Abstract
We investigated the association of individual-level ambient exposure to black carbon (spatiotemporal model-based estimate for latitude and longitude of residential address) with individual, household, and census tract socioeconomic measures among a study sample comprised of 1757 US urban working class white, black and Latino adults (age 25-64) recruited for two studies conducted in Boston, MA (2003-2004-2008-2010). Controlling for age, study, and exam date, the estimated average annual black carbon exposure for the year prior to study enrollment at the participants' residential address was directly associated with census tract poverty (beta = 0.373; 95% confidence interval (CI) 0.322, 0.423) but not with annual household income or education; null associations with race/ethnicity became significant only after controlling for socioeconomic position.
Keywords: air pollution, black carbon, poverty, race/ethnicity, socioeconomic
Introduction
Despite growing awareness of the need to integrate social epidemiologic and environmental health analyses (Morello-Frosch, 2002; Payne-Sturges et al, 2006; Brulle and Pellow, 2006; Krieger, 2011), only four studies (3 European, 1 US) have simultaneously investigated the association of contextual, household, and individual-level socioeconomic position with residential exposure to air pollution. These studies, all of urban populations, all found that exposure to air pollution -- whether nitrogen dioxide (NO2)(Chaix et al, 2006; Hajat et al, 2013), nitrogen oxides (NOx)(Goodman et al, 2011), fine particulate matter ≤ 2.5 micrometers in diameter (PM2.5) (Hajat et al, 2013), or traffic indicators (Cesaroni et al, 2011) – was more strongly associated with neighborhood-level compared to individual- and household-level measures of socioeconomic position.
We add to this limited literature by investigating the association of exposure to black carbon with individual, household, and census tract socioeconomic measures among US urban working class white, black and Latino adults. We focus on black carbon because it is a major component of traffic-related air pollution, a key contributor to urban air pollution (Gryparis et al 2007). Informed by the ecosocial theory of disease distribution and its approach to analyzing the adverse impact of multiple types of social injustice at diverse levels and spatiotemporal scales (Krieger, 2011), our a priori hypothesis was that the observed social patterning of exposure to black carbon would depend on both the level of measurement of socioeconomic position and race/ethnicity.
Materials and methods
To test our hypothesis, we linked 3 data sets, each geocoded to latitude-longitude based on exact street address of residence: two with data on the study participants, and the third with spatiotemporal data on black carbon exposure. Our investigation was approved as exempt by the Harvard School of Public Health Institutional Review Board (Protocol #23169-101), effective November 5, 2012.
Study population
The two Boston-based studies included the same socioeconomic measures. The first was the United for Health (UFH) study (2003-2004), which recruited 1202 employed working class adults, age 25-64, who worked in wholesale meat and meat production, retail grocery stores, lighting fixtures manufacturing, and school bus services; the study response rate was 72% (Barbeau et al, 2007). The second was the My Body, My Story (MBMS) study (2008-2010), comprised of a random sample of 1005 black and white non-Hispanic US-born members, age 35-64, from four Boston community health centers; the study response rate was 82% (Krieger et al, 2011). The proportion of participants geocoded to latitude-longitude based on residential street address were, respectively, 93% for UFH and 95% for MBMS. In both studies, race/ethnicity – conceptualized as a social construct arising from inequitable race relations that shape living and working conditions and hence population health (Winant, 2000; Krieger, 2012) – was measured based on self-report using pre-specified categories employed in the US census (US Census, 2013).
Socioeconomic measures
We conceptualized socioeconomic position as an inherently multidimensional construct, whose manifest dimensions (e.g., educational attainment, occupational class, and income) can each be measured at different levels (e.g., individual, household, neighborhood) and at different points in time (e.g., childhood, adulthood) (Krieger et al, 1997; Lynch and Kaplan, 2000; Shaw et al, 2007). Logically and materially consequent to social class, these manifest socioeconomic variables arise from interdependent economic relationships determined by a society's forms of property, ownership, and labor, as well as their connections through production, distribution, and consumption of goods, services, and information (Krieger et al, 1997; Shaw et al, 2007; Grusky and Szelenyi, 2011). Table 1 details the validated self-report and census tract socioeconomic measures employed (Krieger et al, 1997; Krieger et al, 2005; Krieger et al, 2006; Krieger et al, 2011; US Census, 2013).
Table 1. Sociodemographic and economic characteristics and average exposure to black carbon among study participants (N = 1757) residing in catchment area for monitoring black carbon exposure and geocoded to latitude-longitude: United for Health (Greater Boston Area, 2003-2004) and My Body, My Story (Boston, 2008-2010).
