | −.01 | −2.34 – 2.06 | ||
| N=2,210 | Home ownership | −.01 | −.01 – .01 | N=2,112 | Home ownership | .02 | |
| Moderate activity minutes/week | Independent variable | β | 95% CI | Vigorous activity minutes/week | Independent variable | β | 95% CI |
| Step 1*** | Age*** | −.08 | −.42 – −.12 | Step 1*** | Age*** | −.13 | −.66 – −.33 |
| Sex*** | −.15 | −20.69 – −11.38 | Sex*** | −.19 | −27.57 – −17.72 | ||
| Education | −.02 | −3.09 – 1.22 | Education | −.02 | −3.35 – 1.21 | ||
| Step 2 | Age*** | −.08 | −.42 – −.12 | Step 2 | Age*** | −.13 | −.66 – −.33 |
| Sex*** | −.15 | −20.68 – −11.37 | Sex*** | −.19 | −27.57 – −17.71 | ||
| Education | −.02 | −2.99 – 1.38 | Education | −.02 | −3.46 – 1.17 | ||
| N=2,122 | Home ownership | −.02 | −.16 – .08 | N=2,129 | Home ownership | .01 | −.10 – .15 |
Note:
= p < .10;
= p < .05;
= p < .01;
= p < .001;
significance at Block 2 indicates a significant F change
95% CI = confidence interval
3. Results
3.1. Descriptive analysis
3.1.1. Analytic sample
The analytic sample (N=2,222) was on average 53.8 years old (SD = 14.77), over half were female (62.3%), with a median household income of $30,001–40,000/year, and at least some college education (54.1%). The study sample is comparable to national averages for African American men and women with regard to income, with the 2015 median income of African American households reported as $36,544[34]. The current sample was similar to the U.S. population in proportion of men (current=37.7%, U.S.=47.7%) and educational attainment (current=25.2% attended 4+ years of college, U.S.=20.3%), however our participants were older than the U.S. African American population where the median age is 33.7 years.
3.1.2. Spatial distribution of sample
Figure 1 shows the distribution of study participants across the census divisions. Most were from the eastern half of the U.S. with higher proportions in the West South Central, South Atlantic, and East South-Central divisions. The spatial distribution of percentage of participants in the 6 U.S. Census divisions included in the analysis are as follows: 34.0% South Atlantic, 19.9% East North Central, 17.3% Middle Atlantic, 13.4% West South Central, 12.0% East South Central, 3.4% West North Central.
Figure 1:
Spatial distribution of African American study participants and study area
3.1.3. Cancer control behaviors
Descriptive data on cancer control behaviors is shown in Table 1. Just over 40% of the sample had consumed alcohol in the previous 30 days, and few engaged in heavy alcohol use (16.6%). Almost half had smoked 100 or more cigarettes in their lifetime. Mean daily fruit and vegetable servings combined was 4.59. For the three levels of physical activity, the mean varied from 40.15 to 48.40 minutes per week. With regard to cancer screening behaviors, roughly three-fourths of women reported having a mammogram in the past year and among men, almost seventy percent reported a prostate specific antigen test the past year.
3.1.4. Neighborhood characteristics
Descriptive data on participants’ neighborhood factors indicate that generally, the neighborhoods had a median resident age in the middle 30’s and the neighborhoods showed modest scores (33.60 out of 100) on the racial/ethnic diversity index. Neighborhood median household income was on average was just over $32,000 and fewer than 50% of housing units were owner-occupied.
3.2. Bivariate associations between neighborhood factors and cancer control behaviors
Table 2 shows correlations between the neighborhood characteristics and cancer control behaviors. For cancer risk behaviors, alcohol use showed a handful of associations with neighborhood factors, but cigarette smoking was consistently and significantly associated with less racially/ethnically diverse neighborhoods and those low in indicators of socioeconomic status. For cancer prevention behaviors, while there were no significant neighborhood associations with fruit consumption and only one for physical activity, vegetable consumption was significantly associated with greater neighborhood age and median income. Cancer screening utilization had significant associations with several neighborhood factors.
