You are here

Article Text

Article menu

Download PDFPDF

Research report
Inequalities in cardiovascular disease mortality: the role of behavioural, physiological and social risk factors
  1. Alison Beauchamp1,
  2. Anna Peeters1,
  3. Rory Wolfe1,
  4. Gavin Turrell2,
  5. Linton R Harriss1,
  6. Graham G Giles1,3,4,
  7. Dallas R English3,4,
  8. John McNeil1,
  9. Dianna Magliano5,
  10. Stephen Harrap6,
  11. Danny Liew7,
  12. David Hunt8,
  13. Andrew Tonkin1
  1. 1Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Australia
  2. 2School of Public Health, Queensland University of Technology, Brisbane, Australia
  3. 3Cancer Epidemiology Centre, Cancer Council Victoria, Melbourne, Australia
  4. 4Centre for Molecular, Environmental, Genetic and Analytic Epidemiology, School of Population Health, University of Melbourne, Melbourne, Australia
  5. 5International Diabetes Institute, Melbourne, Australia
  6. 6Department of Physiology, University of Melbourne, Melbourne, Australia
  7. 7Department of Medicine, University of Melbourne, Melbourne, Australia
  8. 8Department of Cardiology, Royal Melbourne Hospital, Melbourne, Australia
  1. Correspondence to Dr Alison Beauchamp, Department of Epidemiology & Preventive Medicine, Monash University, Alfred Hospital, Melbourne 3004, Australia; alison.beauchamp{at}med.monash.edu.au

Abstract

Background While the relationship between socio-economic disadvantage and cardiovascular disease (CVD) is well established, the role that traditional cardiovascular risk factors play in this association remains unclear. The authors examined the association between education attainment and CVD mortality and the extent to which behavioural, social and physiological factors explained this relationship.

Methods Adults (n=38 355) aged 40–69 years living in Melbourne, Australia were recruited in 1990–1994. Subjects with baseline CVD risk factor data ascertained through questionnaire and physical measurement were followed for an average of 9.4 years with CVD deaths verified by review of medical records and autopsy reports.

Results CVD mortality was higher for those with primary education only, compared with those who had completed tertiary education, with an HR of 1.66 (95% CI 1.10 to 2.49) after adjustment for age, country of birth and gender. Those from the lowest educated group had a more adverse cardiovascular risk factor profile compared with the highest educated group, and adjustment for these risk factors reduced the HR to 1.18 (95% CI 0.78 to 1.77). In analysis of individual risk factors, smoking and waist circumference explained most of the difference in CVD mortality between the highest and lowest education groups.

Conclusions Most of the excess CVD mortality in lower socio-economic groups can be explained by known risk factors, particularly smoking and overweight. While targeting cardiovascular risk factors should not divert efforts from addressing the underlying determinants of health inequalities, it is essential that known risk factors are addressed effectively among lower socio-economic groups.

  • Epidemiology
  • cardiovascular disease
  • primary prevention
  • risk factors
  • socio-economic status
  • epidemiology me
  • heart disease
  • prevention PR
  • social inequalities

https://doi.org/10.1136/jech.2009.094516

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Socio-economically disadvantaged adults are more likely to experience premature death from cardiovascular disease (CVD) and have higher disease prevalence rates than their more advantaged counterparts.1–3 CVD is largely preventable, and the landmark INTERHEART case–control study of acute myocardial infarction in 52 countries showed that the population attributable risk for nine independent risk factors totalled approximately 90%.4 While it is known that these risk factors are socially patterned,3 5 their precise role in generating socio-economic inequalities in CVD is unclear.6 Studies which have examined the role of traditional and non-traditional risk factors as explanations for relative socio-economic differences in CVD have varied in their findings.7–16 Therefore, the extent to which targeting risk factors among disadvantaged groups will reduce social inequalities in CVD is uncertain.16 17

    Analysis of cardiovascular risk factors in the context of their relative position on the pathway between socio-economic status (SES) and CVD is important in understanding the role they play in this association. Behavioural risk factors generally precede physiological abnormalities, and social risk factors are thought to impact on both.18 Identifying the relative contribution of these groups of risk factors may not only help elucidate the mechanisms underlying the SES/CVD gradient but also inform priorities in public health policy. However, few studies have assessed physiological, behavioural or social risk factor groups in the same analysis,13 and no study has been identified that examined the three types of risk factors concurrently in both men and women.

