Abstract
Objectives. We evaluated the independent and joint effects of race, individual socioeconomic status (SES), and neighborhood SES on mortality risk.
Methods. We conducted a prospective analysis involving 52 965 non-Hispanic Black and 23 592 non-Hispanic White adults taking part in the Southern Community Cohort Study. Cox proportional hazards modeling was used to determine associations of race and SES with all-cause and cause-specific mortality.
Results. In our cohort, wherein Blacks and Whites had similar individual SES, Blacks were less likely than Whites to die during the follow-up period (hazard ratio [HR] = 0.78; 95% confidence interval [CI] = 0.73, 0.84). Low household income was a strong predictor of all-cause mortality among both Blacks and Whites (HR = 1.76; 95% CI = 1.45, 2.12). Being in the lowest (vs highest) category with respect to both individual and neighborhood SES was associated with a nearly 3-fold increase in all-cause mortality risk (HR = 2.76; 95% CI = 1.99, 3.84). There was no significant mortality-related interaction between individual SES and neighborhood SES among either Blacks or Whites.
Conclusions. SES is a strong predictor of premature mortality, and the independent associations of individual SES and neighborhood SES with mortality risk are similar for Blacks and Whites.
From birth through approximately age 85 years, there is a mortality rate disparity between Blacks and Whites in the United States that peaks in early adulthood and slowly narrows thereafter.1–4 Most of the excess deaths among Blacks occur in middle-aged adults, given the confluence of rising mortality rates and the disparity at those ages. During much of the 20th century, this disparity was unyielding,4–6 but recent data point to some narrowing of the gap beginning in the 1990s.7–9 Still, in 2011 the highest age-standardized death rate in the United States was that among non-Hispanic Blacks (877.4 per 100 000 standard population), followed by non-Hispanic Whites (738.1 per 100 000 standard population).10 Also, average life expectancies at birth in 2011 were 4.5 years shorter for Black than White men and 3.1 years shorter for Black than White women.10
Although national mortality data are routinely reported by race/ethnicity, their interpretation must consider the determinants of race-specific mortality rates, including behavioral, social, economic, and political factors that determine the resources available to maintain health and prolong life.3 Whether socioeconomic status (SES) completely accounts for mortality differences between Blacks and Whites is not clear. Previous studies have reported that SES alone cannot fully account for the disparity, although in settings where Blacks and Whites are drawn from considerably different SES strata, confounding by SES may be difficult to overcome.11–14 By contrast, in settings where race-specific SES differences are minimal (including the current study), it has been suggested that important health indicators are quite similar by race.15–17 Individual-level SES aside, neighborhood-level SES has also been reported to influence mortality rates,18 but fewer investigations have assessed the joint contribution of individual and neighborhood SES,19–22 and analyses assessing the interplay of these 2 SES domains with race are rare.19,21
We thus took the opportunity, within a large prospective study of non-Hispanic Black and White adults (residing in a large area of the United States, enrolled mainly in low-income settings but also non-low-income settings, and representing a range of SES levels), to evaluate the independent and joint contributions of race, individual SES, and neighborhood SES to overall and cause-specific mortality risk.
METHODS
The Southern Community Cohort Study (SCCS), a prospective cohort study designed to investigate racial disparities in cancer, enrolled more than 85 000 men and women across 12 southeastern states from 2002 to 2009. Comprehensive study details are provided elsewhere.23 Briefly, individuals were eligible for enrollment if they were 40 to 79 years of age, they were English speaking, and they had not been under treatment for cancer within the preceding year. Recruitment took place primarily (86%) at community health centers (CHCs), which provide health services in medically underserved, lower-income areas. The remainder of the SCCS participants (14%) enrolled through the mail in response to population-based mass mailings. Of 85 689 SCCS cohort members, 80 641 (94%) self-reported their race/ethnicity as either non-Hispanic Black or non-Hispanic White and served as the study base for this analysis.
At CHCs, trained interviewers administered a computer-assisted personal interview to collect baseline data on demographic characteristics, body size, medical history, and a wide range of lifestyle factors (e.g., diet, smoking, exercise). Participants who enrolled through the mail completed an identical survey on a scannable, self-administered form (available at http://www.southerncommunitystudy.org).
Mortality Follow-Up
The cohort was followed prospectively for mortality via linkage to the National Death Index through 2010 and the Social Security Administration’s Death Master File through February 2011. Cause of death was ascertained from the National Death Index. Use of these national mortality registries was expected to lead to minimal loss to follow-up, particularly given that Social Security numbers were available for more than 95% of the participants.24,25 Participants were followed from enrollment until their date of death or February 2011, whichever came first. In the case of those whose vital status was reported as unknown by the Social Security Administration in 2011, person-time accrued to the final known date they were confirmed alive. The average length of follow-up was 5.4 years (SD = 2.0), and the maximum was 8.9 years.
Individual and Neighborhood Socioeconomic Status
Annual household income was reported in categories of less than $15 000, $15 000 to $24 999, $25 000 to $49 999, $50 000 to $99 999, and $100 000 or more, with the 2 highest categories combined owing to small numbers. Educational attainment was classified as less than 9 years; 9 to 11 years; high school, general educational development (GED), or vocational school; some college or junior college; and college graduate or beyond. Marital status was classified as married or living as married with a partner, separated or divorced, widowed, and single or never married.
In addition, health insurance coverage was classified as none, Medicaid, Medicare (in 2 categories for those aged 65 years or older and those younger than 65 years to distinguish individuals who aged into coverage from individuals who achieved eligibility through disability at younger ages), private insurance, military insurance, and “other.” Participants also reported the type of job they held for the longest period of time during their adult life in 20 categories. We applied Nam–Powers–Boyd occupational status scores,26 which fall on a scale from 1 (lowest) to 100 (highest) and represent the socioeconomic standing of an occupation, to our occupational categories by assigning each the average of the scores for the individual job examples in that category. Finally, we considered data on household size, the number of close friends or relatives that participants reported would help with their emotional problems or feelings, and the number of close friends or relatives they could ask for help in an emergency or with lending them money.
To estimate neighborhood SES, we computed a neighborhood deprivation index (NDI) using methods described previously27; the index was based on 20 tract-level US census variables in the 7 domains of poverty, housing, occupation, employment, education, residential stability, and racial composition (Table A, available as a supplement to the online version of this article at http://www.ajph.org). The variables were obtained from 2000 US census data28 and linked to the geographical coordinates of SCCS participants’ residential addresses.29 Also, we used Federal Information Processing Standards codes to link county of residence to the 9-level 2003 rural–urban continuum code created by the US Office of Management and Budget.30 We then collapsed this 9-level measure into a dichotomous variable indicating urban (metropolitan) or rural (nonmetropolitan) residence.
We performed an initial principal components analysis involving all census tracts of the SCCS participants, overall as well as by strata of urban or rural residence. Twelve census variables with loadings greater than the root mean square of all loading values in the loading matrix in either the combined population or the rural or urban subsets were initially chosen for retention in the final analysis. One of the 12 variables was percentage of non-Hispanic Blacks, which we subsequently forced out of the analysis to avoid undue influence over Black participants’ deprivation index assignment, a decision that had negligible impact on the variance explained.
Thus, the final principal components analysis was based on 11 census tract-level variables: the percentage of individuals who had less than a high school education; the percentage of individuals who were unemployed; the percentage of men who worked in managerial jobs; the percentages of households with more than one person per room and with renter or owner costs greater than 50% of household income; the percentage of households living below the poverty line; the percentage of households that were headed by women and had dependent children present; the percentages of households that had an annual income below $30 000 per year, that were receiving public assistance, and that had no car; and the median value of owner-occupied homes. The first principal component was retained (explaining 60% of the total variance), and we determined quartiles of this tract-level measure (the NDI) and applied those values to the participant-level data. The first (lowest) quartile of the NDI represents the least deprived areas.
