1
Faculty of Medicine Ramathibodi Hospital,
Mahidol University
RACE 699 Dissertation
Doctor of Philosophy Program in Clinical Epidemiology
(International Program)
Name student: WIN KHAING
ID: RACE/D 5736100
Title:
The mediating roles of education and income on major cardiovascular events
Research Proposal
2
Contents
INVESTIGATOR AND SUPERVISORS 3 1. BACKGROUND AND RATIONALE 4 1.1 Background and Rationale 1.2 Research Question 1.3 Research Objectives 1.3.1 Primary Objective 1.3.2 Secondary Objectives 2. LITERATURE REVIEW 8 2.1 Epidemiologic transition and cardiovascular diseases 2.2 Impact of social determinants of health on
cardiovascular diseases
2.3 Socioeconomic status and cardiovascular diseases 2.4 Effect of education and income on cardiovascular
outcome: systematic review and meta-analysis
2.5 The association between education/income and cardiovascular risk factors
2.6 Conceptual framework 3. METHODOLOGY 24 3.1 Study design and setting 3.2 Study subjects 3.3 Variables & Measurement 3.4 Data Collection 3.5 Sample size estimation 3.6 Data analysis 3.6.1 Data management 3.6.2 Statistical analysis 3.7 Ethics considerations 3.8 Budget 3.9 Time Frame ACKNOWLEDGEMENTS 40 REFERENCES 41 TABLES 58 FIGURES 89 APPENDICES 100 A. Search terms and search strategies B. Newcastle-Ottawa Quality Assessment Scale C. Dummy tables
3
TITLE: The mediating role of education and income on major
cardiovascular events
List of Investigators and Affiliation
Dr. Win Khaing, MBBS MMedSc (Public Health)
Section for Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital,
Mahidol University, Bangkok, Thailand.
Department of Preventive and Social Medicine, University of Medicine, Mandalay, Myanmar.
Supervisors
Dr. Ammarin Thakkinstian, Ph.D.
Section for Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital,
Mahidol University, Bangkok, Thailand.
Dr. Atiporn Ingsathit, Ph.D.
Section for Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital,
Mahidol University, Bangkok, Thailand.
Dr. Sakda Arj-ong Vallipakorn, M.D., Ph.D.
Section for Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital,
Mahidol University, Bangkok, Thailand.
4
CHAPTER 1
BACKGROUND AND RATIONALE
1.1 Background and rationale
1.1.1 Non-communicable diseases and cardiovascular diseases burden
Of the global mortality in 2012 of 56 million annually, 38 million (68%) were due
to non-communicable diseases (NCDs)1. A global epidemic of NCDs strikes hardest to low-
and middle-income countries (LMICs) including Asian countries which accounts for almost
three quarters (28 million) of global NCD deaths1. The World Health Organization (WHO)
estimated that NCD deaths are projected to rise to 52 million in 20301. WHO reported that
NCDs are responsible for more than two thirds of global mortality, of which 82% were
cardiovascular diseases (CVD) followed by cancers, respiratory diseases, and diabetes. CVD
is a major public health problem that accounts for about 30% of the annual global mortality
and 10% of the global disease burden1.
1.1.2 Cardiovascular disease risk factors
The Framingham Heart Study2, the WHO-MONICA Project3, and the INTERHEART4
studies reported evidence for the major risk factors of CVD. Risk factors can be classified as
demographic (e.g., age, sex, race, family history, and etcetera), behavioral (e.g., smoking,
alcohol consumption, physical inactivity, dietary, and etcetera) and metabolic (body mass
index, blood glucose, cholesterol level, and etcetera)5-7. Modification of these risks would lead
to reduced cardiovascular morbidity and mortality. Despite much effort invested in primary
and secondary prevention of CVD, it is still a problem in industrialized and high income
countries, as well as in LMICs1. Understanding of these risk factors is critical to the prevention
of cardiovascular morbidity and mortality. In addition, nontraditional markers (e.g., high-
sensitivity C-reactive protein8, lipoprotein(a)9, homocysteine10, small dense low-density
lipoprotein-C particles11, fibrinogen12, and etcetera) were also identified with advanced
investigations. Despite much effort invested in primary and secondary prevention of CVD, it
is still a problem in industrialized and high income countries, as well as in LMICs1.
5
1.1.3 Cardiovascular diseases and social determinants of health
Recently, the fifth epidemiological transition proposed that social upheaval13
might break down the existing social and health structures and lead to increased CVD
morbidity and mortality. The impacts of these consequences caused much concerned to all
societies and economies, and were particularly devastating in poor and vulnerable LMICs
populations. Since then, many social determinants of health (SDH) have been increasingly
considered and should be included in a causal pathway with other traditional risk factors and
markers.
Many studies show that SDH indirectly influence CVD through behavioral and
metabolic cardiovascular risk factors (CVRFs), psychosocial factors and environmental living
condition14, 15. Some landmark studies16-18 and numerous other epidemiological studies19-22
show an inverse relationship between SDH and CVD morbidity and mortality.
Low educated persons were more likely to have higher CVRFs (e.g., hypertension,
diabetes, dyslipidemias, overweight, smoking and sedentary lifestyle), and had less healthy
dietary habits than high educated persons23-25. Evidence also shows that lower education is
associated with atherosclerosis, ischemic heart disease, cerebrovascular diseases, CVD
mortality and all-cause mortality26-28. Similar to education, the inverse relationship of income
on ischemic heart disease (IHD), coronary events, pre-hospital coronary death and CVD
mortality has also been reported29-33. These effects of education and income are more consistent
in developed countries, but results are still inconclusive in LMICs34, 35
1.1.4 Effects of education/income on cardiovascular diseases
A number of narrative and systematic reviews36-41 have studied the relationship of
socioeconomic status (SES) with CVD including myocardial infarct (MI), strokes, heart failure
(HF), and death. Two meta-analyses have reported the effect of education and income on MI36
and CVD mortality40. In both studies, education and income were roughly categorized as low
and high and SES classes were not uniformly pooled for homogeneity, resulting in an inability
to assess SES gradients. Few studies included participants from LMICs.
We therefore conducted a systematic review and meta-analysis to pool effects of
education and income on various cardiovascular outcomes by including more studies
conducted in developing countries.
Our findings indicated that low to middle education and income carried higher risks of
coronary artery diseases (CAD), cardiovascular events (CVE), strokes and cardiovascular
6
deaths when compared to high education and income. Comparing medium and low versus high
education groups, pooled relative risks (RRs) were 1.21 (1.06, 1.40) & 1.36 (1.11, 1.66) for
CAD, 1.27 (1.09, 1.48) & 1.50 (1.17, 1.92) for CVE, 1.17 (1.01, 1.35) & 1.23 (1.06, 1.43) for
strokes and 1.21 (1.12, 1.30) & 1.39 (1.26, 1.54) for cardiovascular deaths. Pooled RRs for
medium and low versus high income groups were 1.27 (1.10, 1.47) & 1.49 (1.16, 1.91) for
CAD, 1.05 (0.98, 1.13) & 1.17 (0.96, 1.44) for CVE, 1.24 (1.00, 1.53) & 1.30 (0.99, 1.72) for
strokes and 1.34 (1.17, 1.54) & 1.76 (1.45, 2.14) for cardiovascular deaths.
In our systematic review, estimations were deduced from high income countries
(93.1%), mostly from the European region (54.2%). Studies from Asian region were still
lacking, especially, in association of income with CVD outcomes.
1.1.5 Rationale
Results of our systematic review indicated that education and income were associated
with CVD outcomes. Previous evidences have shown that education was also highly associated
with income42, 43 or vice versa44, 45, i.e., higher education provides individuals with higher
income, and both may increase the risk of CVD. Our systematic and previous reviews could
only answer direct effects of education and income on CVDs, but not for a causal relationship
pathway. There was still lack of empirical evidences analyzing the causal pathways between
education/income and CVD outcomes, especially in Asian countries.
In order to answer these questions, a large-scale cohort which has sufficient power to
adjust for all the known CVRFs and follow-up long enough to observe for CVD outcomes
would be necessary, especially in Asian countries. Therefore, this study will be conducted
using data from employees of the Electricity Generating Authority of Thailand (EGAT)
prospective cohort with pre-specified cardiovascular events as primary outcome by following
research questions and objectives.
7
1.2 Research questions
• Does education directly affect major cardiovascular events or is its effect mediated
through income, or vice versa?
• What are the cardiovascular risk factors that are associated with education and income?
1.3 Research objectives
1.3.1 Primary objectives
- To determine the direct and indirect effects of education through income on
major cardiovascular events
- To determine the direct and indirect effects of income through education on
major cardiovascular events
1.3.2 Secondary objectives
- To assess association between education and cardiovascular risk factors
- To assess association between income and cardiovascular risk factors
8
CHAPTER 2
LITERATURE REVIEW
2.1 Epidemiologic transition, cardiovascular diseases
A global pattern of morbidity and mortality of cardiovascular diseases (CVD) has
been observed over time. Historically, CVD are mainly concerned with the cause of infections
like rheumatic fever and syphilis on the heart and cardiomyopathies due to malnutrition, and
death from CVD accounts only less than 10%13, 46. During the second stage of epidemiologic
transition, with more advances in societies, major causes of cardiovascular diseases shifted
from a predominance of infection and nutritional causes to chronic degenerative causes like
diseases related to hypertension, such as hemorrhagic stroke and hypertensive heart disease
with deaths attributed to CVD increased up to 35%13, 46, 47. Because of life expectancy
improvement, during the third stage of transition, cardiovascular diseases related with cigarette
smoking, high-fat diets and sedentary lifestyles have become more common. CVD, most
frequently ischemic heart disease (IHD) and atherosclerotic thrombotic stroke became
prominent especially at ages below 50 years. Not amazingly, deaths accounted to CVD
continued to rise from 35% to 65% of total deaths13. With increased efforts to earlier diagnose,
treat promptly and understand more about preventive measures, cardiovascular diseases had
become able to delay to more advanced ages of CVD during the fourth stage. Therefore, the
relationships between CVD and risk factors such as age, family history, high blood pressure,
tobacco smoking, unhealthy diet, alcohol use, overweight and obesity, diabetes, physical
inactivity, and dyslipidemia have been extensively explored by many researchers in that
periods.
More recently, a fifth epidemiological transition was proposed and was called “age
of health regression and social upheaval”13. We are, in turn, facing resurgence of conditions
seen in the first two stages and also diseases of the third and fourth stages still persist. Social
upheaval or war breaks down the existing social and health structures, leading to increased
deaths due to both cardiovascular and non-cardiovascular causes such as infectious diseases,
violence, accidents. Accordingly, now, many researchers have suggested that social
determinants of health should not only be put together with traditional risk factors acting
directly on CVD, but also be examined as underlying determinants of some CVDs. Actually,
these social risk factors might act along causal chains, influencing the incidence and
9
management of conventional risk factors. Many researchers became increasing awareness that
different socioeconomic factors could affect health at different times in the life cycle, effective
at different levels and through different causal pathways.
2.2 Impact of social determinants of health on cardiovascular diseases
World Health Organization (WHO) defined the term “social determinants of
health” (SDH)48 as “the conditions in which individuals are born, grow, live, work and age,
which are shaped by the distribution of money, power and resources at global, national and
local levels”. WHO comprehended that SDH is the most responsible issue of unfair and
avoidable “health inequities” between groups of people within countries and between
countries. People’s social and economic conditions can effect on their lives and determine their
risk of illness and their decisions to prevent them becoming ill or treat illness when it happens.
In order to push towards progressive achievement of universal health coverage
(UHC), health inequities need to be reduced, and both SDH and UHC need to be take action in
an integrated and systematic manner.
SDH (e.g., the level of education, income, race, ethnicity, culture and language,
health care system, working conditions, employment and job stability, residential environment
and social support or social network) directly or indirectly influenced CVD by impacting
behavioral and metabolic cardiovascular risk factors, psychosocial factors and environmental
living conditions14, 15. The Whitehall study, Whitehall II study, and the Black report are well
known studies that showed evidence of an inverse association between SDH and CVD
morbidity and mortality. The Evans County Study19, the US National Longitudinal Mortality
Study20, the Charleston Heart Study21 and the Alameda Country Study22 also showed the
similar trends. Work-related stress and depression were found to be associated with
hypertension and arthrosclerosis. Negative social relationship was found to be linked with
increased blood pressure. The poor have limited choice of healthy lifestyle and health care
access which may explain the link between socioeconomic status and CVD14, 15.
2.3 Socioeconomic status and cardiovascular diseases
Socioeconomic status (SES) has been widely accepted as the most powerful SDH.
Three common measures of socioeconomic status, i.e., education, income and occupation have
been extensively explored with regard to their relationship to cardiovascular health. In general,
lower socioeconomic status is associated with a higher prevalence of CVD risk factors and a
greater incidence of mortality resulting from CVD41, 49, 50. Researchers showed low education
10
and income are associated with higher mortality from coronary heart disease51-53.
Socioeconomic gradient in stroke is also greatly influenced by traditional stroke risk factors
like diabetes, hypertension and alcohol abuse54, 55.
Education is the most widely accepted measurement of SES because it is relatively easy
to obtain, more likely to respond their education level, and has less recall bias as people tends
to remember their education level accurately. It is also usually fixed in late childhood or early
adulthood, precedes health outcomes, compared to income which is far more likely to change
over the life cycle.
Since, education shapes future earning potential and occupational opportunities, higher
education provides individuals with higher income, and provides better knowledge and life
skills to get more access to information and resources to promote health. Therefore, better-
educated people can utilize better healthcare resources and healthier foods, and can also enable
more leisure time for exercise. Higher level of education and income also tend to stand in higher
social class, status and social network56 from which many positive benefits can be gained like
beneficial behavioral norms, positive materials and emotional support.
Measurement of education is generally favored to use years of schooling, but not reflect
difference in school prestige or resources, which may effect to differences in future earnings.
Measurement of income at individual, family, and community levels remains a great challenge.
In comparison, education is typically established in early adulthood and remains stable
throughout the life, whilst income is dynamic and might change extensively from early
adulthood to middle-age and then into retirement and late old age.
2.3.1. Education and cardiovascular diseases
Lower levels of educational attainment are associated with a higher incidence of CVD,
higher prevalence of cardiovascular risks and greater cardiovascular mortality26, 41. Tromsø
Heart Study23 in a 12,368 Norwegians cohort found that higher educated persons were less
likely to smoke, likely to be overweight, but more physically active and had a healthier diet.
Educational differences in ischemic heart disease, cerebrovascular diseases and CVD mortality
in the US and 11 Western European countries was studied by Mackenbach et al26 who found
that lower education individuals have higher mortality with inequality in smoking and
excessive alcohol consumption in all countries. A study27, which included 308 asymptomatic
women from the Healthy Women Study has shown lower education was associated with greater
early stage atherosclerosis. Stanford Five City Project which included 2,380 participants also
11
showed a consistent trend between lower educational attainment and high exposure to
cardiovascular risk factors. Cirera et al24 studied cardiovascular risk factors and educational
attainment in 3091 Spanish adults, and showed that people without schooling have two to three
times higher prevalence of overweight in women and hypertension in both genders when
compared to people at the university level. Recently, the ATTICA study25 showed that people
with low education were 52% higher risk of developing CVD compared to those with high
education. Low educated people were also higher prevalence of hypertension, diabetes and
dyslipidemias and more likely to be smokers and sedentary, with less healthy dietary habits.
The GREECS longitudinal study28 also shown that all-cause mortality was 2 times higher in
low education group as compared to intermediate and high education groups (40%, vs. 22%
and 19%, respectively, p<0.001). Reversed association in developing countries has also been
observed elsewhere. A study conducted by Fernald & Alder34 in Mexico found that educational
attainment showed an inverse association with systolic blood pressure in low-income rural
women.
2.3.2 Income and cardiovascular diseases
In parallel with education, many studies documented the association of income and
cardiovascular outcome. Andersen et al29 studied income and risk of ischemic heart disease
(IHD) in 22,782 people in Nordic countries. Inverse effect of income on IHD was seen with
hazard ratio for highest versus lowest deciles of income of 0.53 (95%CI: 0.44, 0.65). The
FINMONICA study30 reported low-income men have 2 times higher risk of pre-hospital
coronary death compared to high-income men and 1-year mortality rate was also significantly
higher in low-income patients in those who survived after MI. The FINAMI study31 also
showed that lower income people are 5.21 times and 11.13 times more likely to develop
coronary events than higher income people among 35 to 64 year-old men and women,
respectively. Similar findings were also reported by Alter et al.32 and Rao et al.33, with
opposing, reverse findings which were reported from a study35 in China. This Chinese study
found that people with higher family average income were 1.94 times more likely to develop
strokes compared to those with lower average family income after adjustment for demographic
and traditional risk factors. A review by Harper et al also agreed that there is no evidence of
consistent associations between income inequality and prevalence of CVD risk factors and
outcomes.
12
2.4 Effects of education and income on cardiovascular outcomes: Systematic review and meta-analysis
2.4.1 Methods
The review protocol has been registered with the international prospective register of
systematic review (PROSPERO number CRD42016046615)57.
2.4.1.1 Search strategy
Relevant studies were identified from Medline and Scopus databases since inception to
30th July 2016. Titles and abstracts were screened, and full articles were read if decision of
selection could not be made. Reference lists were also checked for studies that were not
identified by our searching. The following search terms were used for Medline:
"Cardiovascular Disease"[Mesh], "cardiovascular event", "Myocardial Infraction"[Mesh],
"Heart Failure"[Mesh], "Ventricular Function, Left"[Mesh], "Coronary Disease"[Mesh],
"Coronary Restenosis"[Mesh], "restenosis", "re-stenosis", "coronary flow", "coronary blood
flow", "ejection fraction", "stroke", "cardiovascular death", "cardiovascular mortality",
education[Mesh], "education status"[Mesh], "education level" and income[Mesh]. Search
strategies for both databases are described in Appendix A.
2.4.1.2 Selection of studies
Retrospective or prospective cohorts published in English were selected if they met the
following criteria: assessed associations between education/income and cardiovascular
outcomes in either a general or specific types of adult population; measured education (either
education years/groups) or income in terms of money or in category; had at least one of
outcome of interest (i.e., coronary artery diseases (CAD), cardiovascular events (CVE), strokes
and cardiovascular deaths); had contingency data between education/income and
cardiovascular outcomes, or a beta-coefficient. Studies were excluded from the review if data
for education and income were combined; income was assessed based on ownership of
car/house/health insurance/zip-code. In cases of missing data, we made 3 attempts to contact
authors to request additional data.
2.4.1.3 Study factors
Education and income were our study factors; which were assessed and reported
differently across studies. To standardize data for pooling across studies, they were re-
13
categorized into 3 groups as low, medium, and high for education years ≤ 9 (i.e., illiteracy/ no
education/ basic/ primary education), 10 – 12 (i.e., secondary/ high school/ intermediate/
technical/ apprenticed/ trade/ vocation), and > 12 years (i.e., university/ college/ associates/
master/ professional/ PhD), respectively. In addition, income expressed in other currencies was
converted to US currency/year using the reported exchange rates or the exchange rate at the
time of study publication by using online currency converter58. The salary income was re-
categorized as ≤20,000, 20,001 to 40,000, and >40,000 US$ for low, medium, and high,
respectively. If the original studies reported income as quartile, these were re-categorized: 1st
= low, 2nd = medium, and 3rd + 4th quartiles = high income. If studies reported income as
quintile, these were re-categorized: 1st + 2nd = low, 3rd = medium, and 4th + 5th quintiles = high,
respectively.
2.4.1.4 Outcomes
The outcomes of interest were CVDs including CAD (e.g., acute myocardial infarct
(AMI), IHD, coronary heart disease (CHD)), CVE (e.g., HF, hospital admission due to cardiac
causes, revascularization and composite CVDs, e.g., IHD or strokes), strokes (ischemic or
hemorrhagic strokes), and cardiovascular deaths. These were defined according to the original
studies.
2.4.1.5 Data extraction
Two reviewers (WK and SV) independently extracted general information (the first
author’s last name, the publication year) and characteristics of studies/patients (i.e., study
country, mean age, gender, mean body mass index (BMI), diabetes mellitus, physical activity,
smoking, alcohol drinking, hypertension, dyslipidemia and chronic renal failure). In addition,
education and income and type of outcomes were also extracted. Furthermore, cross-tabulated
data between education/income groups and individual outcome were extracted for pooling.
Summary statistics (e.g., odds ratio, risk ratio, or hazard ratio) along with its 95% confidence
interval (CI) were extracted instead, if frequency data were not reported. Authors were
contacted if insufficient data were insufficient. Data entry, cleaning and cross check validation
were performed separately for each study. Entries were compared for accuracy and any
disagreements were solved by consensus.
14
2.4.1.6 Risk of bias assessment
The quality of studies were independently assessed by two reviewers (WK and SV)
using the Newcastle and Ottawa risk of bias criteria (see Appendix B). The following three
domains were evaluated, i.e., selection of study groups (4 subdomains), comparability of
groups (2 subdomains) and ascertainment of exposure and outcome (3 subdomains). Each was
graded as 0 to 1 with a total grade ranging from 0 to 9. A total grade of seven or more was
regarded as higher quality or lower risk of bias.
