Intended for healthcare professionals


Does access to cardiac investigation and treatment contribute to social and ethnic differences in coronary heart disease? Whitehall II prospective cohort study

BMJ 2004; 329 doi: (Published 05 August 2004) Cite this as: BMJ 2004;329:318
  1. Annie Britton, lecturer in epidemiology1,
  2. Martin Shipley, senior lecturer in medical statistics1,
  3. Michael Marmot, professor of epidemiology and public health1,
  4. Harry Hemingway (h.hemingway{at}, reader in clinical epidemiology1
  1. 1 International Centre for Health and Society, Department of Epidemiology and Public Health, University College London Medical School, London WC1E 6BT
  1. Correspondence to: H Hemingway
  • Accepted 25 May 2004


Objective To determine whether access to cardiac procedures and drugs contributes to social and ethnic differences in coronary heart disease in a population setting.

Design Prospective study with follow up over 15 years. Civil service employment grade was used as a measure of individual socioeconomic position. Need for cardiac care was determined by the presence of angina, myocardial infarction, and coronary risk factors.

Setting 20 civil service departments originally located in London.

Participants 10 308 civil servants (3414 women; 560 South Asian) aged 35-55 years at baseline in 1985-8.

Main outcome measures Use of exercise electrocardiography, coronary angiography, and coronary revascularisation procedures and secondary prevention drugs.

Results Inverse social gradients existed in incident coronary morbidity and mortality. South Asian participants also had higher rates than white participants. After adjustment for clinical need, social position showed no association with the use of cardiac procedures or secondary prevention drugs. For example, men in the low versus high employment grade had an age adjusted odds ratio for angiography of 1.87 (95% confidence interval 1.32 to 2.64), which decreased to 1.27 (0.83 to 1.94) on adjustment for clinical need. South Asians tended to be more likely to have cardiac procedures and to be taking more secondary prevention drugs than white participants, even after adjustment for clinical need.

Conclusion This population based study, which shows the widely observed social and ethnic patterning of coronary heart disease, found no evidence that low social position or South Asian ethnicity was associated with lower use of cardiac procedures or drugs, independently of clinical need. Differences in medical care are unlikely to contribute to social or ethnic differences in coronary heart disease in this cohort.


Low social position and South Asian ethnicity are both associated with increased risk of dying from coronary heart disease,1 2 but the impact of differential access to medical care on these inequalities remains uncertain. If access to coronary management matched coronary incidence, then low social position and South Asian ethnicity would be associated with higher rates of use. For social position, most studies,310 but not all,1113 find the opposite—high social deprivation is associated with lower rates of coronary angiography and revascularisation. Several studies, mainly small and retrospective, report less aggressive treatment of South Asian people with coronary disease compared with white patients.1422 Such potential healthcare disparities have stimulated calls in the United States and United Kingdom for remedial action.23 24

Three interrelated questions remain unanswered. Firstly, in a general population that exhibits social and ethnic differences in rates of coronary heart disease, do differences exist in access to care? Population studies are lacking; all but a few studies are confined to patients who have sought hospital care for coronary disease and therefore exclude many people with angina in the community.12 Furthermore, studies have concentrated on invasive procedures, ignoring non-invasive investigation and secondary prevention, which may be particularly relevant in primary care. Secondly, how does the social deprivation of an individual patient, as opposed to an area, influence access to cardiac investigation and treatment? Most previous studies have used neighbourhood measures of social deprivation.311 One of the few prospective studies reporting individual social class, albeit confined to white men, found no effect on access to coronary angiography but lower revascularisation rates among men from non-manual occupations.25 Thirdly, among South Asians, is the apparent lower use of cardiac investigation and treatment independent of or explained by their social position?26 Previous studies have made little attempt to answer this question.

The Whitehall II prospective cohort study of civil servants offers the opportunity to consider each of these questions. Our objective was to determine whether access to cardiac procedures (exercise electrocardiography, coronary angiography, and coronary revascularisation) and secondary prevention drugs contributes to social and ethnic differences in coronary heart disease in a population setting.


