Women's Health Issues
Volume 15, Issue 2 , Pages 80-88, March 2005

The association between educational level and risk of cardiovascular disease fatality among women with cardiovascular disease

  • JiWon R. Lee, MS, RD, MPH

      Affiliations

    • College of Physicians and Surgeons, Columbia University; New York Presbyterian Hospital, New York, New York
    • ,
  • Furcy Paultre, PhD

      Affiliations

    • College of Physicians and Surgeons, Columbia University; New York Presbyterian Hospital, New York, New York
    • ,
  • Lori Mosca, MD, PhD

      Affiliations

    • College of Physicians and Surgeons, Columbia University; New York Presbyterian Hospital, New York, New York
    • Corresponding Author InformationCorrespondence to: Dr. Lori Mosca, New York Presbyterian Hospital, 622 West 168th Street, PH 10-203B, New York, NY 10032.

Received 14 October 2003; received in revised form 27 February 2004; accepted 12 November 2004.

Article Outline

Background

The inverse relation of socioeconomic status with incident cardiovascular diseases (CVDs) has been well established. However, few data are available describing this relation among ethnically diverse women with prevalent CVD. Using education as a proxy for socioeconomic status, we examined its relation to CVD mortality among women with established CVD.

Subjects

Data from 2,157 women with CVD at baseline, who participated in nine long-term U.S. cohort studies, were pooled.

Methods

Cox regression models adjusted for history of diabetes mellitus, total cholesterol, systolic and diastolic blood pressure, body mass index, smoking, race/ethnicity, and age at baseline were used to estimate hazard ratios for CVD mortality between non-high school graduates and high school graduates.

Results

During a mean follow-up time of 11.5 years, 615 CVD deaths were observed. There was an age-dependent (p = .003) inverse association between education and CVD mortality among women with CVD. At age 60, the risk of dying due to CVD among non-high school graduates was more than twice greater than that of high school graduates (hazard ratio = 2.34; 95% CI 1.27–4.29). At age 65, the hazard ratio decreased to 1.31 (95% CI 1.00–1.71). By age 70, there was no difference in the hazard of dying between high school graduates and nongraduates (hazard ratio = 1.01; 95% CI .85–1.21).

Conclusions

Our results show that among women with CVD, educational level was a significant, and age-dependent, predictor of fatal CVD independent of other traditional risk factors. These women are an important high-risk population to target secondary prevention and educational efforts.

 

Back to Article Outline

Introduction 

Socioeconomic status (SES) has been shown to be an independent risk factor for incidence of cardiovascular disease (CVD) since 1960 (Marmot, Shipley, & Rose, 1984; Lynch et al.,. 1996; Diez-Roux et al., 1995). Several studies have found an inverse association between SES and CVD mortality among middle-aged adults (Feldman et al., 1989; Mackenbach et al., 2000; Otten & Bosma, 1997; Salomaa et al., 2000; Shkolnikov et al., 1998). Education is one of the most commonly used proxies for SES (Kaplan & Keil, 1993; Kitagawa & Hauser, 1973) because educational level is simple to collect, is associated with a high response rate, is minimally affected by later health status (reverse causation), and remains relatively constant (Kaplan & Keil, 1993; Kitagawa & Hauser, 1973; Bassuk, Berkman, & Amick, 2002).

Many epidemiological studies have demonstrated the inverse association of education with incident CVD mortality including coronary heart disease (CHD) deaths (Marmot, Shipley, & Rose, 1984; Lynch et al., 1996; Diez-Roux et al., 1995; Feldman et al., 1989; Mackenbach et al., 2000; Otten & Bosma, 1997; Salomaa et al., 2000; Shkolnikov et al., 1998; Kitagawa & Hauser, 1973; Bassuk, Berkman, & Amick, 2002; Elo & Preston, 1996; Hemingway et al., 2000; Mustard et al., 1997; Wamala & Orth-Gomer, 1998; Gonzalez, Rodriguez Artalejo, & Calero, 1998; Tenconi, Devoti, & Comelli, 2000; Liu et al., 1982; Antonovsky, 1967). Gonzalez et al. (1998) and Mackenbach et al. (2000) showed a significant inverse association between education and CVD using internationally pooled data. The inverse association between education and CHD mortality was also shown in middle-aged Italian men (40–69 years; Tenconi, Devoti, & Comelli, 2000) and whites in the United States (40–59 years; Liu et al., 1982). In addition, the protective effect of education on CVD mortality has been demonstrated in large population-based epidemiological studies (Feldman et al., 1989; Otten & Bosma, 1997). In Dutch adults (≥19 years), the protective impact of education on heart disease mortality was more apparent in middle-aged and older people (≥40 years) compared to adults in general (Otten & Bosma, 1997). The inverse association was also shown among U.S. whites aged 45–74 years with a clear dose-response (Feldman et al., 1989). Studies of the relation between education and recurrent CHD are limited.

