Psychosomatic Medicine Faster Service from Outside North America
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Thurston, R. C.
Right arrow Articles by Berkman, L. F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Thurston, R. C.
Right arrow Articles by Berkman, L. F.
Related Collections
Right arrow Social Class
Right arrow Anxiety
Right arrow Depression
Right arrow Other Cardiovascular Medicine
Psychosomatic Medicine 68:25-32 (2006)
© 2006 American Psychosomatic Society

ORIGINAL ARTICLES

Do Depression and Anxiety Mediate the Link Between Educational Attainment and CHD?

Rebecca C. Thurston, PhD, Laura D. Kubzansky, PhD, MPH, Ichiro Kawachi, MD, PhD and Lisa F. Berkman, PhD

From the Department of Society, Human Development, and Health, Harvard School of Public Health, Boston, MA.

Address correspondence and reprint requests to Rebecca C. Thurston, PhD, Department of Psychiatry, University of Pittsburgh, 3811 O’Hara Street, Pittsburgh, PA 15213. E-mail: thurstonrc{at}upmc.edu


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
Objective: Depression and anxiety are frequently hypothesized yet rarely examined pathways linking low socioeconomic status (SES) to coronary heart disease (CHD). This study evaluates depression and anxiety as mediators of the association between educational attainment and incident CHD.

Methods: Subjects (n = 6265, age 25–74) were participants in NHANES I and follow-up studies, a longitudinal, nationally representative study of the US population. Measures of educational attainment and depressive and anxious symptoms (General Well-Being Schedule) were derived from the baseline interview and incident CHD from hospital records and death certificates. Analyses included logistic regression and Cox proportional hazards models.

Results: In fully adjusted models, less than high school (relative risk [RR] = 1.46; 95% confidence interval [CI], 1.15–1.86) and some college (RR = 1.40; 95% CI, 1.05–1.88) education were associated with increased CHD risk relative to a college education. High depressive (RR = 1.31; 95% CI, 1.06–1.61) or anxious (RR = 1.35; 95% CI, 1.13–1.62) symptoms were associated with significantly increased CHD risk relative to low symptoms. Low educational levels were associated with increased risk for high depressive (OR = 3.43; 95% CI, 2.34–5.03) and anxious (OR = 1.71; 95% CI, 1.32–2.22) symptoms. However, depressive and anxious symptoms accounted for little of the association between education and CHD.

Conclusion: Education and depressive and anxious symptoms are associated with each other and risk of incident CHD. Although depressive and anxious symptoms are highest among those with lowest levels of education, they do not appear to mediate the relation between educational attainment and incident CHD. Findings suggest the importance of interventions to reduce socioeconomic disadvantage and negative affect in preventing CHD.

Key Words: socioeconomic status • education • depression • anxiety • coronary heart disease • mediation

Abbreviations: SES = socioeconomic status; CHD = coronary heart disease; NHANES I = First National Health and Nutrition Survey; NHEF = First National Health and Nutrition Epidemiologic Follow-up Surveys; GWB = General Well-Being Schedule; GWB-A = General Well-Being Schedule Anxiety subscale; GWB-D = General Well-Being Schedule Depression subscale; BMI = body mass index; SBP = systolic blood pressure; DBP = diastolic blood pressure; RR = relative risk; CI = confidence interval; MI = myocardial infarction.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
The inverse relation between socioeconomic status (SES) and coronary heart disease (CHD) is well established, with lower levels of educational attainment and income associated with increased CHD morbidity and mortality (1–3). This relation exists across age, gender, racial/ethnic groups, and across virtually all levels of SES, with additional protection afforded at even the highest levels of educational attainment (1–3). The relation between SES and CHD mortality has been demonstrated in a number of studies in representative samples of the US population (3) and in Europe (2,4–6). However, evidence linking SES with incident CHD is more limited (7–10).

Research suggests that standard cardiovascular risk factors, such as smoking, obesity, diabetes, atherogenic lipid profiles, hypertension, and a sedentary lifestyle, increase with decreasing SES (11,12). However, they generally do not fully account for the relation between SES and CHD, with some estimates indicating they account for 20% to 35% of this association (2). Thus, pathways linking SES to CHD risk are not fully understood and are likely to encompass a broader set of exposures than standard risk factors. An exploration of one of these risks, negative affect, is the focus of our current analysis.

Negative affect, such as depression and anxiety, also appears to follow an SES gradient, with lower SES associated with higher prevalence of mood and anxiety disorders and symptoms (13). A growing literature has documented the association between negative affect and increased risk for CHD. Depressive symptoms or disorders are associated with increased risk for incident CHD (14,15) and CHD mortality (16). Other research indicates that anxious symptoms and disorders may be associated with risk of incident CHD (17) and sudden cardiac death (18,19).

Given the links between SES, CHD, and depression and anxiety, Gallo and Matthews (20) recently hypothesized that depression and anxiety may be partial mediators of the link between SES and CHD. Some research has evaluated the mediational role of other psychological factors, such as hostility (21) and job control (22), in the relationship between SES and cardiovascular outcomes. However, despite the frequency by which negative affect has been cited as a link between SES and health (20,23–26), depression and anxiety have not been empirically evaluated as possible mediators between SES and CHD.