Statistical Analysis
Among the 80 641 SCCS participants eligible for this analysis, we excluded 4084 (5.1%) who had missing information on one or more of the variables of a priori interest. This left 76 557 participants available for our analysis (52 965 Blacks and 23 592 Whites).
We calculated race- and gender-specific crude mortality rates as the number of deaths divided by the corresponding person-time, and we used 5-year age increments to age standardize these rates according to the US 2000 standard population. We used Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause mortality and cause-specific mortality (divided into 3 groups reflecting death from cardiovascular disease [CVD], cancer, and all other nonexternal causes combined), with age used as the underlying time metric. To account for clustering of participants within census tracts and guard against biased variance estimates and confidence intervals if data within each tract were not independent, we computed robust variances based on the sandwich estimator in the Cox models.31 However, the resulting confidence intervals for all exposures of interest were minimally different from those observed with standard maximum likelihood variances, so only the latter are presented.
We constructed multivariate models considering all of the covariates shown in Table 1. Seven categories were used to model smoking (never, former smoker in tertiles of pack-year exposure, current smoker in tertiles of pack-year exposure). Occupational scores were categorized in approximate quartiles based on the entire study population distribution, with Nam–Powers–Boyd score cutoffs of 22, 37.3, and 49. We found that calendar year of enrollment, household size, and number of people participants could ask for help in an emergency or with lending them money (the latter 2 had null associations with mortality) did not affect the results, and thus we omitted these variables from our final models. In addition, adjustment for comorbidities (history of heart disease, diabetes, and hypertension) changed the hazard ratio for race by only 2.3%, the hazard ratio for the highest level of household income by 6.3%, and the hazard ratio for the lowest NDI quartile by 1.5%, and these variables were potentially in the causal pathway; thus, they were also excluded from the final models.
TABLE 1—
Baseline Characteristics and Crude and Age-Standardized Mortality Rates, by Race, Gender, and Enrollment Type: 76 557 Southern Community Cohort Study Participants, Southeastern United States, 2002–2009
| Community Health Center Women |
Community Health Center Men |
General Population Women |
General Population Men |
|||||
| Black (n = 28 639) | White (n = 12 092) | Black (n = 20 577) | White (n = 6476) | Black (n = 2186) | White (n = 2408) | Black (n = 1563) | White (n = 2616) | |
| No. of deaths | 1749 | 813 | 2289 | 809 | 78 | 93 | 111 | 172 |
| Crude mortality rate (per 100 000 person-years) | 1056.0 | 1380.3 | 1983.0 | 2701.9 | 662.4 | 725.5 | 1330.9 | 1241.7 |
| Age standardized mortality ratea (per 100 000 person-years) | 1373.9 | 2126.6 | 2848.0 | 3762.1 | 599.9 | 667.1 | 1330.7 | 1073.9 |
| Age at enrollment, y, no. (%) | ||||||||
| 40–49 | 13 904 (48.6) | 4835 (40.0) | 10 928 (53.1) | 2971 (45.9) | 741 (33.9) | 675 (28.0) | 410 (26.2) | 502 (19.2) |
| 50–59 | 9458 (33.0) | 4127 (34.1) | 6877 (33.4) | 2060 (31.8) | 987 (45.2) | 1047 (43.5) | 701 (44.9) | 1056 (40.4) |
| 60–69 | 3885 (13.6) | 2358 (19.5) | 2176 (10.6) | 1069 (16.5) | 395 (18.1) | 589 (24.5) | 381 (24.4) | 885 (33.8) |
| 70–79 | 1392 (4.9) | 772 (6.4) | 596 (2.9) | 376 (5.8) | 63 (2.9) | 97 (4.0) | 71 (4.5) | 173 (6.6) |
| Annual household income, $, no. (%) | ||||||||
| < 15 000 | 17 805 (62.2) | 6851 (56.7) | 13 083 (63.6) | 3742 (57.8) | 627 (28.7) | 408 (16.9) | 301 (19.3) | 165 (6.3) |
| 15 000–24 999 | 6624 (23.1) | 2470 (20.4) | 4474 (21.7) | 1358 (21.0) | 454 (20.8) | 364 (15.1) | 269 (17.2) | 215 (8.2) |
| 25 000–49 999 | 3220 (11.2) | 1673 (13.8) | 2290 (11.1) | 894 (13.8) | 641 (29.3) | 658 (27.3) | 443 (28.3) | 666 (25.5) |
| ≥ 50 000 | 990 (3.5) | 1098 (9.1) | 730 (3.6) | 482 (7.4) | 464 (21.2) | 978 (40.6) | 550 (35.2) | 1570 (60.0) |
| Education, no. (%) | ||||||||
| < 9 y | 2212 (7.7) | 1067 (8.8) | 1900 (9.2) | 743 (11.5) | 44 (2.0) | 48 (2.0) | 52 (3.3) | 49 1.9) |
| 9–11 y | 6881 (24.0) | 2372 (19.6) | 5354 (26.0) | 1239 (19.1) | 174 (8.0) | 101 (4.2) | 148 (9.5) | 69 (2.6) |
| High school/GED/vocational | 11 289 (39.4) | 4913 (40.6) | 8624 (41.9) | 2575 (39.8) | 675 (30.9) | 725 (30.1) | 483 (30.9) | 616 (23.6) |
| Some college/junior college | 5720 (20.0) | 2389 (19.8) | 3363 (16.3) | 1213 (18.7) | 562 (25.7) | 578 (24.0) | 398 (25.5) | 536 (20.5) |
| College | 2537 (8.9) | 1351 (11.2) | 1336 (6.5) | 706 (10.9) | 731 (33.4) | 956 (39.7) | 482 (30.8) | 1346 (51.5) |
| Marital status, no. (%) | ||||||||
| Married/living as married | 7526 (26.3) | 5206 (43.1) | 5899 (28.7) | 2634 (40.7) | 720 (32.9) | 1302 (54.1) | 1035 (66.2) | 2168 (82.9) |
| Divorced | 9864 (34.4) | 4278 (35.4) | 7092 (34.5) | 2385 (36.8) | 817 (37.4) | 659 (27.4) | 329 (21.1) | 272 (10.4) |
| Widowed | 4044 (14.1) | 1650 (13.7) | 765 (3.7) | 269 (4.2) | 253 (11.6) | 273 (11.3) | 40 (2.6) | 53 (2.0) |
| Single, never married | 7205 (25.2) | 958 (7.9) | 6821 (33.2) | 1188 (18.3) | 396 (18.1) | 174 (7.2) | 159 (10.2) | 123 (4.7) |
| No. of household members, no. (%) | ||||||||
| 1 | 6705 (23.4) | 3139 (26.0) | 6298 (30.6) | 2149 (33.2) | 573 (26.2) | 656 (27.2) | 295 (18.9) | 319 (12.2) |
