Influence of Life-Course Socioeconomic Position on Incident Heart Failure in Blacks and Whites

Abstract

The influence of early-life socioeconomic position (SEP) on incident heart failure in blacks and whites is unknown. The authors examined the relation between early-life SEP and incident, hospitalized heart failure among middle-aged US participants (2,503 black and 8,519 white) in the Atherosclerosis Risk in Communities (ARIC) Study. Early-life SEP indicators assessed included parental education, occupation, and home ownership. From 1987 to 2004, 221 and 537 incident heart failure events were identified in blacks and whites, respectively. In Cox proportional hazards regression, early-life SEP was inversely related to incident heart failure after adjustment for age, gender, and study center (for blacks, hazard ratio (HR) = 1.39, 95% confidence interval (CI): 1.00, 1.95; for whites, HR = 1.32, 95% CI: 1.06, 1.64). Additional adjustment for young and mid-to-older adulthood SEP and established heart failure risk factors attenuated this association towards the null in both blacks and whites. Of the SEP measures, mid-to-older adulthood SEP showed the strongest association with incident heart failure in both blacks (HR = 1.32, 95% CI: 0.90, 1.96) and whites (HR = 1.39, 95% CI: 1.11, 1.75). SEP over the life course is related to the risk of incident heart failure, with SEP later in adulthood having a more prominent role than earlier SEP.

Heart failure is the leading principal diagnosis for hospitalization among men aged 65 years or older in the United States, as well as a major contributor to health-care expenditures (1). There is growing evidence that mid-to-older adulthood measures of socioeconomic position (SEP) are inversely related to the risk of incident heart failure (2–7). Considering that heart failure is the culmination of disease processes developing over the life course (8), early-life SEP may also play a role in predicting the risk of incident heart failure. Numerous studies have observed that low SEP in childhood is associated with worse heart failure risk factors in adulthood, such as smoking (9, 10), high blood pressure (11–13), obesity (14–16), and coronary heart disease (17, 18). However, it is currently unknown whether early-life SEP directly contributes to the diagnosis of incident heart failure later in life and whether it independently influences incident heart failure beyond adjustment for adulthood SEP and traditional risk factors for heart failure.

Elucidating these associations is especially germane for African Americans, given that black men and women, on average, develop risk factors for heart failure such as obesity (19) and hypertension (20) at earlier ages than their white counterparts and have the highest incidence of heart failure in the United States (21). Moreover, it is unknown whether childhood SEP affects health outcomes differently in blacks and whites (22–24). Thus, SEP prior to middle age may provide additional insights into disparities surrounding incident heart failure. Ultimately, assessing the impact of SEP across the life course on incident heart failure could inform future research on how to adequately measure and monitor socioeconomic inequalities in heart failure.

Data from the biracial Atherosclerosis Risk in Communities (ARIC) Study offered us a unique opportunity to examine whether early-life SEP is significantly related to the risk of incident heart failure; whether the effect of early-life SEP on incident heart failure persists after adjustment for young adulthood SEP, mid-to-older adulthood SEP, and traditional heart failure risk factors; and whether early-life SEP differently affects the risk of incident heart failure in blacks and whites.

MATERIALS AND METHODS

Study population

The ARIC Study is an ongoing, biethnic, longitudinal cohort study designed to investigate the etiology of atherosclerosis and cardiovascular disease. At baseline (1987–1989), 15,792 men and women aged 45–64 years were enrolled in the study from 4 US communities: Forsyth County, North Carolina; Jackson, Mississippi; Minneapolis, Minnesota (suburbs); and Washington County, Maryland. Follow-up examinations occurred every 3 years after baseline until 1996–1998. Telephone calls have been made annually since 1987 to maintain contact with the participants and to ascertain their health status.

The Life Course Socioeconomic Status, Social Context, and Cardiovascular Disease (LC-SES) Study is an ancillary study to ARIC that was conducted from 2000 to 2001 to evaluate the association between socioeconomic exposures over the life course and cardiovascular diseases. Of the 15,792 original ARIC cohort study members, 12,716 (80.5%) were recruited into the LC-SES Study. Because of a small sample size (n = 35), nonblacks and nonwhites were excluded. Blacks who did not reside in Jackson, Mississippi, or Forsyth County, North Carolina, were also not included because of a small sample size (n = 55). Additionally, we excluded participants with prevalent heart failure (n = 752) (defined using the Gothenburg criterion (25, 26)) or with unknown heart-failure status (n = 287) at the initial ARIC visit in order to limit the analyses to recorded incident heart failure events. Further, participants with missing data for any of the covariates (464 blacks and 468 whites) were also excluded for this complete case analysis. Excluding participants with missing data did not bias the results of this study. The final sample consisted of 2,503 blacks and 8,519 whites.