| Characteristic | United for Health (UFH) (n = 807) | My Body, My Story (MBMS) (n = 950) | |||
|---|---|---|---|---|---|
|
| |||||
| N | valuea | N | valuea | ||
| SOCIODEMOGRAPHIC CHARACTERISTICS | |||||
| Age (yrs): mean (SD) (range: UFH = 25-64; MBMS = 35-64) | 807 | 43.6 (9.7) | 950 | 48.9 (7.9) | |
| Race/ethnicity + nativity: n (%) | |||||
| White (non-Hispanic) | US-born | 134 | 16.6 | 485 | 51.1 |
| not US-born | 160 | 19.8 | |||
| nativity unknown | 8 | 1.0 | |||
| Black (non-Hispanic) | US-born | 184 | 22.8 | 465 | 48.9 |
| not US-born | 20 | 2.5 | |||
| nativity unknown | 1 | 0.1 | |||
| Latino/Hispanic | US-born | 50 | 6.2 | ||
| not US-born | 124 | 15.4 | |||
| nativity unknown | 21 | 2.6 | |||
| Additional race/ethnicities | US-born | 27 | 3.4 | ||
| not US-born | 47 | 5.8 | |||
| nativity unknown | 4 | 0.5 | |||
| (missing data: n and %) | (27) | (3.4) | |||
| Gender: n (%) | |||||
| Women | 303 | 37.5 | 626 | 65.9 | |
| Men | 489 | 60.6 | 324 | 34.1 | |
| (missing: n and %) | (15) | (1.9) | (0) | (0) | |
| Sexuality: n (%) | |||||
| Heterosexual | 642 | 80.0 | 845 | 89.0 | |
| Lesbian/gay/bisexual/transgender | 49 | 6.1 | 105 | 11.0 | |
| Other | 68 | 8.4 | 0 | 0 | |
| (missing: n and %) | (48) | (6.0) | (0) | (0) | |
| ECONOMIC CHARACTERISTICS | |||||
| CURRENT: INDIVIDUAL | |||||
| Occupational class: n (%) | |||||
| Working class: non-supervisory employee | 453 | 56.1 | 300 | 37.0 | |
| Not working class: supervisory employee | 206 | 25.5 | 141 | 17.4 | |
| self-employed/freelance | 53 | 6.6 | 67 | 8.3 | |
| own or run business | 38 | 4.7 | 47 | 5.8 | |
| Not in the paid labor force | 0 | 0.0 | 255 | 31.5 | |
| (missing: n and %) | (57) | (7.1) | (140) | (14.7) | |
| Educational attainment: n (%) | |||||
| < high school (HS)/12 years/General Education Development (GED) | 175 | 21.7 | 130 | 13.8 | |
| >= HS/GED and < 4 yrs college | 480 | 59.5 | 609 | 64.3 | |
| >= 4 yrs college | 78 | 9.7 | 207 | 21.9 | |
| (missing: n and %) | (74) | (9.2) | (4) | (0.4) | |
| CURRENT: HOUSEHOLD | |||||
| Annual household income: n (%) | |||||
| <$12,000 | 241 | 29.9 | 171 | 20.0 | |
| $12,000 to <$36,000 | 306 | 37.9 | 242 | 28.3 | |
| $36,000 to <$48,000 (US median household income in 2006b) | 67 | 8.3 | 44 | 5.2 | |
| $48,000 to <$72,000 | 53 | 6.6 | 206 | 24.0 | |
| $72,000 to <$120,000 | 19 | 2.4 | 100 | 11.7 | |
| $120,000 to <$144,000 | 10 | 1.2 | 31 | 3.6 | |
| >=$144,000 (3× US median household income in 2006b) | 27 | 3.4 | 61 | 7.1 | |
| (missing: n and %) | (84) | (10.4) | (95) | (10.0) | |
| Poverty level (household): n (%) | |||||
| Below poverty (<100% poverty line) | 333 | 41.3 | 241 | 28.3 | |
| Above poverty: 100 – 199% poverty line | 176 | 21.8 | 174 | 20.4 | |
| >=200% poverty line | 210 | 26.0 | 438 | 51.3 | |
| (missing: n and %) | (88) | (10.9) | (97) | (10.2) | |
| Household economic deprivation (occurred at least 2 times in last year), by type and number of types: n (%) | |||||
| Not enough money for food, rent, or mortgage | 183 | 22.7 | 349 | 36.8 | |
| Had to borrow money for medical expenses | 103 | 12.8 | 126 | 13.2 | |
| Not enough money to make ends meet | 198 | 24.5 | 392 | 41.2 | |
| Received public assistance or welfare | 95 | 11.8 | 349 | 36.7 | |
| Experienced 0 of these 4 types of economic deprivation | 481 | 59.6 | 386 | 40.8 | |
| Experienced 1-2 of these 4 types of economic deprivation | 194 | 24.0 | 352 | 37.1 | |
| Experienced 3-4 of these 4 types of economic deprivation | 91 | 11.3 | 210 | 22.2 | |
| (missing: n and %) | (41) | (5.1) | (0) | (0) | |
| CURRENT: CENSUS TRACT (2006-2010c) | |||||
| Census tract poverty level: n (%) | |||||
| >=20% below poverty (poverty area) | 414 | 51.30 | 426 | 44.84 | |
| 10 to 19% below poverty | 192 | 23.79 | 257 | 27.05 | |
| 5 to 9% below poverty | 140 | 17.35 | 163 | 17.16 | |
| <5% below poverty | 61 | 7.56 | 104 | 10.95 | |
| (missing: n and %) | (0) | (0) | (0) | (0) | |
| CHILDHOOD: HOUSEHOLD | |||||
| Highest educational attainment of mother, father, or guardian: n (%) | |||||
| < high school (HS)/12 years/General Education Development (GED) | 228 | 28.3 | 166 | 19.9 | |
| >= HS/GED and < 4 yrs college | 296 | 36.7 | 454 | 54.4 | |
| >= 4 yrs college | 77 | 9.5 | 215 | 25.9 | |
| (missing: n and %) | (206) | (25.5) | (115) | (12.1) | |
| EXPOSURE TO BLACK CARBON (prior to exam) | |||||
| 1-year cumulative average exposure (μg/m-3): mean (SD) | 0.68 | 0.17 | 0.64 | 0.14 | |
| 24-hour average exposure (μg/m-3): mean (SD) | 0.64 | 0.36 | 0.63 | 0.35 | |
| Day prior to exam | 0.64 | 0.39 | 0.67 | 0.34 | |
| 4 weeks prior to exam | 0.60 | 0.23 | 0.63 | 0.18 | |
| 8 weeks prior to exam | 0.60 | 0.22 | 0.63 | 0.17 | |
| 12 weeks prior to exam | 0.61 | 0.22 | 0.63 | 0.16 | |
| (missing: n and %) | (0) | (0) | (0) | (0) | |
observed percent based on participants with no missing values (percent missing separately reported)
note: 2006 is the mid-point of the years encompassed by UFH and MBMS; source of US household median income data (in current dollars): US Census Bureau, Current Population Survey (available at: http://www.census.gov/hhes/www/income/data/historical/household/; accessed: November 29, 2013)
source: US Census Bureau, American Community Survey (available at: https://www.census.gov/acs/www/; accessed: November 29, 2013)
Exposure to black carbon
We obtained the black carbon exposure from a new Boston-based spatiotemporal data set that enables precise estimation, to latitude and longitude, of time-specific ambient exposure to traffic-related air pollution, reflected by black carbon concentrations in PM2.5 (Gryparis et al, 2007). Using this model, we estimated each individual's 1-year cumulative average exposure to ambient black carbon exposure at the longitude-latitude of their residential address in the year prior to their exam; we also estimated the corresponding 24-hour average exposure for the day prior to the exam and also for the 4, 8, and 12 weeks prior to the exam.
