OUP user menu

Ethnic differences in myocardial infarction in patients with hypertension: effects of diabetes mellitus

J.V. Patel, H.S. Lim, A. Gunarathne, I. Tracey, P.N. Durrington, E.A. Hughes, G.Y.H. Lip
DOI: http://dx.doi.org/10.1093/qjmed/hcm151 231-236 First published online: 19 January 2008


Background: It has been reported that hypertension carries a greater risk of myocardial infarction (MI) in South Asians living in the UK than in the indigenous British population. This has been attributed to some specifically Asian susceptibility factor.

Design: Using a longitudinal approach, we investigated the relationship between coronary heart disease (CHD) risk factors amongst hypertension patients attending Sandwell and City Hospitals, and the onset of cardiovascular events over a 5-year follow-up period.

Results: A total of 350 Caucasian (83.7% male) and 104 South Asian (66.3% male) patients with hypertension [age 63.7 (7.6) years and 57.1 (11.1) years respectively, P < 0.001] were followed-up for a mean (SD) period of 64.7(12.1) months. There were 11 (6.4/1000 patient years) cases of MI in Caucasian patients vs. 11 (17.8/1000 patient years) in South Asians, with event-free survival times being significantly lower amongst South Asians (log-rank test P = 0.04). The prevalence of diabetes mellitus was 22.9% higher amongst South Asians (P < 0.001), whilst mean serum cholesterol and fasting triglyceride levels were higher amongst Caucasians (P = 0.001). There were no ethnic differences in HDL cholesterol concentrations, the use of tobacco, statin therapy or anti-platelet therapies (all P = NS), or in composite endpoint (MI, angina, peripheral vascular disease, stroke, revascularization or death; P = 0.74). On Cox regression analysis of all independent cardiovascular risk variables, associated treatments and ethnicity, MI risk was associated with diabetes mellitus (odds ratio 3.77, 95%CI 1.55–9.15, P = 0.003) but not ethnicity per se (P = 0.26).

Conclusion: Increased risk of MI in hypertensive South Asians in the United Kingdom appears to be the result of a higher prevalence of diabetes mellitus. Further work is required to understand the pathophysiological basis with which diabetes increases CHD risk in this ethnic group.


The causes of the excess coronary heart disease (CHD) risk among South Asians living in Britain are unclear, and this increased burden continues to be a source of the disparity in equitable healthcare for this ethnic group.1,,2 Recently published guidelines (JBS 2) on the assessment and management of people with established cardiovascular disease (CVD) and persons at high risk of developing CVD, have dismissed the need for specific guidelines for risk factor targets for South Asians.3 However, a paucity of longitudinal data (cause preceding effect) is likely to hamper an evidence-based consensus approach for ethnically specific CHD management and prevention.1

In the UK, the disproportionately higher CHD mortality rates in South Asians have been related to hypertension.4 However, mean blood pressure levels, the use of antihypertensive therapy5 and secular trends in blood pressure6 amongst South Asians appear to be comparable with European Caucasian counterparts.7 Longitudinal studies also show that hypertension is a strong risk factor for CHD in Indians living elsewhere.8,,9 The possibility remains that migration from the Indian subcontinent to the UK, and UK residence therein, confers an adverse exposure to blood pressure in this group,10,,11 which may explain a greater prevalence of hypertension-related morbidity in South Asians.12

Amongst UK patients with essential hypertension, South Asian ethnicity was an independent predictor of cardiovascular outcomes, with an excess of CHD.5 The UK Prospective Diabetes Study (UKPDS), which was a 9-year follow-up assessment of diabetes patients, showed that South Asian ethnicity was not associated with incident CHD.13 The disparity between findings between the post hoc analyses in these studies suggest that diabetes could explain the excess CHD risk in South Asians but further evidence was clearly needed from prospective data. Also, it is not clear whether ethnic differences in CHD are explained by variations in macrovascular risk factors associated with diabetes.

Using a 5-year follow-up analysis of patients treated for hypertension, we tested the hypothesis that rates of myocardial infarction (MI) would be higher amongst South Asian patients compared to white European Caucasian counterparts, who were being managed for hypertension, diabetes and CHD risk therein. Determinants of this ethnic difference were related to CHD risk factors measured at baseline.


Patients who were attending our hypertension and cardiovascular risk screening clinics (Sandwell and City Hospitals, Sandwell and West Birmingham Hospitals NHS Trust, West Midlands UK) between January 1998 and September 2000 were assessed for CHD risk. All patients were residing in the Sandwell and West Birmingham area (West Midlands). For the present analysis, only hypertensive patients were included (hypertension was defined as systolic blood pressure ⩾140 or diastolic ⩾90 mmHg or existing antihypertensive therapy) and that there was documented case note evidence that these patients were receiving antihypertensive treatment. Patients of an ethnicity that was not deemed to be of South Asian or European Caucasian origin (as reported by the patient), or those with clinical evidence of secondary hypertension or malignant phase hypertension and major concomitant non-CVD were excluded.