    It is also important to improve our understanding of the current contribution of individual risk factors to the SES/CVD gradient. Although several studies have found that smoking is the most important health behaviour explaining socio-economic differences in CVD mortality,7 9–12 14 16 others have found that physical activity is more important.19 20 Among physiological risk factors, findings are also inconsistent, with the relative importance of blood pressure, body mass index (BMI) and serum cholesterol varying between studies.10 13 19 Trends in population levels of these risk factors are changing,21 22 and so contemporary evidence concerning their impact upon health inequalities is required.

    We aimed to examine whether a socio-economic gradient in CVD mortality exists in a large cohort of both men and women, with accurate ascertainment of both endpoints and risk factors, and to describe and quantify the separate effects of behavioural, physiological and social risk factors on the relationship between SES and CVD.

    Methods

    Subjects

    The Melbourne Collaborative Cohort Study (MCCS) is a prospective study of 41 514 subjects aged 27–80 years, recruited between 1990 and 1994. Twenty-four per cent of subjects were southern European migrants, deliberately oversampled to extend the range of lifestyle factors. Details of the design, recruitment and study procedures have been published elsewhere.23 In brief, subjects were volunteers from metropolitan Melbourne, recruited using electoral rolls, telephone books, multicultural media outlets, community centres and churches. All subjects provided written informed consent, and the study was approved by The Cancer Council Victoria's Human Research Ethics Committee. Baseline examination included face-to-face interviews and questionnaires conducted in the subjects' preferred language (English, Greek or Italian). Subjects outside the age range of 40–69 years were excluded from analyses (n=377), as were those who self-reported a history of CVD (n=2443), and those missing data on education (n=10) or covariates (n=329), leaving 38 355 subjects for final analysis (15 261 men and 23 094 women). A total of 62 participants were known to have left Australia before the end of follow-up of 31 December 2002. Of these, 10 had primary education only, nine had some secondary education, nine had completed secondary education, and 34 had tertiary education.

    Because of deliberate oversampling of southern European born migrants, we categorised participants according to their country of birth as follows: (1) Australia/New Zealand (NZ)/northern Europe (the latter comprising UK and The Netherlands), or (2) southern Europe (comprising Italy, Greece and Malta).

    Assessment of socio-economic status

    Socio-economic status was measured using educational attainment and was ascertained from the question ‘What is the highest level of education you completed?’ Education levels were categorised as (1) primary only (comprising no school, or primary school only), (2) some secondary (comprising some secondary education only), (3) completed secondary (comprising completed secondary education, trade certificate or some study towards a tertiary degree) and (4) completed tertiary (comprising degree, diploma or higher). These categories are considered to represent hierarchical stages of education, each of which has important socio-economic implications.24

    Baseline risk factors

    Behavioural risk factors

    Smoking status was ascertained from questions modified from the Medical Research Council 1986 Respiratory Symptoms Questionnaire,25 26 and categorised as either currently smoking at least seven cigarettes a week for a year, or not currently smoking. Alcohol intake was ascertained by asking participants their usual quantity and frequency of alcohol for the current decade and analysed according to quintiles of g/day, assuming a dose–response relationship. Subjects were classified as drinkers if they responded ‘yes’ to the question: ‘Have you ever drunk at least 12 alcoholic drinks in a year (sips and tastes don't count)?’ Physical activity over the previous 6 months was based on the number of times per week that exercise for recreation or sport was undertaken at a vigorous, less vigorous or walking level only. These data were combined to give an overall score of relative energy expenditure in four categories, based on the Compendium of Physical Activities.27 Dietary information was collected using a self-administered food frequency questionnaire (FFQ) specifically developed for use in the MCCS.28 The FFQ assessed the average intake of 121-items over the previous 12 months with this information used to calculate average daily nutrient intake, including saturated fat intake.29 The number of times per day that fruit and vegetables were eaten was also ascertained.