2.4.1.7 Statistical analysis
Relative risks (RR) of having each outcome between low versus high (RR1) and
medium versus high (RR2) education/income groups were recalculated from frequency data for
studies whose frequency data were available. These were then appended with reported
summary statistics where frequency data were not available. A multivariate random-effect
meta-analysis59 was applied for pooling two RRs simultaneously. Variance-covariance
between RR1 and RR2 was assumed to be zero for those studies reporting summary RRs without
frequency data.
Heterogeneity and degree of heterogeneity were assessed by Cochrane’s Q test and I-
squared statistic, respectively. Heterogeneity was considered to be present if the p value of Q
test was <0.1 or I-squared ≥25%.
Subgroup analyses were conducted to examine potential sources of heterogeneity by
fitting each of the co-variables (i.e., country, country income level60, number of co-variables
adjustment, age group, BMI, percentage of males, diabetes, obesity, hypertension, high
physical activity, smoking, alcohol drinking, dyslipidemia and chronic kidney disease) in a
multivariate meta-regression model.
Finally, exploration of potential publication bias was visualized using a funnel plot and
Egger's test. If any of these indicated asymmetry, a contour enhanced funnel plot was
constructed to distinguish whether the cause of the asymmetry was due to publication bias or
heterogeneity.
All analyses were performed using STATA61 version 14.1. P-values <0.05 were
considered as statistically significant, except for the test of heterogeneity where p <0.10 was
used.
15
2.4.2 Results
We identified 354 and 1335 studies from Medline and Scopus databases with 11
additional studies from reference lists. Of these 1700 studies, 115 were duplicates, leaving 1585
studies to be screened. After screening titles and abstracts, 1399 studies did not address our
primary question, leaving 72 studies for inclusion. Reasons for exclusion of the studies are
presented in Figure 2.1 following the Preferred Reporting Items for Systematic Review and
Meta-analysis (PRISMA) guideline.
2.4.2.1 General Characteristics of included studies
Characteristics of the 72 included cohorts published between 1982 and 2016 are shown
in Table 2.1. Among them, 14, 39 and 19 studies were conducted in Asia, Europe, and the
United States, respectively. Most studies were from high-income countries (93.1%); mean age
and mean BMI ranged from 38.5 to 78 years and 23.02 to 30.33 kg/m2, respectively. Percentage
of males and proportion with diabetes, smoking and hypertension varied from 35.9% to 78%,
1.3% to 42%, 7.28% to 72.64%, and 6.25% to 72.5% respectively. Thirty-three studies assessed
association between education and cardiovascular outcomes, 10 studies assessed effect of
income, and 29 studies assessed effects of both education and income, with a sample size
ranging from 128 to 4,157,202.
2.4.2.2 Risk of bias assessment
Results of the “risk of bias” assessment of the included studies are shown in Table 2.2.
Total scores ranged from 5 to 9 with a median of 7. Among the included studies, 45 out of 72
(62%) had low risk of bias and 27 out of 72 (38%) had high risk of bias.
2.4.2.3 Education and cardiovascular outcomes
A total of 62 studies assessed association between education and cardiovascular deaths
(N = 35 and 31 for low and medium vs high), CAD (N = 21 and 18 for low and middle vs
high), CVE (N = 13 and 15 for low and middle vs high) and strokes (N = 15 and 13 for low
and middle vs high).
Among them, there were very few studies (4 in cardiovascular death and CAD, 3 in
CVE, and 2 in strokes) where risks were estimated from unadjusted or raw frequency data. To
be consistent, only co-variates adjusted studies were pooled to see the effects of education.
Results are presented in Table 2.3. Effects of education on these outcomes were heterogeneous
across studies with the I2 ranging from 83% to 99%, see Table 2.3. Multivariate meta-analysis
16
was applied indicating significant educational effects on all outcomes, see Table 2.3 & Figure
2.2. The strongest education effect was on CVE, where low and medium education increased
CVE by 50% (RR 1.50, 95% CI: 1.17, 1.92) and 27 % (RR 1.27, 95% CI: 1.09, 1.48) compared
to high education. A similar trend occurred for cardiovascular deaths, in which the risks for
these education levels were 39% (RR 1.39, 95% CI: 1.26, 1.54) and 21% (RR 1.21, 95% CI:
1.12, 1.30). In addition, patients with low education showed 36% (RR 1.36, 95% CI: 1.11,
1.66) higher risk, and patients with medium education showed 21% (RR 1.21, 95% CI: 1.06,
1.40) higher risks for CAD. Furthermore, low and medium education levels were associated
with 23% (RR 1.23, 95% CI: 1.06, 1.43) and 17% (RR 1.17, 95% CI: 1.01, 1.35) higher risks,
respectively, for developing strokes when compared to high education level.
Sources of heterogeneity were next explored by meta-regression or subgroup analyses,
see Table 2.4 – 2.8. Geographical regions were grouped as Asia, Europe, and US, but only a
small number of studies in the Asian setting were available for most outcomes. Effects of both
low/middle education still remained on all 4 cardiovascular outcomes for pooling within
Europe and US, but not for Asia, likely due to small numbers of studies, see Table 2.4.
We performed subgroup analyses by co-variables including number of adjusted
variables, age (≤60 vs >60 years), BMI (<25 kg/m2 vs ≥ 25 kg/ m2), percentages of male,
diabetes, and smoking (see Table 2.5 – 2.8), and none of these was identified as a source of
heterogeneity. However, education levels were associated with all four CVD outcomes in the
subgroup younger than 60 years (see Table 2.5 – 2.8). The risk of cardiovascular deaths and
CAD outcomes was higher in the studies comprising a higher percentage of male participants.
Likewise, the risk of CVD outcomes (except CAD) was higher in the studies with a higher
proportion of diabetic participants. The association between BMI and CVE was detected in the
BMI subgroup ≥ 25 kg/m2 (see Table 2.5 – 2.8).
There was no evidence of publication bias using Egger’s test except for low versus high
education level on CVD outcomes (Egger’s test: β=2.33, p=0.008) which corresponded with
funnel plots showing asymmetry (see Figure 2.3 & 2.4). A contour enhanced funnel plot
showed that some studies fell in both non-significant and significant areas, so asymmetry was
more likely due to heterogeneity (see Figure 2.5 & 2.6). No individual study significantly
changed the overall estimates based on the results of the sensitivity analysis.
17
2.4.2.4 Income and cardiovascular outcomes
A total of 39 studies assessed association between income and cardiovascular deaths
(N = 22 and 13 for low and middle vs high), respectively, CAD (N = 13 and 14 for low and
middle vs high), CVE (both N = 8 for low and middle vs high) and strokes (both N = 7 for low
and middle vs high). Amongst, risk estimations from unadjusted or raw frequency data (1 in
CVD, 4 in CAD, 2 in CVE and 1 in strokes) were excluded in order to pool the effects of
income from co-variates adjusted studies. Results are shown in Table 2.3. Effects of income
on these outcomes were pooled using multivariate meta-analysis with random-effect models,
see Table 2.3 & Figure 2.2. Substantial heterogeneity across studies was found with I2 ranging
from 95% to 99%, see Table 2.3. The largest income effect was also on cardiovascular deaths,
with 76% (RR 1.76, 95% CI: 1.45, 2.14) and 34% (RR 1.34, 95% CI: 1.17, 1.54) higher risk
of cardiovascular death for low and medium versus high income, respectively. Comparable
effects were seen on CAD, with 49% (RR 1.49, 95% CI: 1.16, 1.91) and 27% (RR 1.27, 95%
CI: 1.10, 1.47) higher risks, respectively. Furthermore, patients with low income showed 17%
(RR 1.17, 95% CI: 0.96, 1.44) higher risk, and patients with middle income showed 5% (RR
1.05, 95% CI: 0.98, 1.13) higher risks for CVE. Additionally, low and medium incomes showed
about 30% (RR 1.30, 95% CI: 0.99, 1.72) and 24% higher risks (RR 1.24, 95% CI: 1.00, 1.53)
of developing strokes when compared to high income.
Sources of heterogeneity were next explored by meta-regression or subgroup analyses
(see Table 2.4 – 2.8). By geographical regions, European studies showed effects of income
similar to the overall effect, see Table 2.4.
In subgroup analyses performed by age group and percentages of males, low income
was associated with higher risks for cardiovascular deaths, CAD and CVE, in the studies
consisting of participants aged 60 years and younger (see Table 2.5 – 2.8).
No publication bias was identified by Egger’s test except in medium versus high income
level with CAD outcome (Egger’s test: β=2.98, p=0.009), but funnel plots showed asymmetry
(see Figure 2.8 & 2.9). The contour enhanced funnel plots suggested that asymmetry was more
likely due to heterogeneity (see Figure 2.10 & 2.11). Overall estimates were similar to the
sensitivity analyses.
2.4.3 Discussion
We performed a systematic review and meta-analysis to pool effects of education and
income on CVDs. Our findings indicated that low to middle education and income carried
18
higher risks of CAD, CVE, strokes and cardiovascular death when compared to high education
and income. The pooled RRs for low and middle versus high education were 1.36 and 1.21 for
CAD, 1.50 and 1.27 for CVE, 1.23 and 1.17 for strokes, and 1.39 and 1.21 for cardiovascular
death. The pooled RRs for low and middle versus high income for these corresponding
outcomes were 1.49 and 1.27, 1.17 and 1.05, 1.30 and 1.24, and 1.76 & 1.34, respectively.
Direct or indirect mechanisms between education and income on CVD have been
described in which behavioral risk factors52, lifestyle or living environment condition62, health
literacy63 and psychological factors64, 65 play important roles. Low education and low income
persons had a higher prevalence of risk behaviors (smoking, obesity, physical inactivity,
unhealthy diet, and etcetera), and were more likely to live in poor polluted environment, have
poor health literacy (ability to read/understand comprehend medical information, lack of
awareness of impact of lifestyle behavior, poor adherence/incorrect medication, ignorance of
medical checkup), and have higher prevalence of depression with poorer coping in response to
cumulative stress. Consequently, mortality was high, potentially due to delayed access to
medical care, poor understanding in disease progress management, and lack of post-disease
cardiac rehabilitation66-71.
Moreover, education and income have mutual causal influences on CVD morbidity and
mortality and one should not rely on a single, potentially biased parameters72. Combined effects
of education and income have been studied previously57, and persons with low income and
education had the highest risk of incident CHD, when compared to those with high
education/low income, low education/high income, and high education/high income. However,
some researchers have suggested that education and income should not be combined and should
not be interchangeable45, because they may affect CVD outcomes through different, potentially
independent, causal pathways. For example, Ahmed et al73 found that low income was a
significant independent predictor of HF regardless of education level in community-dwelling
older adults age ≥65 years population. To prove this hypothesis, individual patient data
containing education and income variables are required, and mediation analysis should be
applied.
Many studies45, 74 have compared the difference between the highest and the lowest
strata of socioeconomic measure. This approach does not make maximal use of the data and
one loses the ability to see a “dose-response curve”18, 75, 76. In this study, to increase
comparability across the studies and to study the full gradient of exposure, the medium-level
education and income categories were maintained. This approach confirmed the social gradient
effect of education and income. Although there was high heterogeneity in the results, statistical
19
significance was seen, except in effects of income on CVE and strokes outcomes. This may be
due to the possibility of different definitions and classifications of education and income
categories between individual studies, and between different geographical regions, economies,
educational systems and cultures. Differences in study periods over time could lead to
variability in the scales used to classify the exposure.
Strength and limitation
Our meta-analysis has some strengths. To the best of our knowledge, it is the first meta-
analysis assessing levels of education and income effects on major CVD outcomes. To increase
comparability across the studies and to study the social gradient effects, three strata of
education and income were categorized and considered to yield more details than previous
meta-analyses36, 40. Effects of education/income were simultaneously pooled using multivariate
meta-analyses. In addition, we included only cohort studies that could provide more reliable
effects of education and income on CVD outcomes. This review was also conducted in
accordance with PRISMA guidelines77.
However, our study has also some limitations. Pooled estimates were highly
heterogeneous, which may be due to differences in characteristics of the study populations,
differences in definitions and classifications of education and income in both developed and
developing countries, and differences in timing of measurement of education and income
categories across the studies. Although many efforts were made to explore the heterogeneity,
we could not identify the sources. We also did not have access to primary data and many of the
studies did not adjust and report for confounding variables, and thus the estimated risk might
be confounded.
Clinical Implications and further research
Braveman et al45 pointed out that education can influence general and health-related
knowledge, health literacy, and problem-solving skills, which can change one’s health
outcome. The results of our meta-analysis provided some evidence of the effects of education
and income on CVD outcomes. However, whether education or income is directly associated
with CVD outcomes72, or education is indirectly associated with CVD outcomes through
income as mediator78, or both education and income are indirectly associated with CVD
outcomes through other risk factors such as BMI79, diabetes, smoking as mediators has not
been clearly answered in studies.
20
Further research should focus on the causal pathway between education and income on
CVD outcomes with more advanced statistical analysis, such as a mediation/moderation
analysis80.
2.4.4 Conclusion
Low and middle income countries have seen a rising tide of NCDs and CVD in recent
years1, 5, 6. This analysis identified an increased risk of CAD, CVE, strokes and cardiovascular
death among medium and low education or income. The findings in the current study, therefore,
confirmed the preexisting evidences supporting an association between socioeconomic
gradient and major cardiovascular events. This may explain why CVD may severely affect
LMICs, where a large proportion of citizens do not get an opportunity for education and live
under the national poverty line60. Furthermore, a large-scale cohort which has sufficient power
to adjust for all the known CVRFs and follow-up long enough to observe for CVD outcomes
would be necessary, especially in Asian countries.
2.5 The association between education/income and cardiovascular risk factors
Winkley and colleague49 studied the relationship between education/ income/
occupation and cardiovascular risk factors (cigarette smoking, blood pressure, and total and
HDL cholesterol) on people aged 25 to 64 and showed that only education was significantly
associated with these risk factors after adjustment for age and time of survey. Hoeymans et al81
also provided evidence that there was inverse association between educational level and
prevalence of smoking, physical inactivity, obesity, hypertension, hypercholesterolemia and
low HDL-cholesterol, but not with alcohol. The Minnesota Heart Survey82 conducted in 7781
adults aged 25 to 74 years shown that education was inversely related to blood pressure (BP),
cigarette smoking and BMI in both men and women while serum cholesterol was inversely
related to education in women only. However, for household income, the results showed less
consistency in magnitude and direction. A study from India83 showed that higher education
level was associated with overweight, physical inactivity, family history of CVD, higher fruit
intake and lower alcohol intake in both men and women, but higher diabetes and hypertension
prevalence was found only in men. BMI and waist circumference were also greater in those
with higher educational level for both sexes. The poorer had less diabetes, less overweight in
both sexes and less likely to have a family or established history of CVD, and smoked more in
poor men. They concluded that some biological cardiovascular risk factors were worse in
21
higher SES peoples while some behavioral risk factors were worse in lower SES peoples with
little knowledge about risk factors and screening practice. In the SESAMI study by Alter et
al32, income was inversely associated with 2-year mortality rate in unadjusted model (HR=0.45,
95%CI: 0.35, 0.57; P<0.001), but after adjustment for age and preexisting cardiovascular
events or cardiovascular risk factors, the effect was attenuated (HR=0.77, 95%CI: 0.54, 1.10;
P= 0.150).
2.6 Causal pathways between education/income and cardiovascular diseases
Many researchers agreed that socioeconomic indicators should not be used
interchangeably45 because they may effect outcome in different causal pathways and may
represent independent separate important risk factors of cardiovascular diseases. Although
education and income correlated to each other, it is not strong enough to use education and
income as proxies for each other45. They measure different phenomena and tap into different
causal mechanisms. Income can vary at similar education levels, mainly across different social
(eg., age, sex, race, ethnic) groups.
Many researchers have accepted that education/income should be included and
considered alongside with standard risk factors in risk prediction. They recognized that, even
in the well-known Framingham risk score, it was underestimated in low SES and overestimated
in the highest SES groups84, 85. Molshatzki and colleagues86 found that long-term post-MI
prediction model considerably gained improvement when education/income was considered
into the model. Gerber and colleague87 studied income-by-education interaction in post-MI
patients which showed that low income with low education patients had higher mortality risk,
because they failed to attend cardiac rehabilitation, and did not adhered to post-discharge
medication and lifestyle recommendations.
22
2.6 Conceptual framework
• Education may directly affect major cardiovascular events and its effect may also be
indirectly mediated through income effect.
• Income may directly affect major cardiovascular events and its effect may also be
indirectly mediated through education effect.
23
Education Income
Health Literacy
- Adherence of medication
- Awareness of importance of medical check up
- Ability to read or understand comprehend medical information - Awareness of impact of life style behavior
Psychological
- Depression
- Coping of cumulative stress
Environment/ Neighborhood - Living
conditions
- Environment conditions
Health Service Utilization
- Understanding of Post-cardiac rehabilitation
- Ability to navigate health care
- Understanding in management of disease progress
Co-morbid condition
- Dyslipidemia
- Hypertension
- Diabetes
Cardiovascular Events
Behavior or lifestyles
- Diet pattern
- Smoking
- Physical Activity
- Alcohol
- Obesity
Age, Sex, Race
Conceptual framework
24
CHAPTER 3
METHODOLOGY
3.1 Study design and setting
Data from a prospective cohort of Electricity Generating Authority of Thailand
(EGAT) will be used88. Briefly, EGAT cohort is collaboration between Ramathibodi
Hospital and EGAT, which has included 3 parallel cohorts: EGAT1, EGAT2, EGAT3.
All EGAT’s employees were invited to participate by sending invitation letters. The
total number of 9,082 participants were randomly enrolled. EGAT1 (n=3,499), EGAT2
(n=2,999) and EGAT3 (n=2,584), in 1985, 1998 and 2009, respectively. All participants
gave written informed consent and underwent complete medical examination with self-
administered questionnaires and thorough laboratory tests. Then, they were regularly
resurveyed in every 5 years, except 12 years follow-up gap between first (1985) and
second (1997) survey between EGAT1/1 and EGAT1/2.
This study will use data of the EGAT1 cohort (see, Figure 3.1) because it mainly
covered the detailed information about education and income with cardiovascular risk
factors. The data from the second survey (EGAT1/2 in 1997) will be used as baseline
data, and they were followed up in 2002 (EGAT1/3), 2007 (EGAT1/4) and 2012
(EGAT1/5). The 15 years duration of follow-up is considered to be long enough to get
sufficient number of interested outcomes (i.e., myocardial infarct, death and stroke).
Figure 3.1 Time frame of first EGAT cohort and follow-up
EGAT1/1 1985
n = 3499
EGAT1/2 1997
n = 2967
EGAT1/3 2002
n = 2360
EGAT1/4 2007
n = 1958
EGAT1/5 2012
Time-frame of this study
Baseline 5-years 5-years 5-years
25
3.2 Study subjects
All studied subjects of second survey of EGAT1/2 will be included in the study.
The studied subjects will be excluded if subjects had outcome of interest (i.e.,
myocardial infarct, strokes or transient ischemic attack and cardiovascular death) before
the date of enrollment in 1997.
3.3 Variables and measurement
3.3.1 Interested outcomes
Primary outcome
The primary outcome of interest is incidence of major cardiovascular
events (MCVE) which are combined endpoints of CVD death, myocardial infarction
(MI), strokes, and transient ischemic attack (TIA).
Secondary outcomes
The secondary outcomes of interest are
• Cardiovascular death
• Non-fatal ischemic stroke and TIA
• Non-fatal MI
Definition of outcomes
Myocardial Infarction
Myocardial infarction is defined as “a condition of myocardial necrosis that
occurs due to myocardial ischemia”. The third universal definition of myocardial
infarction89 will be used for diagnosis of myocardial infarct in this study.
Stroke
Stroke is defined as “a neurological deficit attributed to an acute focal injury of
the central nervous system by a vascular cause, including cerebral infarction,
intracerebral hemorrhage and subarachnoid hemorrhage”. AHA/ASA expert consensus
document and an updated definition of stroke for 21st century90 will be used for diagnosis
of stroke in this study.
26
Transient Ischemic Attack
Transient ischemic attack is defined as “brief episodes of neurological
dysfunction resulting from focal cerebral ischemia not associated with permanent
cerebral infraction”. AHA/ASA scientific statement and definition and evaluation of
transient ischemic attack91 will be used for diagnosis of TIA in this study.
Cardiovascular death
Cardiovascular death is defined as “death from coronary artery disease including
myocardial infarction, sudden cardiac death or ischemic stroke”.
All-cause mortality
All-cause mortality is defined as “any death from any cause”.
3.3.2 Outcomes Verification
The outcomes wer verified in a previous study92 with various methods as
follows:
MCVE outcomes
All MCVE outcomes and dates of occurrence were detected and were checked
from the following sources:
a. Documents of EGAT 1/3, 1/4 and 1/5 surveys, which were taken at 5 years
intervals. The surveys were consisted of thorough participants’ history with
physical examination records and investigation results including EKG, CXR,
and etcetera.
b. The reimbursement information of EGAT cohort will be requested from EGAT
office and hospital.
c. The 3 government reimbursement schemes data from Comptroller General’s
Department, the National Health Security Office and the Social Security Office,
which cover >99% of health reimbursement of Thailand people will be obtained
by contacting to Central Office for Health Care Information
27
d. If in doubt about MCVE outcomes, telephone interview will be made and
hospital medical records will be acquired. If essential, other necessary
investigations (e.g., CT brain, MRI brain, coronary angiography) results will be
retrieved.