Individual-level socioeconomic indicators

Data on measures of SEP in early life and young adulthood were collected retrospectively during the LC-SES Study (2000–2001). Data on the socioeconomic indicators for mid-to-older adulthood were obtained at ARIC baseline (1987–1989). Early-life SEP was represented by 3 measures: the highest educational level (≤8th grade, 9th–11th grade, or ≥12th grade), occupation (nonprofessional or managerial/professional), and home ownership status (yes or no) of the participant's parent(s) or caretaker(s) when the participant was 10 years of age. For parental SEP indicators, data from the parent or caretaker with the highest level were utilized. Occupation was defined on the basis of the 5-class National Statistics Socio-economic Classification schema used in the United Kingdom (27), which couples information on occupation (e.g., technician) and employment status (e.g., manager). Because of study-size limitations, we could not exactly replicate the 5-class National Statistics Socio-economic Classification indicator. Instead of creating a categorical measure with 5 levels, we dichotomized parental occupation into 2 groups: managerial/professional workers (i.e., professionals, managers, or employers) and nonprofessionals (i.e., homemakers, retired persons, and service and manual laborers). Young adulthood SEP at age 30 years consisted of the participant's education (less than high school, high school graduate and/or vocational school, or some college or greater), occupation (managerial/professional, nonprofessional, or homemaker), and home ownership status (yes or no). Mid-to-older adulthood SEP consisted of annual household income (in quartiles: ≤$25,000, >$25,000–≤$50,000, or >$50,000), occupation (managerial/professional, nonprofessional, homemaker, or retired), and home ownership status (yes or no). Homemakers and the retired were assigned a rank according to their risk of incident heart failure relative to persons in other occupations.

Summary SEP exposures

The above individual-level socioeconomic indicators were summed within each life epoch to test whether early-life SEP independently predicted the risk of incident heart failure. For each individual-level socioeconomic indicator, the least disadvantaged category constituted the referent group and was assigned a value of 0. The other categories were incrementally assigned a value of 1 to represent an escalation in negative exposure. The summary SEP scores ranged from 0 to 4 for early-life SEP, 0 to 5 for young adulthood SEP, and 0 to 6 for mid-to-older adulthood SEP. Each summary SEP score was classified into 3 categories. Given the uneven distributions of the 3 SEP scores, the 25th and 75th percentiles were used as cutpoints. The summary SEP scores were categorized separately for blacks and whites because of large differences in the SEP distributions in both groups, which may have been related to historical inequities in the workplace such as occupational segregation (28), wage discrimination (29), and/or harmful exposure to toxins (29). The least-disadvantaged participants were in the lowest group (i.e., ≤25th percentile), followed by the 25th–75th percentile and the ≥75th percentile. These SEP categories were labeled high, medium, and low, respectively.

Covariates

Age at baseline, ethnicity, and ARIC study center comprised the demographic risk factors. Age was modeled as categorical (<50, ≥50–<60, or ≥60 years). Race/ethnicity was self-reported as either black or white. Trained study personnel and research technicians took all physical measurements and administered all questionnaires, following a standardized protocol that included quality control measures at the baseline visit (30). Body mass index was calculated as weight (kilograms) divided by height squared (meters) and was categorized into normal (<25), overweight (≥25–30), and obese (≥30). Type 2 diabetes mellitus was defined as having a fasting blood glucose level ≥126 mg/dL, a nonfasting blood glucose level ≥200 mg/dL, use of hypoglycemic medication, or a self-reported physician's diagnosis. Hypertension was defined as systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg or use of antihypertensive medication. Prevalent coronary heart disease was determined by the presence of at least 1 of the following: Q waves on the visit 1 electrocardiogram, history of myocardial infarction diagnosed by a physician, coronary bypass, or angioplasty of coronary arteries. Cornell voltage left ventricular hypertrophy was evaluated by electrocardiography (31). Current smoking status (ever or never smoker), alcohol drinking status (ever or never drinker), and not having health insurance (yes or no) were self-reported.

Outcome

Incident heart failure was defined as the first occurrence of either an International Classification of Diseases, Ninth Revision (ICD-9) code 428 hospital discharge diagnosis or an underlying cause of death code of 428 or International Classification of Diseases, Tenth Revision (ICD-10) code 150 (26) occurring from baseline (1987–1989) through December 31, 2004, among persons without a previous record at baseline of a hospitalization with an International Classification of Diseases, Ninth Revision, Clinical Modification code 428. Incident heart failure was ascertained via annual contacts and standardized review of hospital records.

Statistical analysis

Demographic factors, individual-level socioeconomic indicators, and traditional risk factors for incident heart failure were compared between blacks and whites. The prevalence of traditional heart failure risk factors according to early-life SEP ranking is presented by race/ethnicity. Age-adjusted incidence rates of incident hospitalized heart failure were calculated using Poisson regression.

