Background
Cigarette smoking is a major risk factor for a multitude of diseases [
1], and despite declines in smoking in recent decades, an estimated 24% of men and 18% of women in the United States were smokers in 2009 [
2]. Low individual socioeconomic status (SES) is strongly associated with increased smoking prevalence across race and gender lines [
3], and recent work has begun to examine whether socioeconomic characteristics of the neighborhood in which a person resides influence smoking behavior independently from individual-level SES [
4‐
11]. Several plausible mechanisms have been suggested to explain how neighborhood-level factors might affect smoking behavior including the influence of neighborhood cultural or normative standards [
4], geographic distribution of tobacco advertising [
12], and psychosocial stress related to disadvantaged neighborhood settings [
5]. If neighborhood SES characteristics affect smoking behaviors above and beyond the influences of individual SES through these mechanisms or other pathways yet to be determined, novel public health interventions to reduce smoking initiation and encourage smoking cessation may be developed to target high-risk neighborhoods as well as individuals. However, before such interventions can be developed and appropriately tailored, research is needed to determine whether differences exist in the effects of neighborhood characteristics across race and gender groups. To date, however, most studies that have examined neighborhood SES in relation to smoking have had limitations regarding sample composition that have prevented robust comparisons of associations between neighborhood-level characteristics and smoking by race and gender. For example, in the United States, various measures of lower neighborhood-level SES have been associated with increased smoking prevalence in a study of young black and white adults (age 18-30) [
5], in small study populations in North Carolina [
7] and Illinois [
4], in a large national sample of black women [
8], and in participants residing in four communities (one of which included black participants) in the Atherosclerosis Risk in Communities (ARIC) study [
13]. To improve upon the limited comparisons across race and gender groups in these studies, we examined associations between current cigarette smoking and both individual-level and neighborhood-level characteristics in a large group of black and white adults age 40-79 living in twelve states in the southeastern US.
Results
Of the 72, 615 participants enrolled in the SCCS via CHCs from 2002-2009, 64, 960 (19, 561 black males, 27, 412 black females, 6, 231 white males, and 11, 756 white females) were included in the final analysis. Exclusions included 2, 953 (4.1%) participants who reported their race as being other than 'White' or 'Black/African American'; 139 (0.2%) with missing information on cigarette smoking; 1, 854 (2.6%) with missing information on individual-level characteristics; 82 (0.1%) whose address could not be geocoded; 344 (0.5%) whose address was outside of the 12-state enrollment area; 1, 780 (2.5%) who resided in such small block groups that the area measures were deemed to be unreliable (population < 300, housing units < 30, or > 33% of individuals living in group quarters); and 503 (0.7%) missing block group owner-occupied housing status.
The location and participant count of the 10, 168 block groups for the 64, 960 SCCS participants' home addresses at SCCS enrollment are shown in Figure
1. A mean of 6.4 participants resided in each block group (range 1-245). Individual-level household income and educational attainment were generally low among both blacks and whites (Table
1). As expected based on the large proportion of low-income participants, smoking prevalences were high among cohort members, and males were more likely to be current smokers than females. In contrast to the relatively similar distribution of individual-level education and income between the race groups, large differences were observed in the distribution of neighborhood-level SES characteristics with blacks being much more likely than whites to reside in block groups of lower SES (Table
1).
Cross-tabulations of participants across individual-level income and education and the neighborhood advantage summary score showed that, as expected, large numbers of participants with low individual income and education lived in low advantage neighborhoods, and similarly, large numbers of individuals of high individual income and education lived in high advantage neighborhoods (Table
2). Notably, however, meaningful numbers of participants were found across all categories of individual- by neighborhood-level SES. Differences by race were evident in the cross-tabulations; among individuals with household income < $15, 000/year, 13.4% and 11.9% of black males and females, respectively, lived in the most advantaged neighborhoods while 26.7% of white males and 26.0% of white females resided in the highest advantaged neighborhoods.