Several potential mechanisms may explain the association between education and CVD mortality. Education is inversely associated with established CVD risk factors such as blood pressure, lipid levels, and obesity (Feldman et al., 1989; Gran, 1995; Luepker et al., 1993; Garrison et al., 1993; Cirera et al., 1998; Jacobsen & Thelle, 1988; Matthews et al., 1989; Gupta, Gupta, & Ahluwalia, 1994; Pekkanen et al., 1995). In addition, the association between education and CHD may be mediated through psychosocial factors. Psychosocial factors are differentially distributed across education (Hemingway et al., 2000; Wamala et al., 1999). Individuals with low educational levels tend to have jobs with low responsibility and high demand, which may increase stress levels. Release of stress hormones could alter hemodynamics, lipid metabolism, homeostasis, and other factors, increasing susceptibility to CVD (Feldman et al., 1989; Hemingway et al., 2000; Wamala et al., 1999). Lastly, education as a proxy for SES may be related to access to medical care (Feldman et al., 1989).

Most studies of educational level and CVD have been limited to populations free of CVD at baseline (Feldman et al., 1989; Mackenbach et al., 2000; Otten & Bosma, 1997; Salomaa et al., 2000; Shkolnikov et al., 1998; Bassuk, Berkman, & Amick, 2002). Few data have estimated this relation in women with prevalent CVD. The purpose of this study was to evaluate the relation between education and CVD mortality among ethnically diverse adult women with CVD and to determine if differences in traditional risk factors explain any difference in CVD mortality based on educational status. Prevalent CVD, in this study, was composed of a history of CHD, defined by a baseline record of angina and/or myocardial infarction (MI), and/or a history of stroke at baseline according to cohort-specific definitions and their diagnostic criteria (Appendix 1).

Appendix 1. Diagnostic criteria for history of MI, angina, and stroke in each cohort
CohortPrevious MIPrevious AnginaPrevious Stroke
ARICECG evidence and medical historyRose questionnaireDoctor’s diagnosis, “Has a doctor ever told you had stroke?”
CharlestonSelf-reported history and ECG evidenceSelf-reported historyNA
Evans’ CountyReview panelReview panelReview panel
FraminghamReview panelChest pain questionnaire and review panelPhysician or hospital record
NHEFSSelf reported, “Have a doctor ever told you—?”NADoctor’s diagnosis, “Has a doctor told that you had condition”
Framingham OffspringSame method as Framingham Same method as Framingham
Rancho BernadoSelf-reported, “Have you ever been hospitalized for a heart attack?” and review panel validated the resultsNASelf-reported, “Have you ever had a stroke?”
San AntonioSelf-reported, “Has a doctor ever told you—” or Minnesota-coded ECGRose questionnaireSelf-reported stroke
TecumsehSelf-reported, “Has a doctor ever told you had a heart attack?” or diagnosis from interview data or ECGChest pain questionnaire and review panelSelf-reported, “Have you ever had a stroke?”

Abbreviations: MI, myocardial infarction; ECG, electrocardiogram; NA, not applicable.

Back to Article Outline

Methods 

Study sample 

Data were from the Women’s Pooling Project (WPP), which is the combined data from nine U.S. population-based prospective cohort studies (Atherosclerosis Risk in Communities Study [ARIC], Charleston Heart Study, Evans County Study, Framingham Heart Study, Framingham Offspring Study, National Health and Nutrition Examination Survey Epidemiologic Follow-up Study [NHEFS], Rancho Bernardo Study, San Antonio Heart Study, and Tecumseh Community Health Study). Details of sampling procedures, study designs, and methods for each cohort have been published elsewhere (The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives, 1989; Jackson et al., 1996; National Heart, Lung and Blood Institute, 1987; Bolye, 1970; Keil et al., 1984; Johnson et al., 1986; McDonough et al., 1965; Gordon & Shurtleff, 1973; Cohen et al., 1987; National Center for Health Statistics, 1973; Criqui, Barrett-Connor, & Austin, 1978; Hazuda et al., 1986; Epstein et al., 1965). The sample included in this analysis includes 2,157 women with CVD. Women were included in this study if at baseline they were 30 years or older with prevalent CVD and had data available on history of diabetes, blood pressure, total cholesterol, smoking status, height, and weight.