To examine whether negative affect may represent a pathway between SES and CHD, Gallo and Matthews (20) underscore the value of performing a mediational analysis (27). They note that many previous reports have implied mediation without performing formal mediational analyses. According to criteria set forth by Baron and Kenny (27), the following relationships need to be evaluated in considering negative affect as a mediator in the association between SES and CHD: (1) SES and incident CHD, (2) SES and negative affect, and (3) negative affect and incident CHD. Assuming that negative affect accounts for variations in levels of both SES and CHD incidence, the analysis would then examine whether including negative affect in a model with SES and incident CHD eliminates or sizably attenuates the relationship between SES and CHD.

Following the procedure outlined above, this study will evaluate symptoms of depression and anxiety as partial mediators of the relation between SES, as indexed by educational attainment, and incident CHD in the First National Health and Nutrition Survey (NHANES I) and the NHANES I Epidemiologic Follow-up Surveys (NHEF, 1982–1992). These studies together compose a nationally representative, longitudinal, population-based study. We hypothesize that (1) lower SES will be associated with increased risk of incident CHD, (2) elevated depressive and anxious symptoms will each be associated with increased risk of incident CHD, (3) lower SES will be associated with increased depressive and anxious symptoms, and (4) depressive and anxious symptoms will each act as partial mediators in the relation between SES and CHD. Moreover, given evidence of differences by gender in the psychosocial experience of socioeconomic disadvantage (28), in rates of affective disorders (29) and rates of CHD (30), and in associations between SES and CHD (1), we will examine the role of negative affect in the relation between SES and incident CHD by gender in an exploratory fashion.


   METHODS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 
Sample and Study Design
Study participants are respondents to NHANES I, a multistage, national probability survey conducted between 1971 and 1975 on the US civilian noninstitutionalized population aged 1 to 74. The study included oversampling of women of childbearing age, persons living in poverty areas, and elderly persons. The baseline assessment, including a medical examination, blood draw, and an in-person structured interview, was conducted on the full cohort. A detailed medical examination and selected psychological measures were obtained on a representative subsample of noninstitutionalized adults aged 25 to 74 (n = 6913) (31), which comprised the sample for the present investigation. Of those initially contacted for participation, the interview nonresponse rate was 1.4% and the examination nonresponse rate was 30.5%. Interview nonresponders did not differ from participants on any demographic characteristics. However, older age, lower education, and residence in large urban centers were associated with examination nonresponse. Details of study design and sampling procedures are published elsewhere (32).

Follow-up studies (NHEFS) were conducted in 1982, 1987, and 1992 on the entire surviving NHANES I cohort aged 25 to 74 at baseline examination (33–35). The NHEFS 1986 was conducted on only those members aged 55 to 74 at baseline (36). Assessments included either in-person interviews (NHEFS 1982) or automated telephone interviews (NHEFS 1986, 1987, 1992) with respondent or proxy (for decedents), blood pressure and weight measurements (NHEFS 1982), tracking of all members via the National Death Index and obtainment of death certificates (at all follow-ups), and obtainment of records of reported hospital and nursing home stays (at all follow-ups).

This study included members of the detailed subsample (n = 6913). All members of the detailed subsample were traced at one or more of the follow-ups. Of these 6,913 participants, 444 had baseline evidence of cardiovascular disease by self-report or physical examination and were excluded from the analysis. An additional 204 had missing values for one or more covariates. The final sample available for analyses included 6,265 participants (2,853 men, 3,412 women).

SES
Educational attainment was selected as our indicator of SES, given that it is attained relatively early in life and is stable over time. Participants reported their highest level of educational attainment at NHANES I and NHEF 1982 interviews. Education was categorized into four categories: less than high school, high school graduate, some college, and college graduate. Educational levels reported in this sample were unchanged between the interviews. Participants lacking educational data (n = 28) did not significantly differ from the rest of the sample with respect to levels of depressive or anxious symptoms or to risk for CHD.

Psychological Measures
Depressive and anxious symptoms were measured using the General Well-Being Schedule (GWB), a validated measure with known psychometric properties (37). Participants were assessed during NHANES I by trained interviewers blind to study hypotheses. The GWB contains six subscales, two of which were used in these analyses: cheerful versus depressed mood (General Well-Being Schedule Depression subscale, GWB-D) and relaxed versus tense/anxious (General Well-Being Schedule Anxiety subscale, GWB-A). The GWB-D and GWB-A each yield subscale scores ranging from 0 to 25, with low values indicating more depressive or anxious symptoms. Each scale was grouped into three levels utilized (38–40) and validated by previous investigators against CESD clinical cut points (41): scale scores of 0 to 12 indicated high, 13 to 18 indicated moderate, and 19 to 25 indicated low symptomatology.

The overall GWB has sound psychometric properties (37). The subscales, used in the present study, showed strong internal consistency in the present investigation (GWB-D, {alpha} = 0.82; GWB-A, {alpha} = 0.85). They have demonstrated validity, with studies showing that the GWB-D correlates highly with the CESD (r = 0.71 (41)), Zung Depression Scale (r = 0.62 (37)), Personal Feelings Inventory Depression subscale (r = 0.67 (37)), and Psychiatric Symptom Scale Depression subscale (r = 0.70 (37)), whereas the GWB-A correlates highly with Personal Feelings Inventory Anxiety subscale (r = 0.62 (37)) and Psychiatric Symptom Scale Anxiety subscale (r = 0.76 (37)). The GWB-D and GWB-A subscales were correlated at 0.75.