Informing the black carbon model are data collected over the period of 1999-2008, involving over 8700 daily observations obtained from 134 sites, most of which monitored black carbon continuously using aethalometers; some sites collected particles on a filter over 24 hours and measured elemental carbon using reflectance analysis (Gryparis et al, 2007). Covariates in the prediction model included cumulative traffic density within 100 m, geographic information system (GIS) location (latitude, longitude), daily meteorological factors (apparent temperature, wind speed, and height of the planetary boundary layers), and other characteristics (day of week, day of season) (Alexeeff et al, 2011), and separate models were fit for warm and cold seasons. Exposure levels are predicted using semi-parametric models that included regression splines which allow for non-linear main effects, and thin-plate splines which measure the residual spatial variability not explained by the spatial predictors. Using this model, predicted daily concentrations showed over a 3-fold variation in exposure levels across measurement sites (adjusted R2 = 0.83), and a validation sample at an additional 30 monitoring sites found an average correlation of 0.59 between the predicted and observed black carbon levels, indicating the model is appropriate (Gryparis et al, 2007).
Analytic methods
We restricted the analytic data set to the 1757 participants (UFH: 807; MBMS: 905) with records geocoded to latitude-longitude who resided in the air monitor catchment area (Gryparis et al, 2007). We first analyzed the distribution of the included participants' sociodemographic and economic characteristics and their black carbon exposure, overall and in relation to these social characteristics. We then conducted multivariable linear regression to quantify the association between individual, household, and census tract socioeconomic measures and annual average black carbon exposure, controlling for relevant covariates.
Results and discussion
The 1757 UFH and MBMS participants included in this investigation (Table 1) were, as per the total study populations (Krieger et al, 2006; Krieger et al, 2011), predominantly working class adults who, like their parents, typically had less than a college education. Overall, 46% and 28% of the UFH and MBMS participants, respectively, lived in households below the poverty line, and ∼40% of participants lived in high poverty census tracts (≥20% below poverty) and ∼12% lived in low poverty census tracts (<5% below poverty); the risk of living in a poor household or census tract was 1.4 to 2.2 times higher among black and Latino compared to white participants. The mean 1-year cumulative average black carbon exposure (μg/m-3) at residential latitude-longitude equaled 0.68 (standard deviation (SD): 0.17) among the UFH participants and equaled 0.64 (SD 0.14) among the MBMS participants (mean difference: 0.04; 95% confidence interval (CI) 0.03, 0.05); results were similar for cumulative exposure 4, 8, and 12 weeks prior to the exam, as was the mean exposure for 24-hours prior to the exam (albeit with a greater standard deviation).
In bivariate analyses (Table 2), within each racial/ethnic group the annual average black carbon exposure at residential latitude-longitude was consistently associated with age (inversely) and census tract poverty (positively). Only among the white participants, however, was this black carbon exposure associated with education (inverse, for both the participants' and that of their parents/guardian), annual household income (inverse), and household poverty (positive); no associations existed among any racial/ethnic group for occupational class or self-reported household economic deprivation (Table 2). Inconsistent associations with black carbon exposure also existed for gender (white: higher among women compared to men; black: higher among men compared to women) and nativity (black only: higher among US- compared to foreign-born); no differences existed comparing heterosexual versus lesbian/gay/bisexual/transgender participants in any racial/ethnic group.
Table 2. Average annual black carbon exposure by sociodemographic and economic characteristics among white (non-Hispanic), black (non-Hispanic), and Latino study participants residing in catchment area for monitoring black carbon exposure and geocoded to latitude-longitude: United for Health (Greater Boston Area, 2003-2004) and My Body, My Story (Boston, 2008-2010).