| 2 | 8702 (30.4) | 4723 (39.1) | 5965 (29.0) | 2333 (36.0) | 706 (32.3) | 1055 (43.8) | 611 (39.1) | 1506 (57.6) |
| 3–4 | 9478 (33.1) | 3277 (27.1) | 6081 (29.6) | 1532 (23.7) | 696 (31.8) | 580 (24.1) | 503 (32.2) | 656 (25.1) |
| 5–6 | 2888 (10.1) | 751 (6.2) | 1717 (8.3) | 361 (5.6) | 159 (7.3) | 101 (4.2) | 127 (8.1) | 120 (4.6) |
| > 6 | 866 (3.0) | 202 (1.7) | 516 (2.5) | 101 (1.6) | 52 (2.4) | 16 (0.7) | 27 (1.7) | 15 (0.6) |
| Occupational status score, no. (%) | ||||||||
| Quartile 1 (lowest status) | 5993 (22.2) | 2717 (23.4) | 7583 (39.0) | 2171 (35.4) | 288 (14.4) | 396 (17.4) | 298 (19.9) | 245 (9.8) |
| Quartile 2 | 8683 (32.1) | 2667 (22.9) | 5298 (27.3) | 1433 (23.3) | 354 (17.7) | 167 (7.3) | 272 (18.2) | 274 (10.9) |
| Quartile 3 | 5085 (18.8) | 3206 (27.6) | 4563 (23.5) | 1562 (25.4) | 444 (22.2) | 526 (23.1) | 376 (25.1) | 499 (19.9) |
| Quartile 4 (highest status) | 7301 (27.0) | 3039 (26.1) | 1988 (10.2) | 976 (15.9) | 911 (45.6) | 1185 (52.1) | 552 (36.9) | 1493 (59.5) |
| Neighborhood deprivation index, no. (%) | ||||||||
| Quartile 1 (least deprived) | 1475 (5.2) | 2052 (17.0) | 1008 (4.9) | 1065 (16.5) | 255 (11.7) | 775 (32.2) | 232 (14.9) | 980 (37.5) |
| Quartile 2 | 2942 (10.3) | 3596 (29.7) | 1742 (8.5) | 1673 (25.9) | 338 (15.5) | 716 (29.7) | 278 (17.8) | 729 (27.9) |
| Quartile 3 | 5096 (17.8) | 3768 (31.2) | 3089 (15.0) | 1827 (28.3) | 548 (25.1) | 625 (26.0) | 385 (24.7) | 602 (23.0) |
| Quartile 4 (most deprived) | 19 124 (66.8) | 2675 (22.1) | 14 726 (71.6) | 1901 (29.4) | 1045 (47.8) | 292 (12.1) | 667 (42.7) | 305 (11.7) |
| No. of people to help with emotional problems,b no. (%) | ||||||||
| 0 | 2316 (8.1) | 886 (7.3) | 1964 (9.6) | 805 (12.5) | 148 (7.0) | 113 (4.8) | 142 (9.6) | 156 (6.1) |
| 1–2 | 8668 (30.4) | 3435 (28.5) | 5950 (29.0) | 2015 (31.2) | 489 (23.2) | 424 (17.8) | 308 (20.8) | 495 (19.4) |
| 3–4 | 6807 (23.8) | 2903 (24.1) | 4545 (22.2) | 1411 (21.8) | 490 (23.2) | 538 (22.6) | 282 (19.0) | 612 (24.0) |
| ≥ 5 | 10 765 (37.7) | 4842 (40.1) | 8047 (39.2) | 2229 (34.5) | 982 (46.6) | 1303 (54.8) | 752 (50.7) | 1284 (50.4) |
| Missing | 83 | 26 | 71 | 16 | 77 | 30 | 79 | 69 |
| No. of people to help with emergencies/lending money,c no. (%) | ||||||||
| 0 | 3075 (10.8) | 1273 (10.6) | 2335 (11.4) | 875 (13.6) | 233 (10.9) | 208 (8.8) | 140 (9.5) | 203 (8.0) |
| 1 | 5205 (18.2) | 2212 (18.4) | 3392 (16.5) | 1172 (18.2) | 270 (12.7) | 232 (9.8) | 164 (11.1) | 243 (9.6) |
| 2 | 6287 (22.0) | 2731 (22.7) | 4152 (20.2) | 1335 (20.7) | 482 (22.6) | 447 (18.9) | 267 (18.1) | 415 (16.3) |
| 3 | 3999 (14.0) | 1552 (12.9) | 2751 (13.4) | 791 (12.3) | 305 (14.3) | 319 (13.5) | 190 (12.9) | 261 (10.3) |
| 4 | 2648 (9.3) | 1088 (9.0) | 1815 (8.9) | 517 (8.0) | 192 (9.0) | 244 (10.3) | 139 (9.4) | 267 (10.5) |
| ≥ 5 | 7323 (25.7) | 3195 (26.5) | 6073 (29.6) | 1765 (27.3) | 647 (30.4) | 920 (38.8) | 577 (39.1) | 1151 (45.3) |
| Missing | 102 | 41 | 59 | 21 | 57 | 38 | 86 | 76 |
| Health insurance coverage, no. (%) | ||||||||
| None | 11 229 (39.2) | 5031 (41.6) | 10 231 (49.8) | 3119 (48.2) | 388 (18.4) | 312 (13.2) | 234 (15.8) | 162 (6.3) |
| Medicaid | 3432 (12.0) | 1489 (12.3) | 2539 (12.4) | 1019 (15.7) | 255 (12.1) | 190 (8.1) | 187 (12.6) | 167 (6.5) |
| Medicare < age 65 y | 6251 (21.8) | 1920 (15.9) | 2963 (14.4) | 814 (12.6) | 187 (8.9) | 116 (4.9) | 77 (5.2) | 77 (3.0) |
| Medicare ≥ age 65 y | 1367 (4.8) | 1114 (9.2) | 596 (2.9) | 527 (8.1) | 109 (5.2) | 226 (9.6) | 134 (9.0) | 427 (16.6) |
| Private | 5900 (20.6) | 2341 (19.4) | 3234 (15.7) | 713 (11.0) | 1064 (50.5) | 1406 (59.6) | 710 (47.8) | 1580 (61.4) |
| Military | 144 (0.5) | 91 (0.8) | 732 (3.6) | 231 (3.6) | 45 (2.1) | 56 (2.4) | 98 (6.6) | 121 (4.7) |
| Other | 303 (1.1) | 101 (0.8) | 263 (1.3) | 51 (0.8) | 60 (2.9) | 55 (2.3) | 45 (3.0) | 41 (1.6) |
| Missing | 13 | 5 | 19 | 2 | 78 | 47 | 78 | 41 |
| Baseline smoking status, no. (%) | ||||||||
| Current | 9714 (34.0) | 4768 (39.5) | 12 314 (59.9) | 3418 (52.8) | 422 (20.2) | 495 (20.9) | 388 (26.1) | 403 (15.8) |
| Former | 5433 (19.0) | 2957 (24.5) | 3914 (19.0) | 1798 (27.8) | 547 (26.2) | 722 (30.5) | 590 (39.7) | 1181 (46.3) |
| Never | 13 466 (47.1) | 4359 (36.1) | 4336 (21.1) | 1257 (19.4) | 1122 (53.7) | 1148 (48.5) | 507 (34.1) | 968 (37.9) |
| Missing | 26 | 8 | 13 | 3 | 95 | 43 | 78 | 64 |
| Baseline health history, no. (%) | ||||||||
| Diabetes | 7143 (25.0) | 2626 (21.7) | 3623 (17.6) | 1261 (19.5) | 473 (21.6) | 324 (13.5) | 323 (20.7) | 388 (14.8) |
| Heart attack/bypass surgery | 1478 (5.2) | 908 (7.5) | 1303 (6.3) | 951 (14.7) | 97 (4.6) | 105 (4.5) | 107 (7.2) | 293 (11.6) |
| Hypertension | 17 902 (62.5) | 6207 (51.4) | 10 168 (49.5) | 3262 (50.4) | 1292 (59.1) | 926 (38.5) | 882 (56.4) | 1120 (42.8) |
| BMI at enrollment (kg/m2), mean (SD) | 32.5 (8.0) | 31.1 (8.2) | 27.6 (6.0) | 28.8 (6.7) | 32.8 (8.1) | 28.8 (7.4) | 28.9 (5.9) | 28.4 (5.4) |
| BMI at age 21 y (kg/m2), mean (SD) | 23.1 (5.3) | 22.7 (5.7) | 23.8 (4.5) | 24.0 (4.8) | 22.9 (5.1) | 22.1 (4.9) | 23.5 (3.8) | 23.6 (3.7) |
| Physical activity expenditure (MET hours per day), mean (SD) | 21.0 (15.9) | 20.7 (16.1) | 26.4 (23.5) | 23.7 (22.6) | 21.3 (15.8) | 21.5 (15.3) | 24.7 (21.3) | 22.0 (16.3) |
| No. of sedentary h/d, mean (SD) | 9.4 (5.1) | 8.7 (4.5) | 9.3 (5.3) | 8.8 (4.9) | 11.6 (5.8) | 9.7 (4.5) | 10.7 (5.4) | 10.0 (4.8) |
Note. BMI = body mass index; GED = general educational development (high school equivalent); MET = metabolic equivalent.