Mortality outcomes
Death or alive condition of all participants was checked by requesting data from
the Bureau of Policy and Strategy, Ministry of Public Health and death certificate
databases from the Ministry of Interior.
The causes of death were confirmed with patients’ death certificates and by
telephone interview to the passed away patients’ relatives. If the patient died at hospital,
the patients’ medical records will be retrieved. The causes of death were determined by
the consensus of the outcome verification team.
3.3.3 Study factors
Education
Education status was extracted from a self-administered questionnaire.
Education will be categorized into 3 groups as primary (0-6 years of education),
secondary (7 – 12 years of education) and tertiary (>12 years of education).
Income
Income was extracted from a self-administered questionnaire. Income will be
categorized into 3 groups: low (<20,000 Baht), middle (20,000 – 50,000 Baht), and high
(>50,000 Baht).
Diabetes
Diabetes is diagnosed if the participant had history of diabetes or had fasting
blood sugar ≥126 mg/dl, or had documented taking anti-diabetes medication93.
Hypertension
Blood pressure measurement was obtained from patient records. Blood pressure
was measured twice after 5 minutes rest. Hypertension was diagnosed if the participant
28
had history of hypertension, systolic blood pressure (SBP) ≥140 mmHg or diastolic
blood pressure (DBP) ≥90 mmHg, or had been taking prescribed blood pressure
lowering medication94.
Body Mass Index
Body mass index (BMI) will be calculated from the recorded weight in
kilograms divided by squared height in meters. BMI will be categorized as: underweight
(<18.5 kg/m2), normal (18.5 – 24.9 kg/m2), overweight (25 – 29.9 kg/m2) and obese
(>30 kg/m2)95.
Smoking
Smoking will be extracted from self-administered questionnaire. Smoking will
be categorized as: non-smokers (persons who had never smoked or had smoked fewer
than 100 cigarettes in their lifetime), current smokers (persons having smoked at least
100 cigarettes in their lifetime and currently smoked cigarettes every day or some days),
and former smokers (persons having smoked at least 100 cigarettes in their lifetime and
did not currently smoke at the time of examination)96.
Hypercholesterolemia
Hypercholesterolemia was diagnosed if the participants had fasting serum
cholesterol ≥240 mg/dL, or had history of hypercholesterolemia or had documented
taking cholesterol-lowering medication97.
Hypertriglyceridemia
Hypertriglyceridemia was diagnosed if the participants had fasting serum
triglyceride ≥150 mg/dL, or had history of hypertriglyceridemia or had documented
taking triglyceride-lowering medication97.
High Density Lipoprotein
High density lipoprotein (HDL) is classified as low (participants had fasting
serum HDL <40mg/dL or had history of low HDL or taking HDL-raising medication),
normal (participants had fasting serum HDL between 40 – 59 mg/dL without taking any
29
HDL-raising medication), and high (participants had fasting serum HDL ≥60 mg/dL
without taking any HDL-raising medication)97.
3.4 Data collection
From EGAT1 survey, data recorded at baseline in 1997 will be retrieved. They
were collected by self-administered questionnaires, which consisted of general
demographic data (age, gender, educational level, income, place of living), behavioral
data (smoking status, alcohol consumption), family history of illness, underlying
diseases (diabetes, hypertension, stroke/TIA, chronic kidney disease, dyslipidemia), and
use of medication. In addition, data of physical examination done by clinicians,
cardiologists and trained personnel from Ramathibodi Hospital will be collected. These
include weight, height, waist circumference, hip circumference, SBP, DBP. Blood was
collected after fasting overnight for 12 hours before blood examination. Complete blood
count, fasting plasma glucose, lipid profile (total cholesterol, LDL, HDL, triglyceride),
creatinine, and uric acid will be collected. Electrocardiography and chest X-rays data
will also be collected.
3.5 Sample size estimation
Sample size estimation was calculated based on more than two groups of
proportions calculation technique. Our meta-analysis showed that the proportion of
CVE in high, medium and low education groups were ranged from 2% to 28%, 3% to
27%, and 5% to 38%, respectively. Distributions of subjects with high versus medium
education, and high to low education were 1:3 and 1:5, respectively. Type I and Type II
error were set at 5% and 20%, respectively. Assigning rate of CVE in high education
was 2%, minimal detectable sizes were 4% and 1% for low and medium versus high
education, respectively. Sample size was calculated using STATA version 14.1, with
the following commands98:
. artbin, pr(.02 .03 .05) ngroup(3) aratio(1 3 5) alpha(.05) power(.8)
ART - ANALYSIS OF RESOURCES FOR TRIALS (version 1.0.0, 3 March 2004)
------------------------------------------------------------------------------
A sample size program by Abdel Babiker, Patrick Royston & Friederike Barthel,
MRC Clinical Trials Unit, London NW1 2DA, UK.
30
------------------------------------------------------------------------------
Type of trial Superiority - binary outcome
Statistical test assumed Unconditional comparison of 3
binomial proportions
Number of groups 3
Allocation ratio 1.00:3.00.00:5.00
Anticipated event probabilities 0.020, 0.030, 0.050
Alpha 0.050 (two-sided)
Power (designed) 0.800
Total sample size (calculated) 2775
Expected total number of events 111
------------------------------------------------------------------------------
For income, the proportion of CVE in high, medium and low income groups
were ranged from 2% to 24%, 3% to 25%, and 5% to 39%, respectively and ratio of
subjects with high to medium income and high to low income were 1:1 and 1:2,
respectively. Type I and Type II error were set at 5% and 20%, respectively. Assigning
rate of CVE in high income was 2%, minimal detectable sizes were 4% and 1% for low
and medium versus high income, respectively. Sample size was calculated using
STATA version 14.1, with the following commands98:
. artbin, pr(.02 .03 .05) ngroup(3) aratio(1 1 2) alpha(.05) power(.8)
ART - ANALYSIS OF RESOURCES FOR TRIALS (version 1.0.0, 3 March 2004)
------------------------------------------------------------------------------
A sample size program by Abdel Babiker, Patrick Royston & Friederike Barthel,
MRC Clinical Trials Unit, London NW1 2DA, UK.
------------------------------------------------------------------------------
Type of trial Superiority - binary outcome
Statistical test assumed Unconditional comparison of 3
binomial proportions
Number of groups 3
Allocation ratio 1.00:1.00.00:2.00
Anticipated event probabilities 0.020, 0.030, 0.050
Alpha 0.050 (two-sided)
Power (designed) 0.800
Total sample size (calculated) 2061
Expected total number of events 78
------------------------------------------------------------------------------
31
From these two calculations, total sample size of 2775 will be included in this study
because this sample size does not exceed the number of participants of EGAT 1/2
(n=2967) and was the larger sample size among two calculations.
3.6 Data analysis
3.6.1 Data management plan
Baseline database will be extracted from EGAT data in 1997 (EGAT 1/2) which
contained demography, behavioral, underlying disease, family history, physical
examination, laboratory and imaging information. Then, outcomes databases (type of
outcomes and date of events) will be retrieved as described above. After that, these
outcome databases will then be merged with baseline database for analysis.
Data will be summarized and checked for missing and outliers for each variable.
Data validation will be cross-checked with relevant factors, e.g., with underlying
diseases and current medication. The consistency of data will be checked with 3 time
points of survey (EGAT1/3, 1/4 and 1/5). If inconsistency presents, source documents
(case recode form) will be explored to validate data. If needed, the health-section of
EGAT will be contacted or participants will be contacted to get more accurate data. If
missing or outliers cannot be solved in these ways, imputation of data will be done.
3.6.2 Data imputation
Imputation methods
Missing data will be imputed using a multivariate imputation with chained
equations (MICE) method99, 100 with the assumption that data are missing at random
(MAR). By assuming that all variables in data set were missing, each missing variable
will be modeled conditionally on the remaining variables in the data set until no missing
variable remains.
Variables selection
To preserve all the main characteristics of the observed data and to avoid bias
and gain precision, the imputation model will include all variables that are in the analysis
model, i.e., predictors (education, income), outcomes (MCVE) and covariates (age,
32
gender, smoking, alcohol, height, weight, systolic blood pressure, diastolic blood
pressure, fasting plasma glucose, cholesterol, triglyceride, HDL levels, and etcetera).
Imputation modeling
MICE method will involve a series of univariate models, i.e., impute the data on
a variable by variable basis by specifying an imputation model per variable. Education
and income levels will be imputed using ordinal regression. Smoking and alcohol status
will be imputed using multi-logit regression. Continuous variables such as height,
weight, systolic blood pressure, diastolic blood pressure, fasting plasma glucose,
cholesterol, triglyceride, and HDL levels will be imputed using linear regression. The
missing BMI variable will be calculated from imputed height and weight. The other
missing categorical variables (e.g., diabetes, hypertension, and dyslipidemia) will be
classified and imputed using data from imputed continuous variables such as systolic
blood pressure, diastolic blood pressure, fasting plasma glucose, cholesterol,
triglyceride, and HDL levels. This entire process of iteration will be repeated until no
missing variable is left. The observed data and the final set of imputed values will then
constitute one “complete” data set (one imputation).
Numbers of imputations
First, 5 to 10 imputations101 will be done. Then, it will be checked with the
largest fraction of missing information (FMI). The proper number of imputations should
be ≥ 100 × FMI, otherwise, many more imputations will be imputed102.
Multiple imputation diagnostics
MAR assumption will be assessed by performing diagnostic plots103 using the
“midiagplots” command in STATA. Distribution of observed and imputed data will be
explored and if they are not different, we will assume that selected models show less
bias.
33
3.6.3 Statistical analysis
Baseline characteristics of the cohort in EGAT1/2 will be described using mean
and standard deviation or median and range where appropriate for continuous data, and
using frequency and percentage for categorical data.
In this study, analysis will be done in 2 parts, mediation analysis and time-to-
event analysis as follows:
3.6.3.1 Mediation analysis
Mediation analysis will be conducted using rationale and statistical procedures
outlined by MacKinnon and his colleagues104. Two mediation analyses will be
performed according to two separate pathways as shown in Figure 3.2 below:
Figure 3.2 Pathway A (solid line) – effect of education on CVE through income as
the mediator; Pathway B (dashed line) – effect of income on CVE through
education as the mediator
In pathway A (solid line), the effect of education on MCVE will be determined
by setting them as the study factor and the interested outcome, respectively. Income will
be considered as mediator. Briefly, income is said to be mediated if (i) education has a
statistically significant effect on the income (ii) the income is associated with the MCVE
after controlling for the education effect (iii) the mediated effect is statistically
significant. Two equations from causal pathway will be constructed.
Cardiovascular risk factors (Z)
Income
Cardiovascular Events (Y) Education
b a c'
b c'
a
M1
X1
X2
M2
34
Figure 3.3 Statistical Diagram – effect of education on MCVE through income as
the mediator
First, income is regressed on education using multinomial logistic regression,
called path a, see equation 1 & 2.
ln �𝑃𝑃(𝑖𝑖𝑖𝑖𝑖𝑖1|𝑒𝑒𝑒𝑒𝑒𝑒1)𝑃𝑃(𝑖𝑖𝑖𝑖𝑖𝑖1|𝑒𝑒𝑒𝑒𝑒𝑒2)
� = 𝑎𝑎01 + 𝑎𝑎11𝑒𝑒𝑒𝑒𝑒𝑒1+ 𝑎𝑎12𝑒𝑒𝑒𝑒𝑒𝑒2 (1)
ln �𝑃𝑃(𝑖𝑖𝑖𝑖𝑖𝑖2|𝑒𝑒𝑒𝑒𝑒𝑒1)𝑃𝑃(𝑖𝑖𝑖𝑖𝑖𝑖2|𝑒𝑒𝑒𝑒𝑒𝑒2)
� = 𝑎𝑎02 + 𝑎𝑎21𝑒𝑒𝑒𝑒𝑒𝑒1+ 𝑎𝑎22𝑒𝑒𝑒𝑒𝑒𝑒2 (2)
where,
𝑒𝑒𝑒𝑒𝑒𝑒1 Low versus high education
𝑒𝑒𝑒𝑒𝑒𝑒2 Medium versus high education
𝑖𝑖𝑖𝑖𝑖𝑖1 Low versus high income
𝑖𝑖𝑖𝑖𝑖𝑖2 Medium versus high income
Second, the MCVE will be set by fitting MCVE on education and income
mediator using logistic regression model, called path b, c’, see equation 3.
Income1
(L vs H)
Income2
(M vs H)
Education1
(L vs H)
Education2
(M vs H)
a12 a11 b1
b2 a21
a22
c'1
MCVE c'2
35
𝑙𝑙𝑙𝑙𝑙𝑙𝑖𝑖𝑙𝑙(𝑌𝑌) = 𝑏𝑏0 + 𝑏𝑏1𝑖𝑖𝑖𝑖𝑖𝑖1 + 𝑏𝑏2𝑖𝑖𝑖𝑖𝑖𝑖2 + 𝑖𝑖′1𝑒𝑒𝑒𝑒𝑒𝑒𝑖𝑖1 + 𝑖𝑖′2𝑒𝑒𝑒𝑒𝑒𝑒2 (3)
Hence, substituting in equation for income1 and income2:
ln � 𝑃𝑃(𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀+)1−𝑃𝑃(𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀+)
� = 𝑏𝑏0 + 𝑏𝑏1(𝑎𝑎01 + 𝑎𝑎11𝑒𝑒𝑒𝑒𝑒𝑒1+ 𝑎𝑎12𝑒𝑒𝑒𝑒𝑒𝑒2) + 𝑏𝑏2(𝑎𝑎02 + 𝑎𝑎21𝑒𝑒𝑒𝑒𝑒𝑒1+ 𝑎𝑎22𝑒𝑒𝑒𝑒𝑒𝑒2) + 𝑖𝑖′1𝑒𝑒𝑒𝑒𝑒𝑒𝑖𝑖1 + 𝑖𝑖′2𝑒𝑒𝑒𝑒𝑒𝑒2
= 𝑏𝑏0 + 𝑎𝑎01𝑏𝑏1 + 𝑎𝑎11𝑏𝑏1𝑒𝑒𝑒𝑒𝑒𝑒1+ 𝑎𝑎12𝑏𝑏1𝑒𝑒𝑒𝑒𝑒𝑒2 + 𝑎𝑎02𝑏𝑏2 + 𝑎𝑎21𝑏𝑏2𝑒𝑒𝑒𝑒𝑒𝑒1+ 𝑎𝑎22𝑏𝑏2𝑒𝑒𝑒𝑒𝑒𝑒2 + 𝑖𝑖′1𝑒𝑒𝑒𝑒𝑒𝑒𝑖𝑖1 + 𝑖𝑖′2𝑒𝑒𝑒𝑒𝑒𝑒2
= 𝑏𝑏0 + 𝑎𝑎01𝑏𝑏1 + 𝑎𝑎02𝑏𝑏2 + 𝑎𝑎11𝑏𝑏1𝑒𝑒𝑒𝑒𝑒𝑒1 + 𝑎𝑎21𝑏𝑏2𝑒𝑒𝑒𝑒𝑒𝑒1+ 𝑖𝑖′1𝑒𝑒𝑒𝑒𝑒𝑒𝑖𝑖1 + 𝑎𝑎12𝑏𝑏1𝑒𝑒𝑒𝑒𝑒𝑒2+ 𝑎𝑎22𝑏𝑏2𝑒𝑒𝑒𝑒𝑒𝑒2 + 𝑖𝑖′2𝑒𝑒𝑒𝑒𝑒𝑒2
ln � 𝑃𝑃(𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀+)1−𝑃𝑃(𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀+)
� = (𝑏𝑏0 + 𝑎𝑎01𝑏𝑏1 + 𝑎𝑎02𝑏𝑏2) + (𝑎𝑎11𝑏𝑏1 + 𝑎𝑎21𝑏𝑏2+𝑖𝑖′1) 𝑒𝑒𝑒𝑒𝑒𝑒𝑖𝑖1 + (𝑎𝑎12𝑏𝑏1+ 𝑎𝑎22𝑏𝑏2 + 𝑖𝑖′2) 𝑒𝑒𝑒𝑒𝑒𝑒2
After that, by using Iacobucci105 and MacKinnon & Cox106 proposed equations, using
the parameter estimates a and b and their standard errors, sa and sb, the standardized
coefficient za from equation (1 & 2) and zb from equation (3) will be computed as
follow:
𝑧𝑧𝑎𝑎 = 𝑎𝑎/𝑠𝑠𝑎𝑎,
𝑧𝑧𝑏𝑏 = 𝑏𝑏/𝑠𝑠𝑏𝑏,
Because, za and zb are estimated in separate equations, there is possible presence of a
correlation between the two coefficients. Considered together with correlation, the final
zmediation will be computed as follow:
𝑧𝑧𝑚𝑚𝑒𝑒𝑒𝑒𝑖𝑖𝑎𝑎𝑚𝑚𝑖𝑖𝑚𝑚𝑖𝑖 = 𝑧𝑧𝑧𝑧𝑎𝑎𝑧𝑧𝑏𝑏 = 𝑧𝑧𝑎𝑎𝑧𝑧𝑏𝑏𝜎𝜎𝑧𝑧𝑎𝑎𝑏𝑏
= 𝑧𝑧𝑎𝑎𝑧𝑧𝑏𝑏+𝑟𝑟
�𝑧𝑧𝑎𝑎2+𝑧𝑧𝑏𝑏2+2𝑧𝑧𝑎𝑎𝑧𝑧𝑏𝑏𝑟𝑟+1+𝑟𝑟2
Since, there are four indirect effects of education on MCVE:
𝑎𝑎11𝑏𝑏1 , 𝑎𝑎21𝑏𝑏2 , 𝑎𝑎12𝑏𝑏1 , 𝑎𝑎22𝑏𝑏2
Therefore, zmediation will be:
𝑧𝑧𝑚𝑚𝑒𝑒𝑒𝑒𝑖𝑖𝑎𝑎𝑚𝑚𝑖𝑖𝑚𝑚𝑖𝑖 = 𝑧𝑧𝑧𝑧𝑎𝑎11𝑧𝑧𝑏𝑏1+𝑧𝑧𝑧𝑧𝑎𝑎21𝑧𝑧𝑏𝑏2+𝑍𝑍𝑧𝑧𝑎𝑎12𝑧𝑧𝑏𝑏1+𝑍𝑍𝑧𝑧𝑎𝑎22𝑧𝑧𝑏𝑏2
36
After that, zmediation will then be tested against a standardized normal, i.e., there is a
significant mediated effect if zmediation exceeds |1.96| for 2 tailed tests with α=0.05 level.
A 1000 replicated bootstrapping analysis will be applied without requiring the
assumption of normality. For each bootstrap sample, the mediation effects will be
estimated, then the approximation of mediation effects and their 95% CI will be
determined using bias-corrected bootstrap technique107. The total mediation effects will
be estimated by summation of all mediation effects.
For pathway B (dashed line), the effect of income on CVE will be determined
by setting them as the study factor and the interested outcome, respectively. Education
will be considered as mediator. We will use the same procedure of mediation analyses
as in pathway A, except switching education and income.
Time-to-event analysis
The composite of MCVE including MI, ischemic stroke/TIA, and cardiovascular
death will be considered as outcomes of interest. Time from enrollment from 1997 to
any diagnosis of composite MCVE and death from any causes will be calculated for
each subject. Subjects will be censored if they were free from MCVE at the end of the
study period, or if they were lost to follow-up. Univariate analysis will be conducted
initially by simple Cox regression model. Factors, which have p-value <0.1 from
univariate analysis, will be explored further in multivariate Cox proportional hazard
modeling with forward selection. Finally, the adjusted hazard ratio (HR) and their 95%
CIs will be calculated for each significant variable included in the model. P-value less
than 0.05 will be considered as statistical significance.
If the death rate of cohort ≥ 5%, the competing risk regression analysis with
cause-specific hazard function will be calculated by considering death from other causes
which will be considered as competing risk of MCVE outcome. The risk set is defined
as a group subjects who are free from MCVE or death, and thus they are at risk for both
events at time t. Subjects who have experienced MCVE or death will be removed from
the risk set for further observation after time t.
37
Univariate analysis will be performed using simple competing risk regression
model by fitting MCVE and other co-variables. Co-variables with p-value <0.1 will be
considered simultaneously to include in the multivariate competing risk regression
model. Overall cumulative incidence function (CIF) will be estimated using cause-
specific hazard function. The incidence rate of MCVE, MI, and stroke/TIA by each
group of risk factors will be estimated. Cause-specific hazard ratios (csHR) along with
95% CIs for education, income and other significant variables will be estimated. P-value
<0.05 will be considered as statistical significance.
Furthermore, each component of MCVE will be further explored. The
competing risk regression will also be applied by considering one event as interested
event and the rests as competing risk events.
All statistical analysis will be done using STATA version 14.1 and p-value <0.05
will be considered as statistically significant.
3.7 Ethics considerations
This retrospective study will use the demographic, medical, laboratory data from
EGAT project. The permission to access database will be asked to the principal
investigator (PI) of EGAT project at the Faculty of Medicine, Ramathibodi Hospital,
Mahidol University. They will be clearly informed about the objectives, benefits and
methodology of this study before making a decision about the permission. After
permission, study progression will be monthly reported to PI of EGAT for monitoring
and auditing for correct usage of data. If there is evidences of misuse of data or mismatch
from proposed objectives, PI of EGAT will have full authority to terminate this study.