| Annual average black carbon exposure (μg/m-3) | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
|||||||||||||||||||
| Variable | White (N=670: MBMS=465, UFH=205) | Black (N=787: MBMS=485, UFH=302) | Latino (N=195: MBMS=0, UFH=195) | ||||||||||||||||
|
|
|||||||||||||||||||
| N | Mean (SD) | Median | IQR | Min | Max | N | Mean (SD) | Median | IQR | Min | Max | N | Mean (SD) | Median | IQR | Min | Max | ||
| SOCIODEMOGRAPHIC CHARACTERISTICS | |||||||||||||||||||
| Age (years) | |||||||||||||||||||
| 25 – 44 yrs | 242 | 0.69 (0.16) | 0.68 | 0.21 | 0.2 | 1.24 | 300 | 0.66 (0.15) | 0.65 | 0.16 | 0.29 | 1.61 | 129 | 0.71 (0.15) | 0.71 | 0.19 | 0.29 | 1.14 | |
| 45 – 54 yrs | 258 | 0.65 (0.19) | 0.64 | 0.22 | 0.13 | 1.45 | 309 | 0.64 (0.14) | 0.62 | 0.16 | 0.23 | 1.21 | 52 | 0.67 (0.14) | 0.68 | 0.19 | 0.47 | 0.96 | |
| 55 – 65 yrs | 168 | 0.64 (0.16) | 0.62 | 0.19 | 0.19 | 1.15 | 176 | 0.63 (0.11) | 0.62 | 0.13 | 0.37 | 1.04 | 14 | 0.58 (0.15) | 0.58 | 0.29 | 0.37 | 0.85 | |
| F test (exact p-value) | 0.008 | 0.010 | 0.007 | ||||||||||||||||
| Nativity | |||||||||||||||||||
| US-born | 649 | 0.66 (0.17) | 0.65 | 0.21 | 0.13 | 1.45 | 619 | 0.66 (0.13) | 0.64 | 0.13 | 0.23 | 1.61 | 50 | 0.70 (0.14) | 0.71 | 0.16 | 0.29 | 0.96 | |
| Not US-born | 20 | 0.66 (0.16) | 0.66 | 0.27 | 0.37 | 0.94 | 160 | 0.60 (0.16) | 0.54 | 0.22 | 0.30 | 1.12 | 124 | 0.69 (0.16) | 0.70 | 0.24 | 0.37 | 1.14 | |
| F test (exact p-value) | 0.952 | <.0001 | 0.879 | ||||||||||||||||
| Gender | |||||||||||||||||||
| Women | 302 | 0.68 (0.18) | 0.68 | 0.23 | 0.13 | 1.45 | 335 | 0.63 (0.14) | 0.63 | 0.18 | 0.23 | 1.12 | 112 | 0.70 (0.16) | 0.70 | 0.23 | 0.35 | 1.14 | |
| Men | 367 | 0.64 (0.16) | 0.63 | 0.20 | 0.17 | 1.24 | 443 | 0.66 (0.14) | 0.64 | 0.14 | 0.29 | 1.61 | 80 | 0.68 (0.14) | 0.68 | 0.19 | 0.29 | 0.96 | |
| F test (exact p-value) | 0.017 | 0.001 | 0.346 | ||||||||||||||||
| Sexuality | |||||||||||||||||||
| Heterosexual | 572 | 0.66 (0.17) | 0.65 | 0.21 | 0.13 | 1.45 | 702 | 0.65 (0.14) | 0.63 | 0.15 | 0.23 | 1.61 | 135 | 0.70 (0.14) | 0.71 | 0.19 | 0.35 | 1.12 | |
| Lesbian/gay/bisexual/transgender | 84 | 0.64 (0.16) | 0.62 | 0.19 | 0.29 | 1.06 | 45 | 0.67 (0.10) | 0.66 | 0.14 | 0.43 | 0.89 | 18 | 0.67 (0.20) | 0.64 | 0.25 | 0.38 | 1.14 | |
| Other | 9 | 0.77 (0.19) | 0.74 | 0.11 | 0.53 | 1.23 | 22 | 0.65 (0.18) | 0.60 | 0.26 | 0.30 | 0.96 | 23 | 0.65 (0.16) | 0.70 | 0.27 | 0.29 | 0.92 | |
| F test (exact p-value) | 0.119 | 0.594 | 0.323 | ||||||||||||||||
| ECONOMIC CHARACTERISTICS | |||||||||||||||||||
| CURRENT: INDIVIDUAL | |||||||||||||||||||
| Occupational class | |||||||||||||||||||
| Working class: non-supervisory employee | 262 | 0.67 (0.18) | 0.66 | 0.24 | 0.13 | 1.45 | 343 | 0.64 (0.14) | 0.62 | 0.17 | 0.29 | 1.22 | 91 | 0.70 (0.15) | 0.70 | 0.23 | 0.37 | 1.14 | |
| Not working class: supervisory employee | 131 | 0.68 (0.18) | 0.67 | 0.22 | 0.30 | 1.28 | 147 | 0.66 (0.16) | 0.64 | 0.16 | 0.36 | 1.61 | 37 | 0.71 (0.15) | 0.71 | 0.21 | 0.39 | 1.12 | |
| self-employed/freelance | 52 | 0.64 (0.13) | 0.63 | 0.15 | 0.34 | 1.01 | 39 | 0.64 (0.11) | 0.65 | 0.15 | 0.47 | 1.04 | 22 | 0.64 (0.17) | 0.67 | 0.22 | 0.29 | 0.94 | |
| own or run business | 40 | 0.65 (0.16) | 0.64 | 0.18 | 0.19 | 1.01 | 29 | 0.64 (0.18) | 0.64 | 0.16 | 0.23 | 1.22 | 12 | 0.68 (0.13) | 0.70 | 0.19 | 0.49 | 0.96 | |
| Not in the paid labor force | 125 | 0.65 (0.18) | 0.63 | 0.21 | 0.19 | 1.24 | 130 | 0.66 (0.12) | 0.65 | 0.13 | 0.34 | 1.21 | 0 | -- | -- | -- | -- | -- | |
| F test (exact p-value) | 0.620 | 0.391 | 0.349 | ||||||||||||||||
| Educational attainment | |||||||||||||||||||
| < high school (HS)/12 years/GED | 85 | 0.67 (0.22) | 0.63 | 0.26 | 0.19 | 1.45 | 136 | 0.65 (0.14) | 0.64 | 0.15 | 0.34 | 1.22 | 67 | 0.70 (0.16) | 0.70 | 0.22 | 0.35 | 1.14 | |
| > = HS/GED and < 4 yrs college | 409 | 0.67 (0.17) | 0.67 | 0.20 | 0.19 | 1.22 | 536 | 0.65 (0.14) | 0.63 | 0.15 | 0.23 | 1.61 | 80 | 0.69 (0.14) | 0.71 | 0.21 | 0.29 | 0.96 | |
| >= 4 yrs college | 167 | 0.63 (0.15) | 0.61 | 0.17 | 0.13 | 1.28 | 95 | 0.65 (0.12) | 0.64 | 0.15 | 0.39 | 0.99 | 12 | 0.61 (0.13) | 0.61 | 0.19 | 0.40 | 0.83 | |
| F test (exact p-value) | 0.007 | 0.801 | 0.139 | ||||||||||||||||
| CURRENT: HOUSEHOLD | |||||||||||||||||||
| Annual household income | |||||||||||||||||||
| < $12,000 | 117 | 0.69 (0.17) | 0.69 | 0.20 | 0.23 | 1.24 | 200 | 0.67 (0.14) | 0.66 | 0.16 | 0.29 | 1.21 | 67 | 0.70 (0.17) | 0.72 | 0.23 | 0.37 | 1.14 | |