Incident CHD
CHD events were identified by hospital/nursing home discharge reports and death certificates. At each follow-up, participants reported all hospital or nursing home stays since last study contact. Hospitals/nursing homes were then contacted with participant permission and discharge reports obtained for all visits in the study period. Participants were also tracked via the National Death Index, and death certificates were obtained for decedents. A CHD event was coded if International Classification of Diseases, Ninth Revision codes 410 (acute myocardial infarction (MI)), 411 (other acute and subacute ischemic heart disease), and 414 (other forms of chronic ischemic heart disease) were listed on the hospital/nursing home discharge report or as the cause of death on the death certificate. The date of nonfatal CHD events was coded as the discharge date, and if no discharge date was available, the event date was admission date. The date of a fatal CHD event was date of death on the death certificate. For participants with more than one event (e.g., MI followed by CHD death), the earliest event was used.

Covariates
Participants underwent a physical examination, blood draw, and in-person interview at the NHANES I examination. Participants in the detailed sample underwent a more extensive cardiovascular examination. Gender, marital status, smoking status (current versus never/former), leisure time physical activity (sedentary-light, moderate, or regular exercise), and alcohol use (none, up to 2, >2 serving/day) were obtained from NHANES I interview. Age and race/ethnicity (white, nonwhite) obtained in NHANES I interview were updated/corrected in NHEF 1982 to resolve discrepancies between interviews (33), with corrected values used in present analyses. Body mass index (BMI) was calculated as ratio of weight (in kilograms) to standing height (in meters squared) from measures taken in NHANES I physical examination. Diabetes and hypertension status were based on self-reported past or present doctor-diagnosed health conditions and/or past or present medication use for the condition. Systolic blood pressure (SBP) and diastolic blood pressure (DBP) values were derived from one seated measurement taken during NHANES I physical examination.

Statistical Analyses
Follow-up time was calculated as date of baseline interview to date of CHD event, non-CHD death, or date last known alive. Relative risks (RRs) of incident CHD and 95% confidence intervals (95% CIs) associated with education and depressive or anxious symptoms were each estimated separately in multivariate Cox proportional hazards regression (SAS PROC PHREG) to account for unequal follow-up time. The odds of negative affect by educational level were first estimated utilizing multinomial logistic regression in which each of the educational categories (with college as reference) are associated with each of the negative affect categories (with low symptoms as reference). Because nonlinearities were evident, with risk of negative affect associated with lower education concentrated in the high symptom category (data not shown), symptoms were dichotomized as high versus medium/low and examined in relation to educational level using binary logistic regression. Models testing mediation estimated the relation between education and CHD in multivariate Cox proportional hazards regression, controlling for depressive and anxious symptoms each in separate models. The percent of relation between education and incident CHD accounted for by depressive or anxious symptoms was estimated as 1 – log(HRadjusted)/log(HRunadjusted) (42). Each model was estimated adjusting for age and subsequently adjusting for all covariates. Effect modification by gender was examined in all models, and gender-stratified results were reported when a significant or marginally significant interaction was evident. Fully adjusted multivariate Cox proportional hazards regression models included covariates age, gender, marital status, race/ethnicity, smoking status, aerobic exercise, alcohol use, SBP, DBP, BMI, cholesterol, hypertension, and diabetes status. Fully adjusted logistic regression models included covariates age, gender, race/ethnicity, and marital status. Analyses were conducted using SAS V8.2 (SAS Institute, Cary, NC). Models were subsequently estimated to account for the complex survey design, incorporating sample weights, clustering, and stratification within SAS callable version of SUDAAN (Research Triangle Institute, Research Triangle Park, NC). Because findings were largely unchanged, for ease of interpretability results unadjusted for the complex survey design are presented here. In analyses not shown here, we examined effects for fatal and nonfatal events separately. Because the results are similar, we have presented results for fatal and nonfatal events together.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
DISCUSSION
 NOTES
 REFERENCES
 
Descriptive Analyses
Over the follow-up period (M = 15.1, SD = 5.9, range = 0–21.9 years), 1,082 incident CHD events (419 fatal, 663 nonfatal) were recorded via death certificate or hospital/nursing home records. Baseline demographic and psychological characteristics are presented in Table 1 and CHD events by education and negative affect in Table 2.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Characteristics of Study Participants at Baseline by Educational Attainment

 

View this table:
[in this window]
[in a new window]
  		TABLE 2. CHD Events (n,Rate) by Education and Negative Affect (n = 6265)

 

Education and Incident CHD
In Cox proportional hazards models, educational attainment showed a significant and graded inverse relation with incident CHD. In age- and gender-adjusted models, those with less than a high school (RR = 1.82; 95% CI, 1.44–2.31), high school (RR = 1.39; 95% CI, 1.08–1.78), or some college (RR = 1.47; 95% CI, 1.10–1.96) education showed significantly increased risk of incident CHD relative to those with a college education (see Figure 1 for plot of Kaplan-Meier survival function). Education remained a significant predictor of incident CHD in fully adjusted models (less than high school versus college: RR = 1.52; 95% CI, 1.18–1.96; some college versus college: RR = 1.53; 95% CI, 1.13–2.06). A stronger effect was evident among women (interaction between education and gender: p = .03). Relative to those with a college education, the age-adjusted RR of CHD among those with less than a high school education was 2.09 (95% CI, 1.42–3.08) among women and 1.72 (95% CI, 1.26–2.32) among men.