Definitions of baseline variables 

Educational attainment for each subject was dichotomized as high school graduate or not. This was done because high school graduate was available in all cohorts. All cohorts defined subject completion of 12 years of education as a high school graduate except for the Charleston and Evans county studies, where 11 years was required for high school graduation. History of diabetes was identified by serum glucose level or treatment in all cohorts except NHEFS and Charleston study, which used self-reported physician diagnosis, and ARIC study, which combined both criteria. Smoking was classified as current smoker or not. Nonsmokers included those who never or formerly smoked. Baseline age and race/ethnicity were assessed by cohort-specific questionnaires. Total cholesterol was measured according to cohort-specific methods. Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were assessed as the mean of the last two readings unless only one reading was available, in which case the single reading was used. Body mass index (BMI) was calculated as the weight in kilograms divided by the height in meters squared.

Assessment of CVD death 

CVD mortality was identified based on the International Classification of Disease (ICD) codes 401–448 (National Center for Health Statistics, 1967; World Health Organization, 1977). Most of the studies used ICD code version 9, except the Charleston, Evans County, and Framingham studies whose baseline examination were done in the 1960s and used ICD code version 8.

Statistical analysis 

Baseline characteristics of high school graduates and non-high school graduates were compared using χ2 tests for categorical variables and two-sided t-tests for continuous variables. A p value <.05 was considered significant.

To examine the association between educational level and age with CVD mortality, Cox regression models using age as the time axis with left truncation at baseline age were used (Allison, 1995). Women not recorded as dying from CVD were treated as censored on the last day recorded alive. First, Cox proportional hazards regression models were used to estimate age-specific hazard ratios (HR) of non-high school graduates to high school graduates for the midpoint (N) of 5-year intervals. Each group included all women who participated in the study during the 5-year interval age group N ± 2.5 years. Events that occurred after this interval were treated as censored at the end of the interval.

Next we used nonproportional hazards regression models by introducing an interaction term between age and educational status in Cox models to account for the age-dependent effect of education on CVD mortality risk. The addition of an interaction between education and age in the model allowed us to calculate the age-specific effect of education on CVD fatality continuously. The nonproportional regression model was restricted to age groups N = 60 to N = 90 years, because of the small number of CVD deaths (<30 events) in the extreme age groups (N ≤ 55 years and N ≥ 95 years). The −2 log likelihood ratio statistic, a measure of model agreement with data, was used to identify the best-fitted nonproportional Cox regression model (Collett, 1994). The significance of the interaction term was determined by the difference between the −2 log likelihood (which has a χ2 distribution with 1 degree of freedom) in models before and after the interaction term was added. Both baseline age-adjusted HR and adjusted HR (controlled for history of diabetes, total cholesterol, SBP, DBP, BMI, smoking and race/ethnicity, in addition to baseline age) were calculated for both proportional and nonproportional hazards regression models. To identify the significance of association, the HR and its 95% confidence interval (CI) were used. When the 95% CI did not include 1, it was considered significant.

Lastly, secondary analyses were done to examine potential cohort effects on the association using stratification in the nonproportional regression model (Collett, 1994). Also, general trends of the relationship between education and CVD mortality in women with CVD were explored in each cohort individually to identify any specific cohort influence on the overall pattern of the association. Cohorts with less than 5% of CVD deaths in the age groups between N = 60 and N = 90 years were excluded for this analysis (Charleston Heart Study, Evans County Study, Framingham Offspring, and San Antonio Heart Study) because of their small number of events. All analyses were done using SAS version 8.01 (SAS Institute Inc., 1999).

Back to Article Outline

Results 

Table 1 describes baseline characteristics of 2,157 women with established CVD by educational status. During a mean follow-up of 11.5 ± 7.4 years (24,863 person-time years), 615 CVD deaths were observed. High school graduates, compared to non-high school graduates, were younger (57 ± 9 versus 59 ± 10 years; p < .0001), were less likely to be diabetic (12% versus 18%; p < .0001), and had lower means for traditional risk factors: total cholesterol (6.09 ± 1.27 versus 6.28 ± 1.28 mmol/L; p = .0007), SBP (135 ± 28 versus 146 ± 31 mm Hg; p < .0001), DBP (79 ± 15 versus 84 ± 16 mm Hg; p < .0001), and BMI (27 ± 6 versus 29 ± 7 kg/m2; p < .0001). Moreover, the distribution of race/ethnicity between high school graduates and nongraduates were significantly different (p < .0001); specifically, high school graduates were more likely to be white (86% versus 70%; p < .0001).