The CHD risk assessment was preceded with an invitation to attend one of our clinics upon fasting from 10.00 pm the previous evening. During the clinic appointment, patients were interviewed by a trained fieldworker. Demography, medical history and place of origin were assessed on each individual. Venous blood was collected from all fasting participants. Plasma glucose, serum cholesterol, serum triglycerides and HDL cholesterol were determined by routine auto analyser assay within the Biochemistry Departments of Sandwell and City Hospitals (each under daily quality assurance monitoring schemes). Diabetes mellitus was diagnosed by clinical history and elevated fasting and post challenge glucose levels according to contemporary criteria (WHO guidelines).13

Blood pressure was measured after the patient was seated in a quite room for >5 min using the OMRON 705CP (Omron Healthcare Europe, Mannheim, Germany), a minimum of three readings were performed and the average of the last two readings was used. Every participant provided written or witnessed consent to be involved within the research, which had local research ethics committee approval.

The study was of a prospective design, with no provision for disease stratification (e.g. known diabetes), or matching by age or gender. Using a standardized case note analysis across ethnic groups, details of the following endpoints (and dates of occurrence) were included, (i) angina (supported by recorded symptoms and electrocardiographic evidence), (ii) MI (electrocardiographic evidence and tropoinin elevation), (iii) non-haemorrhagic stroke (supporting evidence from computed tomography when possible), (iv) coronary intervention (documented evidence of percutaneous coronary angioplasty and coronary artery bypass graft), (v) peripheral vascular disease (low ankle brachial pressure index (<0.9) or intermittent claudication) and (vi) death. A composite endpoint, including all of these listed endpoints was also generated.

Power calculations

We hypothesized a higher incidence of MI in South Asians compared to European Caucasians on 5-year follow-up. Statistical power calculations were based on previous data reported by Khattar et al.,5 in which a 5-year follow-up study of 100 South Asian and 300 Caucasian hypertensive patients would allow more than 90% power to detect ethnic differences in hazard ratios reported in that study.

Statistical analyses

The parametric distribution of the data collected was tested using the Kolmogorov–Smirnov method. Blood pressure, body mass index (BMI), serum cholesterol, HDL cholesterol and plasma glucose were normally distributed and presented as mean (SD), whereas serum triglycerides were non-normally distributed and were presented as median [interquartile (IQR) range]. Data between the two ethnic groups were compared using unpaired t-tests or the Mann–Whitney U test (as appropriate) for continuous variables and the χ2- test for categorical variables. Survival curves were developed using the Kaplan–Meier method and differences in endpoint rates by ethnic group were analysed using the log-rank test. Multivariate analysis of disease outcomes was analysed using Cox regression modelling. Statistical significance was accepted at the 0.05 level (two-sided). Statistical analyses were undertaken using SPSS software v12 (SPSS Inc., Chicago, Illinois).


In total, 372 European Caucasians and 122 South Asian patients [age 63.7 (7.6) years and 57.1 (11.1) years, respectively, P < 0.001] with hypertension were studied. Of the Caucasians, all were treated for hypertension and cardiovascular risk, but 22 individuals were lost to follow-up. Of the South Asian patients, 18 were newly diagnosed for hypertension but had no documented evidence that they were subsequently treated for hypertension. Hence, a total of 350 Caucasians and 104 South Asian patients with hypertension were included in the final analysis. The mean follow-up period in these patients was 64.7 (SD 12.1) months.

Distribution of demography, medical history and CHD risk

On comparison of risk factors at initial assessment, Caucasian patients with hypertension were older (P < 0.001), more likely to smoke (P < 0.001), more obese (greater BMI, P < 0.001) and more hyperlipidaemic (P = 0.001) than South Asians, who had more diabetes (P < 0.001). The use of antihypertensive therapies was more common amongst the Caucasians, who were also more likely to be using two or more different blood pressure therapies (P < 0.001) and had higher mean systolic blood pressure compared to South Asians (P < 0.001) (Table 1).