| $12,000 to <$36,000 | 201 | 0.67 (0.17) | 0.65 | 0.21 | 0.31 | 1.45 | 231 | 0.64 (0.14) | 0.63 | 0.17 | 0.34 | 1.12 | 79 | 0.69 (0.14) | 0.70 | 0.21 | 0.35 | 0.99 | |
| $36,000 to <$48,000 | 45 | 0.71 (0.16) | 0.73 | 0.24 | 0.45 | 1.21 | 50 | 0.63 (0.19) | 0.60 | 0.17 | 0.36 | 1.61 | 7 | 0.70 (0.22) | 0.73 | 0.29 | 0.29 | 0.95 | |
| $48,000 to <$72,000 | 127 | 0.63 (0.19) | 0.62 | 0.21 | 0.13 | 1.28 | 120 | 0.64 (0.12) | 0.63 | 0.12 | 0.23 | 1.22 | 4 | 0.58 (0.06) | 0.55 | 0.06 | 0.54 | 0.66 | |
| $72,000 to <$120,000 | 77 | 0.63 (0.12) | 0.62 | 0.17 | 0.34 | 0.96 | 41 | 0.64 (0.12) | 0.62 | 0.15 | 0.38 | 0.94 | 1 | 0.81 (--) | 0.81 | 0 | 0.81 | 0.81 | |
| $120,000 to <$144,000 | 27 | 0.64 (0.16) | 0.61 | 0.17 | 0.40 | 1.12 | 13 | 0.68 (0.15) | 0.63 | 0.25 | 0.49 | 0.95 | 0 | -- | -- | -- | -- | -- | |
| >= $144,000 | 35 | 0.63 (0.18) | 0.62 | 0.29 | 0.19 | 1.07 | 43 | 0.63 (0.11) | 0.61 | 0.14 | 0.37 | 0.91 | 4 | 0.65 (0.08) | 0.67 | 0.09 | 0.53 | 0.71 | |
| F test (exact p-value) | 0.024 | 0.190 | 0.648 | ||||||||||||||||
| Poverty level (household) | |||||||||||||||||||
| < 100% poverty | 152 | 0.68 (0.17) | 0.69 | 0.19 | 0.23 | 1.24 | 287 | 0.66 (0.14) | 0.65 | 0.16 | 0.29 | 1.21 | 98 | 0.69 (0.16) | 0.69 | 0.21 | 0.35 | 1.14 | |
| 100 – 199% poverty | 139 | 0.67 (0.19) | 0.65 | 0.23 | 0.19 | 1.45 | 149 | 0.64 (0.16) | 0.62 | 0.15 | 0.35 | 1.61 | 41 | 0.71 (0.15) | 0.72 | 0.16 | 0.4 | 0.99 | |
| > = 200% poverty | 336 | 0.64 (0.17) | 0.63 | 0.20 | 0.13 | 1.23 | 260 | 0.63 (0.13) | 0.63 | 0.14 | 0.23 | 1.22 | 22 | 0.63 (0.13) | 0.66 | 0.17 | 0.29 | 0.85 | |
| F test (exact p-value) | 0.034 | 0.074 | 0.191 | ||||||||||||||||
| Household economic deprivation score | |||||||||||||||||||
| 0 types of economic deprivation | 368 | 0.66 (0.18) | 0.65 | 0.21 | 0.13 | 1.45 | 347 | 0.64 (0.15) | 0.62 | 0.17 | 0.34 | 1.61 | 89 | 0.69 (0.16) | 0.71 | 0.24 | 0.29 | 1.01 | |
| 1-2 types | 181 | 0.65 (0.15) | 0.65 | 0.18 | 0.17 | 1.15 | 291 | 0.65 (0.13) | 0.64 | 0.14 | 0.23 | 1.22 | 52 | 0.70 (0.15) | 0.72 | 0.18 | 0.35 | 1.12 | |
| 3-4 types | 117 | 0.66 (0.18) | 0.63 | 0.26 | 0.29 | 1.24 | 137 | 0.67 (0.12) | 0.66 | 0.13 | 0.39 | 1.00 | 32 | 0.69 (0.16) | 0.68 | 0.18 | 0.47 | 1.14 | |
| F test (exact p-value) | 0.741 | 0.103 | 0.974 | ||||||||||||||||
| CURRENT: CENSUS TRACT | |||||||||||||||||||
| Census tract poverty level (2006-2010) | |||||||||||||||||||
| >= 20% below poverty (poverty area) | 221 | 0.73 (0.15) | 0.73 | 0.17 | 0.29 | 1.45 | 466 | 0.67 (0.13) | 0.66 | 0.15 | 0.43 | 1.21 | 90 | 0.74 (0.12) | 0.74 | 0.14 | 0.44 | 1.01 | |
| 10 to 19% below poverty | 191 | 0.67 (0.16) | 0.63 | 0.22 | 0.34 | 1.23 | 196 | 0.64 (0.15) | 0.61 | 0.13 | 0.29 | 1.61 | 45 | 0.72 (0.17) | 0.72 | 0.15 | 0.29 | 1.14 | |
| 5 to 9% below poverty | 153 | 0.61 (0.15) | 0.60 | 0.20 | 0.19 | 0.95 | 89 | 0.57 (0.13) | 0.55 | 0.16 | 0.30 | 0.98 | 44 | 0.60 (0.13) | 0.57 | 0.15 | 0.37 | 0.96 | |
| < 5% below poverty | 105 | 0.56 (0.19) | 0.55 | 0.21 | 0.13 | 1.28 | 36 | 0.52 (0.17) | 0.52 | 0.18 | 0.23 | 1.22 | 16 | 0.56 (0.10) | 0.55 | 0.16 | 0.38 | 0.70 | |
| F test (exact p-value) | <0.0001 | <.0001 | <.0001 | ||||||||||||||||
| CHILDHOOD: HOUSEHOLD | |||||||||||||||||||
| Highest level of education for mother, father, or guardian | |||||||||||||||||||
| < high school (HS)/12 years/GED | 91 | 0.68 (0.18) | 0.67 | 0.23 | 0.33 | 1.24 | 208 | 0.65 (0.14) | 0.64 | 0.18 | 0.34 | 1.22 | 76 | 0.68 (0.15) | 0.70 | 0.23 | 0.29 | 1.14 | |
| > = HS/GED and < 4 yrs college | 338 | 0.66 (0.17) | 0.66 | 0.22 | 0.19 | 1.45 | 339 | 0.65 (0.14) | 0.63 | 0.14 | 0.23 | 1.61 | 33 | 0.67 (0.16) | 0.66 | 0.24 | 0.37 | 0.96 | |
| >= 4 yrs college | 183 | 0.62 (0.15) | 0.60 | 0.18 | 0.13 | 1.06 | 92 | 0.63 (0.12) | 0.62 | 0.12 | 0.41 | 1.21 | 7 | 0.62 (0.13) | 0.66 | 0.2 | 0.44 | 0.78 | |
| F test (exact p-value) | 0.010 | 0.687 | 0.534 | ||||||||||||||||
Figure 1 illustrates the interplay between census tract poverty, race/ethnicity, and black carbon exposure, whereby symbols indicating level of exposure to black carbon by race/ethnicity are superimposed on a dot density depiction of census tract poverty. As shown by this map, among participants in the top two quintiles of exposure, the white compared to the black and Latino participants lived in different neighborhoods comprised of less impoverished census tracts.
Figure 1.
Average annual black carbon exposure (μg/m-3) by quintile for black, Latino, and white study participants (United for Health, 2003-2004; My Body, My Story, 2008-2010), and average annual census tract poverty level (2006-2010), Boston, MA air monitoring catchment area.