Figure 15
View larger version (23K):
[in this window]
[in a new window]
  		Figure 1. Relative risk of incident CHD associated with educational attainment.

 

Negative Affect and Incident CHD
High depressive and anxious symptoms were each associated with significantly increased risk of CHD. In age- and gender-adjusted models, individuals with high (RR = 1.57; 95% CI, 1.29–1.92) or moderate (RR = 1.20; 95% CI, 1.05–1.37) depressive symptoms or high (RR = 1.60; 95% CI, 1.34–1.90) or moderate (RR = 1.18; 95% CI, 1.03–1.36) anxious symptoms each had significantly increased CHD risk relative to those with low symptoms (Figures 2 and 3 for plots of Kaplan-Meier survival functions). Relations remained significant in fully adjusted models for those with high (RR = 1.30; 95% CI, 1.05–1.61) and moderate depressive symptoms (RR = 1.15; 95% CI, 1.01–1.32), as well as high anxious symptoms (RR = 1.39; 95% CI, 1.16–1.67), relative to those with low symptoms. No significant interaction between depressive and anxious symptoms was observed in age-adjusted (p = .69) or fully adjusted (p = .80) models. Interactions between gender and affect were marginally significant (p = .05). High depressive symptoms were associated with somewhat stronger CHD risk among women (RR = 1.94; 95% CI, 1.49–2.54) versus men (RR = 1.27; 95% CI, 0.93–1.73), adjusting for age.


Figure 25
View larger version (20K):
[in this window]
[in a new window]
  		Figure 2. Relative risk of incident CHD associated with anxious symptoms.

 


Figure 35
View larger version (19K):
[in this window]
[in a new window]
  		Figure 3. Relative risk of incident CHD associated with depressive symptoms.

 
Education and Negative Affect
Lower educational attainment was associated with increased risk of depressive and anxious symptoms. Education was inversely and significantly related to odds of high depressive and anxious symptoms, particularly for those with less than high school education (see Table 3). Anxious symptoms appeared to be associated with educational attainment only among women (less than high school versus college: odds ratio (OR) = 2.45; 95% CI, 1.73–3.47; high school versus college: OR = 1.57; 95% CI, 1.11–2.22, age-adjusted model), with no significant associations among men (interaction between education and gender: p = .0004). Gender differences showed a similar pattern, although not significant, for depressive symptoms (data not shown).


View this table:
[in this window]
[in a new window]
  		TABLE 3. Odds Ratios and Confidence Intervals for High Negative Affect Associated With Education (n = 6265)

 

Negative Affect as a Mediator Between SES and CHD
To evaluate negative affect as a mediator of the effect of education on CHD, depressive and anxious symptoms (high, medium, low) were each entered separately in the models with education and CHD. There was little evidence of mediation. For example, in age- and gender-adjusted models, the RR of CHD associated with less than a high school education was 1.76 and 1.80 when adjusted for depressive and anxious symptoms, respectively, compared with 1.82 in models unadjusted for negative affect (see Table 4). Depressive symptoms accounted for only 4.8% and anxious symptoms less than 1% of the effect of education on CHD incidence in fully adjusted models. Stratifying by gender revealed that mediation did not substantively vary by gender. Interactions between education and depressive (p = .23) and anxious (p = .18) symptoms on CHD risk were not significant.


View this table:
[in this window]
[in a new window]
  		TABLE 4. Association Between Educational Attainment and Risk of CHD (n = 6265)

 

Additional Analyses
To address the possibility that poorer baseline health status among those with lower education and higher negative affect leads to increases in CHD events, all analyses of incident CHD events were repeated excluding the first 3 years of follow-up. Results were largely unchanged (data not shown). Moreover, all analyses were conducted utilizing the continuous measure of depressive and anxious symptoms. In fully adjusted Cox proportional hazards models, depressive (RR = 1.03; 95% CI, 1.01–1.04) and anxious (RR = 1.02; 95% CI, 1.01–1.04) symptoms were associated with increased CHD risk. In linear regression models, less than high school (b = 0.23, p < .0001) and high school (b = 0.08, p = .006) education were significantly associated with increased (log-transformed) depressive symptoms. Significant associations with education were not observed for anxiety, although nonlinearities were notable. In multinomial logistic regression models, increased risk was observed for high anxious symptoms (OR = 1.41; 95% CI, 1.32–1.51) and only for the lowest educational group.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
NOTES
 REFERENCES
 
Low levels of educational attainment and high levels of anxious or depressive symptoms significantly predicted CHD incidence in this study. Moreover, we found lower educational levels to be associated with increased risk of high depressive and anxious symptoms. Thus, three of the criteria set forth by Baron and Kenny (27) were met. However, whereas both educational level and negative affect were predictors of incident CHD, depressive and anxious symptoms did not appear to be mediators by which educational attainment influenced CHD risk.