Table 2
Cross-tabulation of Individual-level income and education by neighborhood-level Neighborhood Advantage Summary score among participants enrolled in the Southern Community Cohort Study
Black males |
Individual-level household income |
< $15, 000 | 29.9 | 23.5 | 18.5 | 14.6 | 13.4 |
$15, 000-$24, 999 | 21.8 | 23.6 | 19.9 | 17.9 | 16.8 |
$25, 000-$49, 999 | 16.6 | 20.2 | 20.9 | 20.1 | 22.2 |
≥$50, 000 | 10.1 | 15.0 | 14.9 | 21.4 | 38.6 |
Individual-level education (years) |
< 9 years | 31.9 | 24.3 | 19.6 | 13.5 | 10.7 |
9- < 12 years | 29.7 | 23.5 | 18.8 | 15.8 | 12.3 |
12- < 16 | 24.2 | 22.9 | 19.3 | 16.6 | 17.0 |
≥16 | 18.2 | 17.9 | 15.7 | 18.2 | 30.0 |
Black females |
Individual-level household income |
< $15, 000 | 27.9 | 24.1 | 20.2 | 15.9 | 11.9 |
$15, 000-$24, 999 | 19.9 | 22.1 | 21.5 | 20.2 | 16.3 |
$25, 000-$49, 999 | 13.3 | 18.4 | 21.5 | 22.4 | 24.4 |
≥$50, 000 | 6.5 | 13.3 | 16.3 | 21.3 | 42.6 |
Individual-level education (years) |
< 9 years | 29.6 | 24.5 | 22.3 | 14.2 | 9.5 |
9- < 12 years | 30.3 | 23.7 | 20.4 | 14.9 | 10.7 |
12- < 16 | 21.6 | 22.9 | 20.4 | 19.1 | 16.0 |
≥16 | 14.0 | 16.2 | 20.5 | 20.3 | 28.9 |
White males |
Individual-level household income |
< $15, 000 | 12.5 | 14.8 | 19.6 | 26.4 | 26.7 |
$15, 000-$24, 999 | 9.2 | 13.1 | 19.5 | 27.0 | 31.1 |
$25, 000-$49, 999 | 4.3 | 10.6 | 17.5 | 29.7 | 37.8 |
≥$50, 000 | 1.5 | 8.8 | 15.9 | 21.1 | 52.7 |
Individual-level Education (years) |
< 9 years | 12.9 | 18.6 | 25.0 | 25.4 | 18.1 |
9- < 12 years | 13.2 | 14.8 | 20.6 | 27.5 | 24.0 |
12- < 16 | 8.9 | 13.1 | 18.4 | 27.4 | 32.1 |
≥16 | 5.7 | 7.3 | 13.1 | 21.7 | 52.3 |
White females |
Individual-level household income |
< $15, 000 | 9.5 | 14.1 | 21.8 | 28.5 | 26.0 |
$15, 000-$24, 999 | 6.7 | 12.3 | 22.6 | 28.0 | 30.4 |
$25, 000-$49, 999 | 2.7 | 10.0 | 20.0 | 29.0 | 38.2 |
≥$50, 000 | 1.5 | 8.2 | 14.3 | 21.8 | 54.2 |
Individual-level education (years) |
< 9 years | 12.1 | 18.1 | 22.1 | 28.2 | 19.4 |
9- < 12 years | 10.5 | 15.2 | 22.8 | 30.4 | 21.1 |
12- < 16 | 6.2 | 12.1 | 21.6 | 28.0 | 32.2 |
≥16 | 3.6 | 7.3 | 14.3 | 22.4 | 52.4 |
In regression models including only individual-level measures, both individual-level income and education were strongly associated with smoking in each race and sex group. In the lowest v. highest category of household income, PRs (95% CIs) for smoking were 1.71 (1.52-1.91) for black males, 1.78 (1.53-2.08) for black females, 1.63 (1.39-1.91) for white males, and 2.06 (1.78-2.38) for white females. Similarly for categories of education comparing lowest to highest levels, PRs (95% CIs) for smoking were 1.14 (1.05-1.23) for black males, 1.30 (1.17-1.45) for black females, 1.28 (1.13-1.45) for white males, and 1.62 (1.44-1.83) for white females.
Next, each neighborhood-level characteristic was examined individually in relation to current smoking in robust Poisson regression models accounting for within-neighborhood correlation with adjustment for individual-level characteristics (Table
3). Lower quartiles of neighborhood-level household income, percentage of adults with a high school education, percentage of owner-occupied housing units, and percentage of households earning interest, dividends, or rental income as well as higher quartiles of percentage in poverty were all associated with increased smoking in each sex and race group except white women. Lower quartiles of percentage of adults with a college education and percentage of adults employed in professional occupations were both associated with increased prevalence of smoking only among blacks, and the effects were strongest among black women. Decreasing neighborhood advantage summary score was associated with increased prevalence of smoking most clearly in black women with evidence of a similar but more modest trend being apparent for the other race and sex groups (Table
3).