All 3 models for the multivariable regression analyses (Table 3) controlled for age, study, and date of exam. In Model 1, race/ethnicity was not associated with annual black carbon exposure, but significant associations (95% CI excluded 0) occurred for age (inverse), study (higher in MBMS compared to UFH), and exam date (lower exposure with more recent date); together, these variables explained little of the observed variance (R2 = 0.0474). In Model 2, which included socioeconomic but not racial/ethnic data, the R2 increased to 0.1638, but black carbon exposure was associated only census tract poverty (beta = 0.373; 95% confidence interval (CI) 0.322, 0.443) and not annual household income (beta = -0.002; 95% CI -0.006, 0.002). Finally, in Model 3, which included all variables (R2 = 0.1699), the association for census tract poverty remained unchanged (beta = 0.385; 95% CI 0.335, 0.436) and the associations for race/ethnicity became significant, whereby compared to the white participants, exposures were lower among black participants (beta = -0.024; 95% CI -0.041, -0.007) and Latino participants (beta = -0.034; 95% CI -0.061, -0.0006).
Table 3. Regression of annual black carbon exposure (μg/m-3) against economic variables and covariates among white (non-Hispanic), black (non-Hispanic), and Latino study participants residing in catchment area for monitoring black carbon exposure and geocoded to latitude-longitude: United for Health (Greater Boston Area, 2003-2004) and My Body, My Story (Boston, 2008-2010).
| Variable | Outcome: annual average black carbon exposure (μg/m-3) | |||||
|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | ||||
|
|
||||||
| beta | (95% CI) | beta | (95% CI) | beta | (95% CI) | |
| Age (year; continuous) | -0.00211 | (-0.00297, -0.00124) | -0.00173 | (-0.00253, -0.0009) | -0.00191 | (-0.00271, -0.00110) |
| Race/ethnicity | ||||||
| Black | 0.00132 | (-0.01677, 0.01940) | -0.02380 | (-0.04102, -0.00658) | ||
| Latino | -0.01792 | (-0.04683, 0.01100) | -0.03373 | (-0.06096, -0.00651) | ||
| Other | -0.02117 | (-0.06120, 0.01886) | -0.03742 | (-0.07486, 0.00003) | ||
| White (referent) | 0.0 | 0.0 | ||||
| Census tract poverty (continuous) | 0.37251 | (0.32241, 0.42261) | 0.38540 | (0.33466, 0.43615) | ||
| Household income (annual)a | -0.00201 | (-0.00634, 0.00231) | -0.00271 | (-0.00705, 0.00164) | ||
| Study | ||||||
| MBMS | 0.17899 | (0.08686, 0.27111) | 0.26845 | (0.19076, 0.34614) | 0.22823 | (0.14191, 0.31455) |
| UFH (referent) | 0.0 | 0.0 | 0.0 | |||
| Exam date (continuous) | -0.00010 | (-0.000140, -0.0000551) | -0.00014 | (-0.00017, -0.000101) | -0.000121 | (-0.000160, -0.000082) |
| R-square | 0.0474 | 0.1638 | 0.1699 | |||
Note: parameter estimates in bold have 95% CI that exclude 0
Household income categories: 1= < $12,000; 2= $12,000 to < $36,000; 3= $36,000 to <$48,000; 4= $48,000 to <$72,000; 5= $72,000 to <$120,000; 6= $120,000 to <$144,000; 7= >=$144,000
Consequently, our study offers several important contributions to the small literature (n = 4 studies) documenting that exposure to air pollution is more strongly associated with area-based versus household- or individual-level socioeconomic measures. Thus, ours is the first investigation to focus on black carbon and to diversify the range of study participants by investigating associations among US working class black, Latino, and white adults age 25-64 residing in a major US city (Boston, MA; 2003-2004 and 2008-2010). This is because the prior four investigations focused on: (1) NO2 exposure (in 2001) among children (age 7-15) in Malmö in 2001 (Chaix et al, 2006); (2) NOx exposure (in 2003) among London civil servants (age 50-74; Whitehall 2 cohort) examined in 2002-2004 (Goodman et al, 2011); (3) 2005 data on traffic indicators and a 2001 random sample (age <1 to ≥75) of the population of Rome (Cesaroni et al, 2010); and (4) NO2 and PM2.5 data (in 2000) among a population-based sample of adults age 45 to 84 recruited in 2000-2002 from 5 US cities and 1 county (Baltimore, MD, Chicago, IL; Forsyth County, NC; Los Angeles, CA; New York, NY; and St. Paul, MN) (Hajat et al, 2013). Our additional novel finding was that controlling for socioeconomic position revealed a lower on-average exposure among the black and Latino compared to white participants, which, as suggested by Figure 1, was likely due to the highly exposed white participants residing in several non-impoverished census tracts, in which none of the black and Latino participants lived.