    Social connection

    Social connection30 was determined by asking the number of people living in the household, and modelled as living alone versus living with others. In addition, the number of hours per week participants spent in social activities outside home or work was analysed according to the following categories: 0, 1–2, 3–4, 5–9 or 10+ h.

    Physiological risk factors

    Self-reported history of diabetes was ascertained from the baseline questionnaire. Standard methods were used to measure height, weight, and waist and hip circumferences.31 Blood pressure was measured three times after supine rest for 5 min, and the average of the second and third readings used in analysis. Blood samples for total plasma cholesterol were collected into 15 ml lithium–heparin vacutainers and measured immediately using a Kodak Ektachem analyser (Rochester, New York). In all, 68% of subjects were fasting when samples were taken.

    Ascertainment of deaths

    CVD-related deaths in the cohort were verified through medical record review and adjudication. In brief, CVD-related deaths occurring between participant baseline attendance date and 31 December 2002 were identified through linkage with the Victorian Registry of Births, Deaths and Marriages and the Australian National Death Index. Participant's medical records and autopsy reports were reviewed and categorised by panels of expert cardiologists and neurologists. Where no medical record or autopsy report was available, death certificates (n=30) were used.

    Fatal CVD events were classified as related to coronary heart disease (CHD), stroke or ‘other’ CVD cause, comprising indeterminate fatal CVD event, non-coronary cardiac death, other vascular death and heart failure.

    Statistical analysis

    Analyses were performed using Stata 9.2 (Stata Corp, College Station, Texas). Baseline risk factors and demographics are reported as means or proportions. The significance of any trend in baseline variables across education categories was assessed using logistic regression (for binary variables) or linear regression (for continuous variables) with education as a continuous variable.

    Survival was described using crude mortalities. Cox proportional hazards models were constructed, and HRs and their 95% CIs calculated to assess the association between education and CVD mortality, with completed tertiary education used as the reference group. For these analyses, time at risk began at recruitment to the cohort and ended at the date of death, emigration from Australia or 31 December 2002, whichever came first. These models were extended to adjust for age, sex and country of birth, with subsequent models adjusting for behavioural, social or physiological risk factor groups and finally for all risk factors simultaneously. Tests of interaction between education and sex were not significant, and analyses are presented with males and females combined. Two-sided p values are presented, with p values <0.05 regarded as significant. The proportional hazards assumption was tested and met for all variables used in analysis.

    The percentage reduction in excess risk of CVD attributable to risk factors was calculated as:9 11 13 32

    100×(HRadjusted for age, sex, country of birth−HRadjusted for age, sex, country of birth+risk factors)(HRadjusted for age, sex, country of birth−1)

    This measure provides an estimate of the explanatory contribution of risk factors to educational inequalities in CVD mortality.7

    Results

    Baseline characteristics

    Relevant baseline characteristics are presented in table 1. Compared with more highly educated groups, those subjects with primary education only were more likely to be female, older and born in a southern European country. Inverse gradients were seen for most risk factors across all education categories apart from fruit, saturated fat intake, living alone, current alcohol drinking and average alcohol intake, for which positive gradients were seen. There was no significant gradient for the frequency of vegetable consumption.

    View this table:
    Table 1

    Characteristics of 38 355 subjects in the Melbourne Collaborative Cohort Study at baseline (1990–1994)

    Cardiovascular mortality and educational attainment

    There were 392 adjudicated CVD deaths during a mean follow-up time of 9.4 years per person. Of these, 239 related to CHD, 77 to stroke and 76 to ‘other’ CVD events. The crude mortality of CVD for those with primary education was more than twice that of those who had completed tertiary education (table 2). There was a trend towards an inverse gradient between education and fatal CVD events, although rates for the two intermediate categories were similar, with overlapping 95% CIs. Similar patterns were seen in population subgroups according to age, sex and country of birth.

    View this table:
    Table 2

    Crude cardiovascular disease mortalities in the Melbourne Collaborative Cohort Study overall and by subgroups

    Adjustment for risk factor groups

    After adjusting for age, sex and country of birth, CVD mortality for those with primary education only was significantly higher than for those who had completed tertiary education (HR 1.66, 95% CI 1.10 to 2.49; table 3). For those who had completed some or all secondary education, the CVD mortality was intermediate between the primary and tertiary education categories, and an inverse gradient was observed.