Table 3
Prevalence ratios for current cigarette smoking according to categories of neighborhood-level SES characteristics from race- and sex-stratified robust Poisson regression modelsa
Percent poverty |
< 10% | 0.92 | (0.87-0.97) | 0.78 | (0.73-0.84) | 0.81 | (0.74-0.89) | 0.99 | (0.87-1.13) |
10 - < 20% | 0.95 | (0.91-0.99) | 0.84 | (0.79-0.90) | 0.80 | (0.73-0.88) | 1.01 | (0.89-1.14) |
20 - < 50% | 0.99 | (0.95-1.02) | 0.91 | (0.86-0.96) | 0.82 | (0.75-0.90) | 1.00 | (0.88-1.13) |
≥ 50% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
0zHousehold income |
< $18, 879 | 1.08 | (1.04-1.13) | 1.24 | (1.17-1.31) | 1.10 | (1.04-1.18) | 0.99 | (0.91-1.07) |
$18, 879 - < $26, 094 | 1.05 | (1.01-1.10) | 1.18 | (1.12-1.26) | 0.96 | (0.90-1.03) | 1.00 | (0.94-1.06) |
$26, 094 - < $34, 583 | 1.02 | (0.98-1.07) | 1.07 | (1.00-1.14) | 1.00 | (0.94-1.05) | 1.00 | (0.96-1.06) |
≥$34, 583 | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent adults with ≥ HS educationb
|
< 57.8% | 1.05 | (1.01-1.09) | 1.20 | (1.13-1.26) | 1.06 | (1.00-1.13) | 1.03 | (0.97-1.11) |
57.8% - < 67.5% | 1.01 | (0.97-1.05) | 1.12 | (1.06-1.18) | 0.95 | (0.89-1.02) | 1.01 | (0.95-1.07) |
67.5% - < 77.3% | 1.02 | (0.98-1.06) | 1.06 | (1.00-1.12) | 1.01 | (0.96-1.07) | 1.03 | (0.98-1.09) |
≥77.3% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent adults with ≥ college graduationb
|
< 5.8% | 1.05 | (1.01-1.08) | 1.22 | (1.15-1.28) | 1.02 | (0.96-1.08) | 1.04 | (0.98-1.11) |
5.8 - < 10.3% | 1.05 | (1.01-1.08) | 1.11 | (1.05-1.18) | 0.97 | (0.91-1.04) | 1.03 | (0.97-1.09) |
10.3 - < 17.1% | 1.01 | (0.97-1.05) | 1.04 | (0.98-1.10) | 0.97 | (0.91-1.03) | 1.03 | (0.97-1.10) |
≥17.1% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent households owner occupied |
< 43% | 1.20 | (1.15-1.25) | 1.41 | (1.33-1.49) | 1.11 | (1.05-1.18) | 1.01 | (0.94-1.08) |
43 - < 63% | 1.13 | (1.08-1.18) | 1.25 | (1.18-1.33) | 1.04 | (0.98-1.11) | 1.03 | (0.98-1.09) |
63 - < 80% | 1.10 | (1.04-1.15) | 1.17 | (1.10-1.24) | 0.97 | (0.91-1.03) | 0.97 | (0.92-1.02) |
≥80% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Median owner-occupied household value |
< $44, 300 | 0.97 | (0.94-1.01) | 1.04 | (0.99-1.10) | 0.99 | (0.93-1.06) | 1.05 | (0.99-1.12) |
$44, 300 - < $57, 300 | 0.98 | (0.95-1.02) | 1.04 | (0.98-1.10) | 0.99 | (0.93-1.06) | 0.98 | (0.92-1.05) |
$57, 300 - < $77, 400 | 1.00 | (0.96-1.04) | 1.06 | (1.00-1.12) | 1.02 | (0.96-1.08) | 1.05 | (1.00-1.11) |
≥$77, 400 | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent households with interest, dividends, or net rental income |
< 8.0% | 1.08 | (1.04-1.12) | 1.23 | (1.16-1.31) | 1.17 | (1.10-1.24) | 1.03 | (0.96-1.12) |
8.0 - < 14.7% | 1.07 | (1.03-1.12) | 1.13 | (1.07-1.21) | 1.07 | (1.01-1.13) | 1.02 | (0.95-1.09) |
14.7 - < 24.1% | 1.03 | (0.98-1.07) | 1.04 | (0.98-1.11) | 1.01 | (0.95-1.07) | 1.04 | (0.99-1.09) |
≥24.1% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent Employed |
< 49.5% | 1.03 | (0.99-1.06) | 1.14 | (1.08-1.20) | 1.00 | (0.94-1.07) | 0.98 | (0.92-1.05) |
49.5% - < 56.9% | 0.99 | (0.96-1.03) | 1.10 | (1.04-1.17) | 0.96 | (0.90-1.02) | 0.99 | (0.93-1.05) |
56.9% - < 64.1% | 0.99 | (0.95-1.03) | 1.06 | (1.00-1.12) | 0.97 | (0.92-1.03) | 1.04 | (0.99-1.10) |