Table 1. Baseline characteristics among women with cardiovascular disease
Non-HS Graduate (n = 1027)HS Graduate (n = 1130)p value
Baseline age (years) (mean ± SD)59±1057±9<.0001
≥65 years, n (%)335(33)226(20)<.0001
Race/ethnicity
White, n (%)723(70)971(86)
Black, n (%)212(21)123(10.9)
Hispanic, n (%)89(8.7)35(3)
Other, n (%)3(0.3)1(0.1)<.0001a
Disease history
Coronary heart disease, n (%)804(84)915(86).28
Stroke, n (%)300(30)289(26).037
Diabetes mellitus, n (%)186(18)136(12)<.0001
Cholesterol (mmol/L) (mean ± SD)6.28±1.286.09±1.27.0007
SBP (mm Hg) (mean ± SD)146±31135±28<.0001
DBP (mm Hg) (mean ± SD)84±1679±15<.0001
BMI (kg/m2) (mean ± SD)29±727±6<.0001
Smoker, n (%)286(28)301(27).53

Abbreviations: HS, high school; SD, standard deviation; CHD, coronary heart disease; SBP, systolic blood pressure, DBP, diastolic blood pressure; BMI, body mass index.

a The proportion of race/ethnicity between HS graduates and non-HS graduates.

Age-specific HR from Cox proportional hazards regression models, adjusted for baseline age and traditional risk factors (total cholesterol, SBP, DBP, history of diabetes, BMI, smoking status, and race/ethnicity) in addition to baseline age, are shown in Table 2. The age-specific effect was calculated in each 5-year interval age group as an estimate for the HR of the midpoint (N) in each group. The association between education and CVD fatality was greatest in age group N = 60 in both age-adjusted (HR = 2.16; 95% CI 1.09–4.27) and fully adjusted models (HR = 2.11; 95% CI 1.03–4.33). But, it became not very different from a hazard ratio of 1 by age group N = 70.

Table 2. Age-specific cardiovascular disease mortality hazard ratios of non-high school graduates to high school graduates among women with cardiovascular disease at baseline
Age Group (Na)CVD DeathsNumber of SubjectsHR (95% CI)bHR (95% CI)c
452263——
50105671.04(0.29–3.69)0.80(0.21–2.99)
55178840.81(0.30–2.19)0.64(0.23–1.75)
603610872.16(1.09–4.27)2.11(1.03–4.33)
654711461.85(1.02–3.36)1.76(0.94–3.29)
708610701.05(0.69–1.61)0.97(0.62–1.52)
751078081.00(0.68–1.47)0.91(0.61–1.37)
801266040.95(0.67–1.35)0.89(0.61–1.30)
851063720.76(0.52–1.12)0.71(0.47–1.07)
90571590.97(0.57–1.65)0.88(0.49–1.60)
9518360.84(0.33–2.14)0.74(0.24–2.26)
10034——

Abbreviations: CVD, cardiovascular disease; HR, hazards ratio; CI, confidence interval.

a Age group N included all women who participated in the study during the 5-year interval N ± 2.5 years. Events that occurred after the interval were treated as censored at the end of the interval.

b Baseline age-adjusted hazard ratio.

c Hazard ratio adjusted for age at baseline, history of diabetes mellitus, total cholesterol, systolic and diastolic blood pressure, body mass index, smoking status, and race/ethnicity.

We next modeled the change in association between education and CVD fatality in a continuous manner using an interaction term between education and age in a Cox nonproportional hazards model. Because few CVD deaths (<30 events) occurred in the extreme age groups (N ≤ 55 and N ≥ 95 years), the analysis was restricted to age groups N = 60 to N = 90 years. The decrease in the −2 log likelihood statistics was significant (χ2 = 9.3 with 1 degree of freedom; p = .002) when the interaction term was included. Therefore, we used the nonproportional hazards model to identify a continuous association between education and CVD fatality in women with CVD. The inverse association of education with CVD fatality in women with established CVD was summarized using both proportional and nonproportional models, adjusted for traditional risk factors in age groups N = 60 to N = 90 years in Figure 1. The nonproportional Cox regression model (solid line) with its 95% CI (dotted line) shows the continuous change of the effect of education on CVD mortality with age, while the results of the proportional model, for each 5-year age group, is represented by solid circles for each midpoint N. The inverse association of education with CVD mortality among women with CVD was age-dependent. In the continuous model, non-high school graduates had 2.34 times higher risk of dying due to CVD than high school graduates (95% CI 1.27–4.29) at age 60 years. By age 65 years the hazard ratio decreased to 1.31 (95% CI 1.00–1.71), and by age 70 years, the risk of dying due to CVD was similar for non-high school graduates and high school graduates (HR = 1.01; 95% CI .85–1.21). These results remained similar when women with angina only (n = 948) were excluded, with hazard ratios of 3.03, 1.27, and .86 for ages 60, 65, and 70, respectively (p = .0021 for the inclusion of the interaction term).

  • View full-size image.
  • Figure 1. 