This study is notable for several reasons. First, it demonstrates increased risk of incident CHD associated with lower SES in a nationally representative, longitudinal study of the US population. Previous research in US samples examined CHD mortality only (3) or examined incident CHD in selected, nonrepresentative samples (7–10). Second, this study is one of few to show depressive symptoms associated with increased risk of incident CHD among both women and men. Third, it is the first to show anxious symptoms associated with increased risk of incident CHD among women using a standardized measure of general anxious symptoms. Finally, this study is the first to examine a mediational role of depressive or anxious symptoms in the association between SES and incident CHD.

Negative affect is a frequently cited pathway linking SES to CHD or other health outcomes (20,23–26). This research is based on the known links between SES and negative affect (13), negative affect and CHD (14,15,17), and SES and CHD (1–3). However, very few studies have included any psychological factors in models examining SES and any health outcome (4,21,43–47). Two of these studies have suggested a partial mediating role of psychological factors for self-rated health (44), incident ulcer (48), or accelerated age-related functional declines associated with low SES (43), although not for total mortality (45). In the case of CHD, one investigation suggested a partial mediational role for a group of psychosocial factors that included depressive symptoms in the association between income and cardiovascular mortality (4) and another study for hostility in the association between SES and cardiovascular reactivity among African American adolescents (21). Two reports among men with CHD participating in the Beta Blocker and Heart Attack Trial indicated a very small mediational role of a group of psychosocial factors in the association between education and functional recovery (46). A larger role was demonstrated for life stress and social isolation, but not depressive symptoms, in the association between education and survival (47). However, most of these studies considered only groups of psychosocial factors together (4,43,44,46,48), and the individual impact of negative affect could not be determined. Many included men only (4,46,47), some did not perform a full mediational analysis (4,48), and none of these studies have examined incident CHD. In short, the present investigation is the first to explicitly consider depression and anxiety as mediators of the association between SES and incident CHD.

Given that our findings satisfied three of the four criteria set forth by Baron and Kenny (27) for mediation, it is somewhat surprising that negative affect failed to sizably attenuate the relation between education and CHD. In the interpretation and explanation of these results, several methodological and conceptual issues merit consideration.

First, the mechanisms by which educational attainment influences CHD development are undoubtedly multifactorial, and it may be unlikely that any single factor would emerge as a strong mediator. For example, cigarette smoking, strongly and inversely associated with SES (11,12) and a leading risk factor for CHD (49), accounted for only 11.7% of CHD risk associated with low education.

It is also possible that the model may not be fully specified. Recent critiques of standard approaches for demonstrating direct versus indirect effects (i.e., evaluating the "direct" effect of educational attainment on incident CHD, after controlling for the "indirect" effect of educational attainment on incident CHD via the pathway negative affect) suggest that they may be flawed (50) or at least valid only under the assumption that negative affect is not a common effect of educational attainment and an unobserved variable. Notably, the Baron and Kenny (27) guidelines for mediation were initially put forth in an experimental context, where a higher degree of control of confounding variables is possible.

The Baron and Kenny (27) guidelines were also initially proposed in the context of linear regression, with linear relations between predictor, mediator, and outcome. In the present study, the relation between education and negative affect was nonlinear. High negative affect occurred primarily at the lowest educational levels. However, such nonlinearity was not evident in the relation between negative affect and CHD. Nonlinearity in one of the criterion relationships is likely to affect our ability to observe mediational effects using models that assume consistent and often linear forms of relationships among its component variables. Thus, this work raises questions regarding the appropriateness of widely used tests of mediation in the context of observational studies with multiple complex and often nonlinear relationships.

Kraemer and colleagues (51) highlight problems with traditional tests of mediation and propose a set of guidelines in the context of multiple complex causal relationships. This framework suggests considering whether negative affect could be a "proxy risk factor" for education and thereby be set aside. Proxy risk factors are determined by comparing several correlated risk factors (e.g., education, negative affect). If these risk factors have no clear temporal sequence and one risk factor has lower predictive power for the outcome than the other, then the weaker risk factor may be considered a proxy to the stronger risk factor. However, conceptually, negative affect cannot be considered a component of education in its relation to CHD. Moreover, in the present study, education and negative affect had independent effects on CHD risk. Thus, this approach does not increase insight into the nature of the relationships under study.

Our findings should be interpreted in light of several limitations. Incident CHD, based on hospital/nursing home discharge reports and death certificates, may have been misclassified. Diagnoses on discharge reports or death certificates also may have been inaccurate, increasing error in the measurement of CHD and downwardly biasing estimates. Moreover, given the examination nonresponse rate, the elderly, less educated, and urban-dwelling individuals participating in this study may be somewhat less representative of the population. Furthermore, the measure of SES chosen was educational attainment because it is attained relatively early in life and is stable throughout life and was the only individual level SES index in the present study. It is possible that income or occupational status, other SES measures, may have a stronger impact on emotional functioning at a given point in time. However, these measures may be more problematic among women (52) and are more vulnerable to reverse causality. Finally, although the GWB subscales have been validated against other symptom measures, it has not been validated against clinical diagnosis and may have measured these constructs with substantial error. The GWB measures depressive and anxious symptoms over the past month, as opposed to clinical depression or anxiety disorders, or lifetime experience of depressive or anxious symptoms. Therefore, one might expect depressive and anxious symptoms in any given month to have weak relations at best to CHD risk over subsequent decades. In this light, the significant relation observed between symptom levels in a given month to incident CHD over 20 years is impressive.