Respect for human rights and autonomy
EGAT cohort participants were absolutely voluntary and has already given
written informed consent including blood analysis. All participants had the right to ask
for further information or withdraw their participation from this study at any time.
This study will be use the answered questionnaires, received physical
examinations and blood test performed data in 1997. The outcome data will be extracted
from the EGAT cohort outcome monitoring. No additional blood tests or additional
38
investigation will be performed for this study apart from validation of unclear
information in the original EGAT protocol.
This study protocol will be submitted and will ask for approval by the
Institutional Review Board of Ramathibodi’s Ethical Committee.
Confidentiality
The information provided by patient will be kept confidential. Their personal
information will be concealed and only authorized persons will be able to see this
information.
Beneficence
The result of this study may not contain benefit directly to individual
participants. However, participants who join to this project will receive health education
and health advice throughout the study period. They are allowed to participate in any
health activities arranged by the EGAT investigators team. The result of this study will
provide benefits to their society.
Non-maleficence
This study is an observational study and there will be no additional potential
intervention are provided to participants. In addition to this, they can withdraw from this
study at any time. Therefore, there is no more than minimal risk.
Justice
Standard hospital operational procedures are provided to all participants the
same as those that are provided to those who do not participate or patients who
withdrawal from this research.
39
3.8 Budget
Cost Unit No. of
unit
Unit costs
(Baht)
Total
(Baht)
A. Study related cost
1. Project management Project 1 5% 30,500
2. Research Project Director Month 6 20,000 120,000
3. Research Coordinator Month 6 15,000 90,000
4. Research Assistant Month 6 10,000 60,000
5. Ethical approval Site 1 5,000 5,000
6. Statistical plan and report Project 1 150,000 150,000
7. Manuscript publication Article 1 50,000 50,000
B. Patient related cost
1. Patient data confirmation for
inconsistent cases
Records 3000 20 60,000
C. Data related cost
1. Database Programming Project 1 15,000 15,000
2. Data cleaning, checking Project 1 15,000 15,000
3. Data analysis Project 1 15,000 15,000
D. Site related cost
1. Meeting Meeting 3 10,000 30,000
TOTAL 640,500
40
3.9 Time frame
Task 2016 2017
Dec Jan Feb Mar Apr May June
Ask permission to access EGAT data
Proposed to Ethics Committee
Retrieve data from 2nd EGAT cohort,
and identify eligible subjects
Data Cleaning, Checking, Imputation
Statistical analysis
Writing manuscript
Manuscript submission
ACKNOWLEDGEMENTS
This study proposal is a part of the dissertation for WIN KHAING’s training in
PhD (Clinical Epidemiology), Faculty of Medicine Ramathibodi Hospital and Faculty
of Graduates Studies, Mahidol University, Thailand.
41
REFERENCES
1. Shanthi Mendis, Tim Armstrong, Douglas Bettcher, Francesco Branca, Jeremy
Lauer, Cecile Mace, et al. WHO: Global status report on noncommunicable diseases
2014. Switzerland: World Health Organization; 2014.
2. Kannel WB, McGee DL. Diabetes and cardiovascular disease: the Framingham
study. Jama. 1979;241(19):2035-8.
3. Keil U, Kuulasmaa K. WHO MONICA Project: risk factors. Int J Epidemiol.
1989;18(3 Suppl 1):S46-55.
4. Yusuf S, Hawken S, Ôunpuu S, Dans T, Avezum A, Lanas F, et al. Effect of
potentially modifiable risk factors associated with myocardial infarction in 52 countries
(the INTERHEART study): case-control study. The Lancet. 2004;364(9438):937-52.
5. Murray CJ, Lopez AD. Measuring the global burden of disease. New England
Journal of Medicine. 2013;369(5):448-57.
6. Feigin VL, Roth GA, Naghavi M, Parmar P, Krishnamurthi R, Chugh S, et al.
Global burden of stroke and risk factors in 188 countries, during 1990–2013: a
systematic analysis for the Global Burden of Disease Study 2013. The Lancet
Neurology. 2016.
7. Lim SS, Vos T, Flaxman AD, Danaei G, Shibuya K, Adair-Rohani H, et al. A
comparative risk assessment of burden of disease and injury attributable to 67 risk
factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the
Global Burden of Disease Study 2010. The lancet. 2013;380(9859):2224-60.
8. Hemingway H, Philipson P, Chen R, Fitzpatrick NK, Damant J, Shipley M, et
al. Evaluating the quality of research into a single prognostic biomarker: a systematic
review and meta-analysis of 83 studies of C-reactive protein in stable coronary artery
disease. PLoS Med. 2010;7(6):e1000286.
9. Collaboration L-PS. Lipoprotein-associated phospholipase A 2 and risk of
coronary disease, stroke, and mortality: collaborative analysis of 32 prospective studies.
The Lancet. 2010;375(9725):1536-44.
10. Collaboration HS. Homocysteine and risk of ischemic heart disease and stroke:
a meta-analysis. Jama. 2002;288(16):2015-22.
42
11. Hoogeveen RC, Gaubatz JW, Sun W, Dodge RC, Crosby JR, Jiang J, et al. Small
Dense LDL Cholesterol Concentrations Predict Risk for Coronary Heart Disease: the
Atherosclerosis Risk in Communities (ARIC) Study. Arteriosclerosis, thrombosis, and
vascular biology. 2014;34(5):1069.
12. Ernst E, Resch KL. Fibrinogen as a cardiovascular risk factor: a meta-analysis
and review of the literature. Annals of Internal Medicine. 1993;118(12):956-63.
13. Yusuf S, Reddy S, Ôunpuu S, Anand S. Global burden of cardiovascular
diseases part I: general considerations, the epidemiologic transition, risk factors, and
impact of urbanization. Circulation. 2001;104(22):2746-53.
14. Lang T, Lepage B, Schieber A-C, Lamy S, Kelly-Irving M. Social determinants
of cardiovascular diseases. Public Health Reviews. 2012;33(2):601-22.
15. McKee M, Chow CK. The Social Determinants of Cardiovascular Disease.
Evidence-Based Cardiology, Third Edition. 2010:211-20.
16. Rose G, Marmot M. Social class and coronary heart disease. British heart
journal. 1981;45(1):13-9.
17. Smith GD, Bartley M, Blane D. The Black report on socioeconomic inequalities
in health 10 years on. BMJ: British Medical Journal. 1990;301(6748):373.
18. Marmot MG, Stansfeld S, Patel C, North F, Head J, White I, et al. Health
inequalities among British civil servants: the Whitehall II study. The Lancet.
1991;337(8754):1387-93.
19. 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(2):209-
20.
20. Lin CC, Rogot E, Johnson NJ, Sorlie PD, Arias E. A further study of life
expectancy by socioeconomic factors in the National Longitudinal Mortality Study.
Ethnicity & disease. 2002;13(2):240-7.
21. Nietert PJ, Sutherland SE, Keil JE, Bachman DL. Demographic and biologic
influences on survival in whites and blacks: 40 years of follow-up in the Charleston
heart study. International journal for equity in health. 2006;5(1):1.
43
22. Beebe-Dimmer J, Lynch JW, Turrell G, Lustgarten S, Raghunathan T, Kaplan
GA. Childhood and Adult Socioeconomic Conditions and 31-Year Mortality Risk in
Women. American Journal of Epidemiology. 2004;159(5):481-90.
23. Jacobsen BK, Thelle DS. Risk factors for coronary heart disease and level of
education the tromsø heart study. American Journal of Epidemiology. 1988;127(5):923-
32.
24. Cirera L, Tormo M-J, Chirlaque M-D, 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(8):755-63.
25. Panagiotakos D, Georgousopoulou E, Notara V, Pitaraki E, Kokkou E,
Chrysohoou C, et al. Education status determines 10-year (2002-2012) survival from
cardiovascular disease in Athens metropolitan area: the ATTICA study, Greece. Health
& social care in the community. 2016;24(3):334-44.
26. Mackenbach JP, Cavelaars A, Kunst AE, Groenhof F. Socioeconomic
inequalities in cardiovascular disease mortality. An international study. European heart
journal. 2000;21(14):1141-51.
27. Gallo LC, Matthews KA, Kuller LH, Sutton-Tyrrell K, Edmundowicz D.
Educational attainment and coronary and aortic calcification in postmenopausal women.
Psychosomatic medicine. 2001;63(6):925-35.
28. Notara V, Panagiotakos D, Kogias Y, Stravopodis P, Antonoulas A, Zombolos
S, et al. The impact of education status on the 10-year (2004-2014) cardiovascular
disease incidence and all cause mortality, among Acute Coronary Syndrome patients:
the GREECS longitudinal study. Journal of Preventive Medicine and Public Health.
2016.
29. Andersen I, Osler M, Petersen L, Grønbæk M, Prescott E. Income and risk of
ischaemic heart disease in men and women in a Nordic welfare country. International
Journal of Epidemiology. 2003;32(3):367-74.
30. Salomaa V, Niemelä M, Miettinen H, Ketonen M, Immonen-Räihä 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(16):1913-8.
44
31. Lammintausta A, Immonen-Räihä P, Airaksinen JKE, Torppa J, Harald K,
Ketonen M, et al. Socioeconomic Inequalities in the Morbidity and Mortality of Acute
Coronary Events in Finland: 1988 to 2002. Annals of Epidemiology. 2012;22(2):87-93.
32. Alter DA, Chong A, Austin PC, Mustard C, Iron K, Williams JI, et al.
Socioeconomic status and mortality after acute myocardial infarction. Annals of internal
medicine. 2006;144(2):82-93.
33. Rao SV, Schulman KA, Curtis LH, Gersh BJ, Jollis JG. Socioeconomic status
and outcome following acute myocardial infarction in elderly patients. Archives of
internal medicine. 2004;164(10):1128-33.
34. Fernald LC, Adler NE. Blood pressure and socioeconomic status in low-income
women in Mexico: a reverse gradient? J Epidemiol Community Health. 2008;62(5):e8.
35. Xu F, Tse LA, Yin X, Yu IT-s, Griffiths S. Impact of socio-economic factors on
stroke prevalence among urban and rural residents in Mainland China. BMC Public
Health. 2008;8(1):1.
36. Manrique-Garcia E, Sidorchuk A, Hallqvist J, Moradi T. Socioeconomic
position and incidence of acute myocardial infarction: a meta-analysis. Journal of
Epidemiology and Community Health. 2011;65(4):301-9.
37. Cox AM, McKevitt C, Rudd AG, Wolfe CD. Socioeconomic status and stroke.
The Lancet Neurology. 2006;5(2):181-8.
38. Marshall IJ, Wang Y, Crichton S, McKevitt C, Rudd AG, Wolfe CD. The effects
of socioeconomic status on stroke risk and outcomes. The Lancet Neurology.
2015;14(12):1206-18.
39. Chen R, Hu Z, Chen R-L, Zhang D, Xu L, Wang J, et al. Socioeconomic
deprivation and survival after stroke in China: a systematic literature review and a new
population-based cohort study. BMJ open. 2015;5(1):e005688.
40. Vathesatogkit P, Batty GD, Woodward M. Socioeconomic disadvantage and
disease-specific mortality in Asia: systematic review with meta-analysis of population-
based cohort studies. J Epidemiol Community Health. 2014;68(4):375-83.
41. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a
review of the literature. Circulation. 1993;88(4):1973-98.
42. Gerber Y, Goldbourt U, Drory Y. Interaction between income and education in
predicting long-term survival after acute myocardial infarction. European journal of
45
cardiovascular prevention and rehabilitation : official journal of the European Society
of Cardiology, Working Groups on Epidemiology & Prevention and Cardiac
Rehabilitation and Exercise Physiology. 2008;15(5):526-32.
43. Rasmussen JN, Rasmussen S, Gislason GH, Buch P, Abildstrom SZ, Kober L,
et al. Mortality after acute myocardial infarction according to income and education. J
Epidemiol Community Health. 2006;60(4):351-6.
44. Lemstra M, Rogers M, Moraros J. Income and heart disease Neglected risk
factor. Canadian Family Physician. 2015;61(8):698-704.
45. Braveman PA, Cubbin C, Egerter S, Chideya S, Marchi KS, Metzler M, et al.
Socioeconomic status in health research: one size does not fit all. Jama.
2005;294(22):2879-88.
46. Levenson JW, Skerrett PJ, Gaziano JM. Reducing the global burden of
cardiovascular disease: the role of risk factors. Preventive cardiology. 2002;5(4):188-
99.
47. Olshansky SJ, Ault AB. The fourth stage of the epidemiologic transition: the age
of delayed degenerative diseases. The Milbank Quarterly. 1986:355-91.
48. World Health Organization. WHO | What are social determinants of health? :
World Health Organization; 2015
(http://www.who.int/social_determinants/sdh_definition/en/).
49. Winkleby MA, Jatulis DE, Frank E, Fortmann SP. Socioeconomic status and
health: how education, income, and occupation contribute to risk factors for
cardiovascular disease. Am J Public Health. 1992;82(6):816-20.
50. Havranek EP, Mujahid MS, Barr DA, Blair IV, Cohen MS, Cruz-Flores S, et al.
Social Determinants of Risk and Outcomes for Cardiovascular Disease A Scientific
Statement From the American Heart Association. Circulation. 2015;132(9):873-98.
51. Winkleby MA, Jatulis DE, Frank E, Fortmann SP. Socioeconomic status and
health: how education, income, and occupation contribute to risk factors for
cardiovascular disease. American journal of public health. 1992;82(6):816-20.
52. 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(10):934-42.
46
53. Hallqvist J, Lundberg M, Diderichsen F, Ahlbomb A. Socioeconomic
differences in risk of myocardial infarction 1971–1994 in Sweden: time trends, relative
risks and population attributable risks. International Journal of Epidemiology.
1998;27(3):410-5.
54. Hart CL, Hole DJ, Smith GD. The contribution of risk factors to stroke
differentials, by socioeconomic position in adulthood: the Renfrew/Paisley Study.
American Journal of Public Health. 2000;90(11):1788.
55. Avendano M, Kawachi I, Van Lenthe F, Boshuizen HC, Mackenbach JP, Van
den Bos GA, et al. Socioeconomic status and stroke incidence in the US elderly: the role
of risk factors in the EPESE study. Stroke. 2006;37(6):1368-73.
56. Berkman LF, Glass TA. Social integration, social networks, social support and
health. In: Berkman LF, Kawachi I, editors. Soical Epidemiology. New York: Oxford
University Press; 2000. p. 137 - 73.
57. Khaing W, Vallipakorn SA-O, Thakkinstian A. The effect of education and
income on cardiovascular outcomes: a systematic review and meta-analysis of
registration in PROSPERO 2016
(http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42016046615).
58. PELE L. Currency converter in the past with official exchange rates from 1953
2016 (http://fxtop.com/en/currency-converter-past.php). (Accessed 1st August 2016).
59. White IR. Multivariate random-effects meta-regression: updates to mvmeta.
Stata Journal. 2011;11(2):255.
60. World Bank. World Bank Country and Lending Groups World Bank Data Help
Desk 2016 (https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-
bank-country-and-lending-groups).
61. StataCorp. Stata Statistical Software: Release 14. College Station, TX:
StataCorp LP. 2015.
62. Naska A, Katsoulis M, Trichopoulos D, Trichopoulou A. The root causes of
socioeconomic differentials in cancer and cardiovascular mortality in Greece. European
Journal of Cancer Prevention. 2012;21(5):490-6.
63. Bostock S, Steptoe A. Association between low functional health literacy and
mortality in older adults: longitudinal cohort study. Bmj. 2012;344:e1602.
47
64. Meijer A, Conradi HJ, Bos EH, Thombs BD, van Melle JP, de Jonge P.
Prognostic association of depression following myocardial infarction with mortality and
cardiovascular events: a meta-analysis of 25 years of research. General hospital
psychiatry. 2011;33(3):203-16.
65. Myers V, Gerber Y, Benyamini Y, Goldbourt U, Drory Y. Post-myocardial
infarction depression: increased hospital admissions and reduced adoption of secondary
prevention measures—a longitudinal study. Journal of psychosomatic research.
2012;72(1):5-10.
66. Nielsen KM, Faergeman O, Foldspang A, Larsen ML. Cardiac rehabilitation:
health characteristics and socio-economic status among those who do not attend. The
European Journal of Public Health. 2008;18(5):479-83.
67. Kilander L, Berglund L, Boberg M, Vessby B, Lithell H. Education, lifestyle
factors and mortality from cardiovascular disease and cancer. A 25-year follow-up of
Swedish 50-year-old men. International Journal of Epidemiology. 2001;30(5):1119-26.
68. Steptoe A, Marmot M. The role of psychobiological pathways in socio-
economic inequalities in cardiovascular disease risk. European heart journal.
2002;23(1):13-25.
69. Suadicani P, Hein HO, Gyntelberg F. Strong mediators of social inequalities in
risk of ischaemic heart disease: a six-year follow-up in the Copenhagen Male Study.
International Journal of Epidemiology. 1997;26(3):516-22.
70. Blair AS, Lloyd-Williams F, Mair FS. What do we know about socioeconomic
status and congestive heart failure? A review of the literature. The Journal of family
practice. 2002;51(2):169-.
71. Wilkinson RG, Pickett KE. Income inequality and population health: a review
and explanation of the evidence. Social science & medicine. 2006;62(7):1768-84.
72. Lahelma E, Martikainen P, Laaksonen M, Aittomäki A. Pathways between
socioeconomic determinants of health. Journal of Epidemiology and Community
Health. 2004;58(4):327-32.
73. Ahmed AA, Zhang Y, Bourge RC, Kilgore ML, Williams B, Sawyer P, et al.
Abstract 12064: Low Income, Regardless of Education Level, is a Significant
Independent Predictor of Incident Heart Failure in Community-Dwelling, Medicare-
Eligible Older Adults. Circulation. 2011;124(Suppl 21):A12064-A.
48
74. Shavers VL. Measurement of socioeconomic status in health disparities
research. Journal of the national medical association. 2007;99(9):1013.
75. Marmot M, Ryff CD, Bumpass LL, Shipley M, Marks NF. Social inequalities in
health: next questions and converging evidence. Social science & medicine.
1997;44(6):901-10.
76. Marmot MG, Rose G, Shipley M, Hamilton PJ. Employment grade and coronary
heart disease in British civil servants. Journal of epidemiology and community health.
1978;32(4):244-9.
77. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for
systematic reviews and meta-analyses: the PRISMA statement. Annals of internal
medicine. 2009;151(4):264-9.
78. Andersen I, Gamborg M, Osler M, Prescott E, Diderichsen F. Income as
mediator of the effect of occupation on the risk of myocardial infarction: does the
income measurement matter? Journal of epidemiology and community health.
2005;59(12):1080-5.
79. Lu Y, Hajifathalian K, Rimm EB, Ezzati M, Danaei G. Mediators of the effect
of body mass index on coronary heart disease: decomposing direct and indirect effects.
Epidemiology. 2015;26(2):153-62.
80. MacKinnon DP, Fairchild AJ, Fritz MS. Mediation analysis. Annual review of
psychology. 2007;58:593.
81. Hoeymans N, Smit H, Verkleij H, Kromhout D. Cardiovascular risk factors in
relation to educational level in 36 000 men and women in The Netherlands. European
Heart Journal. 1996;17(4):518-25.
82. 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(5):2172-9.
83. Zaman MJ, Patel A, Jan S, Hillis GS, Raju PK, Neal B, et al. Socio-economic
distribution of cardiovascular risk factors and knowledge in rural India. International
journal of epidemiology. 2012:dyr226.
84. Ramsay SE, Morris RW, Whincup PH, Papacosta AO, Thomas MC,
Wannamethee SG. Prediction of coronary heart disease risk by Framingham and
SCORE risk assessments varies by socioeconomic position: results from a study in
49
British men. European Journal of Cardiovascular Prevention & Rehabilitation.
2011;18(2):186-93.
85. Tunstall-Pedoe H, Woodward M. By neglecting deprivation, cardiovascular risk
scoring will exacerbate social gradients in disease. Heart. 2006;92(3):307-10.
86. Molshatzki N, Drory Y, Myers V, Goldbourt U, Benyamini Y, Steinberg DM,
et al. Role of socioeconomic status measures in long-term mortality risk prediction after
myocardial infarction. Medical care. 2011;49(7):673-8.
87. Gerber Y, Goldbourt U, Drory Y. Interaction between income and education in
predicting long-term survival after acute myocardial infarction. European Journal of
Cardiovascular Prevention & Rehabilitation. 2008;15(5):526-32.
88. Vathesatogkit P, Woodward M, Tanomsup S, Ratanachaiwong W, Vanavanan
S, Yamwong S, et al. Cohort profile: the electricity generating authority of Thailand
study. Int J Epidemiol. 2012;41(2):359-65.
89. Thygesen K, Alpert JS, Jaffe AS, Simoons ML, Chaitman BR, White HD. Third
universal definition of myocardial infarction. Circulation. 2012;126(16):2020-35.
90. Sacco RL, Kasner SE, Broderick JP, Caplan LR, Culebras A, Elkind MS, et al.
An updated definition of stroke for the 21st century a statement for healthcare
professionals from the American Heart Association/American Stroke Association.
Stroke. 2013;44(7):2064-89.