≥64.1% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent employed in management, professional, and related occupations |
< 15.1% | 1.05 | (1.01-1.08) | 1.24 | (1.17-1.30) | 1.02 | (0.96-1.09) | 1.04 | (0.98-1.11) |
15.1 - < 21.2% | 1.05 | (1.01-1.09) | 1.09 | (1.03-1.15) | 1.03 | (0.97-1.10) | 1.06 | (1.00-1.13) |
21.2 - < 28.3% | 1.00 | (0.96-1.04) | 1.03 | (0.98-1.10) | 0.99 | (0.93-1.06) | 0.99 | (0.93-1.05) |
≥28.3% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Neighborhood advantage summary score |
< -4.2 | 1.05 | (1.01-1.09) | 1.24 | (1.17-1.32) | 1.10 | (1.03-1.18) | 1.07 | (0.99-1.16) |
-4.2 - < -1.9 | 1.05 | (1.00-1.09) | 1.17 | (1.10-1.25) | 1.00 | (0.93-1.08) | 1.04 | (0.97-1.11) |
-1.9 - < 0.4 | 0.99 | (0.94-1.03) | 1.03 | (0.96-1.10) | 0.96 | (0.89-1.03) | 1.01 | (0.95-1.08) |
0.4 - < 3.7 | 1.01 | (0.97-1.06) | 1.05 | (0.98-1.12) | 1.05 | (0.99-1.11) | 1.06 | (1.00-1.12) |
≥3.7 | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Models examining the summary neighborhood advantage score in relation to smoking were further stratified by individual-level household income (< $25, 000/year versus > $25, 000/year) (data not shown). There was some indication that individual-level household income modified the association between smoking and neighborhood advantage score. Decreasing neighborhood advantage was associated with increased smoking mainly among those in the higher individual-level income group. PRs (95% CI) for current smoking in the lowest v. highest quintile of neighborhood advantage were 1.30 (1.13-1.50) for those with individual-level income > $25, 000/year compared to 1.02 (0.98-1.06) for those with income < $25, 000/year among black males, 1.41 (1.17-1.70) versus 1.22 (1.15-1.3) in black females, 1.19 (0.89-1.58) versus 1.10 (1.02-1.17) in white males, and 1.35 (1.000-1.81) versus 1.05 (0.97-1.14) in white females.
Table
4 shows prevalence ratios from a single robust Poisson regression model for each race and gender group that included all individual-level characteristics as well as all neighborhood-level characteristics (except for the summary z-score which was highly correlated with its individual components, and percent living in poverty and percentage in professional occupations which had high correlations,
ρ > 0.8, with other neighborhood-level measures). Individual-level measures of household income, education, employment status, and marital status were associated with current smoking in these models and the magnitude of the associations was essentially unchanged from models including only individual-level characteristics. In these models, many of the associations between neighborhood-level characteristics and smoking seen in models including each neighborhood-level characteristic individually were attenuated. Among the neighborhood-level characteristics, the lowest quartiles of percentage with a college education and percentage of owner-occupied households were each associated with increased smoking among black men and women but not whites. Unexpectedly, increasing quartiles of median household value were associated with increased risk of smoking in blacks. Percentage of households with interest, dividends, or net rental income was associated with smoking in white males only while percent employed was associated with smoking only in black females.