Strengths of the study include its use of validated measures of socioeconomic position employed in two population-based studies with high response rates whose participants' residential addresses were geocoded to latitude-longitude (Krieger et al, 2006; Krieger et al, 2011) and also validated model-based spatiotemporal estimates, for latitude-longitude, of ambient black carbon exposure (Gryparis et al, 2007). Limitations include the restricted socioeconomic composition and geographic location of the study populations (Krieger et al, 2008; Krieger et al, 2013). Even so, similar results pertaining to the stronger association between air pollution and area-based compared to individual- and household-level socioeconomic measures were obtained in the one analogous US study, whose population-based sample included a higher proportion of affluent and college-educated participants (Hajat et al, 2013) compared to the UFH and MBMS participants.
In conclusion, our brief report underscores the salience of residential location, and not just individual- and household-level characteristics, for analyzing the socioeconomic patterning of exposure to air pollution and their contribution to health inequities. An additional implication is that, at least in the US context, attention to not only racial/ethnic residential segregation (Lopez, 2002; Morello-Frosch, 2002; Payne-Sturges et al, 2006; Brulle and Pellow, 2006) but also its complex interplay with residential economic segregation requires further analysis as co-determinants of exposure to air pollution.
Highlights.
The study included 1757 black, Latino, and white working class adults in Boston, MA.
Census tract poverty was associated with annual average black carbon exposure.
Annual household income was not associated with black carbon exposure.
Individual-level education was not associated with black carbon exposure.
The observed socioeconomic patterns varied by race/ethnicity.
Acknowledgments
With permission, we thank Steve Melly for assistance with the black carbon data, and Anna Kosheleva for her assistance with data management.
Sponsors: This study was supported in part by Pilot Project funding from the HSPH-NIEHS Center for Environmental Health (ES000002) and EPA R-834798, and used augmented data from studies supported, respectively, by: NIOSH 1 R01 OHO7366-01 and NIH/NIA 1 R01 AG027122. None of the sponsors had any role: in study design; in collection, analysis, and interpretation of data; in writing of the report; and in the decision to submit the article for publication.
Footnotes
Conflict of Interest: None of the authors have any conflicts of interest to declare.
Author Contributions: NK and BC designed the study and its analyses, which were implemented by PDW, using black carbon data provided by AG; NK drafted the manuscript, BC, PDW, and AG contributed to the manuscript, and all 4 authors reviewed and approved the final version prior to submission.
IRB: This study was approved as exempt [B4] by the Harvard School of Public Health Institutional Review Board (IRB) as Protocol #23169-101, effective November 5, 2012.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
Nancy Krieger, Department of Social and Behavioral Sciences Harvard School of Public Health, Kresge 717, 677 Huntington Avenue, Boston, MA 02115 (USA).
Pamela D. Waterman, Email: pwaterma@hsph.harvard.edu, Department of Social and Behavioral Sciences Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115 (USA).
Alexandros Gryparis, Email: al.grip@gmail.com, Department of Hygiene, Epidemiology and Medical Statistics, University of Athens Medical School, Athens (Greece).
Brent A. Coull, Email: bcoull@hsph.harvard.edu, Department of Biostatistics and Department of Environmental Health, 655 Huntington Avenue, Building II, Room 413, Boston, Massachusetts 02115 (USA).