91. Easton JD, Saver JL, Albers GW, Alberts MJ, Chaturvedi S, Feldmann E, et al.
Definition and Evaluation of Transient Ischemic Attack A Scientific Statement for
Healthcare Professionals From the American Heart Association/American Stroke
Association Stroke Council; Council on Cardiovascular Surgery and Anesthesia;
Council on Cardiovascular Radiology and Intervention; Council on Cardiovascular
Nursing; and the Interdisciplinary Council on Peripheral Vascular Disease: The
American Academy of Neurology affirms the value of this statement as an educational
tool for neurologists. Stroke. 2009;40(6):2276-93.
92. Nakarin Sansanayudh. The association between mean platelet volume and risk
of cardiovascular events: Mahidol University; 2015.
93. Gavin III JR, Alberti K, Davidson MB, DeFronzo RA. Report of the expert
committee on the diagnosis and classification of diabetes mellitus. Diabetes care.
1997;20(7):1183.
50
94. Chalmers J, MacMahon S, Mancia G, Whitworth J, Beilin L, Hansson L, et al.
1999 World Health Organization-International Society of Hypertension Guidelines for
the management of hypertension. Guidelines sub-committee of the World Health
Organization. Clinical and experimental hypertension (New York, NY: 1993).
1998;21(5-6):1009-60.
95. WHO EC. Appropriate body-mass index for Asian populations and its
implications for policy and intervention strategies. Lancet (London, England).
2004;363(9403):157.
96. Control CfD, Prevention. Cigarette smoking among adults--United States, 1992,
and changes in the definition of current cigarette smoking. MMWR Morbidity and
mortality weekly report. 1994;43(19):342.
97. Reiner Ž, Catapano AL, De Backer G, Graham I, Taskinen M-R, Wiklund O, et
al. ESC/EAS Guidelines for the management of dyslipidaemias. European heart journal.
2011;32(14):1769-818.
98. Barthel FM-S, Royston P, Babiker A. A menu-driven facility for complex
sample size calculation in randomized controlled trials with a survival or a binary
outcome: update. Stata J. 2005;5(1):123-9.
99. White IR, Royston P, Wood AM. Multiple imputation using chained equations:
issues and guidance for practice. Statistics in medicine. 2011;30(4):377-99.
100. Rubin DB, Schenker N. Multiple imputation in health‐are databases: An
overview and some applications. Statistics in medicine. 1991;10(4):585-98.
101. Van Buuren S, Boshuizen HC, Knook DL. Multiple imputation of missing blood
pressure covariates in survival analysis. Statistics in medicine. 1999;18(6):681-94.
102. Graham JW. Missing Data: Analysis and Design: Springer New York; 2012.
103. Eddings W, Marchenko Y. Diagnostics for multiple imputation in Stata. Stata
Journal. 2012;12(3):353.
104. MacKinnon DP. Analysis of mediating variables in prevention and intervention
research. NIDA research monograph. 1994;139:127-.
105. Iacobucci D. Mediation analysis and categorical variables: The final frontier.
Journal of Consumer Psychology, Forthcoming. 2012.
51
106. MacKinnon DP, Cox MC. Commentary on “Mediation analysis and categorical
variables: The final frontier” by Dawn Iacobucci. Journal of consumer psychology: the
official journal of the Society for Consumer Psychology. 2012;22(4):600.
107. Williams J, MacKinnon DP. Resampling and distribution of the product methods
for testing indirect effects in complex models. Structural Equation Modeling.
2008;15(1):23-51.
108. Arrich J, Lalouschek W, Müllner M. Influence of socioeconomic status on
mortality after stroke: Retrospective cohort study. Stroke. 2005;36(2):310-4.
109. Rehkopf DH, Eisen EA, Modrek S, Mokyr Horner E, Goldstein B, Costello S,
et al. Early-Life State-of-Residence Characteristics and Later Life Hypertension,
Diabetes, and Ischemic Heart Disease. Am J Public Health. 2015;105(8):1689-95.
110. Geyer S, Hemström Ö, Peter R, Vågerö D. Education, income, and occupational
class cannot be used interchangeably in social epidemiology. Empirical evidence against
a common practice. Journal of epidemiology and community health. 2006;60(9):804-
10.
111. Honjo K, Iso H, Inoue M, Tsugane S, Group JS. Education, Social Roles, and
the Risk of Cardiovascular Disease Among Middle-Aged Japanese Women The JPHC
Study Cohort I. Stroke. 2008;39(10):2886-90.
112. Rawshani A, Svensson AM, Rosengren A, Eliasson B, Gudbjörnsdottir S.
Impact of socioeconomic status on cardiovascular disease and mortality in 24,947
individuals with type 1 diabetes. Diabetes Care. 2015;38(8):1518-27.
113. Thurston RC, Kubzansky LD, Kawachi I, Berkman LF. Is the association
between socioeconomic position and coronary heart disease stronger in women than in
men? American Journal of Epidemiology. 2005;162(1):57-65.
114. Hetemaa T, Manderbacka K, Reunanen A, Koskinen S, Keskimäki I.
Socioeconomic inequities in invasive cardiac procedures among patients with incident
angina pectoris or myocardial infarction. Scandinavian Journal of Public Health.
2006;34(2):116-23.
115. Peter R, Gässler H, Geyer S. Socioeconomic status, status inconsistency and risk
of ischaemic heart disease: A prospective study among members of a statutory health
insurance company. Journal of Epidemiology and Community Health. 2007;61(7):605-
11.
52
116. Honjo K, Tsutsumi A, Kayaba K, Group JMSCS. Socioeconomic indicators and
cardiovascular disease incidence among Japanese community residents: the Jichi
Medical School Cohort Study. International journal of behavioral medicine.
2010;17(1):58-66.
117. Roux AVD, Merkin SS, Arnett D, Chambless L, Massing M, Nieto FJ, et al.
Neighborhood of residence and incidence of coronary heart disease. New England
Journal of Medicine. 2001;345(2):99-106.
118. Fujino Y, Tamakoshi A, Iso H, Inaba Y, Kubo T, Ide R, et al. A nationwide
cohort study of educational background and major causes of death among the elderly
population in Japan. Preventive medicine. 2005;40(4):444-51.
119. Lee Y-T, Lin RS, Sung FC, Yang C-Y, Chien K-L, Chen W-J, et al. Chin-Shan
Community Cardiovascular Cohort in Taiwan–baseline data and five-year follow-up
morbidity and mortality. Journal of clinical epidemiology. 2000;53(8):838-46.
120. Weikert C, Stefan N, Schulze MB, Pischon T, Berger K, Joost H-G, et al. Plasma
fetuin-a levels and the risk of myocardial infarction and ischemic stroke. Circulation.
2008;118(24):2555-62.
121. Hippe M, Vestbo J, Hein HO, Borch-Johnsen K, Jensen G, Sørensen T. Familial
predisposition and susceptibility to the effect of other risk factors for myocardial
infarction. Journal of epidemiology and community health. 1999;53(5):269-76.
122. Huisman M, Van Lenthe F, Avendano M, Mackenbach J. The contribution of
job characteristics to socioeconomic inequalities in incidence of myocardial infarction.
Social science & medicine. 2008;66(11):2240-52.
123. 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. American Journal of Epidemiology. 1992;135(8):854-64.
124. Bosma H, Appels A, Sturmans F, Grabauskas V, Gostautas A. Educational level
of spouses and risk of mortality: the WHO Kaunas-Rotterdam Intervention Study
(KRIS). International Journal of Epidemiology. 1995;24(1):119-26.
125. Chaix B, Rosvall M, Merlo J. Neighborhood socioeconomic deprivation and
residential instability: effects on incidence of ischemic heart disease and survival after
myocardial infarction. Epidemiology. 2007;18(1):104-11.
53
126. Kuper H, Adami H-O, Theorell T, Weiderpass E. Psychosocial determinants of
coronary heart disease in middle-aged women: a prospective study in Sweden.
American Journal of Epidemiology. 2006;164(4):349-57.
127. Lapidus L, Bengtsson C. Socioeconomic factors and physical activity in relation
to cardiovascular disease and death. A 12 year follow up of participants in a population
study of women in Gothenburg, Sweden. British heart journal. 1986;55(3):295-301.
128. Braig S, Peter R, Nagel G, Hermann S, Rohrmann S, Linseisen J. The impact of
social status inconsistency on cardiovascular risk factors, myocardial infarction and
stroke in the EPIC-Heidelberg cohort. BMC Public Health. 2011;11.
129. Jakobsen L, Niemann T, Thorsgaard N, Thuesen L, Lassen JF, Jensen LO, et al.
Dimensions of socioeconomic status and clinical outcome after primary percutaneous
coronary intervention. Circulation Cardiovascular interventions. 2012;5(5):641-8.
130. Rasmussen JN, Rasmussen S, Gislason GH, Abildstrom SZ, Schramm TK,
Torp-Pedersen C, et al. Persistent socio-economic differences in revascularization after
acute myocardial infarction despite a universal health care system - A Danish study.
Cardiovascular Drugs and Therapy. 2007;21(6):449-57.
131. Senan M, Petrosyan A. The relationship between socioeconomic status and
cardiovascular events. Georgian medical news. 2014(227):42-7.
132. Bosma H, Van Jaarsveld C, Tuinstra J, Sanderman R, Ranchor A, Van Eijk JTM,
et al. Low control beliefs, classical coronary risk factors, and socio-economic
differences in heart disease in older persons. Social science & medicine.
2005;60(4):737-45.
133. Masoudkabir F, Toghianifar N, Talaie M, Sadeghi M, Sarrafzadegan N,
Mohammadifard N, et al. Socioeconomic status and incident cardiovascular disease in
a developing country: Findings from the Isfahan cohort study (ICS). International
Journal of Public Health. 2012;57(3):561-8.
134. Van Minh H, Huong DL, Wall S, Byass P, Chuc NTK. Peer Reviewed:
Cardiovascular Disease Mortality and Its Association With Socioeconomic Status:
Findings From a Population-based Cohort Study in Rural Vietnam, 1999–2003.
Preventing chronic disease. 2006;3(3).
54
135. Hirokawa K, Tsutusmi A, Kayaba K. Impacts of educational level and
employment status on mortality for Japanese women and men: the Jichi Medical School
cohort study. European journal of epidemiology. 2006;21(9):641-51.
136. SIEGEL D, KULLER L, LAZARUS NB, BLACK D, FEIGAL D, HUGHES G,
et al. Predictors of cardiovascular events and mortality in the Systolic Hypertension in
the Elderly Program pilot project. American journal of epidemiology. 1987;126(3):385-
99.
137. He J, Ogden LG, Bazzano LA, Vupputuri S, Loria C, Whelton PK. Risk factors
for congestive heart failure in US men and women: NHANES I epidemiologic follow-
up study. Archives of internal medicine. 2001;161(7):996-1002.
138. Christensen S, Mogelvang R, Heitmann M, Prescott E. Level of education and
risk of heart failure: a prospective cohort study with echocardiography evaluation.
European heart journal. 2011;32(4):450-8.
139. Borné Y, Engström G, Essén B, Sundquist J, Hedblad B. Country of birth and
risk of hospitalization due to heart failure: a Swedish population-based cohort study.
European journal of epidemiology. 2011;26(4):275-83.
140. Philbin EF, Dec GW, Jenkins PL, DiSalvo TG. Socioeconomic status as an
independent risk factor for hospital readmission for heart failure. The American journal
of cardiology. 2001;87(12):1367-71.
141. Schwarz KA, Elman CS. Identification of factors predictive of hospital
readmissions for patients with heart failure. Heart & Lung: The Journal of Acute and
Critical Care. 2003;32(2):88-99.
142. Sui X, Gheorghiade M, Zannad F, Young JB, Ahmed A. A propensity matched
study of the association of education and outcomes in chronic heart failure. International
journal of cardiology. 2008;129(1):93-9.
143. Rosvall M, Engström G, Hedblad B, Janzon L, Göran B. The role of preclinical
atherosclerosis in the explanation of educational differences in incidence of coronary
events. Atherosclerosis. 2006;187(2):251-6.
144. Engström G, Tydén P, Berglund G, Hansen O, Hedblad B, Janzon L. Incidence
of myocardial infarction in women. A cohort study of risk factors and modifiers of
effect. Journal of epidemiology and community health. 2000;54(2):104-7.
55
145. Avendano M, Glymour MM. Stroke disparities in older Americans: is wealth a
more powerful indicator of risk than income and education? Stroke. 2008;39(5):1533-
40.
146. Li C, Hedblad B, Rosvall M, Buchwald F, Khan FA, Engström G. Stroke
incidence, recurrence, and case-fatality in relation to socioeconomic position a
population-based study of middle-aged swedish men and women. Stroke.
2008;39(8):2191-6.
147. van Rossum CT, van de Mheen H, Breteler MM, Grobbee DE, Mackenbach JP.
Socioeconomic differences in stroke among Dutch elderly women the Rotterdam Study.
Stroke. 1999;30(2):357-62.
148. Gillum R, Mussolino ME. Education, poverty, and stroke incidence in whites
and blacks: the NHANES I Epidemiologic Follow-up Study. Journal of clinical
epidemiology. 2003;56(2):188-95.
149. Kuper H, Adami H-O, Theorell T, Weiderpass E. The socioeconomic gradient
in the incidence of stroke a prospective study in middle-aged women in Sweden. Stroke.
2007;38(1):27-33.
150. Jackson CA, Jones M, Mishra GD. Educational and homeownership inequalities
in stroke incidence: a population-based longitudinal study of mid-aged women. The
European Journal of Public Health. 2014;24(2):231-6.
151. Andersen KK, Steding-Jessen M, Dalton SO, Olsen TS. Socioeconomic position
and incidence of ischemic stroke in denmark 2003-2012. A nationwide hospital-based
study. Journal of the American Heart Association. 2014;3(4).
152. Jakovljević D, Sarti C, Sivenius J, Torppa J, Mähönen M, Immonen-Räihä P, et
al. Socioeconomic status and ischemic stroke: The FINMONICA stroke register. Stroke.
2001;32(7):1492-8.
153. Zhou G, Liu X, Xu G, Liu X, Zhang R, Zhu W. The effect of socioeconomic
status on three-year mortality after first-ever ischemic stroke in Nanjing, China. BMC
Public Health. 2006;6.
154. Kim C, Eby E, Piette JD. Is education associated with mortality for breast cancer
and cardiovascular disease among black and white women? Gender Medicine.
2005;2(1):13-8.
56
155. Qureshi AI, Suri MFK, Saad M, Hopkins LN. Educational attainment and risk
of stroke and myocardial infarction. Medical Science Monitor. 2003;9(11):CR466-
CR73.
156. Pednekar MS, Gupta R, Gupta PC. Illiteracy, low educational status, and
cardiovascular mortality in India. BMC Public Health. 2011;11(1):1.
157. Coady SA, Johnson NJ, Hakes JK, Sorlie PD. Individual education, area income,
and mortality and recurrence of myocardial infarction in a Medicare cohort: the National
Longitudinal Mortality Study. BMC Public Health. 2014;14:705.
158. Gallo V, Mackenbach JP, Ezzati M, Menvielle G, Kunst AE, Rohrmann S, et al.
Social inequalities and mortality in Europe–results from a large multi-national cohort.
PLoS One. 2012;7(7):e39013.
159. Bucher HC, Ragland DR. Socioeconomic indicators and mortality from
coronary heart disease and cancer: a 22-year follow-up of middle-aged men. Am J
Public Health. 1995;85(9):1231-6.
160. Tonne C, Schwartz J, Mittleman M, Melly S, Suh H, Goldberg R. Long-term
survival after acute myocardial infarction is lower in more deprived neighborhoods.
Circulation. 2005;111(23):3063-70.
161. Ito S, Takachi R, Inoue M, Kurahashi N, Iwasaki M, Sasazuki S, et al. Education
in relation to incidence of and mortality from cancer and cardiovascular disease in Japan.
The European Journal of Public Health. 2008;18(5):466-72.
162. Van Minh H, Byass P, Wall S. Mortality from cardiovascular diseases in Bavi
District, Vietnam. Scandinavian Journal of Public Health. 2003;31(6 suppl):26-31.
163. 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(6):1308.
164. Rawshani A, Svensson A-M, Zethelius B, Eliasson B, Rosengren A,
Gudbjörnsdottir S. Association Between Socioeconomic Status and Mortality,
Cardiovascular Disease, and Cancer in Patients With Type 2 Diabetes. JAMA Internal
Medicine. 2016.
57
165. Rosvall M, Chaix B, Lynch J, Lindström M, Merlo J. The association between
socioeconomic position, use of revascularization procedures and five-year survival after
recovery from acute myocardial infarction. BMC Public Health. 2008;8(1):1.
166. Khang Y-H, Lynch J, Jung-Choi K, Cho H-J. Explaining age specific
inequalities in mortality from all causes, cardiovascular disease and ischaemic heart
disease among South Korean male public servants: relative and absolute perspectives.
Heart. 2007.
167. Rosvall M, Gerward S, Engström G, Hedblad B. Income and short-term case
fatality after myocardial infarction in the whole middle-aged population of Malmö,
Sweden. The European Journal of Public Health. 2008;18(5):533-8.