Table 4
Prevalence ratios for current cigarette smoking according to categories of individual-level and neighborhood-level SES characteristics from race- and sex-stratified multivariate robust Poisson regression modelsa
Individual-level variables |
Household income |
< $15, 000 | 1.67 | (1.49-1.88) | 1.68 | (1.43-1.98) | 1.60 | (1.36-1.87) | 2.04 | (1.76-2.36) |
$15, 000-$25, 000 | 1.50 | (1.33-1.68) | 1.48 | (1.26-1.73) | 1.52 | (1.29-1.78) | 1.86 | (1.60-2.15) |
$25, 000-$50, 000 | 1.31 | (1.16-1.48) | 1.24 | (1.05-1.47) | 1.35 | (1.14-1.59) | 1.45 | (1.24-1.69) |
≥$50, 000 | 1.00 | | 1.00 | | 1.00 | | 1.00 | |
Education (years) |
< 9 | 1.14 | (1.06-1.24) | 1.27 | (1.14-1.41) | 1.31 | (1.15-1.49) | 1.60 | (1.42-1.81) |
9- < 12 | 1.26 | (1.18-1.35) | 1.47 | (1.35-1.60) | 1.49 | (1.33-1.67) | 1.54 | (1.37-1.72) |
12- < 16 | 1.19 | (1.11-1.27) | 1.27 | (1.17-1.38) | 1.34 | (1.20-1.49) | 1.38 | (1.24-1.54) |
≥16 | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Marital status |
Married | 0.81 | (0.78-0.84) | 0.81 | (0.78-0.85) | 0.79 | (0.74-0.85) | 0.94 | (0.87-1.02) |
Separated/Divorced | 1.05 | (1.02-1.07) | 0.90 | (0.87-0.93) | 1.10 | (1.04-1.16) | 1.17 | (1.09-1.26) |
Widowed | 0.99 | (0.93-1.06) | 0.87 | (0.81-0.92) | 1.23 | (1.10-1.38) | 1.11 | (1.01-1.23) |
Single/Never married | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Currently working |
Yes | 0.95 | (0.92-0.97) | 0.80 | (0.77-0.83) | 0.95 | (0.90-1.00) | 0.84 | (0.80-0.89) |
No | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Neighborhood-level variables |
Household income |
< $18, 879 | 0.99 | (0.93-1.06) | 0.97 | (0.88-1.08) | 0.98 | (0.87-1.10) | 0.92 | (0.81-1.04) |
$18, 879 - < $26, 094 | 1.00 | (0.94-1.06) | 1.02 | (0.94-1.11) | 0.92 | (0.84-1.01) | 0.95 | (0.87-1.04) |
$26, 094 - < $34, 583 | 0.99 | (0.94-1.04) | 0.99 | (0.92-1.06) | 0.98 | (0.92-1.05) | 0.97 | (0.91-1.03) |
≥$34, 583 | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent adults with ≥ HS education b
|
< 57.8% | 0.97 | (0.92-1.03) | 1.02 | (0.93-1.10) | 1.00 | (0.90-1.12) | 1.02 | (0.92-1.13) |
57.8% - < 67.5% | 0.97 | (0.92-1.02) | 1.02 | (0.95-1.10) | 0.95 | (0.86-1.04) | 1.00 | (0.92-1.09) |
67.5% - < 77.3% | 0.99 | (0.95-1.04) | 1.00 | (0.94-1.07) | 1.02 | (0.95-1.09) | 1.02 | (0.96-1.09) |
≥77.3% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent adults with ≥ college graduationb
|
< 5.8% | 1.06 | (1.01-1.12) | 1.15 | (1.07-1.24) | 1.00 | (0.92-1.10) | 1.02 | (0.94-1.12) |
5.8 - < 10.3% | 1.08 | (1.03-1.13) | 1.08 | (1.01-1.16) | 0.98 | (0.90-1.07) | 1.01 | (0.94-1.10) |
10.3 - < 17.1% | 1.03 | (0.99-1.08) | 1.03 | (0.97-1.10) | 0.98 | (0.91-1.05) | 1.02 | (0.96-1.09) |
≥17.1% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent households owner occupied |
< 43% | 1.20 | (1.14-1.26) | 1.34 | (1.25-1.44) | 1.08 | (0.99-1.17) | 1.05 | (0.96-1.14) |
43 - < 63% | 1.14 | (1.08-1.19) | 1.22 | (1.15-1.31) | 1.04 | (0.97-1.12) | 1.05 | (0.99-1.12) |
63 - < 80% | 1.10 | (1.05-1.16) | 1.15 | (1.08-1.23) | 0.98 | (0.92-1.04) | 0.97 | (0.92-1.03) |
≥80% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Median owner-occupied household value |
< $44, 300 | 0.91 | (0.87-0.95) | 0.88 | (0.82-0.94) | 0.95 | (0.86-1.04) | 1.07 | (0.98-1.17) |
$44, 300 - < $57, 300 | 0.94 | (0.90-0.98) | 0.92 | (0.86-0.98) | 0.98 | (0.90-1.06) | 0.99 | (0.92-1.08) |