    Adjusted age-specific cardiovascular disease mortality hazard ratios of non-high school graduates to high school graduates among women with cardiovascular disease. Hazard ratios calculated from Cox regression models adjusted for baseline age, history of diabetes mellitus, body mass index, total cholesterol, systolic and diastolic blood pressure, smoking status, and race/ethnicity at baseline. Continuous line derived from nonproportional hazards model containing an interaction term between education and age. Dotted line represents the 95% confidence interval for nonproportional hazards model. Solid circles derived from proportional hazards models for the midpoint (N) in 5-year interval age groups.

The potential cohort effects on the association of education with CVD fatality among women with CVD were examined using stratification in the nonproportional Cox regression model. The inverse relationship between education and CVD mortality remained significant (p = .02). In addition, we examined the association in each cohort individually to identify any specific cohort influence on the overall trend. Each cohort had similar patterns of the inverse association (see Appendix 2 and Figure 2). The significant age-dependent inverse relationship of education with CVD fatality among women with CVD seems to be determined by the general trend of association in this study, not due to cohort effects.

  • View full-size image.
  • Figure 2. 

    General trends of the association between educational level and cardiovascular disease mortality among women with cardiovascular disease across cohorts. Six lines including overall trend derived from cohort-specific nonproportional hazard models containing an interaction term between education and age and adjusted for baseline age, history of diabetes mellitus, body mass index, total cholesterol, systolic and diastolic blood pressure, smoking status, and race/ethnicity at baseline.

Back to Article Outline

Discussion 

This is one of the first studies to determine the association between education and CVD fatality among women with CVD. In this study, there was a significant interaction between education and age in predicting CVD mortality, with high school graduates showing a protective effect at age 60 years that decreased and disappeared by age 70 years. The age-dependent inverse association between education and CVD mortality persisted across race/ethnicity.

Recently, two studies showed a significant inverse association of SES with CHD mortality regardless of baseline disease status (Salomaa et al., 2000; Hemingway et al., 2000). First, Hemingway et al. (2000) identified the inverse association between occupational gradient and CHD mortality among middle-aged (40–69 years) British civil servants. This study demonstrated that the inverse association was more significant among civil servants without CHD at baseline than those with. The risk of CHD mortality in the lowest grade of occupation was increased by 72% (95% CI 1.4–2.1) compared to that in the highest among men without CHD, while the increase was 56% (95% CI 1.1–2.1) among those with CHD. In addition, Salomaa et al. (2000) confirmed the finding from Hemingway et al. (2000) using the data from the Finnish contribution to the World Health Organization (WHO) multinational monitoring of trends and determinants of cardiovascular disease (FINMONICA) MI registry during 1983–1992. Among middle-aged (35–64 years) Finnish adults, there was a significant inverse association between education and MI death, regardless of prevalence of MI. In contrast to Hemingway et al. (2000), the protective effect of education in this study was more significant among people with MI than without across gender. Furthermore, this study showed that the protective effect of education on MI case-fatality was much stronger in women than in men. Less-educated men had 87% higher risk for a 1-year case-fatality than high-educated men (95% CI 1.71–2.05), while the risk in less-educated women was increased by 134% compared to the risk among high-educated women (95% CI 1.88–2.92).

Our data are consistent with other studies showing that the impact of education on CVD mortality is age-dependent. Antonovsky et al. (1967) evaluated that the impact of SES on health was most prominent in young and middle-aged adults. In addition, the impact disappeared with advancing age (aged ≥70 years), regardless of race/ethnicity. The decrease in the hazard ratio of education with advanced age may be due to the increasing relative importance of the background rate of CVD and other risk factors.

Relative risk peaked at aged 30–44 years, then started declining and disappeared at age 65 years and older. Also, the inverse association has been found among the middle-aged and then disappeared after age 65 years (Elo & Preston, 1996; Mustard et al., 1997; Kitagawa & Hauser, 1973). But a recent study showed that the protective effect of education on overall mortality persisted beyond middle-age, even though the strength of association was weaker than that in middle-aged (Bassuk, Berkman, & Amick, 2002). Community-living elderly men (aged ≥65 years) with low education (≤7 years) had 44% increased risk of dying, compared to those with the highest (≥13 years) (95% CI 1.07–1.94).

Study strength and limitations 

Although our dataset had several advantages, compared to previous studies, which allowed us to identify the association of education with CVD fatality among adult women with CVD and to test the significance of the interaction between education and age among them, there were also several limitations. Due to the small number of CVD deaths (<30 events) in the youngest and oldest age groups (N ≤ 55 and N ≥ 95 years), they were not included in this analysis. Measurement errors due to differences in methodologies between cohorts might have biased our results, but when data were examined within cohorts, the results were similar. Lastly, although commonly used, education may not be a good proxy for SES.