58
Table 2.1. Characteristics of included studied
Author Year
Country, Setting (income level)
Outcome Risk measure
Study factors (Categories)
Relative Risks (95%CI)
NC
Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2)
(a) Coronary artery diseases
Arrich108 2005 Austria, Europe IHD RR Education
(Medium vs High) 1.02 a (0.82, 1.28) 0 66.71 54.28 25.27 26.46 66.98 NA
(High) Education (Low vs High)
1.29 a (1.01, 1.65)
Income (Medium vs High)
1.07 a (0.92, 1.24) 0 65.5 54.35 27.03 27.6 44.95 NA
Income (Low vs High)
0.79 a (0.41, 1.50)
Rehkopf109 2015 US (High) IHD OR Education
(Medium vs High) 1.01 (1.01, 1.02) 9 47 78 8 NA 24 NA
Income (Medium vs High)
0.99 (0.98,1.00)
Geyer110 2006 Germany, Europe MI RR Education
(Medium vs High) 3.41 (2.18, 5.35) 1 42.5 72.4 NA NA NA NA
(High)
Education (Low vs High)
4.06 (2.14, 7.67)
Income (Medium vs High)
1.48 (1.24, 1.76)
Income (Low vs High)
2.02 (1.83, 2.23)
Honjo111 2008 Japan, Asia CHD HR Education (Medium vs High)
1.75 b (0.52, 5.88) 11 NA 0 2.36 7.28 14.82 NA
(High)
Education (Low vs High)
1.28 b (0.67, 2.35)
59
Table 2.1. Characteristics of included studied (continued)
Author Year
Country, Setting (income level)
Outcome Risk measure
Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2)
(a) Coronary artery diseases
Rawshani112 2015 Sweden, Europe CHD HR Education
(Medium vs High) 1.16 b, c (1.09, 1.23) 14 39.27 53.82 100 11.81 NA 25.61
(High)
Education (Low vs High)
1.35 b, c (1.10, 1.64)
Income (Medium vs High)
1.39 c (1.04, 1.86) 14 39.19 53.78 100 12.25 NA 25.6
Income (Low vs High)
1.86 c (1.54, 2.24)
Thurston113 2005 US (High) CHD HR Education
(Medium vs High) 1.22 c (0.95, 1.55) 13 47.42 45.64 3.75 38.11 6.54 25.59
Education (Low vs High)
1.40 c (1.10, 1.77)
Income (Medium vs High)
1.24 c (1.05, 1.46)
Income (Low vs High)
1.23 c (1.05, 1.43)
Salomaa30 2000 Finland, Europe (High)
MI RR Education (Low vs High)
1.49 c (1.42, 1.56) 2 NA NA NA NA NA NA
Income (Low vs High)
1.72 c (1.65, 1.79)
Andersen29 2003 Demark, Europe IHD HR Income
(Medium vs High) 1.18 c (1.05, 1.32) 9 52.71 46.41 NA 36.15 NA 24.96
(High)
Income (Low vs High)
1.45 c (1.30, 1.63)
60
Table 2.1. Characteristics of included studied (continued)
Author Year
Country, Setting (income level)
Outcome Risk measure
Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2)
(a) Coronary artery diseases
Hetemaa114 2006 Finland, Europe MI HR Education
(Medium vs High) 0.77 c (0.68, 0.86) 13 67.34 61.83 16.25 NA 32.08 NA
(High) Education (Low vs High)
0.67 c (0.60, 0.75)
Income (Medium vs High)
0.83 c (0.77, 0.90)
Income (Low vs High)
0.67 c (0.61, 0.73)
Peter115 2007 Germany, Europe IHD HR Education
(Medium vs High) 0.29 c (0.24, 0.35) 0 38.85 53.58 NA NA NA NA
(High) Education (Low vs High)
0.61 c (0.49, 0.76)
Income (Medium vs High)
1.81 c (1.39, 2.36)
Income (Low vs High)
2.98 c (2.17, 4.10)
MI HR Education (Medium vs High)
0.25 c (0.18, 0.35)
Education (Low vs High)
0.64 c (0.45, 0.91)
Income (Medium vs High)
2.39 c (1.55, 3.67)
Income (Low vs High)
4.06 c (2.36, 6.97)
Lammintausta31 2012 Finland, Europe MI RR Income
(Medium vs High) 1.90 c (1.69, 2.14) 2 56.74 44.94 NA NA NA NA
(High) Income (Low vs High)
2.82 c (2.56, 3.10)
61
Table 2.1. Characteristics of included studied (continued)
Author Year
Country, Setting (income level)
Outcome Risk measure
Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2)
(a) Coronary artery diseases
Honjo116 2010 Japan, Asia CHD HR Education
(Medium vs High) 1.03 b, c (0.92, 1.15) 11 54.78 38.8 3.65 23.32 14.36 NA
(High) Education (Low vs High)
0.65 b, c (0.30, 1.40)
Roux117 2001 US (High) CHD RR Education
(Medium vs High) 1.41 c (1.15, 1.73) 1 NA NA NA NA NA NA
Income (Medium vs High)
1.41 c (1.19, 1.68)
Fujino118 2005 Japan, Asia IHD RR Education
(Medium vs High) 0.88 c (0.68, 1.14) 5 66.08 NA NA 21.72 NA NA
(High) Education (Low vs High)
0.85 c (0.68, 1.07)
Andersen78 2005 Denmark, Europe MI HR Income
(Medium vs High) 1.05 (0.84, 1.31) 11 49.5 57.34 NA 36.23 NA NA
(High) Income (Low vs High)
1.17 (0.85, 1.61)
Lynch52 1996 Finland, Europe MI HR Income
(Medium vs High) 1.91 (0.79, 4.63) 23 NA 100 NA NA NA NA
(High) Income (Low vs High)
2.30 (1.21, 4.37)
Lee119 2000 Taiwan,
Asia (High)
CAD OR Education (Low vs High)
1.25 b (0.83, 1.67) 0 NA 47.28 8.55 31.62 28.4 23.84
Weikert120 2008 Germany, Europe MI RR Education
(Medium vs High) 1.18 (0.85, 1.63) 2 54.5 64.5 10.4 22.9 58.5 26.9
(High) Education (Low vs High)
1.22 (0.91, 1.62)
62
Table 2.1. Characteristics of included studied (continued)
Author Year
Country, Setting (income level)
Outcome Risk measure
Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2)
(a) Coronary artery diseases
Hippe121 1999 Demark, Europe MI RR Education
(Medium vs High) 1.36 b, c (1.25, 1.49) 2 66 53.49 NA NA NA NA
(High) Education (Low vs High)
1.71 b, c (1.41, 2.07)
Huisman122 2008 Netherlands,
Europe (High)
MI RR Education (Low vs High)
1.72 (1.06, 2.80) 10 42.33 67.5 NA NA NA NA
Eaker123 1992 US (High) MI HR Education
(Medium vs High) 1.60 (0.70, 3.70) 6 54 0 NA NA NA NA
Education (Low vs High)
2.5 (1.00, 6.10)
Bosma124 1995 Lithuania, Europe MI RR Education
(Medium vs High) 1.40 (0.80, 2.46) 2 51.6 100 NA 72.64 NA 27.19
(High) Education (Low vs High)
1.42 (0.83, 2.45)
Netherlands, Europe Education
(Medium vs High) 0.78 (0.46, 1.31) 2 52.4 100 NA 92.23 NA 25.5
(High) Education (Low vs High)
0.68 (0.38, 1.23)
Chaix125 2007 Sweden, Europe IHD HR Education
(Low vs High) 1.38 (1.24, 1.53) 10 NA NA NA NA NA NA
(High) Income (Medium vs High)
1.30 (1.10, 1.52)
Income (Low vs High)
1.65 (1.38, 1.97)
63
Table 2.1. Characteristics of included studied (continued)
Author Year
Country, Setting (income level)
Outcome Risk measure
Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2)
(a) Coronary artery diseases
Kuper126 2006 Sweden, Europe MI HR Education
(Medium vs High) 1.70 (1.10, 2.50) 7 40.24 0 1.34 59.27 9.27 23.49
(High) Education (Low vs High)
1.90 (1.30, 2.80)
Lapidus & Bengtsson127 1986
Sweden, Europe (High)
MI RR Education (Low vs High)
1.50 (0.60, 3.50) 1 NA 0 NA NA NA NA
(b) Cardiovascular events
Braig128 2011 Germany, Europe
CVE (MI with OR Education
(Medium vs High) 1.22 (0.94, 1.59) 5 50 59.61 1.69 24.17 NA 26.13
(High) Stroke) Education (Low vs High)
1.83 (1.52, 2.21)
Jakobsen129 2012 Demark, Europe CVE HR Education
(Medium vs High) 0.94 (0.77, 1.15) 25 NA 75 6.5 48.72 NA NA
(High) Education (Low vs High)
0.87 (0.71, 1.07)
Income (Medium vs High)
1.15 (0.97, 1.36) 25 NA 73.19 9.5 46.34 NA NA
Income (Low vs High)
1.06 (0.87, 1.29)
Panagiotakos25 2016 Greek, Europe CVE HR Education
(Medium vs High) 0.78 (0.41, 1.51) 9 45.45 49.75 8.96 54.6 31.78 26.32
(High) Education (Low vs High)
1.31 (0.63, 2.74)
64
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (b) Cardiovascular events
Rasmussen130 2007 Demark, Europe RV HR Education
(Medium vs High) 1.04 b (1.02, 1.05) 2 60.8 71.07 4.26 NA NA NA
(High) (CVE) Education (Low vs High)
1.01 b (0.95, 1.06)
Income (Medium vs High)
0.96 b (0.95, 0.97) 2 60.66 71.05 9.5 NA NA NA
Income (Low vs High)
0.89 b (0.85, 0.94)
Senan & Petrosyan131 2014 India, Asia
(Lower- CVE RR Education (Low vs High)
1.22 b, c (1.19, 1.25) 4 NA 75.14 NA NA NA NA
middle) Income (Low vs High)
1.15 b, c (1.14, 1.16)
Bosma132 2005 Netherlands, Europe CVE HR Education
(Medium vs High) 1.15 (0.84, 1.57) 10 69.69 41.69 5.43 NA 20.37 NA
(High) Education (Low vs High)
1.24 (0.92, 1.68)
Income (Medium vs High)
1.30 (0.94, 1.79)
Income (Low vs High)
1.21 (0.89, 1.64)
Masoudkabir133 2012 Iran,
Asia (Upper- CVE
(IHD with HR Education (Medium vs High)
1.14 b (0.88, 1.47) 5 58.81 45.4 21.8 25.7 59.8 27.28
middle) stroke) Education (Low vs High)
0.99 b (0.52, 1.89)
Income (Medium vs High)
1.05 b (0.98, 1.12)
Income (Low vs High)
1.18 b (0.81, 1.70)
65
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (b) Cardiovascular events
Minh134 2006 Vietnam, Asia (Lower- CVE RR Education
(Low vs High) 4.50 (3.40, 5.80) 2 41.6 53.61 NA NA Na NA
middle) Income (Low vs High)
1.25 b (0.83, 1.67)
Hirokawa135 2006 Japan, Asia (High) CVE HR Education
(Medium vs High) 1.67 c (0.90, 3.09) 13 54.98 61.19 NA 24.99 35.1 23.02
Education (Low vs High)
2.01 c (1.04, 3.91)
Siegel136 1987 US (High) CVE HR Education (Medium vs High)
0.59 (0.26, 1.33) 11 72.85 36.48 NA 11.3 NA NA
He137 2001 US (High) HF RR Education (Medium vs High)
1.22 c (1.05, 1.41) 14 49.77 40.64 3.82 35 28.2 25.6
Christensen138 2011 Demark, Europe HF HR Education
(Low vs High) 1.27 b (1.19, 1.36) 12 52.4 45.25 2.87 63.43 6.25 25.15
(High) Income (Medium vs High)
1.13 b, c (1.08, 1.17) 0 52.4 45.25 2.87 63.43 6.25 25.15
Income (Low vs High)
1.51 b, c (1.28, 1.78)
Borne139 2011 Sweden, Europe HF HR Income
(Medium vs High) 0.97 b ,c (0.96, 0.98) 4 60.8 44.4 NA NA NA NA
(High) Income (Low vs High)
1.67 b, c (1.61, 1.73)
Philbin140 2001 US (High) HF OR Income (Medium vs High)
1.08 (1.01, 1.16) 6 74 43 33 NA 45 NA
Income (Low vs High)
1.18 (1.10, 1.26)
Schwarz & Elman141 2003 US (High) HF HR Education
(Medium vs High) 0.51 (0.26, 1.02) 0 78 50 42 NA 33 NA
66
67
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (b) Cardiovascular events
Sui142 2008 US (High) HF HR Education (Medium vs High)
1.63 (0.94, 2.81) 0 63.7 78 27.52 NA 46.48 NA
CVE HR Education (Medium vs High)
1.55 (1.05, 2.30)
Rosvall143 2006 Sweden, Europe CVE HR Education
(Medium vs High) 1.59 (0.89, 2.80) 2 59.18 44.41 8.14 30.28 18.16 25.4
(High) Education (Low vs High)
2.19 (1.29, 3.73)
Engstrom144 2000 Sweden, Europe CVE HR Education
(Medium vs High) 2.77 b (1.46, 5.27) 6 51 0 13 72 31 24.6
(High) Education (Low vs High)
2.86 b (0.91, 9.09)
Notara28 2016 Greek, Europe CVE HR Education
(Medium vs High) 1.61 (1.23, 2.08) 9 66.11 75.97 31.54 NA 53.64 NA
(High) Education (Low vs High)
1.25 (0.88, 1.78)
(c) Strokes
Weikert120 2008 Germany, Europe Stroke RR Education
(Medium vs High) 1.66 (1.13, 2.45) 2 55.9 64.5 13.8 12.4 63.8 26.8
(High) Education (Low vs High)
1.63 (1.14, 2.33)
Lapidus & Bengtsson127 1986
Sweden, Europe (High)
Stroke RR Education (Low vs High)
1.30 (0.40, 4.10) 1 NA 0 NA NA NA NA
68
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (c) Strokes Avendano & Glymour145 2008 US (High) Stroke HR Education
(Medium vs High) 1.07 b, c (0.97, 1.18) 17 67.7 43 21 22 21 27
Education (Low vs High)
0.96 b, c (0.86, 1.08)
Income (Medium vs High)
1.00 b, c (0.99, 1.01)
Income (Low vs High)
1.08 b, c (1.01, 1.16)
Rawshani112 2015 Sweden, Europe Stroke HR Education
(Medium vs High) 1.42 b, c (1.31, 1.55) 14 39.27 53.82 100 11.81 NA 25.61
(High) Education (Low vs High)
1.82 b, c (1.33, 2.50)
Income (Medium vs High)
1.29 c (0.88, 1.69) 14 39.19 53.78 100 12.25 NA 25.60
Income (Low vs High)
2.09 c (1.62, 2.69)
Li146 2008 Sweden, Europe Stroke RR Income
(Medium vs High) 1.41 c (1.21, 1.63) 4 62.7 48.88 NA NA NA NA
(High) Income (Low vs High)
1.45 c (1.24, 1.70)
Rossum147 1999 Netherlands, Europe Stroke RR Education
(Medium vs High) 0.90 b (0.83, 25.0) 12 71 0 4.5 18.7 35.8 26.8
(High) Education (Low vs High)
4.79 b (1.48, 15.5)
Income (Medium vs High)
1.42 b (1.04, 1.96)
Income (Low vs High)
1.75 b (0.81, 3.85)
69
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (c) Strokes Gillum & Mussolino148 2003 US (High) Stroke RR Education
(Medium vs High) 0.85 b, c (0.84, 0.86) 9 62 47.3 7.48 29.81 19.47 NA
Education (Low vs High)
1.16 b, c (0.90, 1.49)
Kuper149 2007 Sweden, Europe Stroke HR Education
(Medium vs High) 1.20 (0.90, 1.80) 7 40.27 0 1.3 59.42 9.24 23.49
(High) Education (Low vs High)
1.5 (1.00, 2.20)
Jackson150 2014 Australia, Asia Stroke OR Education
(Medium vs High) 1.57 (0.95, 2.61) 11 49.5 0 3.68 16.91 24.53 25.94
(High) Education (Low vs High)
1.52 (0.99, 2.33)
Honjo116 2010 Japan, Asia (High) Stroke HR Education
(Medium vs High) 1.38 b, c (1.24, 1.54) 11 54.78 38.8 3.65 23.32 14.36 NA
Education (Low vs High)
1.04 b, c (0.65, 1.67)
Fujino118 2005 Japan, Asia (High) Stroke RR Education
(Medium vs High) 1.14 c (0.98, 1.33) 5 66.08 NA NA 21.72 NA NA
Education (Low vs High)
1.22 c (1.01, 1.47)
Lee119 2000 Taiwan, Asia (High) Stroke OR Education
(Low vs High) 1.67 b (0.91, 2.50) 0 NA 47.28 8.55 31.62 28.4 23.84
Honjo111 2008 Japan, Asia (High) Stroke HR Education
(Medium vs High) 0.74 b (0.51, 1.09) 11 NA 0 2.36 7.28 NA NA
Education (Low vs High)
1.10 b (0.96, 1.28)
70
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (c) Strokes
Arrich108 2005 Austria, Europe Stroke RR Education
(Medium vs High) 1.90 a (1.56, 2.31) 0 66.71 54.28 25.27 26.46 66.98 NA
(High) Education (Low vs High)
2.07 a (1.68, 2.55)
Income (Medium vs High)
1.29 a (1.16, 1.42) 0 65.5 54.35 27.03 27.6 44.95 NA
Income (Low vs High)
1.25 a (0.89, 1.76)
Andersen151 2014 Denmark, Europe Stroke RR Education
(Medium vs High) 1.17 b (1.15, 1.18) 4 71.9 52.5 12.48 27.43 46.9 NA
(High) Education (Low vs High)
1.17 b (1.15, 1.22)
Income (Medium vs High)
1.51 b (1.49, 1.52)
Income (Low vs High)
1.35 b (1.34, 1.37)
Avendano55 2006 US (High) Stroke HR Education (Medium vs High)
0.89 (0.54, 1.47) 12 NA NA NA NA NA NA
Education (Low vs High)
0.73 (0.51, 1.04)
Income (Medium vs High)
0.74 (0.44, 1.26)
Income (Low vs High)
0.70 (0.47, 1.06)
(d) Cardiovascular deaths
Lynch52 1996 Finland, Europe
Death due to CVE HR Income
(Medium vs High) 0.34 (0.13, 0.93) 23 NA 100 NA NA NA NA
(High) Income (Low vs High)
0.72 (0.39, 1.34)
71
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (d) Cardiovascular deaths
Jakovljevic152 2001 Finland, Europe
Death due to stroke HR Education
(Low vs High) 1.19 (1.05, 1.33) 4 NA 71.31 NA NA NA NA
(High) Income (Medium vs High)
1.36 (1.13, 1.64)
Income (Low vs High)
1.72 (1.45, 2.05)
Zhou153 2006 China, Asia (High)
Death due to stroke HR Education
(Medium vs High) 1.02 (0.57, 1.83) 9 77.2 54.8 26.5 27.7 NA NA
Education (Low vs High)
0.88 (0.46, 1.68)
Income (Low vs High)
3.37 (2.34, 4.87)
Beebe-Dimmer22 2004 US (High) Death due
to IHD or HR Education (Medium vs High)
0.99 (0.80, 1.21) 7 44 0 NA 40.3 NA NA
stroke Education (Low vs High)
1 (0.79, 1.28)
Income (Low vs High)
1.45 (1.20, 1.74)
Jakobsen129 2012 Demark, Europe
Death due to CVE HR Education
(Medium vs High) 0.82 (0.57, 1.16) 25 NA 75 6.5 48.72 NA NA
(High) Education (Low vs High)
0.74 (0.52, 1.04)
Income (Medium vs High)
1.05 (0.77, 1.44) 25 NA 73.19 9.5 47.25 NA NA
Income (Low vs High)
1.22 (0.88, 1.69)
Kim154 2005 US (High) Death due to CVE OR Education
(Low vs High) 1.41 (1.28, 1.56) 3 45.03 0 NA NA NA NA
72
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (d) Cardiovascular deaths
Geyer110 2006 Sweden, Europe
Death due to MI RR Education
(Medium vs High) 1.22 (1.14, 1.30) 1 47.8 49.3 NA NA NA NA
(High) Education (Low vs High)
1.41 (1.31, 1.50)
Income (Medium vs High)
1.38 (1.27, 1.51)
Income (Low vs High)
2.20 (2.09, 2.31)
Qureshi155 2003 US (High) Death due to stroke RR Education
(Medium vs High) 1.37 c (1.09, 1.71) 9 50.73 42.4 3.98 26.45 NA 25.82
Death due to MI RR Education
(Medium vs High) 1.36 c (1.18, 1.57)
Pednekar156 2011 India, Asia (Lower-
Death due to CVE HR Education
(Low vs High) 1.15 b, c (1.14, 1.17) 5 51.72 59.83 NA 9.93 NA NA
middle) Death due to IHD HR Education
(Low vs High) 1.05 b, c (1.04, 1.07)
Death due to stroke HR Education
(Low vs High) 2.31 b, c (1.98, 2.68)
Rawshani112 2015 Sweden, Europe
Death due to CVE HR Education
(Medium vs High) 1.54 c (1.41, 1.68) 11 39.27 53.82 100 11.81 NA 25.61
(High) Education (Low vs High)
1.27 c (0.94, 1.70)
Income (Medium vs High)
1.92 c (1.31, 2.81) 14 39.19 53.78 NA 12.25 NA 25.6
Income (Low vs High)
3.40 c (2.64, 4.37)
73
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (d) Cardiovascular deaths
Coady157 2014 US (High) Death due to MI HR Education
(Medium vs High) 1.04 c (0.96, 1.13) 9 78 49.64 28.31 NA 54.95 NA
Education (Low vs High)
1.08 c (0.99, 1.17)
Gallo158 2012 Europe (High)
Death due to CVE HR Education
(Medium vs High) 1.21 b, c (1.20, 1.23) 7 51.97 35.9 NA 27.29 NA NA
Education (Low vs High)
1.63 b, c (1.46, 1.82)
Death due to IHD HR Education
(Medium vs High) 1.24 b, c (1.22, 1.26)
Education (Low vs High)
1.87 b, c (1.60, 2.19)
Death due to stroke HR Education
(Medium vs High) 1.16 b, c (1.12, 1.19)
Education (Low vs High)
1.43 b, c (1.12, 1.83)
Rasmussen43 2006 Demark, Europe
Death due to MI RR Education
(Medium vs High) 1.10 c (1.02, 1.19) 6 61.02 70.87 3.6 NA NA NA
(High) Education (Low vs High)
1.15 c (1.06, 1.25)
Income (Medium vs High)
1.14 c (1.08, 1.20) 6 61 60.2 3.58 NA NA NA
Income (Low vs High)
1.42 c (1.35, 1.50)
Salomaa30 2000 Finland, Europe
Death due to MI RR Education
(Low vs High) 1.96 c (1.84, 2.08) 2 NA NA NA NA NA NA
(High) Income (Low vs High)
2.61 c (2.47, 2.76)
74
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (d) Cardiovascular deaths Bucher & Ragland159 1995 US (High) Death due
to CHD RR Education (Medium vs High)
1.54 (1.13, 2.09) 5 46.2 100 NA NA NA NA
Income (Low vs High)
2.07 (0.94, 4.57)
Death due to stroke RR Education
(Medium vs High) 1.27 (0.97, 1.66)
Income (Low vs High)
1.08 (0.56, 2.08)
Tonne160 2005 US (High) Death due to MI RR Education
(Medium vs High) 1.21 (1.05, 1.39) 13 69 58.1 31 NA 63.5 NA
Education (Low vs High)
1.32 (1.15, 1.52)
Income (Medium vs High)
1.25 (1.04, 1.52)
Income (Low vs High)
1.38 (1.14, 1.67)
Chen39 2015 China, Asia (High)
Death due to stroke HR Education
(Low vs High) 1.88 (1.05, 3.36) 14 73.4 46.1 10.2 22.6 72.5 NA
Income (Low vs High)
1.64 (0.97, 2.78)
Chaix125 2007 Sweden, Europe
Death due to IHD HR Education
(Low vs High) 1.46 (1.24, 1.73) 10 NA NA NA NA NA NA
(High) Income (Medium vs High)
1.85 (1.43, 2.43)
Income (Low vs High)
2.83 (2.16, 3.82)
75
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (d) Cardiovascular deaths
Ito161 2008 Japan, Asia (High)
Death due to CVE HR Education
(Medium vs High) 1.15 c (0.77, 1.70) 14 NA 48.28 NA 28.75 NA NA
Education (Low vs High)
1.33 c (0.90, 1.97)
Lee119 2000 Taiwan, Asia (High)
Death due to CVE OR Education
(Low vs High) 1.25 b (0.83, 2.00) 0 NA 47.28 8.55 31.62 28.4 23.84
Minh162 2003 Vietnam,
Asia (Lower-middle)
Death due to CVE RR Education
(Low vs High) 1.00 (0.32, 3.13) 4 NA 75.2 NA NA NA NA
Liu163 1982 Chicago, CHA, US
Death due to CHD RR Education
(Medium vs High) 2.12 (1.15, 3.89) 1 48.9 100 NA 40.2 NA NA
(High) Education (Low vs High)
3.6 (1.99, 6.60)
Death due to CVE RR Education
(Medium vs High) 2.49 (1.40, 4.44)
Education (Low vs High)
4.08 (2.31, 7.21)
Chicago, WEPG, US
Death due to CHD RR Education
(Medium vs High) 1.00 (0.63, 1.59) 1 48.56 100 NA 70.2 NA NA
(High) Education (Low vs High)
1.62 (1.08, 2.44)
Death due to CVE RR Education
(Medium vs High) 0.97 (0.62, 1.42)
Education (Low vs High)
1.52 (1.09, 2.11)
Kilander67 2001 Sweden, Europe
Death due to CVE and HR Education
(Medium vs High) 0.78 (0.48, 1.24) 17 NA 100 NA 50.59 NA 25.03
(High) stroke Education (Low vs High)
1.01 (0.67, 1.52)
76
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (d) Cardiovascular deaths Schwarz & Elman141 2003 US (High) Death due
to HF HR Education (Medium vs High)
0.58 (0.10, 3.43) 0 78.67 61.9 NA NA NA NA
Sui142 2008 US (High) Death due to HF HR Education
(Medium vs High) 1.58 (0.66, 3.78) 0 63.7 78 27.52 NA 46.48 NA
Death due to CVE HR Education
(Medium vs High) 1.60 (0.90, 2.84)
Bosma124 1995 Lithuania, Europe
Death due to CHD RR Education
(Medium vs High) 1.06 (0.60, 1.90) 2 51.6 100 NA 72.64 NA 27.19
(High) Education (Low vs High)
1.08 (0.62, 1.88)
Death due to CVE RR Education
(Medium vs High) 1.12 (0.67, 1.86)
Education (Low vs High)
1.16 (0.72, 1.88)
Netherlands, Europe
Death due to CHD RR Education
(Medium vs High) 1.78 (0.85, 3.70) 2 52.4 100 NA 92.23 NA 25.5
(High) Education (Low vs High)
1.06 (0.46, 2.43)
Death due to CVE RR Education
(Medium vs High) 1.56 (0.88, 2.77)
Education (Low vs High)
1.40 (0.76, 2.58)
Lapidus & Bengtsson127 1986
Sweden, Europe (High)
Death due to CVE RR Education
(Low vs High) 1.2 (0.7, 2.0) 1 NA 0 NA NA NA NA
Notara28 2016 Greece, Europe
Death due to ACS HR Education
(Medium vs High) 1.72 (1.35, 2.22) 9 66.11 75.97 31.54 NA 53.64 NA
(High) Education (Low vs High)
1.33 (0.93, 1.92)
77
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (d) Cardiovascular deaths
Rawshani164 2016 Sweden, Europe
Death due to CVE HR Education
(Medium vs High) 1.08 b (1.04, 1.10) 16 58.25 60.38 100 17.16 NA 30.33
(High) Education (Low vs High)
1.19 b (1.10, 1.28)
Income (Medium vs High)
1.54 (1.42, 1.68)
Income (Low vs High)
1.87 (1.76, 1.99)
Lammintausta31 2012 Finland, Europe
Death due to MI RR Income
(Medium vs High) 2.68 c (2.12, 3.24) 2 56.74 44.94 NA NA NA NA
(High) Income (Low vs High)
4.78 c (4.13, 5.54)
Li146 2008 Sweden, Europe
Death due to stroke RR Income
(Medium vs High) 1.32 c (0.90, 1.93) 4 62.7 48.88 NA NA NA NA
(High) Income (Low vs High)
1.90 c (1.32, 2.72)
Rosvall165 2008 Sweden, Europe
Death due to MI HR Income
(Medium vs High) 1.09 b, c (1.05, 1.14) 1 70.39 65.43 7.38 NA 8.04 NA
(High) Income (Low vs High)
1.26 b, c (1.22, 1.30)
Khang166 2007 South Korea, Asia
Death due to CVE RR Income
(Low vs High) 1.35 c (1.25, 1.45) 5 43.14 100 NA 57.19 NA NA
(High) Death due to IHD RR Income
(Low vs High) 1.20 c (1.05, 1.36)
78
Table 2.1. Characteristics of included studied (continued)
Author Year Country, Setting
(income level) Outcome Risk
measure Study factors (Categories)
Relative Risks (95%CI)
NC Mean Age
(years)
Male (%)
DM (%)
Smoking (%)
HT (%)
Mean BMI
(kg/m2) (d) Cardiovascular deaths
Rosvall167 2008 Sweden, Europe (High)
Death due to MI OR Income
(Low vs High) 1.23 c (1.18, 1.28) 2 63.8 70.54 NA NA NA NA
Arrich108 2005 Austria, Europe
Death due to stroke HR Education
(Medium vs High) 0.85 (0.60, 1.19) 9 66.71 54.28 25.27 26.46 66.98 NA
(High) Education (Low vs High)
0.86 (0.56, 1.32)
Income (Medium vs High)
1.64 (1.23, 2.17) 0 65.5 54.35 27.03 27.6 44.95 NA
Income (Low vs High)
1.07 (0.26, 4.39)
RR, Relative risk, OR, Odds ratio; HR, Hazard ratio; NC; Number of controlled variable; DM, Diabetes Mellitus; HT, Hypertension; BMI, Body Mass Index; IHD, Ischemic Heart Disease; MI, Myocardial infarction; CHD, Coronary heart disease; CAD, Coronary artery disease; CVE, Cardiovascular events; RV, Revascularization; HF, Heart failure; ACS, Acute coronary syndrome; NA, US, United States, CHA, Chicago Heart Association Detection Project; WEPG, Chicago Western Electric Company Study and Peoples Gas Company Studies not available (or) not reported a, RR (95%CI) was recalculated based on raw/frequency data reported in original article; b, RR (95%CI) was recalculated by reversing original RR if the middle or lowest category of education or income was used as a reference group; c, RR (95%CI) was recalculated by pooling separate subgroup RRs (weighted by inverse of their variance) to obtain a single estimate from each study
79
Table 2.2. Risk of bias assessment of included studies.