    View this table:
    Table 3

    Fatal cardiovascular disease outcomes: HRs from different Cox proportional hazards models for 38 355 Melbourne Collaborative Cohort Study subjects free of cardiovascular disease at baseline, with adjustment for behavioural, social and physiological risk factor combinations

    After further adjustment for the combined behavioural risk factors, the association between education and CVD mortality was attenuated, with a HR of 1.41 (95% CI 0.94 to 2.12) for those with only primary education, a 38% reduction in excess risk of CVD mortality. Adjustment for social connection reduced the HR for primary education by 2%, while adjustment for physiological risk factors reduced it by 45% (table 3). Adjustment for all risk factors combined reduced the HR for those with primary education only to 1.18 (95% CI 0.78 to 1.77), accounting for 73% of the excess risk of CVD mortality. Adjustment for all risk factors combined also eliminated differences in CVD mortality between the two intermediate education categories compared with those with tertiary education.

    Because risk-factor distribution differed by country of birth, analyses were repeated for a subgroup comprising only those born in Australia, NZ or northern Europe. We were unable to do this analysis for those born in southern Europe because there were only two deaths due to CVD among those with tertiary education. The risk associated with primary education only was higher than that seen in the total cohort with an HR adjusted for sex and age of 1.94 (95% CI 1.21 to 3.13). In this subgroup, adjustment for behavioural risk factors reduced the HR for CVD mortality comparing the highest and lowest education groups to a greater extent than adjustment for physiological risk factors (54% and 34% reduction respectively), opposite findings to those in the total cohort. Adjustment for social connection had a greater effect than for the total cohort, with a 13% reduction in HR. Simultaneous adjustment for all three risk factor groups reduced the HR for CVD mortality for the lowest compared with the highest education group to 1.18 (95% CI 0.72 to 1.92), a reduction in excess risk of 81%.

    Individual risk factors

    To determine the relative explanatory role of individual risk factors on the SES/CVD gradient for the total cohort, we adjusted separately for each risk factor (table 4). Adjustment for smoking reduced the HR for CVD mortality in the lowest compared with the highest education group by the greatest amount, followed by waist circumference and systolic blood pressure. Similar patterns were seen for those born in Australia, NZ and northern Europe only, although in this subgroup, the percentage reduction in excess risk was greater for smoking, and slightly smaller for waist circumference (33% and 19% respectively).

    View this table:
    Table 4

    Fatal cardiovascular disease outcomes: HRs from different Cox proportional hazards models for 38 355 Melbourne Collaborative Cohort Study subjects free of cardiovascular disease at baseline, with adjustment for individual risk factors

    Discussion

    In this very large contemporary Australian cohort, we found an inverse relationship between education attainment and CVD mortality. In the total cohort, physiological risk factors contributed more to this gradient than behavioural factors, although when southern European born migrants were excluded, behavioural factors made the larger contribution. In combination, behavioural, social and physiological factors explained almost all of the difference in CVD mortality between the tertiary and primary education categories in the total cohort, with smoking and waist circumference contributing most to this.

    Behavioural, physiological and social risk-factor groups

    Our finding that traditional risk factors can explain a substantial proportion of the difference in CVD mortality between low and high socio-economic groups has been reported by a limited number of studies.10–13 In one study, the risk of excess CHD among a low-income group of Finnish men was reduced by 118% after adjustment for 23 cardiovascular risk factors.13 Most other studies have found that combined risk factors play a smaller explanatory role.8 9 14 16 Differences in findings may relate to variations in the underlying social distribution of risk factors, the risk factor combinations examined and the health of the populations studied.9

    The relative explanatory effect of physiological factors was greater when southern European migrants were included in the analysis. This may be because this group, which comprised 85% of those with primary education, typically has a high prevalence of diabetes and other adverse physiological risk factors.33 Despite this, southern Europeans in our study had lower CVD mortalities than those born in Australia, NZ and northern Europe. The underlying reasons for this are unclear, although the Mediterranean diet, predominant among southern Europeans, may have been cardioprotective.34

    Although social connection is associated with lower rates of recurrent events in those with prevalent CVD,35 36 its role in incident CVD is less clear.30 37 We found that social factors had little contribution to CVD mortality in our study. This may be because the measures used were too crude to detect an association in this particular cohort.