Age standardized according to the US 2000 standard population (5-year age increments).
Participants were asked the following question: “How many close friends or relatives would help you with your emotional problems or feelings if you needed it?”
Participants were asked the following question: “How many people could you ask for help in an emergency or with lending you money?”
We examined models excluding the first 1, 2, and 3 years of follow-up to identify evidence of reverse causation with regard to SES (i.e., that underlying fatal disease resulted in poverty). We verified the proportionality assumptions of the Cox models by dividing the period of follow-up into the first 4 and subsequent years and comparing hazard ratios for the 2 time periods. As a means of comparing models with and without the relevant interaction terms, we used the likelihood ratio test to assess interactions. All P values are 2-tailed, and SAS/STAT version 9.3 (SAS Institute Inc, Cary, NC) was used in conducting the statistical analyses.
RESULTS
A total of 6114 deaths occurred during a follow-up period representing 416 677 person-years, for a crude mortality rate of 1467 per 100 000 person-years. The gender- and race-specific mortality rates for CHC-enrolled participants were markedly higher than those for participants who enrolled in the SCCS through the mail, reflecting the different demographic characteristics of the 2 study bases (Table 1). Among CHC-enrolled participants, household income was similarly distributed across race and gender strata, suggesting only minor racial differences in individual SES; despite these similarities, Blacks tended to reside in communities with higher NDI values. Among the mail-enrolled participants with income distributions more reflective of a general volunteer sample, Blacks were also more likely (relative to their individual SES level) than Whites to reside in areas of high deprivation.
Overall, Blacks were less likely than Whites to die during the follow-up period (HR = 0.78; 95% CI = 0.73, 0.84; Table 2). The absence of excess mortality among Blacks was evident at each level of household income, within each quartile of neighborhood deprivation, and among those enrolled from the general population and from CHCs alike (Table B, available as a supplement to the online version of this article at http://www.ajph.org). Associations between individual- and neighborhood-level SES factors and all-cause mortality are shown in Table 2. These results were not modified by race, gender, or source of participant enrollment (Table 2).
TABLE 2—
Multivariate Hazard Ratios for Associations Between Race, Socioeconomic Status Indicators, and All-Cause Mortality: Southern Community Cohort Study, Southeastern United States, 2002–2009
| Model 1, Adjusted HRa (95% CI) | Model 2, Adjusted HRb (95% CI) | |
| Racec | ||
| Black | 0.84 (0.79, 0.89) | 0.78 (0.73, 0.84) |
| White (Ref) | 1.00 | 1.00 |
| Annual household income, $d | ||
| < 15 000 | 3.79 (3.26, 4.41) | 1.76 (1.45, 2.12) |
| 15 000–24 999 | 2.46 (2.10, 2.89) | 1.48 (1.22, 1.78) |
| 25 000–49 999 | 1.61 (1.36, 1.90) | 1.20 (0.99, 1.45) |
| ≥ 50 000 (Ref) | 1.00 | 1.00 |
| Neighborhood deprivation indexe | ||
| Quartile 4 (most deprived) | 1.73 (1.55, 1.92) | 1.26 (1.12, 1.42) |
| Quartile 3 | 1.42 (1.27, 1.59) | 1.19 (1.06, 1.35) |
| Quartile 2 | 1.34 (1.19, 1.51) | 1.17 (1.03, 1.33) |
| Quartile 1 (least deprived) (Ref) | 1.00 | 1.00 |
| Education | ||
| < 9 y | 1.20 (1.10, 1.30) | 0.97 (0.88, 1.07) |
| 9–11 y | 1.22 (1.15, 1.30) | 1.06 (0.98, 1.13) |
| High school/GED/vocational (Ref) | 1.00 | 1.00 |
| Some college/junior college | 0.91 (0.85, 0.99) | 0.98 (0.91, 1.07) |
| College | 0.63 (0.57, 0.70) | 0.95 (0.84, 1.07) |
| Marital status | ||
| Married/living as married (Ref) | 1.00 | 1.00 |
| Separated/divorced | 1.50 (1.41, 1.60) | 1.16 (1.07, 1.24) |
| Widowed | 1.53 (1.39, 1.67) | 1.18 (1.07, 1.30) |
| Single, never married | 1.67 (1.55, 1.80) | 1.21 (1.11, 1.32) |
| NPB occupational status score | ||
| Quartile 4 (highest) (Ref) | 1.00 | 1.00 |
| Quartile 3 | 1.20 (1.10, 1.30) | 0.99 (0.90, 1.08) |
| Quartile 2 | 1.27 (1.17, 1.38) | 0.98 (0.89, 1.07) |
| Quartile 1 (lowest) | 1.39 (1.28, 1.51) | 1.01 (0.93, 1.11) |
| No. of close friendsf | ||
| 0 (Ref) | 1.00 | 1.00 |
| 1–2 | 0.87 (0.79, 0.95) | 0.96 (0.87, 1.06) |
| ≥ 3 | 0.71 (0.66, 0.77) | 0.88 (0.81, 0.97) |
| Health insurance coverage | ||
| Private (Ref) | 1.00 | 1.00 |
| None | 2.12 (1.92, 2.34) | 1.38 (1.24, 1.55) |
| Medicare < age 65 y | 3.55 (3.20, 3.94) | 2.17 (1.92, 2.45) |
| Medicaid | 3.29 (2.96, 3.65) | 1.90 (1.68, 2.14) |
| Medicare ≥ age 65 y | 1.90 (1.64, 2.18) | 1.36 (1.16, 1.59) |
| Military | 2.29 (1.91, 2.76) | 1.72 (1.41, 2.10) |
| Other | 1.83 (1.40, 2.40) | 1.38 (1.03, 1.86) |
Note. CI = confidence interval; GED = general educational development (high school equivalent); HR = hazard ratio; NPB = Nam–Powers–Boyd.
Hazard ratio from a Cox proportional hazards model adjusting for gender, race, and enrollment type (community health center or general population).
Hazard ratio from a Cox proportional hazards model adjusting for gender, race, and enrollment type (community health center or general population), plus all of the covariates shown in Table 2, smoking, body mass index (BMI), BMI at age 21 years, physical activity (in metabolic equivalent hours per day), and sedentary time (number of hours sitting per day).
For race, there were no statistically significant interactions with enrollment source (interaction P = .92), household income (interaction P = .31), or neighborhood deprivation (interaction P = .65).
For household income, there were no statistically significant interactions with enrollment source (interaction P = .13), race (interaction P = .31), or neighborhood deprivation (interaction P = .6).