Authors Year
Selection Comparability Outcome
Total stars Representativeness
of cohort
Selection of non-exposed cohort
Ascertainment of exposure
Outcome of interest
Comparability of cohorts
Assessment of outcome
Adequate duration of follow up
Adequate follow up of cohort
Andersen et al.29 2003 a(*) a(*) a(*) a(*) a(*), b(*) b(*) a(*) c 8
Andersen et al.78 2005 a(*) a(*) a(*) a(*) a(*), b(*) b(*) a(*) b(*) 9
Andersen et al.151 2014 a(*) a(*) a(*) a(*) a(*) b(*) a(*) b(*) 8
Arrich et al.108 2005 b(*) a(*) b(*) b a(*), b(*) b(*) b c 6
Avendano et al.55 2006 a(*) a(*) b(*) a(*) a(*), b(*) b(*) a(*) b(*) 9
Avendano & Glymour145 2008 a(*) a(*) b(*) a(*) a(*) b(*) a(*) b(*) 8
Beebe-Dimmer et al.22 2004 c (women) a(*) c b a(*), b(*) b(*) a(*) b(*) 6
Borné et al.139 2011 a(*) a(*) c a(*) a(*) b(*) a(*) b(*) 7
Bosma et al.124 1995 c (men) a(*) a(*) b a(*) b(*) a(*) b(*) 6
Bosma et al.132 2005 a(*) a(*) c a(*) a(*), b(*) b(*) a(*) d 7
Braig et al.128 2011 a(*) a(*) c a(*) a(*), b(*) c b b(*) 6
Bucher & Ragland159 1995 c (men) a(*) a(*) a(*) a(*) b(*) a(*) d 6
Chaix et al.125 2007 a(*) a(*) a(*) a(*) a(*) b(*) a(*) c 7
Chen et al.39 2015 a(*) a(*) b(*) b a(*), b(*) c a(*) b(*) 7
80
Table 2.2. Risk of bias assessment of included studies (continued)
Authors Year
Selection Comparability Outcome
Total stars Representativeness
of cohort
Selection of non-exposed cohort
Ascertainment of exposure
Outcome of interest
Comparability of cohorts
Assessment of outcome
Adequate duration of follow up
Adequate follow up of cohort
Christensen et al.138 2011 a(*) a(*) b(*) a(*) a(*), b(*) b(*) a(*) b(*) 9
Coady et al.157 2014 a(*) a(*) a(*) a(*) a(*), b(*) b(*) a(*) b(*) 9
Eaker et al.123 1992 c (women) a(*) b(*) b a(*), b(*) a(*) a(*) b(*) 7
Engström et al.144 2000 c (women) a(*) a(*) b a(*), b(*) b(*) a(*) d 6
Fujino et al.118 2005 a(*) a(*) b(*) a(*) a(*) b(*) a(*) c 7
Gallo et al.158 2012 a(*) a(*) b(*) b a(*), b(*) b(*) a(*) b(*) 8
Geyer et al.110 2006 a(*) a(*) a(*) b a(*) b(*) a(*) d 6
Gillum & Mussolino148 2003 a(*) a(*) b(*) a(*) a(*), b(*) b(*) a(*) b(*) 9
He et al.137 2001 a(*) a(*) b(*) a(*) a(*), b(*) b(*) a(*) b(*) 9
Hetemaa et al.114 2006 a(*) a(*) a(*) a(*) a(*) b(*) b d 6
Hippe et al.121 1999 a(*) a(*) a(*) a(*) a(*) b(*) a(*) b(*) 8
Hirokawa et al.135 2006 a(*) a(*) c a(*) a(*), b(*) b(*) a(*) c 7
Honjo et al.111 2008 c (women) a(*) c a(*) a(*), b(*) b(*) a(*) b(*) 7
Honjo et al.116 2010 a(*) a(*) c a(*) a(*), b(*) a(*) a(*) c 7
Huisman et al.122 2008 a(*) a(*) c a(*) a(*) b(*) a(*) c 6
81
Table 2.2. Risk of bias assessment of included studies (continued)
Authors Year
Selection Comparability Outcome
Total stars Representativeness
of cohort
Selection of non-exposed cohort
Ascertainment of exposure
Outcome of interest
Comparability of cohorts
Assessment of outcome
Adequate duration of follow up
Adequate follow up of cohort
Ito et al.161 2008 a(*) a(*) c a(*) a(*), b(*) b(*) a(*) c 7
Jackson et al.150 2014 c (women) a(*) b(*) a(*) a(*), b(*) b(*) b c 6
Jakobsen et al.129 2012 a(*) a(*) a(*) b a(*), b(*) b(*) a(*) b(*) 8
Jakovljević et al.152 2001 a(*) a(*) a(*) a(*) a(*) b(*) a(*) a(*) 8
Khang et al.166 2007 c (men) a(*) a(*) a(*) a(*), b(*) b(*) a(*) b(*) 8
Kilander et al.67 2001 c (men) a(*) b(*) b a(*), b(*) b(*) a(*) b(*) 7
Kim et al.154 2005 c (women) a(*) b(*) b a(*) b(*) a(*) b(*) 6
Kuper et al.126 2006 c (women) a(*) c b a(*), b(*) b(*) a(*) b(*) 6
Kuper et al.149 2007 c (women) a(*) c a(*) a(*), b(*) b(*) a(*) c 6
Lammintausta et al.31 2012 a(*) a(*) a(*) a(*) - a(*) a(*) d 6
Lapidus & Bengtsson127 1986 c (women) a(*) b(*) a(*) a(*) a(*) a(*) b(*) 7
Lee et al.119 2000 a(*) a(*) b(*) b - a(*) a(*) b(*) 6
Li et al.146 2008 a(*) a(*) b(*) a(*) a(*) b(*) a(*) b(*) 8
Liu et al.163 1982 c (men) a(*) b(*) b a(*) b(*) a(*) b(*) 6
82
Table 2.2. Risk of bias assessment of included studies (continued)
Authors Year
Selection Comparability Outcome
Total stars Representativeness
of cohort
Selection of non-exposed cohort
Ascertainment of exposure
Outcome of interest
Comparability of cohorts
Assessment of outcome
Adequate duration of follow up
Adequate follow up of cohort
Lynch et al.52 1996 c (men) a(*) b(*) a(*) a(*), b(*) b(*) a(*) c 7
Masoudkabir et al.133 2012 a(*) a(*) b(*) a(*) a(*), b(*) a(*) a(*) b(*) 9
Minh et al.162 2003 a(*) a(*) b(*) b a(*) a(*) b d 5
Minh et al.134 2006 a(*) a(*) b(*) b a(*) a(*) a(*) d 6
Notara et al.28 2016 a(*) a(*) b(*) a(*) a(*), b(*) a(*) a(*) b(*) 9
Panagiotakos et al.25 2016 a(*) a(*) b(*) a(*) a(*), b(*) b(*) a(*) c 8
Pednekar et al.156 2011 a(*) a(*) b(*) a(*) a(*) b(*) a(*) b(*) 8
Peter et al.115 2007 a(*) a(*) a(*) a(*) - b(*) b d 5
Philbin et al.140 2001 a(*) a(*) a(*) b a(*) b(*) b b(*) 6
Qureshi et al.155 2003 a(*) a(*) a(*) a(*) a(*), b(*) b(*) a(*) b(*) 9
Rasmussen et al.43 2006 a(*) a(*) a(*) a(*) a(*) b(*) a(*) b(*) 8
Rasmussen et al.130 2007 a(*) a(*) a(*) a(*) a(*) b(*) a(*) b(*) 8
Rawshani et al.112 2015 c (dm) a(*) a(*) a(*) a(*), b(*) b(*) a(*) b(*) 8
Rawshani et al.164 2016 c (dm) a(*) a(*) d a(*) b(*) a(*) b(*) 6
83
Table 2.2. Risk of bias assessment of included studies (continued)
Authors Year
Selection Comparability Outcome
Total stars Representativeness
of cohort
Selection of non-exposed cohort
Ascertainment of exposure
Outcome of interest
Comparability of cohorts
Assessment of outcome
Adequate duration of follow up
Adequate follow up of cohort
Rehkopf et al.109 2015 b(*) a(*) a(*) b a(*), b(*) b(*) a(*) d 7
Rossum et al.147 1999 c (women) a(*) b(*) a(*) a(*), b(*) b(*) b c 6
Rosvall et al.143 2006 a(*) a(*) c a(*) a(*) b(*) a(*) b(*) 7
Rosvall et al.165 2008 a(*) a(*) a(*) a(*) a(*) b(*) a(*) d 7
Rosvall et al.167 2008 a(*) a(*) a(*) a(*) a(*) b(*) a(*) d 7
Roux et al.117 2001 a(*) a(*) b(*) a(*) a(*) b(*) a(*) c 7
Salomaa et al.30 2000 a(*) a(*) a(*) a(*) a(*) b(*) a(*) b(*) 8
Schwarz & Elman141 2003 a(*) a(*) b(*) b - b(*) b b(*) 5
Senan & Petrosyan131 2014 b(*) a(*) c a(*) a(*) c a(*) b(*) 6
Siegel et al.136 1987 c (elderly) a(*) b(*) a(*) a(*), b(*) a(*) b d 6
Sui et al.142 2008 a(*) a(*) a(*) b - b(*) b b(*) 5
Thurston et al.113 2005 a(*) a(*) b(*) a(*) a(*), b(*) b(*) a(*) b(*) 9
Tonne et al.160 2005 a(*) a(*) a(*) b a(*) a(*) a(*) d 6
Weikert et al.120 2008 a(*) a(*) b(*) a(*) a(*) b(*) a(*) b(*) 8
Zhou et al.153 2006 a(*) a(*) b(*) a(*) a(*), b(*) a(*) b a(*) 8
84
Table 2.3. Estimations of pooled effects of education and income on cardiovascular outcomes (co-variates adjusted studies only)
Coronary Artery Diseases Cardiovascular Events
n RR
(95% CI) Q p-value
I2 (%)
n
RR (95% CI)
Q p-value
I2 (%)
Education Medium vs High 15 1.21 (1.06, 1.40) 0.005 96 12 1.27 (1.09, 1.48) 0.003 83
Low vs High 17 1.36 (1.11, 1.66) 0.002 94 13 1.50 (1.17, 1.92) 0.001 99 Income
Medium vs High 10 1.27 (1.10, 1.47) 0.001 95 7 1.05 (0.98, 1.13) 0.131 99 Low vs High 10 1.49 (1.16, 1.91) 0.002 98 6 1.17 (0.96, 1.44) 0.117 97
Strokes Cardiovascular Deaths
n RR (95% CI) Q p-value
I2 (%)
n
RR (95% CI)
Q p-value
I2 (%)
Education Medium vs High 12 1.17 (1.01, 1.35) 0.034 99 28 1.21 (1.12, 1.30) <0.001 98
Low vs High 13 1.23 (1.06, 1.43) 0.005 83 34 1.39 (1.26, 1.54) <0.001 98 Income
Medium vs High 6 1.24 (1.00, 1.53) 0.049 99 12 1.34 (1.17, 1.54) <0.001 96 Low vs High 6 1.30 (0.99, 1.72) 0.061 98 21 1.76 (1.45, 2.14) <0.001 99
85
Table 2.4. Pooled education and income effects on cardiovascular outcomes by regions
Education Income
n RR (95% CI) Q p-value I2 n RR (95% CI)
Q p-value I2
Cardiovascular deaths
Asia Medium vs High 2 1.12 (0.78, 1.60) 0.540 5 0 NA NA NA Low vs High 8 1.34 (1.04, 1.72) 0.024 99 4 1.69 (1.07, 2.67) 0.024 96
Europe Medium vs High 15 1.17 (1.06, 1.29) 0.001 99 12 1.40 (1.18, 1.67) <0.001 97 Low vs High 19 1.32 (1.17, 1.49) <0.001 91 14 1.89 (1.47, 2.44) <0.001 99
US Medium vs High 14 1.30 (1.14, 1.49) <0.001 72 1 NA NA NA Low vs High 8 1.69 (1.28, 2.22) <0.001 95 4 NA NA NA
CAD
Asia Medium vs High 3 1.03 (0.85, 1.25) 0.750 28 0 NA NA NA Low vs High 4 1.03 (0.79, 1.33) 0.839 45 0 NA NA NA
Europe Medium vs High 11 1.04 (0.72, 1.50) 0.852 99 11 1.39 (1.18, 1.63) <0.001 92 Low vs High 15 1.24 (0.97, 1.60) 0.086 96 12 1.74 (1.31, 2.32) <0.001 98
US Medium vs High 4 1.21 (0.97, 1.51) 0.085 75 3 NA NA NA Low vs High 2 1.51 (0.93, 2.45) 0.099 47 1 NA NA NA
CVE
Asia Medium vs High 2 1.47 (0.82, 2.63) 0.191 61 2 NA NA NA Low vs High 4 1.85 (0.93, 3.70) 0.081 96 2 NA NA NA
Europe Medium vs High 8 1.26 (1.06, 1.49) 0.090 76 5 1.05 (0.95, 0.37) 0.368 99 Low vs High 9 1.36 (1.07, 1.72) 0.011 95 5 1.24 (0.98, 1.58) 0.080 98
US Medium vs High 5 1.07 (0.69, 1.66) 0.758 78 1 NA NA NA Low vs High 0 NA NA NA 1 NA NA NA
Strokes
Asia Medium vs High 4 1.22 (0.91, 1.65) 0.192 87 0 NA NA NA Low vs High 5 1.27 (1.07, 1.50) 0.006 34 0 NA NA NA
Europe Medium vs High 6 1.46 (1.23, 1.72) <0.001 87 5 1.37 (1.24, 1.52) <0.001 70 Low vs High 7 1.61 (1.28, 2.02) <0.001 76 5 1.54 (1.33, 1.79) <0.001 64
US Medium vs High 3 0.98 (0.81, 1.19) 0.848 89 2 0.89 (0.62, 1.27) 0.514 49 Low vs High 3 0.99 (0.83, 1.20) 0.957 53 2 0.91 (0.58, 1.41) 0.661 78
n, Number of studies; RR, Relative risk; CI, Confidence Interval; Q p-value, p value for Q test for heterogeneity, I2, I2 statistics (%); CAD, Coronary Artery Diseases; CVE, Cardiovascular Events; US, United States; NA, Not available or insufficient data;
86
Table 2.5. Pooled education and income effect on coronary artery diseases (subgroup analyses)
Education Income
n RR (95% CI) Q p-value I2 n RR (95% CI)
Q p-value I2
Number of adjusted variables ≤ 5 Medium vs High 10 0.97 (0.65, 1.45) 0.888 97 6 1.57 (1.30, 1.91) <0.001 87 Low vs High 12 1.22 (0.94, 1.57) 0.130 93 6 2.12 (1.52, 2.96) <0.001 98 > 5 Medium vs High 8 1.14 (0.98, 1.32) 0.085 95 8 1.14 (0.99, 1.31) 0.059 92 Low vs High 9 1.28 (1.02, 1.61) 0.035 89 7 1.29 (0.98, 1.68) 0.066 95 Age (years) ≤ 60 Medium vs High 12 1.05 (0.70, 1.57) 0.817 99 9 1.42 (1.32, 1.52) <0.001 95 Low vs High 12 1.28 (0.99, 1.65) 0.058 85 8 1.83 (1.82, 1.84) <0.001 97 > 60 Medium vs High 4 1.00 (0.77, 1.30) 0.999 92 2 0.94 (0.73, 1.20) 0.600 89 Low vs High 4 1.05 (0.69, 1.59) 0.821 95 2 0.72 (0.51, 1.01) 0.060 18 Male percentage ≤ 60 Medium vs High 10 0.94 (0.64, 1.38) 0.759 99 8 1.43 (1.16, 1.76) 0.001 91 Low vs High 12 1.25 (0.99, 1.58) 0.060 85 8 1.82 (1.30, 2.56) <0.001 96 > 60 Medium vs High 6 1.26 (0.85, 1.86) 0.246 97 4 1.16 (0.85, 1.59) 0.356 98 Low vs High 6 1.25 (0.78, 2.01) 0.359 92 3 1.28 (0.70, 2.32) 0.419 99 Diabetes percentage ≤ 8 Medium vs High 5 1.16 (0.95, 1.42) 0.136 83 2 NA NA NA Low vs High 4 1.25 (0.83, 1.88) 0.295 64 1 NA NA NA > 8 Medium vs High 4 1.03 (0.87, 1.22) 0.733 87 2 0.94 (0.73, 1.20) 0.600 89 Low vs High 5 1.11 (0.84, 1.46) 0.465 89 2 0.72 (0.51, 1.01) 0.060 18 BMI (kg/m2) < 25 Medium vs High 1 NA NA NA 1 NA NA NA Low vs High 2 NA NA NA 1 NA NA NA ≥ 25 Medium vs High 5 1.16 (1.10, 1.23) <0.001 0 2 1.31 (1.07, 1.59) 0.007 29 Low vs High 5 1.30 (1.15, 1.47) <0.001 0 2 1.50 (1.00, 2.26) 0.050 91 Smoking percentage < 30 Medium vs High 6 1.07 (0.97, 1.19) 0.166 48 2 1.21 (0.90, 1.62) 0.203 69 Low vs High 6 1.13 (0.94, 1.37) 0.192 58 2 1.26 (0.53, 2.98) 0.600 85 ≥ 30 Medium vs High 4 1.21 (0.89, 1.63) 0.225 61 3 1.17 (1.07, 1.29) 0.001 0 Low vs High 5 1.32 (0.98, 1.77) 0.066 68 3 1.32 (1.14, 1.53) <0.001 51
n, Number of studies; RR, relative risk; CI, Confidence Interval; Q p-value, p value for Q test for heterogeneity, I2, I2 statistics (%);BMI, Body Mass Index; NA, Not available or insufficient data;
87
Table 2.6. Pooled education and income effect on cardiovascular events (subgroup analyses)
Education Income
n RR (95% CI) Q p-value I2 n RR (95% CI)
Q p-value I2
Number of adjusted variables ≤ 5 Medium vs High 7 1.25 (1.03, 1.52) 0.027 71 5 1.05 (0.97, 1.13) 0.249 99 Low vs High 6 1.69 (1.07, 2.68) 0.025 99 5 1.31 (1.01, 1.69) 0.039 98 > 5 Medium vs High 8 1.28 (1.03, 1.60) 0.028 74 3 1.11 (1.00, 1.23) 0.052 9 Low vs High 7 1.22 (0.98, 1.51) 0.074 72 3 1.16 (1.05, 1.28) 0.004 7 Age (years) ≤ 60 Medium vs High 7 1.35 (1.06, 1.70) 0.014 61 2 1.09 (1.01, 1.16) 0.018 68 Low vs High 8 1.93 (1.35, 2.76) <0.001 92 3 1.34 (1.10, 1.64) 0.004 38 > 60 Medium vs High 7 1.17 (0.90, 1.53) 0.248 79 4 1.01 (0.93, 1.09) 0.900 99 Low vs High 3 1.09 (0.90, 1.31) 0.393 65 4 1.21 (0.92, 1.58) 0.167 99 Male percentage ≤ 60 Medium vs High 9 1.21 (0.95, 1.55) 0.128 79 5 1.07 (1.00, 1.14) 0.054 89 Low vs High 8 1.61 (1.10, 2.37) 0.015 95 6 1.35 (1.18, 1.55) <0.001 88 > 60 Medium vs High 6 1.31 (1.17, 1.48) <0.001 81 3 1.08 (0.94, 1.23) 0.282 99 Low vs High 5 1.17 (1.00, 1.38) 0.047 94 2 0.99 (0.83, 1.19) 0.955 65 Diabetes percentage ≤ 8 Medium vs High 6 1.11 (1.02, 1.21) 0.020 51 3 1.05 (0.94, 1.17) 0.394 95 Low vs High 6 1.23 (0.99, 1.53) 0.058 95 3 1.18 (0.86, 1.63) 0.308 92 > 8 Medium vs High 7 1.46 (1.07, 1.99) 0.016 74 3 1.07 (1.02, 1.12) 0.005 0 Low vs High 4 1.35 (0.83, 2.19) 0.231 72 3 1.17 (1.10, 1.24) <0.001 0 BMI (kg/m2) < 25 Medium vs High 2 2.14 (1.26, 3.63) 0.005 29 0 NA NA NA Low vs High 2 2.26 (1.17, 4.37) 0.016 8 0 NA NA NA ≥ 25 Medium vs High 5 1.20 (1.06, 1.35) 0.003 5 2 1.09 (1.01, 1.18) 0.025 73 Low vs High 5 1.50 (1.16, 1.93) 0.002 74 2 1.35 (1.01, 1.81) 0.043 54 Smoking percentage < 30 Medium vs High 4 1.18 (0.98, 1.42) 0.078 8 1 NA NA NA Low vs High 3 1.61 (1.06, 2.47) 0.027 44 1 NA NA NA ≥ 30 Medium vs High 5 1.29 (0.95, 1.75) 0.099 77 1 NA NA NA Low vs High 5 1.39 (0.93, 2.09) 0.109 88 1 NA NA NA