View this table:
Table 1

Demography, medical history and cardiovascular risk factors amongst European Caucasian and South Asians patients with hypertension

Caucasian (n = 350)South Asian (n = 104)P
Age (years)63.7 (7.6)57.1 (11.1)<0.001
Male (%)83.7 (293)66.3 (69)<0.001
History of ischaemic heart disease (%)6.00 (21)7.70 (8)0.53
History of cerebrovascular disease (%)6.60 (23)2.90 (3)0.16
History of target organ damage (%)11.4 (40)22.5 (9)0.045
Smoking history (%)20.0 (77)17.3 (18)<0.001
Diabetes history (%)19.4 (68)42.3 (44)<0.001
ACE therapy (%)49.3 (172)34.5 (29)0.015
Beta blocker therapy (%)44.7 (156)29.8 (25)0.013
On two or more antihypertensives (%)47.0 (164)22.1 (23)<0.001
Statin therapy (%)59.6 (208)50.0 (42)0.20
Aspirin therapy (%)41.4 (144)32.1 (27)0.12
Insulin therapy (%)3.7 (13)4.3 (3)0.69
Metformin therapy (%)56.9 (41)34.3 (24)0.07
Sulphonylurea therapy (%)44.4 (32)31.4 (22)0.11
Glitazone therapy (%)1.4 (1)4.3 (2)0.30
Body mass index (kg/m2)29.5 (4.6)28.3 (5.0)0.016
Systolic blood pressure (mmHg)166 (17)156 (20)<0.001
Diastolic blood pressure (mmHg)90.1 (10.2)88.8 (10.4)0.21
Fasting plasma glucose (mmol/l)6.40 (2.28)7.14 (3.07)0.008
Serum cholesterol (mmol/l)6.08 (1.16)5.63 (1.09)<0.001
Serum triglycerides (mmol/l)a1.70 (1.20–2.50)1.20 (0.99–1.73)0.001
HDL cholesterol (mmol/l)1.37 (0.90)1.35 (0.39)0.82
  • Data are percent (n) or mean (SD), aexcept where median (IQR) are shown.

Variations in disease rates

During the 5-year follow-up period the rate of MI was 17.8 per 1000 patient years in South Asian patients (11 cases, including 2 females), which double that observed in European Caucasians, 6.5 per 1000 patient years (11 cases, including 1 female). The rate of all cause mortality in European Caucasians was 16.1/1000 patient years (15 deaths) and 12.2/1000 patient years (seven deaths) in South Asians.

Survival analysis

On univariate analysis using the Kaplan–Meier method, South Asians had a poorer event-free survival rate for MI compared to the European Caucasians (log rank chi square = 4.12, P = 0.043) (Figure 1). On similar analysis across all patients with hypertension, patients with diabetes had a poorer event-free survival rate for MI compared to those without diabetes (Figure 2). Using a broader composite endpoint of angina, MI, non-haemorrhagic stroke, coronary intervention, peripheral vascular disease and death, there were a total of 52 events in European Caucasians compared with 16 events in South Asians, with no ethnic variation (P = 0.74). Compared to South Asians, Caucasian patients had poorer event-free survival for non-haemorrhagic stroke (log rank chi square = 3.15, P = 0.007) and non-significantly higher rates of angina (log rank chi square = 3.15, P = 0.007). There were no ethnic variations in rates of all cause mortality, revascularization or peripheral vascular disease. Using a multivariate approach, in a Cox regression analysis model that included baseline independent CVD risk factors, medical history, ethnicity and CVD therapy, the onset of MI was associated with diabetes (odds ratio, 95% CI: 3.77, 1.55–9.15, P = 0.003) and non-significantly, with anti-platelet therapy (3.9, 1.5–10.3, P = 0.07) (Table 2).

Figure 1.

Myocardial infarction in patients with hypertension by ethnicity.

Figure 2.

Myocardial infarction in patients with hypertension by diabetes status.

View this table:
Table 2

Cox regression analysis of baseline factors and the onset of myocardial infarction

Variableβ 95%(CI)P-value
Diabetes3.77 (1.55–9.15)0.003
Male gender0.30
ACE therapy0.90
Statin therapy0.39
Beta blocker use0.50
Antiplatelet use0.07
Systolic blood pressure0.42
Fasting plasma glucose0.85


This study suggests that the increased CHD risk in South Asians is explained by conventional risk factors in this population, most notably an excess of diabetes mellitus. This susceptibility of diabetes amongst South Asians promotes an adverse CHD risk, even in patients treated for high blood pressure and cardiovascular risk. Importantly, further work is required to understand the pathophysiological basis with which diabetes increases CHD risk in this ethnic group.

While the association between diabetes and MI is not novel, the implication of this longitudinal approach is that a susceptibility to diabetes mellitus, rather than ‘ethnicity’ per se is likely to explain the excess of CHD affecting South Asians.5 Diabetes was equally common between South Asians and Caucasians who had MI on follow-up. Indeed, our data, albeit with fewer events, are consistent with those published in the UKPDS analysis of ethnicity and cardiovascular outcomes,14 where rates of MI amongst diabetic South Asians were no different to other ethnic groups with diabetes. Previously, Khattar et al.5 reported that South Asian ethnicity was a predictor of CHD, independently of diabetes amongst a cohort of hypertensive patients; however, their analysis included an African Caribbean cohort that had high rates of diabetes, but low rates of CHD. Neither of these longitudinal studies have assessed variations in disease rates by gender. In the present study, there were more females in our South Asian patient group, but the impact of this on disease rates remains unknown. Previously, we reported no gender variations in the diagnosis of hypertension and the treatment of hypertension and glucose intolerance between Indian Gujaratis.11 Further work is warranted to highlight gender differences in the magnitude of impact from various cardiovascular risk factors and efficacy of cardiovascular medications therein amongst South Asians.