For neighborhood deprivation, there were no statistically significant interactions with enrollment source (interaction P = .8), race (interaction P = .65), or household income (interaction P = .6).
Participants were asked the following question: “How many close friends or relatives would help you with your emotional problems or feelings if you needed it?”
Household income was a strong predictor of mortality (as shown in the survival curve in Figure A, available as a supplement to the online version of this article at http://www.ajph.org). Those in the lowest income group (< $15 000) were about 75% as likely to die during the follow-up period as those with incomes of $50 000 or more, an association that was even larger among those who had never smoked (HR = 2.52; 95% CI = 1.75, 3.63). After adjustment for household income, education and occupational status were not significantly associated with mortality. Independent of household income, however, there was a significant (P < .001) trend of increasing mortality risk with increasing NDI quartile. We also observed a significantly elevated mortality risk among individuals who had never married and who had any type of health insurance other than private insurance; individuals who had 3 or more close friends exhibited a significantly reduced mortality risk.
The combined effect of being in the least (vs most) advantageous category with respect to both individual and neighborhood SES was a striking increase in mortality risk (HR = 2.76; 95% CI = 1.99, 3.84; Table 3), an effect that was similar for Blacks and Whites (data not shown). We detected no significant interaction between individual SES and neighborhood SES, with the effect of one being similar across categories of the other for both Blacks (interaction P = .98) and Whites (interaction P = .58).
TABLE 3—
Hazard Ratios for the Joint Effects of Household Income and Neighborhood Deprivation on All-Cause Mortality: Southern Community Cohort Study, Southeastern United States, 2002–2009
| Neighborhood Deprivation Index |
||||
| Annual Household Income, $ | Quartile 4 (Most Deprived), HRa (95% CI) | Quartile 3, HRa (95% CI) | Quartile 2, HRa (95% CI) | Quartile 1 (Least Deprived), HRa (95% CI) |
| < 15 000 | 2.76 (1.99, 3.84) | 2.68 (1.92, 3.74) | 2.54 (1.81, 3.56) | 2.40 (1.69, 3.42) |
| 15 000–24 999 | 2.37 (1.69, 3.31) | 2.16 (1.52, 3.06) | 2.18 (1.53, 3.12) | 1.79 (1.20, 2.68) |
| 25 000–49 999 | 1.93 (1.36, 2.74) | 1.76 (1.21, 2.56) | 1.79 (1.23, 2.62) | 1.41 (0.93, 2.13) |
| ≥ 50 000 | 1.86 (1.22, 2.82) | 1.31 (0.82, 2.10) | 1.69 (1.13, 2.55) | 1.00 (Ref) |
Note. CI = confidence interval; HR = hazard ratio.
Hazard ratio for combinations of household income and neighborhood deprivation derived from a single Cox proportional hazards model adjusting for race, gender, education, marital status, occupational status, number of close friends, health insurance type, enrollment type, smoking, body mass index (BMI), BMI at age 21 years, physical activity (in metabolic equivalent hours per day), and sedentary time (number of hours sitting per day).
The adverse impact of low individual SES was observed for mortality from CVD and other nonmalignant diseases but not for cancer mortality (Table 4). By contrast, the highest NDI quartile was associated with an approximate 30% increase in mortality risk regardless of cause. Relative to Whites, Blacks had an equivalent risk of cancer mortality but a lower risk of mortality from CVD and other nonmalignant diseases during follow-up.
TABLE 4—
Multivariate Hazard Ratios for Associations Between Individual- and Neighborhood-Level Socioeconomic Status and Cause-Specific Mortality: Southern Community Cohort Study, Southeastern United States, 2002–2009
| Cardiovascular Disease Mortality (n = 1869), HRa (95% CI) | Cancer Mortality (n = 1435), HRa (95% CI) | Mortality From All Other Nonexternal Causes (n = 2076), HRa (95% CI) | |
| Race | |||
| Black | 0.88 (0.78, 1.00) | 1.00 (0.87, 1.16) | 0.70 (0.62, 0.78) |
| White (Ref) | 1.00 | 1.00 | 1.00 |
| Annual household income, $ | |||
| < 15 000 | 2.08 (1.45, 2.98) | 1.08 (0.78, 1.50) | 2.38 (1.62, 3.49) |
| 15 000–24 999 | 1.87 (1.30, 2.68) | 1.00 (0.72, 1.38) | 1.79 (1.22, 2.64) |
| 25 000–49 999 | 1.53 (1.06, 2.20) | 0.80 (0.57, 1.11) | 1.43 (0.96, 2.12) |
| ≥ 50 000 (Ref) | 1.00 | 1.00 | 1.00 |
| Education | |||
| < 9 y | 0.99 (0.84, 1.18) | 1.01 (0.82, 1.22) | 1.00 (0.85, 1.18) |
| 9–11 y | 1.03 (0.90, 1.17) | 1.09 (0.94, 1.27) | 1.10 (0.97, 1.23) |
| High school/GED/vocational (Ref) | 1.00 | 1.00 | 1.00 |
| Some college/junior college | 0.99 (0.85, 1.14) | 1.04 (0.88, 1.24) | 0.96 (0.84, 1.11) |
| College | 0.82 (0.66, 1.03) | 1.02 (0.80, 1.31) | 0.89 (0.71, 1.10) |
| Marital status | |||
| Married/living as married (Ref) | 1.00 | 1.00 | 1.00 |
| Separated/divorced | 1.09 (0.96, 1.24) | 1.02 (0.87, 1.18) | 1.26 (1.11, 1.43) |
| Widowed | 1.03 (0.87, 1.23) | 1.18 (0.98, 1.43) | 1.29 (1.08, 1.53) |
| Single, never married | 1.13 (0.97, 1.31) | 1.12 (0.93, 1.35) | 1.36 (1.18, 1.58) |
| NPB occupational status score | |||
| Quartile 4 (highest) (Ref) | 1.00 | 1.00 | 1.00 |
| Quartile 3 | 0.91 (0.77, 1.07) | 1.17 (0.97, 1.41) | 0.89 (0.76, 1.05) |
| Quartile 2 | 0.89 (0.76, 1.04) | 1.12 (0.93, 1.36) | 0.93 (0.80, 1.09) |
| Quartile 1 (lowest) | 0.88 (0.75, 1.04) | 1.18 (0.97, 1.43) | 1.00 (0.86, 1.17) |
| No. of close friendsb | |||
| 0 (Ref) | 1.00 | 1.00 | 1.00 |
| 1–2 | 0.95 (0.80, 1.14) | 1.08 (0.86, 1.36) | 0.96 (0.81, 1.13) |
| ≥ 3 | 0.81 (0.69, 0.96) | 1.14 (0.92, 1.41) | 0.87 (0.75, 1.02) |
| Health insurance coverage | |||
| Private (Ref) | 1.00 | 1.00 | 1.00 |
| None | 1.57 (1.27, 1.93) | 1.35 (1.09, 1.68) | 1.37 (1.11, 1.69) |
| Medicare < age 65 y | 2.43 (1.94, 3.04) | 1.49 (1.16, 1.91) | 2.75 (2.21, 3.44) |
| Medicaid | 2.08 (1.66, 2.61) | 1.40 (1.10, 1.78) | 2.35 (1.88, 2.93) |
| Medicare ≥ age 65 y | 1.64 (1.24, 2.17) | 1.10 (0.82, 1.47) | 1.46 (1.08, 1.96) |
| Military | 1.83 (1.26, 2.65) | 1.48 (1.00, 2.20) | 2.03 (1.42, 2.91) |
| Other | 1.66 (0.98, 2.80) | 0.94 (0.48, 1.85) | 1.70 (1.03, 2.80) |
| Neighborhood deprivation index | |||
| Quartile 4 (most deprived) | 1.28 (1.03, 1.58) | 1.29 (1.01, 1.64) | 1.31 (1.06, 1.61) |
| Quartile 3 | 1.23 (0.98, 1.54) | 1.32 (1.02, 1.69) | 1.20 (0.96, 1.49) |
| Quartile 2 | 1.19 (0.94, 1.51) | 1.15 (0.88, 1.50) | 1.20 (0.95, 1.50) |
| Quartile 1 (least deprived) (Ref) | 1.00 | 1.00 | 1.00 |
Note. HR = hazard ratio; CI = confidence interval; GED = general educational development (high school equivalent); NPB = Nam–Powers–Boyd. Cardiovascular disease (CVD) death was defined as International Classification of Diseases (9th edition; ICD-9) codes 390–459 and 798 and ICD-10 codes I00–I9.32,33 Cancer death was defined as ICD-9 codes 140–239 and ICD-10 codes C00–C97. All other nonexternal causes of death were non-CVD and noncancer deaths that excluded external causes.