Cox proportional hazards regression was used to estimate hazard ratios and 95% confidence intervals for the association between early-life SEP indicators and incident heart failure. To test whether the association between incident heart failure and summary early-life SEP or its components differed by race/ethnicity, we evaluated terms for interaction between race/ethnicity and early-life SEP. Interaction terms between race/ethnicity and summary young adulthood SEP and mid-to-adulthood SEP were also tested for heterogeneity. No significant differences were observed between blacks and whites using an a priori P value of <0.10, except for parental home ownership. Race/ethnicity-specific regression models were adjusted for age, gender, study center, summary SEP indicators, and traditional risk factors for incident heart failure in sequential order. We used Spearman correlations to evaluate the correlation between the summary-level SEP indicators. The summary-level SEP measures over the life course were associated but not strongly correlated (r = 0.21–0.38), which allowed for simultaneous adjustment of early-life SEP and adulthood SEP. Terms for interaction between the summary-level SEP exposures were tested for heterogeneity using an a priori P value of <0.10. Among white participants, statistically significant interaction was found between young adulthood SEP and mid-to-older adulthood SEP. However, the inclusion of the interaction term between young adulthood and mid-to-older adulthood SEP did not alter the association between early-life SEP and the risk of incident heart failure; therefore, the interaction term was not included. Log (−log) survival plots were examined and continuous time-covariate interactions were created to ensure that the proportional hazards assumption was not violated, and no violations were observed.

For the majority of the individual-level socioeconomic variables, fewer than 3% of data were missing. Single imputation was used to substitute the average for the missing data. However, for parental education, data were missing for approximately 18% of observations. Multiple imputation was used to reduce potential bias because the data may not have been missing at random (32, 33). A Bayesian Gibbs sampling algorithm was used to perform the multiple imputation, utilizing 15 covariates and 500 iterations from 10 imputations. All statistical analyses were conducted using SAS, version 9.1.3 (SAS Institute, Inc., Cary, North Carolina). The institutional review board at the University of North Carolina approved the research.

RESULTS

During a mean follow-up period of 15.9 years, 221 (8.8%) blacks and 537 (6.3%) whites were diagnosed with incident heart failure (Table 1). At baseline, blacks and whites were of similar age (P = 0.78); however, a higher proportion of whites were male (45.8%) in comparison with blacks (36.1%) (P < 0.001). The distributions of individual-level SEP indictors in early life, young adulthood, and mid-to-older adulthood were statistically significantly different in blacks and whites, with whites having more education, higher income, greater home ownership, and higher occupational status. The prevalences of most of the traditional heart failure risk factors at baseline, namely body mass index, type 2 diabetes, hypertension, left ventricular hypertrophy, and current smoking status, were greater in blacks than in whites, whereas prevalent coronary heart disease and current alcohol drinking status were more common in whites.

Table 1.

Selected Demographic, Socioeconomic, and Health Characteristics of Participants at the ARIC Baseline Examination, by Racial/Ethnic Group, ARIC LC-SES Study, 1987–2004^a

Characteristic	Blacks (n = 2,503)	Whites (n = 8,519)	P Valueb
Median duration of follow-up, years (interquartile range)	16.0 (15.4–16.9)	16.2 (15.5–17.0)	<0.001
No. with incident heart failure	221	537	<0.001
Total duration of follow-up, person-years	39,554.3	136,510.0
Demographic factors
Mean age, years (standard deviation)	52.8 (5.7)	53.9 (5.6)	0.78
Male gender, %	36.1	45.8	<0.001
ARIC study center, %
Jackson, Mississippi	90.4	0.0	<0.001
Forsyth County, North Carolina	9.6	30.5
Minneapolis, Minnesota (suburbs)	0.0	36.0
Washington County, Maryland	0.0	33.5
Individual-level childhood SEP indicators
Parental education, %
≤8th grade	35.2	29.0	<00001
9th–11th grade	48.3	30.2
≥12th grade	16.5	40.8
Parental occupationc
Nonprofessional	57.1	34.2	<0.001
Managerial/professional	42.9	65.9
Home ownership
No	54.3	26.6	<0.001
Yes	45.7	73.4
Individual-level young adulthood SEP indicators
Education
Less than high school	35.5	14.4	<0.001
High school graduation and/or vocational school	28.7	46.1
Some college or greater	35.8	39.6
Occupationc
Nonprofessional	45.2	19.8	<0.001
Homemaker	10.7	29.6
Managerial/professional	44.1	50.6
Home ownership
No	52.0	27.2	<0.001
Yes	48.0	72.9
Individual-level mid-to-older adulthood SEP indicators
Quartile of annual household income
≤$25,000	69.4	22.2	<0.001
>$25,000–≤$50,000	23.0	45.7
>$50,000	7.6	32.0
Occupation
Nonprofessional	46.9	26.7	<0.001
Retired	10.1	13.7
Homemaker	8.2	9.1
Managerial/professional	34.9	50.5
Home ownership
No	16.5	6.7	<0.001
Yes	83.5	93.3
Heart failure risk factors
Body mass indexd
≥25–<30	39.1	40.5	<0.001
≥30	39.7	20.9
Type 2 diabetes, mellitus, %	14.6	7.0	<0.001
Hypertension, %	50.5	23.6	<0.001
Prevalent coronary heart disease, %	2.2	3.3	<0.01
Left ventricular hypertrophye, %	5.3	0.9	<0.001
Current smoker, %	25.5	21.0	<0.001
Current alcohol drinker, %	31.8	66.3	<0.001
No health insurance, %	20.7	4.1	<0.001