n, Number of studies; RR, relative risk; CI, Confidence Interval; Q p-value, p value for Q test for heterogeneity, I2, I2 statistics (%);BMI, Body Mass Index; NA, Not available or insufficient data;
88
Table 2.7. Pooled education and income effect on strokes (subgroup analyses)
Education Income
n RR (95% CI) Q p-value I2 n RR (95% CI)
Q p-value I2
Number of adjusted variables ≤ 5 Medium vs High 4 1.43 (1.15, 1.77) 0.001 90 3 1.41 (1.27, 1.56) <0.001 76 Low vs High 6 1.48 (1.17, 1.87) 0.001 85 3 1.38 (1.24, 1.52) <0.001 5 > 5 Medium vs High 9 1.13 (0.93, 1.36) 0.221 95 4 1.10 (0.86, 1.40) 0.463 78 Low vs High 9 1.23 (0.99, 1.53) 0.055 80 4 1.32 (0.79, 2.20) 0.292 93 Age (years) ≤ 60 Medium vs High 5 1.41 (1.32, 1.50) <0.001 0 1 NA NA NA Low vs High 5 1.54 (1.30, 1.83) <0.001 0 1 NA NA NA > 60 Medium vs High 6 1.23 (0.93, 1.63) 0.147 99 5 1.31 (1.19, 1.45) <0.001 99 Low vs High 6 1.31 (0.97, 1.75) 0.073 95 5 1.26 (1.19, 1.33) <0.001 83 Male percentage ≤ 60 Medium vs High 10 1.26 (1.05, 1.50) 0.011 99 6 1.32 (1.14, 1.53) <0.001 99 Low vs High 12 1.37 (1.15, 1.63) <0.001 87 6 1.40 (1.16, 1.68) <0.001 94 > 60 Medium vs High 1 NA NA NA 0 NA NA NA Low vs High 1 NA NA NA 0 NA NA NA Diabetes percentage ≤ 8 Medium vs High 6 1.12 (0.83, 1.51) 0.445 93 1 NA NA NA Low vs High 6 1.28 (1.05, 1.57) 0.014 33 1 NA NA NA > 8 Medium vs High 5 1.37 (1.15, 1.63) <0.001 95 4 1.25 (1.06, 1.47) 0.008 99 Low vs High 6 1.48 (1.15, 1.89) 0.002 93 4 1.31 (1.30, 1.31) <0.001 98 BMI (kg/m2) < 25 Medium vs High 1 NA NA NA 0 NA NA NA Low vs High 2 NA NA NA 0 NA NA NA ≥ 25 Medium vs High 5 1.35 (1.12, 1.64) 0.002 76 3 1.19 (0.94, 1.51) 0.139 72 Low vs High 5 1.55 (1.07, 2.23) 0.019 80 3 1.61 (1.01, 2.55) 0.044 90 Smoking percentage < 30 Medium vs High 11 1.26 (1.07, 1.50) 0.007 99 5 1.29 (1.09, 1.54) 0.003 99 Low vs High 11 1.35 (1.13, 1.60) 0.001 89 5 1.41 (1.11, 1.79) 0.006 96 ≥ 30 Medium vs High 1 NA NA NA 0 NA NA NA Low vs High 2 NA NA NA 0 NA NA NA
n, Number of studies; RR, relative risk; CI, Confidence Interval; Q p-value, p value for Q test for heterogeneity, I2, I2 statistics (%);BMI, Body Mass Index; NA, Not available or insufficient data;
89
Table 2.8. Pooled education and income effect on cardiovascular deaths (subgroup analyses)
Education Income
n RR (95% CI) Q p-value I2 n RR (95% CI)
Q p-value I2
Number of adjusted variables ≤ 5 Medium vs High 14 1.29 (1.15, 1.44) <0.001 51 6 1.34 (1.13, 1.60) 0.001 95 Low vs High 18 1.53 (1.31, 1.79) <0.001 99 12 1.81 (1.40, 2.34) <0.001 99 > 5 Medium vs High 17 1.16 (1.07, 1.26) 0.001 98 7 1.37 (1.07, 1.76) 0.013 96 Low vs High 17 1.28 (1.14, 1.44) <0.001 88 10 1.73 (1.30, 2.30) <0.001 97 Age (years) ≤ 60 Medium vs High 18 1.26 (1.16, 1.38) <0.001 98 4 1.43 (1.10, 1.87) 0.008 96 Low vs High 18 1.53 (1.31, 1.78) <0.001 99 9 1.94 (1.40, 2.71) <0.001 99 > 60 Medium vs High 9 1.18 (1.00, 1.39) 0.047 81 5 1.26 (1.08, 1.47) 0.004 93 Low vs High 7 1.21 (1.05, 1.40) 0.009 78 8 1.65 (1.30, 2.09) <0.001 98 Male percentage ≤ 60 Medium vs High 13 1.19 (1.00, 1.29) <0.001 98 6 1.55 (1.25, 1.92) <0.001 88 Low vs High 18 1.35 (1.20, 1.53) <0.001 99 9 2.31 (1.72, 3.10) <0.001 95 > 60 Medium vs High 18 1.25 (1.08, 1.45) 0.002 88 6 1.14 (1.01, 1.28) 0.028 93 Low vs High 15 1.41 (1.17, 1.69) <0.001 91 11 1.35 (1.19, 1.54) <0.001 96 Diabetes percentage ≤ 8 Medium vs High 4 1.15 (0.93, 1.42) 0.205 85 2 1.12 (1.06, 1.18) <0.001 61 Low vs High 2 1.22 (0.78, 1.90) 0.375 89 2 1.33 (1.18, 1.50) <0.001 93 > 8 Medium vs High 9 1.23 (1.05, 1.45) 0.012 94 4 1.47 (1.20, 1.81) <0.001 83 Low vs High 9 1.24 (1.11, 1.40) <0.001 65 6 1.76 (1.32, 2.36) <0.001 88 BMI (kg/m2) < 25 Medium vs High 0 NA NA NA 0 NA NA NA Low vs High 1 NA NA NA 0 NA NA NA ≥ 25 Medium vs High 9 1.27 (1.09, 1.47) 0.002 84 2 1.71 (1.29, 2.27) <0.001 54 Low vs High 7 1.21 (1.07, 1.36) 0.002 2 2 2.49 (1.39, 4.47) 0.002 95 Smoking percentage < 30 Medium vs High 10 1.21 (1.10, 1.33) <0.001 99 3 1.65 (1.39, 1.97) <0.001 34 Low vs High 12 1.39 (1.18, 1.65) <0.001 99 5 2.38 (1.71, 3.33) <0.001 83 ≥ 30 Medium vs High 11 1.18 (0.95, 1.47) 0.129 64 0 NA NA NA Low vs High 12 1.42 (1.06, 1.89) 0.018 81 3 1.32 (1.21, 1.44) <0.001 38
n, Number of studies; RR, relative risk; CI, Confidence Interval; Q p-value, p value for Q test for heterogeneity, I2, I2 statistics (%);BMI, Body Mass Index; NA, Not available or insufficient data;
90
PubMed search (n=354) Scopus search (n=1335)
Duplicates (n=115)
Record screened (n=1585)
Full text articles assessed for eligibility (n = 186)
Records excluded based on titles and abstract review
941 Non-CVD374 Not include study factors36 Non-English16 Narrative review14 Systematic review7 Commentary5 No full-text available3 Letter 1 Protocol1 Guidelines1 Book
Studies included in qualitative synthesis
(n = 72)
Full text articles excluded 41 Not cardiovascular outcomes studies 35 Non-cohort design23 No outcome of interest12 Study factors as co-variate/control factors3 Not sufficient for data extraction
References lists (n=11)
Studies included in meta-analysis (n = 72)
Education (n = 62)
Income (n = 39 )
Coronary Diseases (n = 23)
Cardiovascular Events (n = 18)
Cardiovascular Death (n = 42)
Cerebrovascular Diseases (n = 15)
Coronary Diseases (n = 15)
Cardiovascular Events (n = 9)
Cardiovascular Death (n = 22)
Cerebrovascular Diseases (n = 7)
Figure 2.1. Flow diagram for selection of studies
91
Figure 2.2. Pooling effects of education on cardiovascular outcomes
92
Figure 2.3. Funnel plots of relative risks of cardiovascular outcomes among medium versus high education levels
93
Figure 2.4. Funnel plots of relative risks of cardiovascular outcomes among low versus high education levels
94
Figure 2.5. Contour-enhanced plots of relative risks of cardiovascular outcomes among medium versus high education levels
95
Figure 2.6. Contour-enhanced plot of relative risks of cardiovascular outcomes among low versus high education levels
96
Figure 2.7. Pooling effects of income on cardiovascular outcomes
97
Figure 2.8. Funnel plots of relative risks of cardiovascular outcomes among medium versus high income levels
98
Figure 2.9. Funnel plots of relative risks of cardiovascular outcomes among low versus high income levels
99
Figure 2.10. Contour-enhanced plots of relative risks of cardiovascular outcomes among medium versus high income levels
100
Figure 2.11. Contour-enhanced plots of relative risks of cardiovascular outcomes among low versus high income levels
101
Appendix A
PubMed Search
((((((((((((((((("Cardiovascular Diseases"[Mesh])) OR ("cardiovascular events")) OR
("Myocardial Infarction"[Mesh])) OR ("Heart Failure"[Mesh])) OR ("Ventricular Function,
Left"[Mesh])) OR ("Coronary Restenosis"[Mesh])) OR (restenos*)) OR (re-stenos*)) OR
("Coronary Disease"[Mesh])) OR ("coronary flow")) OR ("coronary blood flow")) OR
("ejection fraction")) OR ("stroke")) OR ("cardiovascular death")) OR ("cardiovascular
mortality"))) AND (((((("Education"[Mesh])) OR ("Educational Status"[Mesh])) OR
("education level"))) OR ("Income"[Mesh]))
Scopus Search
( ( ( TITLE-ABS-KEY ( "cardiovascular disease*" ) ) OR ( TITLE-ABS-KEY (
"cardiovascular event*" ) ) OR ( TITLE-ABS-KEY ( "myocardial infarction" ) ) OR (
TITLE-ABS-KEY ( restenos* ) ) OR ( TITLE-ABS-KEY ( re-stenos* ) ) OR ( TITLE-
ABS-KEY ( "cardiovascular death" ) ) OR ( TITLE-ABS-KEY ( "cardiovascular mortality"
) ) OR ( TITLE-ABS-KEY ( "heart failure" ) ) ) OR ( ( TITLE-ABS-KEY ( "left
ventricular function" ) ) OR ( TITLE-ABS-KEY ( "ejection fraction" ) ) OR ( TITLE-ABS-
KEY ( "coronary flow" ) ) OR ( TITLE-ABS-KEY ( "coronary blood flow" ) ) OR (
TITLE-ABS-KEY ( "stroke" ) ) ) ) AND ( ( TITLE-ABS-KEY ( education ) ) OR ( TITLE-
ABS-KEY ( income ) ) )
102
Appendix B
NEWCASTLE - OTTAWA QUALITY ASSESSMENT SCALE (COHORT STUDIES)
Note: A study can be awarded a maximum of one star for each numbered item within the
Selection and Outcome categories. A maximum of two stars can be given for Comparability
Selection
1) Representativeness of the exposed cohort
a. truly representative of the average in the community*
b. somewhat representative of the average in the community*
c. selected group of users e.g. nurses, volunteers
d. no description of the derivation of the cohort
2) Selection of the non-exposed cohort
a. drawn from the same community as the exposed cohort*
b. drawn from a different source*
c. no description of the derivation of the non-exposed cohort
3) Ascertainment of exposure
a. secure record (e.g. surgical records, medical records, census registration)*
b. structured interview*
c. written self-report
d. no description
4) Demonstration that outcome of interest was not present at start of study
In the case of mortality studies, outcome of interest is still the presence of a
disease/incident, rather than death. That is to say that a statement of no history of
disease or incident earns a star.
a. yes*
b. no
Comparability
1) Comparability of cohorts on the basis of the design or analysis. A maximum of 2
stars can be allotted in this category.
a. study controls for age/sex *
103
b. study controls for any three of the following cardiovascular risk factors:
Diabetes, BMI, Obesity, Physical activity, Hypertension, Smoking, Alcohol
drinking, Dyslipidemia and Chronic Kidney Disease *
Outcome
1. Assessment of outcome
a. independent or blind assessment stated in the paper, or confirmation of the
outcome by reference to secure records (x-rays, medical records, etc.)*
b. record linkage (e.g. identified through ICD codes on database records)*
c. self-report (i.e. no reference to original medical records or x-rays to confirm
the outcome)
d. no description.
2. Was follow-up long enough for outcomes to occur
Minimum required follow-up period is ≥ 5 years.
a. yes*
b. no
If the follow-up period is reported with a mean and a range, and the mean is longer
than the required minimum, rate it as ‘yes.’
3. Adequacy of follow-up of cohorts
a. complete follow-up, all subjects accounted for*
b. subjects lost to follow-up are unlikely to introduce bias – small number lost
<20%
c. follow-up rate <80% and no description of those lost
d. no description or unclear
104
Appendix C: Dummy tables
Table 4.1 Baseline characteristics
Factors MCVE N (%)
Non-MCVE N (%)
p-value
Age, years (mean ±SD)
Sex Male Female
Education (years) Primary (0-6) Secondary (7-12) Tertiary (>12)
Income (Baht) Low (<20,000) Middle (20,000 – 50,000) High (>50,000)
Smoking Non smoker Former smoker Current smoker
Diabetes Mellitus (DM) DM Non-DM
Hypertension (HT) HT Non-HT
Obesity Normal (BMI <25) Overweight (BMI= 25- 29.9) Obesity (BMI >=30)
Total Cholesterol Normal High
Triglyceride Normal High
HDL Normal Low HDL
105
Table 4.2 The incidence rate of MCVE stratified by category of risk factors
Factors Person-year
Cardiovascular events p-value No. of
events Incidence rate/1000
Age, years (mean ±SD)
Sex Male Female
Education (years) Primary (0-6) Secondary (7-12) Tertiary (>12)
Income (Baht) Low (<20,000) Middle (20,000 – 50,000) High (>50,000)
Smoking Non smoker Former smoker Current smoker
Diabetes Mellitus (DM) DM Non-DM
Hypertension (HT) HT Non-HT
Obesity Normal (BMI <25) Overweight (BMI= 25- 29.9) Obesity (BMI >=30)
Total Cholesterol Normal High
Triglyceride Normal High
HDL Normal Low HDL
106
Table 4.3 Univariate analysis of risk factors of MCVE
Factors Univariate analysis
p-value Hazard ratio 95% CI
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Factor 6
Factor 7
Table 4.4 Multivariate analysis of risk factors of MCVE
Factors Multivariate analysis
p-value Hazard ratio 95% CI
Factor 1
Factor 2
Factor 3
Factor 4
107
Table 4.5 Mediation analysis of MCVE (pathway A)
Equation Factors b SE 95%CI p-value
Income Education
Age
Gender
Obesity
Smoking
Diabetes
Hypertension
Dyslipidemia
MCVE Education
Income
Age
Gender
Obesity
Smoking
Diabetes
Hypertension
Dyslipidemia
Table 4.6 Causal association between education and MCVE (pathway A)
Parameter Model Pathway Beta 95% CI
ACEM Income mediator Education → Income → MCVE
(ab)
Direct effect Education → MCVE (c’)
Total effect ab + c’
108
Table 4.7 Mediation analysis of MCVE (pathway B)
Equation Factors b SE 95%CI p-value
Education Income
Age
Gender
Obesity
Smoking
Diabetes
Hypertension
Dyslipidemia
MCVE Education
Income
Age
Gender
Obesity
Smoking
Diabetes
Hypertension
Dyslipidemia
Table 4.8 Causal association between education and MCVE (pathway B)
Parameter Model Pathway Beta 95% CI
ACEM Education
mediator
Income → Education → MCVE
(ab)
Direct effect Income → MCVE (c’)
Total effect ab + c’