This study has a variety of key methodological strengths. It is a large, longitudinal, nationally representative study of the US population. It included measurement of both depressive and anxious symptoms using a validated measure. It allowed examination of incident CHD, established by hospital/nursing home records and death certificates as opposed to self-report, as well as baseline disease status determined by physical examination.

Our findings suggest that negative affect and educational attainment may represent separate, although related, pathways to increased risk of CHD. Although the strength of the relationship between negative affect and CHD risk was consistent across educational groups, both negative affect and risk of CHD were highest among low-SES groups. Findings suggest that depression, anxiety, and low SES may each be key targets for interventions aimed at the primary and secondary prevention of CHD. Moreover, given the concentration of psychosocial and biomedical risk among low-SES populations, a population perspective would suggest the importance of interventions aimed at these groups.


   NOTES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
REFERENCES
 

This work was supported by the Robert Wood Johnson Foundation (Health and Society Scholars Implementation Grant 045821; Drs. Berkman (PI), Kawachi, Kubzansky, and Thurston) and the Russell Sage Foundation (831003; Dr. Berkman).

DOI:10.1097/01.psy.0000195883.68888.68


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 REFERENCES
 

  1. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation 1993;88:1973–98.[Abstract/Free Full Text]
  2. Rose G, Marmot MG. Social class and coronary heart disease. Br Heart J 1981;45:13–9.[Abstract/Free Full Text]
  3. Rogot E, Sorlie PD, Johnson NJ, Schmitt C. A mortality study of 1.3 million persons by demographic, social, and economic factors: 1979–1985 follow-up, U.S. National Longitudinal Mortality Study, National Institutes of Health. 1992;1–481.
  4. Lynch JW, Kaplan GA, Cohen RD, Tuomilehto J, Salonen JT. Do cardiovascular risk factors explain the relation between socioeconomic status, risk of all-cause mortality, cardiovascular mortality, and acute myocardial infarction? Am J Epidemiol 1996;144:934–42.[Abstract/Free Full Text]
  5. Salomaa V, Niemela M, Miettinen H, Ketonen M, Immonen-Raiha P, Koskinen S, Mahonen M, Lehto S, Vuorenmaa T, Palomaki P, Mustaniemi H, Kaarsalo E, Arstila M, Torppa J, Kuulasmaa K, Puska P, Pyorala K, Tuomilehto J. Relationship of socioeconomic status to the incidence and prehospital, 28-day, and 1-year mortality rates of acute coronary events in the FINMONICA myocardial infarction register study. Circulation 2000;101:1913–8.[Abstract/Free Full Text]
  6. Wamala SP, Mittleman MA, Schenck-Gustafsson K, Orth-Gomer K. Potential explanations for the educational gradient in coronary heart disease: a population-based case-control study of Swedish women. Am J Public Health 1999;89:315–21.[Abstract/Free Full Text]
  7. Frank JW, Cohen R, Yen I, Balfour J, Smith M. Socioeconomic gradients in health status over 29 years of follow-up after midlife: the Alameda county study. Soc Sci Med 2003;57:2305–23.[CrossRef][Medline]
  8. Eaker ED, Pinsky J, Castelli WP. Myocardial infarction and coronary death among women: psychosocial predictors from a 20-year follow-up of women in the Framingham Study. Am J Epidemiol 1992;135:854–64.[Abstract/Free Full Text]
  9. Siegel D, Kuller L, Lazarus NB, Black D, Feigal D, Hughes G, Schoenberger JA, Hulley SB. Predictors of cardiovascular events and mortality in the Systolic Hypertension in the Elderly Program pilot project. Am J Epidemiol 1987;126:385–99.[Abstract/Free Full Text]
  10. Hinkle LE Jr, Whitney LH, Lehman EW, Dunn J, Benjamin B, King R, Plakun A, Flehinger B. Occupation, education, and coronary heart disease: risk is influenced more by education and background than by occupational experiences, in the Bell System. Science 1968;161:238–46.[Free Full Text]
  11. Jacobsen BK, Thelle DS. Risk factors for coronary heart disease and level of education: the Tromso Heart Study. Am J Epidemiol 1988;127:923–32.[Abstract/Free Full Text]
  12. Matthews KA, Kelsey SF, Meilahn EN, Kuller LH, Wing RR. Educational attainment and behavioral and biologic risk factors for coronary heart disease in middle-aged women. Am J Epidemiol 1989;129:1132–44.[Abstract/Free Full Text]
  13. Lorant V, Deliege D, Eaton W, Robert A, Philippot P, Ansseau M. Socioeconomic inequalities in depression: a meta-analysis. Am J Epidemiol 2003;157:98–112.[Abstract/Free Full Text]
  14. Anda R, Williamson D, Jones D, Macera C, Eaker E, Glassman A, Marks J. Depressed affect, hopelessness, and the risk of ischemic heart disease in a cohort of U.S. adults. Epidemiology 1993;4:285–94.[Medline]
  15. Barefoot JC, Schroll M. Symptoms of depression, acute myocardial infarction, and total mortality in a community sample. Circulation 1996;93:1976–80.[Abstract/Free Full Text]
  16. Barefoot JC, Helms MJ, Mark DB, Blumenthal JA, Califf RM, Haney TL, O’Connor CM, Siegler IC, Williams RB. Depression and long-term mortality risk in patients with coronary artery disease. Am J Cardiol 1996;78:613–7.[CrossRef][Medline]
  17. Kubzansky LD, Kawachi I, Spiro A 3rd, Weiss ST, Vokonas PS, Sparrow D. Is worrying bad for your heart? a prospective study of worry and coronary heart disease in the Normative Aging Study. Circulation 1997;95:818–24.[Abstract/Free Full Text]
  18. Kawachi I, Colditz GA, Ascherio A, Rimm EB, Giovannucci E, Stampfer MJ, Willett WC. Prospective study of phobic anxiety and risk of coronary heart disease in men. Circulation 1994;89:1992–7.[Abstract/Free Full Text]
  19. Kawachi I, Sparrow D, Vokonas PS, Weiss ST. Symptoms of anxiety and risk of coronary heart disease: the Normative Aging Study. Circulation 1994;90:2225–9.[Abstract/Free Full Text]
  20. Gallo LC, Matthews KA. Understanding the association between socioeconomic status and physical health: do negative emotions play a role? Psychol Bull 2003;129:10–51.[CrossRef][Medline]