Abbreviations: ARIC, Atherosclerosis Risk in Communities; LC-SES, Life Course Socioeconomic Status, Social Context, and Cardiovascular Disease; SEP, socioeconomic position.

^aAll demographic and heart failure risk factors were measured at the ARIC baseline examination (1987–1989). Data on the SEP indicators were self-reported at follow-up during the LC-SES Study (2001).

^bTwo-sided P values were generated by chi-squared test for categorical variables and t test for age and the Kruskal-Wallis test for follow-up.

^cOccupation was defined on the basis of the National Statistics Socio-economic Classification (27). The nonprofessional class represents intermediate occupations, lower supervisory and technical occupations, and semiroutine to routine occupations.

^dWeight (kg)/height (m)².

^eLeft ventricular hypertrophy was determined by electrocardiography.

Demographic characteristics and heart failure risk factors according to summary early-life SEP group are presented in Table 2 for blacks and whites; the specific SEP score for each life epoch and race/ethnicity group is detailed in Table 3. A greater percentage of blacks with low SEP in early life were overweight, obese, had coronary heart disease, and lacked health insurance at baseline as compared with blacks with medium or high early-life SEP. The prevalence of type 2 diabetes, hypertension, left ventricular hypertrophy, and current smoking did not vary in blacks by level of SEP in early life. In contrast, among whites, demographic and heart failure risk factors at baseline (with the exception of left ventricular hypertrophy) significantly varied by level of early-life SEP, with participants of medium or high SEP showing more favorable measures than participants with low early-life SEP (Table 2).

Table 2.

Selected Demographic and Health Characteristics^a of Participants by Racial/Ethnic Group and Summary Early-Life Socioeconomic Position^b, ARIC LC-SES Study, 1987–2004

Characteristic	Early-Life SEP
	Blacks (n = 2,503)				Whites (n = 8,519)
	Low (n = 833)	Medium (n = 893)	High (n = 777)	P Valuec	Low (n = 3,405)	Medium (n = 2,551)	High (n = 2,563)	P Valuec
Demographic factors
Male gender, %	32.2	37.7	38.5	0.01	44.5	45.6	47.8	0.03
Mean age, years (standard deviation)	53.0 (5.7)	53.1 (5.7)	52.1 (5.6)	<0.001	54.9 (5.5)	53.9 (5.6)	52.6 (5.6)	<0.001
Heart failure risk factors
Body mass indexd, %
≥25–<30	36.0	38.8	42.9	<0.001	40.7	41.1	39.5	0.01
≥30	41.8	42.2	34.5		22.4	20.5	19.5
Type 2 diabetes mellitus, %	14.9	14.6	14.4	0.96	7.8	7.7	5.2	<0.001
Hypertension, %	50.7	51.7	49.0	0.54	25.6	24.4	20.1	<0.001
Prevalent coronary heart disease, %	2.9	2.7	1.0	0.02	3.8	3.4	2.6	0.03
Left ventricular hypertrophye, %	5.9	4.9	5.2	0.66	1.1	0.8	0.6	0.12
Current smoker, %	24.5	25.2	26.8	0.56	22.0	22.3	18.2	<0.001
Current alcohol drinker, %	28.5	29.9	37.7	<0.001	58.6	65.7	77.1	<0.001
No health insurance, %	22.6	24.3	14.5	<0.001	5.4	4.1	2.5	<0.001

Abbreviations: ARIC, Atherosclerosis Risk in Communities; LC-SES, Life Course Socioeconomic Status, Social Context, and Cardiovascular Disease; SEP, socioeconomic position.

^aAll demographic and heart failure risk factors were measured at the ARIC baseline examination (1987–1989).

^bSummary early-life SEP was generated by summing information on parental educational level, class, and home ownership status when the participant was 10 years of age and dividing the total at the 25th and 75th percentiles. “Low,” “medium,” and “high” represent the most disadvantaged to the least disadvantaged, respectively. See Table 3 for the specific SEP score for each group. Data on individual-level SEP indicators were self-reported during the LC-SES Study (2001).

^cP values were generated by chi-squared test for categorical variables and the Kruskal-Wallis test for continuous exposures.

^dWeight (kg)/height (m)².

^eLeft ventricular hypertrophy was determined by electrocardiography using the Cornell criteria (31).

Table 3.