Hazard ratios derived from 3 separate Cox proportional hazards models (one for each cause of death) adjusting for all covariates shown in Table 2 in addition to gender, race, enrollment type (community health center or general population), smoking, body mass index (BMI), BMI at age 21 years, physical activity (in metabolic equivalent hours per day), and sedentary time (number of hours sitting per day).
Participants were asked the following question: “How many close friends or relatives would help you with your emotional problems or feelings if you needed it?”
Results excluding the first 1, 2, and 3 years of follow-up were similar to those presented in the tables. For example, after exclusion of the first 3 years, the hazard ratio for low income (< $15 000) was 1.66 (95% CI = 1.28, 2.15), the hazard ratio for NDI quartile 4 was 1.25 (95% CI = 1.07, 1.47), and the hazard ratio for Black race was 0.77 (95% CI = 0.70, 0.85). Also, among participants who were relatively healthy at baseline (i.e., no self-reported history of myocardial infarction, coronary artery bypass surgery, stroke, diabetes, or cancer), household income (< $15 000 vs ≥ $50 000; HR = 1.55; 95% CI = 1.19, 2.02) and NDI (quartile 4 vs quartile 1; HR = 1.30; 95% CI = 1.10, 1.54) were still significantly associated with mortality.
DISCUSSION
This is among the first and largest investigations to examine race, individual SES, and neighborhood SES jointly as determinants of mortality. We did not find evidence that individual and neighborhood SES acted synergistically; rather, both played an independent role in predicting mortality, with a particularly strong link between household income and longevity. We did not observe a higher mortality risk among Blacks (vs Whites) in any stratum of SES, and in fact we observed an overall better mortality profile for Blacks. This was apparent in both segments of SCCS participants (i.e., those drawn from CHCs and those drawn from the general population). We had expected mortality parity for Blacks and Whites among the CHC-enrolled participants (with these groups having similar individual SES and arguably similar access to medical care) and would not have been surprised to also find mortality parity in the general population participants (a group for which we had detailed information on demographic and lifestyle factors that would allow for good control of confounding).
The reason for the lower risk of mortality from CVD and other nonmalignant causes among Blacks in the SCCS is unclear. The observed mortality deficit among Blacks persisted despite adjustment for additional factors that might benefit Blacks in the South, such as church-based social or spiritual support (data not shown). A possibility is the influence of confounding from unmeasured factors, perhaps even those related to resiliency in US Black communities,34–36 or residual confounding from factors included in the analysis (e.g., cigarette smoking). Despite similar smoking prevalence, White SCCS smokers had an average smoking history of 33.7 pack-years (SD = 23.7), as compared with 18.8 pack-years (SD = 16.2) among Black SCCS smokers, consistent with national figures showing higher numbers of cigarettes per day smoked among White than Black smokers.37 The makeup of the SCCS may be another factor. According to standardized mortality ratios (SMRs) comparing SCCS participants with the general US population of the same race, gender, and calendar year period within the 12 enrollment states, CHC-enrolled SCCS participants had, as expected, higher mortality rates than their general population counterparts, although this excess was relatively greater among Whites (SMRs of 2.36 for men and 1.77 for women) than among Blacks (SMRs of 1.45 for men and 1.13 for women). Volunteer participants from the general population are typically a select, healthier group, and standardized mortality ratios for the SCCS general population participants reflect this situation; however, this “healthy volunteer” effect was somewhat stronger for Blacks (SMRs of 0.80 for men and 0.88 for women) than for Whites (SMRs of 0.81 for men and 0.96 for women).
The link between low SES and premature mortality is well established.11,19,38–42 In our analyses, the effect of low income on all-cause mortality was on par with that of heavy smoking, with a hazard ratio of 1.75 (95% CI = 1.60, 1.92) for current smokers with 27 or more pack-years of exposure (vs those who had never smoked). The reason SES is such a strong predictor of mortality, independent of proximate risk factors such as smoking and obesity, is less understood and may involve life resources that escape quantification in epidemiologic analyses. Link and Phelan43 postulated that such key resources may include money, knowledge, power, prestige, and beneficial social connections; moreover, the advantages provided by higher SES are not only wide ranging and broadly applicable to overall health enhancement but are also “adaptable to changing health-related conditions and can be used to protect health no matter…the current risks, treatments, or diseases.”3(p267) Within the SCCS, we can show empirically that education is associated with health insurance benefits (8% of individuals with less than a high school education vs 33% of individuals with a college education reported having private health insurance). However, it is impossible to capture the full scope of relevant advantages, some of which may be quite subtle, in any study.
In line with some previous reports,19,21 we found that individual SES was a stronger predictor of all-cause mortality than neighborhood SES. We did not, however, find an interaction between individual and neighborhood SES, in contrast to a few prior reports suggesting that those of low individual SES exhibit the highest mortality rates in areas of high neighborhood SES.20,22 It is important to understand how residential communities can influence mortality, particularly given that mortality inequalities between deprived and nondeprived US communities appeared to increase from 1969 to 1998.44 The general living environment may influence or shape health exposures and outcomes through its infrastructure, values placed on health and health-associated factors, availability of health care, exposure to pollutants, and stress related to crime, transportation options, or lack of social cohesion.19,45
Some studies have documented, longitudinally, detrimental changes in physical functioning and in self-rated health among adults living in socioeconomically disadvantaged or otherwise burdened (e.g., excessive noise, heavy traffic) neighborhoods.46,47 In our study, it is noteworthy that whereas associations between individual income and mortality were limited to CVD and other nonmalignant diseases, trends in risk associated with the NDI were similar regardless of cause of death. Reasons for the differences by cause of death are not clear but raise the possibility of neighborhood environmental influences on cancer risk above those associated with low SES.
In analyses involving US vital statistics that could not account for individual-level characteristics such as SES, it has been estimated that a substantial proportion of the higher mortality among Blacks than Whites nationally arises from mortality that is “medical care amenable” or “avoidable,” that is, mortality from conditions that can often be controlled with quality health care and appropriate therapies.7,8,48 Examples of these conditions include CVD, type 2 diabetes, and intestinal and respiratory infections. On the basis of our cause-specific mortality findings and evidence that low SES, as opposed to race, is a primary barrier to obtaining preventive health services,49 we surmise that these reported observations are due predominantly to SES.