All non-industrial civil servants aged 35-55 years working in the London offices of 20 departments were invited to participate in the Whitehall II study. Recruitment took place during 1985-8, and the final cohort consisted of 10 308 participants (3414 women). Between 1985 and 1999 self completed questionnaires were obtained at five phases of data collection and physical examination was conducted at three phases. Full details are published elsewhere.27

Socioeconomic position

We used civil service employment grade as a measure of socioeconomic position. On the basis of salary and work role, the civil service defines a hierarchy of employment grades, which we analysed in three levels: unified grades 1-7 (high), executive officers (medium), and clerical and support staff (low), as described previously.28


We defined ethnicity according to the Office for National Statistics 1991 census categories. Of participants with known ethnicity, 9162 (89.7%) described themselves as white; 560 participants were grouped as South Asian, of whom 74% were Indian, 13% Sri Lankan, 10% Pakistani, and 3% Bangladeshi.

Clinical need

We obtained evidence of incident angina and non-fatal myocardial infarction from questionnaire items at each follow up phase (symptoms of chest pain, recall of diagnosis by doctor), resting electrocardiograms (done at screening phases 1, 3 (1991-3), and 5 (1997-9) and classified according to the Minnesota code), and clinical records. Full details of ascertainment and classification are published elsewhere.29

A total of 10 300 (99.9%) participants were flagged at the NHS Central Registry. The registry notified us of the date and cause of death up to the end of 2001.

Risk factors

At baseline, standardised assessments were made of smoking, blood pressure, total cholesterol, body mass index, self reported diagnosis of diabetes, and family history of onset of coronary heart disease before age 55 years. In addition, the presence of the metabolic syndrome was assessed at phases 3 and 5 on the basis of oral glucose tolerance tests, plasma triglycerides and high density lipoprotein, waist:hip ratio, and systolic blood pressure.30


Exercise electrocardiography, angiography, and revascularisation

A total of 7830 participants completed a phase 5 questionnaire (76% response rate from phase 1). Of these, 994 participants reported that they had had an exercise electrocardiogram, 403 reported having a coronary angiogram, and 124 reported having had coronary angioplasty or a coronary artery bypass graft operation (revascularisation procedure).

To investigate the possibility that employment grade or ethnicity may influence the reporting of medical care, we examined hospital discharge data from the NHS-wide clearing service. All 10 308 participants were flagged with their NHS number for coronary angiography (code K634, K635, K636), coronary bypass grafting (K401-89) and angioplasty procedures (K491-501) until the end of 2001. As the results obtained by using discharge data did not differ by grade or ethnicity from those obtained by using self reported data, we combined both sources. The final numbers used in the analysis were 525 angiograms (122 identified from discharge data alone) and 179 revascularisations (55 identified from discharge data alone).

Secondary prevention drugs

At the phase 5 follow up participants reported their use of prescribed drugs over the previous 14 days. We coded all drugs according to the British National Formulary and defined secondary prevention drugs as aspirin, β blockers, lipid lowering agents, or angiotensin converting enzyme inhibitors.

Statistical analysis

We based the analyses of use of exercise electrocardiography on those participants who answered the question in the phase 5 questionnaire (n = 7805). We based the analyses of coronary angiography and revascularisation on all 10 308 participants and adjusted all these analyses for data source. We calculated the age adjusted rates of event outcomes and prevalence of having had a procedure (exercise electrocardiography, angiography, or revascularisation) by using direct standardisation. We used Cox's proportional hazards model to calculate adjusted rates and 95% confidence intervals for event outcomes. We used logistic regression to calculate odds ratios of having had a procedure for each employment grade, with high grades as the reference group. We made statistical adjustments for age, evidence of coronary heart disease, diabetes, risk factors at baseline, and ethnicity. We similarly calculated odds ratios of having a procedure by ethnicity, with “white” as the reference group, adjusting in addition for employment grade. For the angiography and revascularisation analyses, we included an indicator of phase 5 attendance in order to minimise the potential bias from loss to follow up. To assess the quality of medical care we calculated the age adjusted prevalence of use of secondary prevention drugs among participants who attended phase 5 and had a history of myocardial infarction or angina.


Table 1 shows that incident coronary morbidity and mortality were higher among lower employment grades than among higher grades and higher among South Asians than among white participants. For example, in men, the age adjusted rate ratio for incident angina and myocardial infarction was 1.66 (95% confidence interval 1.32 to 2.10) for low versus high employment grades and 1.95 (1.28 to 2.96) for South Asians compared with white participants. Risk factors tended to be adverse in low employment grades and among South Asians. South Asians were less likely to be in a high employment grade than were white participants.