3.3. Multilevel models including individual and neighborhood factors
3.3.1. Neighborhood resident median age
Tables 3 and 4 show multi-level logistic and linear regressions, respectively, from multilevel models where individual demographics were entered in step 1 and then neighborhood factors in step 2, to evaluate whether they contributed uniquely to the outcomes. The logistic regression models indicate that average age of neighborhood residents made marginal additional contributions to alcohol use in the past 30 days and smoking history (ps < .10), in which as average resident median age increases, likelihood of these behaviors decreases. Average age of neighborhood residents did not contribute significantly beyond individual factors for heavy drinking, fruit or vegetable consumption, or physical activity. While average age of neighborhood residents did not significantly contribute to mammography, it did contribute beyond individual factors to men’s greater likelihood (OR = 1.04) of having a PSA exam in the previous year (p < .05).
3.3.2. Neighborhood racial/ethnic diversity
Tables 3 and 4 show multi-level logistic and linear regressions, respectively, for the neighborhood diversity models. Neighborhood racial/ethnic diversity made a significant additional contribution, beyond individual factors, to smoking history (p < .05), in which as racial/ethnic diversity in a neighborhood increases, likelihood of smoking decreases. Neighborhood racial/ethnic diversity contributed marginally to heavy drinking again where greater diversity was associated with lower heavy drinking (p < .10). Neighborhood racial/ethnic diversity did not make a significant contribution beyond individual factors for any other of the behaviors.
3.3.3. Neighborhood median income
Tables 3 and 4 show multi-level logistic and linear regressions, respectively, for the neighborhood median income models. Median income contributed uniquely to smoking history where participants living in neighborhoods with higher income were less likely (OR = 0.93) to report smoking (p < .05). For cancer screening, neighborhood median income contributed marginally to mammography (p < .10) and significantly to prostate specific antigen testing (p < .01; OR = 1.25), such that those living in neighborhoods higher in income were more likely to report screening. Median neighborhood income contributed uniquely to heavy drinking, where those living in neighborhoods with greater income were less likely to report heavy drinking (p < .01). Neighborhood median income did not make a significant contribution beyond individual factors for any other of the behaviors.
3.3.4. Neighborhood home ownership
Tables 3 and 4 show multi-level logistic and linear regressions, respectively, for the neighborhood home ownership models. Home ownership contributed uniquely to smoking history and marginally to heavy drinking in which participants living in neighborhoods where more people own their homes were less likely to smoke or engage in binge drinking (p < .01, p < .10, respectively). Home ownership made a significant contribution to men’s reporting prostate specific antigen testing whereby those living in neighborhoods where more people owned their homes were more likely to report screening (p < .01). Neighborhood home ownership did not make a significant contribution beyond individual factors for any other of the behaviors.
3.3.5. Regional analyses
Only in the Midwest region was neighborhood’s average resident age negatively associated with odds of smoking, beyond the demographic variables (p < .05) (see Supplemental Table 1). Only in the Southeast region was neighborhood racial/ethnic diversity negatively associated with odds of smoking, beyond the demographic variables (p < .05) (see Supplemental Table 2). Only in the Northeast region was neighborhood median income negatively associated with odds of smoking, beyond the demographic variables (p < .05, OR = 0.89) (see Supplemental Table 3). Home ownership in the Midwest was marginally negatively associated with odds of smoking, beyond the demographic variables (p < .10) while in the other two regions home ownership was non-significant (see Supplemental Table 4).
4. Discussion
This study makes an important contribution to understanding the role of neighborhood social environment in cancer through examination of the unique contribution of neighborhood factors historically rooted in structural racism and discrimination, above and beyond key individual demographics, to a variety of cancer control behaviors. The study’s focus on African American men and women is notable given the disproportionate burden of cancer impacting this group, and the study’s findings have implications for cancer control interventions aiming to close the disparity. Examining cancer risk, prevention, and screening behaviors provides an opportunity to determine whether neighborhood factors contribute similarly or differently to lifestyle behaviors from smoking and physical activity, to cancer screening.