Age-Adjusted Incidence Rates^a of Heart Failure by Racial/Ethnic Group and Socioeconomic Position Across the Life Course, ARIC LC-SES Study, 1987–2004

Life Epochb	Blacks (n = 2,503)						Whites (n = 8,519)
	SEP Score	No. of Heart Failure Events	No. of Subjects	Total PY of Follow-up	Incidence Rate/1,000 PY	95% CI	SEP Score	No. of Heart Failure Events	No. of Subjects	Total PY of Follow-up	Incidence Rate/1,000 PY	95% CI
Overall		221	2,503	39,554.30	5.23	4.32, 6.33		537	8,519	136,510	3.18	2.81, 3.60
Early lifec
Low	3–4	88	833	13,123.3	6.09	4.97, 7.45	2–4	265	3,405	54,397.6	3.99	3.52, 4.53
Medium	2	77	893	14,068.5	5.16	4.48, 5.95	1	150	2,551	40,808.7	3.40	3.08, 3.75
High	0–1	56	777	12,362.6	4.38	3.46, 5.55	0	122	2,563	41,303.7	2.90	2.47, 4.00
P-trendd					0.0540						0.0028
Young adulthoode
Low	4–5	101	868	13,541.0	6.67	5.47, 8.13	3–5	203	2,211	35,149.8	4.56	3.96, 5.26
Medium	3	44	446	7,037.8	5.27	4.57, 6.06	2	115	2,075	33,396.3	3.61	3.29, 3.97
High	0–2	76	1,189	18,975.5	4.16	3.37, 5.14	0-1	219	4,233	67,963.9	2.86	2.51, 3.27
P-trend					0.0018						<0.0001
Mid-to-older adulthoodf
Low	4–6	87	683	10,602.5	7.20	5.81, 8.91	3–6	265	2,607	41,347.9	4.73	4.11, 5.43
Medium	3	79	791	12,506.5	5.30	4.60, 6.10	2	101	1,858	29,897.3	3.57	3.25, 3.92
High	0–2	55	1,029	16,445.3	3.89	3.10, 4.89	0–1	171	4,054	65,264.8	2.70	2.34, 3.11
P-trend					0.0004						<0.0001

Abbreviations: ARIC, Atherosclerosis Risk in Communities; CI, confidence interval; LC-SES, Life Course Socioeconomic Status, Social Context, and Cardiovascular Disease; PY, person-years; SEP, socioeconomic position.

^aIncidence rates were estimated and adjusted for mean age at baseline for blacks and whites, separately.

^bEach categorical early life epoch measure was generated by summing the individual-level SEP factors within each life epoch and dividing the total at the 25th and 75th percentiles. “Low,” “medium,” and “high” represent the most disadvantaged to the least disadvantaged, respectively.

^cSummary early-life SEP was generated by summing parental educational level, occupation, and home ownership status when the participant was 10 years of age. Data on individual-level SEP indicators were self-reported during the LC-SES Study (2001).

^dP value from a linear trend test.

^eYoung adulthood SEP consisted of the educational level, occupation, and home ownership status of the participant at age 30 years. Data on individual-level SEP indicators were collected during the LC-SES Study (2001).

^fMid-to-older adulthood SEP was a composite of the participant's income, occupation, and home ownership status at ARIC baseline (1987–1989).

Over the 18 years of follow-up (1987–2004), the overall age-adjusted incidence rate of heart failure was 5.23 events per 1,000 person-years in blacks and 3.18 events per 1,000 person-years in whites (Table 3). A graded linear association was observed between the incidence rate of heart failure and summary early-life SEP in both blacks (linear trend test: P = 0.0540) and whites (linear trend test: P = 0.0028). Blacks with a high, medium, or low level of summary early-life SEP had a greater incidence rate of heart failure than whites within each respective level. Notably, blacks with the least exposure to negative early-life SEP had a greater incidence rate of heart failure (incidence rate = 4.38 per 1,000 person-years, 95% confidence interval (CI): 3.46, 5.55) than whites with the most exposure to negative early-life SEP (incidence rate = 3.99 per 1,000 person-years, 95% CI: 3.52, 4.53).