References
- Alexeeff SE, Coull BA, Gryparis A, Suh H, Sparrow D, Vokonas PS, Schwartz J. Medium-term exposure to traffic-related air pollution and markers of inflammation and endothelial function. Environ Health Perspect. 2011;119:481–486. doi: 10.1289/ehp.1002560. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Barbeau EM, Hartman C, Quinn MM, Stoddard AM, Krieger N. Methods for recruiting white, black, and Hispanic working class women and men to a study of physical and social hazards at work: the United for Health Study. Int J Health Services. 2007;37:127–144. doi: 10.2190/B0N2-5850-6467-0230. [DOI] [PubMed] [Google Scholar]
- Brulle RJ, Pellow DN. Environmental justice: human health and environmental inequalities. Annu Rev Public Health. 2006;27:103–124. doi: 10.1146/annurev.publhealth.27.021405.102124. [DOI] [PubMed] [Google Scholar]
- Cesaroni G, Badaloni C, Romano V, Donato E, Perucci CA, Forastiere F. Socioeconomic position and health status of people who live near busy roads: the Rome Longitudinal Study (RoLS) Environ Health. 2010;9:41. doi: 10.1186/1476-069X-9-41. http://www.ehjournal.net/content/9/1/41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chaix B, Gustafsson S, Jerrett M, Kristersson H, Lithman T, Boalt Å, Merlo J. Children's exposure to nitrogen dioxide in Sweden: investigating environmental injustice in an egalitarian country. J Epidemiol Community Health. 2006;60:234–241. doi: 10.1136/jech.2005.038190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Goodman A, Wilkinson P, Stafford M, Tonne C. Characterising socio-economic inequalities in exposure to air pollution: a comparison of socio-economic markers and scales of measurement. Health Place. 2011;17:767–774. doi: 10.1016/j.healthplace.2011.02.002. [DOI] [PubMed] [Google Scholar]
- Grusky D, Szelenyi S, editors. The Inequality Reader: Contemporary and Foundational Readings in Race, Class, and Gender. Westview Press; Boulder, CO: 2011. [Google Scholar]
- Gryparis A, Coull BA, Schwartz J, Suh HH. Semiparametric latent variable regression models for spatiotemporal modeling of mobile source particles in the greater Boston area. Appl Statist. 2007;56:183–209. [Google Scholar]
- Hajat A, Diez-Roux AV, Adar SD, Auchincloss AH, Lovasi GS, O'Neill MS, Sheppard L, Kaufman JD. Air pollution and individual and neighborhood socioeconomic status: evidence from the Multi-Ethnic Study of Atherosclerosis (MESA) Environ Health Perspect. 2013 doi: 10.1289/ehp.1206337. Epub ahead of print (2013 Sept 27). http://dx.doi.org/10.1289/ehp.1206337. [DOI] [PMC free article] [PubMed]
- Krieger N, Chen JT, Waterman PD, Hartman C, Stoddard AM, Quinn MM, Sorensen G, Barbeau E. The inverse hazard law: blood pressure, sexual harassment, racial discrimination, workplace abuse and occupational exposures in the United for Health study of US low-income black, white, and Latino workers (Greater Boston Area, Massachusetts, United States, 2003-2004) Soc Sci Med. 2008;67:1970–1981. doi: 10.1016/j.socscimed.2008.09.039. [DOI] [PubMed] [Google Scholar]
- Krieger N, Chen JT, Waterman PD, Rehkopf DH, Subramanian SV. Painting a truer picture of US socioeconomic and racial/ethnic health inequalities: the Public Health Disparities Geocoding Project. Am J Public Health. 2005;95:312–323. doi: 10.2105/AJPH.2003.032482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krieger N, Waterman PD, Hartman C, Bates LM, Stoddard AM, Quinn MM, Sorensen G, Barbeau EM. Social hazards on the job: workplace abuse, sexual harassment, and racial discrimination -- a study of black, Latino, and white low-income women and men workers (US) Int J Health Services. 2006;36:51–85. doi: 10.2190/3EMB-YKRH-EDJ2-0H19. [DOI] [PubMed] [Google Scholar]
- Krieger N, Waterman PD, Kosheleva A, Chen JT, Carney DR, Smith KW, Bennett GG, Williams DR, Freeman E, Russell B, Thornhill G, Mikolowsky K, Rifkin R, Samuel L. Exposing racial discrimination: implicit & explicit measures—the My Body, My Story study of 1005 US-born black & white community health center members. PLoS ONE. 2011;6(11):e27636. doi: 10.1371/journal.pone.0027636. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krieger N, Waterman PD, Kosheleva A, Chen JT, Smith KS, Carney DR, Bennett G, Williams DR, Thornhill G, Freeman E. Racial discrimination & cardiovascular disease risk: My Body My Story study of 1005 US-born black and white community health center participants (US) PLoS ONE. 2013;8(10):e77174. doi: 10.1371/journal.pone.0077174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krieger N, Williams D, Moss N. Measuring social class in US public health research: concepts, methodologies and guidelines. Annu Rev Public Health. 1997;18:341–378. doi: 10.1146/annurev.publhealth.18.1.341. [DOI] [PubMed] [Google Scholar]
- Krieger N. Epidemiology and The People's Health: Theory and Context. Oxford University Press; New York: 2011. [Google Scholar]
- Krieger N. Methods for the scientific study of discrimination and health: from societal injustice to embodied inequality – an ecosocial approach. Am J Public Health. 2012;102:936–945. doi: 10.2105/AJPH.2011.300544. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lopez R. Segregation and black/white differences in exposure to air toxics 1990. Environ Health Perspectives. 2002;110(suppl 2):289–295. doi: 10.1289/ehp.02110s2289. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lynch J, Kaplan G. Socioeconomic position. In: Berkman L, Kawachi I, editors. Social Epidemiology. Oxford University Press; New York: 2000. pp. 13–35. [Google Scholar]
- Morello-Frosch RA. Discrimination and the political economy of environmental inequality. Env Planning C-Gov Policy. 2002;20:477–496. [Google Scholar]
- Payne-Sturges D, Gee GC, Crowder K, Hurley BJ, Lee C, Morello-Frosch R, Rosenbaum A, Schulz A, Wells C, Woodruff T, Zenick H. Workshop summary: connecting social and environmental factors to measure and track environmental health disparities. Environ Res. 2006;102:146–153. doi: 10.1016/j.envres.2005.11.001. [DOI] [PubMed] [Google Scholar]
- Shaw M, Galobardes B, Lawlor DA, Lynch J, Wheeler B, Davey Smith G. The Handbook of Inequality and Socioeconomic Position: Concepts and Measures. The Policy Press; Bristol, UK: 2007. [Google Scholar]
- US Census Bureau, 2013. American Community Survey, 2006-2010. [accessed: December 1 2013]; Available at: http://www.census.gov/acs/www/
- Winant H. Race and race theory. Annu Rev Sociol. 2000;26:169–185. [Google Scholar]