The current findings suggest that the neighborhood social environment contributes statistically, yet modestly in magnitude, beyond key individual demographic factors, to smoking status in African American men and women and to men’s prostate cancer screening. Our participants living in neighborhoods with greater racial/ethnic diversity, median income, and home ownership were less likely to report smoking. It is possible that the social environments in this type of neighborhood may facilitate anti-tobacco norms, however there may be unmeasured factors such as lower stress that help explain these effects. For African American men, neighborhood income and home ownership were also associated with prostate cancer screening, yet neighborhood resident age was also a factor, again suggesting a normative context or information sharing between residents that may foster screening. Our findings are consistent with others who reported that neighborhood makes a unique contribution health indicators such as soda consumption and weight[35] and prostate cancer survival[5], or may interact with individual factors, such as the interaction of individual race/ethnicity and neighborhood racial composition in women’s development of high-grade cervical lesions[6]. However, there were also a number of null findings, such as for diet and physical activity. This highlights the importance of considering the applied implications of neighborhood social environment for different types of cancer control outcomes. Behaviors like diet or physical activity may be more rooted in the neighborhood built environment rather than the social environment.
In bivariate models, though alcohol use showed sporadic and modest associations with neighborhood factors, cigarette smoking was consistently and significantly associated with less racially/ethnically diverse neighborhoods and those low in socioeconomic indicators. Though these associations were small, they could be due to patterns of tobacco advertising in predominately African American neighborhoods and the availability of tobacco, including individual cigarette sales[36]. Alternatively, findings suggest that greater neighborhood racial/ethnic diversity may be a protective factor against smoking as people from different races/ethnicities bring different social/cultural norms and acceptance of smoking. The finding around neighborhood diversity and smoking is consistent with the body of literature suggesting that residential segregation has a predominately negative impact on health-related outcomes.
It is noteworthy that neighborhood characteristics contributed more consistently to cancer risk and screening behaviors than to cancer prevention behaviors such as diet and physical activity. Results suggest that individual factors may be more important for maintaining positive lifestyle behaviors while one’s neighborhood environment is more likely to play a role in unhealthy behaviors, yet other neighborhood factors may be supportive of cancer screening. It is possible that people living in neighborhoods with older residents get exposed to positive social influences and role modeling by others getting screened, or that providers strategically locate in neighborhoods with more people age-eligible for cancer screening.
The regional analysis suggested that different neighborhood factors appeared to play a role in smoking status, whereby in the Northeast, neighborhood income played a small role, in the Midwest, neighborhood resident age played a small role, and in the Southeast, neighborhood racial/ethnic diversity came into play. These results indicate the need for future research to determine why these patterns exist as well as conduct regional analyses for additional cancer control behaviors. The findings have implications for regionally targeted cancer interventions for African American men and women.
The current findings are to be taken in light of some limitations. Though the study took advantage of data from a national probability sample, there were too few African American households contacted in several census divisions, particularly those in the Western U.S., for data analysis. Findings are not to be taken as representing all African American men and women due to response biases in the sample such as older age. Second, there are additional individual- and interpersonal-level factors that also impact cancer control behaviors but were beyond the scope of the analysis. Researchers should consider examining more complex models of multiple influences on cancer-relevant outcomes, particularly among populations impacted by health disparities, including models that assess potential moderation or mediation effects. Third, though some studies examine neighborhood factors in health using an index comprised of multiple related indicators, we intended to evaluate the role of individual indicators neighborhood social environment. Use of an index can result in masking the role of individual indicators and makes the development of recommendations for targeted interventions imprecise. Finally, the effect sizes were modest for the neighborhood-level factors, however this may be due to small units of measurement. These findings call for further research to explicate the relative contributions of multi-level factors in cancer control behaviors in this and other populations.
5. Conclusions
The current analyses provide insights as to the modest yet unique contributions of neighborhood characteristics to cancer control behaviors in a group of people historically plagued by cancer disparities. Findings contribute to a more nuanced understanding of the interplay of multilevel social determinants of health among African American men and women. The findings have implications for the development of more effective cancer control interventions to eliminate the historically rooted cancer disparities impacting this group. This includes smoking cessation interventions that aim to minimize the negative impact of segregation or interventions that foster supportive relationships between mature African American men to encourage informed decision making for prostate cancer screening.
Supplementary Material
Acknowledgements
The team would like to acknowledge the work of Opinion America and Tina Madison who conducted participant recruitment/retention and data collection activities for the study.