In both blacks and whites, summary SEP from early life continued to show an inverse graded association with incident heart failure after adjustment for age, gender, and study center (Table 4). The effect estimates for summary early-life SEP (low vs. high) were comparable in blacks (hazard ratio (HR) = 1.39, 95% CI: 1.00, 1.95) and whites (HR = 1.31, 95% CI: 1.06, 1.64) (model 1). In both race/ethnicity groups, young adulthood SEP and mid-to-older adulthood SEP were also significantly related to the risk of incident heart failure, with the latter showing the strongest link with the outcome. The association between summary early-life SEP and incident heart failure was attenuated after adjustment for summary young adulthood SEP in both blacks (HR = 1.25, 95% CI: 0.89, 1.77) and whites (HR = 1.20, 95% CI: 0.96, 1.51) (model 2). Adjustment for summary mid-to-older adulthood SEP also reduced the magnitude of the association (model 3). Further adjustment for heart failure risk factors in the fully adjusted model slightly reduced the effect estimate for early-life SEP in both race/ethnicity groups (model 4). In the fully adjusted model, young adulthood SEP and mid-to-older adulthood SEP persisted in showing an inverse association with the risk of incident heart failure. In blacks, body mass index, type 2 diabetes, hypertension, prevalent coronary heart disease, left ventricular hypertrophy, current smoking status, and lack of health insurance were independently related to an increased risk of incident heart failure, and current alcohol drinking status was inversely related to the outcome (model 4). For whites, each of the heart failure risk factors, with the exception of current alcohol drinking and health insurance status, was predictive of incident heart failure (model 4).

Table 4.

Adjusted Hazard Ratios for the Association Between Socioeconomic Position Across the Life Course and Incident Heart Failure, by Racial/Ethnic Group, ARIC LC-SES Study, 1987–2004^a

Race/Ethnicity and Characteristicb	Model 1		Model 2		Model 3		Model 4
	HR	95% CI	HR	95% CI	HR	95% CI	HR	95% CI
Blacks
Early-life SEPc
Low	1.39	1.00, 1.95	1.25	0.89, 1.77	1.20	0.85, 1.69	1.19	0.84, 1.69
Medium	1.11	0.78, 1.57	1.02	0.72, 1.44	0.98	0.69, 1.39	0.97	0.68, 1.38
High	1.00		1.00		1.00		1.00
Young adulthood SEPd
Low	1.63	1.20, 2.20	1.57	1.15, 2.14	1.33	0.95, 1.87	1.13	0.81, 1.57
Medium	1.48	1.02, 2.14	1.45	1.00, 2.10	1.28	0.87, 1.89	1.14	0.77, 1.68
High	1.00		1.00		1.00		1.00
Mid-to-older adulthood SEPe
Low	1.95	1.37, 2.78			1.68	1.14, 2.47	1.32	0.90, 1.96
Medium	1.62	1.15, 2.31			1.38	0.94, 2.03	1.20	0.82, 1.76
High	1.00				1.00		1.00
Heart failure risk factors
BMIf ≥25–30 vs. <25							1.69	1.06, 2.69
BMI ≥30 vs. <25							2.39	1.51, 3.78
Type 2 diabetes mellitus (yes vs. no)							2.91	2.19, 3.86
Hypertension (yes vs. no)							1.42	1.06, 1.90
Prevalent coronary heart disease (yes vs. no)							2.90	1.71, 4.92
Left ventricular hypertrophy (yes vs. no)							2.12	1.41, 3.19
Current smoker (yes vs. no)							2.06	1.53, 2.78
Current alcohol drinker (yes vs. no)							0.66	0.47, 0.93
No health insurance (yes vs. no)							1.44	1.06, 1.93
Whites
Early-life SEPc
Low	1.32	1.06, 1.64	1.20	0.96, 1.51	1.16	0.92, 1.46	1.08	0.86, 1.35
Medium	1.10	0.86, 1.39	1.05	0.82, 1.34	1.03	0.81, 1.32	0.95	0.74, 1.21
High	1.00		1.00		1.00		1.00
Young adulthood SEPd
Low	1.53	1.25, 1.87	1.47	1.20, 1.81	1.33	1.07, 1.64	1.14	0.91, 1.42
Medium	1.12	0.89, 1.41	1.09	0.86, 1.39	1.05	0.83, 1.34	0.97	0.76, 1.23
High	1.00		1.00		1.00		1.00
Mid-to-older adulthood SEPe
Low	1.77	1.43, 2.18			1.60	1.28, 1.99	1.39	1.11, 1.75
Medium	1.21	0.94, 1.55			1.10	0.85, 1.42	1.09	0.85, 1.41
High	1.00				1.00		1.00
Heart failure risk factors
BMI ≥25–30 vs. <25							1.57	1.24, 1.98
BMI ≥30 vs. 25							2.89	2.27, 3.68
Type 2 diabetes mellitus (yes vs. no)							2.14	1.71, 2.67
Hypertension (yes vs. no)							1.61	1.34, 1.92
Prevalent coronary heart disease (yes vs. no)							3.50	2.71, 4.53
Left ventricular hypertrophy (yes vs. no)							2.34	1.37, 4.01
Current smoker (yes vs. no)							2.34	1.93, 2.84
Current alcohol drinker (yes vs. no)							0.89	0.74, 1.07
No health insurance (yes vs. no)							0.99	0.76, 1.16

Abbreviations: ARIC, Atherosclerosis Risk in Communities; BMI, body mass index; CI, confidence interval; HR, hazard ratio; LC-SES, Life Course Socioeconomic Status, Social Context, and Cardiovascular Disease; SEP, socioeconomic position.