Back to Article Outline

Conclusions 

We found that an inverse association between educational level and CVD mortality is evident among women with prevalent CVD. Moreover, there was a significant interaction between education and age among women with CVD. The inverse association between educational level and CVD mortality was greatest in women at age 60 years, decreased with age, and disappeared by age 70 years. Considering that women have higher prevalence of CVD and CVD case-fatality than men with increasing age (≥65 years; American Heart Association, 2003; Lerner & Kannel, 1986; Marrugat et al, 2001; Vaccarino, Berkman, & Krumholz, 2000), the significant protective effect of education from our study might give a novel strategy to reduce CVD mortality among women with CVD. Especially for middle-aged women with low education, increasing heart health education program may improve their awareness of CVD risk and then result in modifying lifestyle change by continuous education, which further leads to decrease CVD mortality gap between high school graduate and non-high school graduate among middle-aged women. More intensive preventive and educational efforts may be needed to target less-educated women across race/ethnicity by health professionals and public health policy makers to maximize survival from CVD in women. This may include an emphasis on blood pressure, blood lipid, and weight control. In addition, more studies are needed to identify this association across race/ethnicity among adult women.

Back to Article Outline

Acknowledgements 

This study was supported by a grant from the National Heart, Lung and Blood Institute (KO8 HL 03681) and by a grant from the American Heart Association (9750703N). The authors would like to thank the steering committee members of the Women’s Pooling Project for their participation in the project, including Ralph D’Agostino, Boston University; Elizabeth Barrett-Connor, University of California San Diego; Victor Hawthorne, University of Michigan; Millicent Higgins, University of Michigan; William Kannel, Framingham Study; Julian Keil, Medical University of South Carolina; Michael Stern, University of Texas, University of Maryland School of Medicine; Susan Sutherland, Research Institute at Mission St. Joseph’s; H. Al Tyroler, University of North Carolina; and Aaron Folson, University of Minnesota.

Back to Article Outline

Appendix 2. 

Assessing general patterns of the association across cohorts 

General trends of the inverse association between educational level and CVD fatality in women with CVD were examined to identify any specific cohort influence on the overall pattern (Figure 2). Cohorts with at least 5% of CVD deaths in the age groups N = 60 to N = 90 years (ARIC, NHEFS, Rancho Bernardo, and Tecumseh studies) were included in this analysis, except for the Framingham study, which only included N = 65 to N = 90 years because of the small number of events in the age group N = 60 years (7/247 = 2.8%). Nonproportional hazard models containing an interaction term between education and age derived from each cohort including overall trend and adjusted for baseline age, history of diabetes mellitus, body mass index, total cholesterol, systolic and diastolic blood pressure, smoking status, and race/ethnicity at baseline. All five cohorts had similar patterns of the inverse association between education and CVD fatality and the variance of curvature in each cohort might be due to their different events and sample sizes in any given age.

In conclusion, the overall nonproportional hazard model seems to well illustrate the general pattern of the relationship of education with CVD fatality among women with CVD in our study, not due to the any specific cohort influence.