Strengths and Limitations
This study has a number of strengths, including its base in a large, diverse, well-characterized cohort with essentially complete follow-up for mortality. Despite the generally low SES of many SCCS participants, our study had sufficient numbers of both Black and White individuals across SES strata to enable well-powered analyses in an SES range relevant to the general US population. The simultaneous examination of race, individual SES, and neighborhood SES also provided needed insights and clarification into questions of interaction between individual and neighborhood SES and whether SES was a stronger mediator of mortality for Blacks than for Whites. In addition, the SCCS was unique in that there was substantial SES overlap between Blacks and Whites, affording opportunities for racial comparisons less confounded by SES than in other studies.
Our assessment of individual and neighborhood SES was limited by data at one point in time, and a life-course assessment may have provided richer insights into the impact of SES on adult mortality risk. In addition, poor health can contribute to loss of jobs, income, and private health insurance. The same dynamic may also shift those in poor health into poor neighborhoods. Although this was a prospective study, these phenomena, if they occurred prior to enrollment, may have created spurious associations between SES and mortality. We have some evidence, however, that our results did not stem from such bias: restricting analyses to individuals with more than 3 years of follow-up did not change the findings, and participants residing in the most deprived areas reported having lived in their current home for an average of nearly 9 years prior to enrollment (the average was identical for those who subsequently died and those who did not die). Finally, although we aimed to characterize “neighborhood” SES, census tracts were not devised to delineate neighborhood boundaries; however, the census tract is a widely used metric of convenience for this purpose.19,20,38,50,51
Conclusions
Our data support the strong role of SES in predicting premature mortality in the United States. Our findings also suggest that individual and community SES act independently to affect mortality risk in a similar manner for Blacks and Whites. Because they are new, our findings of lower mortality from noncancer causes among Blacks than Whites require replication in other studies before conclusive interpretations can be made.
Acknowledgments
This work was supported by grant R01 CA092447 from the National Cancer Institute.
We acknowledge the work of Jennifer Sonderman, who extracted the census variables needed for our analysis.
Human Participant Protection
The Southern Community Cohort Study was approved by the Vanderbilt University and Meharry Medical College institutional review boards. All participants provided written informed consent.
References
- 1.Hummer RA, Chinn JJ. Race/ethnicity and US adult mortality: progress, prospects, and new analyses. Du Bois Rev. 2011;8(1):5–24. doi: 10.1017/S1742058X11000051. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Stewart QT. The shape of inequality: racial disparities in age-specific mortality. Biodemography Soc Biol. 2008;54(2):152–182. doi: 10.1080/19485565.2008.9989140. [DOI] [PubMed] [Google Scholar]
- 3.Phelan JC, Link BG, Diez-Roux A, Kawachi I, Levin B. “Fundamental causes” of social inequalities in mortality: a test of the theory. J Health Soc Behav. 2004;45(3):265–285. doi: 10.1177/002214650404500303. [DOI] [PubMed] [Google Scholar]
- 4.Satcher D, Fryer GE, Jr, McCann J, Troutman A, Woolf SH, Rust G. What if we were equal? A comparison of the black-white mortality gap in 1960 and 2000. Health Aff (Millwood) 2005;24(2):459–464. doi: 10.1377/hlthaff.24.2.459. [DOI] [PubMed] [Google Scholar]
- 5.Levine RS, Foster JE, Fullilove RE et al. Black-white inequalities in mortality and life expectancy, 1933–1999: implications for Healthy People 2010. Public Health Rep. 2001;116(5):474–483. doi: 10.1093/phr/116.5.474. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Sloan FA, Ayyagari P, Salm M, Grossman D. The longevity gap between Black and White men in the United States at the beginning and end of the 20th century. Am J Public Health. 2010;100(2):357–363. doi: 10.2105/AJPH.2008.158188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Harper S, Lynch J, Burris S, Davey Smith G. Trends in the black-white life expectancy gap in the United States, 1983–2003. JAMA. 2007;297(11):1224–1232. doi: 10.1001/jama.297.11.1224. [DOI] [PubMed] [Google Scholar]
- 8.Macinko J, Elo IT. Black-white differences in avoidable mortality in the USA, 1980–2005. J Epidemiol Community Health. 2009;63(9):715–721. doi: 10.1136/jech.2008.081141. [DOI] [PubMed] [Google Scholar]
- 9.Harper S, Rushani D, Kaufman JS. Trends in the black-white life expectancy gap, 2003–2008. JAMA. 2012;307(21):2257–2259. doi: 10.1001/jama.2012.5059. [DOI] [PubMed] [Google Scholar]
- 10.Hoyert DL, Xu JQ. Deaths: preliminary data for 2011. Natl Vital Stat Rep 6. 2012 No. 61:52. [PubMed] [Google Scholar]
- 11.Sorlie PD, Backlund E, Keller JB. US mortality by economic, demographic, and social characteristics: the National Longitudinal Mortality Study. Am J Public Health. 1995;85(7):949–956. doi: 10.2105/ajph.85.7.949. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Davey Smith G, Neaton JD, Wentworth D, Stamler R, Stamler J. Mortality differences between black and white men in the USA: contribution of income and other risk factors among men screened for the MRFIT. Lancet. 1998;351(9107):934–939. doi: 10.1016/s0140-6736(00)80010-0. [DOI] [PubMed] [Google Scholar]
- 13.Franks P, Muennig P, Lubetkin E, Jia H. The burden of disease associated with being African-American in the United States and the contribution of socio-economic status. Soc Sci Med. 2006;62(10):2469–2478. doi: 10.1016/j.socscimed.2005.10.035. [DOI] [PubMed] [Google Scholar]
- 14.Williams DR, Mohammed SA, Leavell J, Collins C. Race, socioeconomic status, and health: complexities, ongoing challenges, and research opportunities. Ann N Y Acad Sci. 2010;1186:69–101. doi: 10.1111/j.1749-6632.2009.05339.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.LaVeist T, Thorpe R, Jr, Bowen-Reid T et al. Exploring health disparities in integrated communities: overview of the EHDIC study. J Urban Health. 2008;85(1):11–21. doi: 10.1007/s11524-007-9226-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.LaVeist T, Pollack K, Thorpe R, Jr, Fesahazion R, Gaskin D. Place, not race: disparities dissipate in southwest Baltimore when blacks and whites live under similar conditions. Health Aff (Millwood) 2011;30(10):1880–1887. doi: 10.1377/hlthaff.2011.0640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Signorello LB, Schlundt DG, Cohen SS et al. Comparing diabetes prevalence between African Americans and Whites of similar socioeconomic status. Am J Public Health. 2007;97(12):2260–2267. doi: 10.2105/AJPH.2006.094482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Meijer M, Rohl J, Bloomfield K, Grittner U. Do neighborhoods affect individual mortality? A systematic review and meta-analysis of multilevel studies. Soc Sci Med. 2012;74(8):1204–1212. doi: 10.1016/j.socscimed.2011.11.034. [DOI] [PubMed] [Google Scholar]
- 19.Winkleby MA, Cubbin C. Influence of individual and neighbourhood socioeconomic status on mortality among black, Mexican-American, and white women and men in the United States. J Epidemiol Community Health. 2003;57(6):444–452. doi: 10.1136/jech.57.6.444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yen IH, Kaplan GA. Neighborhood social environment and risk of death: multilevel evidence from the Alameda County Study. Am J Epidemiol. 1999;149(10):898–907. doi: 10.1093/oxfordjournals.aje.a009733. [DOI] [PubMed] [Google Scholar]
- 21.Borrell LN, Diez Roux AV, Rose K, Catellier D, Clark BL. Neighbourhood characteristics and mortality in the Atherosclerosis Risk in Communities Study. Int J Epidemiol. 2004;33(2):398–407. doi: 10.1093/ije/dyh063. [DOI] [PubMed] [Google Scholar]
- 22.Winkleby M, Cubbin C, Ahn D. Effect of cross-level interaction between individual and neighborhood socioeconomic status on adult mortality rates. Am J Public Health. 2006;96(12):2145–2153. doi: 10.2105/AJPH.2004.060970. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Signorello LB, Hargreaves MK, Blot WJ. The Southern Community Cohort Study: investigating health disparities. J Health Care Poor Underserved. 2010;21(suppl 1):26–37. doi: 10.1353/hpu.0.0245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Calle EE, Terrell DD. Utility of the National Death Index for ascertainment of mortality among Cancer Prevention Study II participants. Am J Epidemiol. 1993;137(2):235–241. doi: 10.1093/oxfordjournals.aje.a116664. [DOI] [PubMed] [Google Scholar]
- 25.Lash TL, Silliman RA. A comparison of the National Death Index and Social Security Administration databases to ascertain vital status. Epidemiology. 2001;12(2):259–261. doi: 10.1097/00001648-200103000-00021. [DOI] [PubMed] [Google Scholar]
- 26.Nam CB, Boyd M. Occupational status in 2000: over a century of census-based measurement. Popul Res Policy Rev. 2004;23(4):327–358. [Google Scholar]
- 27.Messer LC, Laraia BA, Kaufman JS et al. The development of a standardized neighborhood deprivation index. J Urban Health. 2006;83(6):1041–1062. doi: 10.1007/s11524-006-9094-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.US Census Bureau. 2000 Census of Population and Housing, summary file 3: technical documentation. Available at: http://www.census.gov/prod/cen2000/doc/sf3.pdf. Accessed August 19, 2014.