    Individual risk factors

    The prevalence of current smoking at baseline was 11.5% compared with 27.6% in Victoria in 1989–1990.38 Despite this low prevalence, smoking was still found to be the most significant behavioural factor contributing to CVD inequality. This is consistent with many other contemporary studies.9 10 12 39 As expected, it is also consistent with another study from the same population demonstrating a substantial contribution of smoking to social inequalities in total mortality among men.15 The socio-economic gradient in smoking has widened in many western countries,40 and it is essential that this important risk factor continues to be addressed among disadvantaged groups in order to reduce their burden of CVD.

    Other contemporary studies have reported that overweight or obesity influences health inequality less than blood pressure.9 12 13 In this present study, waist circumference had a stronger effect than blood pressure and was the most important physiological risk factor both in analysis of the total cohort and in the subgroup born in Australia, NZ and northern Europe. Waist circumference may be a more sensitive anthropometric measure, as preliminary analysis of these data found that adjustment for waist-to-hip ratio or BMI (used in the majority of other studies) had a smaller effect. An additional explanation is that the importance of weight as an explanatory risk factor for CVD inequalities has increased because of the increasing prevalence of obesity, highlighting the importance of studies such as this to provide contemporary evidence.

    Methodological considerations and limitations

    This study was based on a large sample with accurate ascertainment of cause of death and lengthy follow-up. Education was the only SES indicator used in this study. It is possible that other indicators such as occupation or income may have contributed independently to the relationship between SES and CVD in this cohort. However, education is considered to be a robust measure of SES,41 and our findings are consistent with studies using other indicators.10 12 13 Although it is unlikely that education level changed during follow-up of this older cohort, changes in social, behavioural and physiological risk factors are possible, leading to attenuation of their effects.9–12 Because risk factors were measured at one time point only, no conclusions about temporal relationships between different groups of risk factors are possible. Despite this, our findings suggest that traditional risk factors do play an underlying mechanistic role in the gradient in CVD mortality related to SES.

    This study was conducted in Melbourne Australia, and findings may not be applicable to other regions. Generalisability is also affected by the oversampling of southern European migrants. Further, the MCCS has a low standardised mortality ratio for fatal CVD (0.41 relative to mortality levels in the state of Victoria).42 However, the observation of an inverse gradient in this relatively healthy cohort clearly underlines the strength of the association between education attainment and CVD.

    There was a significantly greater proportion of southern European born migrants in the group with primary education only. Some of this difference could be due to measurement bias. Although differential misclassification due to language barriers is unlikely (as questionnaires were in the subjects' preferred language), educational attainment may have different socio-economic meanings within the countries represented in this cohort. Analyses that excluded southern European born migrants showed a stronger educational gradient in CVD mortality than that seen in the total cohort. This may either reflect true differences between the two groups or be due to an underestimation of effects in the total cohort because of the inclusion of southern Europeans.

    Caution needs to be used when interpreting the percentage reduction in excess risk. This concept relies on an assumption that there is no confounding of the relationship between the intermediary variable and the outcome.43 44 This is difficult to assess from these data. This measure should therefore be considered as an approximate indication of relative importance only. In addition, the fact that there was still some association between education and CVD mortality in our study after adjustment for risk factors does not preclude the possibility of residual confounding.43 This may have occurred because of measurement bias, particularly as behavioural and social factors were self-reported.

    Conclusion

    The decline in CVD mortality over the past decades45 has not occurred equitably, with a widening of relative inequality in many countries, including Australia.2 It has been suggested that because traditional risk factors do not account for all of the SES/CVD gradient, addressing them within disadvantaged groups will not significantly reduce health inequalities.1 While we agree that health inequalities cannot be fully reduced until the underlying determinants of social disadvantage are addressed,1 these data suggest that addressing traditional risk factors (particularly smoking and central adiposity) may strongly influence the current gradient in CVD. These findings reinforce the need for public health policies and strategies to effectively address both behavioural and physiological risk factors among lower socio-economic groups.