Back to Article Outline

References 

  1. Allison P . Survival analysis using the SAS system. A practical guide . Cary, NC: SAS Institute; 1995;
  2. American Heart Association, 2003. 2003 Heart and stroke statistical update. Available at http://www.americanheart.org.
  3. Antonovsky A . Social class, life expectancy and overall mortality . Milbank Memorial Fund Quarterly . 1967;43:31–37
  4. Bassuk SS , Berkman LF , Amick BC . Socioeconomic status and mortality among the elderly (findings from four US communities) . American Journal of Epidemiology . 2002;155:520–533
  5. Bolye EJ . Biological patterns in hypertension by race, sex, body weight and skin color . Journal of the American Medical Association . 1970;213:1637–1643
  6. Cirera L , Tormo MJ , Chirlaque MD , Navarro C . Cardiovascular risk factors and educational attainment in Southern Spain (a study of a random sample of 3091 adults) . European Journal of Epidemiology . 1998;14:755–763
  7. Cohen CS , Barbano HE , Cox CS , Feldman JJ , Finucase FF , Kleinman JC , et al.   In: Plan and operation of the NHAES I Epidemiologic Followup Study: 1982–84. Vital and Health Statistics Series 1, No 22 . Hyattsville, MD: National Center for Health Statistics; 1987;p. 87–1324 DHHS publication (PHS)
  8. Collett D . Modeling survival data in medical research . London: Chapman & Hall; 1994;
  9. Criqui MH , Barrett-Connor E , Austin M . Differences between respondents and nonrespondents in a population-based cardiovascular disease study . American Journal of Epidemiology . 1978;108:367–372
  10. Diez-Roux AV , Nieto FJ , Tyroler HA , Crum LD , Szklo M . Social inequalities and atherosclerosis. The atherosclerosis risk in communities study . American Journal of Epidemiology . 1995;141:960–972
  11. Elo IT , Preston SH . Educational differentials in mortality (United States, 1979–85) . Social Science and Medicine . 1996;42:47–57
  12. Epstein F , Ostrander L , Johnson B , Payne MW , Hayner NS , Keller JB , et al.   Epidemiological studies of cardiovascular disease in a total community-Tecumseh, Michigan . Annals of Internal Medicine . 1965;62:1170–1187
  13. Feldman JJ , Makuc DM , Kleinman JC , Cornoni-Huntley J . National trends in educational differentials in mortality . American Journal of Epidemiology . 1989;129:919–933
  14. Garrison RJ , Gold RS , Wilson PW , Kannel WB . Educational attainment and coronary heart disease risk (the Framingham Offspring Study) . Preventive Medicine . 1993;22:54–64
  15. Gonzalez MA , Rodriguez Artalejo F , Calero JR . Relationship between socioeconomic status and ischaemic heart disease in cohort and case-control studies (1960–1993) . International Journal of Epidemiology . 1998;27:350–358
  16. Gordon T , Shurtleff D . Means at each examination and interexamination variation of specified characteristics: Framingham Study. Exam 1 to exam 10 . In:  Kannel WB ,  Gordon T editor. The Framingham Study . Washington, DC: US Government Printing Office; 1973;p. 74–478 DHEW publication (NIH)
  17. Gran B . Major differences in cardiovascular risk indicators by educational status. Results from a population based screening program . Scandinavian Journal of Social Medicine . 1995;23:9–16
  18. Gupta R , Gupta VP , Ahluwalia NS . Educational status, coronary heart disease, and coronary risk factor prevalence in a rural population of India . British Medical Journal . 1994;309:1332–1336
  19. Hazuda HP , Comeaux PJ , Stern MP , Haffner SM , Eifler CW , Rosenthal M . A comparison of three indicators for identifying Mexican Americans in epidemiologic research. Methodological findings from the San Antonio Heart Study . American Journal of Epidemiology . 1986;123:96–112
  20. Hemingway H , Shipley M , Macfarlane P , Marmot M . Impact of socioeconomic status on coronary mortality in people with symptoms, electrocardiographic abnormalities, both or neither (the original Whitehall study 25 year follow up) . Journal of Epidemiology and Community Health . 2000;54:510–516
  21. Jackson R , Chambless LE , Yang K , Byrne T , Watson R , Folsom A , et al.   Differences between respondents and nonrespondents in a multicenter community-based study vary by gender ethnicity. The Atherosclerosis Risk in Communities (ARIC) Study Investigators . Journal of Clinical Epidemiology . 1996;49:1441–1446
  22. Jacobsen BK , Thelle DS . Risk factors for coronary heart disease and level of education. The Tromso Heart Study . American Journal of Epidemiology . 1988;127:923–932
  23. Johnson JL , Heineman EF , Heiss G , Hames CG , Tyroler HA . Cardiovascular disease risk factors and mortality among black women and white women aged 40–64 years in Evans County, Georgia . American Journal of Epidemiology . 1986;123:209–220
  24. Kaplan GA , Keil JE . Socioeconomic factors and cardiovascular disease (a review of the literature) . Circulation . 1993;88:1973–1998
  25. Keil JE , Loadholt CB , Weinrich MC , Sandifer SH , Boyle E . Incidence of coronary heart disease in blacks in Charleston, South Carolina . American Heart Journal . 1984;108:779–786
  26. Kitagawa EM , Hauser PM . Differential mortality in the United States (a study in socioeconomic epidemiology) . Cambridge, MA: Harvard University Press; 1973;
  27. Lerner DJ , Kannel WB . Patterns of coronary heart disease morbidity and mortality in the sexes (a 26-year follow-up of the Framingham population) . American Heart Journal . 1986;111:383–390