Funding This work was supported by a grant from the National Cancer Institute, [grant number #1 R01 CA 105202]. Dr. Knott is supported by the Maryland Department of Health’s Cigarette Restitution Fund Program and by the University of Maryland Greenebaum Comprehensive Cancer [grant number P30CA134274]. Funding sources played no role in the study design, collection analysis and interpretation of data, writing the report, or the decision to submit the article for publication.
Footnotes
Declarations of interest: None
References
- 1.Nemeth JM, Thomson TL, Lu B, et al. A social–contextual investigation of smoking among rural women: multi-level factors associated with smoking status and considerations for cessation. Rural And Remote Health. 2018;18(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Calo WA, Vernon SW, Lairson DR, Linder SH. Associations between contextual factors and colorectal cancer screening in a racially and ethnically diverse population in Texas. Cancer epidemiology. 2015;39(6):798–804. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Leader AE, Michael YL. The association between neighborhood social capital and cancer screening. American Journal of Health Behavior. 2013;37(5):683–692. [DOI] [PubMed] [Google Scholar]
- 4.Zhou Y, Bemanian A, Beyer KM. Housing discrimination, residential racial segregation, and colorectal cancer survival in southeastern Wisconsin. Cancer Epidemiology, Biomarkers and Prevention. 2017;26(4):561–568. [DOI] [PubMed] [Google Scholar]
- 5.DeRouen MC, Schupp CW, Koo J, et al. Impact of individual and neighborhood factors on disparities in prostate cancer survival. Cancer epidemiology. 2018;53:1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Waggaman C, Julian P, Niccolai LM. Interactive effects of individual and neighborhood race and ethnicity on rates of high-grade cervical lesions. Cancer epidemiology. 2014;38(3):248–252. [DOI] [PubMed] [Google Scholar]
- 7.Smith JA, Zhao W, Wang X, et al. Neighborhood characteristics influence DNA methylation of genes involved in stress response and inflammation: The Multi-Ethnic Study of Atherosclerosis . Epigenetics. 2017;12(8):662–673. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hiatt RA. New directions in cancer control and population sciences. Cancer Epidemiology and Prevention Biomarkers. 2017;26(8):1165–1169. [DOI] [PubMed] [Google Scholar]
- 9.Krieger N Follow the north star: Why space, place, and power matter for geospatial approaches to cancer control and health equity. Cancer Epidemiology, Biomarkers and Prevention. 2017;26(4):476–479. [DOI] [PubMed] [Google Scholar]
- 10.Schootman M, Gomez SL, Henry KA, et al. Geospatial approaches to cancer control and population sciences. Cancer Epidemiology, Biomarkers and Prevention. 2017;26(4):472–475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Freeman HP. Cancer in the socioeconomically disadvantaged. CA: A Cancer Journal for Clinicians. 1989;39(5):266–288. [DOI] [PubMed] [Google Scholar]
- 12.Freeman HP. Poverty, Culture, and Social Injustice: Determinants of Cancer Disparities. CA: A Cancer Journal for Clinicians. 2004;54(2):72–77. [DOI] [PubMed] [Google Scholar]
- 13.Kramer MR, Hogue CR. Is segregation bad for your health? Epidemiologic Reviews. 2009;31:178–194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Fitzpatrick KM, LaGory M. Unhealthy places: The ecology of risk in the urban landscape. New York, NY: Routledge; 2000. [Google Scholar]
- 15.Kwate NO. Fried chicken and fresh apples: Racial segregation as a fundamental cause of fast food density in black neighborhoods. Health and Place. 2008;14(1):32–44. [DOI] [PubMed] [Google Scholar]
- 16.Morland K, Filomena S. Disparities in the availability of fruits and vegetables between racially segregated urban neighbourhoods. Public Health Nutrition. 2007;10(12):1481–1489. [DOI] [PubMed] [Google Scholar]
- 17.Moore LV, Diez Roux AV, Evenson KR, McGinn AP, Brines SJ. Availability of recreational resources in minority and low socioeconomic status areas. American Journal of Preventive Medicine. 2008;34(1):16–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Gomez SL, Shariff‐Marco S, DeRouen M, et al. The impact of neighborhood social and built environment factors across the cancer continuum: current research, methodological considerations, and future directions. Cancer. 2015;121(14):2314–2330. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.National Cancer Institute. Cancer Disparities. Vol 2019 2019. [Google Scholar]
- 20.Anand P, Kunnumakkara AB, Sundaram C, et al. Cancer is a preventable disease that requires major lifestyle changes. Pharmaceutical research. 2008;25(9):2097–2116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.U.S. Census Bureau. Geographic Terms and Concepts. https://www.census.gov/programs-surveys/geography.html. Published 2010. Accessed August 14, 2019.