Table 1

Age adjusted and mean coronary risk factors at baseline by employment grade and ethnicity and rates of subsequent coronary events

View this table:

Social position

Men and women in the low employment grades reported higher use of exercise electrocardiography (age adjusted proportion 17.5% in men and 10.6% in women) than did those in the high employment grades (14.7% in men and 8.6% in women) (table 2). When we adjusted for history of coronary heart disease during follow up and baseline risk factors, we found no evidence of an overall grade gradient in either men or women. Men and women in the low employment grades had the highest use of angiography with adjustment for age alone, but no grade differences existed when we added clinical need and other risk factors to the models. Similarly, participants in the low grades had the highest use of revascularisation when we adjusted for age alone, but among men these differences were removed by further adjustment. Employment grade was not associated with taking secondary prevention drugs among the subgroup of participants with a history of angina or myocardial infarction (table 3).

Table 2

Use of exercise electrocardiography, coronary angiography, and coronary revascularisation by employment grade

View this table:
Table 3

Age adjusted use of secondary prevention drugs by employment grade among participants with a history of angina or myocardial infarction. Values are percentages unless stated otherwise

View this table:


South Asian men and women were more likely to have an exercise electrocardiogram or coronary angiography than white participants, even after adjustment for clinical need and employment grade (table 4). We found less evidence for ethnic differences in revascularisation procedures. Further adjustment for presence of the metabolic syndrome or diabetes or abnormality on resting electrocardiogram did not attenuate any of these effects (data not shown). South Asians also tended to be more likely to take secondary prevention drugs than white participants—for example, among men with a history of angina or myocardial infarction, 34% of South Asians were taking β blockers compared with 14% of white men (see table on

Table 4

Use of exercise electrocardiography, coronary angiography, and revascularisation by ethnicity

View this table:


Women were less likely to have coronary investigations and treatments than men within each grade (table 2) and ethnic group. When adjusted for age, coronary heart disease, employment grade, and ethnicity, the odds ratios of women having an exercise electrocardiogram, angiogram, and revascularisation compared with men were 0.51 (95% confidence intervals 0.42 to 0.62), 0.48 (0.38 to 0.62), and 0.25 (0.15 to 0.40).


Social position

In this population based cohort, which exhibits the ubiquitous inverse social gradient in coronary morbidity and mortality, we found no evidence of a social gradient in use of cardiac procedures and drugs after adjustment for clinical need. This suggests that use of medical care broadly reflects clinical need and does not explain the well documented social gradient in coronary morbidity and mortality. Differential access to medical care has previously been proposed as an explanation for the relation between social position and health events31; we found no evidence to support this.


South Asian civil servants, predominantly Indian, had a twofold increased risk of coronary morbidity and mortality compared with white civil servants. Contrary to the hypothesis that South Asians might be disadvantaged in their access to cardiac care, we found some evidence of higher rates of use of procedures in South Asians than in white participants, even after adjustment for coronary morbidity, risk factors, and the metabolic syndrome. This may be interpreted as South Asian patients and their doctors responding to the widely perceived increased risk of heart disease with lower thresholds for action. Indeed, a survey on attitudes to health seeking behaviour found that South Asians were more likely to seek immediate care for (hypothetical) anginal symptoms than white people.32 In the appropriateness of coronary revascularisation study, South Asians had more consultations with a general practitioner in the year before angiography than did white patients.22

In socioeconomic terms, the South Asians working in the civil service represent a relatively advantaged and homogeneous group. This offers the opportunity to separate the potential confounding link between social position and ethnicity,26 thus strengthening our finding that, in this study, ethnic differences in care do not explain ethnic differences in coronary heart disease.

Limitations of the study

Clearly, civil servants do not represent the extremes of social position, nor the diversity of South Asian communities. However, the Whitehall II study shows marked social and ethnic differences in coronary heart disease, similar in magnitude to those reported in many other general population studies, which warrant exploration for the determinants. Although largely based in the southeast of England, by 1995, owing to progressive relocation of civil service departments, 44% of participants lived outside Greater London. Furthermore, many Whitehall participants live in areas that are among the most socially deprived in the country. On the basis of the Townsend deprivation index (based on unemployment, overcrowding, non-car ownership, and non-home ownership) approximately 20% of men and 40% of women participants live in the most deprived fifth of electoral wards in England.