^aIn model 1, results for each life epoch were adjusted for age, gender, and study center. In model 2, results were adjusted for all variables in model 1 + young adulthood SEP. In model 3, results were adjusted for all variables in model 2 + mid-to-older adulthood SEP. In model 4, results were adjusted for all variables in model 3 + heart failure risk factors.

^bEach life epoch SEP measure was generated by summing the individual SEP factors within each life epoch and dividing the total at the 25th and 75th percentiles. “Low,” “medium,” and “high” represent the most disadvantaged to the least disadvantaged, respectively.

^cEarly-life SEP was composed of parents’ educational level, class, and home ownership status when the participant was 10 years of age. Data on individual-level SEP indicators for early life and young adulthood were collected during the LC-SES Study (2001).

^dYoung adulthood SEP consisted of the educational level, class, and home ownership status of the participant at age 30 years. Data on individual-level SEP indicators for young adulthood were collected during the LC-SES Study (2001).

^eMid-to-older adulthood SEP was a composite of the participant's income, occupation, and home ownership status at ARIC baseline (1987–1989).

^fWeight (kg)/height (m)².

To evaluate the influence of each early-life SEP indicator, we individually assessed the impact of parental education, occupation, and home ownership status on the risk of incident heart failure (Table 5). In both blacks and whites, parental education (≤8th grade vs. ≥12th grade) was the strongest individual-level SEP indicator after adjustment for age, gender, and study center. After accounting for summary young adulthood SEP, mid-to-older adulthood SEP, and heart failure risk factors, blacks and whites whose parents had an educational level of 8th grade or less still had a higher risk of incident heart failure than participants whose parents had attained at least a 12th-grade education, but the results were imprecise. A lack of parental home ownership in early life was also associated with an increased risk of incident heart failure among black participants, but not among whites. In adjusted analyses, having a nonprofessional parent in childhood versus a professional parent did not predict the risk of incident heart failure in either blacks or whites.

Table 5.

Adjusted Hazard Ratios for the Association Between Individual-Level Socioeconomic Exposures Across the Life Course and Incident Heart Failure, by Racial/Ethnic Group, ARIC LC-SES Study, 1987–2004^a

Individual-Level Socioeconomic Indicator From Early Lifeb	Blacks (n = 2,503)							Whites (n = 8,519)
	No. of Heart Failure Events	Model 1		Model 2		Model 3		No. of Heart Failure Events	Model 1		Model 2		Model 3
		HR	95% CI	HR	95% CI	HR	95% CI		HR	95% CI	HR	95% CI	HR	95% CI
Parental education
≤8th grade	78	1.80	1.14, 2.84	1.52	0.95, 2.42	1.31	0.82, 2.09	191	1.44	1.16, 1.79	1.31	1.06, 1.64	1.20	0.96, 1.49
9th–11th grade	121	1.50	0.93, 2.41	1.33	0.82, 2.14	1.24	0.76, 2.01	187	1.41	1.14, 1.74	1.26	1.01, 1.57	1.19	0.95, 1.48
≥12th grade	22	1.00		1.00		1.00		159	1.00		1.00		1.00
Parental occupationc
Nonprofessional	60	1.11	0.83, 1.50	1.03	0.76, 1.39	1.08	0.80, 1.46	45	1.24	0.91, 1.69	1.18	0.87, 1.60	1.04	0.77, 1.42
Managerial and professional	161	1.00		1.00		1.00		492	1.00		1.00		1.00
Home ownership
No	141	1.42	1.08, 1.87	1.24	0.93, 1.66	1.18	0.89, 1.58	150	0.90	0.74, 1.09	0.85	0.70, 1.04	0.83	0.68, 1.01
Yes	80	1.00		1.00		1.00		387	1.00		1.00		1.00

Abbreviations: ARIC, Atherosclerosis Risk in Communities; CI, confidence interval; HR, hazard ratio; LC-SES, Life Course Socioeconomic Status, Social Context, and Cardiovascular Disease; SEP, socioeconomic position.

^aIn model 1, results were adjusted for demographic characteristics (age, gender, and study center). In model 2, results were adjusted for all of the factors in model 1 + young adulthood SEP + mid-to-older adulthood SEP. In model 3, results were adjusted for all of the factors in model 2 + body mass index, type 2 diabetes, hypertension, left ventricular hypertrophy, prevalent coronary heart disease, current smoking status, current alcohol drinking status, and health insurance (yes or no).

^bEarly-life individual socioeconomic indicators reflect parents’ educational level, class, and home ownership status when the participant was 10 years of age. Data on indicators of early-life SEP were collected retrospectively during the LC-SES Study (2001).

^cOccupation was defined on the basis of the National Statistics Socio-economic Classification (27).