  28. Liu K , Cedres LB , Stamler J , Dyer A , Stamler R , Nanas S , et al.   Relationship of education to major risk factors and death from coronary heart disease, cardiovascular diseases and all causes. Findings of three Chicago epidemiologic studies . Circulation . 1982;66:1308–1314
  29. Luepker RV , Rosamond WD , Murphy R , Sprafka JM , Folsom AR , McGovern PG , et al.   Socioeconomic status and coronary heart disease risk factor trends. The Minnesota Heart Survey . Circulation . 1993;88:2172–2179
  30. 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? . American Journal of Epidemiology . 1996;144:934–942
  31. Mackenbach JP , Cavelaars AE , Kunst AE , Groenhof F . Socioeconomic inequalities in cardiovascular disease mortality; an international study . European Heart Journal . 2000;21:1141–1151
  32. Marmot MG , Shipley MJ , Rose G . Inequalities in death—specific explanations of a general pattern? . Lancet . 1984;1:1003–1006
  33. Marrugat J , Gil M , Masia R , Sala J , Elosua R , Anto JM . Role of age and sex in short-term and long term mortality after a first Q wave myocardial infarction . Journal of Epidemiology and Community Health . 2001;55:487–493
  34. 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 . American Journal of Epidemiology . 1989;129:1132–1144
  35. McDonough JR , Hames CG , Stulb SC , Garrison GE . Coronary heart disease among negroes and whites in Evans County, Georgia . Journal of Chronic Diseases . 1965;18:443–468
  36. Mustard CA , Derksen S , Berthelot JM , Wolfson M , Roos LL . Age-specific education and income gradients in morbidity and mortality in a Canadian province . Social Science and Medicine . 1997;45:383–397
  37. National Center for Health Statistics . International Classification of Diseases, Adapted for Use in the United States. 8th rev . vol 1. Washington, DC: U.S. Government Printing Office; 1967;
  38. National Center for Health Statistics . In: Plan and operation of the Health and Nutrition Examination Survey, United States, 1971–73. Vital and Health Statistics Series 1, No 10a . Washington, DC: US Government Printing Office; 1973;p. 73–1310 DHEW publication (HSM)
  39. National Heart, Lung and Blood Institute . Atherosclerosis Risk in Community (ARIC) Study. Operations manual, no 8. Lipid and lipoprotein determinations. Version 1.0 . Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987;
  40. Otten FW , Bosma HH . The socio-economic distribution of heart diseases (changing gradients in The Netherlands) . Social Science and Medicine . 1997;44:1349–1356
  41. Pekkanen J , Uutela A , Valkonen T , Vartiainen E , Tuomilehto J , Puska P . Coronary risk factor levels (differences between educational groups in 1972–87 in eastern Finland) . Journal of Epidemiology and Community Health . 1995;49:144–149
  42. Salomaa V , Niemela M , Miettinen H , Ketonen M , Immonen-Raiha P , Koskinen S , et al.   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;101:1913–1918
  43. SAS Institute Inc . SAS/STAT user’s guide, version 8.01 . Cary, NC: SAS Institute; 1999;
  44. Shkolnikov VM , Leon DA , Adamets S , Andreev E , Deev A . Educational level and adult mortality in Russia (an analysis of routine data 1979 to 1994) . Social Science and Medicine . 1998;47:357–369
  45. Tenconi MT , Devoti G , Comelli M . Role of socioeconomic indicators in the prediction of all causes and coronary heart disease mortality in over 12,000 men—The Italian RIFLE pooling project . European Journal of Epidemiology . 2000;16:565–571
  46. The ARIC Investigators . The Atherosclerosis Risk in Communities (ARIC) Study (design and objectives) . American Journal of Epidemiology . 1989;129:687–702
  47. Vaccarino V , Berkman LF , Krumholz HM . Long-term outcome of myocardial infarction in women and men (a population perspective) . American Journal of Epidemiology . 2000;152:965–973
  48. 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) . American Journal of Public Health . 1999;89:315–321
  49. Wamala SP , Orth-Gomer K . Interfaces of human biology and social organization (challenges for future research) . Inequality in health—A Swedish Perspective (Contributions from Swedish researchers in the social, behavioral and medical sciences for a national research programme on inequality in health) . Stockholm, Sweden: Swedish Social Research Council (SFR); 1998;
  50. World Health Organization . Manual of the international statistical classification of diseases, injuries, and causes of death. 9th rev . vol 1. Geneva, Switzerland: WHO; 1977;

Dr. Lori Mosca is Director of Preventive Cardiology at New York-Presbyterian Hospital, Associate Professor of Medicine at Columbia University, and Principal Investigator of the NIH-sponsored Women’s Pooling Project.

Dr. Furcy Paultre is a research scientist at Columbia University and the Preventive Cardiology Program at New York-Presbyterian Hospital.

Ms. Ji Won R. Lee was a trainee in the Preventive Cardiology Program at New York-Presbyterian Hospital and a Master in Public Health Student at Columbia University at the time this research was conducted.

PII: S1049-3867(04)00118-5

doi:10.1016/j.whi.2004.11.004

Women's Health Issues
Volume 15, Issue 2 , Pages 80-88, March 2005

Article Tools

* Image Usage & Resolution