  21. Gump BB, Matthews KA, Raikkonen K. Modeling relationships among socioeconomic status, hostility, cardiovascular reactivity, and left ventricular mass in African American and white children. Health Psychol 1999;18:140–50.[CrossRef][Medline]
  22. Marmot MG, Bosma H, Hemingway H, Brunner E, Stansfeld S. Contribution of job control and other risk factors to social variations in coronary heart disease incidence. Lancet 1997;350:235–9.[CrossRef][Medline]
  23. Adler NE, Ostrove JM. Socioeconomic status and health: what we know and what we don’t. Ann N Y Acad Sci 1999;896:3–15.[CrossRef][Medline]
  24. Taylor SE, Seeman TE. Psychosocial resources and the SES-health relationship. Ann N Y Acad Sci 1999;896:210–25.[CrossRef][Medline]
  25. Kaplan GA. Where do shared pathways lead? some reflections on a research agenda. Psychosom Med 1995;57:208–12.[Abstract/Free Full Text]
  26. House JS, Williams DR. Psychosocial Pathways Linking SES and CVD: Chartbook of U.S. National Data on Socioeconomic Status and Cardiovascular Health and Disease. Bethesda, MD: National Heart, Lung, and Blood Institute; 1995:119–24.
  27. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173–82.[CrossRef][Medline]
  28. Krieger N. Discrimination and health. In: Berkman LF, Kawachi I, eds. Social Epidemiology. New York: Oxford University Press; 2000:36–75.
  29. Nolen-Hoeksema S. Sex differences in unipolar depression: evidence and theory. Psychol Bull 1987;101:259–82.[CrossRef][Medline]
  30. Mosca L, Manson JE, Sutherland SE, Langer RD, Manolio T, Barrett-Connor E. Cardiovascular disease in women: a statement for healthcare professionals from the American Heart Association Writing Group. Circulation 1997;96:2468–82.[Free Full Text]
  31. Engel A, Murphy R, Mauer K, Collins E. Plan and Operation of the NHANES I Augmentation Survey of Adults 25–74 Years, United States, 1971–1975, Vital Health Statistics 1, No. 14. Hyattsville, MD: National Center for Health Statistics; 1978:1–110.
  32. Plan and Operation of the Health and Nutrition Examination Survey, United States, 1971–1973, Vital Health Statistics 1, No. 10a. Hyattsville, MD: National Center for Health Statistics; 1973:1–76.
  33. Cohen B, Barbano H, Cox C, Feldman J, Finucane F, Kleinman J, Madans J. Plan and Operations of the NHANES I Epidemiologic Follow-up Study 1982–84, Series 1, No. 22. Hyattsville, MD: National Center for Health Statistics; 1987:1–141.
  34. Cox C, Mussolino M, Rothwell S, Lane M, Golden C, Madans J, Feldman J. Plan and Operations of the NHANES I Epidemiologic Follow-up Study 1992, Vital Health Statistics 1, No. 35. Hyattsville, MD: National Center for Health Statistics; 1997:1–231.
  35. Cox C, Rothwell S, Madans J, Finucane F, Freid V, Kleinman J, Barbano H, Feldman J. Plan and Operations of the NHANES I Epidemiologic Follow-up Study 1987, Vital Health Statastics 1, No. 27. Hyattsville, MD: National Center for Health Statistics; 1992:1–190.
  36. Finucane F, Freid V, Madans J, Cox C, Kleinman J, Rothwell S, Barbano H, Feldman J. Plan and Operations of the NHANES I Epidemiologic Follow-up Study 1986, Series 1, No. 25. Hyattsville, MD: National Center for Health Statistics; 1990:1–154.
  37. Fazio AF. A concurrent validational study of the NCHS General Well-Being Schedule. Vital Health Stat 1977;2:1–53.
  38. Carnethon MR, Kinder LS, Fair JM, Stafford RS, Fortmann SP. Symptoms of depression as a risk factor for incident diabetes: findings from the National Health and Nutrition Examination Epidemiologic Follow-up Study, 1971–1992. Am J Epidemiol 2003;158:416–23.[Abstract/Free Full Text]
  39. Jonas BS, Lando JF. Negative affect as a prospective risk factor for hypertension. Psychosom Med 2000;62:188–96.[Abstract/Free Full Text]
  40. Jonas BS, Mussolino ME. Symptoms of depression as a prospective risk factor for stroke. Psychosom Med 2000;62:463–71.[Abstract/Free Full Text]
  41. Zonderman AB, Costa PT Jr, McCrae RR. Depression as a risk for cancer morbidity and mortality in a nationally representative sample. JAMA 1989;262:1191–5.[Abstract/Free Full Text]
  42. Lin DY, Fleming TR, De Gruttola V. Estimating the proportion of treatment effect explained by a surrogate marker. Stat Med 1997;16:1515–27.[CrossRef][Medline]
  43. House JS, Lepkowski JM, Kinney AM, Mero RP, Kessler RC, Herzog AR. The social stratification of aging and health. J Health Soc Behav 1994;35:213–34.[CrossRef][Medline]
  44. Cohen S, Kaplan GA, Salonen JT. The role of psychological characteristics in the relation between socioeconomic status and perceived health. J Appl Soc Psychol 1999;29:551–74.
  45. Fiscella K, Franks P. Does psychological distress contribute to racial and socioeconomic disparities in mortality? Soc Sci Med 1997;45:1805–9.
  46. Ickovics JR, Viscoli CM, Horwitz RI. Functional recovery after myocardial infarction in men: the independent effects of social class. Ann Intern Med 1997;127:518–25.[Abstract/Free Full Text]
  47. Ruberman W, Weinblatt E, Goldberg JD, Chaudhary BS. Psychosocial influences on mortality after myocardial infarction. N Engl J Med 1984;311:552–9.[Abstract]
  48. Levenstein S, Kaplan GA. Socioeconomic status and ulcer: a prospective study of contributory risk factors. J Clin Gastroenterol 1998;26:14–7.[CrossRef][Medline]
  49. Jonas MA, Oates JA, Ockene JK, Hennekens CH. Statement on smoking and cardiovascular disease for health care professionals: American Heart Association. Circulation 1992;86:1664–9.[Free Full Text]
  50. Cole SR, Hernan MA. Fallibility in estimating direct effects. Int J Epidemiol 2002;31:163–5.[Abstract/Free Full Text]
  51. Kraemer HC, Stice E, Kazdin A, Offord D, Kupfer D. How do risk factors work together? mediators, moderators, and independent, overlapping, and proxy risk factors. Am J Psychiatry 2001;158:848–56.[Abstract/Free Full Text]
  52. Berkman LF, Macintyre S. The measurement of social class in health studies: old measures and new formulation. In: Kogevinas M, Pearce N, Susser M, Boffetta P, eds. Social Inequalities and Cancer. Lyon: International Agency for Research on Cancer; 1997.