- 29.Sonderman JS, Mumma MT, Cohen SS, Cope EL, Blot WJ, Signorello LB. A multi-stage approach to maximizing geocoding success in a large population-based cohort study through automated and interactive processes. Geospat Health. 2012;6(2):273–284. doi: 10.4081/gh.2012.145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.US Department of Agriculture. 2003 rural-urban continuum codes. Available at: http://www.ers.usda.gov/data-products/rural-urban-continuum-codes.aspx. Accessed August 19, 2014.
- 31.Williams RL. A note on robust variance estimation for cluster-correlated data. Biometrics. 2000;56(2):645–646. doi: 10.1111/j.0006-341x.2000.00645.x. [DOI] [PubMed] [Google Scholar]
- 32.International Classification of Diseases, Ninth Revision. Geneva, Switzerland: World Health Organization; 1980. [Google Scholar]
- 33.International Classification of Diseases, 10th Revision. Geneva, Switzerland: World Health Organization; 1992. [Google Scholar]
- 34.McCubbin HI, Thompson EA, Thompson AI, Futrell JA, editors. Resiliency in African-American Families. Thousand Oaks, CA: Sage Publications; 1998. [Google Scholar]
- 35.Thomas CL. Exploring resiliency factors of older African American Katrina survivors. J Evid Based Soc Work. 2012;9(4):351–368. doi: 10.1080/15433714.2010.525411. [DOI] [PubMed] [Google Scholar]
- 36.DeNisco S. Exploring the relationship between resilience and diabetes outcomes in African Americans. J Am Acad Nurse Pract. 2011;23(11):602–610. doi: 10.1111/j.1745-7599.2011.00648.x. [DOI] [PubMed] [Google Scholar]
- 37.Caraballo RS, Holiday DB, Stellman SD et al. Comparison of serum cotinine concentration within and across smokers of menthol and nonmenthol cigarette brands among non-Hispanic black and non-Hispanic white US adult smokers, 2001–2006. Cancer Epidemiol Biomarkers Prev. 2011;20(7):1329–1340. doi: 10.1158/1055-9965.EPI-10-1330. [DOI] [PubMed] [Google Scholar]
- 38.Doubeni CA, Schootman M, Major JM et al. Health status, neighborhood socioeconomic context, and premature mortality in the United States: The National Institutes of Health-AARP Diet and Health Study. Am J Public Health. 2012;102(4):680–688. doi: 10.2105/AJPH.2011.300158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Pappas G, Queen S, Hadden W, Fisher G. The increasing disparity in mortality between socioeconomic groups in the United States, 1960 and 1986. N Engl J Med. 1993;329(2):103–109. doi: 10.1056/NEJM199307083290207. [DOI] [PubMed] [Google Scholar]
- 40.Singh GK, Siahpush M. Widening socioeconomic inequalities in US life expectancy, 1980–2000. Int J Epidemiol. 2006;35(4):969–979. doi: 10.1093/ije/dyl083. [DOI] [PubMed] [Google Scholar]
- 41.Smith GD, Wentworth D, Neaton JD, Stamler R, Stamler J. Socioeconomic differentials in mortality risk among men screened for the Multiple Risk Factor Intervention Trial: II. Black men. Am J Public Health. 1996;86(4):497–504. doi: 10.2105/ajph.86.4.497. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Smith GD, Neaton JD, Wentworth D, Stamler R, Stamler J. Socioeconomic differentials in mortality risk among men screened for the Multiple Risk Factor Intervention Trial: I. White men. Am J Public Health. 1996;86(4):486–496. doi: 10.2105/ajph.86.4.486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Link BG, Phelan J. Social conditions as fundamental causes of disease. J Health Soc Behav. 1995;35:80–94. [PubMed] [Google Scholar]
- 44.Singh GK. Area deprivation and widening inequalities in US mortality, 1969–1998. Am J Public Health. 2003;93(7):1137–1143. doi: 10.2105/ajph.93.7.1137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Cohen DA, Farley TA, Mason K. Why is poverty unhealthy? Social and physical mediators. Soc Sci Med. 2003;57(9):1631–1641. doi: 10.1016/s0277-9536(03)00015-7. [DOI] [PubMed] [Google Scholar]
- 46.Glymour MM, Mujahid M, Wu Q, White K, Tchetgen Tchetgen EJ. Neighborhood disadvantage and self-assessed health, disability, and depressive symptoms: longitudinal results from the Health and Retirement Study. Ann Epidemiol. 2010;20(11):856–861. doi: 10.1016/j.annepidem.2010.08.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Balfour JL, Kaplan GA. Neighborhood environment and loss of physical function in older adults: evidence from the Alameda County Study. Am J Epidemiol. 2002;155(6):507–515. doi: 10.1093/aje/155.6.507. [DOI] [PubMed] [Google Scholar]
- 48.Schwartz E, Kofie VY, Rivo M, Tuckson RV. Black/white comparisons of deaths preventable by medical intervention: United States and the District of Columbia 1980–1986. Int J Epidemiol. 1990;19(3):591–598. doi: 10.1093/ije/19.3.591. [DOI] [PubMed] [Google Scholar]
- 49.Sambamoorthi U, McAlpine DD. Racial, ethnic, socioeconomic, and access disparities in the use of preventive services among women. Prev Med. 2003;37(5):475–484. doi: 10.1016/s0091-7435(03)00172-5. [DOI] [PubMed] [Google Scholar]
- 50.Chen JT, Rehkopf DH, Waterman PD et al. Mapping and measuring social disparities in premature mortality: the impact of census tract poverty within and across Boston neighborhoods, 1999–2001. J Urban Health. 2006;83(6):1063–1084. doi: 10.1007/s11524-006-9089-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Krieger N, Chen JT, Waterman PD, Soobader MJ, Subramanian SV, Carson R. Choosing area based socioeconomic measures to monitor social inequalities in low birth weight and childhood lead poisoning: the Public Health Disparities Geocoding Project (US) J Epidemiol Community Health. 2003;57(3):186–199. doi: 10.1136/jech.57.3.186. [DOI] [PMC free article] [PubMed] [Google Scholar]