The clinical “careers” of people with angina or myocardial infarction evolve over time, punctuated by changes in morbidity and in healthcare response. The Whitehall II study, in common with other population based studies, is not able to characterise the detailed history of appropriateness of each management step in this career, nor were we able to explore waiting times.

Strengths of the study

Civil service employment grade has proved an informative tool with which to investigate the influence of social position on health. Access to use of a car is a measure of social position available in wider settings and also may influence travel to healthcare facilities. We further tested our associations with car access and found higher rates of coronary heart disease and greater use of procedures in those participants without access to a car (data not shown). This mirrors our findings with employment grade.

The Whitehall II study offers, as far as we are aware, unique data with which to explore the impact of access to cardiac care. We carried out detailed, repeated assessments of clinical need extending beyond symptoms and resting electrocardiograms to include the metabolic syndrome. The metabolic syndrome is associated with low social position and South Asian ethnicity and may influence decisions about testing and treatment. Furthermore, whereas other studies have examined hospital admissions for revascularisation, the Whitehall II study is alone in reporting the more prevalent, earlier stages of non-invasive and invasive investigation.

Quality of care and secondary prevention

Our demonstration that the quantity of medical care in terms of procedure rates is not lower by social position or South Asian ethnicity does not exclude the possibility that the quality of care differs.33 34 Procedures might be less appropriate,35 and the outcomes may be worse, among people of low social position or South Asian ethnicity. One of the few studies to test this hypothesis directly found no evidence that this was the case when examining ethnic differences in revascularisation.22 In Whitehall II, use of secondary prevention, a direct marker of the quality of care, showed no differences by social position and tended to be higher among South Asians. Since these data were collected, government policy has emphasised the equitable increase in secondary prevention on the basis of disease registers in primary care.


The findings on social position and ethnicity were consistent in men and women; however, women had substantially lower rates of use of cardiac procedures and secondary prevention than men, consistent with other studies.36 37 Reasons for this may include less severe disease among women, less willingness to undergo procedures, differences in language used to describe symptoms,38 or bias by physicians.

Implications for research and policy

If medical care is not important in explaining social and ethnic differences in coronary disease, this strengthens the case for investigating other behavioural, biological, and psychosocial factors. Healthcare systems should monitor social and ethnic differences in treatment rates and prognosis; explaining differences in incidence of disease demands population based cohort studies in which people are recruited regardless of contact with medical care.


In this population based study we found no evidence that low social position or South Asian ethnicity was associated with lower use of cardiological investigation or treatment independent of clinical need. Differences in access to medical care are unlikely to explain the social and ethnic differences in coronary heart disease.

What is already known on this topic

Low social position and South Asian ethnicity are associated with higher risk of developing coronary heart disease

Previous studies which suggested social and ethnic differences in access to care were limited by ecological measures of social position, poor ability to distinguish ethnic from social effects, and lack of population base

What this study adds

Social position was not associated with use of cardiac procedures or secondary prevention, once account was taken of clinical need

South Asians tended to be more likely to undergo cardiac procedures and be on secondary prevention drugs than were white participants

Access to medical care is unlikely to explain the social or ethnic differences in coronary morbidity


  • Embedded Image An additional table is on

    Professor Peter W Macfarlane of the Royal Infirmary, Glasgow, reviewed all the electrocardiographs. We thank all participating civil service departments and their welfare, personnel, and establishment officers; the Occupational Health and Safety Agency; the Council of Civil Service Unions; all participating civil servants in the Whitehall II study; and all members of the Whitehall II study team.

  • Contributors AB and HH had the idea for the study. AB wrote the first draft and incorporated comments from all co-authors. MS did all statistical analyses. AB is the guarantor.

  • Funding The Whitehall II study has been supported by grants from the Medical Research Council; British Heart Foundation; Health and Safety Executive; Department of Health; National Heart Lung and Blood Institute (HL36310), US, NIH; National Institute on Aging (AG13196), US, NIH; Agency for Health Care Policy Research (HS06516); and the John D and Catherine T MacArthur Foundation Research Networks on Successful Midlife Development and Socioeconomic Status and Health. AB and MS are supported by the British Heart Foundation. HH is supported by a public health career scientist award from the Department of Health. MM is supported by an MRC research professorship.

  • Competing interests None declared.

  • Ethical approval Each phase of the Whitehall II study has received ethical approval from the research ethics committee of UCL Hospitals.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.