    What is already known on this subject

    • The association between social disadvantage and increased risk of cardiovascular disease (CVD) is well established, but studies examining the contribution of risk factors to this association have shown inconsistent findings.

    • This study explores the relationship between education and CVD mortality within a large, contemporary and ethnically diverse sample of both men and women using a wide variety of risk factors.

    What this study adds

    • Traditional risk factors can explain a large part of the association between socioeconomic disadvantage and CVD. Consistent with several other studies, our findings show that smoking is the most important risk factor in this relationship; however, our study also shows that central adiposity is the second most important factor in explaining the increased risk of CVD among lower-educated groups.

    • The relative importance of risk factors also differs between those born in southern Europe, and those born in northern Europe/Australia and New Zealand, and dietary factors may contribute to this difference.

    Policy implications

    • This study provides contemporary evidence for the need for public health policies to effectively address both behavioural and physiological risk factors among disadvantaged groups.

    Acknowledgments

    This study was made possible by the contribution of many people, including the original investigators and the team who recruited the participants. We would like to express our gratitude to the many thousands of Melbourne residents who continue to participate in the study.

    References

    1. ↵
      1. Marmot M
      . Achieving health equity: from root causes to fair outcomes. Lancet 2007;370:1153–63.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. AIHW,
      2. Moon L,
      3. Waters A
      . Socioeconomic inequalities in cardiovascular disease in Australia: current picture and trends since the 1990s. Bulletin No. 37. AIHW cat. no. AUS 74. Canberra: AIHW, 2006.
    3. ↵
      World Health Organization. Social determinants of health: the solid facts. 2nd edn. Denmark: WHO, 2003.
    4. ↵
      1. Yusuf S,
      2. Hawken S,
      3. Ôunpuu S,
      4. et al
      . Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case–control study. Lancet 2004;364:937–52.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. O'Brien K
      AIHW:O'Brien K. Living dangerously: australians with multiple risk factors for cardiovascular disease. Bulletin No. 24. AIHW Cat. No. AUS 57. Canberra: AIHW, 2005.
    6. ↵
      World Health Organization. The social determinants of health: developing an evidence base for political action. Geneva: WHO, 2007.
    7. ↵
      1. Khang Y,
      2. Lynch J,
      3. Jung-Choi K,
      4. et al
      . 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 2008;94:75–82.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Kivimaki M,
      2. Lawlor D,
      3. Davey Smith G,
      4. et al
      . Socioeconomic position, co-occurrence of behavior-related risk factors, and coronary heart disease: the finnish public sector study. Am J Public Health 2007;97:874–9.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Laaksonen M,
      2. Talala K,
      3. Martelin T,
      4. et al
      . Health behaviours as explanations for educational level differences in cardiovascular and all-cause mortality: a follow-up of 60 000 men and women over 23 years. Eur J Public Health 2008;18:38–43.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Harald K,
      2. Pajunen P,
      3. Jousilahti P,
      4. et al
      . Modifiable risk factors have an impact on socio-economic differences in coronary heart disease events. Scand Cardiovasc J 2006;40:87–95.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Strand B,
      2. Tverdal A
      . Can cardiovascular risk factors and lifestyle explain the educational inequalities in mortality from ischaemic heart disease and from other heart diseases? 26 year follow up of 50 000 Norwegian men and women. J Epidemiol Community Health 2004;58:705–9.
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Woodward M,
      2. Oliphant J,
      3. Lowe G,
      4. et al
      . Contribution of contemporaneous risk factors to social inequality in coronary heart disease and all causes mortality. Prev Med 2003;36:561–8.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Lynch J,