Metabolic risk factors such as HDL cholesterol and fasting triglycerides were not associated with the onset of MI in this study, but the pathophysiological impact of diabetes on CHD risk amongst South Asians is worthy of further attention. Indeed, there are a number of molecular mechanisms that may explain this clinical association.15,,16 The comparatively lower rates of MI reported amongst African Caribbean hypertension patients, who have comparable rates of diabetes (and higher triglycerides) to the South Asians, would imply that a putative protective factor is evoked in the former. Furthermore, the cardiovascular risk profile amongst the South Asian hypertensive group here is devoid of many risk factors on comparison to Caucasian contemporaries. Further research is needed to address the pathophysiological mechanism by which diabetes confers an adverse cardiovascular risk in South Asians, and whether there is ethnic variation therein.

Migration studies that have compared migrants from the Indian subcontinent living in Britain to contemporaries in India give an indication of the change of CHD risk factors that the migrant South Asian have been exposed to. These studies highlight that CHD risk factors are markedly higher amongst Indian Punjabi17 and Gujarati11 migrants compared to contemporaries still living in places of origin. Whilst absolute levels of CHD risk factors do not appear ethnically different to the general UK population,3 mean body mass indices, blood pressure, serum cholesterol and serum triglycerides deteriorate with migration.11 It is likely that the impact of cardiovascular risk factors in this and other migrant populations in Britain underestimate insidious changes.

The limitations of this study are the short follow-up period, which precludes any discussion on whether the disparity in rates of MI continues to widen between the South Asian and indigenous population of the UK. Moreover, given the number of events and time of follow-up, it was not possible to analyse predictors of event-free survival within ethnic groups or by gender. There were a number of Caucasian patients who were lost to follow-up despite a comprehensive approach that included contact with primary and secondary care associated with the management of risk within patients. Moreover, despite an initial diagnosis of hypertension, many South Asians remained untreated. The reasons for this were unclear, but may relate to barriers to healthcare,2 our approach to exclude these individuals from the analysis provides an ethnically sensitive bias in this hospital-based study – adding to confounding factors. Haemoglobin A1c levels are predictive of cardiovascular events in patients with diabetes18 but these measures were not consistently available at recruitment. Proportions of patients on diuretic therapy were not available.

Given the proposed 40% increment for the cardiovascular risk score estimate by JBS 2, and migration-related deterioration of major CHD risk factors (save diabetes), it may be reasonable to develop a lower threshold for blood pressure and serum cholesterol for the primary prevention of CVD amongst male and female South Asians in Britain. For example, a 5% reduction in optimal targets for blood pressure and cholesterol would be needed to manage an absolute risk estimate that is increased by 40%. However, a prospective study that charts the development of CHD amongst a representative sample of South Asians living in Britain is urgently needed to develop a true ethnically sensitive approach to CVD prevention in this population. Important considerations are gender and the mixture of geographical and cultural backgrounds amongst the South Asian population in Britain, which is changing.6 Also, there are likely to be differences between the first generation (those born in the Indian Subcontinent) and the second generation (those born in UK or outside of the Indian Subcontinent), with CHD risk factors likely to be relatively worse amongst the latter, as the risk of CHD mortality amongst migrants from the Indian subcontinent increases with duration of living in Britain.19

In this study, incident stroke was more common amongst the Caucasian group, the reasons for this remain unclear. In an epidemiological study of stroke admissions to Sandwell and West Birmingham Hospitals, over a 9-year period, we reported that compared to other ethnic groups, South Asian patients were significantly younger on admission and poorer survival at 30 days.20 However, on multivariate survival analysis, diabetes rather than ethnicity was an independent predictor of mortality.21

In conclusion, the susceptibility to diabetes mellitus amongst South Asians promotes an adverse CHD risk, even in patients treated for high blood pressure and cardiovascular risk. Further investigation is needed to assess whether lower thresholds for CHD risk management are needed in this population.


We are grateful for the financial support of Sandwell Medical Research Unit and the Haemostasis Thrombosis and Vascular Biology Unit by the Sandwell and West Birmingham Hospitals NHS Trust research and development programme.

Conflict of interest: None declared.


View Abstract