- 22.Henry KA, Sherman RL, McDonald K, et al. Associations of census-tract poverty with subsite-specific colorectal cancer incidence rates and stage of disease at diagnosis in the United States. Journal of Cancer Epidemiology. 2014;2014:823484. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Eschbach K, Mahnken JD, Goodwin JS. Neighborhood composition and incidence of cancer among Hispanics in the United States. Cancer. 2005;103(5):1036–1044. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Chang ET, Yang J, Alfaro-Velcamp T, So SK, Glaser SL, Gomez SL. Disparities in liver cancer incidence by nativity, acculturation, and socioeconomic status in California Hispanics and Asians. Cancer Epidemiology, Biomarkers and Prevention. 2010;19(12):3106–3118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Freedman VA, Grafova IB, Rogowski J. Neighborhoods and chronic disease onset in later life. American Journal of Public Health. 2011;101(1):79–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Community Health Awareness M, & Prevention (CHAMP) Lab. Community Health Awareness, Messages, & Prevention (CHAMP). University of Maryland School of Public Health; https://sph.umd.edu/department/bch/lab/43501. Published 2009. Accessed 8/22/2019, 2019. [Google Scholar]
- 27.Holt CL, Le D, Calvanelli J, et al. Participant retention in a longitudinal national telephone survey of African American men and women. Ethnicity and Disease. 2015;25(2):187–192. [PMC free article] [PubMed] [Google Scholar]
- 28.Roth DL, Mwase I, Holt CL, Clark EM, Lukwago S, Kreuter MW. Religious involvement measurement model in a national sample of African Americans. Journal of Religion and Health. 2012;51(2):567–578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Centers for Disease Control and Prevention. Behavioral Risk Factors Surveillance System (BRFSS): Prevalence data. U.S. Department of Health and Human Services; http://www.cdc.gov/brfss. Published 2004. Accessed2004. [Google Scholar]
- 30.Stein AD, Lederman RI, Shea S. The Behavioral Risk Factor Surveillance System questionnaire: its reliability in a statewide sample. American Journal of Public Health. 1993;83(12):1768–1772. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Block G, Hartman AM, Dresser CM, Carroll MD, Gannon J, Gardner LA. A data-based approach to diet questionnaire design and testing. American Journal of Epidemiology. 1986;124(3):453–469. [DOI] [PubMed] [Google Scholar]
- 32.Kreuter MW, Skinner CS, Holt CL, et al. Cultural tailoring for mammography and fruit and vegetable consumption among low-income African American women in urban public health centers. Preventive Medicine. 2005;41(1):53–62. [DOI] [PubMed] [Google Scholar]
- 33.Craig CL, Marshall AL, Sjöström M, et al. International physical activity questionnaire: 12-country reliability and validity. Medicine and Science in Sports and Exercise. 2003;35(8):1381–1395. [DOI] [PubMed] [Google Scholar]
- 34.U.S. Census Bureau. 2015 American Community Survey 1-Year Estimates. U.S. Census Bureau; https://factfinder.census.gov/faces/nav/jsf/pages/index.xhtml.Published 2015. Accessed 11/01/17. [Google Scholar]
- 35.Wong MS, Chan KS, Jones-Smith JC, Colantuoni E, Thorpe RJ Jr, Bleich SN. The neighborhood environment and obesity: understanding variation by race/ethnicity. Preventive medicine. 2018;111:371–377. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Yerger VB, Przewoznik J, Malone RE. Racialized geography, corporate activity, and health disparities: Tobacco industry targeting of inner cities. Journal of Health Care for the Poor and Underserved. 2007;18(4 Suppl):10–38. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.