$57, 300 - < $77, 400 | 0.97 | (0.93-1.01) | 0.98 | (0.92-1.04) | 1.03 | (0.96-1.10) | 1.05 | (0.98-1.11) |
≥$77, 400 | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent households with interest, dividends, or net rental income |
< 8.0% | 1.04 | (0.98-1.09) | 1.05 | (0.96-1.13) | 1.20 | (1.08-1.33) | 1.02 | (0.91-1.13) |
8.0 - < 14.7% | 1.05 | (1.00-1.11) | 1.02 | (0.95-1.10) | 1.12 | (1.03-1.21) | 1.00 | (0.92-1.08) |
14.7 - < 24.1% | 1.03 | (0.98-1.08) | 0.99 | (0.93-1.06) | 1.03 | (0.97-1.09) | 1.02 | (0.97-1.08) |
≥24.1% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
Percent employed |
< 49.5% | 1.02 | (0.98-1.07) | 1.07 | (1.00-1.15) | 0.99 | (0.91-1.08) | 0.99 | (0.91-1.07) |
49.5% - < 56.9% | 1.00 | (0.95-1.04) | 1.07 | (1.00-1.15) | 0.97 | (0.90-1.04) | 0.98 | (0.91-1.05) |
56.9% - < 64.1% | 0.99 | (0.95-1.03) | 1.04 | (0.98-1.11) | 0.97 | (0.92-1.04) | 1.04 | (0.98-1.09) |
≥64.1% | 1.0 | | 1.0 | | 1.0 | | 1.0 | |
For comparison with the Black Women's Health Study (BWHS) [
8], we conducted additional analyses among black females that excluded all former smokers as was done in the BWHS report. In the SCCS, the PR (95% CI) for smoking comparing > 20% v.5% neighborhood poverty was 1.17 (1.07-1.29) and in the BWHS, the odds ratio was 1.6 (1.5-1.8).
Discussion
In this large sample of black and white adults, several measures representing decreased neighborhood advantage were associated with increased prevalence of cigarette smoking after adjustment for individual-level SES although the associations varied to some extent by race and gender. The overall associations between smoking and neighborhood-level SES in our study were consistent with those among black women enrolled in the large BWHS [
8] and the CARDIA study of young black and white adults [
5] as well as from other smaller US-based studies [
4,
7,
13]. Collectively, our findings as well as those from other studies point to an overall modest but significant negative effect of lower neighborhood-level SES on cigarette smoking after adjustment for individual-level SES measures such as education and income that are known to be associated with smoking behavior. Much speculation has been made for the potential mechanisms behind these effects and include factors related to neighborhood context (such as social norms, psychosocial stress, and exposure to tobacco advertising) as well as potential influences of individuals and their behavior on other individuals within neighborhoods, sometimes called the contagion perspective [
4,
5,
8].
Our study fills a sizeable gap in the literature by examining smoking in relation to neighborhood-level characteristics in a population of both black and white men and women over a wide age range where neighborhood poverty was common, an especially important population to study because of the high prevalence of cigarette smoking [
3]. Interventions to prevent smoking initiation and increase smoking cessation are desperately needed to reduce the morbidity and mortality associated with smoking, and the results from this study as well as others that have examined neighborhood characteristics in relation to smoking indicate that the development of interventions that target high-risk neighborhoods may be beneficial. Further, this work indicates that these interventions may be tailored to specific subgroups of race or gender that might be especially affected by aspects of the area in which they reside.