DISCUSSION

In this biracial, community-based study, we found that worse SEP in early life was significantly associated with an increased risk of incident heart failure. However, accounting for summary young adulthood SEP and mid-to-older adulthood SEP attenuated the effect of early-life SEP on the risk of incident heart failure, attenuating it towards the null value. Despite black participants’ having a higher burden of adverse exposures than whites during childhood, the effects of early-life SEP on incident heart failure were similar in blacks and whites.

To our knowledge, this study is the first report on the association between early-life SEP and the risk of incident heart failure. Prior studies have consistently shown similar inverse relations between early-life SEP and other cardiovascular diseases (17, 34–39), which were also largely mitigated by adulthood SEP and/or cardiovascular risk factors (34, 36–38). Childhood SEP may not have a direct effect on the incidence of heart failure or other cardiovascular diseases because it occurs decades prior to organ damage or its diagnosis. Instead, it may exert an influence on physical development and health behaviors in youth, which then in turn affect risk factors downstream. Thus, our findings lend support to previous studies (34, 36–38) and the pathway model (18), which suggest that early-life SEP is an upstream determinant that shapes risk factors, which in turn directly affect health outcomes later in life. This interpretation must be considered in the context of our limitations. As in most studies of this nature, we lacked information regarding “episodic factors” and “economic shocks” incurred during childhood, such as parental unemployment and other hardships, which may also play a pivotal role in children's long-term health.

There has been recent interest in understanding whether life-course SEP differentially explains health outcomes in blacks and whites, given that in the United States blacks are disproportionately disadvantaged, and it is well-established that poor health outcomes in adulthood are linked to level of material wealth across the life course. Unfortunately, there have been few such studies (22, 24, 40, 41). Congruent with our findings for incident heart failure, the reported effects of negative SEP in childhood have been found to be similar in blacks and whites with regard to self-rated health (22), physical, mental, and cognitive well-being (24), and risk of diabetes mellitus (40).

In both blacks and whites, parental education was the strongest early-life predictor of heart failure. Interestingly, black participants whose parent(s) did not own their homes were at increased risk of incident heart failure. These factors are worth further exploration. They may bring to light pathways which may have an adverse effect on health outcomes.

Several limitations of our study should be noted. Inherent in collecting SEP data retrospectively is the chance of recall bias, especially for childhood SEP indicators. The impact of the recall bias should have been minimal, however, because a recent study found that 89% of ARIC participants’ recall of their father's occupation matched data from birth certificates and census records (κ = 0.86) (42). Although 80.5% of the ARIC participants completed the LC-SES survey, there was a chance of selection bias. However, in a sensitivity analysis of the cohort members who did not complete the LC-SES survey, we found that they were older, more frequently male, and had a lower educational level and family income at ARIC baseline. Thus, our results may be biased toward the null. As an additional consideration, the SEP of older participants may have changed since the baseline visit, which could potentially have affected their risk of heart failure. We also note that only Southern blacks were included in the study; thus, our results may not be generalizable to African Americans living in other regions of the United States.

There are many methodological and conceptual challenges in using occupational data as an exposure in this context (43). Beyond the usual complexities of using occupational data, there were additional challenges that had to be met in this study, because in addition to contemporaneous data, we also incorporated remote, historical data to approximate SEP. In this respect, our approach of categorizing occupations into distinct groups using the National Statistics Socio-economic Classification criteria (27) as a guide represents a practical satisfactory solution. Additionally, because of a lack of literature on the subject matter, adult occupation may have been limited by our method of ranking homemakers and retired participants. Lastly, heart failure events were not validated by physician review of medical records but by heart failure diagnosis codes 428 (ICD-9) and I50 (ICD-10), which were by far the most frequent and specific documented codes for the heart failure events in this study (44, 45).

Despite these limitations, our study had a number of strengths. To our knowledge, we are the first investigators to report on the association between SEP across the life course and incident heart failure in African Americans, a group particularly afflicted with this disease. In addition, we were able to examine the effect of early-life SEP in the presence of SEP in mid-to-older adulthood as well as young adulthood. Incident heart failure events were ascertained from hospital discharge records rather than from self-reports, reducing the opportunity for bias. Further, retention of the ARIC participants was high (95%) at the time of the LC-SES Study, thus minimizing survivorship bias.

In conclusion, we found that in both blacks and whites, early-life SEP was inversely related to the risk of incident heart failure, but its effect was reduced after accounting for adulthood SEP and traditional risk factors for heart failure. SEP over the life course is associated with the risk of incident heart failure, with SEP later in adulthood showing a more prominent role than SEP from childhood or young adulthood in both blacks and whites. These findings call attention to the need for additional bi-/multiethnic studies to elucidate the mechanisms by which SEP over the life course influences risk factors and health outcomes later in life.

Glossary

Abbreviations

ARIC

Atherosclerosis Risk in Communities

CI

confidence interval

HR

hazard ratio

ICD

International Classification of Diseases

LC-SES

Life Course Socioeconomic Status, Social Context, and Cardiovascular Disease

SEP

socioeconomic position