      2. Kaplan G,
      3. Cohen R,
      4. et al
      . Do cardiovascular risk factors explain the relation between socioeconomic status, risk of all-cause mortality, cardiovascular mortality, and acute myocardial infarction? Am J Epidemiol 1996;144:934–42.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Marmot M,
      2. Davey Smith G,
      3. Stansfeld S,
      4. et al
      . Health inequalities among British civil servants: the Whitehall II study. Lancet 1991;991:1387–93.
      OpenUrl
    15. ↵
      1. Shiahpush M,
      2. English DE,
      3. Powles J
      . The contribution of smoking to socioeconomic differentials in mortality: results from the Melbourne collaborative cohort study, Australia. J Epidemiol Community Health 2006;60:1077–9.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Emberson J,
      2. Whincup P,
      3. Morris R,
      4. et al
      . Social class differences in coronary heart disease in middle-aged British men: implications for prevention. Int J Epidemiol 2004;33:289–96.
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Marmot M
      . Commentary: risk factors or social causes? Int J Epidemiol 2004;33:297–8.
      OpenUrlFREE Full Text
    18. ↵
      AIHW. Australia's health 2006. AIHW cat. no. AUS 73. Canberra: AIHW, 2006.
    19. ↵
      1. Schrijver C,
      2. Stronks K,
      3. van de Mheen,
      4. et al
      . Explaining educational differences in mortality: the role of behavioral and material factors. Am J Public Health 1999;89:535–40.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Lantz P,
      2. House J,
      3. Lepkowski J,
      4. et al
      . Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults. JAMA 1998;279:1703–8.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Kanjilal S,
      2. Gregg E,
      3. Cheng Y,
      4. et al
      . Socioeconomic status and trends in disparities in 4 major risk factors for cardiovascular disease among US adults, 1971–2002. Arch Intern Med 2006;166:2348–55.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Lyratzopoulos G,
      2. Heller R,
      3. McElduff P,
      4. et al
      . Deprivation and trends in blood pressure, cholesterol, body mass index and smoking among subjects of a UK primary care-based cardiovascular risk factor screening programme: both narrowing and widening in cardiovascular risk factor inequalities. Heart 2006;92:1198–206.
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Giles G,
      2. English D
      . The Melbourne collaborative cohort study IARC scientific publications No 156. Lyon: International Agency for Research on Cancer, 2002:69–70.
    24. ↵
      1. Dutton T,
      2. Turrell G,
      3. Oldenburg BF
      . Measuring socioeconomic position in population health monitoring and health research. Brisbane, Australia: Queensland University of Technology, 2005.
    25. ↵
      Medical Research Council's Committee on Environmental and Occupational Health. Questionnaire on respiratory symptoms. London, UK: Medical Research Council, 1986.
    26. ↵
      1. Cotes J,
      2. Chinn D
      . MRC questionnaire (MRCQ) on respiratory symptoms. Occup Med 2007;57:388.
      OpenUrlFREE Full Text
    27. ↵
      1. Ainsworth B,
      2. Haskell W,
      3. Leon A,
      4. et al
      . Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc 1993;25:71–80.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Hodge A,
      2. Patterson A,
      3. Brown W,
      4. et al
      . The anti cancer council of victoria FFQ: relative validity of nutrient intakes compared with weighed food records in young to middle-aged women in a study of iron supplementation. Aust N Z J Public Health 2000;24:576–83.
      OpenUrlPubMedWeb of Science
    29. ↵
      1. Ireland P,
      2. Jolley D,
      3. Giles GG
      . Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving an ethnically diverse cohort. Asia Pac J Clin Nutr 1994;3:19–31.
      OpenUrl
    30. ↵
      1. Kaplan GA,
      2. Salonen JT,
      3. Cohen RD,
      4. et al
      . Social connections and mortality from all causes and from cardiovascular disease: prospective evidence from eastern Finland. Am J Epidemiol 1998;128:370–80.
      OpenUrl
    31. ↵
      1. Lohman T,
      2. Roche A,
      3. Martorell R
      . Anthropometric Standardization Reference Manual. Champaign, IL: Kinetics Books, 1988.
    32. ↵
      1. van Oort F,
      2. van Lenthe F,
      3. Mackenbach J
      . Material, psychosocial, and behavioural factors in the explanation of educational inequalities in mortality in The Netherlands. J Epidemiol Community Health 2005;59:214–20.
      OpenUrlAbstract/FREE Full Text
    33. ↵
      1. Hodge A,
      2. English D,
      3. O'Dea K,
      4. et al
      . Increased diabetes incidence in Greek and Italian migrants to Australia: how much can be explained by known risk factors? Diabetes care 2004;27:2330–4.
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Harriss LR,
      2. English DR,
      3. Powles G,
      4. et al
      . Dietary patterns and cardiovascular mortality in the Melbourne collaborative cohort study. Am J Clin Nutr 2007;86:221–9.
      OpenUrlAbstract/FREE Full Text
    35. ↵
      1. Brummett B,
      2. Barefoot J,
      3. Seigler I,
      4. et al
      . Characteristics of socially isolated patients with coronary artery disease who are at elevated risk for mortality. Psychosom Med 2001;63:267–72.
      OpenUrlAbstract/FREE Full Text
    36. ↵
      1. Rutledge T,
      2. Reis S,
      3. Olsen M,
      4. et al
      . Social networks are associated with lower mortality rates among women with suspected coronary disease: the National Heart, Lung, and Blood Institute-sponsored women's ischemia syndrome evaluation study. Psychosom Med 2004;66:882–8.
      OpenUrlAbstract/FREE Full Text
    37. ↵
      1. Lett H,
      2. Blumenthal JA,
      3. Babyak MA,
      4. et al
      . Social support and coronary heart disease: epidemiologic evidence and implications for treatment. Psychosom Med 2005;67:869–78.
      OpenUrlAbstract/FREE Full Text
    38. ↵
      Australian Bureau of Statistics. Australian social trends, Cat. No. 4102.0. 1994. http://www.abs.gov.au/ (accessed 22 Dec 2008).
    39. ↵
      1. Jha P,
      2. Chaloupka F,
      3. Corrao M,
      4. et al
      . Reducing the burden of smoking world-wide: effectiveness of interventions and their coverage. Drug Alcohol Rev 2006;25:597–609.
      OpenUrlCrossRefPubMedWeb of Science
    40. ↵
      1. White V,
      2. Hill D,
      3. Siahpush M,
      4. et al
      . How has the prevalence of cigarette smoking changed among Australian adults? trends in smoking prevalence between 1980 and 2001. Tob Control 2003;12:67ii–74.
      OpenUrlAbstract/FREE Full Text
    41. ↵
      1. Galobardes B,
      2. Shaw M,
      3. Lawlor D,
      4. et al
      . Indicators of socioeconomic position (part 1). J Epidemiol Community Health 2006;60:7–12.
      OpenUrlAbstract/FREE Full Text
    42. ↵
      1. Harriss L,
      2. English D,
      3. Hopper J,
      4. et al
      . Alcohol consumption and cardiovascular mortality accounting for possible misclassification of intake: 11-year follow-up of the Melbourne collaborative cohort study. Addiction 2007;102:1574–85.
      OpenUrlCrossRefPubMedWeb of Science
    43. ↵
      1. Szklo M,
      2. Nieto JF
      . Epidemiology: beyond the basics. 2nd edn. Toronto: Jones and Bartlett, 2006:161.
    44. ↵
      1. Cole SR,
      2. Hernan M
      . Fallibility in estimating direct effects. Int J Epidemiol 2002;31:163–5.
      OpenUrlAbstract/FREE Full Text
    45. ↵
      1. Ford E,
      2. Ajani U,
      3. Croft J,
      4. et al
      . Explaining the decrease in U.S. deaths from coronary disease, 1980–2000. N Engl J Med 2007;356:2388–98.
      OpenUrlCrossRefPubMedWeb of Science

    Footnotes

    • Funding This work was supported by the National Health and Medical Research Council (NHMRC) (ID No 209057, 334032, 396414). Further infrastructure support was provided by The Cancer Council Victoria and Monash University. AB is a PhD scholar funded by the NHMRC (ID No 465352). GT is a Senior Research Fellow funded by the NHMRC (ID No 390109). AP is funded by a VicHealth Fellowship. Cohort recruitment was funded by VicHealth and The Cancer Council Victoria.

    • Competing interests None.

    • Ethics approval Ethics approval was provided by the Cancer Council Victoria Human Research Ethics Committee.

    • Provenance and peer review Not commissioned; externally peer reviewed.