With respect to differences observed by race, individual-level household income and educational attainment were similar between black and white SCCS participants, but blacks were considerably more likely than whites to live in more disadvantaged neighborhoods with more poverty and lower percentages of highly educated and professional residents. This type of residential segregation has been described previously [
26] and indicates that further investigation is warranted into as-yet unmeasured aspects of neighborhood settings that may differentially affect smoking behavior such as racial differences in social support and cultural norms. In the SCCS population, there was a significant inverse association between neighborhood-level percentage of adults with a college education and smoking behavior among blacks but not whites. The opposite was observed in the CARDIA study [
5] and no association was seen for this measure among black females in the BWHS [
8]; these inconsistencies may be related to different distributions of individual-level and neighborhood-level education levels in the SCCS compared with other studies.
The lack of association between neighborhood SES and smoking prevalence among white women in the SCCS was noTable in this analysis. The individual measures of income and education were most strongly associated with smoking in white women, and these effects may have overwhelmed small effects of neighborhood SES in the statistical models. Unmeasured aspects of both the individual and neighborhood environment are also possible explanations for differences in white women from other groups such as stress, peer behavior, and social support for quitting.
Additionally, we found some evidence that the association between smoking and living in a disadvantaged neighborhood (as measured by the neighborhood advantage summary score) was greater in individuals with higher rather than lower individual-level household income. Diez Roux et al. observed a similar effect among blacks (combined over gender) in their analysis of young adults in the CARDIA study [
5]. These findings are contrary to the often-hypothesized notion that individuals of lower SES are more susceptible to the negative effects of living in disadvantaged neighborhoods due to increased pressure to engage in negative-health behaviors or lack of resources and positive supports. These results suggest that neighborhood pressures may be stronger in individuals of higher individual SES, but future work to determine why and how individual-level factors such as income might differentially affect the impact of neighborhood context on smoking behavior is needed.
Despite general trends indicating an inverse association between neighborhood SES and smoking behavior, many inconsistencies exist in the current literature for specific SES characteristics, particularly across race and gender lines. Some of the inconsistencies across studies may be related to the specification of the smoking measure. In our analysis, we compared current smokers to non-current smokers, a group which consisted of both former and never smokers; the same measure was used in at least two other studies [
4,
5]. Other metrics have included comparisons of ever v. never smokers and current v. former smokers [
7] and current v. never smokers after excluding former smokers [
8]. Inconsistencies across studies could also be related to the use of census tracts versus census block groups, but comparisons of effects using these two geographic entities showed little variation in at least one comparison study [
5].
A major strength of this investigation was the utilization of the SCCS cohort which includes large numbers of black and white participants of generally similar individual-level socioeconomic and geographic situation, enhancing comparability between race and gender groups. While a majority of SCCS participants are of low SES, the large sample also includes sizeable numbers of individuals of higher SES allowing for robust comparisons across the spectrum of education and income levels. Additionally, there was sufficient overlap in this study population of individuals across all categories of individual-level and neighborhood-level SES to assess these measures together. We also used robust modeling techniques which allowed for the assessment of the relative contributions of individual- and neighborhood-level characteristics as well as the estimation of more accurate standard errors than those produced using standard modeling techniques. Limitations should also be considered. First, the SCCS is not a strictly population-based sample; because of the unique recruitment of participants through southeastern CHCs and the resulting high proportion of low SES and other factors (such as high smoking prevalence), the results observed here may not be generalizable to the entire US population. However, it should be emphasized that while generalizability is a limitation, the SCCS design uniquely increases internal validity when making comparisons of effects across race groups. A second limitation of this study is that the cross-sectional nature of the data limits our ability to make temporal inferences about the association between individual and neighborhood-level characteristics and current cigarette smoking. However, as has previously been observed, current neighborhood characteristics may exert influence on smoking quitting patterns even if it did not influence its initiation [
27]. Additionally, the use of census block groups as proxies for neighborhoods requires assumptions which could not be evaluated including that census block group characteristics uniformly affect all individuals within the group and that block group boundaries adequately represent an individual's neighborhood. A final limitation related to the use of the 2000 census data is that SES characteristics within block groups may have changed over the 2002-2009 SCCS enrollment period although it should be noted that half of the cohort was enrolled by 2004 and only 14% enrolled after 2007.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
All authors conceived the study. SSC supervised the study, supervised the statistical analyses, and led the writing. MTT performed the data linkage to the US Census. JSS and MTT performed the statistical analyses. All authors contributed critical revisions for content to the manuscript. All authors read and approved the final manuscript.