This article has been cited by other articles:


Home page
 Psychosom. Med.Home page
R. C. Thurston and L. D. Kubzansky
Women, Loneliness, and Incident Coronary Heart Disease
Psychosom Med, October 1, 2009; 71(8): 836 - 842.
[Abstract] [Full Text] [PDF]

Home page
 AJPHHome page
B. Mezuk, W. W. Eaton, S. H. Golden, and Y. Ding
The Influence of Educational Attainment on Depression and Risk of Type 2 Diabetes
Am J Public Health, August 1, 2008; 98(8): 1480 - 1485.
[Abstract] [Full Text] [PDF]

Home page
 Psychosom. Med.Home page
S. S. Pedersen, O. R. F. Smith, J. D. Vries, A. Appels, and J. Denollet
Course of Anxiety Symptoms Over an 18-Month Period in Exhausted Patients Post Percutaneous Coronary Intervention
Psychosom Med, April 1, 2008; 70(3): 349 - 355.
[Abstract] [Full Text] [PDF]

Home page
 Am J EpidemiolHome page
P. Rajan, K. T. Kelsey, J. D. Schwartz, D. C. Bellinger, J. Weuve, D. Sparrow, A. Spiro III, T. J. Smith, H. Nie, H. Hu, et al.
Lead Burden and Psychiatric Symptoms and the Modifying Influence of the {delta}-Aminolevulinic Acid Dehydratase (ALAD) Polymorphism: The VA Normative Aging Study
Am. J. Epidemiol., December 15, 2007; 166(12): 1400 - 1408.
[Abstract] [Full Text] [PDF]

Home page
 J. Epidemiol. Community HealthHome page
J. Ahnquist, P. Fredlund, and S. P Wamala
Is cumulative exposure to economic hardships more hazardous to women's health than men's? A 16-year follow-up study of the Swedish Survey of Living Conditions
J Epidemiol Community Health, April 1, 2007; 61(4): 331 - 336.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Thurston, R. C.
Right arrow Articles by Berkman, L. F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Thurston, R. C.
Right arrow Articles by Berkman, L. F.
Related Collections
Right arrow Social Class
Right arrow Anxiety
Right arrow Depression
Right arrow Other Cardiovascular Medicine

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS