Ethnic inequalities in cardiovascular disease:

incidence, prognosis, and health care use

Louise van Oeffelen

Ethnic inequalities in cardiovascular disease: incidence, prognosis, and health care use PhD thesis, Utrecht University, The

ISBN: 978-90-393-6137-5 Author: Louise van Oeffelen Cover illustration: Olga Oliynik, www.nl123rf.com Cover design: Wendy Schoneveld, www.wenziD.nl Lay-out: Louise van Oeffelen Printed by: Wöhrmann Print Service B.V.

Ethnic inequalities in cardiovascular disease: incidence, prognosis, and health care use

Etnische verschillen in hart- en vaatziekten: incidentie, prognose en gebruik van gezondheidszorg (met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof.dr. G.J. van der Zwaan, ingevolge het besluit van het college door promoties in het openbaar te verdedigen op dinsdag 3 juni 2014 des middags te 4.15 uur

door

Aloysia Adriana Maria van Oeffelen

geboren op 31 juli 1984 te Terneuzen

Promotoren: Prof.dr. M.L. Bots Prof.dr. K. Stronks

Copromotoren: Dr. I. Vaartjes Dr. C.O. Agyemang

The research described in this thesis was supported by a grant of the Dutch Heart Foundation (grant number: 2010B296). Financial support by the Dutch Heart Foundation for the publication of this thesis is gratefully acknowledged.

Additional financial support by the Academic Medical Center Amsterdam, ChipSoft B.V., and Servier Nederland Farma B.V. for the publication of this thesis is also gratefully acknowledged. TABLE OF CONTENTS

Chapter 1 General introduction 7

Chapter 2 Inequalities in incidence of acute myocardial infarction 17 2.1 Ethnic inequalities in acute myocardial infarction incidence in first 19 and second generation ethnic minority groups 2.2 Sex disparities in acute myocardial infarction incidence by ethnic 39 group 2.3 Time trends in acute myocardial infarction incidence by country of 55 birth 2.4 Incidence of first acute myocardial infarction over time by age, sex, 69 and country of birth 2.5 Socioeconomic inequalities in acute myocardial infarction incidence 87 by ethnic group

Chapter 3 Inequalities in incidence of stroke 105 3.1 Ethnic inequalities in stroke incidence by stroke subtype and sex 107 3.2 Socioeconomic inequalities in stroke incidence by country of birth 123

Chapter 4 Inequalities in prognosis after cardiovascular disease 139 4.1 Ethnic inequalities in prognosis after acute myocardial infarction 141 and congestive heart failure 4.2 Socioeconomic inequalities in short-term mortality after acute 157 myocardial infarction

Chapter 5 Inequalities in cardiovascular health care use 173 5.1 Ethnic inequalities in cardiovascular drug dispense and quitting 175 rates in primary care 5.2 Ethnic inequalities in revascularisation procedure rate after an ST- 193 elevation myocardial infarction 5.3 Ethnic inequalities in cardiovascular drug dispense and quitting 205 rates after a first acute myocardial infarction

Chapter 6 General discussion 215

Chapter 7 Summary 237 Samenvatting 241 Dankwoord 246 Curriculum vitae 248 List of publications 249

CHAPTER 1

GENERAL INTRODUCTION

General introduction

BACKGROUND Although cardiovascular disease (CVD) has declined markedly over the past 30 years, it is still the main contributor to morbidity and mortality. In 2012, 20-22% of all European deaths were caused by coronary heart disease (CHD) and 10-15% by stroke. Other CVDs encompassed 12-15% of all deaths.1 CVD was also the largest single contributor to Disability Adjusted Life Years (DALYs) lost during 2012; 23% of all DALYs lost was due to CVD, mainly CHD (14%) and stroke (9%). CVD places therefore a huge burden on both patients and health care expenditure. Overall, CVD is estimated to cost the EU economy nearly 200 billion a year, of which 54% is due to direct health care costs, 24% to productivity loss, and 22% to informal care of people with CVD.1 Over the last decades it has become clearer that incidence and prognosis of cardiovascular disease differs between specific populations within a country.2 The growing number of multi-ethnic populations has led to the introduction of ethnic health research. Especially in the USA, where African minorities comprise a substantial part of the population for centuries, ethnic inequalities in cardiovascular disease incidence, prognosis, and health care use have been widely investigated. Higher acute myocardial infarction (AMI) incidence and mortality rates among African-American minorities than among their White American counterparts have been consistently reported. Some underlying factors described are a higher prevalence of traditional risk factors (hypertension, diabetes, hypercholesterolemia), a lower cardiovascular drug adherence, and a lower likelihood to receive effective and timely treatment in the acute care setting.3 In , evidence on ethnic inequalities in CVD is less extensive. Difficulties in including ethnic minorities in large cohort studies may underlie the limited evidence so far.4 The majority of European research available originates from the UK where especially South-Asian minorities have been investigated. Compared with the UK majority population, South-Asian minorities are at higher risk for ischemic heart disease (IHD). Found explanations include a higher prevalence of the insulin resistance syndrome (characterised by hyperinsulinaemia, hyperglycaemia, diabetes, dyslipidaemia, elevated triglyceride, and reduced high density lipoprotein level), a more severe type of disease (higher prevalence of triple vessel disease, several lesions, and non-discrete lesions), and a lower likelihood of receiving (timely) revascularisation procedures.5 Also Scandinavian countries and the Netherlands have published some important findings regarding ethnic inequalities in CVD, although they mainly focused on differences in cardiovascular risk factors, and less often on incidence, prognosis, and explanatory factors such as health care use.6-22 Furthermore, some major ethnic minority groups in Europe, such as Moroccans, Turkish, and Chinese, remain significantly underexposed in ethnic health research. In this thesis we will therefore investigate ethnic inequalities in CVD incidence, prognosis, and health care use in a wide range of ethnic minority groups.

9

Chapter 1

ETHNIC MINORITY GROUPS IN EUROPE AND THE NETHERLANDS

During the last few decades there has been a steady rise in migration flows towards Europe. Since the late 1980s the number of first generation ethnic minority groups (henceforth, migrants) living in European countries has doubled.23 In 2010, more than 47 million migrants lived in the 27 European Union (EU) member states, equivalent to 9.4% of the total population.24 Two of the largest migrant groups are from Turkish and Moroccan origin. The biggest new-flow of migrants originate from Romania. Other increasing migration flows since 2001 are from Polish and Chinese descent.23 The Netherlands is an increasingly ethnically diverse country. In 2012, the Dutch population consisted of 11.2% migrants, which is almost 2% higher than in the total EU27 population. Furthermore, 9.8% of the population was born in the Netherlands with at least one parent born abroad (second generation ethnic minorities). In total, 3.5 million Dutch citizens (21% of the total Dutch population) belonged to a first or second generation ethnic minority group in 2012.25 This percentage doubled since 1972, when only 9% of the Dutch population belonged to an ethnic minority group.26 Of the ethnic minorities, 9% is of Western origin and 12% of non-Western origin (mainly from Turkey, Morocco, , and the Netherlands Antilles).25 In the upcoming decennia, the ethnic minority population is expected to grow even further. Prognostic analyses show an increase from 21% in 2012 towards 31% in 2060. The growth will be most profound among non- Western ethnic minorities; from 12% in 2012 towards 18% in 2060. Currently, the age distribution among non-Western ethnic minority groups is relatively low; only 4% of them are ≥65 years of age. Prognostic studies estimate that in 2060 22% of all non-Western ethnic minority groups are ≥65 years of age, which will be only slightly less than within the ethnic Dutch population.27 Since cardiovascular disease is a condition of the elderly, reducing ethnic inequalities in cardiovascular disease will become even more important in the future.

OBJECTIVES OF THIS THESIS

The high and rising number of ethnic minority groups in European countries in addition to the limited evidence regarding ethnic inequalities in CVD incidence, prognosis and health care use in Europe has resulted in the following research objectives: 1. To assess inequalities in the incidence of non-fatal and fatal cardiovascular disease between ethnic minority groups and ethnic Dutch. 2. To assess inequalities in mortality and readmission rate after a first cardiovascular event between ethnic minority groups and ethnic Dutch. 3. To assess inequalities in cardiovascular health care use (drugs, procedures) between ethnic minority groups and ethnic Dutch.

10

General introduction

DEFINITION OF ETHNIC BACKGROUND: THIS THESIS

In this thesis ethnic background is defined by using the country of birth and parental country of birth, as described previously.28 A first generation ethnic minority is born abroad with at least one of the parents born abroad, whereas a second generation ethnic minority is born in the Netherlands with at least one of the parents born abroad. Those with both parents born in the Netherlands are indicated as ethnic Dutch throughout this thesis. In all studies, ethnic minorities from Surinamese, Turkish, and Moroccan origin are incorporated, since these are the three largest non-Western ethnic minority groups living in the Netherlands. Suriname is a previous Dutch colony located in South America. The Surinamese population is ethnically diverse, and consists of people originating from West Africa (30%), India (37%), Java (15%), China (1.5%), and people of mixed origin.29 In some studies in this thesis, surname analyses were used to disaggregate the Hindustani Surinamese (those with a South Asian background) from the non-Hindustani Surinamese. Whenever possible, also ethnic minority groups originating from the Netherlands Antilles (a previous Dutch colony located in the Caribbean) and Indonesia (a previous Dutch colony located in East Asia) were included, since they are the fourth and fifth largest ethnic minority groups in the Netherlands. Furthermore, if numbers allowed, Chinese and South-Asian minority groups were included. The South-Asian group incorporates those originating from India, Sri Lanka, and Pakistan. This thesis mainly focuses on non-Western ethnic minority groups, although few studies also incorporate some major Western ethnic minority groups. Several studies in this thesis combine first and second generation ethnic minority groups, but most of them only include first generation ethnic minorities. Figure 1 presents the number of persons belonging to the ethnic minority groups studied in this thesis in the Netherlands on 1 January 2012, stratified by generational status.25

300000 250000 First generation 200000 Second generation 150000 100000 50000 0 Number of persons on January 1st 2012

Country of origin

Figure 1 Number of 1st and 2nd generation ethnic minority groups by country of origin in the Netherland in 2012

11

Chapter 1

LINKGAGE OF DUTCH REGISTERS

To investigate ethnic inequalities in CVD incidence and prognosis (the first and second objective of this thesis), several nationwide Dutch registers were linked. Linkage of nationwide registers is a promising and powerful tool to study health and disease among ethnic minorities.30;31 They are an inexpensive and easily accessible source of information, giving the opportunity to study individual ethnic minority groups and even subgroups with sufficient power. The studies presented in this thesis were conducted by linking the Hospital Discharge Register, Population Register, Cause of Death Register, and the Regional Income Survey which were available from 1995 until 2007/2010. The registers were used to obtain information regarding cardiovascular events, comorbidity, demographic factors and income (used as proxy for socioeconomic status). Methods for linkage of these nationwide registers have been described previously.32 To investigate ethnic inequalities in cardiovascular health care use (the third objective of this thesis), data from the Achmea Health Database between 2006 and 2011 were used. The Achmea Health Database records payments for the provision of all medical care to patients insured with the Achmea health insurance company (a major health insurance company in the Netherlands).

OUTLINE OF THIS THESIS

In Chapter 2, inequalities in the incidence of acute myocardial infarction (AMI) between ethnic minority groups and ethnic Dutch, stratified for several factors, are studied. Chapter 2.1 reports on the overall difference in AMI incidence between a wide range of ethnic minority groups and ethnic Dutch. Furthermore, if numbers allow, results are presented for first generation and second generation ethnic minority groups separately. In Chapter 2.2 the difference in AMI incidence between men and women is investigated among several ethnic minority groups and ethnic Dutch. Chapter 2.3 examines trends in acute myocardial infarction incidence over the time period 1998- 2007 among ethnic minority groups and ethnic Dutch, and gives insight into ethnic differences in these trends. Chapter 2.4 shows absolute incidence rates of acute myocardial infarction among ethnic minority groups and ethnic Dutch during the time periods 2000-2004 and 2005-2010, specific for age and sex. In Chapter 2.5 socioeconomic inequalities in AMI incidence among ethnic minority groups and ethnic Dutch are investigated. In Chapter 3, inequalities in the incidence of stroke between ethnic minority groups and ethnic Dutch are studied. Chapter 3.1 reports on ethnic differences in the incidence of total stroke and subtypes of stroke (intracerebral haemorrhage, subarachnoid haemorrhage and ischemic stroke). In Chapter 3.2 socioeconomic inequalities in total stroke incidence among ethnic minority groups and ethnic Dutch are investigated. In Chapter 4, ethnic and socioeconomic inequalities in prognosis after a first cardiovascular event are studied. Chapter 4.1 explores ethnic inequalities in short- and long-term prognosis

12

General introduction

(mortality and readmission) after a first hospitalisation for AMI and congestive heart failure. Chapter 4.2 reports on socioeconomic inequalities in short-term mortality (pre-hospital mortality and case- fatality) after a first AMI event. In Chapter 5, inequalities in cardiovascular health care use between ethnic minority groups and ethnic Dutch are studied. In Chapter 5.1, ethnic inequalities in cardiovascular drug dispense and quitting rates are investigated in the primary care setting. Chapter 5.2 reports on ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction. Chapter 5.3 describes ethnic inequalities in cardiovascular drug dispense and quitting rates after a first AMI event. In Chapter 6, the general discussion, the main findings of this thesis are discussed. Furthermore, implications and recommendations for future research and practice are considered. Finally, a summary in English and Dutch is presented.

13

Chapter 1

REFERENCES 1. Nichols M, Towsed N, Luengo-Fernandez R, Leal J, Loray A, Scarborough P, Rayner M. European Cardiovascular Disease Statistics 2012. European Heart Network, Brussels, European Society of Cardiology, Sophia Antipolis. 2013. 2. Yancy CW, Benjamin EJ, Fabunmi RP, Bonow RO. Discovering the full spectrum of cardiovascular disease: Minority Health Summit 2003: executive summary. Circulation 2005;111(10):1339-49. 3. Mody P, Gupta A, Bikdeli B, Lampropulos JF, Dharmarajan K. Most important articles on cardiovascular disease among racial and ethnic minorities. Circ Cardiovasc Qual Outcomes 2012;5(4):e33-e41. 4. Sue S, Dhindsa MK. Ethnic and racial health disparities research: issues and problems. Health Educ Behav 2006;33(4):459-69. 5. Chaturvedi N. Ethnic differences in cardiovascular disease. Heart 2003;89(6):681-6. 6. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7. 7. Agyemang C, Ujcic-Voortman J, Uitenbroek D, Foets M, Droomers M. Prevalence and management of hypertension among Turkish, Moroccan and native Dutch ethnic groups in Amsterdam, the Netherlands: The Amsterdam Health Monitor Survey. J Hypertens 2006;24(11):2169-76. 8. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 9. Agyemang C, Nicolaou M, Boateng L, Dijkshoorn H, van de Born BJ, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Ghanaian population in Amsterdam, the Netherlands: the GHAIA study. Eur J Prev Cardiol 2013;20(6):938-46. 10. Borne Y, Engstrom G, Essen B, Sundquist J, Hedblad B. Country of birth and risk of hospitalization due to heart failure: a Swedish population-based cohort study. Eur J Epidemiol 2011;26(4):275-83. 11. Gadd M, Johansson SE, Sundquist J, Wandell P. Morbidity in cardiovascular diseases in immigrants in Sweden. J Intern Med 2003;254(3):236-43. 12. Gadd M, Sundquist J, Johansson SE, Wandell P. Do immigrants have an increased prevalence of unhealthy behaviours and risk factors for coronary heart disease? Eur J Cardiovasc Prev Rehabil 2005;12(6):535-41. 13. Hedlund E, Lange A, Hammar N. Acute myocardial infarction incidence in immigrants to Sweden. Country of birth, time since immigration, and time trends over 20 years. Eur J Epidemiol 2007;22(8):493-503. 14. Hempler NF, Diderichsen F, Larsen FB, Ladelund S, Jorgensen T. Do immigrants from Turkey, Pakistan and Yugoslavia receive adequate medical treatment with beta-blockers and statins after acute myocardial infarction compared with Danish-born residents? A register-based follow-up study. Eur J Clin Pharmacol 2010;66(7):735-42. 15. Hempler NF, Larsen FB, Nielsen SS, Diderichsen F, Andreasen AH, Jorgensen T. A registry-based follow- up study, comparing the incidence of cardiovascular disease in native Danes and immigrants born in Turkey, Pakistan and the former Yugoslavia: do social inequalities play a role? BMC Public Health 2011;11:662. 16. Khan FA, Zia E, Janzon L, Engstrom G. Incidence of stroke and stroke subtypes in Malmo, Sweden, 1990-2000: marked differences between groups defined by birth country. Stroke 2004;35(9):2054-8. 17. Ringback WG, Ericsson O, Lofroth E, Rosen M. Equal access to treatment? Population-based follow-up of drugs dispensed to patients after acute myocardial infarction in Sweden. Eur J Clin Pharmacol 2008;64(4):417-24.

14

General introduction

18. Sundquist K, Li X. Coronary heart disease risks in first- and second-generation immigrants in Sweden: a follow-up study. J Intern Med 2006;259(4):418-27. 19. Ujcic-Voortman JK, Schram MT, Jacobs-van der Bruggen MA, Verhoeff AP, Baan CA. Diabetes prevalence and risk factors among ethnic minorities. Eur J Public Health 2009;19(5):511-5. 20. Ujcic-Voortman JK, Bos G, Baan CA, Uitenbroek DG, Verhoeff AP, Seidell JC. Ethnic differences in total and HDL cholesterol among Turkish, Moroccan and Dutch ethnic groups living in Amsterdam, the Netherlands. BMC Public Health 2010;10:740. 21. Ujcic-Voortman JK, Baan CA, Verhoeff AP, Krol A, Seidell JC. Ethnic differences in systemic inflammation: an investigation of C-reactive protein levels among Moroccan, Turkish and Dutch groups in the Netherlands. Atherosclerosis 2011;218(2):511-6. 22. Ujcic-Voortman JK, Bos G, Baan CA, Verhoeff AP, Seidell JC. Obesity and body fat distribution: ethnic differences and the role of socio-economic status. Obes Facts 2011;4(1):53-60. 23. Salt J. Trends in Europe's international migration. In: Rechel B, Mladovsky P, Devillé W, Rijks B, Petrova-Benedict R, McKee M, editors. Migration and health in the European Union. 1 ed. England: Open University Press; 2012. p. 23. 24. Vasileva K. 6.4% of the EU population are foreigners and 9.4% are born abroad. Population and Social Conditions, Statistics in Focus 34/2011. Luxembourg: Eurostat, European Commission. 2011. 25. Sanderse C, Verweij A, de Beer J. Etniciteit: Wat is de huidige situatie? In: Volksgezondheid Toekomst Verkenning, Nationaal Kompas Volksgezondheid. Bilthoven: RIVM, 2012. 26. Verweij A, Sanderse C, de Beer J. Etniciteit: Wat waren de belangrijkste ontwikkelingen in het verleden? In: Volksgezondheid Toekomst Verkenning, Nationaal Kompas Volksgezondheid. Bilthoven: RIVM, 2012. 27. Sanderse C, Verweij A, de Beer J. Etniciteit: Wat zijn de verwachtingen voor de toekomst? In: Volksgezondheid Toekomst Verkenning, Nationaal Kompas Volksgezondheid. Bilthoven: RIVM, 2012. 28. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69. 29. Bos V, Kunst AE, Keij-Deerenberg IM, Garssen J, Mackenbach JP. Ethnic inequalities in age- and cause- specific mortality in The Netherlands. Int J Epidemiol 2004;33(5):1112-9. 30. Agyemang C, Bhopal R. Hypertension and cardiovascular disease endpoints by ethnic group: the promise of data linkage. Heart 2013;99(10):675-6. 31. Rafnsson SB, Bhopal RS. Large-scale epidemiological data on cardiovascular diseases and diabetes in migrant and ethnic minority groups in Europe. Eur J Public Health 2009;19(5):484-91. 32. Reitsma JB, Kardaun JW, Gevers E, de Bruin A, van der Wal J, Bonsel GJ. [Possibilities for anonymous follow-up studies of patients in Dutch national medical registrations using the Municipal Population Register: a pilot study]. Ned Tijdschr Geneeskd 2003;147(46):2286-90.

15

CHAPTER 2

INEQUALITIES IN INCIDENCE OF ACUTE MYOCARDIAL INFARCTION

CHAPTER 2.1

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups

van Oeffelen AAM, Vaartjes I, Stronks K, Bots ML, Agyemang C. Incidence of acute myocardial infarction in first and second generation minority groups: does the second generation converge towards the majority population?

Int J Cardiol 2013;168(6):5422-9. Chapter 2.1

ABSTRACT Background Differences in acute myocardial infarction (AMI) incidence between ethnic minority groups and the majority population have been reported. Health differences may converge towards the majority population over generations. We assessed whether AMI incidence differences between ethnic minority groups living in the Netherlands and the ethnic Dutch population exist, and whether the incidence converges towards ethnic Dutch over generations.

Methods A nationwide register-based cohort study was conducted from 1997-2007. Using Cox proportional hazard models, AMI incidence differences between ethnic minorities and ethnic Dutch were estimated. When possible, analyses were stratified by generation.

Results AMI incidence differences between ethnic minorities and ethnic Dutch depended on the country of origin, and often varied between minorities originating from the same geographical region. For example, among North-African and Mediterranean minorities, incidence was higher in Turkish (Hazard Ratio (HR): 1.34; 95% Confidence Interval (95% CI): 1.28-1.41), but lower in Moroccans (HR: 0.46; 95% CI: 0.40-0.52) compared with ethnic Dutch. Most ethnic minorities had a similar or lower incidence than ethnic Dutch, which remained similar or converged towards the incidence of ethnic Dutch over generations. In contrast, among minorities from the former Dutch colonies (Suriname, Indonesia, and Netherlands Antilles) beneficial intergenerational changes were observed.

Conclusion Health care professionals and policy makers should be aware of substantial AMI incidence differences between ethnic minority groups and the majority population, and the often unbeneficial change over generations. Future research should be cautious when clustering ethnic minority groups based on geographical region of the country of origin.

20

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups

INTRODUCTION

The Netherlands is an ethnically diverse country; ten to 20 percent of the population is of foreign origin. Of all ethnic minority groups residing in the Netherlands more than half originates from non- Western countries, mainly from Turkey, Morocco, Suriname, Indonesia and the Netherlands Antilles. Of the Western minorities, about one third originates from countries surrounding the Netherlands (Germany, Belgium, and the United Kingdom). In addition, there is a variety of smaller groups from all over the world. Previous research in several Western countries reported differences in coronary heart disease (CHD) between ethnic minority groups and the majority population.1;2 However, evidence is scarce and often related to mortality instead of incidence.3 Studies regarding incidence differences mainly reported a higher CHD incidence in ethnic minority groups compared with the majority population.4-9 Factors that cohere with immigration (stress, poverty, low socioeconomic status (SES), language barriers), preservation of an unfavourable risk factor profile, and genetics have been suggested as potential factors that may underlie this higher incidence.10 Yet, in some ethnic minority groups CHD incidence was lower.6;8 The ‘healthy migrant effect’, characterised by superior health and financial status of migrant populations relative to populations that stay behind, is often seen as the underlying explanatory factor.11 It has been suggested that health differences between ethnic minority groups and the majority population might be more profound in those who migrated than in their offspring due to acculturation towards the majority population.12 Factors that coincide with immigration and the healthy migrant effect diminish over time and generations. To our knowledge, only one study investigated this trend over generations with respect to CHD incidence. Results showed that the higher incidence in first generation ethnic minorities converged towards the incidence of the majority population in the second generation, but in women only.8 Unfortunately, analyses were limited to European ethnic minorities and, because of small numbers, minority groups from different countries of origin were merged which complicates interpretation of results. In the present follow-up study, nationwide registers were used to determine the incidence of acute myocardial infarction (AMI) in the ethnic Dutch population and in a wide range of ethnic minority groups living in the Netherlands from all over the world. The first aim was to investigate AMI incidence differences between ethnic minority groups and ethnic Dutch. The second aim was to investigate whether incidence differences seen in first generation ethnic minorities converge towards ethnic Dutch over generations.

21

Chapter 2.1

METHODS

Cohort enrolment A nationwide register-based cohort study was conducted. The Population Register (PR), Hospital Discharge Register (HDR), Cause of Death Register (CDR), and Regional Income Survey (RIS) were used to obtain information regarding demographic factors, AMI hospitalisations, fatal AMI events, and comorbidities. The registers are described in detail previously.13 The overall quality of Dutch national registers proved to be high.14 By linking previous registers with a personal identifier a cohort was built, starting at 1 January 1997. On 1 January 1997, 15,045,392 Dutch citizens (16.4% ethnic minorities) with no previous hospital admission for AMI (ICD-9-code 410) or old AMI (ICD-9 code 412) in 1995 and 1996 were registered in the PR. Only persons registered during the whole period between 1 January 1995 and 1 January 1997 were included to ensure a minimal residing period in the Netherlands of two years, and to enable taking medical history data into account for every person under study. Due to the absence of a personal identifier in the HDR, the PR and HDR could only be linked via the combination of date of birth, sex and four digits of the postal code as identifying key. In case of multiple persons with an identical identifying key (non-uniqueness), PR and HDR could not be validly linked. Persons who were not present or not unique in the PR between 1 January 1995 and 1 January 1997 were excluded (3,071,969 persons, 24.2% ethnic minorities). As interest is in AMI, persons younger than 30 years of age were excluded (4,371,638 persons, 17.3% ethnic minorities). The final cohort comprised 7,601,785 persons.

Follow-up From 1 January 1997 persons were followed until their first AMI event, comprising a hospital admission with AMI as primary or secondary diagnosis (ICD-9 code 410), or an out-of-hospital death with AMI as primary or secondary cause (ICD-10 code I21). The validity of these ICD codes proved to be good.15 Persons were censored in case of death, non-uniqueness, emigration, or the end of the study period at 31 December 2007, whichever came first.

Determinants Ethnic background Ethnic minority groups were constructed based on the country of birth of the resident and his/her parents, according to the definition of Statistics Netherlands.16 A person is considered an ethnic minority if he/she was born abroad and at least one of the parents was born abroad (first generation ethnic minority), or if he/she was born in the Netherlands with at least one of the parents born abroad (second generation ethnic minority). A person was considered ethnic Dutch when both parents were born in the Netherlands.

22

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups

Neighbourhood socioeconomic status Neighbourhood socioeconomic status (SES) was based on income data registered in the RIS.17 The RIS started in 1994, when a representative sample of 1.9 million Dutch residents was selected. Every year, the sample was corrected for emigration and mortality on one hand, and immigration and birth on the other hand. All residents belonging to the households of the sample population (about 5 million residents) were included in the RIS. Mean disposable income of the residents with income data available in each neighborhood was calculated for 1997, and assigned to all persons living in that neighborhood on 1 January 1997. Neighbourhood income was divided in tertiles, with the first tertile representing the lowest income group.

Comorbidity Presence and extent of comorbidity were determined with the Charlson index score based on previous hospital admissions,18 which proved to be a reliable and valid method to measure comorbidity in clinical research.19 The Charlson index score ranges from zero to six (cut-off value), with zero representing no comorbidity.

Data analysis In order to perform the analyses with sufficient power, only the ethnic Dutch and ethnic minority groups with at least ten events were included (n=7,570,510). In ethnic minority groups with at least ten events per generation, analyses were stratified by generation (n=759,576). Baseline characteristics (age, sex, generation, neighbourhood SES, Charlson comorbidity index) were analysed on 1 January 1997 in ethnic Dutch, in ethnic minorities in total, and in first and second generation ethnic minorities separately. Using Cox proportional hazard regression analyses, adjusted for age at baseline and sex, AMI incidence differences between ethnic minority groups and ethnic Dutch (reference) were investigated. Additionally, adjustments were made for neighbourhood SES and comorbidity. Using the same procedure, but now stratified by generation, the AMI incidence differences with ethnic Dutch were investigated for first and second generation ethnic minority groups separately. To determine whether the change in AMI incidence over generations was statistically significant, a Cox proportional hazard model was built comparing second generation ethnic minorities with first generation ethnic minorities (reference). When the confidence interval did not incorporate one, the intergenerational change was considered statistically significant. Three additional analyses were performed. First, because of possible sex differences in the relations under study, analyses were stratified by sex (only in ethnic minority groups with ≥10 persons in both sexes). Second, to get insight into potential selection bias due to the exclusion of non-unique persons, the relation between country of origin and uniqueness and between AMI mortality and uniqueness were addressed using logistic regression analyses. Third, as in ethnic

23

Chapter 2.1 minorities’ elderly are less well represented, adjustment for age only may not be enough to remove age effects. A lower AMI incidence in ethnic minority groups compared with ethnic Dutch may be provoked by their young age distribution. In ethnic minority groups with a lower incidence compared with ethnic Dutch, analyses were stratified by age (<55 years, ≥55 years) to explore whether relations remained. Log-minus-log plots showed no violation of the proportional hazards assumption. Results were expressed as Hazard Ratios (HR) with accompanying 95% confidence intervals (95% CI).We used SPSS software, version 14.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation Netherlands.

RESULTS

Baseline characteristics After exclusion of non-unique persons, the final cohort comprised 7,570,510 unique persons, of which 944,280 (12.5%) ethnic minorities (Table 1). Non-uniqueness was related to both ethnic minority status (ethnic minorities were less often unique) and AMI mortality (those who died from AMI were more often unique). However, after correcting for age and sex, being unique was not related to AMI mortality anymore (Odds Ratio (OR): 1.01; 95% CI: 0.98-1.04). Overall, ethnic minorities were younger, had a lower neighbourhood SES, and less comorbidities than ethnic Dutch. In the analysis stratified by generation 7,385,806 persons were included, of which 759,576 (10.3%) ethnic minorities (375,168 first generation and 384,408 second generation). First generation ethnic minorities encompassed fewer men and had a lower neighbourhood SES than ethnic Dutch, whereas their age and Charlson comorbidity index were similar. Second generation ethnic minorities were younger and had a lower Charlson comorbidity index than ethnic Dutch, whereas their sex distribution and neighbourhood SES were similar.

Incidence differences between ethnic minority groups and ethnic Dutch AMI incidence differed between ethnic minority groups and ethnic Dutch, with HRs ranging from 0.40 (95% CI: 0.22-0.74) in minorities from the Philippines to 2.04 (95% CI: 1.62-2.55) in Pakistani minorities (Figure 1). Most ethnic minority groups had a similar or lower AMI incidence than ethnic Dutch. Results shown are adjusted for age and sex; additional adjustment for neighbourhood SES and comorbidity index did not change the findings. Differences in AMI incidence compared with ethnic Dutch often varied between ethnic minority groups originating from the same geographical region (Table 2). Differences were mixed among minority groups from Northwest Europe, South Europe, East Europe, North Africa and the Mediterranean, Latin America and the Caribbean, and Sub-Saharan Africa. Among minorities from East Asia, the AMI incidence was predominantly lower, except for Indonesians (HR: 1.03; 95% CI: 1.01-1.06). Among North-American, South-Asian, and

24

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups

West-Asian minority groups there was a consistent pattern in incidence differences; North Americans had a lower incidence, South Asians had a higher incidence and West Asians had a similar incidence compared with ethnic Dutch. Additional analyses showed no differences in these relations between men and women, except among Turkish minorities (men HR: 1.43; 95% CI: 1.35-1.51, women HR: 1.01; 95% CI: 0.90-1.13) (Supplementary table 1). In age stratified analyses among ethnic minority groups with a lower AMI incidence compared with ethnic Dutch, the lower incidence remained in those below and above 55 years of age (Supplementary table 2).

Table 1 Baseline characteristics of ethnic minority groups and ethnic Dutch in the Netherlands, 1997-2007 Ethnic Dutch Ethnic 1st generation 2nd generation minorities minorities minorities Persons 6,626,230 944,280 375,168 384,408 Person-years at risk 56,012,604 7,726,074 2,973,570 3,280,809 AMI events 241,074 28,980 13,360 12,292 Incidence ratea 244 238 241 241 Mean age in years (sd) 53.1 (15.6) 50.1 (14.2) 53.0 (15.3) 50.3 (13.5) Men % 48.0 47.5 42.5 49.2 First generation % - 58.7 - - Neighbourhood SES % Tertile 1 (lowest income) 30.1 41.8 42.4 31.8 Tertile 2 (medium income) 34.4 29.4 27.9 34.2 Tertile 3 (highest income) 35.5 28.7 29.6 34.0 Mean Charlson index score (sd) 0.7 (1.4) 0.6 (1.3) 0.6 (1.4) 0.6 (1.3) Charlson index score > 0 % 27.3 23.4 26.0 24.7 a Incidence rate per 100,000 person-years at risk standardised to the age distribution of the European population in ten-year age bands sd: standard deviation; SES: socioeconomic status

Incidence differences between first and second generation ethnic minorities and ethnic Dutch Seventeen ethnic minority groups were included for analyses stratified by generation (Table 3). Among all European minority groups, except for Italians, the difference in AMI incidence with ethnic Dutch did not change or changed unbeneficial over generations (Figure 2). (Borderline) statistically significant intergenerational changes were found among minorities from Switzerland, where the lower incidence in the first generation converged towards the incidence of ethnic Dutch in the second generation; and among minorities from Belgium, Germany, and Poland, where the similar or higher incidence in the first generation exceeded that of ethnic Dutch in the second generation. Minorities from South-American countries (Netherlands Antilles, Suriname) experienced a beneficial change over generations, which was statistically significant among the Surinamese. Among East- Asian minorities, Chinese turned from a lower incidence in the first generation to a higher incidence in the second generation; among Indonesians the opposite occurred. There were no differences in these intergenerational changes between men and women (Supplementary table 1).

25

Chapter 2.1

Figure 1 Relative difference in AMI incidence (HR (95% CI)) between ethnic minority groups and ethnic Dutch (reference) in the Netherlands 1997-2007, adjusted for age and sex

26

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups

Table 2 AMI incidence differences (HR (95% CI)) between ethnic minority groups and ethnic Dutch (reference) in the Netherlands, 1997-2007, adjusted for age and sex Geographical region Country of origin Populationa AMIb HR (95% CI) Northwest Europe Switzerland 3,117 70 0.77 (0.61-0.98)* Scandinavia 5,253 88 0.78 (0.63-0.96)* Ireland 1,822 21 0.78 (0.51-1.20) Great Britain 22,997 423 0.80 (0.73-0.88)* Austria 8,434 271 0.92 (0.82-1.04) France 9,377 210 0.92 (0.80-1.05) Belgium 60,620 2,551 1.07 (1.03-1.12)* Germany 266,847 11,032 1.10 (1.08-1.13)* South Europe Portugal 4,506 49 0.44 (0.33-0.58)* Spain 9,969 207 0.85 (0.75-0.98)* Italy 12,056 343 0.95 (0.86-1.06) Greece 3,316 99 1.18 (0.97-1.44) East Europe Romania 1,671 21 0.56 (0.37-0.86)* Czechoslovakia 3,398 81 0.76 (0.61-0.95)* Soviet Union 4,915 121 1.05 (0.88-1.26) Poland 12,059 349 1.07 (0.96-1.18) Yugoslavia 17,095 346 1.09 (0.98-1.21) Hungary 6,520 243 1.12 (0.99-1.28) North Africa and Morocco 30,874 238 0.46 (0.40-0.52)* Mediterranean Tunisia 1,691 24 0.85 (0.57-1.26) Algeria 1,309 18 0.90 (0.57-1.43) Turkey 66,241 1,631 1.34 (1.28-1.41)* Egypt 3,632 77 1.50 (1.20-1.88)* North America USA 6,437 95 0.65 (0.52-0.79)* Canada 1,479 16 0.67 (0.41-1.10) Latin America and Colombia 1,246 <10 0.24 (0.08-0.75)* the Caribbean Chile 1,135 <10 0.42 (0.21-0.84)* Argentina 1,030 20 0.76 (0.49-1.17) Dominican Republic 1,182 11 0.92 (0.51-1.67) Netherlands Antilles 23,396 361 0.95 (0.85-1.05) 1,561 21 0.97 (0.63-1.48) British Guiana 1,099 28 1.32 (0.91-1.90) Suriname 91,799 2,601 1.41 (1.35-1.46)* Sub-Saharan Africa Ghana 1,958 <10 0.34 (0.13-0.90)* Cape Verde 5,354 55 0.54 (0.41-0.70)* Ethiopia 1,783 <10 0.55 (0.29-1.04) South Africa 2,290 56 0.89 (0.69-1.16) Somalia 1,242 <10 1.24 (0.59-2.61) East Asia Japan 1,012 <10 0.34 (0.14-0.80)* Philippines 2,316 10 0.40 (0.22-0.74)* Hong Kong 5,741 49 0.44 (0.33-0.58)* China 8,386 122 0.57 (0.48-0.68)* Singapore 1,216 12 0.61 (0.35-1.07) Vietnam 3,769 39 0.64 (0.47-0.88)* Thailand 1,455 <10 0.67 (0.35-1.28) Indonesia 213,965 6,691 1.03 (1.01-1.06)* South Asia India 2,311 48 1.35 (1.01-1.79)* Sri Lanka 1,624 32 1.49 (1.06-2.11)* Pakistan 3,055 75 2.04 (1.62-2.55)* West Asia Iran 4,785 71 0.96 (0.76-1.21) Iraq 2,346 33 0.97 (0.69-1.36) Oceania Australia 1,420 21 0.91 (0.59-1.39) a Number of persons; b Number of persons with a first AMI event; *p<0.05

27

Chapter 2.1

d

generation generation 2007, adjusted adjusted 2007, - generation vs. generation st

HR (95% CI) HR 1 nd 1.58 (0.99;2.52) 1.05 (0.85;1.30) 1.15 (0.88;1.51) 1.10 (0.86;1.39) 1.09 (0.99;1.19) 0.75 (0.54;1.04) 1.44 (0.92;2.26) 1.23 (0.86;1.78) 0.98 (0.75;1.28) 1.28 (0.86;1.92) 0.67 (0.42;1.06) 1.01 (0.64-1.57) 1.01 (0.64-1.57) 1.08 (1.03;1.12)* 1.36 (1.10;1.68)* 0.62 (0.50;0.77)* 2.69 (1.71;4.23)* 0.75 (0.71;0.80)* 2

c 1.34) 1.21) 1.16) 1.01) 1.39) 1.61) 1.38) 1.02) 1.11) 2.04) 0.99)* 1.14)* 1.15)* 1.52)* 0.98)* 0.87)* ------

------

(0.90 HR (95% CI) HR 0.98 (0.71 0.78 (0.54-1.13) 0.99 (0.81 0.97 (0.82 0.75 (0.55 0.97 (0.68 1.20 (0.90 1.11 (0.89 0.65 (0.42 0.89 (0.72 1.36

0.83 (0.69 1.09 (1.05 1.13 (1.10 1.29 (1.10 0.74 (0.56 0.82 (0.78

b

37 28 95 41 30 46 79 47 19 84 23 113 128 144 AMI

1,962 8,246 1,170

a generation vs. ethnic vs. ethnic Dutch generation

nd 2 827 1,323 1,748 5,889 3,779 4,654 3,534 1,150 5,336 2,296 3,427 2,043 1,730 5,121

42,836 97,948 200,767 Population

c 1.03) 1.04) 1.09) 1.11) 1.09) 1.22) 1.32) 1.08) 0.88)* 0.88)* 1.09)* 0.89)* 0.76)* 1.49)* 0.62)* 1.12)* ------

HR (95% CI) HR 0.86 (0.72 0.89 (0.75 1.01 (0.93 0.99 (0.89 0.95 (0.83 0.98 (0.78 1.13 (0.97 0.97 (0.87 0.62 (0.44 0.77 (0.60-0.99)* 0.79 (0.71 1.05 (1.01 0.68 (0.51 0.58 (0.43 1.44 (1.38 0.51 (0.42 1.09 (1.06

b

33 60 51 75 48 99 310 115 143 589 302 205 164 342 AMI 2,786 2,517 5,521

a

generation vs. ethnic Dutch ethnic vs. generation

st

1 3,505 3,505 5,598 3,780 8,522 2,248 6,723 2,619 3,093 4,394 7,559 1,794 1,794 17,108 17,784 66,080 21,666 86,678

116,017 Population between first and second generation ethnic minority groups and ethnic Dutch in the Netherlands, 1997 Netherlands, in the Dutch and ethnic groups minority ethnic second generation and first between

(HR (95% CI))

Country of origin of Country Switzerland Scandinavia Great Britain France Austria Belgium Germany Italy Czechoslovakia Poland Soviet Union Hungary USA Netherlands Antilles Suriname China Indonesia

events

AMI incidence difference

Geographical Geographical region Northwest Europe South Europe East Europe America North South America East Asia Reference: first generationethnic minorities Reference: ethnic Dutch Number of AMI AMI of Number Number of persons persons of Number

Table 3 Table agefor and sex a b c d *p<0.05 *p<0.05

28

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups

Figure 2 Relative difference in AMI incidence (HR (95% CI)) between first and second generation ethnic minority groups and ethnic Dutch (reference) in the Netherlands 1997-2007, adjusted for age and sex * p<0.05

29

Chapter 2.1

DISCUSSION

Main findings Differences in AMI incidence between ethnic minority groups living in the Netherlands and ethnic Dutch depended on the country of origin, and varied between ethnic minority groups originating from the same geographical region except among South-Asian, West-Asian, and North-American minority groups. Ethnic minority groups with a lower AMI incidence in the first generation remained similar or converged towards the AMI incidence of ethnic Dutch over generations. In ethnic minority groups with a similar or higher AMI incidence in the first generation, the intergenerational changes were more diverse; there were beneficial changes among minorities from the three former Dutch colonies (Netherlands Antilles, Suriname, and Indonesia) and Italy, and no or unbeneficial changes in the other ethnic minority groups.

Incidence differences between ethnic minority groups and ethnic Dutch Former European studies found CHD incidence differences between ethnic minority groups and the majority population.4-9 Isolated ethnic minority groups investigated in those studies showed a higher incidence in minorities from Poland, Turkey, Hungary, Pakistan and India, and a lower incidence in minorities from the USA and China compared with the majority population, which is in agreement with our results. Previous studies often clustered ethnic minority groups based on geographical region of the country of origin. We demonstrate considerable differences in AMI incidence between ethnic minority groups originating from countries within the same geographical region. Combining these groups may obscure results. The majority of ethnic minority groups had a lower AMI incidence compared with ethnic Dutch, most likely explained by the ‘healthy migrant effect’,11 which states that particularly the healthiest persons migrate to a foreign country. Age bias due to the young age distribution of these ethnic minority groups was ruled out as a potential explanation, since the AMI incidence was lower in younger as well as older ethnic minorities. Also adjustment for neighbourhood SES and comorbidity did not alter found incidence differences. This is in line with other studies investigating ethnic CHD incidence differences, where adjustment for several SES indicators did not markedly influence results.4-8

Northwest-European minorities Belgian and German minorities had a higher AMI incidence, whereas British, Scandinavian, and Swiss minorities had a lower incidence than ethnic Dutch. Explanations for these differences are unclear. However, Belgium and Germany are the immediate neighbouring countries with quite similar cultural traditions, which makes it relatively easy to migrate to the Netherlands. In contrast, minorities from Great Britain, Scandinavia, and Switzerland are more likely to be highly educated

30

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups persons in good health. A higher CHD incidence among minorities originating from countries surrounding the host country was also observed in previous studies.6;8

North-African and Mediterranean minorities Turkish minorities had a higher AMI incidence, whereas Moroccan minorities had a lower incidence compared with ethnic Dutch. The findings in Turkish minorities are in line with international literature.5-8 A higher prevalence of smoking, overweight, physical inactivity, high cholesterol, and diabetes in Turkish minorities compared with ethnic Dutch may underlie the higher incidence.20;21 After stratifying by sex, results only remained in men, which could be due to the considerably higher smoking prevalence in Turkish men than in Turkish women. CHD incidence studies among Moroccan minorities are scarce; a Dutch report showed a lower AMI incidence in Moroccan minorities equivalent to our study,22 possibly due to the lower prevalence of hypertension and smoking compared with ethnic Dutch.20;21;23

Latin-American and Caribbean minorities Of the Latin American and Caribbean minority groups, only the Surinamese had a statistically significantly higher AMI incidence. Suriname is a former Dutch colony, and migration to the Netherlands was mainly related to the political developments in Suriname between 1975 and 1980. The Surinamese population is ethnically diverse and consists of people originating from India (37%), West Africa (30%), Java (15%), China (1.5%), and people of mixed origin.24 A substantial part of the Surinamese residing in the Netherlands comprises people from Indian origin. As minorities from India had a higher AMI incidence (Table 2), this could explain the higher incidence found in Surinamese minorities. A high prevalence of hypertension and type 2 diabetes among Surinamese minorities in the Netherlands may drive the higher AMI incidence.25;26

East-Asian minorities Compared with ethnic Dutch, AMI incidence was considerably lower among all East-Asian minority groups, except among Indonesians (HR: 1.03; 95% CI: 1.01-1.06). Indonesia is also a former Dutch colony, and migration took place between 1945 and 1949 in the aftermath of World War II and the Indonesian War of Independence. Indonesians are well integrated in the Dutch society with respect to language and culture.27 This possibly results in more similarity to the ethnic Dutch population compared with other East-Asian minority groups, and may underlie the more similar AMI incidence.

South-Asian minorities South Asian minorities had a higher AMI incidence compared with ethnic Dutch. This is in agreement with other studies from the UK and Sweden.4;6;7;9 Potential explanations are the higher prevalence of

31

Chapter 2.1 diabetes and a more aggressive coronary wall calcification process, possibly due to genetic predispositions.28;29

Incidence differences between first and second generation ethnic minorities and ethnic Dutch One previous study, limited to minorities from European origin, concluded that the higher CHD incidence found in most first generation ethnic minority groups converged towards the majority population in the second generation, but in women only.8 In our study this convergence was found in some, but not all, ethnic minority groups with no marked differences between men and women (Supplementary table 1), implying that findings in one country are not necessarily generalizable to the same ethnic minority groups residing in other countries. Changes in AMI incidence over generations were often unbeneficial, and most profound among Polish and Chinese minorities. This is worrying since the number of Polish and Chinese minorities is not only expanding in the Netherlands (Polish born 1997: 0.09%, 2011: 0.40%; Chinese born 1997: 0.11%, 2011: 0.27% of the total Dutch population),30 but also in other Western countries.31 First generation Polish minorities had a similar AMI incidence as ethnic Dutch, whereas the second generation had a higher incidence. This could be explained by the loss of beneficial health behaviours from the country of origin, the adoption of unhealthy behaviours in the host country,32 and by the weakening of the healthy migrant effect over generations.11 However, research regarding cardiovascular health among Polish minorities is lacking, and risk factors that may drive this intergenerational change remain unclear. Among Chinese, the lower AMI incidence in the first generation compared with ethnic Dutch, turned into an almost significantly higher incidence in the second generation. This is tremendously interesting as almost all studies show a better health outcome in Chinese minorities compared with the majority population. Our results suggest that this health advantage vanishes over generations. Possibly, healthy behaviours from China have been lost (e.g. traditional dietary patterns33), while unhealthy behaviours have been adopted (e.g. physical inactivity,34 obesity35). In contrast, beneficial changes over generations were seen in minorities from the three former Dutch colonies, a finding previously reported for cardiovascular mortality.27 These ethnic minority groups are more closely related to ethnic Dutch because of colonial heritage. Therefore, second generation ethnic minorities may have an advantage in adopting healthy lifestyle factors in the Netherlands, while retaining healthy lifestyle factors of their parents. This is reflected in a previous Dutch study, which showed a higher overall diet quality in Surinamese minorities than in ethnic Dutch. Those who had more social contacts with ethnic Dutch (as parameter for integration into Dutch society) had the same fruit and vegetable intake, but a lower salt intake.36 Furthermore, offspring of migrants from the former Dutch colonies may have a stronger social cohesion with ethnic Dutch and a better language proficiency compared with offspring of other migrant groups,

32

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups which might improve their ability to adequately use health care facilities and adhere to prescribed therapy.37

Considerations There were several strengths in our study. We presented data regarding changes in AMI incidence in ethnic minorities over generations, which is very rarely studied. To our knowledge, we are among the first presenting such data for minorities from non-European countries like China, Indonesia, and the USA. Unlike previous studies, we avoided clustering of ethnic minority groups by geographical region, to prevent mixing of effects that may obscure results. Furthermore, we included a wide range of ethnic minority groups, creating the opportunity for cross-country comparison of results. There are some limitations in our study that need to be addressed. Non-unique persons in the PR (with respect to the variables date of birth, gender, and four digits of the postal code) were excluded in our cohort.38 This could potentially lead to selection bias. Since uniqueness was not related to the outcome under study, exclusion of non-unique persons will only have led to precision loss, not to biased results. Furthermore, as in numerous studies the classification of the various groups was based on country of birth. Country of birth may reflect ethnicity reasonably well among some ethnic minority groups but might be an unreliable measure of ethnicity in other groups such as Surinamese.16 Since previous studies reported considerable differences in cardiovascular disease between African Surinamese and Asian Surinamese, efforts should be made to distinguish these groups in future national registers.28

Conclusion Health care professionals and policy makers should be aware of substantial AMI incidence differences between ethnic minority groups and the majority population, and the often unbeneficial change over generations, most striking in some of Europe’s most expanding ethnic minority groups (Polish, Chinese). Future research should be cautious when clustering ethnic minority groups based on geographical region of the country of origin.

33

Chapter 2.1

REFERENCES 1. Anand SS, Yusuf S, Vuksan V, Devanesen S, Teo KK, Montague PA, et al. Differences in risk factors, atherosclerosis, and cardiovascular disease between ethnic groups in Canada: the Study of Health Assessment and Risk in Ethnic groups (SHARE). Lancet 2000;22;356(9226):279-84. 2. Mak KH, Chia KS, Kark JD, Chua T, Tan C, Foong BH, et al. Ethnic differences in acute myocardial infarction in Singapore. Eur Heart J 2003;24(2):151-60. 3. Rafnsson SB, Bhopal RS. Large-scale epidemiological data on cardiovascular diseases and diabetes in migrant and ethnic minority groups in Europe. Eur J Public Health 2009;19(5):484-91. 4. Fischbacher CM, Bhopal R, Povey C, Steiner M, Chalmers J, Mueller G, et al. Record linked retrospective cohort study of 4.6 million people exploring ethnic variations in disease: myocardial infarction in South Asians. BMC Public Health 2007;7:142. 5. Gadd M, Johansson SE, Sundquist J, Wandell P. Morbidity in cardiovascular diseases in immigrants in Sweden. J Intern Med 2003;254(3):236-43. 6. Hedlund E, Lange A, Hammar N. Acute myocardial infarction incidence in immigrants to Sweden. Country of birth, time since immigration, and time trends over 20 years. Eur J Epidemiol 2007;22(8):493-503. 7. Hempler NF, Larsen FB, Nielsen SS, Diderichsen F, Andreasen AH, Jorgensen T. A registry-based follow- up study, comparing the incidence of cardiovascular disease in native Danes and immigrants born in Turkey, Pakistan and the former Yugoslavia: do social inequalities play a role? BMC Public Health 2011;11:662. 8. Sundquist K, Li X. Coronary heart disease risks in first- and second-generation immigrants in Sweden: a follow-up study. J Intern Med 2006;259(4):418-27. 9. Bhopal RS, Bansal N, Fischbacher C, Brown H, Capewell S. Ethnic variations in chest pain and angina in men and women: Scottish Ethnicity and Health Linkage Study of 4.65 million people. Eur J Cardiovasc Prev Rehabil 2012;19(6):1250-7. 10. Uniken Venema HP, Garretsen HF, van der Maas PJ. Health of migrants and migrant health policy, The Netherlands as an example. Soc Sci Med 1995;41(6):809-18. 11. Fennelly K. The "healthy migrant" effect. Minn Med 2007;90(3):51-3. 12. Landrine H, Klonoff EA. Culture change and ethnic-minority health behavior: an operant theory of acculturation. J Behav Med 2004;27(6):527-55. 13. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 14. Harteloh P, de Bruin K, Kardaun J. The reliability of cause-of-death coding in The Netherlands. Eur J Epidemiol 2010;25(8):531-8. 15. Merry AH, Boer JM, Schouten LJ, Feskens EJ, Verschuren WM, Gorgels AP, et al. Validity of coronary heart diseases and heart failure based on hospital discharge and mortality data in the Netherlands using the cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009;24(5):237-47. 16. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69. 17. van Oeffelen AAM, Agyemang C, Bots ML, Stronks K, Koopman C, van Rossem L, et al. The relation between socioeconomic status and short-term mortality after acute myocardial infarction persists in the elderly: results from a nationwide study. Eur J Epidemiol 2012;27(8):605-13. 18. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57(12):1288-94.

34

Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups

19. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56(3):221-9. 20. Hosper K, Nierkens V, Nicolaou M, Stronks K. Behavioural risk factors in two generations of non- Western migrants: do trends converge towards the host population? Eur J Epidemiol 2007;22(3):163- 72. 21. Ujcic-Voortman JK, Baan CA, Seidell JC, Verhoeff AP. Obesity and cardiovascular disease risk among Turkish and Moroccan migrant groups in Europe: a systematic review. Obes Rev 2012;13(1):2-16. 22. Klijs B, Kunst A, Mackenbach JP, Nijhof G, Reulings P, Stirbu I, et al. Ongelijkheid in gezondheid, is gezondheidszorg van belang? Sociaaleconomische en etnische verschillen in gezondheidszorguitkomsten op het terrein van hart- en vaatziekten in Nederland. Inspectie voor de Gezondheidszorg; 2011. 23. Yusuf S, Hawken S, Ounpuu 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. Lancet 2004;364(9438):937-52. 24. Bos V, Kunst AE, Keij-Deerenberg IM, Garssen J, Mackenbach JP. Ethnic inequalities in age- and cause- specific mortality in The Netherlands. Int J Epidemiol 2004;33(5):1112-9. 25. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7. 26. Bindraban NR, van Valkengoed IB, Mairuhu G, Holleman F, Hoekstra JB, Michels BP, et al. Prevalence of diabetes mellitus and the performance of a risk score among Hindustani Surinamese, African Surinamese and ethnic Dutch: a cross-sectional population-based study. BMC Public Health 2008;8:271. 27. Ho L, Bos V, Kunst AE. Differences in cause-of-death patterns between the native Dutch and persons of Indonesian descent in the Netherlands. Am J Public Health 2007;97(9):1616-8. 28. Chaturvedi N. Ethnic differences in cardiovascular disease. Heart 2003;89(6):681-6. 29. Koulaouzidis G, Nicoll R, Charisopoulou D, McArthur T, Jenkins PJ, Henein MY. Aggressive and diffuse coronary calcification in South Asian angina patients compared to Caucasians with similar risk factors. Int J Cardiol 2013;167(6):2472-6. 30. Statistics Netherlands, Statline 2012. 31. Salt J. Trends in Europe's international migration. In: Rechel B, Mladovsky P, Devillé W, Rijks B, Petrova-Benedict R, McKee M, editors. Migration and health in the European Union. 1 ed. England: Open University Press; 2012. p. 23. 32. Landrine H, Klonoff EA. Culture change and ethnic-minority health behavior: an operant theory of acculturation. J Behav Med 2004;27(6):527-55. 33. Harrison GG, Kagawa-Singer M, Foerster SB, Lee H, Pham KL, Nguyen TU, et al. Seizing the moment: California's opportunity to prevent nutrition-related health disparities in low-income Asian American population. Cancer 2005;104(12 Suppl):2962-8. 34. Fable-Munsuz A, Ponce NA, Rodriguez M, Perez-Stable EJ. Immigrant generation and physical activity among Mexican, Chinese & Filipino adults in the U.S. Soc Sci Med 2010 ;70(12):1997-2005. 35. Smith NR, Kelly YJ, Nazroo JY. The effects of acculturation on obesity rates in ethnic minorities in England: evidence from the Health Survey for England. Eur J Public Health 2012;22(4):508-13. 36. Nicolaou M, van Dam RM, Stronks K. Acculturation and education level in relation to quality of the diet: a study of Surinamese South Asian and Afro-Caribbean residents of the Netherlands. J Hum Nutr Diet 2006;19(5):383-93.

35

Chapter 2.1

b

37. Gushulak BD, Pottie K, Hatcher RJ, Torres S, DesMeules M. Migration and health in Canada: health in the global village. CMAJ 2011;183(12):E952-E958.

------

38. Reitsma JB, Kardaun JW, Gevers E, de Bruin A, van der Wal J, Bonsel GJ. [Possibilities for anonymous Women follow-up studies of patients in Dutch national medical registrations using the Municipal Population 0.95 (0.64;1.42) 0.95 1.42 (0.94;2.13) 0.98 (0.67;1.45) 0.74 (0.42;1.30) 1.04 (0.54;1.99) 1.27 (0.90;1.80) 1.34 (0.68;2.65) Register: a pilot study]. Ned Tijdschr Geneeskd 2003;147(46):2286-90. 1.15 (1.01;1.31)* 1.17 (1.10;1.24)* 0.72 (0.63;0.81)*

generation ethnic minorities ethnic generation

------st

Men vs. 1 vs.

nd 1.42 (0.98;2.05) 1.42 1.07 (0.94;1.22) 1.01 (0.70;1.47) 1.09 (0.84;1.41) 1.04 (0.76;1.43) 0.72 (0.49;1.06) 0.78 (0.43;1.44) 2 1.08 (1.01;1.15)* 0.74 (0.69;0.80)* 1.34 (1.02;1.76)*

a

1.20) 1.61) 1.03) 1.28) 1.34) 1.77) 1.22)* 1.21)* 0.83)* 1.82)* ------

------(0.52

Women 2007, stratified by generation and sex, adjusted for for adjusted sex, and generation by stratified 2007, - 0.84 (0.59 0.84 1.17 (0.84 0.73 0.77 (0.46 0.78 (0.45 0.98 (0.54 1.14 (1.06 1.17 (1.13 0.74 (0.66 1.37 (1.04

1.26) 1.17) 1.08) 1.57) 1.04) 1.09)

1.14)* 1.15)* 0.90)* 1.52)* ------

------

Dutch ethnic vs. generation

Men nd 2 the Netherlands, 1997 Netherlands, the 1.03 (0.84 1.03 0.91 (0.70 0.87 (0.70 1.23 (0.96 0.72 (0.50 0.68 (0.42 1.07 (1.01 1.12 (1.09 0.84 (0.78 1.25 (1.02

a 1.08) 1.10) 1.06) 1.04) 1.32) 1.08) 1.07) 1.32) 1.03) 0.90)* ------

-

------

Women

(0.73 0.89 0.99 (0.88 0.82 (0.64 1.00 (0.95 1.04 (0.82 1.03 (0.99 0.75 (0.52 1.08 (0.88 0.73 (0.52 0.75 (0.62

1.13) 1.17) 1.10) 1.44) 1.13) 1.12) 1.27)

0.99)* 0.92)* 1.17)* ------

------generation vs. ethnic Dutch ethnic vs. generation Men st 1 1.00 (0.89 0.90 (0.68 1.03 (0.97 1.18 (0.96 1.00 (0.89 0.93 (0.77 0.87 (0.60 0.73 (0.53 0.73 0.80 (0.69 1.13 (1.09

a

1.92) 1.04) 2.65) 1.18) 1.13) 1.13) 0.88) 1.74) 1.21) 1.03) 1.18) 1.02) 1.10) 1.38) 1.05) 1.16)* 0.93)* 0.92)* 0.88)* 0.54)* 3.97)* ------(0.78 Women 1.03 (0.56 (0.74 0.88 1.62 (1.00 0.86 (0.63 0.93 (0.76 1.10 (1.06 0.74 (0.63 1.05 (0.63 0.98 (0.79 0.99 (0.95 0.65 (0.36 0.76 (0.56 0.93 1.17 (0.99 0.78 (0.58 1.09 (1.03 0.56 (0.34 0.68 (0.50 0.52 (0.31 0.40 (0.29 2.26 (1.28

1.47) 1.09) 1.96) 1.09) 1.47) 1.30) 1.08) 1.09) 1.20) 1.05)

1.12)* 0.72)* 0.67)* 0.95)* 1.13)* 0.92)* 1.40)* 0.57)* 1.09)* 0.54)* 2.39)* ------

------Men

(0.58 Ethnic minorities vs. ethnic Dutch ethnic vs. minorities Ethnic 0.81 (0.45 (0.78 0.92 1.11 (0.63 0.90 (0.75 1.18 (0.96 1.01 (0.78 0.97 (0.86 0.96 (0.84 1.05 (0.92 0.78

1.06 (1.01 0.53 (0.39 0.54 (0.43 0.70 (0.52 1.10 (1.07 0.82 (0.73 1.20 (1.02 0.41 (0.29 1.05 (1.02 0.47 (0.41 1.87 (1.46

36

Kong

Supplementary table 1 AMI incidence differences between ethnic minority groups and ethnic Dutch (HR (95% CI)) in CI)) in (95% (HR Dutch ethnic and groups minority ethnic between differences incidence AMI 1 table Supplementary age origin of Country Australia Austria Belgium British Verde Cape China Czechoslovakia France Germany Britain Great Greece Hungary Hong Indonesia Iran Italy Morocco Antilles Netherlands Pakistan Poland Scandinavia Ethnic inequalities in acute myocardial infarction incidence in first and second generation ethnic minority groups

b

------Women 0.95 (0.64;1.42) 1.15 (1.01;1.31) 1.42 (0.94;2.13) 1.17 (1.10;1.24) 0.98 (0.67;1.45) 0.74 (0.42;1.30) 0.72 (0.63;0.81) 1.04 (0.54;1.99) 1.27 (0.90;1.80) 1.34 (0.68;2.65) 1.14 (0.58;2.23) 0.67 (0.46;0.97) 1.43 (0.65;3.13)

generation ethnic minorities

------st Men (0.69;0.80) vs. 1

nd 1.42 (0.98;2.05) 1.07 (0.94;1.22) 1.01 (0.70;1.47) 1.08 (1.01;1.15) 1.09 (0.84;1.41) 1.04 (0.76;1.43) 0.74 0.72 (0.49;1.06) 1.34 (1.02;1.76) 0.78 (0.43;1.44) 1.28 (0.72;2.28) 0.59 (0.45;0.77) 1.25 (0.78;2.00) 2

a

1.20) 1.22) 1.61) 1.21) 1.03) 1.28) 0.83) 1.34) 1.82) 1.77) 2.06) 1.31) 1.09) ------

------2007, stratified by generation and sex, adjusted for age (1.04 - Women 0.84 (0.59 1.14 (1.06 1.17 (0.84 1.17 (1.13 0.73 (0.52 0.77 (0.46 0.74 (0.66 0.78 (0.45 1.37 0.98 (0.54 1.11 (0.60 0.90 (0.62 0.63 (0.37

1.26) 1.14) 1.17) 1.15) 1.08) 1.57) 0.90) 1.04) 1.52) 1.09) 1.67) 1.15) 1.11)

------

------generation vs. Dutchethnic Men nd 2 1.03 (0.84 1.07 (1.01 0.91 (0.70 1.12 (1.09 0.87 (0.70 1.23 (0.96 0.84 (0.78 0.72 (0.50 1.25 (1.02 0.68 (0.42 1.20 (0.87 0.89 (0.68 0.80 (0.57

a

1.08) 1.10) 1.06) 1.04) 0.90) 1.32) 1.08) 1.07) 1.32) 1.03) 1.26) 1.45) 0.78) ------

------(0.82 Women 0.89 (0.73 0.99 (0.88 0.82 (0.64 1.00 (0.95 0.75 (0.62 1.04 1.03 (0.99 0.75 (0.52 1.08 (0.88 0.73 (0.52 0.97 (0.75 1.35 (1.27 0.44 (0.25

0.99) 1.13) 1.17) 1.10) 0.92) 1.44) 1.17) 1.13) 1.12) 1.27) 1.51) 1.57) 0.88)

------

------generation ethnicvs. Dutch Men

(1.43 (0.46 st 1 0.73 (0.53 1.00 (0.89 0.90 (0.68 1.03 (0.97 0.80 (0.69 1.18 (0.96 1.13 (1.09 1.00 (0.89 0.93 (0.77 0.87 (0.60 0.94 (0.58 1.50 0.64

a

1.92) 1.04) 1.16) 2.65) 0.93) 0.92) 1.18) 1.13) 1.13) 0.88) 1.74) 1.21) 0.88) 1.03) 1.18) 1.02) 0.54) 1.10) 3.97) 1.38) 1.05) 1.21) 1.26) 1.17) 1.42) 0.88) 0.78) 1.33) 1.13) ------(1.00 (0.29 Women 1.03 (0.56 0.88 (0.74 1.09 (1.03 1.62 0.56 (0.34 0.68 (0.50 0.86 (0.63 0.93 (0.76 1.10 (1.06 0.74 (0.63 1.05 (0.63 0.98 (0.79 0.52 (0.31 0.99 (0.95 0.65 (0.36 0.76 (0.56 0.40 0.93 (0.78 2.26 (1.28 1.17 (0.99 0.78 (0.58 0.76 (0.48 0.99 (0.78 0.88 (0.66 1.33 (1.25 0.61 (0.42 0.53 (0.36 1.10 (0.91 1.01 (0.90

1.47) 1.09) 1.12) 1.96) 0.72) 0.67) 0.95) 1.09) 1.13) 0.92) 1.47) 1.40) 0.57) 1.09) 1.30) 1.08) 0.54) 1.09) 2.39) 1.20) 1.05) 1.37) 1.45) 0.98) 1.54) 1.28) 0.89) 1.23) 1.51)

------Men (0.73 Ethnic minorities vs. ethnic Dutch 0.81 (0.45 0.92 (0.78 1.06 (1.01 1.11 (0.63 0.53 (0.39 0.54 (0.43 0.70 (0.52 0.90 (0.75 1.10 (1.07 0.82 1.18 (0.96 1.20 (1.02 0.41 (0.29 1.05 (1.02 1.01 (0.78 0.97 (0.86 0.47 (0.41 0.96 (0.84 1.87 (1.46 1.05 (0.92 0.78 (0.58 1.00 (0.72 1.10 (0.84 0.84 (0.72 1.46 (1.39 0.94 (0.69 1.43 (1.35 0.70 (0.56 1.08 (0.95 AMI incidence differences between ethnic minority groups and ethnic Dutch (HR (95% CI)) in the Netherlands, 1997

t generationt ethnic minorities

groups categorygroups each in with at least cases ten included are Country of origin Australia Austria Belgium GuyanaBritish Cape Verde China Czechoslovakia France Germany Great Britain Greece Hungary Hong Kong Indonesia Iran Italy Morocco Netherlands Antilles Pakistan Poland Scandinavia South Africa Soviet Union Spain Suriname Switzerland Turkey USA Yugoslavia Reference: firs Reference: Dutch ethnic Only Supplementary table 1 b a

37

Chapter 2.1

Supplementary table 2 AMI incidence differences between ethnic minority groups and ethnic Dutch (HR (95 % CI)) in the Netherlands, 1997-2007, stratified by age group, adjusted for age and sex Country of origin Totala <55 yearsa ≥55 yearsa Cape Verde 0.54 (0.41;0.70)* 0.65 (0.46;0.92)* 0.44 (0.29;0.66)* China 0.57 (0.48;0.68)* 0.41 (0.29;0.58)* 0.67 (0.55;0.83)* Czechoslovakia 0.76 (0.61;0.95)* 0.88 (0.59;1.30) 0.72 (0.55;0.93)* Great Britain 0.80 (0.73;0.88)* 0.83 (0.72;0.95)* 0.77 (0.68;0.88)* Hong Kong 0.44 (0.33;0.58)* 0.43 (0.28;0.64)* 0.47 (0.32;0.69)* Morocco 0.46 (0.40;0.52)* 0.54 (0.45;0.64)* 0.38 (0.31;0.47)* Portugal 0.44 (0.33;0.58)* 0.50 (0.32;0.80)* 0.39 (0.28;0.56)* Scandinavia 0.78 (0.63;0.96)* 0.72 (0.50;1.05) 0.83 (0.64;1.07) Spain 0.85 (0.75;0.98)* 0.92 (0.74;1.14) 0.76 (0.64;0.91)* Switzerland 0.77 (0.61;0.98)* 0.78 (0.49;1.26) 0.77 (0.59;1.01) USA 0.65 (0.52;0.79)* 0.65 (0.48;0.90)* 0.65 (0.50;0.85)* Vietnam 0.64 (0.47;0.88)* 0.79 (0.54;1.16) 0.53 (0.31;0.91)* a Reference: ethnic Dutch *p<0.05 Only ethnic minority groups with a statistically significantly lower AMI incidence compared with ethnic Dutch in the unstratified analyses, and with at least ten cases in each category are included.

38

CHAPTER 2.2

Sex disparities in acute myocardial infarction incidence by ethnic group

van Oeffelen AAM, Vaartjes I, Stronks K, Bots ML, Agyemang C. Sex disparities in acute myocardial infarction incidence: Do ethnic minority groups differ from the majority population?

Eur J Prev Cardiol 2013 [Epub ahead of print]. Chapter 2.2

ABSTRACT Background The incidence of acute myocardial infarction (AMI) in men exceeds that in women. The extent of this sex disparity varies widely between countries. Variations may also exist between ethnic minority groups and the majority population, but scientific evidence is lacking.

Methods A nationwide register-based cohort study was conducted (n=7,601,785) between 1997 and 2007. Cox proportional hazard models were used to estimate sex disparities in AMI incidence within the ethnic Dutch population and within ethnic minority groups, stratified by age (30-54, 55-64, ≥65 years).

Results AMI incidence was higher in men than in women in all groups under study. Compared with ethnic Dutch (Hazard Ratio (HR): 2.23; 95% confidence interval (95% CI): 2.21-2.25), sex disparities were similar among ethnic minorities originating from the immediate surrounding countries (Belgium, Germany), whereas they were greater in most other ethnic minority groups. Most pronounced results were found among minorities from Morocco (HR: 3.48; 95% CI: 2.48-4.88), South Asia (HR: 3.92; 95% CI: 2.45-6.26), and Turkey (HR: 3.98; 95% CI: 3.51-4.51). Sex disparity differences were predominantly evident in those below 55 years of age, and were mainly provoked by a higher AMI incidence in ethnic minority men compared with ethnic Dutch men.

Conclusion Sex disparities in AMI incidence clearly varied between ethnic minorities and ethnic Dutch. Health prevention strategies may first target at a reduction of AMI incidence in young ethnic minority men, especially those originating from Turkey and South Asia. Furthermore, an increase in AMI incidence in their female counterparts should be prevented.

40

Sex disparities in acute myocardial infarction incidence by ethnic group

INTRODUCTION

It is well known that the incidence of acute myocardial infarction (AMI) is much higher in men than in women, and that this difference diminishes with age. Sex disparities in cardiovascular risk factors and the protective effect of female hormones in premenopausal women are considered the most important explanatory factors.1;2 Although the adverse position of men regarding AMI incidence is universal, the extent varies considerably between countries.3 Consequently, it seems plausible that sex disparities in AMI incidence differ between ethnic minority groups and the majority population. This may be of relevance to enable specifically targeted cardiovascular preventive efforts. Literature regarding sex disparities in coronary heart disease (CHD) among ethnic minorities relative to the majority population is scarce, especially in Europe. Available studies focused on CHD mortality instead of CHD incidence, and were performed in the encompassing ethnic minority groups dissimilar from those residing in European countries. Ho et al demonstrated a threefold higher sex disparity in CHD mortality among than among African Americans in the younger age groups.4 This is in agreement with previous Dutch studies reporting significantly smaller sex disparities of several cardiovascular risk factors (hypertension and the metabolic syndrome, including all of its components) in Surinamese migrants, compared with the ethnic Dutch population.5;6 However, it is unclear whether the smaller sex disparity in CHD risk factors among some Dutch minority groups reflects on CHD incidence. The aim of the present study was to determine sex disparities in AMI incidence in ethnic minority groups, and the difference of these sex disparities with the ethnic Dutch population. Furthermore, we investigated whether the magnitude of this difference changed with increasing age. In case of a significant difference in sex disparity compared with ethnic Dutch, we explored whether this was caused by AMI incidence differences between ethnic minority men and ethnic Dutch men, or between ethnic minority women and ethnic Dutch women.

METHODS

Cohort enrolment A nationwide register-based cohort study was conducted. The Population Register (PR), Hospital Discharge Register (HDR), Cause of Death Register (CDR), and Regional Income Survey (RIS) were used to obtain information regarding demographic factors, AMI hospitalisations, fatal AMI events, and comorbidities. The registers are described in detail previously.7 The overall quality of Dutch national registers proved to be adequate.8;9 By linking previous registers with a personal identifier a cohort was built, starting at 1 January 1997. On 1 January 1997, 15,045,392 Dutch citizens (16.4% migrants), with no previous hospital admission for AMI (ICD-9 code 410) or old AMI (ICD-9 code 412) between 1995 and 1996, were registered in the PR. Only persons registered during the whole period between 1 January 1995

41

Chapter 2.2 and 1 January 1997 were included to ensure a minimal residing period in the Netherlands of two years, and to enable taking medical history data into account for every person under study. Due to the absence of a personal identifier in the HDR, the PR and HDR could only be linked via the combination of date of birth, sex and four digits of the postal code as identifying key. In case of multiple persons with an identical identifying key (non-uniqueness), PR and HDR could not be validly linked. Persons who were not present or not unique in the PR between 1 January 1995 and 1 January 1997 were excluded (3,071,969 persons, 24.2% migrants). As the outcome of interest is AMI, persons younger than 30 years of age were excluded (4,371,638 persons, 17.3% migrants). The final cohort contained 7,601,785 persons.

Follow-up Persons were followed from 1 January 1997 until their first AMI event, comprising a hospital admission with AMI as primary or secondary diagnosis (ICD-9 code 410), or a fatal event with AMI as primary or secondary cause of death (ICD-10 code I21). The validity of these ICD-codes proved to be good (sensitivity: 84%, positive predictive value: 97%).9 Persons were censored in case of death, non-uniqueness, immigration, or the end of the study period at 31 December 2007, whichever came first.

Determinants Ethnic background Ethnic minority groups were constructed based on the country of birth of the resident and his/her parents, according to the definition of Statistics Netherlands.10 A person was considered an ethnic minority if at least one of the parents was born abroad. The biggest Western (Belgium, Germany, Great Britain) and non-Western ethnic minority groups (Cape Verde, Indonesia, Morocco, Netherlands Antilles, Suriname, Turkey) residing in the Netherlands were included, as well as some other common ethnic minority groups residing in Europe (China, Poland, South Asia, and USA). Persons with both parents born in the Netherlands were indicated as ethnic Dutch.

Neighbourhood socioeconomic status Neighbourhood income was used as proxy for socioeconomic status (SES), and was based on income data registered in the RIS.11 In each neighbourhood, the mean disposable income of all residents with income data available (about one third of the Dutch population) was calculated for 1997 and subsequently assigned to all residents living in that neighbourhood on 1 January 1997. Neighbourhood SES was divided into tertiles, with the first tertile representing the lowest income group.

42

Sex disparities in acute myocardial infarction incidence by ethnic group

Comorbidity Presence and extent of comorbidity were determined with the Charlson index score, based on previous hospital admissions in the HDR.12 It proved to be a reliable and valid method to measure comorbidity in clinical research.13 The Charlson index score ranges from zero to six (cut-off value), with zero representing no comorbidity.

Data analysis Baseline characteristics (age, neighbourhood SES, comorbidity) were analysed on 1 January 1997 in the ethnic Dutch population and in all ethnic minority groups under study, for men and women separately. Absolute incidence rates were calculated as the number of AMI events, fatal and non- fatal combined, per 100,000 person-years at risk. As ethnic minorities are in general younger than the ethnic Dutch population, absolute incidence rates in minorities may be underestimated. To prevent age bias, the absolute incidence rates were age-standardised with a direct method using the age distribution of the European population with ten-year age bands. We investigated the difference in AMI incidence between men and women for every group under study using Cox proportional hazard regression analyses, adjusted for age at baseline (with women as reference group). These analyses were repeated in several age groups (30-55, 55-65, and ≥65 years). By adding interaction terms between ethnicity and sex to the model, the statistical significance of the difference in sex disparity between ethnic minority groups and ethnic Dutch was investigated. In ethnic minority groups where the sex disparity significantly differed from ethnic Dutch, Cox proportional hazard regression analyses were used to determine the AMI incidence difference between ethnic minorities and ethnic Dutch in men and women separately (with ethnic Dutch as reference group). Doing so, we assessed whether the deviation in sex disparity from ethnic Dutch was due to a difference in AMI incidence among ethnic minority men, or to a difference in AMI incidence among ethnic minority women. Results were expressed as Hazard Ratios (HR) with accompanying 95% confidence intervals (95% CI). We used SPSS software, version 14.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation of Statistics Netherlands.

RESULTS

After exclusion of the ethnic minority groups not under study (N=160,004), the cohort comprised 7,441,781 persons (11.0% ethnic minorities), of which 446,591 were from non-Western (48.9% men), and 368,960 from Western origin (45.0% men). In all groups under study, baseline characteristics were similar among men and women with respect to age, neighbourhood SES, and comorbidity (Table 1).

43

Chapter 2.2

c

75 40 92 349 159 210 190 127 361 157 120 294 139 417 194 403 131 345 116 264 162 666 173 366 175 285 118 337 186 223 323 172

standardised IR -

Age

c

66 37 97 IR 557 317 193 244 120 509 259 156 304 126 558 213 422 105 474 114 395 204 483 393 639 378 308 156 584 216 302 577 358

years at risk - 20,676 22,816 61,496 61,964 77,292 31,976 18,976 89,941 93,760 35,639 65,230 23,198 25,886 831,225 949,491 126,628 110,055 100,052 304,573 423,155 281,603 259,199 216,693 281,532 996,600 1,235,890 1,707,204 1,918,458 1,362,071 1,702,298 26,497,833 29,514,771

Person

40 15 81 41 41 20 70 25 197 235 126 901 135 296 277 146 208 141 4,230 2,461 1,700 1,355 6,751 3,905 1,444 1,107 6,365 4,667 7,864 6,086 N AMI 93,574

147,500

4.0 4.0 7.8 7.3 T3% 35.9 35.2 20.9 22.2 38.7 39.2 15.0 14.0 22.0 23.0 18.8 19.0 17.5 21.4 31.0 31.3 29.1 29.3 42.1 45.8 28.5 31.1 44.0 48.6 25.9 27.3 30.6 30.9

b

SES (T1: lowest income tertile, T3: highest income tertile) ;

T1% 29.6 30.6 87.7 87.6 51.8 51.7 31.5 31.5 61.0 62.5 54.4 51.7 62.2 61.3 63.7 59.9 72.1 73.4 27.7 27.0 34.4 35.4 30.2 27.5 39.3 37.2 31.5 27.3 49.3 47.9 33.1 33.5

7.2 8.9 %>0 26.9 27.7 12.6 15.5 15.8 15.4 21.1 23.2 15.5 21.3 20.7 24.3 13.3 16.9 18.1 19.8 18.8 21.8 29.6 29.3 29.9 30.6 16.6 20.8 26.6 20.4 16.7 17.7 28.6 29.3

Comorbidity

)

a

63) 67) 53) 49) 55) 54) 58) 63) 52) 47) 49) 51) 52) 53) 45) 45) 54) 51) 65) 69) 62) 67) 52) 55) 63) 53) 53) 53) 55) 56) 62) 67) ------Age 50 (40 52 (40 44 (38 40 (35 46 (38 44 (36 46 (38 47 (38 41 (34 39 (34 41 (35 42 (36 43 (36 42 (36 40 (35 39 (34 41 (34 41 (34 53 (43 56 (43 54 (46 56 (47 45 (38 46 (38 47 (40 44 (38 45 (37 44 (37 44 (36 44 (36 53 (44 55 (45 Median (IQRMedian

ased on mean disposable income on neighborhood level N 2,698 2,656 4,261 4,125 4,360 2,216 4,465 7,594 3,118 3,319 . 17,461 13,413 10,796 12,600 40,978 50,821 36,276 29,965 26,888 33,732 11,515 11,482 101,418 112,547 218,248 228,343 119,981 146,866 165,967 202,993 3,178,693 3,447, 537

years at risk -

Socioeconomic status b status Socioeconomic

Sex Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women Men Women b

ange;

ate per 100,000 person 100,000 per ate

Britain Baseline characteristics of ethnic minority groups and ethnic Dutch ≥30 years, stratified sexby

Western ethnic minority groups

- Country of origin Netherlands Non Cape Verde China Indonesia Morocco Netherlands Antilles Suriname South Asia Turkey Total Western ethnic minority groups Belgium Germany Great Poland USA Total incidence r IQR: interquartile r

Table 1 a c

44

Sex disparities in acute myocardial infarction incidence by ethnic group

Sex disparity in AMI incidence within ethnic minority groups and ethnic Dutch AMI incidence was significantly higher in men than in women in ethnic Dutch (HR: 2.23; 95% CI: 2.21-2.25) as well as in all ethnic minority groups under study (Figure 1). Compared with ethnic Dutch, sex disparities were significantly more pronounced in minorities from Indonesia, Netherlands Antilles, and Suriname, and most striking in minorities from Morocco (HR: 3.48; 95% CI: 2.48-4.88), South Asia (HR: 3.92; 95% CI: 2.45-6.26), and Turkey (HR: 3.98; 95% CI: 3.51-4.51). Sex disparities were also greater in minorities from Great Britain and the USA, although they did not differ significantly from ethnic Dutch. In the other ethnic minority groups, sex disparities were more similar to ethnic Dutch, especially in those originating from the immediate surrounding countries (Belgium and Germany). Overall, sex disparities were most pronounced in the young (<55 years of age), and declined gradually with increasing age (Table 2).

Figure 1 Difference in AMI incidence between men and women ≥ 30 years (HR (95% CI)) in ethnic minority groups and the ethnic Dutch population, adjusted for age. Women are used as reference group. The dotted line represents the Hazard Ratio of the ethnic Dutch population.

45

Chapter 2.2

- value 0.64 0.48 0.31 0.35 0.06 0.73 0.09 - 0.00* 0.02* 0.04* 0.00* 0.02* 0.00* p ethnicity*sex

a 1.78) 1.78) 1.83) 2.50) 1.68) 2.78) 1.93) 1.83) 2.47) 2.20) 4.15) 10.20) 21.00) 12.35) ------

(1.20

HR (95% CI) 1.76 (1.75 1.04 (0.61 1.71 (1.60 1.53 (0.93 1.44 (1.23 1.83 (1.21 1.74 (1.57 1.73 (1.64 1.73 1.64 (1.23 2.14 (1.11 2.40 (0.57 4.91 (1.15 2.22 (0.40

≥65 years

N 212 968 118 686 1,175 1,285 8,435 1,996 2,552 2,184 42,772 18,457 66,351 1,695,759

a 2.76) 8.43) 5.19) 3.18) 2.53) 3.47) 2.50) 4.01) 3.96) 3.26) 2.60) 5.44) 3.67) 10.18) ------

(0.42 (2.26

HR (95% CI) 2.71 (2.65 2.35 (0.66 2.44 (1.15 2.83 (2.52 1.51 (0.91 2.22 (1.42 2.12 (1.80 1.29 3.17 (2.54 2.71 2.42 (2.25 3.41 (2.14 1.92 (1.01 3.35 (1.10 65 years - 55

N 756 296 987 728 1,135 4,048 2,371 2,467 30,141 11,316 11,375 11,804 72,266

1,086,982

a

8.77) 3.64) 5.74) 5.33) 5.44) 4.78) 8.14) 5.76) 3.83) 3.78) 4.46) 4.73) - 18.13) 13.90) ------

HR (95% CI)

3.57 (3.50 2.64 (1.21 4.74 (4.21 3.95 (2.87 4.22 (3.72 4.68 (2.69 4.87 (4.11 3.10 (3.52 3.44 (2.13 3.18 (2.26 3.28 (2.27 5.40 (3.32 6.35 (2.22 5.81 (2.43 55 years -

30

N 4,392 6,283 6,576 8,888 5,023 25,651 19,740 72,048 52,870 30,632 17,978 141,052 128,230

3,843,489 in ethnic minority groups ethnicand Dutch ≥30 years, stratified and foradjusted age

a 2.25) 4.68) 2.84) 2.57) 4.88) 3.57) 2.95) 6.26) 4.51) 2.31) 2.32) 3.33) 2.69) 5.35) ------

(2.32

HR (95% CI) 2.23 (2.21 2.59 (1.43 1.95 (1.34 2.44 3.48 (2.48 2.87 (2.31 2.72 (2.51 3.92 (2.45 3.98 (3.51 2.13 (1.97 2.23 (2.15 2.71 (2.20 2.17 (1.75 3.38 (2.14 Total

eenand men women

N 5,354 8,386 6,576 6,437 30,874 23,396 91,799 66,241 60,620 22,997 12,059 213,965 266,847 6,626,230 minority groups

AMI incidence difference betw

Western ethnic

- Country of origin Netherlands Non Cape Verde China Indonesia Morocco Netherlands Antilles Suriname South Asia Turkey Western ethnic minority groups Belgium Germany Great Britain Poland USA Sex disparity significantly different from the ethnic Dutch population

Reference: men * Table 2 a

46

Sex disparities in acute myocardial infarction incidence by ethnic group

Differences in AMI incidence between ethnic minority groups and ethnic Dutch stratified by sex Figure 2 shows the AMI incidence difference between ethnic minority groups from Indonesia, Morocco, Netherlands Antilles, Suriname, South Asia, and Turkey (the groups with significantly deviating sex disparities in AMI incidence from ethnic Dutch) and ethnic Dutch stratified by sex. Indonesian, Surinamese, South-Asian and Turkish minority men had a statistically significant higher AMI incidence compared with ethnic Dutch men. This was less pronounced or absent in their female counterparts. Minorities from Morocco had a significantly lower AMI incidence compared with ethnic Dutch in both sexes, which was slightly less pronounced in men. Minorities from the Netherlands Antilles had a similar AMI incidence as their ethnic Dutch counterparts in both sexes. Results were most striking in the youngest age group (Table 3).

Figure 2 Difference in AMI incidence between ethnic minority groups and the ethnic Dutch population ≥ 30 years (HR (95% CI)), stratified by sex, and adjusted for age. Only ethnic minority groups with a significant difference in sex disparity from ethnic Dutch are shown. Ethnic Dutch men and women are used as reference (horizontal line).

47

Chapter 2.2

Table 3 AMI incidence difference between ethnic minority groups and ethnic Dutch ≥30 years (HR (95% CI)), stratified by sex and age, and adjusted for agea Country of origin Totalb 30-55 yearsb 55-64 yearsb ≥65 yearsb Men Indonesia 1.05 (1.02-1.09)* 1.19 (1.13-1.25)* 1.08 (1.02-1.15)* 0.97 (0.92-1.02) Morocco 0.47 (0.41-0.54)* 0.58 (0.48-0.69)* 0.39 (0.30-0.51)* 0.30 (0.20-0.46)* Netherlands 0.96 (0.84-1.09) 1.10 (0.94-1.28) 0.87 (0.65-1.12) 0.69 (0.47-1.03) Antilles Suriname 1.46 (1.39-1.54)* 1.95 (1.83-2.07)* 1.25 (1.13-1.38)* 0.90 (0.80-1.01) South Asia 1.63 (1.38-1.93)* 2.05 (1.71-2.45)* 1.01 (0.58-1.78) 0.69 (0.26-1.83) Turkey 1.43 (1.35-1.51)* 1.76 (1.64-1.89)* 1.20 (1.09-1.31)* 0.69 (0.52-0.89)* Women Indonesia 0.99 (0.95-1.03) 0.88 (0.79-0.98)* 1.04 (0.94-1.15) 1.00 (0.96-1.05) Morocco 0.40 (0.29-0.54)* 0.38 (0.24-0.60)* 0.73 (0.47-1.12) 0.11 (0.03-0.45)* Netherlands 0.93 (0.78-1.10) 0.97 (0.74-1.29) 1.12 (0.80-1.58) 0.86 (0.64-1.16) Antilles Suriname 1.33 (1.25-1.42)* 1.63 (1.46-1.82)* 1.62 (1.42-1.84)* 1.10 (0.99-1.22) South Asia 1.25 (0.81-1.94) 1.51 (0.90-2.56) 2.16 (0.81-5.75) 0.59 (0.15-2.34) Turkey 1.01 (0.90-1.13) 1.28 (1.10-1.50)* 1.06 (0.86-1.30) 0.67 (0.49-0.91)* a Only ethnic minority groups with a significantly different sex disparity from ethnic Dutch are included b Reference: ethnic Dutch * p<0.05

DISCUSSION

This study demonstrates a higher AMI incidence in men than in women in the ethnic Dutch population as well as in all ethnic minority groups under study. Sex disparities were strikingly more pronounced in minorities from Morocco, South Asia, and Turkey compared with ethnic Dutch. This was mainly attributed to a more disadvantaged position of ethnic minority men compared with ethnic Dutch men. Differences in sex disparity were most prominent in the young (<55 years of age).

Discussion of key findings The previously observed smaller sex disparities in major CHD risk factors (hypertension, metabolic syndrome) in Surinamese minorities compared with ethnic Dutch, was not reflected in a smaller sex disparity in AMI incidence.5;6 Our results are also in contrast to an American study which showed a smaller sex disparity in CHD mortality in African Americans compared with White Americans.4 Explanations for this lack of agreement are unclear. However, several factors may underlie the observed discrepancies. First, risk factors may have a differential effect in men and women. It has been demonstrated that men develop more disease with the same level of cholesterol and blood pressure than women.14 Consequently, the previously reported smaller sex disparities in cholesterol and blood pressure levels in Surinamese minorities might not be reflected in a smaller sex disparity in

48

Sex disparities in acute myocardial infarction incidence by ethnic group

AMI incidence. Moreover, it takes time before risk factors, like hypertension and the metabolic syndrome, induce a myocardial infarction. Therefore, the broad sex disparities in AMI incidence in Surinamese minorities may narrow over time. The same principle may apply to other non-Western ethnic minority groups. Second, literature indicates narrowing sex disparities in CHD risk factors with increasing length of stay in the host country, due to the loss of the advantaged position regarding CHD in ethnic minority women. For instance, obesity prevalence in ethnic minority women may increase over time in the host country.15-17 Consequently, the current relatively low AMI incidence in ethnic minority women may increase in the future, leading to smaller sex disparities. This is supported by an American study showing considerably smaller sex disparities in CHD mortality in African Americans compared with White Americans. The main cause was the higher CHD mortality in African-American women compared with White American women.4 Ethnic minorities in the Netherlands migrated very recently,18 whereas African Americans have been residing in America for centuries. It is plausible that African-American women have lost some of their advantaged position towards AMI incidence over time, leading to a lower sex disparity. If this phenomenon also occurs in the Netherlands, the more pronounced sex disparity in ethnic minority groups may decline in the future. Third, our results might reflect on the epidemiological transition in ethnic minority groups. During the epidemiological transition, the major causes of death and disability shift from a predominance of nutritional deficiencies and infectious diseases towards degenerative diseases (such as CHD). The course of this transition towards CHD differs between the sexes.19 In Western countries, CHD mortality was equivalent in men and women before the 1920s. Thereafter, a steep increase in men occurred until the 1970s, whereas it remained much more stable in women. From the 1980s onwards, CHD mortality decreased more rapidly in men than in women. As a result, sex disparities were greatest between 1970 and 1980 after which they have been narrowing. Ethnic minorities from non-Western countries may have a time-lag in the epidemiological transition, resulting in greater sex disparities in AMI incidence.20 Fourth, sex disparities in smoking, one of the most important risk factors for AMI, have not yet been explicitly investigated. However, sex stratified data concerning smoking prevalence showed huge differences between ethnic minority men and women, whereas this was much more similar between ethnic Dutch men and women. Surinamese minority men had a twofold higher smoking prevalence than Surinamese minority women.6 Minority men from Morocco and Turkey, two groups with the most striking sex disparity in AMI incidence, had an advantaged position with respect to hypertension,21 obesity,21-23 and physical inactivity,21;24 but a seriously disadvantaged position with respect to smoking. Among Turkish minorities smoking prevalence was approximately two times higher in men than in women; and in Moroccan minorities, smoking prevalence was even ten times higher in men than in women.24;25 There is no information regarding smoking prevalence in South-

49

Chapter 2.2

Asian minorities by sex. However, among those living in South Asia, men had a twofold higher smoking prevalence than their female counterparts.26 It has been shown that smoking affects the absolute risk of CHD more in men than in women, which may further expand the sex disparity in AMI incidence.27 As our study indicated men as the main contributors to the more pronounced sex disparity in ethnic minority groups, their high smoking prevalence might be an important driver. Fifth, differences in access to health care and therapy compliance could influence sex disparities in AMI incidence. After migration, non-Western ethnic minorities suddenly acquire more access to health care facilities. It has been proposed that ethnic minority women are more likely to adequately use these facilities than ethnic minority men, for example regarding GP use.28 It has also been suggested that ethnic minority women have a better therapy compliance than ethnic minority men, for example regarding drug intake for hypertension control.29;30 Both factors could boost the disadvantaged position in men and increase the sex disparity.

Considerations Since data concerning sex disparities in CHD among ethnic minority groups are limited, the present study contributes importantly to this knowledge gap. An American study provided some evidence regarding sex disparities in CHD mortality between African Americans and White Americans, but not regarding CHD incidence. To our knowledge, there are no European studies concerning sex disparity differences in CHD between ethnic minority groups and the majority population. This study is unique in providing those data, not only for the major Western and non-Western ethnic minority groups in the Netherlands, but also for some important ethnic minority groups in other European countries (e.g. minorities from South Asia, Poland and China). The use of register-based data enabled us to build a large cohort, minimised the risk on information bias, and limited the number of missing data. Inevitably our study has some limitations. To be able to validly link all Dutch residents with the Hospital Discharge Register, citizens who were not unique with respect to the combination of the variables sex, birth date, and four digits of the postal code were excluded.31 Non-uniqueness was more prevalent among ethnic minorities than among the ethnic Dutch population, but since non- uniqueness was not related to the health outcomes under study it unlikely biased our findings.32 Bias could have been introduced by combining migrants and their offspring (first and second generation ethnic minorities). As additional analyses only demonstrated small and non-significant differences between HRs in the combined group and the HRs in first generation ethnic minorities only, validity of results remain unaffected. Some degree of non-differential misclassification of the explanatory variables Charlson comorbidity index and neighbourhood SES might be present, which may have led to incomplete adjustment for confounders.33 Since there is no evidence that this misclassification affected the various ethnic groups and sexes differently, conclusions of our study probably remain

50

Sex disparities in acute myocardial infarction incidence by ethnic group unchanged. Finally, we were unable to adjust our analyses for lifestyle factors, which could have given us more insight in explanations for found results.

Conclusion Compared with the ethnic Dutch population, sex disparities in AMI incidence were strikingly more pronounced in most non-Western ethnic minority groups, especially among those originating from Morocco, Turkey, and South Asia, and were most marked below 55 years of age. This high sex disparity was mainly attributed to the relatively high AMI incidence in ethnic minority men compared with ethnic Dutch men. Since Moroccan minorities already have a very low AMI incidence (in both men and women), health prevention strategies may first target at a reduction of AMI incidence in young men of Turkish and South-Asian origin. Furthermore, based on previous literature it seems plausible that, without intervention, non-Western ethnic minority women may lose some of their advantaged position with increasing time in the host country, leading to a higher AMI incidence. Therefore, early intervention strategies should also be targeted at these young ethnic minority women to prevent the loss of their beneficial risk factor profile. Countries encompassing other ethnic minority groups should be aware of possible differences in sex disparity in AMI incidence between these groups and the majority population, and take this into account when identifying target groups for cardiovascular preventive efforts.

51

Chapter 2.2

REFERENCES 1. Jousilahti P, Vartiainen E, Tuomilehto J, Puska P. Sex, age, cardiovascular risk factors, and coronary heart disease: a prospective follow-up study of 14 786 middle-aged men and women in Finland. Circulation 1999;99(9):1165-72. 2. Kalin MF, Zumoff B. Sex hormones and coronary disease: a review of the clinical studies. Steroids 1990;55(8):330-52. 3. Lawlor DA, Ebrahim S, Davey SG. Sex matters: secular and geographical trends in sex differences in coronary heart disease mortality. BMJ 2001;323(7312):541-5. 4. Ho JE, Paultre F, Mosca L. The gender gap in coronary heart disease mortality: is there a difference between blacks and whites? J Womens Health (Larchmt) 2005;14(2):117-27. 5. Agyemang C, de Munter J, van Valkengoed I, van den Born BJ, Stronks K. Gender disparities in hypertension among different ethnic groups in Amsterdam, The Netherlands: the SUNSET study. Am J Hypertens 2008;21(9):1001-6. 6. Agyemang C, van Valkengoed I, van den Born BJ, Bhopal R, Stronks K. Heterogeneity in sex differences in the metabolic syndrome in Dutch white, Surinamese African and South Asian populations. Diabet Med 2012;29(9):1159-64. 7. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 8. Harteloh P, de Bruin K, Kardaun J. The reliability of cause-of-death coding in The Netherlands. Eur J Epidemiol 2010;25(8):531-8. 9. Merry AH, Boer JM, Schouten LJ, Feskens EJ, Verschuren WM, Gorgels AP, et al. Validity of coronary heart diseases and heart failure based on hospital discharge and mortality data in the Netherlands using the cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009;24(5):237-47. 10. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69. 11. van Oeffelen AA, Agyemang C, Bots ML, Stronks K, Koopman C, van Rossem L., et al. The relation between socioeconomic status and short-term mortality after acute myocardial infarction persists in the elderly: results from a nationwide study. Eur J Epidemiol 2012;27(8):605-13. 12. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57(12):1288-94. 13. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56(3):221-9. 14. Isles CG, Hole DJ, Hawthorne VM, Lever AF. Relation between coronary risk and coronary mortality in women of the Renfrew and Paisley survey: comparison with men. Lancet 1992;339(8795):702-6. 15. Dagevos J, Dagevos H. Minderheden meer gewicht. Over overgewicht bij Turken, Marokkanen, Surinamers en Antillianen en het belang van integratiefactoren. Sociaal en Cultureel Planbureau Den Haag 2008. 16. Garcia L, Gold EB, Wang L, Yang X, Mao M, Schwartz AV. The relation of acculturation to overweight, obesity, pre-diabetes and diabetes among U.S. Mexican-American women and men. Ethn Dis 2012;22(1):58-64. 17. Oza-Frank R, Cunningham SA. The weight of US residence among immigrants: a systematic review. Obes Rev 2010;11(4):271-80. 18. Suurmond J, Stronks K, Foets M, Schuster J. Allochtone groepen in Nederland. Gezondheids(zorg) onderzoek onder allochtone bevolkingsgroepen.Amsterdam: Uitgeverij Aksant 2012; p. 15-28.

52

Sex disparities in acute myocardial infarction incidence by ethnic group

19. Nikiforov SV, Mamaev VB. The development of sex differences in cardiovascular disease mortality: a historical perspective. Am J Public Health 1998;88(9):1348-53. 20. Yusuf S, Reddy S, Ounpuu 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. 21. Agyemang C, Ujcic-Voortman J, Uitenbroek D, Foets M, Droomers M. Prevalence and management of hypertension among Turkish, Moroccan and native Dutch ethnic groups in Amsterdam, the Netherlands: The Amsterdam Health Monitor Survey. J Hypertens 2006;24(11):2169-76. 22. Dijkshoorn H, Nierkens V, Nicolaou M. Risk groups for overweight and obesity among Turkish and Moroccan migrants in The Netherlands. Public Health 2008;122(6):625-30. 23. Ujcic-Voortman JK, Bos G, Baan CA, Verhoeff AP, Seidell JC. Obesity and body fat distribution: ethnic differences and the role of socio-economic status. Obes Facts 2011;4(1):53-60. 24. Hosper K, Nierkens V, Nicolaou M, Stronks K. Behavioural risk factors in two generations of non- Western migrants: do trends converge towards the host population? Eur J Epidemiol 2007;22(3):163- 72. 25. Nierkens V, De Vries H, Stronks K. Smoking in immigrants: do socioeconomic gradients follow the pattern expected from the tobacco epidemic? Tob Control 2006;15(5):385-91. 26. Sinha DN, Palipudi KM, Rolle I, Asma S, Rinchen S. Tobacco use among youth and adults in member countries of South-East Asia region: review of findings from surveys under the Global Tobacco Surveillance System. Indian J Public Health 2011;55(3):169-76. 27. Teo KK, Ounpuu S, Hawken S, Pandey MR, Valentin V, Hunt D, et al. Tobacco use and risk of myocardial infarction in 52 countries in the INTERHEART study: a case-control study. Lancet 2006;368(9536):647-58. 28. Gerritsen AA, Deville WL. Gender differences in health and health care utilisation in various ethnic groups in the Netherlands: a cross-sectional study. BMC Public Health 2009;9:109. 29. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7. 30. Agyemang C, Nicolaou M, Boateng L, Dijkshoorn H, van de Born BJ, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Ghanaian population in Amsterdam, the Netherlands: the GHAIA study. Eur J Prev Cardiol 2013;20(6):938-46. 31. Reitsma JB, Kardaun JW, Gevers E, de Bruin A, van der Wal J, Bonsel GJ. [Possibilities for anonymous follow-up studies of patients in Dutch national medical registrations using the Municipal Population Register: a pilot study]. Ned Tijdschr Geneeskd 2003;147(46):2286-90. 32. De Bruin A, de Bruin E, Gast A, Kardaun J, van Sijl M, Verweij G. Koppeling van LMR- en GBA-gegevens: methode, resultaten en kwaliteitsonderzoek. Centraal Bureau voor de Statistiek Voorburg/Heerlen 2003. 33. Greenland S. The effect of misclassification in the presence of covariates. Am J Epidemiol 1980;112(4):564-9.

53

CHAPTER 2.3

Time trends in acute myocardial infarction incidence by country of birth

van Oeffelen AAM, Agyemang C, Koopman C, Stronks K, Bots ML, Vaartjes I. Downward trends in acute myocardial infarction incidence: how do migrants fare with the majority population? Results from a nationwide study.

Eur J Prev Cardiol 2013 [Epub ahead of print]. Chapter 2.3

ABSTRACT Background In previous decades a steep decline in acute myocardial infarction (AMI) incidence occurred in Western countries. We assessed whether this decline was also present among first generation ethnic minority groups (henceforth, migrant groups) living in the Netherlands.

Methods Nationwide registers were linked between 1998 and 2007. Poisson regression analyses were used to calculate the biannual percentage change (bi-APC) in AMI incidence within major non-Western migrant groups, and the differences in these changes with the ethnic Dutch population.

Results Within ethnic Dutch, AMI incidence significantly declined in men (bi-APC: -12%) and women (bi- APC:-9.5%). Incidence also declined among most migrant groups under study, ranging from -12% to -4.0% in men, and from -16% to -9.5% in women. Only in Turkish women and Moroccan men the AMI incidence remained stable over time (-0.3% and 2.8% respectively). There were no statistically significant trend differences between ethnic Dutch and the migrant groups under study. The higher AMI incidence in Turkish men and Surinamese men and women, and the lower AMI incidence in Moroccan men persisted over time.

Conclusion There was a declining AMI incidence rate within ethnic Dutch as well as within most of the major migrant groups living in the Netherlands, except in Turkish women and Moroccan men. Trend patterns in migrant groups did not significantly differ from ethnic Dutch. To reduce ethnic inequalities, primary preventive strategies should be targeted at those migrant groups with a persisting higher incidence.

56

Time trends in acute myocardial infarction incidence by country of birth

INTRODUCTION

In previous decades, the incidence and mortality of coronary heart disease (CHD) steadily declined in Western countries such as the Netherlands.1;2 This positive tendency may be the result of both improvements in primary preventive efforts (predominantly reductions in smoking prevalence, total cholesterol, and blood pressure) and medical treatments (drug use and surgical procedures).3-5 Although trend patterns in CHD are evident in Western populations at large, less is known about CHD trends in ethnic minority groups residing in those countries. As reducing health inequalities is an important goal of health policy, it is essential to monitor health over time in both the majority population and ethnic minority groups. Previous research has suggested that non-communicable disease incidence in ethnic minorities converges towards the level of the majority population over time.6 With respect to CHD mortality, the scant literature available mainly revealed smaller declines in ethnic minority groups, leading to a convergence towards the majority population in those with an initial lower risk, and to widening inequalities in those with an initial higher risk.7;8 With regard to CHD hospitalisations, decreasing, stable, and increasing trends were observed.9 However, detailed information regarding trends in CHD incidence (incorporating both mortality and hospitalisations) in migrant groups and changes in ethnic inequalities over time is lacking. This information is vital in order to design and implement targeted interventions to reduce CHD inequalities. Therefore, the aim of our study was to investigate time trends in acute myocardial infarction (AMI) incidence in ethnic Dutch and the major non-Western first generation ethnic minority groups (henceforth, migrant groups) living in the Netherlands between 1998 and 2007, and to assess differences in trends between these migrant groups and ethnic Dutch.

METHODS

Study population By linking several Dutch nationwide registers (i.e. Population Register, Hospital Discharge Register, Cause of Death Register, Regional Income Survey), provided by Statistics Netherlands, information regarding demographic factors, AMI hospitalisations, fatal AMI events, and comorbidities was obtained. The registers are described in detail previously.10 The overall quality of Dutch nationwide registers proved to be adequate.11 Dutch citizens who were registered in the Population Register between 1 January 1998 and 31 December 2007 were selected. Only those who were unique with respect to the combination of the variables date of birth, sex, and four digits of the postal code in the year of interest and in the prior three years were included. In case of non-uniqueness, persons could not be validly identified in the Hospital Discharge Register. About 30% of the persons did not meet this requirement and were excluded. As the outcome of interest was AMI, persons younger than 30 years of age were excluded.

57

Chapter 2.3

Finally, about half of the Dutch population was included for analyses. The inclusion procedure was performed for each year between 1998 and 2007 separately. Subsequently, five biannual time periods were assigned knowing 1998-1999, 2000-2001, 2002-2003, 2004-2005, and 2006-2007.

Outcome variable AMI incidence was defined as the first AMI event within each year under study (1998-2007), comprising a hospital admission with AMI as primary or secondary diagnosis (ICD-9 code 410), or a fatal event with AMI as primary or secondary cause of death (ICD-10 code I21). The validity of these ICD-codes proved to be good.12 Persons with a previous hospital admission for AMI (ICD-9 code 410) or old AMI (ICD-9 code 412) in the prior three years were excluded. The number of person-years at risk was calculated yearly, from the beginning of the year until the first AMI event, death, non- uniqueness, immigration, or the end of the year at 31 December, whichever came first.

Determinants Ethnic background Because of possible differences in time trends between non-Western migrants (first generation) and their offspring (second generation), and an often insufficient number of second generation ethnic minorities for analysis, only the first generation was included. Migrant groups were constructed based on the country of birth of the resident and his/her parents, according to the definition of Statistics Netherlands.13 A person was considered a migrant if he/she was born abroad and at least one of the parents was born abroad. The major non-Western migrant groups residing in the Netherlands were included, which are those born in Turkey, Suriname, Morocco, Indonesia, and the Netherlands Antilles (comprising about 5% of the Dutch population in 2007). Persons with both parents born in the Netherlands were indicated as ethnic Dutch. Persons born in the Netherlands, but with one or both parents born abroad (second generation ethnic minorities), were excluded.

Neighbourhood socioeconomic status Neighbourhood income was used as a proxy for socioeconomic status (SES), which was based on income data registered in the Regional Income Survey.14 Within each neighborhood, the mean disposable income on household level of all residents with income data available (about one third of the population) was calculated in the year prior to the year of interest. Subsequently, this income was assigned to all residents living in the same neighborhood on 1 January of the year of interest. Neighbourhood SES was divided into tertiles, with the first tertile representing the lowest income group.

58

Time trends in acute myocardial infarction incidence by country of birth

Comorbidity Presence and extent of comorbidity were determined with the Charlson index score,15 based on discharge diagnosis in the previous three years. The Charlson index score, ranging from zero to six (with zero representing no comorbidity), was dichotomised (0, ≥1). It proved to be a reliable and valid method to measure comorbidity in clinical research.16

Data analysis Absolute incidence rates were calculated as the number of first AMI events per 100,000 person- years at risk for every two-year period between 1998 and 2007, and were age-standardised (direct method) using the age distribution of the European population with ten-year age bands. All analyses were performed with Poisson regression modelling, stratified by sex, and adjusted for age category (30-44, 45-54, 55-64, 65-74, ≥75 years). The biannual percentage change in AMI incidence (the average percentage change in AMI incidence over a two-year period) was calculated in ethnic Dutch and the migrant groups under study. The interaction between ‘migrant group’ and ‘period’ was assessed to investigate whether there were significant trend differences between migrant groups and the ethnic Dutch population. Finally, the percentage difference in AMI incidence between ethnic Dutch and migrant groups was calculated in every two-year period. Additional models were built including the explanatory variables neighbourhood SES and Charlson comorbidity index. Results were expressed as percentage differences with accompanying 95% confidence intervals (95% CI). Absolute incidence rates were calculated with SPSS 14.0 (SPSS Inc, Chicago, Illinois, USA). Direct age-standardisation of incidence rates and Poisson regression analyses were executed in STATA version 11.0 (Stata Corp. 2009. STATA Statistical Software: Release 11. College Station, TX: StataCorp LP). All analyses were performed in accordance with privacy legislation Netherlands.

RESULTS

AMI incidence trends Surinamese migrants had the highest and Moroccan migrants had the lowest AMI incidence rate in both sexes (Table 1). Because the number of events in Moroccan women was lower than ten within most time periods, they had to be excluded for analyses due to privacy issues. The incidence rates were much higher in men than in women among all groups under study. AMI incidence declined over the years 1998 to 2007, except in Moroccan men and Turkish women where the AMI incidence remained stable over time (Figure 1). Declining trends were statistically significant in Indonesian, Surinamese, and Turkish men, and in Indonesian, Antillean, and Surinamese women (Table 2). Adjustment for neighbourhood SES and comorbidity did not alter results.

59

Chapter 2.3

b

81 80 31 83 95 88 IR 187 224 100 137 247 220

a

2007 - 86,634 61,198 27,271 91,944 56,654 34,995 98,216 100,217 108,451 123,026 6,697,739 7,209,958 PY at risk 2006

69 60 13 29 90 557 358 286 355 154 Events 24,243 15,044

and the ethnic Dutch population

b c

- 71 95 93 82 85 IR

218 288 231 319 240 124

a

2005 - 90171 32972 54,865 25,743 89,368 94,437 48,911 89,904 111,493 117,997 6,621,477 7,145,153 PY at risk 2004 Western migrant groups groups migrant Western -

52 69 29 77 668 390 296 367 <10 198 Events 27,505 17,055

b c

- 94 85 68 IR 254 312 220 307 242 111 119 131

years at risk in non -

a

2003 - 92,941 47,519 22,897 84,608 87,090 40,878 28,888 80,504 113,929 110,623 6,556,119 7,091,099 PY at risk 2002

36 56 27 58 705 390 297 486 <10 179 rate per 100,000 person

Events 31,303 19,500

b c

- 51 88 incidence IR 279 301 160 317 223 114 117 115 132 AMI

a

the age distribution of the European population 2001

- 95,648 39,063 19,946 77,751 77,676 33,172 24,800 69,772 to

116,978 102,437 6,399,689 6,953,348 PY at risk

2000 standardised -

26 38 31 54 710 350 242 453 <10 170 standardised Events 32,739 19,656

b c

- 98 81 IR 306 308 298 350 300 124 126 126 195 years at and risk, age - years at risk, at risk, years -

a

1999

- 98,888 32,723 17,907 70,844 69,800 27,433 22,164 94,151 61,101 120,580 6,300,354 6,871,683

PY at risk

1998

40 43 28 48 735 340 286 487 <10 195 years at risk Events 35,286 20,883

- year period between 1998 and 2007

Turkish Turkish Number of events, number of person

Antillean Antillean every two- Moroccon Moroccon Indonesian Indonesian

Surinamese Surinamese Ethnic Dutch Ethnic Dutch Men Women Number of person Numberof AMI events per 100,000 person Not given in line with the Dutch data protection guideline as the number of cases was lessthan 10.

Table 1 within a b c

60

Time trends in acute myocardial infarction incidence by country of birth

Indonesian men vs. ethnic Dutch men Indonesian women vs. ethnic Dutch women 350 350 300 300 250 250 200 200 150 150 100 100 50 50 0 0 98-99 00-01 02-03 04-05 06-07 98-99 00-01 02-03 04-05 06-07

Antillean men vs. ethnic Dutch men Antillean women vs. ethnic Dutch women 350 350 300 300 250 250 200 200 150 150 100 100 50 50 0 0 98-99 00-01 02-03 04-05 06-07 98-99 00-01 02-03 04-05 06-07

Surinamese men vs. ethnic Dutch men Surinamese women vs. ethnic Dutch women 350 350 300 300 250 250 200 200 150 150 100 100 50 50 0 0 98-99 00-01 02-03 04-05 06-07 98-99 00-01 02-03 04-05 06-07

Turkish men vs. ethnic Dutch men Turkish women vs. ethnic Dutch women 350 350 300 300 250 250 200 200 150 150 100 100 50 50 0 0 98-99 00-01 02-03 04-05 06-07 98-99 00-01 02-03 04-05 06-07

Moroccan men vs. ethnic Duch men 350 300 250 200 150 100 50 0 98-99 00-01 02-03 04-05 06-07

Figure 1 Age-standardised AMI incidence rate per 100,000 person-years at risk in ethnic Dutch vs. migrant groups over the years 1998-2007, stratified by sex Grey, migrant group; Black, ethnic Dutch population

61

Chapter 2.3

Table 2 Biannual percentage change in AMI incidence within non-Western migrant groups and the ethnic Dutch population between 1998 and 2007, and the significance of trend differences between migrant groups and ethnic Dutch Bi-APCa 95% CI P difference ethnic Dutchb Model 1c Model 2d Men Ethnic Dutch -12 -13;-10* Indonesian -7.2 -10;-4.2* 0.17 0.09 Moroccan 2.8 -11;19 0.16 0.09 Antillean -4.0 -12;5.2 0.37 0.28 Surinamese -7.1 -11;-3.0* 0.27 0.15 Turkish -10 -15;-5.4* 0.84 0.83 Women Ethnic Dutch -9.5 -12;-7.2* Indonesian -9.6 -13;-5.8* 0.95 0.79 Moroccan -e -e -e -e Antillean -16 -29;-0.2* 0.57 0.36 Surinamese -12 -18;-6.2* 0.48 0.29 Turkish -0.3 -10;11 0.30 0.11 a Biannual percentage change b Significance of difference in trend over the years 1998-2007 between migrant groups and the ethnic Dutch population c adjusted for age d adjusted for age, neighbourhood SES and Charlson index score e Not given in line with the Dutch data protection guideline as the number of cases was less than ten * Statistically significant trend in AMI incidence over the period 1998-2007 (p-value<0.05)

Trend differences between migrant groups and the ethnic Dutch population The trend differences in AMI incidence between migrant groups and ethnic Dutch were not statistically significant (Table 2). When we stratified results by time period, differences in AMI incidence between migrants and ethnic Dutch were in the same direction and with the same significance within most time periods. Surinamese men and women and Turkish men had a significantly higher incidence, whereas Moroccan men had a significantly lower incidence compared with ethnic Dutch during the entire study period. There were no differences between the other migrant groups and ethnic Dutch (Table 3). Although the significance of results was mainly the same over the different time periods, some small fluctuations were observed (Figure 2). Adjustment for neighbourhood SES and comorbidity did not alter results.

62

Time trends in acute myocardial infarction incidence by country of birth

Table 3 Percentage difference in AMI incidence with 95% CI in non-Western migrant groups compared with the ethnic Dutch population in every two-year time period between 1998 and 2007, adjusted for age category 1998-1999 2000-2001 2002-2003 2004-2005 2006-2007 % 95% CI % 95% CI % 95% CI % 95% CI % 95% CI Men Ethnic Dutch 0 0 0 0 0 Indonesian 2.5 -14;22 8.0 -8.8;28 14 -3.0;34 21 5.4;40* 17 -1.5;38 Moroccan -58 -80;-11* -74 -89;-38* -68 -84;-36* -54 -73;-25* -40 -63;-3.1* Antillean -13 -58;80 -29 -65;47 -4.5 -46;68 17 -25;82 2.8 -39;72 Surinamese 39 7.1;80* 38 8.9;76* 49 20;85* 60 33;92* 56 26;93* Turkish 40 5.1;85* 15 -14;54 32 2.9;69* 35 8.9;67* 33 4.6;68* Women Ethnic Dutch 0 0 0 0 0 Indonesian 2.3 -17;26 1.9 -16;24 11 -9.7;38 -5.1 -22;15 4.4 -15;28 Moroccan -a -a -a -a -a -a -a -a -57 -85;25 Antillean 14 -51;166 15 45;140 -15 -65;105 -10 -54;76 -8.6 -55;86 Surinamese 48 7.5;105* 24 -9.3;71 21 -14;70 42 10;84* 16 -15;59 Turkish -11 -54;70 -9.1 -48;59 -17 -55;51 8.8 -28;65 25 -17;87 a Not given in line with the Dutch data protection guideline as the number of cases was less than ten *Statistically significant difference compared with the ethnic Dutch population (p-value<0.05)

Indonesian Moroccan Antillean Surinamese Turkish

Figure 2 Percentage difference in AMI incidence between migrants and the ethnic Dutch population over the years 1998-2007, corrected for age category. Reference line: ethnic Dutch.

63

Chapter 2.3

DISCUSSION

Between 1998 and 2007, there was a decline in AMI incidence in men and women within the ethnic Dutch population, and within most major non-Western migrant groups living in the Netherlands. In contrast, among Turkish women and Moroccan men, AMI incidence rates remained stable over time. Trend patterns did not significantly differ between migrant groups and ethnic Dutch. The higher AMI incidence in Surinamese men and women and Turkish men, and the lower incidence in Moroccan men compared with the ethnic Dutch population remained during the entire study period between 1998 and 2007.

Discussion of key findings AMI incidence trends Our results are in line with previous studies showing predominantly declining CHD mortality trends in ethnic minority groups,7;8 but not with a study showing mainly stable CHD hospitalisation trends in migrant groups.9 The deviating results could be due to the fact that our incidence measure incorporated out-of-hospital mortality as well as hospitalisations. A recent Dutch study showed a greater decline in out-of-hospital AMI mortality than in AMI hospitalisations.1 Another time period (1991-1999 vs. 1998-2007), another host country (Sweden vs. the Netherlands), and other migrant groups (only Turkish migrants were included in both studies) could additionally provoke the differences in results. In our study, incidence declined in most non-Western migrant groups, but remained stable over time in Turkish women and Moroccan men. Trends in Moroccan women could not be presented due to the low number of events, but results are suggestive of stable trends over time, equivalent to their male counterparts. The absent decline in Turkish women and Moroccans in our study is presumably caused by a complex combination of factors. Firstly, Moroccans already had a very low AMI incidence, hampering a further decline. Secondly, Indonesia, Suriname, and the Netherlands Antilles are all former Dutch colonies, whereas Morocco and Turkey are not. Migrants from the colonies are presumably more integrated than Moroccan and Turkish migrants,17 who seem to have difficulties with the (especially the women).18 Poor communication complicates the information transmission between GP and patient.19;20 Indeed, a previous Dutch study reported more drug prescriptions in Moroccan and Turkish migrants, but a lower drug compliance.21;22 Improvements in primary preventive efforts and specialised care may therefore be less effective in those migrants. Thirdly, the distribution of risk factors differs across migrant groups. In the Netherlands, most improvements in primary preventive efforts recently accomplished were in in the fields of cardiovascular drug use, smoking cessation and lowering total cholesterol. However, efforts to tackle obesity and diabetes have not been successful yet, as prevalence rates are still rising.23 Obesity (especially in women) and diabetes are far more common in Moroccan and Turkish migrants than in the ethnic Dutch population.24-26

64

Time trends in acute myocardial infarction incidence by country of birth

Trend differences between migrant groups and the ethnic Dutch population Differences in CHD trends between ethnic minority groups and the majority population are rarely investigated. In studies from the USA, England and Wales, all ethnic minority groups with an initially lower CHD mortality converged to the CHD mortality rate of the majority population over time. In the ethnic minority groups with an initially higher CHD mortality, inequalities remained equal or widened.7;8 Overall this is comparable with our results. However, trend differences between migrant groups and ethnic Dutch in our study did not reach statistical significance. Consequently, ethnic inequalities in AMI incidence seem to remain stable over time, which indicates that improvements in primary preventive efforts might have similar effects on migrant groups as on the ethnic Dutch population. Yet, since governments set priority at tackling health inequalities, AMI incidence need to be lowered more extensively in those with a higher incidence compared with ethnic Dutch. Primary preventive strategies should therefore be targeted at Surinamese men and women, and Turkish men.

Considerations The major strength of our study is the ability to use nationwide registers, which enabled us to analyse time trends in AMI incidence in migrant groups, which is very rarely done. Therefore this study contributes importantly to current health inequality research. Inevitably our study has some limitations. Firstly, due to the young age distribution of most migrant groups, the number of events was sometimes small. This makes it difficult to obtain statistically significant results. AMI incidence rates should be monitored over a longer time period to see whether the potential trend differences become statistically significant when the population ages and the number of AMI events increase. Secondly, citizens who were not unique with respect to the combination of the variables sex, birth date, and four digits of the postal code were excluded.27 Non-uniqueness was more prevalent among migrant groups than among the ethnic Dutch population, but since non-uniqueness was not related to the health outcomes under study it is unlikely that it biased our findings. Thirdly, the analyses were adjusted for neighbourhood income as proxy for SES. We acknowledge the chance of non-differential misclassification when using a SES indicator on neighbourhood level instead of individual level. Therefore, it is possible that SES had more influence than could be demonstrated by our results. Fourthly, AMI hospitalisations outside the Netherlands could not be traced down, but we don’t anticipate that this will have a major influence on our results.

65

Chapter 2.3

Conclusion Between 1998 and 2007 a significant decline in AMI incidence was present within the ethnic Dutch population and within most major non-Western migrant groups, except in Moroccan men and Turkish women where AMI incidence remained stable over time. There was no significant difference in trend pattern between migrant groups and ethnic Dutch. The higher incidence in Surinamese men and women and Turkish men, and the lower incidence in Moroccan men in 1998 persisted until 2007. Primary preventive strategies should therefore pay extra attention to men born in Suriname and Turkey and to women born in Suriname. AMI incidence trends in migrants compared with ethnic Dutch should be closely monitored to evaluate its progress over time.

66

Time trends in acute myocardial infarction incidence by country of birth

REFERENCES 1. Koopman C, Bots ML, van Oeffelen AA, van Dis I, Verschuren WM, Engelfriet PM, et al. Population trends and inequalities in incidence and short-term outcome of acute myocardial infarction between 1998 and 2007. Int J Cardiol 2013;168(2):993-8. 2. Levi F, Chatenoud L, Bertuccio P, Lucchini F, Negri E, La Vecchia C. Mortality from cardiovascular and cerebrovascular diseases in Europe and other areas of the world: an update. Eur J Cardiovasc Prev Rehabil 2009;16(3):333-50. 3. Collart P, Coppieters Y, Leveque A. Trends in acute myocardial infarction treatment between 1998 and 2007 in a Belgian area (Charleroi). Eur J Prev Cardiol 2012;19(4):738-45. 4. Unal B, Critchley JA, Capewell S. Explaining the decline in coronary heart disease mortality in England and Wales between 1981 and 2000. Circulation 2004;109(9):1101-7. 5. Wijeysundera HC, Machado M, Farahati F, Wang X, Witteman W, van der Velde G, et al. Association of temporal trends in risk factors and treatment uptake with coronary heart disease mortality, 1994- 2005. JAMA 2010;303(18):1841-7. 6. Kunst A, Stronks K, Agyemang C. Non-communicable diseases. In: Rechel B, Mladovsky P, Devillé W, Rijks B, Petrova-Benedict R, McKee M, editors. Migration and health in the European Union. 1 ed. Open University Press; 2012. p. 116. 7. Cooper R, Cutler J, svigne-Nickens P, Fortmann SP, Friedman L, Havlik R, et al. Trends and disparities in coronary heart disease, stroke, and other cardiovascular diseases in the United States: findings of the national conference on cardiovascular disease prevention. Circulation 2000;102(25):3137-47. 8. Harding S, Rosato M, Teyhan A. Trends for coronary heart disease and stroke mortality among migrants in England and Wales, 1979-2003: slow declines notable for some groups. Heart 2008;94(4):463-70. 9. Gadd M, Johansson SE, Sundquist J, Wandell P. The trend of cardiovascular disease in immigrants in Sweden. Eur J Epidemiol 2005;20(9):755-60. 10. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 11. Harteloh P, de Bruin K, Kardaun J. The reliability of cause-of-death coding in The Netherlands. Eur J Epidemiol 2010;25(8):531-8. 12. Merry AH, Boer JM, Schouten LJ, Feskens EJ, Verschuren WM, Gorgels AP, et al. Validity of coronary heart diseases and heart failure based on hospital discharge and mortality data in the Netherlands using the cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009;24(5):237-47. 13. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69. 14. van Oeffelen AA, Agyemang C, Bots ML, Stronks K, Koopman C, van Rossem L., et al. The relation between socioeconomic status and short-term mortality after acute myocardial infarction persists in the elderly: results from a nationwide study. Eur J Epidemiol 2012;27(8):605-13. 15. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57(12):1288-94. 16. Groot de V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56(3):221-9. 17. Martinovic B, van Tubergen F, Maas I. Changes in immigrants' social integration during the stay in the host country: the case of non-Western immigrants in the Netherlands. Soc Sci Res 2013;38(4):870-82. 18. te Riele S. Marokkanen hebben minder moeite met Nederlands dan Turken [in Dutch]. Centraal Bureau voor Statistiek 2008.

67

Chapter 2.3

19. Harmsen JA, Bernsen RM, Bruijnzeels MA, Meeuwesen L. Patients' evaluation of quality of care in general practice: what are the cultural and linguistic barriers? Patient Educ Couns 2008;72(1):155-62. 20. Uiters E, Deville WL, Foets M, Groenewegen PP. Use of health care services by ethnic minorities in The Netherlands: do patterns differ? Eur J Public Health 2006;16(4):388-93. 21. Stronks K, Ravelli AC, Reijneveld SA. Immigrants in the Netherlands: equal access for equal needs? J Epidemiol Community Health 2001;55(10):701-7. 22. Uiters E, van Dijk L, Deville W, Foets M, Spreeuwenberg P, Groenewegen PP. Ethnic minorities and prescription medication; concordance between self-reports and medical records. BMC Health Serv Res 2006;6:115-21. 23. van Dis I, Poos M, Engelfriet P, Deckers J. Neemt het aantal mensen met een coronaire hartziekte toe of af? In: Volksgezondheid Toekomst Verkenning, Nationaal Kompas Volksgezondheid. Bilthoven:RIVM 2010. 24. Dijkshoorn H, Uitenbroek DG, Middelkoop BJ. [Prevalence of diabetes mellitus and cardiovascular disease among immigrants from Turkey and Morocco and the indigenous Dutch population]. Ned Tijdschr Geneeskd 2003;147(28):1362-6. 25. Hosper K, Nierkens V, Nicolaou M, Stronks K. Behavioural risk factors in two generations of non- Western migrants: do trends converge towards the host population? Eur J Epidemiol 2007;22(3):163- 72. 26. Ujcic-Voortman JK, Bos G, Baan CA, Verhoeff AP, Seidell JC. Obesity and body fat distribution: ethnic differences and the role of socio-economic status. Obes Facts 2011;4(1):53-60. 27. Reitsma JB, Kardaun JW, Gevers E, de Bruin A, van der Wal J, Bonsel GJ. [Possibilities for anonymous follow-up studies of patients in Dutch national medical registrations using the Municipal Population Register: a pilot study]. Ned Tijdschr Geneeskd 2003;147(46):2286-90.

68

CHAPTER 2.4

Incidence of first acute myocardial infarction over time by age, sex, and country of birth

van Oeffelen AAM, Agyemang C, Stronks K, Bots ML, Vaartjes I. Incidence of first acute myocardial infarction over time specific for age, sex, and country of birth.

Neth J Med 2014;72(1):20-7. Chapter 2.4

ABSTRACT Objectives To study the age- and sex-specific incidence rates of first acute myocardial infarction (AMI) among first generation ethnic minority groups (henceforth, migrant groups) and the ethnic Dutch population in the Netherlands during two time periods (2000-2004 and 2005-2010).

Methods Through linkage of Dutch nationwide registers, first AMI events in the ethnic Dutch population and the major migrant groups living in the Netherlands were identified from 2000-2004 and 2005-2010. Absolute incidence rates were calculated within each age-sex-period-country of birth group.

Results Regardless of ethnic background, AMI incidence rates were higher in men than in women and increased with age. Incidence significantly declined over time among the ethnic Dutch population (men: -26.8%, women: -26.7%) and among most migrant groups under study. Only in Moroccan migrants AMI incidence significantly increased over time (men: 25.2%, women: 41.7%). Trends differed between age categories, but did not show a consistent pattern. The higher AMI incidence in Surinamese men and women, and Turkish and Indonesian men compared with ethnic Dutch persisted over time, but decreased with age and became absent after 70 years of age. Moroccans had a significantly lower AMI incidence compared with ethnic Dutch during 2000-2004, which disappeared during 2005-2010.

Conclusion Primary preventive strategies should focus on Surinamese men and women and Turkish and Indonesian men below 70 years of age. Future research is necessary to unravel the factors that provoke the increasing AMI incidence over time among Moroccans.

70

Incidence of first acute myocardial infarction over time by age, sex, and country of birth

INTRODUCTION

Cardiovascular disease (CVD) is one of the main contributors to morbidity and mortality worldwide. In the Netherlands it is the number one cause of death in women and the number two in men, immediately after cancer. Ischemic heart disease (IHD), and in particular acute myocardial infarction (AMI), is responsible for the majority of CVD deaths.1 Information on age- and sex-specific incidence rates of AMI is vital in developing and maintaining preventive strategies. The latest detailed estimates of absolute AMI incidence in the Netherlands stratified by age and sex date back to 2000, and were restricted to the general Dutch population.2 As a decline in AMI incidence has taken place in the Netherlands over the past decade, there is a need for updated AMI incidence estimates.3 Since 10% of the Dutch population was born abroad, and ethnic variations in AMI incidence have been reported, it is important to present results specific for country of birth.4-7 Internationally, only few studies have described absolute age- and sex-specific incidence rates in ethnic minority groups.8-11 Results of these studies suggest that ethnic differences in AMI incidence are particularly present in the young, and that AMI incidence is higher in men than in women irrespective of age and country of birth. In the Netherlands, such estimates are unavailable yet. It is well known that AMI incidence has been declining over time in Western countries.3;12 However, time trends among ethnic minority groups have not been widely investigated, and age- and sex-specific time trends are even more limited.11;13-16 The one study reporting such extensively stratified data showed a larger decline in AMI incidence among White Americans than among African Americans, especially in men and elderly women.11 This information is important to monitor whether preventive strategies have gained effect, and in which groups this occurred. Furthermore, it may target specific groups that need extra attention in future preventive strategies. Therefore this study presents the absolute incidence rate of age- and sex-specific first AMI events (hospitalisations and out-of-hospital deaths) during two time periods (2000-2004 and 2005-2010) within the ethnic Dutch population and within first generation ethnic minority groups (henceforth, migrant groups) living in the Netherlands.

METHODS

Data sources and enrolment New cases of first AMI events in the Dutch population were identified during two time periods: 2000-2004 (period one) and 2005-2010 (period two). This enabled us to investigate trend patterns in AMI incidence during the past decade, while keeping numbers high enough for analyses. First AMI events included hospitalised first AMI patients and out-of-hospital deaths caused by a first AMI. Data on AMI hospitalisations were derived from the Dutch National Hospital Discharge Register (HDR). Data on out-of-hospital deaths caused by AMI were derived from the National Cause of Death Register (CDR). Demographic data (date of birth, country of birth, parental country of birth, sex) were derived from the Dutch Population Register (PR), which contains information on all

71

Chapter 2.4 officially registered persons living in the Netherlands. The registers have been described in detail previously.17 The overall quality of Dutch nationwide registers proved to be adequate.18;19 As the nominator, all persons with a first AMI event in period one and period two were included. A first AMI comprised all first hospitalisations with AMI as principal or secondary diagnosis (ICD-9 code 410) and all out-of-hospital deaths with AMI as primary or secondary cause (ICD-10 code I21). Persons who suffered a previous hospitalisation with AMI as the principal or secondary diagnosis from 1995 onwards were excluded. As denominator, person-years at risk were calculated based on all unique persons in the PR during the years 2000-2010 (if a person was unique in only a part of these years, person-years at risk were based on these years). A person was ascribed PR unique when there was a unique combination of the variables date of birth, sex, and four digits of postal code in the year of interest. In case of non-uniqueness a person could not be validly tracked down in the HDR and had to be excluded.

Ethnic background Only first generation ethnic minority groups (henceforth, migrant groups) were included. Migrant groups were constructed based on the country of birth of the resident and his/her parents, according to the definition of Statistics Netherlands.20 A person is considered a migrant if he/she was born abroad and at least one of the parents was born abroad. Persons in which both parents were born in the Netherlands were indicated as ethnic Dutch. For this study, individuals born in Suriname, Morocco, Turkey, Netherlands Antilles, and Indonesia were included (the five major migrant groups living in the Netherlands).

Data analysis AMI incidence rates in period one and two with 95% confidence intervals (95% CI) were computed by age (30-39, 40-49, 50-59, 60-69, 70-79, 80-89, ≥90 years) and sex within the ethnic Dutch population and within the migrant groups under study. The incidence rates were expressed as number of events per 100,000 person-years at risk. Subsequently, age-standardised incidence rates were calculated using the age distribution of the European population in ten-year age bands. The percent change in AMI incidence rate over the two time periods was calculated within each age-sex- country of birth group. To study the differences between migrant groups and ethnic Dutch, Relative Risks (RR) were calculated within each age-sex-period group (reference: ethnic Dutch). We used SPSS software, version 14.0 (SPSS Inc, Chicago, Illinois, USA) to calculate the number of events and person-years at risk. Incidence rates with 95% CI were calculated with the online program openepi.com.21 Age-standardised incidence rates with 95% confidence intervals were calculated with STATA 11.0 (Stata Corp. 2009. STATA Statistical Software: Release 11. College Station, TX:

72

Incidence of first acute myocardial infarction over time by age, sex, and country of birth

StataCorp LP). Relative risks and percent changes were calculated using Microsoft Excel 2010. All analyses were performed in accordance with privacy legislation of the Netherlands.

RESULTS

From 2000-2004, the mean number of first AMI events was 25,070 per year; from 2005-2010 this number decreased to 18,507 per year. Among migrants, the Antilleans were the smallest and the Surinamese the largest group (Supplementary table 1 and 2). Men had the highest AMI incidence rate within all country of birth groups, age strata, and time periods (Table 1 and 2). Incidence increased with age in all groups and time periods under study.

Trends in AMI incidence among migrants and the ethnic Dutch population There was a decline in AMI incidence over time among the ethnic Dutch population, among Surinamese and Indonesian men and women, and among Turkish and Antillean men (Figure 1). Among Moroccans incidence increased over time, whereas it remained stable among Turkish and Antillean women. After age stratification, the direction of time trends was often similar between age categories, with some variation in magnitude (Figure 2). An exception was in Moroccan men, where age stratified results showed decreased, increased, as well as stable AMI incidence rates over time.

Difference in AMI incidence between migrants and the ethnic Dutch population Surinamese men and women, as well as Turkish and Indonesian men, had a statistically significantly higher AMI incidence compared with their ethnic Dutch counterparts, which remained stable over time (Table 1 and 2). After age stratification, the higher incidence decreased with age and only remained in those younger than 70 years of age. Moroccans had a statistically significantly lower AMI incidence compared with their ethnic Dutch counterparts during period one, whereas there was no difference observed anymore during period two. Among Moroccan men this was mainly due to the fact that the lower incidence in 50-80 year olds disappeared over time. Among Antillean men and women, and Turkish and Indonesian women, there was no difference in incidence with ethnic Dutch. After age stratification, however, AMI incidence was higher in some age groups among Turkish and Indonesian women, but in period two only (Table 2).

73

Chapter 2.4

- - - - RR 1.0 0.9 1.2 1.0 1.1 1.1 1.0 1.0 0.9 1.0 1.6* 1.1* 1.1* 2.3* 1.1* 2.0*

1561) 3280) 1639) 278) 429) 706) 604) 250) 275) 583) 317) - - - 94) 1093) 66) 1063) ------109) - - -

-

b b b b Indonesian - - - - IR

44 (28 88 (81 86 (68 229 (187 383 (342 642 (582 574 (546 235 (220 236 (201 523 (468 297 (279 996 (906 952 (850 1372 (1201 2424 (1748 1335 (1076

------RR 0.9 1.2 1.0 1.1 1.0 1.0 1.0 1.4 0.6 1.1 0.8 0.5* 0.3*

228) 417) 854) 143) 234) 539) 558) 1888) 55) 1577) 63) 88) ------130) - - - -

- Antillean

b b b b b b b ------IR (158

29 (13 51 (40 71 (53 82 (48 172 (127 322 (243 624 (444 123 (106 196 302 (200 304 (148 946 (526 1059 (538

- - - - RR 1.2 1.1 1.3 1.0 0.8 0.8 1.1 1.0 2.0* 2.0* 1.7* 1.2* 0.7* 0.9* 1.1* 0.3*

326) 651) 819) 950) 236) 248) 281) 630) 4297) 20) 76) 70) 1633) 56) 142) ------109) Turkish - - - - -

-

b b b b - - - - IR

61 (48 48 (33 48 (41 12 (6.2 86 (64 106 (78 283 (244 571 (499 726 (641 681 (474 220 (205 228 (208 213 (158 420 (267 785 (318 1579 (402 sex group in theethnic Dutch population and inthe major migrant groups living in the

-

------RR 1.0 0.8 1.1 0.6* 0.5* 0.4* 0.3* 0.5* 0.1* 0.6*

247) 602) 643) 358) 46) 82) 23) 65) 131) - - - -

113) - - - - -

-

b b b b b b b b b b Moroccan ------IR (51 years at risk in every age - 31 (21 71 (61 17 (12 48 (30 year age bands 78 -

111 (92 179 (126 473 (366 392 (223 225 (134

- - - RR 1.1 1.6 1.2 1.2 1.0 0.7

2.6* 2.3* 1.7* 1.2* 0.8* 1.2* 1.3* 1.9* 1.4* 0.7* 1.3*

2155) 357) 641) 807) 976) 312) 292) 188) 372) 719) 1890) - 25) 69) 1263) 117) 120) ------103) - - - -

-

b b b - - - Surinamese IR (124 53 (41 16 (9.3 81 (63 the European population in ten 106 (96 108 (96

318 (281 573 (511 698 (600 782 (618 293 (275 270 (247 154 304 (247 590 (479

978 (744 1060 (539 -years risk at

1609 (1175

1528) 2067) 1467) 145) 340) 590) 250) 206) 217) 509) 148) - - - 1.6) 0.7) 33) 1007) 45) 84) 1002) 87) ------11) ------

-

IR r 100,000 person (1878 10 (9 31 (30 43 (41 81 (78 86 (85 1.4 (1.1 0.5 (0.4 Ethnic Dutch

141 (137 334 (328 581 (572 248 (246 205 (203 211 (206 500 (491 147 (145 993 (979 986 (969 1500 (1472 1971 1423 (1380

a a Incidence ofrate acute myocardial infarction per 100,000 person 39 49 59 69 79 89 39 49 59 69 79 89 ------<30 ≥90 <30 ≥90

tand tand 30 40 50 60 70 80 30 40 50 60 70 80 S S All ages All ages Men Women Significant difference compared with the ethnic Dutch population Not given in line with the Dutch data protection guideline as the number of cases was less than ten Standardised to the age distribution of Netherlands between 2000 and 2004 Table 1 IR=Incidence rate pe RR=Relative Risk of AMI incidence compared with the ethnic Dutch population Dutch ethnic the with compared incidence AMI of Risk RR=Relative a a b *

74

Incidence of first acute myocardial infarction over time by age, sex, and country of birth

- - - - RR 1.1 1.1 1.0 1.0 1.1 1.0 0.9 1.1 1.5* 1.4* 1.2* 2.6* 1.2* 1.2* 1.1* 2.2*

1321) 2088) 243) 403) 551) 784) 544) 192) 207) 448) 815) 1223) 273) - - 62) 95) 72) ------

b b b b Indonesian - - - - IR

37 (20 76 (60 67 (61 179 (128 361 (323 498 (450 711 (643 517 (491 180 (168 175 (147 400 (355 732 (656 256 (240 1002 (813 1180 (1051 1565 (1147

------RR 1.4 0.8 1.0 1.1 0.8 0.9 0.7 0.8 0.7 1.4 1.0 0.6* 0.4*

329) 593) 843) 136) 154) 425) 65) 78) 1621) 52) 75) 174) ------Antillean 132) ------

-

b b b b b b b ------IR 39 (22 48 (28 41 (33 60 (44 95 (66 111 (67 261 (205 464 (357 530 (314 119 (103 132 (110 260 (148 989 (563 populationand inthe major migrant groups living in the

- - - - - RR 1.0 1.1 1.0 0.8 0.9 1.1 Dutch 1.7* 1.9* 1.5* 1.2* 0.8* 1.1* 1.6* 1.5* 0.5*

ethnic

247) 462) 569) 640) 209) 183) 266) 378) 16) 60) 51) 1152) 66) 81) 144) ------Turkish ------

b b b b b - - - - - IR

47 (37 38 (28 58 (51 70 (59 10 (5.5 112 (86 220 (195 403 (350 501 (439 524 (424 196 (183 170 (156 216 (173 275 (194 691 (384 sex group in the

-

------RR 0.9 1.1 1.1 0.8 0.9 0.6 1.5 1.2 1.1 0.6* 0.7* 0.5* 0.5* 0.2*

bands 317) 542) 961) 155) 326) 635) 25) 1540) 34) 79) 32) 87) 113) ------

102) ------

-

b b b b b b Moroccan cases was less than ten than less was cases ------

IR (194 years at risk in every age - year age 19 (10 46 (25 26 (21 68 (49 16 (9.6 79 (60 -

101 (90 250 443 (358 760 (593 139 (122 221 (143 423 (269 906 (491

- - - - RR 0.9 0.9 1.4 1.3 1.1

3.3* 2.1* 1.7* 1.4* 1.4* 1.4* 1.6* 1.6* 0.8* 0.8* 1.2*

3870) 268) 506) 640) 696) 304) 225) 282) 483) 723) 1352) - 59) 139) 81) ------114) 105) - - -

- -

b b b b - - - - Surinamese IR

(67 47 (36 74 91 (72 96 (88 the European population in ten 116 (97

238 (209 458 (415 565 (497 571 (464 287 (270 211 (197 236 (196 398 (324 555 (417

risk years at 1013 (742 -

2227 (1171

1127) 1632) 1173) 119) 270) 414) 675) 200) 151) 10) 149) 357) 745) 118) - - - 1.1) 0.7) 30) 39) 75) 64) ------distribution of ------

-

IR (141 r 100,000 person 28 (27 37 (36 72 (69 63 (63 9.1 (8.2 0.9 (0.8 0.6 (0.5 Ethnic Dutch

116 (113 266 (261 408 (401 665 (654 199 (197 150 (148 145 350 (343 732 (720 117 (116 1106 (1086 1561 (1491 1140 (1107

a a Incidence ofrate acute myocardial infarction per 100,000 person 39 49 59 69 79 89 39 49 59 69 79 89 ------<30 ≥90 <30 ≥90

tand tand 30 40 50 60 70 80 30 40 50 60 70 80 S S All ages All ages Men Women Significant difference compared with the ethnic Dutch population Not given in line with the Dutch data protection guideline as the number of Standardised to the age Netherlands between 2005 and 2010 Table 2 a a b

RR=Relative Risk of AMI incidence compared with the ethnic Dutch population Dutch ethnic the with compared incidence AMI of Risk RR=Relative IR=Incidence rate pe *

75

Chapter 2.4

Men 300

250 Majority population 200 Surinamese migrants years at risk at years - 150 Moroccan migrants Turkish migrants 100 Antillean migrants

standardised AMI incidence rate incidence AMI standardised Indonesian migrants - 50 per 100,000 person 100,000 per Age 0 Period 1 Period 2

Women 120

100 Majority population 80 Surinamese migrants years at risk at years - 60 Moroccan migrants Turkish migrants 40 Antillean migrants

standardised AMI incidence rate incidence AMI standardised Indonesian migrants - 20 per 100,000 person 100,000 per Age 0 Period 1 Period 2

Figure 1 Age-standardised AMI incidence rate in migrants groups and ethnic Dutch during period one and period two Period 1: 2000-2004 Period 2: 2005-2010

76

Incidence of first acute myocardial infarction over time by age, sex, and country of birth

Age- and sex-specific AMI incidence rate in the ethnic Dutch population 2500

2000

1500 Men period 1 Men period 2 1000 Women period 1 500 Women period 2

0 <30 30-39 40-49 50-59 60-69 70-79 80-89 ≥90 Age

% change between period 1 and period 2 Total Men -35.7* -9.7* -17.7* -20.3* -29.8* -33.0* -26.3* -20.8* -26.8* Women 20.0 -9.0* -14.0* -11.0* -31.3* -30.0* -25.8* -19.9* -26.7*

Age- and sex-specific AMI incidence rate in Surinamese migrants

2500

2000

1500 Men period 1

Men period 2 1000 Women period 1 500 Women period 2 0 <30 30-39 40-49 50-59 60-69 70-79 80-89 ≥90 Age % change between period 1 and period 2 Total Men 12.2 -25.0* -20.1* -19.1* -27.0* -37.0* -21.9* Women -12.0 -24.5* -22.5* -32.5* -43.2* -31.5*

Age- and sex-specific AMI incidence rate in Moroccan migrants

1000 900 800 700 600 Men period 1 500 Men period 2 400 300 Women period 1 200 Women period 2 100 0 <30 30-39 40-49 50-59 60-69 70-79 80-89 ≥90 Age

% change between period 1 and period 2 Total Men -48.8* 1.9 39.8* -6.3 93.7 +25.2* Women -2.0 +41.7*

Figure 2 Age- and sex-specific AMI incidence rate per 100,000 person-years at risk during period one and period two by country of birth. * Significant difference between period one and period two.

77

Chapter 2.4

Age- and sex-specific AMI incidence rate in Turkish migrants 1800 1600 1400 1200 1000 Men period 1 800 Men period 2

600 Women period 1 400 Women period 2 200

0 <30 30-39 40-49 50-59 60-69 70-79 80-89 ≥90

Age

% change between period 1 and period 2 Total Men -22.4 -22.1* -29.4* -31.0* -23.1 -34.1* Women -13.1 -21.6 5.3 1.5 -34.4 -12.0 -18.6

Age- and sex-specific AMI incidence rate in Antillean migrants 1200

1000

800 Men period 1 600 Men period 2 400 Women period 1

200 Women period 2

0 <30 30-39 40-49 50-59 60-69 70-79 80-89 ≥90 Age

% change between period 1 and period 2 Total Men 35.7 -44.8* -18.8 -25.7 -44.0 -32.7* Women -41.2* -63.3 -14.5 -6.6 -15.5

Age- and sex-specific AMI incidence rate in Indonesian migrants

3000 2500

2000 Men period 1 1500 Men period 2 1000 Women period 1 500 Women period 2

0 <30 30-39 40-49 50-59 60-69 70-79 80-89 ≥90

Age

% change between period 1 and period 2 Total Men -21.8 -5.9 -22.4* -28.6* -14.0* -35.4* -23.4 Women -15.9 -12.0 -25.8* -23.6* -23.1* -25.0* -23.9*

Figure 2 Continued

78

Incidence of first acute myocardial infarction over time by age, sex, and country of birth

DISCUSSION

In every group under study AMI incidence increased with age and was higher in men than in women. All migrant groups, except Moroccans, showed a decreased or stable AMI incidence over time. The extent varied between age categories, depending on the country of birth. Among migrant groups with a higher incidence compared with ethnic Dutch, the difference diminished with increasing age, and disappeared in those older than 70 years of age. Among migrant groups with a similar or lower incidence compared with ethnic Dutch, there was no clear pattern over age categories. Among most age-sex-migrant groups, the difference with ethnic Dutch remained similar over time.

Discussion of key findings There are some important findings that need to be addressed. Firstly, in migrant groups with a higher overall AMI incidence compared with ethnic Dutch (Surinamese men and women, Turkish and Indonesian men), the difference declined with increasing age. After the age of 70 years, a difference was no longer observed. This is in accordance with international literature studying aboriginal vs. non-aboriginal subjects in an Australian population, South-Asian vs. White subjects in a Canadian population, and African-American vs. White American subjects in a US population.8;10;22 Explanations for this phenomenon may include selective survival, which prevents those in the higher risk groups from reaching old age. Subsequently, elderly migrant groups are healthier compared with their younger counterparts, resulting in diminishing ethnic inequalities in AMI incidence among the elderly.23 Literature also indicates that cardiovascular risk factors present themselves at an earlier age in migrants than in the majority population, which could lead to a shift in AMI incidence towards the young. For example, Surinamese women already had a hypertension prevalence of 35% at age 35, whereas their ethnic Dutch counterparts did not reach this level before the age of 45.24 Similarly, another study showed that the prevalence of diabetes in the ethnic Dutch population started to increase from 55-64 years of age, while in Turkish and Moroccan migrants the same prevalence was already reached 10-20 years earlier.25 Secondly, Moroccans had a lower overall AMI incidence compared with ethnic Dutch in period one, whereas this difference disappeared in period two. In men this was caused by increasing incidence rates over time, especially in the 50-60 and the 70-80 year olds. In women, number of events was too small to distinguish time trends across age groups. One explanation for the increased AMI incidence among Moroccans is the very low initial AMI incidence rate, which may hinder the trend of further decline. Moreover, improvements in primary preventive efforts over the past years may have had less effect in Moroccans due to cultural and language barriers.26-28 In addition, most improvements in the Netherlands were accomplished in the fields of cardiovascular drug use, smoking cessation and lowering total cholesterol, but not in reducing the major risk factors Moroccans have to deal with (obesity and diabetes).29 Finally, previous literature indicates that

79

Chapter 2.4 migrants with an initially lower risk converge towards the risk of the majority population over time, provoked by the loss of healthy lifestyle factors and adoption of adverse lifestyle factors in the host country.15;30 This was reflected in our study, since the AMI risk difference between Moroccans and ethnic Dutch narrowed from 0.5 to 0.9 in men, and from 0.6 to 1.1 in women between the two time periods (Figure 1). The underlying factors of this adverse trend in AMI incidence among Moroccans need further attention in future research. Thirdly, Turkish and Indonesian women had a similar overall age-standardised AMI incidence rate compared with ethnic Dutch. However, after age stratification, there was a significantly higher incidence observed in 50- to 70-year-old Turkish women and 60- to 80-year-old Indonesian women in period two. This was not observed in period one. The higher incidence in Turkish women in period two was provoked by the absent decline in AMI incidence over time, especially in the 50- to 70-year-olds (Figure 2). It is unclear why specifically these Turkish women show an adverse picture. Among Indonesian women between 60 and 80 years the significant difference with ethnic Dutch in period two was only small and less relevant.

Considerations Literature concerning absolute incidence rates of CHD by age, sex, and country of origin is scarce, and in most cases reported results for one time period only.8-10;22 Our study, stratifying by age, sex, country of origin, and time period, expands existing evidence. The nationwide registers yielded a large study population which made it possible to stratify by a wide range of determinants. The inclusion of primary as well as secondary diagnosis and causes of death decreased the chance of missed AMI events. The availability of hospital data from 1995 onwards provided a medical history varying from five to 9 years in the first period and from ten to 15 years in the second period. This long medical history diminished the risk of misclassifying recurrent AMI events as first AMI events. Previous literature reported a five-year risk on recurrent AMI of 2.5% in both men and women, and a 12-year risk of 1.5% in men and 1.0% in women.31 Because of the shorter medical history in period one compared with period two, incidence rates could have been slightly overestimated during period one. However, this overestimation will be minimal since the risk on a recurrent AMI only differed with one percent between a five- and a 12-year medical history. Inevitably our study has some limitations. Firstly, incidence rates may have been underestimated because of misclassification of AMI events. However, validity of AMI registration in the HDR proved to be high with a positive predictive value of 97% and a sensitivity of 84%.19 This means that 97% of all AMI cases in the HDR were correctly coded and that 84% of all AMI events in the Netherlands were registered in the HDR. Furthermore, the validity of AMI death in de cause of death register proved to be one of the highest of all causes, with a maximal misclassification of 10%.18 By additionally including the secondary diagnosis and causes of death, misclassification was

80

Incidence of first acute myocardial infarction over time by age, sex, and country of birth further limited. Secondly, persons who were not PR unique with respect to the combination of the variables date of birth, sex, and four digits of the postal code were excluded. The migrant groups under study have a higher risk on non-uniqueness, mainly due to the absence of the exact date of birth. Therefore, person-years at risk is underestimated, but since non-uniqueness was not related to AMI incidence it did not influence the absolute incidence rates.32;33 Thirdly, a small number of the subjects could not be traced down completely between 1995-2000/2010 because they were not PR unique during the entire study period, or because they immigrated to the Netherlands during the study period. This could have led to a slight overestimation of first AMI events. Migrants are more likely to have migrated to the Netherlands during the study period, but since the majority of those who migrated after 1995 were too young to have suffered an AMI, differential overestimation of first AMI events is unlikely. Fourthly, from 2005 onwards, hospitals are no longer obliged to register in the HDR, which has led to about 10% missing AMI events between 2005 and 2010.34 Subsequently, absolute incidence rates in the second period have been underestimated and declines over time may have been overestimated by maximally 10%. However, this percentage is not high enough to have influenced our final conclusions concerning trends in AMI incidence over time. The risks relative to the ethnic Dutch have also remained unaffected, since missing AMI events were evenly distributed across the entire Netherlands.

Conclusion Regardless of ethnic background, AMI incidence increased with age and was higher among men than among women. All migrant groups, except Moroccans, showed a stable or declining AMI incidence over time. With respect to primary preventive strategies, health care professionals should focus on Surinamese men and women and Turkish and Indonesian men below 70 years of age, because of their high AMI incidence relative to the ethnic Dutch population. Future research should elucidate the factors that provoke the increasing AMI incidence over time among Moroccans.

81

Chapter 2.4

REFERENCES 1. Vaartjes I, van Dis I, Visseren F, Bots M. Hart- en vaatziekten in Nederland. Hart- en vaatziekten in Nederland 2010, cijfers over leefstijl- en risicofactoren, ziekte en sterfte. Den Haag: Hartstichting; 2010. 2. Koek HL, de BA, Gast A, Gevers E, Kardaun JW, Reitsma JB, et al. Incidence of first acute myocardial infarction in the Netherlands. Neth J Med 2007;65(11):434-41. 3. Koopman C, Bots ML, van Oeffelen AA, van Dis I, Verschuren WM, Engelfriet PM, et al. Population trends and inequalities in incidence and short-term outcome of acute myocardial infarction between 1998 and 2007. Int J Cardiol 2013;168(2):993-8. 4. Gadd M, Johansson SE, Sundquist J, Wandell P. Morbidity in cardiovascular diseases in immigrants in Sweden. J Intern Med 2003;254(3):236-43. 5. Hedlund E, Lange A, Hammar N. Acute myocardial infarction incidence in immigrants to Sweden. Country of birth, time since immigration, and time trends over 20 years. Eur J Epidemiol 2007;22(8):493-503. 6. Sundquist K, Li X. Coronary heart disease risks in first- and second-generation immigrants in Sweden: a follow-up study. J Intern Med 2006;259(4):418-27. 7. van Oeffelen AA, Vaartjes I, Stronks K, Bots ML, Agyemang C. Incidence of acute myocardial infarction in first and second generation minority groups: Does the second generation converge towards the majority population? Int J Cardiol 2013;168(6):5422-9. 8. Bradshaw PJ, Alfonso HS, Finn J, Owen J, Thompson PL. A comparison of coronary heart disease event rates among urban Australian Aboriginal people and a matched non-Aboriginal population. J Epidemiol Community Health 2011;65(4):315-9. 9. Fischbacher CM, Bhopal R, Povey C, Steiner M, Chalmers J, Mueller G, et al. Record linked retrospective cohort study of 4.6 million people exploring ethnic variations in disease: myocardial infarction in South Asians. BMC Public Health 2007;7:142. 10. Nijjar AP, Wang H, Quan H, Khan NA. Ethnic and sex differences in the incidence of hospitalized acute myocardial infarction: British Columbia, Canada 1995-2002. BMC Cardiovasc Disord 2010;10:38. 11. Wang H, Steffen LM, Jacobs DR, Zhou X, Blackburn H, Berger AK, et al. Trends in cardiovascular risk factor levels in the Minnesota Heart Survey (1980-2002) as compared with the National Health and Nutrition Examination Survey (1976-2002): A partial explanation for Minnesota's low cardiovascular disease mortality? Am J Epidemiol 2011;173(5):526-38. 12. Levi F, Chatenoud L, Bertuccio P, Lucchini F, Negri E, La Vecchia C. Mortality from cardiovascular and cerebrovascular diseases in Europe and other areas of the world: an update. Eur J Cardiovasc Prev Rehabil 2009;16(3):333-50. 13. Cooper R, Cutler J, svigne-Nickens P, Fortmann SP, Friedman L, Havlik R, et al. Trends and disparities in coronary heart disease, stroke, and other cardiovascular diseases in the United States: findings of the national conference on cardiovascular disease prevention. Circulation 2000;102(25):3137-47. 14. Gadd M, Johansson SE, Sundquist J, Wandell P. The trend of cardiovascular disease in immigrants in Sweden. Eur J Epidemiol 2005;20(9):755-60. 15. Harding S, Rosato M, Teyhan A. Trends for coronary heart disease and stroke mortality among migrants in England and Wales, 1979-2003: slow declines notable for some groups. Heart 2008;94(4):463-70. 16. van Oeffelen AAM, Agyemang C, Koopman C, Stronks K, Bots M, Vaartjes I. Downward trends in acute myocardial infarction incidence: how do migrants fare with the majority population? Results from a nationwide study. Eur J Prev Cardiol 2013 [Epub ahead of print].

82

Incidence of first acute myocardial infarction over time by age, sex, and country of birth

17. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 18. Harteloh P, de Bruin K, Kardaun J. The reliability of cause-of-death coding in The Netherlands. Eur J Epidemiol 2010;25(8):531-8. 19. Merry AH, Boer JM, Schouten LJ, Feskens EJ, Verschuren WM, Gorgels AP, et al. Validity of coronary heart diseases and heart failure based on hospital discharge and mortality data in the Netherlands using the cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009;24(5):237-47. 20. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69. 21. Dean AG, Sullivan KM, Soe MM. OpenEpi: Open Source Epidemiologic Statistics for Public Health, Version. www.OpenEpi.com, updated 2013/04/06, accessed 2013/06/26. 22. Wang OJ, Wang Y, Chen J, Krumholz HM. Recent trends in hospitalization for acute myocardial infarction. Am J Cardiol 2012;109(11):1589-93. 23. Markides KS, Machalek R. Selective survival, aging and society. Arch Gerontol Geriatr 1984;3(3):207- 22. 24. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7. 25. Ujcic-Voortman JK, Schram MT, Jacobs-van der Bruggen MA, Verhoeff AP, Baan CA. Diabetes prevalence and risk factors among ethnic minorities. Eur J Public Health 2009;19(5):511-5. 26. Harmsen JA, Bernsen RM, Bruijnzeels MA, Meeuwesen L. Patients' evaluation of quality of care in general practice: what are the cultural and linguistic barriers? Patient Educ Couns 2008;72(1):155-62. 27. te Riele S. Marokkanen hebben minder moeite met Nederlands dan Turken [in Dutch]. Centraal Bureau voor Statistiek 2008. 28. Uiters E, van DL, Deville W, Foets M, Spreeuwenberg P, Groenewegen PP. Ethnic minorities and prescription medication; concordance between self-reports and medical records. BMC Health Serv Res 2006;6:115-21. 29. van Dis I, Poos M, Engelfriet P, Deckers J. Neemt het aantal mensen met een coronaire hartziekte toe of af? In: Volksgezondheid Toekomst Verkenning, Nationaal Kompas Volksgezondheid. Bilthoven:RIVM 2010. 30. Kunst A, Stronks K, Agyemang C. Non-communicable diseases. In: Rechel B, Mladovsky P, Devillé W, Rijks B, Petrova-Benedict R, McKee M, editors. Migration and health in the European Union. 1 ed. Open University Press; 2012. p. 116. 31. Osler M, Rostgaard K, Sorensen TI, Madsen M. The effect of recurrent events on register-based estimates of level and trends in incidence of acute myocardial infarction. J Clin Epidemiol 1999;52(7):595-600. 32. De Bruin A, de Bruin E, Gast A, Kardaun J, van Sijl M, Verweij G. Koppeling van LMR- en GBA-gegevens: methode, resultaten en kwaliteitsonderzoek. Voorburg/Heerlen: Centraal Bureau voor de Statistiek; 2003. 33. De Bruin A, Kardaun J, Gast A, de Bruin E, van Sijl M, Verweij G. Record linkage of hospital discharge register with population register: experiences at Statistics Netherlands. Statistical Journal of the United Nations Economic Commission for Europe 2013;21:23-32. 34. Centraal Bureau voor de Statistiek. Documentatierapport Landelijke Medische Registratie (LMR) 2009V1. Den Haag, The Netherlands: Centraal Bureau voor de Statistiek 2011.

83

Chapter 2.4

403 978 a a a a 836 456 889 462 327 , 017 700 975 762 239 , - - - - , , , , , , , , , , 515 PY ,

1609 6 42 77 63 43 16 50 78 66 60 32 262 322

Indonesian

groups living in the

506 98 39 22 68 87 , <10 <10 <10 <10 297 410 433 224 158 318 307 960 1 Events

845 525 a a a a a a a a 164 483 , 884 599 , ------, , , , 768 374 290 PY , , , 944 5 1 8

26 16 29 19 105 148

Antillean

lation and in the major migrant 45 53 36 13 14 16 25 10 75 <10 <10 <10 <10 <10 <10 <10 <10 165 Events

569 517 237 531 a a a a a a , 476 816 269 , , 738 481 622 , ------, , , , , , 699 006 PY , , 4 5 65 37 35 55 39 21

115 347 104 327

Turkish

70 32 12 27 42 46 21 <10 <10 <10 <10 <10 <10 185 216 256 764 158 Events sex group in the ethnic Dutch popu

-

782 586 a a a a a a a a a a 747 235 016 119 , , ------, , , , 568 097 PY , ,

3 7 76 32 19 13 237 217

Moroccan

years risk every at in age 10 24 25 34 62 14 16 36 - < <10 <10 <10 <10 <10 <10 <10 <10 <10 169 Events

642 228 973 089 a a a 344 091 166 216 , , , 285 224 771 , - - - , , , , , , , 465 610 625 PY , , , 943 9 2 5 81 84 52 25 57 30 15 319 102 102 387

Surinamese

66 74 42 16 55 88 92 93 55 10 <10 <10 <10 267 299 176 937 411 Events

101 317

618 403 170 756 148 830 , 663 069 235 809 619 979 696 , , , , , , , , , , , , , , 098 247

, 078 , ,

PY 345 158 550 176 247 950 705 894 , 134 106 150 854 813 401 407 , 85 , , , , , , , , , , , , , 735 289 9 4 4 3 2 1 9 4 4 3 2 2 1 27 28

Number of acute myocardial infarction events and person

Ethnic Dutch

199 710 808 023 830 003 875 293 , , , , , , , , 300 983 677 770 123 947 117 50 , , , , , , , 130 408 1 5 1 1 3 5 4 Events 13 15 18 11 67 12 13 41

years at risk 39 49 59 69 79 89 39 49 59 69 79 89

------<30 ≥90 <30 ≥90 30 40 50 60 70 80 30 40 50 60 70 80

All ages All ages Men Women Not given in line with the Dutch data protection guideline as the number of cases was less than ten

PY=Person- Supplementary table 1 Netherlands between 2000 and 2004 a

84

Incidence of first acute myocardial infarction over time by age, sex, and country of birth

541 290 a a a a 276 208 241 412 912 , 726 406 134 341 400 ,

- - - - , , , , , , , , , , 748 180 PY , ,

2 9 21 87 75 55 24 32 93 77 70 44

282 357

Indonesian

460 38 43 12 71 92 , <10 <10 315 375 394 294 <10 <10 135 281 325 916 1 Events

197 093 a a a a a a a 176 734 014 936 , 013 359 , ------, , , , 017 , , 393 415

PY , , , 3 5 1

33 33 26 12 31 15 166 181

Antillean

13 32 68 60 16 15 17 14 14 75 <10 <10 <10 <10 <10 <10 <10 197 Events

792 178 267 261 755 177 a a a a a , , 434 121 378 , , , 828 447 747 , - - - - - , , , , , , 881 PY , 1 49 46 17 51 38 12

128 124 439 132 104 423

Turkish

61 91 13 40 58 83 35 13 <10 273 199 231 <10 <10 861 <10 <10 246

Events sex group in the ethnic Dutch population and in the major migrant groups living in the

-

148 055 307 a a a a a a , 985 585 100 , 990 971 415 , ------, , , , , , 685 325 967 PY , , ,

8 1 4 70 25 20 56 25 10 107 303 291

Moroccan

years risk every at in age 17 56 64 89 66 12 11 12 23 21 76 - <10 <10 306 <10 <10 <10 <10

Events

684 429 541 554 723 a a a a 112 , 309 816 455 , , , 497 394 , - - - - , , , , 245 , , 196 PY , ,

494

4 9 80 85 42 16 49 24 105 385 136 100 479

Surinamese

105 73 94 43 11 64 97 51

, <10 <10 <10 <10 251 391 242 117 117 461 1 Events

464 766 925 870

, 851 196 727 886 665 453 , , 196 151 880 816 814 228 , , , , , , , , , , , , , 524 860

, ,

PY 185 064 661 873

, 334 181 900 878 417 045 , , 265 097 792 924 889 844 , , , , , , , , , , , , , 120 397 4 5 4 3 2 1 4 5 4 3 2 1 11 33 10 33

Number of acute myocardial infarction events and person Ethnic Dutch

019 813 065 566 106 503 654 2 225 013 , , , , 881 910 464 687 , , 535 , 62 , , , , , , , 106 387 1 6 1 1 3 5 4 65688 Events 13 15 16 11 10 13 39

years at risk

39 49 59 69 79 89 39 49 59 69 79 89 ------<30 ≥90 <30 ≥90 30 40 50 60 70 80 30 40 50 60 70 80

All ages All ages Men Women Not given in line with the Dutch data protection guideline as the number of cases was less than ten

Supplementary table Netherlands between 2005 and 2010 a PY=Person-

85

CHAPTER 2.5

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group

Agyemang C, van Oeffelen AAM, Bots ML, Stronks K, Vaartjes I. Socioeconomic inequalities in acute myocardial infarction incidence in migrant groups: has the epidemic arrived? analysis of nation-wide data.

Heart 2014;100(3):239-46. Chapter 2.5

ABSTRACT Objective We assessed socioeconomic inequalities in relation to acute myocardial infarction (AMI) incidence among major ethnic groups in the Netherlands.

Methods A nationwide register-based cohort study was conducted (n=2,591,170) between 1 January 1998 and 31 December 2007 among ethnic Dutch and ethnic minority groups from Suriname, Netherlands Antilles, Indonesia, Morocco, and Turkey. Standardised household disposable income was used a as proxy for socioeconomic status. Cox proportional hazard models were used to estimate the socioeconomic inequalities in AMI incidence.

Results Among ethnic Dutch, the AMI incidence was higher in the low-income group than in the high-income group: adjusted Hazard Ratios (HR) were 2.05 (95% confidence interval (CI): 2.00-2.10) for men and 2.33 (95% CI: 2.23-2.43) for women. Importantly, similar socioeconomic inequalities in AMI incidence were also observed in all ethnic minority groups, with the low socioeconomic group having a higher AMI incidence than the high socioeconomic group: adjusted HRs ranging from 2.07 (95% CI: 1.26-3.40) in Moroccan to 2.73 (95% CI: 1.55-4.80) in Antillean men; and from 2.17 (95% CI: 1.74-2.71) in Indonesian to 3.88 (95% CI: 2.36-6.38) in Turkish women.

Conclusion Our findings demonstrate socioeconomic inequalities in AMI incidence in ethnic minority groups and suggest a convergence towards the ethnic Dutch population. If the AMI incidence rates of the low socioeconomic group could be reduced to the level of the high socioeconomic group, this would represent a major public health improvement for all ethnic groups.

88

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group

INTRODUCTION

It is well established that in many industrialised countries, individuals with low socioeconomic status (SES) (as indicated by educational level, income level or occupational class) have a higher risk of most diseases, including cardiovascular disease (CVD), compared with individuals with high SES.1-3 People with low SES are often exposed to unfavourable levels of multiple risk factors including poor dietary habits, smoking, physical inactivity, poor access to health care and services which, inevitably, increase their risk of adverse health outcomes.3-5 Consequently, reducing socioeconomic health differentials has been an important goal of health policy.6 This negative gradient in relation to health outcomes does not, however, apply to all subgroups in the population. More specifically, among ethnic minority populations, the relationship between socioeconomic disadvantage and the risk of CVD remains uncertain.7-10 The analyses of Marmot et al in the early 1970s, for example, found a clear social class gradient in coronary heart disease (CHD) among the general population in England and Wales, but found no such association among Indian subcontinent migrants living in England and Wales.9 The lack of a socioeconomic gradient in CVD in migrant groups was confirmed by later studies in the UK and the Netherlands.11-14 Bos et al’s study in the 1990s in the Netherlands, for example, found that ethnic Dutch individuals with low SES had a higher risk of CVD mortality than individuals with high SES (Relative Risk (RR)=1.57; 95% confidence interval (CI): 1.51-1.64 for men, and RR=1.67; 95% CI: 1.56-1.78 for women, respectively). However, among migrant populations, no associations were found. For example, the RRs for CVD mortality were 0.94; 95%CI: 0.66-1.34 and 1.15; 95% CI: 0.74-1.80 for Turkish low-SES men and women compared with the high-SES Turkish men and women.13 The lack of association between SES and CVD in ethnic minority groups in the earlier studies largely mimicked the observations in low-and middle-income countries where many of the ethnic minority populations originated from;3 and was generally in line with the ‘diffusion theory’ of the epidemic of CHD.1 The ‘diffusion theory’ postulates that the rise of CHD started in the high socioeconomic groups in high-income countries, because they were the first who could afford the behaviours such as a diet rich in saturated fats and smoking, which increase the risk of CHD. With time, the disease spreads to lower socioeconomic groups in high-income countries as living standards improve, and to low-and middle-income countries as people adopt the unhealthy Western lifestyles. When the CHD epidemic started to decline, the higher socioeconomic groups were once again the first groups to reap the benefit as they were the first to adopt the healthy behavioural changes. Accordingly, the general expectation is that the socioeconomic gradient in CVD will eventually emerge in ethnic minority groups and converge towards the socioeconomic gradient seen in the host populations in industrialised countries.12;14 Data on the current state of socioeconomic inequalities in relation to CVD, particularly, incidence data among ethnic minority populations are lacking. The main aim of this study was,

89

Chapter 2.5 therefore, to examine whether differences exist in acute myocardial infarction (AMI) incidence by socioeconomic strata in ethnic minority groups and ethnic Dutch in the Netherlands. Furthermore, we assessed whether ethnic variations in AMI incidence exist in different socioeconomic groups.

METHODS

Cohort enrolment Data were obtained from Dutch national registers: Population Register (PR), Hospital Discharge Register (HDR), Cause of Death Register (CDR), and Regional Income Survey (RIS). The registers were used to acquire information regarding demographic factors, AMI hospitalisations and comorbidities, fatal AMI events, and income data as SES indicator. The registers have been described in detail previously.15 The overall quality of Dutch national registers has been proved to be adequate.16 In Hartoleth et al’s study, for example, the reliability of the Dutch cause-of-death statistics for major causes of death such as cancers and AMI were shown to be high (>90%).16 By linking previous registers with a personal identifier, a cohort was built, starting at 1 January 1998. On 1 January 1998, 15,431,715 Dutch citizens were registered in the PR, of which 13,421,681 (87%) were unique with respect to the combination of the variables date of birth, sex, and four digits of the postal code in the year of interest. As only PR unique persons can be identified in the HDR we only included PR unique persons in the study. Because a low number of AMI events were expected in the young age group, only persons of 30 years of age and older were included (n=8,185,247 (61.0%)). The SES indicator was based on the disposable household income, which was available for about one third of the Dutch population. After excluding the ethnic minority groups not under study (n=216,247), the final cohort comprised 2,591,170 persons (Figure 1).

Follow-up From 1 January 1998, persons with data on income were followed until their first AMI event, comprising a hospital admission with AMI as primary or secondary diagnosis (ICD-9 code 410), or a fatal event with AMI as primary or secondary cause of death (ICD-10 code I21). The validity of these ICD-codes proved to be good (sensitivity: 84%, positive predictive value: 97%).17 Persons with a previous hospital admission for AMI were excluded from the analyses. Persons were censored in case of death, non-uniqueness, migration outside the Netherlands, or the end of the study period at 31 December 2007, whichever came first.

Determinants Ethnic background Ethnic minority groups were constructed based on the country of birth of the resident and his/her parents, according to the definition of Statistics Netherlands.18 A person is considered a member of

90

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group an ethnic minority group if he/she was born abroad and at least one of the parents was born abroad (first generation), or if he/she was born in the Netherlands with at least one of the parents born abroad (second generation). For this study, ethnic Dutch (n=2,405,276) and minorities from Suriname (n=40,523), Morocco (n=18,031), Turkey (n=36,848), Netherlands Antilles (n=10,080), and Indonesia (n=80,412) were included. Indonesia and Suriname were former Dutch colonies. About 126,000 Indonesians migrated to the Netherlands between 1945 and 1949 in the aftermath of World War II and the Indonesian War of Independence.19 migrated to the Netherlands mainly due to political instability in Suriname in 1975 and 1980. The Surinamese population is ethnically diverse and mainly consists of African Surinamese (West African descent) and South Asian Surinamese (North Indian descent).18;20 The Antillean/Aruban population is predominantly of West African, European and mixed origin, and its migration to the Netherlands has been relatively stable over time. Turkish and Moroccan men came to the Netherlands in the 1960s and 1970s as labour migrants, and were later followed by their families.18

Socioeconomic status Socioeconomic status (SES) was based on income data registered in the RIS, which received income data from tax services.21 The RIS started in 1994, when a randomly representative sample of 1.9 million Dutch residents was selected. Every year, the sample was corrected for emigration and mortality on one hand, and immigration and birth on the other hand. All persons belonging to the households of the sample population (about one third of the Dutch population) were included in the RIS. SES was defined as the standardised disposable income (in euros) at the household level (adjusted for number of household members) in the year preceding the AMI. Among the non-cases we used income in the year preceding the end of follow-up. In this way, the time between SES measurement and end of follow-up was the same for every subject under study. Subsequently, this income was divided into tertiles (the first tertile representing the lowest income group, second tertile representing the medium income group, and third tertile representing highest income group) in two ways: 1) the income of the total cohort (‘total’ SES); and 2) the income within every ethnic group (‘within group’ SES). We divided the income tertiles in two ways because the income distribution differed between the ethnic groups. Thus, we used ‘within ethnic group’ tertiles for the analyses on socioeconomic inequalities in AMI incidence within groups. For the analyses on differences between ethnic Dutch and ethnic minority groups, we used tertiles of the total population to facilitate direct comparability of the results.

91

Chapter 2.5

247

Others

N=216,

N=80,412 Indonesian

Antillean N=10,080

Turkish N=36,848

N=8,185,247 N=2,807,417 N=15,431,715 N=13,421,681 1 January 1998 Unique persons ≥30 years of age of years ≥30 Population Register Population

available data income Household Final cohort cohort Final N=2,591,170

N=18,031 Moroccan

N=40,523 Surinamese

Survey Income Regional of of inclusion procedure

Flowchart 1 1

N=2,405,276 Dutch Ethnic Figure

92

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group

Comorbidity Presence and extent of comorbidity were determined with the Charlson index score,22 based on 16 discharge diagnosis at baseline (congestive heart failure, peripheral vascular disease, cerebral vascular accident, dementia, pulmonary disease, connective tissue disorder, peptic ulcer, liver disease, diabetes, diabetes complications, paraplegia, renal disease, cancer, metastatic cancer, severe liver disease, and HIV). The Charlson index ranges from zero to six (cut-off value), with zero representing no comorbidity. The Charlson index score was subsequently grouped into four categories (0, 1-2, 3-4, and ≥5). The Charlson index score has proven to be a reliable and valid method to measure comorbidity in clinical research.22

Data analysis Baseline characteristics were analysed on 1 January 1998. Absolute AMI incidence rates were calculated as the number of AMI events per 100,000 person-years at risk, and subsequently age- standardised to the age distribution of the European population (using the direct method) with ten- year age bands. For the calculation of relative AMI incidence rates, Cox proportional hazard regression analyses were used stratified by sex because the incidence of AMI differed between men and women. Using the ‘within group’ SES indicator, AMI incidence differences between persons with a low SES and persons with a high SES (reference) in every ethnic minority group and the ethnic Dutch population were investigated. All analyses were adjusted for age (Model 1) and additionally for marital status, degree of urbanisation, generation, and Charlson comorbidity index (Model 2). Also, AMI incidence differences between ethnic minority groups and ethnic Dutch within every SES tertile were investigated with the ‘total’ SES indicator. These analyses were adjusted for age (Model 1) and, additionally, for marital status, degree of urbanisation, Charlson comorbidity index, and disposable income on household level (Model 2). The extra adjustment for disposable income was to overcome the problem of income differences within SES tertiles. Results were expressed as Hazard Ratio (HR) with accompanying 95% CI. We used SPSS software, version 14.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation Netherlands.

RESULTS

Characteristics of the study population Table 1 shows the characteristics of the study population. There were, in total, 20,584,569 person- years at risk with 70,554 AMI events. In general, ethnic minority populations were younger than the ethnic Dutch population. Surinamese and Antilleans were the least likely to be married or to live together with a partner. Except for Indonesians, the majority of the ethnic minority groups were first generation ethnic minorities ranging from 92% in Antilleans to 99.8% in Moroccans. Indonesians had

93

Chapter 2.5

6.3 6.3 3,9 1.0 0.2 66. 49.4 48.5 28.4 33.5 38.2 22.2 31.1 24.9 15.5 94.9 1,944 80,412 21,541 629,260 45 (38;55) Indonesian (16,126, 28,387) (16,126,

8.2 8.2 2.9 4.4 0.7 0.3 108 46.5 45.1 92.0 46.0 30.1 23.9 39.3 31.3 18.3 94.7 10,080 73,604 17,721 Antillean 41 (35;49) (12,508, 23,954) 23,954) (12,508,

9.4 0.9 4.6 0.6 0.1 773 54.5 87.8 99.6 57.2 30.6 12.2 39.5 32.7 17.5 94.6 36,848 15,618 Turkish 284,818 40 (34;51)

20,370) (11,970,

up -

9.4 9.4 2.0 1.6 0.2 0.0 108 55.2 84.5 99.8 62.0 26.4 11.5 39.6 28.8 20.2 98.2 18,031 14,687 133,938 40 (34;49) Moroccan 19,691) (11,375,

4.4 4.4 1.5 5.0 0.9 0.4 922 46.1 47.5 94.7 41.8 33.0 25.2 60.5 20.8 12.8 93.7 40,523 18,523 301,824 42 (36, 50) 42 (36, Surinamese (13,253, 24,418) 24,418) (13,253,

) at baseline: very urban=>2000, urban=1001-2000, urban/rural=501-1000, rural=251-500, very rural=<251 rural=<251 very rural=251-500, urban/rural=501-1000, urban=1001-2000, urban=>2000, very baseline: at )

2

- 4.9 4.9 1.3 0.3 50.0 76.8 32.0 33.7 34.4 13.0 22.6 22.4 24.8 17.3 93.6

66,699 20,610 2,405,276 48 (39, 59) 48 (39, 19,161,125 Ethnic Dutch Ethnic

27,143) (15,482, -up

c 0

>4 b 3-4 3-4 1-2 1-2 onincome population total of inthe year prior end to followof

Rural Rural

e Urban Urban ased ased

rural Very Very urban

Urban/rural Urban/rural

)

a

d

age (IQR age years at risk at years

- Tertile 1(lowest income) Population characteristics between 1998 2007 and characteristics Population vents vents Tertile 3(highest income) Tertile 2(medium income) tertiles % tertiles

N Person AMI e Men % Median % together or living Married % First generation (IQR) Euros in income Median SES % urbanisation of Degree % index Charlson At baseline Socioeconomic status b Populationdensity (number of residentsper km Interquartile range In year prior the end to of follow

1 Table a b c d e

94

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group the highest and Moroccans had the lowest income. Comorbidities were more common in ethnic Dutch than in ethnic minority groups.

AMI incidence by income gradient within each ethnic group Table 2 shows age-standardised absolute incidence rates per 100,000 person-years at risk stratified by sex and income group. In all ethnic groups, individuals in the low-income tertile category had a higher age-standardised AMI incidence rate than individuals in the high-income tertile category in both men and women.

Table 2 Age-standardised absolute incidence rates per 100,000 person-years at risk, stratified by sex and socioeconomic status (based on income within each ethnic group) Low income Medium income High income Events IR (95% CI)a Events IR (95% CI)a Events IR (95% CI)a Men Ethnic Dutch 22,780 464 (453, 476) 13,382 320 (299, 341) 8,506 224 (207, 241) Surinamese 295 534 (466, 602) 214 332 (275, 389) 152 260 (193, 327) Moroccan 40 180 (96, 264) 27 69 (41, 98) 26 58 (32, 85) Turkish 302 428 (371, 486) 212 281 (231, 332) 123 156 (119, 194) Antillean 38 305 (184, 426) 18 110 (37, 184) 19 261 (59, 463) Indonesian 603 459 (421, 497) 454 315 (284, 345) 269 214 (184, 243) Women Ethnic Dutch 13,784 206 (201, 211) 5,395 143 (134, 152) 2,852 89 (82, 96) Surinamese 152 231 (193, 269) 72 163 (117, 210) 37 57 (33, 80) Moroccan <10 -b <10 -b <10 -b Turkish 71 187 (111, 263) 45 88 (46, 130) 20 38 (17;59) Antillean 23 194 (107, 280) <10 -b <10 -b Indonesian 318 206 (179, 233) 192 184 (130, 238) 108 86 (69, 103) a Age-standardised AMI incidence rate: standardised to the age-distribution of the European population b Not given in line with the Dutch data protection guideline as the number of cases was less than ten

Figure 2a shows HRs for AMI incidence by income group in men in each ethnic group. As expected, among ethnic Dutch, the age adjusted incidence of AMI was higher in people with low and medium income than in people with high income; the differences still persisted after further adjustments for marital status, generation, degree of urbanisation, and Charlson index (henceforth, other covariates): HR=1.49 (95% CI, 1.44-1.52) for the medium-income group, and 2.05 (95% CI, 2.00-2.10) for the low-income group, respectively. Among ethnic minority groups, the incidence of AMI was higher among individuals with medium incomes and low incomes compared with individuals with high incomes in the Indonesian, Surinamese, and Turkish groups. The differences hardly changed after adjusting for other covariates, and the magnitude of the income gradient in these ethnic minority groups was largely similar to that of the ethnic Dutch men: HRs for medium income and low income were 1.63 (95% CI: 1.32-2.00) and 2.33 (95% CI: 1.91-2.84) for Surinamese, 1.77 (95% CI:

95

Chapter 2.5

1.41-2.21) and 2.60 (95% CI: 2.10-3.20) for Turkish, and 1.57 (95% CI: 1.35-1.83) and 2.27 (95% CI: 1.96-2.62) for Indonesians, respectively. Antillean and Moroccan men with low income also had a lower AMI incidence than their comparable groups with high income even after other factors had been adjusted for: HR=2.73 (95% CI: 1.55-4.80) and HR=2.07 (95% CI: 1.26-3.40), respectively. However, there were no differences between medium-income groups and high-income groups in Antillean and Moroccan men. Similar patterns were observed among women in all ethnic groups. The age adjusted incidence of AMI was higher in women with medium and low income than in women with high income among ethnic Dutch. The differences persisted after other factors had been adjusted for (Figure 2b). Similarly, the medium-income and low-income groups from Suriname, Turkey and Indonesia had a higher AMI incidence than their high-income counterparts; the differences largely remained the same after further adjustment for other covariates. There were no differences in AMI incidence between low-income and medium-income groups.

Difference in AMI incidence between ethnic Dutch and ethnic minority groups within each income category Table 3 shows the differences in AMI incidence between the ethnic Dutch and the ethnic minority groups stratified by each income tertile. Within the low-income group, age-adjusted incidence of AMI was higher in Surinamese, Turkish and Indonesian men, but lower in Moroccan men compared with ethnic Dutch men. The differences remained largely unchanged after further adjustments for other factors. No differences were found between Antillean men and ethnic Dutch men. Among women, there were no differences in AMI incidence between the ethnic minority women and ethnic Dutch except for the lower incidence in the Moroccan group. Within the medium-income group, Surinamese and Indonesian men had a higher incidence of AMI, while Moroccan and Antillean men had a lower incidence of AMI compared with ethnic Dutch men, even after adjustment for other covariates. Turkish men also tended to have a higher AMI incidence than ethnic Dutch (HR=1.15, 95% CI: 0.98-1.35). Surinamese and Indonesian women had a higher age adjusted incidence of AMI than ethnic Dutch, while Turkish women had lower age adjusted incidence of AMI. Within the high-income group, Surinamese men had a higher incidence of AMI, while Moroccan men had a lower incidence of AMI compared with ethnic Dutch. No further significant differences were observed between the other ethnic minority groups and ethnic Dutch men. Among women, no differences were found between any of the ethnic minority groups and ethnic Dutch.

96

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group

Model 1: Age-adjusted HR (95% CI)

Medium versus high income Ethnic Dutch 1.50 (1.46, 1.54)

Surinamese 1.62 (1.31, 1.99) Moroccan 1.18 (0.69, 2.02)

Turkish 1.80 (1.44, 2.25) Antillean 1.19 (0.62, 2.26)

Indonesian 1.63 (1.40, 1.90)

Low versus high income

Ethnic Dutch 2.08 (2.02, 2.13)

Surinamese 2.30 (1.89, 2.81) Moroccan 2.17 (1.32, 3.57) Turkish 2.63 (2.13, 3.25)

Antillean 2.76 (1.59, 4.81) Indonesian 2.36 (2.05, 2.73)

Model 2: plus marital status, generation, degree of urbanisation, and Charlson index

Medium versus high income Ethnic Dutch 1.49 (1.44, 1.52)

Surinamese 1.63 (1.32, 2.00) Moroccan 1.19 (0,69, 2.04)

Turkish 1.77 (1.41, 2.21) Antillean 1.17 (0.61, 2,24)

Indonesian 1.57 (1.35, 1.83)

Low versus high income Ethnic Dutch 2.05 (2.00, 2.10) Surinamese 2.33 (1.91, 2.84) Moroccan 2.07 (1.26, 3.40)

Turkish 2.60 (2.10, 3.20) Antillean 2.73 (1.55, 4.80) Indonesian 2.27 (1.96, 2.62)

0.5 1.0 2.0 5.0

Figure 2a Hazard Ratios (95% CI) of AMI incidence by income category within each ethnic group in men

97

Chapter 2.5

Model 1: Age-adjusted HR (95% CI)

Medium versus high income

Ethnic Dutch 1.64 (1.57, 1.72)

Surinamese 2.04 (1.37, 3.04)

Moroccan - Turkish 2.30 (1.36, 3.89)

Antillean -

Indonesian 1.60 (1.26, 2.03)

Low versus high income Ethnic 2.36 (2.26, 2.46) Surinamese 3.28 (2.27, 4.73)

Moroccan -

Turkish 3.94 (2.40, 6.48) Antillean -

Indonesian 2.25 (1.80, 2.80)

Model 2: plus marital status, generation, degree of urbanisation, and Charlson index

Medium versus high income

Ethnic Dutch 1.63 (1.56, 1.71) Surinamese 2.03 (1.36, 3.02) Moroccan -

Turkish 2.26 (1.33, 3.83) Antillean -

Indonesian 1.56 (1.23, 1.98)

Low versus high income Ethnic Dutch 2.33 (2.23, 2.43)

Surinamese 3.19 (2.20, 4.61) Moroccan -

Turkish 3.88 (2.36, 6.38) Antillean - Indonesian 2.17 (1.74, 2.71)

0.5 1.0 2.0 5.0 10.0 Figure 2b Hazard Ratios (95% CI) of AMI incidence by income category within each ethnic group in women

98

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group

Table 3 Difference in AMI incidence (HR (95% CI)) between ethnic Dutch and ethnic minority groups ≥30 years of age within each SES tertile (based on income of total population), stratified by sex Men Women Model 1a Model 2b Model 1a Model 2b Low income Ethnic Dutch 1.00 1.00 1.00 1.00 Surinamese 1.42 (1.27;1.57)* 1.36 (1.22;1.51)* 1.11 (0.95;1.30) 1.06 (0.91;1.24) Moroccan 0.36 (0.28;0.45)* 0.35 (0.28;0.45)* -c -c Turkish 1.20 (1.09;1.32)* 1.17 (1.06;1.29)* 0.88 (0.73;1.07) 0.86 (0.71;1.04) Antillean 0.84 (0.63;1.11) 0.81 (0.61;1.08) 0.81 (0.56;1.18) 0.78 (0.54;1.15) Indonesian 1.09 (1.00;1.19) 1.08 (0.99;1.18) 1.01 (0.90;1.14) 1.01 (0.90;1.14) Medium income Ethnic Dutch 1.00 1.00 1.00 1.00 Surinamese 1.45 (1.26;1.67)* 1.38 (1.19;1.60)* 1.51 (1.20;1.91)* 1.47 (1.17;1.86)* Moroccan 0.37 (0.25;0.56)* 0.37 (0.25;0.56)* -c -c Turkish 1.16 (0.99;1.36) 1.15 (0.98;1.35) 0.65 (0.43;0.97)* 0.65 (0.43;0.97)* Antillean 0.47 (0.27;0.81)* 0.45 (0.26;0.78)* -c -c Indonesian 1.14 (1.04;1.25)* 1.13 (1.03;1.23)* 1.18 (1.03;1.35)* 1.18 (1.03;1.35)* High income Ethnic Dutch 1.00 1.00 1.00 1.00 Surinamese 1.48 (1.23;1.79)* 1.44 (1.19;1.74)* 0.92 (0.61;1.38) 0.92 (0.61;1.39) Moroccan 0.30 (0.14;0.68)* 0.31 (0.14;0.69)* -c -c Turkish 0.98 (0.70;1.35) 0.97 (0.70;1.35) 0.80 (0.38;1.68) 0.81 (0.38;1.69) Antillean 0.94 (0.57;1.56) 0.93 (0.56;1.54) -c -c Indonesian 1.02 (0.91;1.14) 1.01 (0.91;1.13) 1.07 (0.89;1.28) 1.08 (0.90;1.29) a Model 1: adjusted for age b Model 2: adjusted for age, disposable income on household level, marital status, degree of urbanisation, Charlson comorbidity index c Not given in line with the Dutch data protection guideline as the number of cases was less than ten * Significant difference between ethnic minority group and ethnic Dutch

DISCUSSION

Key findings Our current findings indicate socioeconomic inequalities in AMI incidence among ethnic groups. Within each income group, ethnic differences in AMI incidence still persist, with sometimes higher and sometimes lower rates among ethnic minority groups.

Discussion of key findings SES and AMI incidence among ethnic minority groups Socioeconomic inequalities in CVD outcomes among ethnic minority groups remain less noticeable although there is a general consensus that the epidemic will eventually emerge. Our current finding of clear socioeconomic inequalities in AMI incidence in all the ethnic groups studied is in line with the European patterns of inequalities and fits with the popular ‘diffusion theory’ of the epidemic of

99

Chapter 2.5

CHD mortality.1 William et al predicted that the socioeconomic gradient would first emerge in health behaviour and anthropometric risk factors, followed by chronic diseases morbidity and mortality.14 In the Netherlands, data on health behaviour by SES in ethnic minority groups are limited, and the limited data only partially support these predictions.23;24 For example, Nierkens et al’s study in the early 2000s found higher smoking rates among Surinamese, Turkish and Moroccan men with lower SES than among those with higher SES.23 By contrast, smoking rates were higher among Surinamese, Turkish and Moroccan women with higher SES than among women with lower SES. These findings suggest the need to assess the current socioeconomic gradient in relation to all potential risk factors including diabetes, high blood pressure, physical inactivity, access to health care, and compliance with therapy, which might predispose low SES individuals to the high risk of AMI incidence.

AMI incidence difference between ethnic Dutch and ethnic minority groups by SES Surinamese men, in particular, had a higher incidence of AMI than the ethnic Dutch in all the income strata. Analyses of ethnic differences in income within class groups showed that within each social class group, most ethnic minority groups have a smaller income than .7 In our current analyses, the ethnic differences persisted even after adjusting for income differentials, which suggests that other unmeasured factors, such as risk factors, and structure characteristics (e.g. access to health care), might play a role in the observed ethnic differences in AMI incidence. For example, some of the important risk factors for AMI (e.g. hypertension and diabetes) are more prevalent in Surinamese men than in ethnic Dutch men.25;26 Besides, poor blood pressure control, an important risk factor for poor CVD outcome, is dismally high in Surinamese men,23 which might further contribute to the observed differences. The high incidence of AMI among Turkish and Indonesian low-income and medium-income groups relative to the ethnic Dutch is also unclear, but it might relate to differences in other risk factors for AMI between low-income and high-income groups. Evidence, for example, suggests that low-income Turkish men have a higher prevalence of smoking than their high-income counterparts, which might contribute to the observed differences.27 Among women, the differences were only observed in medium-income groups in Surinamese and Indonesians. By contrast, Moroccans had a lower AMI incidence than the ethnic Dutch. The low incidence of CVD among Moroccans generally points to their low smoking prevalence and Mediterranean diet,27 which has been shown to have cardio-protective effects.28 The socioeconomic inequalities in health among ethnic groups have largely focused on the role of SES in explaining ethnic differences in health outcomes.7 Our current finding of a clear socioeconomic inequality in AMI incidence within ethnic minority groups on one hand, and the persistent ethnic inequalities in AMI incidence on the other hand, seems to suggest that the current focus on SES as explanatory factor for the ethnic differences in health outcomes needs to be broadened. It is clear from our current findings that individuals in each low socioeconomic group

100

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group have a comparatively high risk of AMI incidence. The magnitudes of the differences between low- income and high-income groups among the ethnic minority groups were generally similar to that of the ethnic Dutch population. If the AMI incidence of the low socioeconomic ethnic minority groups could be reduced to the level of the high socioeconomic group, particularly among ethnic minority groups with high risk, this would represent a major public health improvement for all ethnic groups regardless of our inability to explain ethnic differences in AMI incidence. Failure to acknowledge and act swiftly on socioeconomic inequalities with appropriate preventive measures within ethnic minority groups might exacerbate the existing ethnic inequalities in CVD or wipe out some of the CVD advantage currently being enjoyed by some ethnic minority groups.29 The analyses by Harding et al clearly illustrates the essence of this point. In that study, the CHD mortality declined more rapidly in White populations than in most ethnic minority populations in England and Wales.30 One striking finding was that Jamaican born women who are traditionally known to have a lower CHD risk, for the first time had a higher CHD risk than those born in England and Wales. A similar observation applied in the USA, where African Americans now have a higher CHD incidence than the White Americans, reversing the previous pattern.31

Considerations A major strength of our current study is that it is based on a nationwide database with large sample sizes, which allowed us to study AMI incidence in several ethnic minority groups. Another strength of our study is that we had the opportunity for record linkage of the AMI register data with household income for each individual, which gives a more correct picture on socioeconomic health differentials than the use of other surrogate markers of SES, such as neighbourhood income. It has been demonstrated that a better characterisation of SES leads to the demonstration of health differentials that are considerably wider than those acquired when less precise measures are used.32 There are also limitations to our study.33 Our SES was limited to only household income because of lack of data on other socioeconomic indicators such as educational level and occupational class. It has been emphasised that different measures of SES may affect health through different pathways and causal mechanisms.34 Furthermore, ethnic minority groups were constructed on the basis of country of birth. Country of birth may reflect ethnicity reasonably well among some ethnic groups such as Moroccans,18 but it is likely to be an unreliable proxy measure of ethnicity for other groups such as Surinamese. Nonetheless, our earlier studies showed that AMI risk factors, such as hypertension and diabetes, are higher in both African Surinamese and South Asian Surinamese than in ethnic Dutch people.25,26,35 Additionally, individuals who were not unique with respect to the combination of the variables birth date, sex, and four digits of the postal code were excluded. Non- uniqueness was more common among ethnic minority groups than among ethnic Dutch, which

101

Chapter 2.5 might affect our study conclusions. Nevertheless, since non-uniqueness was not related to AMI mortality, it is unlikely to bias our study conclusions.

Conclusion In conclusion, our current findings demonstrate that socioeconomic inequalities in AMI incidence have surfaced in ethnic minority groups. The magnitudes of the socioeconomic inequalities in ethnic minority groups are largely similar to that of the European general population, and suggest the need to explore the possible mechanisms underlying the observed relationship. If the AMI incidence rates of the low socioeconomic ethnic minority groups could be reduced to the level of the high socioeconomic ethnic minority groups, this would represent a major public health improvement for all ethnic groups, regardless of our inability to explain ethnic differences in AMI incidence.

102

Socioeconomic inequalities in acute myocardial infarction incidence by ethnic group

REFERENCES 1. Mackenbach JP, Cavelaars EJM, Kunst AE, Groenhof F. Socioeconomic inequalities in cardiovascular disease mortality. An international study. Eur Heart J 2000;21:1141–51. 2. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation 1993;88:1973-98. 3. Manrique-Garcia E, Sidorchuk A, Hallqvist J, Moradi T. Socioeconomic position and incidence of acute myocardial infarction: a meta-analysis. J Epidemiol Community Health 2011;65:301-9. 4. Ruberman W, Weinblatt E, Goldberg JD, Chaudhary BS. Psychosocial influences on mortality after myocardial infarction. N Engl J Med 1984;311:552-9. 5. Marmot M, Allen J, Bell R, Bloomer E, Goldblatt P. Consortium for the European Review of Social Determinants of Health and the Health Divide. WHO European review of social determinants of health and the health divide. Lancet 2012;380(9846):1011-29. 6. Geyer S, Hemstrom O, Peter P, Vagero D. Education, income, and occupational class cannot be used interchangeably in social epidemiology. Empirical evidence against a common practice. J Epidemiol Community Health 2006;60:804-10. 7. Nazroo JY. The structuring of ethnic inequalities in health: economic position, racial discrimination, and racism. Am J Public Health 2003;93:277-84. 8. Nazroo JY. South Asian people and heart disease: an assessment of the importance of socioeconomic position. Ethn Dis 2001;11:401-11. 9. Marmot M, Mustard J. Coronary heart disease from a population perspective. In: Evans R, Barer M, Marmor T, eds. Why are some people healthy and others not? New York: Aldin de Gruytere, 1994:189–214. 10. Bhopal R, Hayes L, White M, Unwin N, Harland J, Avis S, et al. Ethnic and socio-economic inequalities in coronary heart disease, diabetes and risk factors in Europeans and South Asians. J Public Health Med 2002;24:95-105. 11. McKeigue PM, Miller GJ, Marmot MG. Coronary heart disease in south Asians overseas: a review. J Clin Epidemiol 1989;42:597-609. 12. McKeigue PM, Marmot MG. Mortality from coronary heart disease in Asian communities in London. BMJ 1988;297(6653):903. 13. Bos V, Kunst AE, Garssen J, Mackenbach JP. Socioeconomic inequalities in mortality within ethnic groups in the Netherlands, 1995-2000. J Epidemiol Community Health 2005;59:329-35. 14. Williams R, Wright W, Hunt K. Social class and health: the puzzling counter-example of British South Asians. Soc Sci Med 1998;47:1277-88. 15. Agyemang C, Vaartjes I, Bots ML, van Valkengoed IG, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95:747-753. 16. Harteloh P, de Bruin K, Kardaun J. The reliability of cause-of-death coding in The Netherlands. Eur J Epidemiol 2010;25:531-538. 17. Merry AH, Boer JM, Schouten LJ, Feskens EJ, Verschuren WM, Gorgels AP, et al. Validity of coronary heart diseases and heart failure based on hospital discharge and mortality data in the Netherlands using the cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009;24:237-247. 18. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14:255-269. 19. Ho L, Bos V, Kunst AE. Differences in cause-of-death patterns between the native Dutch and persons of Indonesian descent in the Netherlands. Am J Public Health 2007;97:1616-8.

103

Chapter 2.5

20. Agyemang C, Bhopal R, Bruijnzeels M. Negro, Black, Black African, African Caribbean, African American or what? Labelling African origin populations in the health arena in the 21st century. J Epidemiol Community Health 2005;59:1014-8. 21. van Oeffelen AA, Agyemang C, Bots ML, Stronks K, Koopman C, van Rossem L, et al. The relation between socioeconomic status and short-term mortality after acute myocardial infarction persists in the elderly: results from a nationwide study. Eur J Epidemiol 2012;27:605-13. 22. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57:1288-1294. 23. Nierkens V, de Vries H, Stronks K. Smoking in immigrants: do socioeconomic gradients follow the pattern expected from the tobacco epidemic? Tob Control 2006;15:385-91. 24. Dijkshoorn H, Nierkens V, Nicolaou M. Risk groups for overweight and obesity among Turkish and Moroccan migrants in The Netherlands. Public Health 2008;122:625-30. 25. Agyemang C, Bindraban N, Mairuhu G, Montfrans Gv, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23:1971-7. 26. Agyemang C, Kunst AE, Bhopal R, Anujuo K, Zaninotto P, Nazroo J, et al. Diabetes prevalence in populations of South Asian Indian and African origins: a comparison of England and the Netherlands. Epidemiology 2011;22:563-7. 27. Ujcic-Voortman JK, Schram MT, Jacobs-van der Bruggen MA, Verhoeff AP, Baan CA. Diabetes prevalence and risk factors among ethnic minorities. Eur J Public Health 2009;19:511-5. 28. Misirli G, Benetou V, Lagiou P, Barnia C, Trichopoulos D, Trichopoulou A. Relation of the traditional Mediterranean diet to cerebrovascular disease in a Mediterranean population. Am J Epidemiol 2012;176:1185-92. 29. van Oeffelen A, Agyemang C, Koopman C, Stronks K, Bots M, Vaartjes I. Downward trends in acute myocardial infarction incidence: how do migrants fare with the majority population? Results from a nationwide study. Eur J Prev Cardiol 2013 [Epub ahead of print]. 30. Harding S, Rosato M, Teyhan A. Trends for coronary heart disease and stroke mortality among migrants in England and Wales, 1979-2003: slow declines notable for some groups. Heart 2008;94:463-70. 31. American Heart Association: 2002 Heart and Stroke Statistical Update. Dallas, Texas: American Heart Association;2001. 32. Davey Smith G, Shipley MJ, Rose G. The magnitude and causes of socioeconomic differentials in mortality: further evidence from the Whitehall Study. J Epidemiol Community Health 1990;44:265- 270. 33. Agyemang C, Bhopal R. Hypertension and cardiovascular disease endpoints by ethnic group: the promise of data linkage. Heart 2013;99:675-6. 34. Geyer S, Peter R. Income, occupational position, qualification and health inequalities--competing risks? (comparing indicators of social status). J Epidemiol Community Health 2000;54:299-305. 35. Agyemang C, Addo J, Bhopal R, Aikins Ade G, Stronks K. Cardiovascular disease, diabetes and established risk factors among populations of sub-Saharan African descent in Europe: a literature review. Global Health 2009;5:7.

104

CHAPTER 3

INEQUALITIES IN INCIDENCE OF STROKE

CHAPTER 3.1

Ethnic inequalities in stroke incidence by stroke subtype and sex

Agyemang C, van Oeffelen AAM, Norredam M, Kappelle LJ, Klijn CJM, Bots ML, Stronks K, Vaartjes I. Ethnic disparities in the incidence of ischemic stroke, intracerebral haemorrhage and subarachnoid haemorrhage in the Netherlands.

Submitted. Chapter 3.1

ABSTRACT Background Data on the incidence of stroke subtypes among ethnic minority groups are limited. We assessed ethnic differences in the incidence of stroke subtypes in the Netherlands.

Methods A Dutch nationwide register-based cohort study (n=7,423,174) was conducted between 1998 and 2010. Persons were followed from 1 January 1998 until their first stroke event. We studied the following stroke subtypes: ischemic stroke (IS), intracerebral haemorrhage (ICH) and subarachnoid haemorrhage (SAH). Cox proportional hazard models were used to estimate incidence differences between first generation ethnic minorities (henceforth, migrants) and the ethnic Dutch population.

Results Compared with ethnic Dutch, Surinamese had an increased risk while Moroccans had a reduced risk of all stroke subtypes. Chinese people had an increased risk of ICH (Hazard Ratio (HR)=2.29, 95% confidence interval (CI): 1.60-3.28 for men and HR=1.61, 95% CI: 0.98-2.63 for women), but a reduced risk of IS (HR=0.66, 95% CI: 0.51-0.86 for men and HR=0.68, 95% CI: 0.51-0.91 for women) compared with ethnic Dutch. Among Indonesians, women had an increased risk of ICH (HR=1.65, 95% CI: 1.55-1.86) and SAH (HR=1.61, 95% CI: 1.38-1.88) while men had an increased risk of ICH (HR=1.70, 95% CI: 1.55-1.86) compared with ethnic Dutch. Antillean men had a higher risk of ICH (HR=1.76, 95% CI: 1.27-2.45) than ethnic Dutch men. Turkish women had a reduced risk of IS (HR=0.89, 95% CI: 0.80-0.99) and SAH (HR=0.62, 95% CI: 0.44-0.87) while Turkish men had an increased risk of ICH (HR=1.48, 95% CI: 1.23-1.77) compared with their ethnic Dutch counterparts.

Conclusion Our findings suggest that Surinamese have an increased risk whilst Moroccans have a reduced risk for all stroke subtypes. Among other migrants, the risk appears to depend on the stroke subtype and sex. These findings underscore the need to identify the root causes of these ethnic differences to assist primary and secondary prevention efforts.

108

Ethnic inequalities in stroke incidence by stroke subtype and sex

INTRODUCTION

Worldwide, stroke is the second most common cause of mortality and the third most common cause of disability.1 Although stroke incidence has remained stable in high-income countries, the absolute number of people with stroke and the global burden of stroke-related disability are high and rising.2 Several studies have demonstrated ethnic variability in stroke mortality in Europe.3-5 However, data on ethnic differences in stroke incidence, in particular for stroke subtypes are very limited.6-10 The few studies that have assessed ethnic differences in the distribution of stroke subtypes in Europe were primarily based on the United Kingdom populations of African and South-Asian descent.8-10 These studies have shown large differences, both within and across ethnic groups. In the South London Ethnicity and Stroke study, intracerebral haemorrhage was over two-fold more common in the ‘black’ African than in African-Caribbean patients even after adjusting for risk factors and social class.10 Gaining insight into the incidence of stroke subtypes among different ethnic groups could help target prevention efforts. This is relevant because stroke is a heterogeneous disease with different risk factor profiles.11 Hence, the aim of this paper was to assess differences in incidence of overall stroke and stroke subtypes among Surinamese, Antillean, Turkish, Moroccan, Indonesian, and Chinese migrants living in the Netherlands. These incidence rates were compared with the ethnic Dutch population.

METHODS

Cohort enrolment Data were obtained from the following Dutch national registers: Population Register (PR), Hospital Discharge Register (HDR), Cause of Death Register (CDR), and Regional Income Survey (RIS). The registers were used to acquire information on demographic factors, hospitalisations for stroke subtypes and comorbidities, fatal stroke events, and income. The registers have been described in detail elsewhere.5 The reliability of Dutch national registers for stroke has been proven to be high.12 By linking these registers with a personal identifier we built a cohort starting at 1 January 1998 when 15,431,715 Dutch citizens were registered in the PR. As only PR unique persons can be identified in the HDR (unique with respect to the combination of the variables date of birth, sex, and four digits of the postal code), non-unique persons were excluded. This left a cohort of 13,421,681 (87.0%) persons. Because a low number of stroke events are expected in the young, persons younger than 30 years of age were excluded (n=5,236,434 (39.0%)).

Follow-up Persons were followed from 1 January 1998 until their first stroke event. We studied the following stroke subtypes: ischemic stroke (IS) (including a transient ischemic attack), intracerebral haemorrhage (ICH), and subarachnoid haemorrhage (SAH). A first event comprised a hospital

109

Chapter 3.1 admission with as primary or secondary diagnosis SAH (ICD-9 code 430), ICH (ICD-9 code 431), or IS (ICD-9 code 434-436), or a fatal event with as primary or secondary cause of death IS (ICD-10 code I63, G45), ICH (ICD-10 code I61), or SAH (ICD-10 code I60). The positive predictive values of these codes have previously shown to be acceptable; ≥75% for ICD-9 code 430, ≥85% for ICD-9 code 431, and ≥85% for ICD-9 codes 434-436.13;14 Persons who already had a previous hospital admission for a stroke from 1995 onwards were not included as having a first stroke. Persons were censored in case of death, emigration, or the end of the study period (31 December 2010), whichever came first.

Determinants Ethnic background Ethnic minority groups were constructed based on the country of birth of the resident and his/her parents, according to the definition of Statistics Netherlands.15 A person is considered an ethnic minority if he/she was born abroad with at least one of the parents born abroad (first generation

ethnic minority), or if he/she was born in the Netherlands with at least one of the parents born procedure procedure inclusion of inclusion Flowchart of Flowchart abroad (second generation ethnic minority). A person was identified as ethnic Dutch if both parents

were born in the Netherlands. In order to investigate a homogeneous group we only included first 1 Figure 1 Figure generation ethnic minority groups (henceforth, migrant groups). The data on the second generation ethnic minority groups were excluded because of smaller numbers. The final cohort comprised 7,756,836 persons (Figure 1). For this study ethnic Dutch (n=7,045,374) and migrants originating from Suriname (n=108,852), Morocco (n=39,859), Turkey (n=79,056), Netherlands Antilles (n=29,309), Indonesia (n=234,290) and China (n=10,130) were included. Migration histories of these populations have been discussed in detail elsewhere.15;16

Neighbourhood socioeconomic status Socioeconomic status (SES) was based on income data registered in the RIS.16 The RIS started in 1994, when a representative sample of 1.9 million Dutch citizens was selected. Every year, the sample was corrected for emigration and mortality on one hand, and immigration and birth on the other hand. All persons belonging to the households of the sample population (about one third of the Dutch population) were included in the RIS. To be able to correct for SES in all study participants, SES was based on the neighbourhood income in the year prior to baseline. In each neighbourhood, the mean disposable income of all residents with income data available was calculated for 1997 and subsequently assigned to all residents living in that neighbourhood on 1 January 1997. Neighbourhood income was divided into tertiles, with the first tertile representing the lowest income group.

110

Ethnic inequalities in stroke incidence by stroke subtype and sex

Others N=333,662

Chinese N=9,398

inclusion procedure

N=120,577 Indonesian

Flowchart inclusion of procedure of

1 Figure

Antillean N=27,033

Turkish N=78,616 Final cohort cohort Final N=7,423,174 N=8,185,247 N=8,185,247 N=7,756,836 N=15,431,715 N=15,431,715 1 January 1998 N=13,421,681 age of years ≥30 Population Register Register Population PR unique persons PR unique

groups migrant and Dutch Ethnic

N=39,714 Moroccan

N=102,462

Surinamese

Flowchart inclusion of procedure

N=7,045,374 Dutch Ethnic Figure 1 Figure

111

Chapter 3.1

Comorbidity Presence and extent of comorbidity were determined with the Charlson index score based on discharge diagnosis within three years prior to baseline (1995-1997).17 The Charlson index score ranges from zero to six (cut-off value), with zero representing no comorbidity. This score was subsequently divided into four categories (0, 1-2, 3-4, ≥5). The Charlson index score had been proven to be a reliable and valid method for measuring comorbidity in clinical research.17

Data analysis Baseline characteristics were analysed at 1 January 1998. Absolute incidence rates of stroke subtypes stratified by ethnic group and sex were calculated as the number of events per 100,000 person-years at risk, and subsequently age-standardised to the age distribution of the European population (using the direct method) with ten-year age bands. For the calculation of incidence differences in stroke subtypes between migrant groups and ethnic Dutch (reference), Cox proportional hazard regression analyses were used. Analyses were stratified by sex because of interaction with ethnicity. To adjust for potential confounders, analyses were corrected for age at baseline, marital status, neighbourhood income, degree of urbanisation and Charlson comorbidity index. Results were expressed as Hazard Ratios (HR) with accompanying 95% confidence intervals (95% CI). The results were considered significant if the confidence intervals did not overlap. We used SPSS software, version 14.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation Netherlands.

RESULTS

Baseline characteristics Table 1 shows the characteristics of the study population. In general, the migrant groups were younger (except for Indonesians), more concentrated in low income urban areas, and had less comorbidity than the ethnic Dutch population. Turkish and Moroccans were the most, and Antilleans and Surinamese were the least to be married or living with a partner. Table 2 shows the absolute number of events and the age-standardised incidence of the various stroke subtypes by sex. Ethnic differences in the incidence of the various stroke subtypes in men and women are shown in Figure 2a and 2b.

All stroke types When all the various stroke subtypes were combined, Surinamese (HR=1.43, 95% CI: 1.35-1.50), Antillean (HR=1.16, 95% CI: 1.03-1.32) and Indonesian (HR=1.04, 95% CI: 1.00-1.07) men had a higher incidence while Moroccan (HR=0.42, 95% CI: 0.36-0.48) and Chinese (HR=0.83, 95% CI: 0.69- 0.99) men had a lower incidence compared with ethnic Dutch men (Figure 2a). Similar findings were

112

Ethnic inequalities in stroke incidence by stroke subtype and sex

54) -

5.7 5.7 2.9 0.6 0.2

50.3 76.6 53.3 27.0 19.6 38.6 26.8 17.1 11.8 96.4

9,398 90,243 Chinese 43 (36

69) very rural=<251 -

,

5.0 5.0 6.1 1.8 0.7

45.4 59.6 33.8 30.0 36.3 25.2 32.1 23.9 13.8 91.5 500 120,577 1,158,498 57 (47 Indonesian

50) -

7.2 7.2 2.3 4.9 0.9 0.3

47.3 38.0 56.9 23.1 20.0 41.8 32.2 16.5 94.0 27,033 244,396 Antillean 42 (36

52) -

8.0 8.0 9.1 0.9 4.8 0.7 0.2

55.5 85.3 72.5 19.5 40.4 32.5 17.1 94.3 between 1998 2010 and between

78,616 Turkish 773,828 40 (34

49) -

8.7 8.7 1.9 2.0 0.2 0.0

58.0 80.4 62.6 23.0 14.4 41.4 28.8 19.2 97.8 39,714 373,476 Moroccan 39 (33

52) -

3.9 3.9 1.3 5.7 1.3 0.6

45.7 41.8 63.5 18.7 17.7 63.1 19.9 11.9 92.4 102,462 973,425 42 (36 Surinamese

) at baseline. Very urban=>2000, urban=1001-2000, urban/rural=501-1000, rural=251- urban/rural=501-1000, urban=1001-2000, urban=>2000, Very baseline. at )

2

64) km the migrant groups and the ethnic Dutch population -

5.9 5.9 1.6 0.5

48.2 69.2 29.3 34.5 36.2 14.8 23.2 22.1 23.7 16.1 92.0 7,045,374 50 (40 70,564,803

Dutch Ethnic

b 0

>4 b 3-4 3-4 1-2 1-2

Rural d Urban Urban

ndex % ndex c

rural Very Very urban

Urban/rural Urban/rural

)

a (lowest income)

(highest income) (highest (medium income) IQR

risk at years - Tertile 1 Baseline population characteristics of Tertile 3

Tertile 2

N Men % ( Median age % together or living Married SES Neighbourhood % % urbanisation of Degree i comorbidity Charlson Person At baseline Population density (number of residents per Neighbourhood SES: socioeconomic status based on neighbourhood income in 1997

Interquartile range Interquartile

1 Table a b c d

113

Chapter 3.1

Table 2 Age-standardiseda absolute incidence rates of stroke subtypes per 100,000 person-years at risk by sex Total Men Women N events IRb (95% CI) N events IRb (95% CI) N events IRb (95% CI) All stroke types Ethnic Dutch 298,177 300 (299-301) 139,794 332 (330-334) 158,383 274 (273-275) Surinamese 3,280 370 (354-387) 1,573 418 (386-449) 1,707 338 (319-358) Moroccan 251 89 (60-118) 173 75 (61-90) 78 84 (40-127) Turkish 1,404 193 (170-216) 868 249 (187-311) 536 154 (129-179) Antillean 541 314 (270-358) 259 386 (289-483) 282 279 (230-328) Indonesian 6,838 312 (304-320) 3,021 339 (325-354) 3,817 291 (281-301) Chinese 201 229 (192-265) 114 248 (194-303) 87 199 (154-244) IS Ethnic Dutch 174,448 165 (164-166) 86,502 189 (188-191) 87,946 146 (145-147) Surinamese 2,111 216 (204-228) 1,040 250 (228-272) 1,071 193 (179-207) Moroccan 145 31 (25-36) 96 36 (28-45) 49 20 (14-27) Turkish 937 117 (101-132) 587 166 (120-212) 350 85 (70-100) Antillean 346 189 (156-221) 169 224 (156-293) 177 170 (133-207) Indonesian 3,669 160 (154-166) 1,755 184 (174-194) 1,914 142 (136-149) Chinese 106 109 (85-134) 59 113 (79-146) 47 99 (68-130) ICS Ethnic Dutch 27,658 26 (26-26) 13,575 30 (30-31) 14,083 23 (23-23) Surinamese 418 42 (37-47) 220 59 (48-69) 198 33 (27-38) Moroccan 31 7 (4-11) 24 8 (4-11) <10c 7 (0-16) Turkish 186 23 (15-30) 122 28 (18-38) 64 15 (7-23) Antillean 58 21 (13-30) 36 40 (18-62) 22 12 (4-20) Indonesian 1,002 42 (39-45) 481 49 (44-54) 521 37 (34-40) Chinese 46 46 (31-61) 30 59 (34-85) 16 34 (16-52) SAH Ethnic Dutch 8,752 8 (7-8) 3,144 6 (6-6) 5,608 9 (9-9) Surinamese 150 11 (9-13) 45 7 (4-9) 105 13 (10-16) Moroccan 15 2 (1-3) <10c 2 (0-3) <10c 2 (0-4) Turkish 60 5 (4-6) 35 5 (3-6) 33 5 (3-7) Antillean 32 6 (4-8) <10c 4 (1-6) 23 8 (5-12) Indonesian 223 10 (9-12) 56 6 (4-8) 167 14 (11-16) Chinese <10c 4 (0-7) <10c 3 (0-7) <10c 5 (0-10) a Standardised to the age distribution of the Dutch population in 1998 b Incidence rate per 100,000 person-years at risk c Actual number of cases not given in line with the Dutch data protection guideline as the cases were less than ten IS: ischemic stroke; ICS: intracerebral haemorrhage; SAH: subarachnoid haemorrhage observed among women, with Surinamese (HR=1.34, 95% CI: 1.28-1.41) and Indonesian (HR=1.07, 95% CI: 1.04-1.11) women having a higher incidence, while Moroccan (HR=0.37, 95% CI: 0.30-0.46) and Chinese (HR=0.78, 95% CI: 0.63-0.97) women having a lower incidence compared with ethnic Dutch (Figure 2b). There were no differences between Antillean, Turkish and ethnic Dutch women.

114

Ethnic inequalities in stroke incidence by stroke subtype and sex

Ischemic stroke Surinamese men had a higher (HR=1.46, 95% CI: 1.37-1.55) whereas Moroccan men (HR=0.34, 95% CI: 0.28-0.41) and Chinese men (HR=0.66, 95% CI: 0.51-0.86) had a lower incidence of IS compared with ethnic Dutch men (Figure 2a). No significant differences were observed between the other migrant groups and ethnic Dutch. Among women, the incidence of IS was higher in Surinamese (HR=1.38, 95% CI: 1.29-1.46), but lower in Moroccans (HR=0.33, 95% CI: 0.25-0.44), Turkish (HR=0.89, 95% CI: 0.80-0.99), and Chinese (HR=0.68, 95% CI: 0.51-0.91) than in ethnic Dutch women (Figure 2b).

Intracerebral haemorrhage Among men, the incidence of ICH was higher in all migrant groups than in ethnic Dutch men, ranging from HR=1.48 (95% CI: 1.23-1.77) in Antilleans to HR=2.29 (95% CI: 1.60-3.28) in Chinese, except for the lower incidence in Moroccans (HR=0.52, 95% CI: 0.37-0.74) (Figure 2a). Among women, Surinamese (HR=1.74, 95% CI: 1.50-2.00), Indonesian (HR=1.65, 95% CI: 1.55-1.86), and Chinese (HR=1.61: 95% CI, 0.98-2.63) had a higher incidence of ICH than ethnic Dutch, while Moroccans (HR=0.34, 95% CI: 0.16-0.72) had a lower incidence of ICH than ethnic Dutch (Figure 2b). There were no significant differences in the incidence of ICH between Antillean, Turkish and ethnic Dutch women.

Subarachnoid haemorrhage Among men, there were no significant differences in the incidence of SAH between migrant groups and ethnic Dutch except for the lower incidence in Moroccans (HR=0.42, 95% CI: 0.20-0.88). Among women, Surinamese (HR=1.26, 95% CI: 1.04-1.54) and Indonesians (HR=1.61, 95% CI: 1.38-1.88) had a higher incidence, whereas Moroccans (HR=0.34, 95% CI: 0.17-0.68) and Turkish (HR=0.62, 95% CI: 0.44-0.87) had a lower incidence of SAH than ethnic Dutch.

115

Chapter 3.1

HR (95% CI)* All stroke types Surinamese 1.43 (1.35, 1.50) Moroccan 0.42 (0.36, 0.48) Turkish 0.99 (0.93, 1.06) Antillean 1.16 (1.03, 1.32) Indonesian 1.04 (1.00, 1.07) Chinese 0.83 (0.69, 0.99)

Ischemic stroke Surinamese 1.46 (1.37, 1.55) Moroccan 0.34 (0.28, 0.41) Turkish 0.98 (0.90, 1.06) Antillean 1.12 (0.96, 1.31) Indonesian 0.98 (0.94, 1.03) Chinese 0.66 (0.51, 0.86)

Intracerebral haemorrhage Surinamese 2.08 (1.82, 2.39) Moroccan 0.61 (0.41, 0.92) Turkish 1.48 (1.23, 1.77) Antillean 1.76 (1.27, 2.45) Indonesian 1.70 (1.55, 1.86) Chinese 2.29 (1.60, 3.28)

Subarachnoid haemorrhage Surinamese 1.25 (0.92, 1.69) Moroccan 0.42 (0.20, 0.88) Turkish 1.01 (0.72, 1.41) Antillean 1.08 (0.56, 2.07) Indonesian 1.04 (0.80, 1.35) Chinese 0.51 (0.13, 2.06)

0.10 0.20 0.50 1.00 2.00 5.00 Figure 2a Ethnic differences in the incidence of stroke subtypes in men (reference is ethnic Dutch men) *Adjusted for age, marital status, degree of urbanisation, neighbourhood SES, Charlson index

116

Ethnic inequalities in stroke incidence by stroke subtype and sex

HR (95% CI)* All stroke types Surinamese 1.34 (1.28, 1.41) Moroccan 0.37 (0.30, 0.46) Turkish 0.92 (0.84, 1.00) Antillean 1.07 (0.95, 1.20) Indonesian 1.07 (1.04, 1.11) Chinese 0.78 (0.63, 0.97)

Ischemic stroke Surinamese 1.38 (1.29, 1.46) Moroccan 0.33 (0.25, 0.44) Turkish 0.89 (0.80, 0.99) Antillean 1.06 (0.92, 1.23)

Indonesian 0.98 (0.94, 1.03) Chinese 0.68 (0.51, 0.91)

Intracerebral haemorrhage 0.44 (0.14, 1.38) Surinamese 1.74 (1.50, 2.00) Moroccan 0.32 (0.16, 0.72) Turkish 1.15 (0.90, 1.48) Antillean 0.93 (0.61, 1.42)

Indonesian 1.65 (1.55, 1.86) Chinese 1.61 (0.98, 2.63)

Subarachnoid haemorrhage Surinamese 1.26 (1.04, 1.54) Moroccan 0.34 (0.17, 0.68) Turkish 0.62 (0.44, 0.87) Antillean 1.16 (0.77, 1.75) Indonesian 1.61 (1.38, 1.88) Chinese 0.44 (0.14, 1.38)

0.10 0.20 0.50 1.00 2.00 5.00

Figure 2b Ethnic differences in the incidence of stroke subtypes in women (reference is ethnic Dutch women) *Adjusted for age, marital status, degree of urbanisation, neighbourhood SES, Charlson index

117

Chapter 3.1

DISCUSSION

Previous studies have generally shown unfavourable risks of stroke subtypes among ethnic minority groups as compared with the majority populations, but we could not confirm this for all migrant groups in our study. This unfavourable pattern was confirmed for the Surinamese population in our study. By contrast, Moroccans appear to have a lower risk for almost all of the various stroke subtypes compared with the ethnic Dutch population. For the other ethnic groups, we found mixed results, with the risks depending on stroke subtype and sex. We can only speculate on the reasons for the large differences in the risk of the various stroke subtypes observed among the ethnic groups in our current study. The ethnic differences reflect on the variations in the underlying risk factors. The relatively high incidence of stroke subtypes in the Surinamese population of predominantly West-African or South-Asian descent, for example, may in part be due to their unfavourable cardiovascular risk profile. In a previous study we found that the prevalence of hypertension was two- to three-fold higher in African Surinamese and South-Asian Surinamese people than in ethnic Dutch people.18 Uncontrolled hypertension is common among Surinamese populations, particularly among the African-Surinamese men.18 In addition, diabetes and obesity are more common among ethnic minority groups as compared with ethnic Dutch people.19-21 The high incidence of the various stroke subtypes found among some of the migrant groups in the current study is consistent with the high incidence found among West- African and South-Asian descent populations in the UK,8-10 African-American and Hispanic people in the United States of America,22-23 and Maori/Pacific people in New Zeeland.24 The lower risk of all stroke subtypes found among Moroccan migrants is in contrast with the aforementioned earlier studies;8-10,22-24 and is in general agreement with the lower incidence rates found for other cardiovascular diseases such as CHD in the Netherlands.16;25 Although type 2 diabetes is highly prevalent among Moroccans,20;21 they have a low prevalence of hypertension compared with ethnic Dutch people.26 In a previous study, the prevalence of hypertension was 26.1% and 19.6% in Moroccan men and women compared with 48.8% and 35.0% in ethnic Dutch men and women.26 The rate of smoking has also been shown to be very low among Moroccans in the Netherlands.20 The Mediterranean diet, which has cardio-protective effects,27 may also contribute to their favourable cardiovascular disease outcomes. The low prevalence of hypertension and smoking and the Mediterranean diet may contribute, at least in part, to the favourable stroke outcomes observed among Turkish women in this study.20;26;27 The low incidence of overall stroke in Chinese migrants is consistent with the low incidence rate found in the Scottish Health and Ethnicity Linkage study.6 However, this current study shows that although the overall stroke incidence is lower, the incidence of ICH is higher in Chinese than in ethnic Dutch emphasizing the need to assess various stroke subtypes in different ethnic groups. The high incidence of ICH among Chinese migrants is consistent with the high incidence found in China.28

118

Ethnic inequalities in stroke incidence by stroke subtype and sex

ICH is one of the leading causes of death and disability in China,28 and has been linked to the apolipoprotein E (APOE) polymorphism.29 In Western populations, for example, APOE usually accounts for about 10% of all strokes, but in many East-Asian populations such as China, it accounts for approximately one third of all strokes.29;30 These findings suggest the need to assess the contribution of genetics to the high incidence of ICH observed in other migrant groups, particularly in men. Our findings have important clinical and public implications and emphasise the need to address the underlying risk factors such as hypertension, in particular among those with a high risk of stroke such as Surinamese and Indonesian migrants. In most of these groups, the underlying modifiable risk factors for stroke have been well characterised.18-21;26 Further studies are needed to identify the root causes of these ethnic differences in stroke outcomes including the various subtypes of ischemic stroke. Such information is crucial for designing effective primary and secondary preventative programmes that target those at high risk.

Considerations A major strength of our current study is that it is based on a nationwide database with large sample sizes, which allowed us to study incidence of the various types of stroke in several migrant groups stratified by sex. There are also limitations to our study. Inherent to many national level databases, we lack data on risk factors such as hypertension and diabetes, and therefore we were unable to assess the direct contribution of the various risk factors to the observed ethnic differences. Nevertheless, we were able to shed light on the potential contributing factors to the observed difference thanks to the previous studies on ethnic differences in risk factors in the Netherlands.18- 21;26 Besides, we lack data on the various subtypes of IS. Moreover, we were unable to further stratify the analyses by age due to small numbers. The observed differences between the ethnic groups warrant a national level effort to identify the root causes of the ethnic differences in the various stroke types including data collection on the subtypes of IS in the Netherlands. Another limitation is that the ethnic groups were constructed on the basis of country of birth. Country of birth may reflect ethnicity fairly well in some ethnic groups such as Moroccans,15 but it is likely to be an unreliable measure of ethnicity for other groups such as Surinamese who are ethnically diverse. Notwithstanding, our earlier studies showed that major risk factors for stroke such as hypertension and type 2 diabetes are higher in both African Surinamese and South-Asian Surinamese than in the ethnic Dutch population.18;19

119

Chapter 3.1

Conclusion In conclusion, our current study has revealed ethnic differences in the risk of stroke subtypes. Surinamese people have an increased risk while Moroccan people have a reduced risk for almost all stroke subtypes as compared to the ethnic Dutch population. Among the other migrant groups, the risk appears to depend on the stroke subtype and sex. These findings underscore the need to identify the root causes of these ethnic differences in stroke outcomes. Such informations is essential for planning effective primary and secondary prevention programmes that target those at high risk such as Surinamese.

120

Ethnic inequalities in stroke incidence by stroke subtype and sex

REFERENCES 1. Lozano R, Naghavi M, Foreman K, Lim S, Shibuya K, Aboyans V, et al. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012;380:2095. 2. Feigin VL. Forouzanfar MH, Krishnamurthi R, Mensah GA, Connor M, Bennett DA, et al; Global Burden of Diseases, Injuries, and Risk Factors Study 2010 (GBD 2010) and the GBD Stroke Experts Group. Global and regional burden of stroke during 1990–2010: findings from the Global Burden of Disease Study 2010. Lancet 2014;383(9913):245-54 3. Bos V, Kunst AE, Garssen J, Mackenbach JP. Socioeconomic inequalities in mortality within ethnic groups in the Netherlands, 1995–2000. J Epidemiol Community Health 2005;59:329-35. 4. Harding S, Rosato M, Teyhan A. Trends for coronary heart disease and stroke mortality among migrants in England and Wales, 1979–2003: slow declines notable for some groups. Heart 2008;94: 463-70. 5. Agyemang C, Vaartjes I, Bots ML, van Valkengoed IG, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95:747-53. 6. Bhopal RS, Bansal N, Fischbacher CM, Brown H, Capewell S; Scottish Health and Ethnic Linkage Study. Ethnic variations in the incidence and mortality of stroke in the Scottish Health and Ethnicity Linkage Study of 4.65 million people. Eur J Prev Cardiol 2012;19:1503-8. 7. Stewart JA, Dundas R, Howard RS, Rudd AG, Wolfe CD. Ethnic differences in incidence of stroke: prospective study with stroke register. BMJ 1999;318:967–71. 8. Wolfe CD, Rudd AG, Howard R Coshall C, Stewart J, Lawrence E, et al. Incidence and case fatality rates of stroke subtypes in a multiethnic population: the South London Stroke Register. J Neurol Neurosurg Psychiatry 2002;72:211–216. 9. Banerjee S, Biram R, Chataway J, Ames D. South Asian strokes: lessons from the St Mary's stroke database. QJM 2010;103:17-21. 10. Markus HS, Khan U, Birns J, Evans A, Kalra L, Rudd AG, et al. Differences in stroke subtypes between black and white patients with stroke: the South London Ethnicity and Stroke Study. Circulation 2007;116: 2157-64. 11. Schulz UGR, Rothwell PM. Differences in vascular risk factors between etiological subtypes of ischemic stroke: importance of population-based studies. Stroke 2003;34:2050–59. 12. Harteloh P, de Bruin K, Kardaun J. The reliability of cause-of-death coding in The Netherlands. Eur J Epidemiol 2010;25:531-8. 13. Ellekjaer H, Holmen J, Kruger O, Terent A. Identification of incident stroke in Norway: hospital discharge data compared with a population-based stroke register. Stroke 1999;30:56-60. 14. Nieuwkamp DJ, Vaartjes I, Algra A, Rinkel GJ, Bots ML. Risk of cardiovascular events and death in the life after aneurysmal subarachnoid haemorrhage: a nationwide study. Int J Stroke 2012 [Epub ahead of print]. 15. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14:255-69. 16. Agyemang C, van Oeffelen AA, Bots ML, Stronks K, Vaartjes I. Socioeconomic inequalities in acute myocardial infarction incidence in migrant groups: has the epidemic arrived? Analysis of nation-wide data. Heart 2014;100:239-46. 17. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57:1288-94.

121

Chapter 3.1

18. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23:1971-7. 19. Agyemang C, Kunst AE, Bhopal R, Anujuo K, Zaninotto P, Nazroo J, et al. Diabetes prevalence in populations of South Asian Indian and African origins: a comparison of England and the Netherlands. Epidemiology 2011;22:563-7. 20. Ujcic-Voortman JK, Baan CA, Seidell JC, Verhoeff AP. Obesity and cardiovascular disease risk among Turkish and Moroccan migrant groups in Europe: a systematic review. Obes Rev 2012;13:2-16. 21. Ujcic-Voortman JK, Schram MT, Jacobs-van der Bruggen MA, Verhoeff AP, Baan CA. Diabetes prevalence and risk factors among ethnic minorities. Eur J Public Health 2009;19:511-5. 22. Sacco RL, Boden-Albala B, Gan R, Chen X, Kargman DE, Shea S, et al. Stroke incidence among white, black, and Hispanic residents of an urban community: the Northern Manhattan Stroke Study. Am J Epidemiol 1998;147:259–68. 23. Broderick JP, Brott T, Tomsick T, Huster G, Miller R. The risk of subarachnoid and intracerebral hemorrhages in blacks as compared with whites. N Engl J Med 1992;326: 733. 24. Feigin V, Carter K, Hackett M, Barber PA, McNaughton H, Dyall L, et al; Auckland Regional Community Stroke Study Group. Ethnic disparities in incidence of stroke subtypes: Auckland Regional Community Stroke Study, 2002-2003. Lancet Neurol 2006;5:130-9. 25. van Oeffelen AA, Vaartjes I, Stronks K, Bots ML, Agyemang C. Incidence of acute myocardial infarction in first and second generation minority groups: does the second generation converge towards the majority population? Int J Cardiol 2013;168:5422-9. 26. Agyemang C, Ujcic-Voortman J, Uitenbroek D, Foets M, Droomers M. Prevalence and management of hypertension among Turkish, Moroccan and native Dutch ethnic groups in Amsterdam, the Netherlands: The Amsterdam Health Monitor Survey. J Hypertens 2006;24:2169-76. 27. Misirli G, Benetou V, Lagiou P, Bamia C, Trichopoulos D, Trichopoulou A. Relation of the traditional Mediterranean diet to cerebrovascular disease in a Mediterranean population. Am J Epidemiol 2012;176:1185-92. 28. Shi FL, Hart RG, Sherman DG, Tegeler CH. Stroke in the People's Republic of China. Stroke 1989;20:1581–1585. 29. Tzourio C, Arima H, Harrap S, Anderson C, Godin O, Woodward M, et al. APOE genotype, ethnicity, and the risk of cerebral haemorrhage. Neurology 2008;70:1322-8. 30. Zhang LF, Yang J, Hong Z, Yuan GG, Zhou BF, Zhao LC, et al. Proportion of different subtypes of stroke in China. Stroke 2003;34:2091–2096.

122

CHAPTER 3.2

Socioeconomic inequalities in stroke incidence by country of birth

Agyemang C, van Oeffelen AAM, Norredam M, Kapelle LJ, Klijn CJM, Bots ML, Stronks K, Vaartjes I. Socioeconomic inequalities in stroke incidence among migrant groups: Analysis of nation-wide data.

Submitted. Chapter 3.2

ABSTRACT Background Data on socioeconomic inequalities in stroke incidence are scarce among ethnic minorities. We assessed socioeconomic inequalities in relation to stroke incidence within major first generation ethnic minority groups (henceforth, migrants). In addition, we assessed the incidence of stroke between migrants and the ethnic Dutch population within income groups.

Methods A nationwide register-based cohort study was conducted (n=2,397,446) between 1 January 1998 and 31 December 2010 among ethnic Dutch and migrant groups. Standardised disposable household income was used as a measure of socioeconomic status. Cox proportional hazard models were used to estimate socioeconomic differentials in stroke incidence.

Results Among ethnic Dutch, the incidence of stroke was higher in the low-income group than in the high- income group (adjusted Hazard Ratio (HR) was 1.18, 95% confidence interval (CI): 1.16-1.20). Similar socioeconomic inequalities in stroke incidence were found among Surinamese (HR=1.36, 95%CI: 1.17-1.58), Indonesians (HR=1.15, 95% CI: 1.03-1.28), Moroccans (HR=1.54, 95% CI: 0.97-2.43), and Turkish (HR=1.19, 95% CI: 0.97-1.46) and to a lesser extent among Antilleans (HR=1.24, 95% CI: 0.84- 1.84). Compared with ethnic Dutch, the incidence of stroke was lower in Moroccans, similar in Turkish, but higher in Surinamese among all income groups. The incidence of stroke was higher in Indonesian low-income and high-income groups than in their ethnic Dutch counterparts. Among Antilleans, the risk of stroke was higher than ethnic Dutch but only in the low-income group.

Conclusion Our findings reveal socioeconomic inequalities in stroke incidence among all ethnic groups. Within each income group, the incidence of stroke was higher in most migrant groups than in ethnic Dutch. Reduction of socioeconomic inequalities in stroke incidence among all ethnic groups may lead to a major public health improvement for all. Policy measures tackling socioeconomic inequalities should take the increased risk of stroke among ethnic minority populations into account.

124

Socioeconomic inequalities in stroke incidence by country of birth

INTRODUCTION

Stroke is a major global health burden.1 According to the World Health Organisation estimates, some 15 million people suffer from stroke worldwide each year.2 Of these, 5 million die and another 5 million are left permanently disabled. Studies in many industrialised countries demonstrate that individuals with low socioeconomic status (SES), whether measured by income, occupational group or by level of educational attainment, have an elevated risk of dying from cardiovascular diseases (CVD) such as stroke.3-5 In a review of SES and stroke by Cox et al in 2006,3 and a recent updated review of SES and stroke in 2012 by Addo et al,4 a consistent pattern of higher stroke incidence and mortality was found in lower SES groups than in high SES groups. The mechanisms through which low SES affects stroke risk and outcomes are unclear, but some studies suggest that the comparatively high prevalence of predisposing risk factors such as hypertension, diabetes, and smoking among poorer people could account for some of the variation.6;7 As a result, reducing socioeconomic inequalities in health has been an important goal of health policy.8 While socioeconomic inequalities in CVD such as stroke have long been documented among European populations, these socioeconomic inequalities in CVD have remained inconsistent among migrant populations.9-13 The general expectation is that a socioeconomic gradient will eventually emerge in migrant groups11;13 in line with the ‘diffusion theory’ of the epidemic of coronary heart disease (CHD).14 This theory suggests that CHD struck more affluent communities of the rich nations first as they were the first who could afford unhealthy behaviours which increase the risk of CHD. However, as the epidemic matures, the disease spread to the poor communities as living standards improved. When the epidemic started to decline, the rich people were again the first to benefit as they were the first to adopt the healthy behavioural changes. Evidence also suggests widening inequalities in both income and CVD risk factors in different populations in recent years, but the effects of these changes on stroke outcomes, particularly among ethnic minority groups, are still unknown.15;16 The main aim of this study was to assess socioeconomic inequalities in stroke incidence among major first generation ethnic minority groups (henceforth, migrants) in the Netherlands (Surinamese, Moroccans, Turkish, Antilleans, Indonesian) and the ethnic Dutch population. In addition, we assessed the difference in incidence of stroke between migrants and ethnic Dutch within income groups.

METHODS

Cohort enrolment Data were obtained from the following Dutch national registers: Population Register (PR), Hospital Discharge Register (HDR), Cause of Death Register (CDR), and Regional Income Survey (RIS). The registers were used to acquire information on demographic factors, stroke hospitalisations and comorbidities, fatal stroke events, and income. The registers have been described in detail

125

Chapter 3.2 previously.17 The reliability of Dutch national registers for stroke has been proven to be high.18 By linking previous registers with a personal identifier a cohort was built, starting at 1 January 1998 when 15,431,715 Dutch citizens were registered in the PR. As only PR unique persons can be identified in the HDR (unique with respect to the combination of the variables date of birth, sex, and four digits of the postal code), non-unique persons were excluded. This left a cohort of 13,421,681 (87.0%) persons. Because a low number of strokes are expected in the young, only persons of 30 years of age and older were included (n=8,185,247 (61.0%)). The SES indicator was based on the disposable income on household level in 1997 (the year prior to baseline), which was available for about one third of the Dutch population. Persons without available information about their income were excluded (n=5,543,796 (67.7%)). After excluding migrant groups not under study (n=106,635), the final cohort comprised 2,397,446 persons (Figure 1).

Follow-up Persons were followed up from baseline until their first stroke. A first event was defined as any hospital admission with a primary or secondary diagnosis stroke (ICD-9 code 430-438), or as a fatal event with stroke as the primary or secondary cause (ICD-10 code I60-I69, G45). The positive predictive values of these codes have been shown acceptable; ≥75% for code 430, ≥85% for code 431, and ≥85% for codes 434-436.19-21 Persons who already had a previous hospital admission for a stroke between 1 January 1995 and 31 December 1997 were not included as having a first event of that specific type. Persons were censored in case of death, emigration, or the end of the study period (31 December 2010), whichever came first.

Determinants Ethnic background The migrant groups were constructed based on the country of birth of the resident and his/her parents, according to the definition of Statistics Netherlands.20 A person is considered a migrant if he/she was born abroad and at least one of the parents was born abroad. For this study migrants from Suriname (n=33,785), Morocco (n=15,114), Turkey (n=31,206), Netherlands Antilles (n=8,145), and Indonesia (n=37,707) were included. Persons of whom both parents were born in the Netherlands were indicated as ethnic Dutch (n=2,271,489). For simplicity, we refer to the first generation ethnic groups in general as migrant groups and the respective countries of origin as Surinamese, Moroccans, Turkish, Antilleans, and Indonesians. The migration histories of these populations have described in detail elsewhere.22;23

126

Socioeconomic inequalities in stroke incidence by country of birth

Others

N=106,635

N=37,707 N=37,707 Indonesian

N=8,145 N=8,145 Antillean

Turkish N=31,206 N=31,206

N=7,756,836 N=2,504,081 N=2,504,081 N=8,185,247 N=15,431,715 N=13,421,681 N=13,421,681 1 January 1998 ≥30 years of age of years ≥30 Unique persons Population Register Register Population

Ethnic Dutch and migrant groups migrant and Dutch Ethnic Household income data available data income Household

cohort Final N=2,397,446

N=15,114 N=15,114 Moroccan

Survey N=33,785 N=33,785 Surinamese

Income Regional

inclusionFlowchart of procedure N=2,271,489 N=2,271,489 Dutch Ethnic

1 Figure

127

Chapter 3.2

Socioeconomic status The socioeconomic status (SES) was based on income data registered in the RIS.24 The RIS started in 1994 with a selected representative sample of 1.9 million Dutch citizens. Each year, the sample was corrected for emigration and mortality on one hand, and immigration and birth on the other hand. All individuals that belonged to the households of the sample population (approximately one third of the Dutch population) were included in the RIS. SES was defined as the standardised disposable household income with adjustment for number of household members in the year prior to baseline. The standardised disposable household income was subsequently categorised into SES tertiles in two ways: 1) by dividing the income of the total cohort in tertiles (‘total’ income), 2) by dividing the income of every migrant group separately in tertiles (‘within group’ income). The first tertile represents the low-income group, the second tertile represents the medium-income group, and the third tertile represents the high-income group.

Comorbidity Charlson index score was used to assess the presence and extent of comorbidity, based on discharge diagnosis within three years prior to baseline (1995-1997).25 The Charlson index score ranges from zero to six (cut-off value), with zero representing no comorbidity. This score was subsequently categorised into four categories (0, 1-2, 3-4, ≥5). The Charlson index score has proven to be a reliable and valid method to measure comorbidity in clinical research.26

Data analysis Baseline characteristics were analysed at 1 January 1998. Absolute incidence of stroke was calculated as the number of events per 100,000 person-years at risk, and subsequently age- standardised to the age distribution of the European population (using the direct method) with ten- year age bands. For the calculation of relative incidence rates of total CVA, Cox proportional hazard regression analyses were used. Using the ‘within group’ SES indicator, incidence differences between persons with a low SES and persons with a high SES (reference) within every migrant group and within ethnic Dutch were investigated. All analyses were adjusted for age, sex, marital status, degree of urbanisation, and Charlson index score. Incidence differences between migrant groups and ethnic Dutch (reference) within every SES tertile were investigated with the ‘total’ SES indicator. These analyses were additionally adjusted for disposable household income to overcome the problem of important income differences between ethnic groups. Results were expressed as Hazard Ratios (HR) with corresponding 95% confidence intervals (95% CI). We used SPSS software, version 14.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation Netherlands.

128

Socioeconomic inequalities in stroke incidence by country of birth

RESULTS

Characteristics of the study population The characteristics of the study population are shown in Table 1. On the whole, the migrant populations were younger than the ethnic Dutch population except for Indonesians. Turkish and Moroccans were the most and Antilleans and Surinamese were the least to be married or to live together with a partner. Indonesians had the highest and Moroccans had the lowest median household income. All the migrant populations tended to be more concentrated in urban centres than the ethnic Dutch.

Incidence rate by income gradient within each ethnic group Table 2 shows age-standardised absolute incidence rates per 100,000 person-years and Figure 2 shows HRs for stroke incidence by income group in each ethnic group. As expected, among ethnic Dutch, the incidence of stroke was higher in people with medium income and low income than in their high-income peers (HR=1.10, 95% CI: 1.08-1.12 for medium income and HR=1.18, 95% CI: 1.16- 1.20 for low income, respectively). Among the migrant groups, similar strong socioeconomic inequalities in stroke incidence were found. The incidence of stroke was higher in people with low income than in their high-income counterparts among Surinamese (HR=1.36, 95% CI: 1.17-1.58) and Indonesians (HR=1.15, 95% CI: 1.03-1.28), and to a lesser extent among Moroccans (HR=1.54, 95% CI: 0.97-2.43), Turkish (HR=1.19, 95% CI: 0.97-1.46) and Antilleans (HR=1.24, 95% CI: 0.84-1.84). Migrants with a medium income, except for Indonesians, also had a higher stroke incidence than their high-income counterparts although the differences were significant only for Turkish (HR=1.39, 95% CI: 1.14-1.69) and Antilleans (HR=1.63, 95% CI: 1.12-2.36).

Differences in stroke incidence between migrant groups and ethnic Dutch within income categories Figure 3 shows the differences in stroke incidence between migrant groups and ethnic Dutch stratified by each income tertile. Among the low-income groups, the incidence of stroke was higher among Surinamese (HR=1.47, 95% CI: 1.35-1.60) and Indonesians (HR=1.09, 95% CI: 1.01-1.19) than among ethnic Dutch. Antilleans also had a higher incidence of stroke than ethnic Dutch although the 95% CIs overlapped (HR=1.19, 95% CI: 0.97-1.46). By contrast, Moroccans had a lower incidence of stroke than ethnic Dutch (HR=0.42, 95% CI: 0.34-0.52). There was no difference between Turkish and ethnic Dutch low-income groups (HR=0.94, 95% CI: 0.85-1.05). Among the medium-income groups, Surinamese (HR=1.36, 95% CI: 1.20-1.55) and Antilleans (HR=1.35, 95% CI: 1.01-1.81) had higher incidence rates of stroke than ethnic Dutch. By contrast, the Moroccan medium-income group had a lower incidence of stroke than their ethnic

129

Chapter 3.2

66) -

5.5 5.5 5.6 1.7 0.6 47.7 68.5 27.3 33.9 38.8 22.9 32.7 24.6 14.2 92.1 1,924 37,707 16,839

371,665 55 (47

Indonesian (12,646-21,631) (12,646-21,631)

50)

-

7.7 7.7 2.6 4.5 0.8 0.3 500, very rural=<251 175 47.6 44.4 52.0 26.6 21.3 40.4 31.4 17.9 94.5 8,145 76,674 12,688 Antillean 43 (36 (8,613-17,818) (8,613-17,818)

53)

-

14,747) 8.5 9.2 0.9 5.1 0.8 0.2 580 55.0 87.9 63.3 28.3 39.9 32.9 17.1 93.9 31,206 11,333 Turkish 309,378 42 (35

(8,273-

-2000, urban/rural=501-1000, rural=251- urban/rural=501-1000, -2000,

50)

-

7.2 7.2 9.2 1.9 1.7 0.3 0.0 112 56.5 84.1 70.4 22.4 39.9 28.9 20.1 98.0

15,114 10,193 144,216 Moroccan 40 (34

(7,811-13,908)

51) -

4.1 4.1 1.3 5.4 1.1 0.5 47.1 48.0 47.6 30.5 22.0 61.8 20.5 12.3 93.0 1,007 33,785 13,498

322,541 43 (36 Surinamese (9,129-18,130)

) at baseline. Very urban=>2000, urban=1001

2

61) -

5.5 5.5 1.4 0.4 49.7 76.2 31.9 33.9 34.2 13.4 22.7 22.3 24.6 17.1 92.7 85,524 15,967

2,271,489 49 (40 23,270,517 Ethnic Dutch Ethnic (11,964-20,785) (11,964-20,785)

1997

0

>4 a 3-4 3-4 1-2 1-2

Rural Rural d

Urban

b rural Very Very urban

Urban/rural Urban/rural

a a c (IQR)

1997

n years at risk at years -

Tertile 1 (lowest income) Population characteristics of migrant groups and ethnic Dutch between 1998 2010 and Dutch ethnic of and groups migrant characteristics Population Tertile 3(highest income)

Tertile 2(medium income) 1

vents ased on household income in in income household on ased N Person E Men % (IQR) Median age % together or living Married Median income % SES Household % urbanisation of Degree % Index Charlson Population density (number of residents per km In euro’s i In euro’s At baseline

B Table a b c d IQR: interquartile range

130

Socioeconomic inequalities in stroke incidence by country of birth

Table 2 Age-standardised incidence rate of stroke per 100,000 person-years at risk, stratified by SESa Low-income group Medium-income group High-income group Events IRb Events IRb Events IRb Ethnic Dutch 36,571 273 (269-278) 25,527 259 (250-268) 23,426 246 (236-255) Surinamese 431 397 (305-489) 288 318 (270-366) 288 366 (151-582) Moroccan 45 108 (63-152) 36 199 (0-416) 31 54 (30-79) Turkish 190 178 (134-222) 218 221 (150-292) 172 145 (92-199) Antillean 58 260 (180-340) 69 309 (209-409) 48 279 (167-391) Indonesian 684 280 (248-312) 620 280 (214-345) 620 242 (212-273) a Socioeconomic status based on household disposable income within each ethnic group b Incidence rate per 100,000 person-years at risk, standardised to the age distribution of the European population

HR (95% CI)*

Medium versus high income

Ethnic Dutch 1.10 (1.08, 1.12)

Surinamese 1.12 (0.95, 1.32)

Moroccan 1.20 (0.74, 1.94)

Turkish 1.39 (1.14, 1.69)

Antillean 1.63 (1.12, 2.36)

Indonesian 1.00 (0.90, 1.12)

Low versus high income

Ethnic Dutch 1.18 (1.16, 1.20)

Surinamese 1.36 (1.17, 1.58)

Moroccan 1.54 (0.97, 2.43)

Turkish 1.19 (0.97, 1.46)

Antillean 1.24 (0.84, 1.84)

Indonesian 1.15 (1.03, 1.28)

0.5 1.0 2.0 5.0

Figure 2 Income inequalities in stroke incidence within each ethnic group *Adjusted for age, sex, marital status, degree of urbanisation, and Charlson index score Results were obtained from analyses using dummy variables for high income, medium income, and low income

131

Chapter 3.2

Dutch counterparts (HR=0.37, 95% CI: 0.24-0.57). There were no differences between ethnic Dutch and Indonesian and Turkish medium-income groups. Among the high-income groups, Surinamese (HR=1.32, 95% CI: 1.14-1.53) and Indonesians (HR=1.08, 95% CI: 1.00-1.16) had higher incidence rates of stroke than ethnic Dutch, whereas Moroccans had a lower incidence of stroke than their ethnic Dutch peers (HR=0.49, 95% CI: 0.25- 0.99). There were no differences between ethnic Dutch and Antilleans and Turkish migrants.

HR (95% CI)*

Low-income group

Surinamese 1.47 (1.35, 1.60)

Moroccan 0.42 (0.34, 0.52)

Turkish 0.94 (0.85, 1.05)

Antillean 1.19 (0.97, 1.46)

Indonesian 1.09 (1.01, 1.19)

Medium-income group

Surinamese 1.36 (1.20, 1.55)

Moroccan 0.37 (0.24, 0.57)

Turkish 0.95 (0.81, 1.11)

Antillean 1.35 (1.01, 1.81)

Indonesian 1.03 (0.95, 1.12)

High-income group

Surinamese 1.32 (1.14, 1.53)

Moroccan 0.49 (0.25, 0.99)

Turkish 0.80 (0.59, 1.10)

Antillean 1.12 (0.81, 1.57)

Indonesian 1.08 (1.00, 1.16)

0.2 0.5 1.0 2.0 Figure 3 Ethnic differences in stroke incidence within each income group *Adjusted for age, sex, marital status, degree of urbanisation, income, and Charlson index score Reference is ethnic Dutch Results were obtained from analyses using dummy variables for high-income, medium-income, and low-income

132

Socioeconomic inequalities in stroke incidence by country of birth

DISCUSSION

Our present findings suggest socioeconomic inequalities in the incidence of stroke in both ethnic Dutch and migrant groups. Within each income group, the incidence of stroke was higher in most migrant groups than in ethnic Dutch The importance of SES as predictors of stroke incidence, mortality, and impact has previously been discussed.3;4 Although socioeconomic inequalities in CVD have been recognised in industrialised communities, the evidence on ethnic minority groups has remained patchy and inconclusive.9-13 Analysis in the early 1970s by Marmot et al, for instance, found a strong social class gradient in CHD among the England and Wales general population, but did not find a relationship among South-Asian migrants living in England and Wales.10 Similar observations were found in the Netherlands in the early 1990s where strong socioeconomic inequalities in cardiovascular mortality were noted among the ethnic Dutch population, but not among migrant groups.12 Consequently, the need to keep track on socioeconomic inequalities with respect to stroke among ethnic minority populations has been emphasised.11;13 The higher risk of stroke among the poor in migrant communities in the present study is consistent with recent findings of socioeconomic inequalities in other CVD outcomes in migrant groups23;27 and fits into the popular ‘diffusion theory’ of the epidemic of cardiovascular mortality.14 Although the reasons for socioeconomic inequalities in stroke incidence are not entirely clear, the distribution of vascular risk factors across socioeconomic groups may play an important role.6-8 In the Netherlands, data on socioeconomic inequalities in risk factors among ethnic minority groups are rare and outdated, and they only partly support the observed socioeconomic inequalities in stroke outcomes.28;29 For example, the study of Nierkens et al in the early 2000s found higher smoking rates only among Surinamese, Turkish, and Moroccan men with lower SES than among those with higher SES.28 Another study on socioeconomic inequalities in metabolic syndrome in early 2000s found low education to be associated with increased prevalence of metabolic syndrome among ethnic Dutch, but not among ethnic minority groups.29 The lack of clear socioeconomic gradients in vascular risk factors among ethnic minority groups in these earlier Dutch studies are consistent with earlier studies from the UK.11;13 The emerging data showing clear socioeconomic inequalities in CVD outcomes among migrant populations support the need to assess the current socioeconomic gradient in relation to vascular risk factors such as hypertension, diabetes as well as access to health care, and compliance with therapy. Such an assessment might explain why low socioeconomic individuals have an elevated risk of stroke. Surinamese had a higher incidence of stroke than ethnic Dutch in all income strata. These differences persisted even after adjusting for income differentials with social strata, suggesting that other unmeasured factors might play a role in the observed ethnic differences in stroke incidence. Some of these factors might be the higher prevalence of hypertension and diabetes as well as

133

Chapter 3.2 poorer blood pressure control in Surinamese than in ethnic Dutch.30-32 Similar mechanisms might be present in Indonesians although data on risk factors among this population are currently lacking. By contrast, Moroccans had a lower stroke incidence than ethnic Dutch in all income groups. The low incidence of stroke among Moroccans as observed in the current study has been linked to the low prevalence of hypertension and smoking,33;34 and possibly in addition to their Mediterranean diet.35 Our current findings have important health implications. Thus far the literature on socioeconomic inequalities in relation to migrant health has largely focused on SES as an explanatory factor for ethnic differences in health outcomes.9;27-29 Our finding of socioeconomic inequalities in stroke incidence clearly suggests the need to pay attention to socioeconomic inequalities within migrant groups. However, the socioeconomic processes that shape health differentials both within and between ethnic groups are evidently multifaceted and context-specific.36 These complexities make the assessment of the role of SES on ethnic differences in health outcomes difficult, if not impossible, particularly for the first generation ethnic minorities in whom information on key SES indicators are generally lacking.36;37 The recent Scottish Health and Ethnicity Linkage Study found that across ethnic groups, SES indicators were inconsistently associated with CVD hospitalisation or mortality.27 Given these complexities, assessing socioeconomic inequalities in health outcomes within ethnic and migrant groups may be a more viable option.37 The magnitudes of the differences between the low-income and high-income groups among most migrant groups were generally greater than among the ethnic Dutch, despite the overlapping 95% confidence intervals for some migrant groups due to small numbers. If the stroke incidence rates of the low-income migrant groups could be reduced to the level of the high-income groups, particularly among migrant groups with high risks such as Surinamese and Indonesians, this would lead to a major health improvement in these groups regardless of the persistent ethnic disparities in stroke outcomes.

Considerations A key strength of our current study is that it is based on nationwide databases with large sample sizes, which allowed us to study incidence of stroke in several migrant groups. There are also limitations to our study. As in many national-level databases, we lack data on risk factors and therefore we were unable to assess the direct contribution of the various risk factors such as hypertension and diabetes to the observed socioeconomic inequalities. Moreover, our SES indicator was based on household income only, because data on other socioeconomic measures such as educational level, occupational class, and childhood SES were not available. It has been emphasised that different measures of SES may affect health through different pathways and causal mechanisms.38 Nevertheless, the disposable household income was based on taxes, which gives a more accurate picture on socioeconomic health differentials than the use of self-reported SES or

134

Socioeconomic inequalities in stroke incidence by country of birth other surrogate markers of SES such as neighbourhood income. It has been established that more accurate characterisation of SES leads to substantially wider health differentials than less precise measures of SES.39;40 Furthermore, we were unable to assess socioeconomic inequalities stratified by sex and in the various stroke subtypes due to smaller numbers. Lastly, the migrant groups were constructed on the basis of country of birth. Country of birth may represent ethnic groups fairly well in some migrant groups,22 but it might be an unreliable measure of ethnicity for other groups such as Surinamese who are ethnically diverse. Nevertheless, our previous studies showed that major risk factors for stroke such as hypertension and type 2 diabetes are higher in both African Surinamese and South-Asian Surinamese than in ethnic Dutch.31,32

Conclusion In conclusion, our findings show socioeconomic inequalities in stroke incidence among all ethnic groups. Within each income group, the incidence of stroke was higher in most migrant groups than in ethnic Dutch. Reduction of socioeconomic inequalities in stroke incidence among all ethnic groups may lead to a public health improvement for all. Policy measures tackling socioeconomic inequalities should take the increased risk of stroke among migrant populations into account.

135

Chapter 3.2

REFERENCES 1. Lozano R, Naghavi M, Foreman K, Lim S, Shibuya K, Aboyans V, et al. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012;380:2095.

2. World Health Organization, World Health Report 2002: Reducing Risks, Promoting Healthy Life. Geneva: World Health Organization 2002. 3. Cox AM, McKevitt C, Rudd AG, Wolfe CD. Socioeconomic status and stroke. Lancet Neurol 2006;5:181–188.

4. Addo J, Ayerbe L, Mohan KM, Crichton S, Sheldenkar A, Chen R, et al. Socioeconomic status and stroke: an updated review. Stroke 2012;43:1186-91.

5. Kunst A, del Rios M, Groenhof F, Mackenbach JP. Socioeconomic inequalities in stroke mortality among middle-aged men: an international overview. Stroke 1998;29:2285–91.

6. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation 1993;88:1973–88.

7. Hays DK, Greenlund KJ, Denny CH, Croft JB, Keenan NL. Racial/ethnic and socio-economic disparities in multiple risk factors for heart disease and stroke – United States, 2003. MMWR Morb Mortal Wkly Rep 2005;54:113–17.

8. Marmot M, Allen J, Bell R, Bloomer E, Goldblatt P. Consortium for the European Review of Social Determinants of Health and the Health Divide. WHO European review of social determinants of health and the health divide. Lancet 2012;380(9846):1011-29.

9. Nazroo JY. South Asian people and heart disease: an assessment of the importance of socioeconomic position. Ethn Dis 2001;11:401-11.

10. Marmot M, Mustard J. Coronary heart disease from a population perspective. In: Evans R, Barer M, Marmor T, eds. Why are some people healthy and others not? New York: Aldin de Gruytere, 1994:189–214.

11. Bhopal R, Hayes L, White M, Unwin N, Harland J, Ayis S, Alberti G. Ethnic and socio-economic inequalities in coronary heart disease, diabetes and risk factors in Europeans and South Asians. J Public Health Med 2002;24:95-105.

12. Bos V, Kunst AE, Garssen J, Mackenbach JP. Socioeconomic inequalities in mortality within ethnic groups in the Netherlands, 1995-2000. J Epidemiol Community Health 2005;59:329-35.

13. Williams R, Wright W, Hunt K. Social class and health: the puzzling counter-example of British South Asians. Soc Sci Med 1998;47:1277-88.

14. Mackenbach JP, Cavelaars EJM, Kunst AE, Groenhof F. Socioeconomic inequalities in cardiovascular disease mortality. An international study. Eur Heart J 2000;21:1141–51.

15. Kanjilal S, Gregg EW, Cheng YJ, Zhang P, Nelson DE, Mensah G, Beckles GL. 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 –2355.

16. Ramsay SE, Whincup PH, Hardoon SL, Lennon LT, Morris RW, Wannamethee SG. Social class differences in secular trends in established coronary risk factors over 20 years: a cohort study of British men from 1978–80 to 1998–2000. PLoS One 2011;6:e19742.

136

Socioeconomic inequalities in stroke incidence by country of birth

17. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53.

18. Harteloh P, de Bruin K, Kardaun J. The reliability of cause-of-death coding in The Netherlands. Eur J Epidemiol 2010;25:531-8

19. Benesch C, Witter DM, Jr., Wilder AL, Duncan PW, Samsa GP, Matchar DB. Inaccuracy of the International Classification of Diseases (ICD-9-CM) in identifying the diagnosis of ischemic cerebrovascular disease. Neurology 1997;49:660-4.

20. Ellekjaer H, Holmen J, Kruger O, Terent A. Identification of incident stroke in Norway: hospital discharge data compared with a population-based stroke register. Stroke 1999;30:56-60.

21. Nieuwkamp DJ, Vaartjes I, Algra A, Rinkel GJ, Bots ML. Risk of cardiovascular events and death in the life after aneurysmal subarachnoid haemorrhage: a nationwide study. Int J Stroke 2012 [Epub ahead of print].

22. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14:255-69.

23. Agyemang C, van Oeffelen AA, Bots ML, Stronks K, Vaartjes I. Socioeconomic inequalities in acute myocardial infarction incidence in migrant groups: has the epidemic arrived? Analysis of nation-wide data. Heart 2014;100:239-46.

24. van Oeffelen AA, Agyemang C, Bots ML, Stronks K, Koopman C, van Rossem L., Vaartjes I. The relation between socioeconomic status and short-term mortality after acute myocardial infarction persists in the elderly: results from a nationwide study. Eur J Epidemiol 2012;27:605-13.

25. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57:1288-94.

26. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56:221-9.

27. Fischbacher CM, Cezard G, Bhopal RS, Pearce J, Bansal N; on behalf of the Scottish Health and Ethnicity Linkage Study. Measures of socioeconomic position are not consistently associated with ethnic differences in cardiovascular disease in Scotland: methods from the Scottish Health and Ethnicity Linkage Study(SHELS). Int J Epidemiol 2013 [Epub ahead of print].

28. Nierkens V, de Vries H, Stronks K. Smoking in immigrants: do socioeconomic gradients follow the pattern expected from the tobacco epidemic? Tob Control 2006;15:385-91.

29. Agyemang C, van Valkengoed I, Hosper K, Nicolaou M, van den Born BJ, Stronks K. Educational inequalities in metabolic syndrome vary by ethnic group: evidence from the SUNSET study. Int J Cardiol 2010;141:266-74.

30. Agyemang C, Bindraban N, Mairuhu G, van Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23:1971-7.

31. Agyemang C, Kunst AE, Bhopal R, Anujuo K, Zaninotto P, Nazroo J, et al. Diabetes prevalence in populations of South Asian Indian and African origins: a comparison of England and the Netherlands. Epidemiology 2011;22:563-7.

137

Chapter 3.2

32. Agyemang C, Kunst A, Bhopal R, Zaninotto P, Nazroo J, Nicolaou M, et al. Dutch versus English advantage in the epidemic of central and generalised obesity is not shared by ethnic minority groups: comparative secondary analysis of cross-sectional data. Int J Obes(Lond) 2011;35:1334-46.

33. Ujcic-Voortman JK, Baan CA, Seidell JC, Verhoeff AP. Obesity and cardiovascular disease risk among Turkish and Moroccan migrant groups in Europe: a systematic review. Obes Rev 2012;13:2-16.

34. Agyemang C, Ujcic-Voortman J, Uitenbroek D, Foets M, Droomers M. Prevalence and management of hypertension among Turkish, Moroccan and native Dutch ethnic groups in Amsterdam, the Netherlands: The Amsterdam Health Monitor Survey. J Hypertens 2006;24:2169-76. 35. Misirli G, Benetou V, Lagiou P, Bamia C, Trichopoulos D, Trichopoulou A. Relation of the traditional Mediterranean diet to cerebrovascular disease in a Mediterranean population. Am J Epidemiol 2012;176:1185-92. 36. Smith GD. Learning to live with complexity: ethnicity, socioeconomic position, and health in Britain and the United States. Am J Public Health 2000;90:1694-8. 37. Kaufman JS, Cooper RS. Seeking causal explanations in social epidemiology. AmJ Epidemiol 1999;150:113-120. 38. Geyer S, Peter R. Income, occupational position, qualification and health inequalities--competing risks? (comparing indicators of social status). J Epidemiol Community Health 2000;54:299-305.

39. Glymour MM, Avendano M, Haas S, Berkman LF. Lifecourse social conditions and racial disparities in incidence of first stroke. Ann Epidemiol 2008;18:904 –912.

40. Davey Smith G, Shipley MJ, Rose G. The magnitude and causes of socioeconomic differentials in mortality: further evidence from the Whitehall Study. J Epidemiol Community Health 1990;44:265– 270.

138

CHAPTER 4

INEQUALITIES IN PROGNOSIS AFTER CARDIOVASCULAR DISEASE

CHAPTER 4.1

Ethnic inequalities in prognosis after acute myocardial infarction and congestive heart failure

van Oeffelen AAM, Agyemang C, Stronks K, Bots ML, Vaartjes I. Prognosis after a first hospitalisation for acute myocardial infarction and congestive heart failure by country of birth.

Heart ‘under revision’. Chapter 4.1

ABSTRACT Objective To investigate differences in 28-day and 5-year mortality and 5-year readmission after a first hospitalisation for acute myocardial infarction (AMI) and congestive heart failure (CHF) between first generation ethnic minority groups (henceforth, migrants) and the ethnic Dutch population.

Methods Nationwide prospective cohorts of first hospitalised AMI (N=213,630) and CHF patients (N=189,069) between 1998 and 2010 were built. Differences in 28-day and 5-year mortality, and in 5-year AMI/CHF readmission between migrants (Surinamese, Moroccan, Turkish, Antillean, Indonesian, Chinese, and South Asian) and the ethnic Dutch population were investigated using Cox proportional hazard regression models.

Results After the first AMI hospitalisation, mortality and AMI/CHF readmission were higher in the majority of migrant groups compared with ethnic Dutch. For example, Hazard Ratios (adjusted for age, sex, marital status, degree of urbanisation, and year of event) with 95% confidence intervals among Surinamese (mainly of African or South-Asian origin) were 1.16 (1.02-1.32) for 28-day mortality, 1.44 (1.30-1.60) for 5-year mortality, 1.33 (1.08-1.63) for AMI readmission, and 2.09 (1.82-2.40) for CHF readmission. After a first CHF hospitalisation, mortality rates among migrants were more diverse, with a lower 28-day mortality among Moroccan and Turkish migrants and a higher 5-year mortality among Surinamese, Chinese, and South Asians. Readmission after CHF was often higher among migrant groups.

Conclusion Prognosis after a first AMI hospitalisation was worse among most migrant groups compared with the ethnic Dutch population. Ethnic inequalities in prognosis after a first CHF hospitalisation were more diverse. Efforts should be made to disentangle the underlying factors of found results.

142

Ethnic inequalities in prognosis after acute myocardial infarction and congestive heart failure

INTRODUCTION

Although there has been a great decline in coronary heart disease (CHD) morbidity and mortality during the last decade, it is still a huge contributor to death and disability. CHD by itself is the single most common cause of death in Europe, accounting for 1.8 million deaths each year, which is about 20% of all men and women. Because CHD damages the heart muscle, it is a common cause of congestive heart failure (CHF).1 Within eight years after an acute myocardial infarction (AMI), more than one-third of patients develops CHF.2 CHF is a chronic disease state with rare recovery and therefore places a huge burden on both patients and health care costs. After a first hospitalisation for CHF, 5-year mortality can be as high as 75%, and readmission rate is among the highest of all diseases.1;3 Timely treatment of AMI cannot only reduce mortality and AMI readmission, but also CHF development. After a patient is diagnosed with AMI or CHF, secondary preventive measures to control concomitant risk factors are important to reduce mortality and readmission risk.1 Previously, ethnic inequalities in mortality after an AMI hospitalisation have been reported, though results were inconsistent. Most recent and largest UK studies showed a lower mortality among South Asians compared with the majority population.4;5 In USA studies mortality was mainly higher in African Americans compared with White Americans,6;7 and in Canadian studies South Asians had a similar short-term mortality but a lower long-term mortality compared with the majority population.8-10 Evidence regarding ethnic inequalities in mortality after a first CHF hospitalisation is more limited, and showed lower or similar mortality rates among South-Asian and African groups compared with the majority population, but a higher mortality among the Chinese.10- 15 In the Netherlands, our research group previously reported similar short- and long-term mortality rates after AMI and CHF among ethnic minorities compared with the ethnic Dutch population.16 Unfortunately, this study was restricted by low numbers and a suboptimal disaggregation of ethnic minority groups. Data concerning some major ethnic minority groups in the Netherlands, such as Moroccans, were missing. Furthermore, ethnic inequalities in readmission rate were not reported. Also international literature concerning ethnic inequalities in readmission after CHD and CHF is scarce, and mainly shows higher readmission rates among African and South-Asian minorities.8;9;17 It is important to expand existing evidence in order to identify high risk groups, so that targeted treatment and management strategies based on the patients’ background can be implemented. This may benefit patients as well as health care providers by decreasing burden of disease and health care costs. Therefore, the aim of our study was to investigate ethnic inequalities in 28-day and 5- year mortality and 5-year AMI/CHF readmission after a first AMI or CHF hospitalisation.

143

Chapter 4.1

METHODS

Cohort enrolment Data were obtained from several Dutch nationwide registers between 1998 and 2010; e.g. Hospital Discharge Register (HDR), Population Register (PR), Cause of Death Register (CDR), and Regional Income Survey (RIS). The registers were used to obtain information regarding AMI/CHF hospitalisations, comorbidities, demographic factors, date and cause of death, and income. By linking previous registers, all persons with a first hospitalisation with as primary diagnosis AMI (ICD-9 code 410) between 1998 and 2010, and who were PR unique at the time of the event (unique with respect to the combination of the variables date of birth, sex, and four digits of the postal code) were selected. Person with a previous AMI hospitalisation between 1995 and 1997 were excluded. The same procedure was performed for the primary diagnosis CHF (ICD-9 codes 402 and 428). Besides the ethnic Dutch population, first generation ethnic minorities (henceforth, migrants) born in Suriname, Morocco, Turkey, Netherlands Antilles, Indonesia, China, and South Asia (India, Pakistan, Sri Lanka) were included. Second generation ethnic minorities were excluded because of the relatively low number of elderly second generation ethnic minorities in the Netherlands, and the possible difference in CVD between first and second generation ethnic minorities. This resulted in two cohorts with 213,630 first hospitalisations for AMI and 189,069 first hospitalisations for CHF between 1998 and 2010. From their first hospitalisation, patients were followed for 28-day and 5- year all-cause mortality, and 5-year AMI/CHF readmission. Persons were censored in case of death, emigration, PR non-uniqueness, or the end of the study period at 31 December 2010, whichever came first. Approval for the use of the anonymised patient data was covered by a general agreement between Statistics Netherlands and Dutch Hospital Data. Additionally, the Dutch association of hospitals (NVZ) approved the use of hospital registration data for this study. No separate ethical approval was necessary for the use of these data. All our analyses were performed within strict privacy rules, i.e. only researchers who received a signed permit were allowed to do analyses within a secured environment at our Institute. Prior to publication, Statistics Netherlands made sure that none of the analyses results showed potential reducibility to the individual level.

Determinants Ethnic background Migrant groups were constructed based on the country of birth of the resident and his/her parents. A person was considered a migrant if he/she was born abroad and at least one of the parents was born abroad. Persons with both parents born in the Netherlands were indicated as the ethnic Dutch population.

144

Ethnic inequalities in prognosis after acute myocardial infarction and congestive heart failure

Neighbourhood socioeconomic status Neighbourhood socioeconomic status (SES) was based on income data registered in the RIS.18 All persons belonging to the households of the sample population (about one third of the Dutch population) were included in the RIS. Within each neighbourhood, the mean standardised disposable income on household level of the residents with income data available was calculated yearly, and subsequently assigned to all residents living in that neighbourhood. The neighbourhood where the patient lived at the time of the event was selected. For those with the event between 1998 and 2005, the neighbourhood income in the year of the event was assigned. For those with the event between 2006 and 2010, the neighbourhood income in 2006 was assigned, since RIS data from 2007 onwards were not available yet. Subsequently, neighbourhood income was divided yearly into SES tertiles, with the first tertile representing the lowest income group.

Comorbidity Presence and extent of comorbidity were determined with the Charlson index score, based on discharge diagnosis in the HDR from 1995 to the date of the hospitalisation. The Charlson index score ranges from zero to six (cut-off value), with zero representing no comorbidity. It proved to be a reliable and valid method to measure comorbidity in clinical research.19

Data analysis Baseline characteristics were analysed using cross-tabs and frequency tables. With Cox proportional hazard regression analyses, adjusted for age, sex, marital status, degree of urbanisation (categorical: very urban to very rural), and year of the event (linear), differences in prognosis after AMI and CHF between migrant groups and ethnic Dutch (reference) were investigated. Additionally, analyses were adjusted for comorbidity and neighbourhood SES to explore to what extent these variables explained found relations. Log-minus-log plots showed no violation of the proportional hazards assumption. In the AMI cohort, analyses were additionally adjusted for AMI type (ST-elevation myocardial infarction (STEMI) and non ST-elevation myocardial infarction (non-STEMI)) as proxy for disease severity. Analyses were executed in SPSS software, version 20.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation Netherlands.

RESULTS

Characteristics of AMI and CHF patients are presented in Tables 1a and 1b. Except for Indonesians, migrants were younger and lived more often in low income urban areas.

145

Chapter 4.1

57)

-

21,157)

e - - 3.9 3.9 8.9 280 89.6 51.8 21.1 19.3 68.6 10.0 56.4 27.5 16.1 10.7

18,763 51 (46 South Asian

(16,963

75)

-

20,761)

e e e e

- - - - - 82 70.7 56.1 25.6 13.4 52.4 12.2 58.5 25.6 15.9 36.6 18,620 Chinese

68 (57 (17,047

77)

- 21,904)

- 4.6 9.4 7.5

66.4 22.5 33.9 26.8 12.2 52.8 16.8 31.4 39.4 29.2 40.1 4,529 20,175 69 (60 Indonesian (17,845

500, very rural=<251

65)

-

-

21,904)

e e - - - 7.9 7.7 482 71.0 37.8 30.9 22.0 39.0 16.8 47.3 32.4 20.3 23.0

19,618 Antillean 57 (49

(17,226

1000, rural=251

-

64)

- 20,526)

- 8.8 8.3 1.0 8.7 8.8

82.4 38.9 33.8 18.0 78.5 13.9 61.7 29.5 20.8

2,137 18,226 Turkish

56 (47 (16,556

2000, urban/rural=501 62)

- -

21,717)

e e - - - 8.3 7.4 5.7

336 85.1 36.6 37.2 17.9 77.7 55.1 28.3 16.7 14.0 19,522 Moroccan 51 (43 (17,634

umber of cases was less than ten than less was cases of umber

66)

- 20,760)

- 1.1 3.5 9.1 8.2 70.5 62.6 13.5 44.3 18.6 58.1 25.5 19.3 26.3 15.6 2,948

18,139 56 (48 Surinamese

(16,134 ery urban=>2000, urban=1001

). V spital admission for acute myocardialinfarction 2

hospitalisation for a diagnosis included in the Charlson comorbidity Index from 1995 until first AMI event 78)

-

21,904)

- 9.0 7.3 66.0 14.3 24.1 24.8 21.0 15.8 58.1 16.6 28.8 45.3 26.0 37.7 20,182 202,836

69 (58 rhood level withtertile representing 1 thelowest income group Ethnic Dutch with a first ho

u (17,904

b Rural d Urban ) dex> least0: at one

a c Tertile 1 Tertile 2 Tertile 3 Tertile

Very rural

) Very urban a Urban/rural

neighbourhood

Characteristics of patients

readmission % ot given in line with the Dutch data protection guideline as the n Number of patients Male % Median age (IQR Degree of urbanisation % Married/living together % Charlson Index > % 0 Median (IQR euros in income Neighbourhood % SES Mortality % readmissionAMI % CHF Socioeconomic status on neighbo Interquartile range Population density (number of residents per km Charlson comorbitiy in

N

1a Table a b c d e

146

Ethnic inequalities in prognosis after acute myocardial infarction and congestive heart failure

72)

-

21,142)

e e - - - 82 70.7 39.0 25.6 19.5 61.0 52.4 51.2 32.9 15.9 45.1 20.7

19,684 58 (49 South Asian

(17,130

83)

-

21,321)

e e - - - 89 60.7 49.4 24.7 14.6 55.1 57.3 44.9 36.0 19.1 76.4 16.9 19,042 Chinese

76 (69 (16,551

84)

- 22,128)

-

4.1 49.9 25.2 33.4 26.4 11.0 41.6 49.7 32.2 38.9 28.9 69.9 16.9 4,113 20,366 78 (71 Indonesian (17,905

500, very rural=<251

71)

- -

21,554)

e e - - - 444 48.9 45.9 31.1 17.3 32.0 43.5 53.8 29.3 16.9 44.1 19.6

18,909 Antillean 63 (54

(16,986

1000, rural=251

-

70)

-

20,574)

- 7.5 0.8 8.4

61.8 40.2 34.3 17.1 73.2 54.7 62.5 29.1 44.0 23.8

1,460 18,390 Turkish en

64 (59 (16,666

Charlson Comorbidity Indexfrom 1995 until first CHF event

2000, urban/rural=501

70)

- -

21,742)

e e - - - 236 60.6 39.0 33.1 20.8 74.2 37.3 58.9 28.0 13.1 26.7 21.6 19,207 Moroccan 61 (48 (17,816

77)

- 20,583)

- 2.5 1.0 46.4 68.1 17.7 10.7 32.2 45.1 62.0 24.0 14.0 58.0 22.5

2,192 18,066 68 (58 Surinamese (16,160

ation for a diagnosis included in the ery urban=>2000, urban=1001

). V 2

hospital admission for congestive heart failure

85)

- 21,888)

- 49.2 17.2 24.6 23.5 20.1 14.7 41.7 49.9 30.3 44.2 25.5 69.9 17.6 20,320 180,453 79 (72 rhood level withtertile representing 1 thelowest income group Ethnic Dutch u (18,023

b

d Rural Urban ) dex> least0: at one hospitalis

a c

Tertile 1 Tertile 2 Tertile Tertile 3

Very rural

) Very urban a Urban/rural

Characteristics of patients with a first

ot given in line with the Dutch data protection guideline as the number of cases was less than t Number of patients Male % Median age (IQR Degree of urbanisation % Married/living together % Charlson Index > % 0 neighhourhoodMedian (IQR euros in income Neighbourhood (%)SES Mortality (%) CHF readmission (%) Population density(number of residents perkm Socioeconomic status on neighbo Interquartile range Charlson comorbitiy in

N

1bTable a b c d e

147

Chapter 4.1

Prognosis after the first AMI hospitalisation by country of birth After a first AMI admission, prognosis was often worse in migrants than in ethnic Dutch (Table 2). 28-day mortality was higher among Surinamese, Antillean, Indonesian, Chinese, and South Asian migrants, whereas Moroccan and Turkish migrants had similar rates as ethnic Dutch. 5-year mortality was higher among Surinamese, Moroccan, Turkish, Antillean, and Chinese migrants. Indonesians had a similar 5-year mortality as ethnic Dutch, whereas it was lower among South Asians. AMI readmission was consistently higher among migrant groups, ranging from a Hazard Ratio (HR) of 1.23 in Indonesians to 1.77 in Chinese. Also CHF readmission was markedly higher among Surinamese, Moroccan, Turkish, Antilleans, and Chinese, ranging from a HR of 1.78 in Chinese to 2.60 in Turkish migrants. Additional adjustment for comorbidity, neighbourhood SES and AMI type (STEMI/non-STEMI) did not influence results.

Prognosis after the first CHF hospitalisation by country of birth Differences in prognosis after a first CHF hospitalisation with ethnic Dutch varied between migrant groups (Table 3). Antillean and Indonesian migrants had similar prognosis (both mortality and readmission) as ethnic Dutch, whereas Chinese and South Asian groups had a worse prognosis (although not statistically significant). While Surinamese migrants had a similar 28-day mortality as ethnic Dutch, their 5-year mortality and readmission rates were significantly higher. Moroccan and Turkish migrants showed a lower 28-day mortality rate, but a higher readmission rate compared with ethnic Dutch. There was no difference in 5-year mortality between ethnic Dutch and migrants from Turkey and Morocco. Additional adjustment for comorbidity and neighbourhood SES did not influence results.

148

Ethnic inequalities in prognosis after acute myocardial infarction and congestive heart failure

Table 2 Difference in risk on mortality and readmission after the first AMI hospitalisation between migrant groups and the ethnic Dutch population (HR (95% CI)) during 1998-2010a Number of events Person-years at risk HR (95% CI) 28-day mortality Ethnic Dutch 20,471 13,921 1.00 Surinamese 180 211 1.16 (1.02-1.32)* Moroccan 12 25 0.87 (0.51-1.47) Turkish 86 156 0.98 (0.82-1.18) Antilleans 27 34 1.25 (0.91-1.72) Indonesian 521 305 1.20 (1.11-1.29)* Chinese 11 6 1.40 (0.83-2.37) South Asian 12 20 1.34 (0.82-2.20) 5-year mortality Ethnic Dutch 35,337 625,849 1.00 Surinamese 362 9,578 1.44 (1.30-1.60)* Moroccan 19 989 1.31 (0.84-2.06) Turkish 223 7,215 1.59 (1.39-1.81)* Antilleans 50 1,528 1.45 (1.10-1.92)* Indonesian 802 13,481 1.06 (0.98-1.13) Chinese 14 236 1.29 (0.77-2.18) South Asian <10b 947 0.59 (0.28-1.23) 5-year AMI readmission Ethnic Dutch 6,029 568,299 1.00 Surinamese 99 8,515 1.33 (1.08-1.63)* Moroccan 10 919 1.50 (0.80-2.79) Turkish 70 6,465 1.33 (1.04-1.68)* Antilleans 17 1,374 1.48 (0.92-2.39) Indonesian 164 12,153 1.23 (1.05-1.43)* Chinese <10b 212 1.77 (0.66-4.71) South Asian <10b 839 1.30 (0.65-2.60) 5-year CHF readmission Ethnic Dutch 12,064 597,600 1.00 Surinamese 212 8,804 2.09 (1.82-2.40)* Moroccan 17 917 2.44 (1.52-3.94)* Turkish 155 6,653 2.60 (2.21-3.05)* Antilleans 34 1,419 2.37 (1.69-3.33)* Indonesian 272 12,807 1.02 (0.90-1.15) Chinese <10b 218 1.78 (0.85-3.74) South Asian <10b 903 1.01 (0.45-2.24) a Adjusted for age, sex, marital status, degree of urbanisation, and year of event b Not given in line with the Dutch data protection guideline as the number of cases was less than ten *p<0.05

149

Chapter 4.1

Table 3 Difference in Risk on mortality and readmission after the first CHF hospitalisation between migrant groups and the ethnic Dutch population (HR (95% CI)) during 1998-2010a Number of events Person-years at risk HR (95% CI) 28-day mortality Ethnic Dutch 31,387 11,985 1.00 Surinamese 232 155 0.94 (0.83-1.06) Moroccan 10 17 0.57 (0.33-0.98)* Turkish 94 106 0.67 (0.55-0.80)* Antilleans 39 32 0.89 (0.66-1.21) Indonesian 718 276 1.02 (0.95-1.10) Chinese 17 6 1.23 (0.78-1.92) South Asian <10b 6 1.22 (0.66-2.26) 5-year mortality Ethnic Dutch 79,988 336,021 1.00 Surinamese 868 4,939 1.18 (1.10-1.26)* Moroccan 45 527 0.89 (0.67-1.20) Turkish 456 3,528 1.03 (0.93-1.13) Antilleans 129 1,106 0.99 (0.84-1.18) Indonesian 1,825 7,552 1.04 (0.99-1.08) Chinese 46 158 1.33 (1.00-1.77) South Asian 25 186 1.28 (0.86-1.89) 5-year CHF readmission Ethnic Dutch 24,640 295,111 1.00 Surinamese 382 4,156 1.24 (1.12-1.38)* Moroccan 39 430 1.32 (0.96-1.81) Turkish 263 2,929 1.20 (1.07-1.36)* Antilleans 68 969 1.04 (0.82-1.32) Indonesian 558 6,575 0.99 (0.91-1.08) Chinese 14 127 1.21 (0.72-2.05) South Asian 12 139 1.23 (0.70-2.16) a Adjusted for age, sex, marital status, degree of urbanisation, and year of event b Not given in line with the Dutch data protection guideline as the number of cases was less than ten *p<0.05

DISCUSSION

After a first AMI admission, mortality and AMI/CHF readmission were higher in the majority of migrant groups compared with the ethnic Dutch population, especially among Surinamese, Antilleans, and Chinese. After a first CHF admission, inequalities in mortality were more diverse, with a lower 28-day mortality among Moroccans and Turkish and a higher 5-year mortality among Surinamese, Chinese, and South Asians. Readmission after CHF was often higher among migrant groups. Results could not be explained by the available measures of neighbourhood SES and comorbidity.

150

Ethnic inequalities in prognosis after acute myocardial infarction and congestive heart failure

Discussion of main findings In 2010 our group studied ethnic inequalities in mortality after AMI and CHF in some major ethnic minority groups in the Netherlands.16 Contrary to our present results, no clear differences were found. This discrepancy may be driven by an earlier and smaller time frame, a lower precision due to smaller numbers, and the inclusion of second generation ethnic minority groups in the earlier work. Internationally, studies have been predominantly executed in South-Asian and African ethnic minorities. Readmission after AMI and CHF was mostly similar or higher among South Asians and Africans, which is in agreement with our results in the South-Asian and Surinamese (mostly from South-Asian or African origin) subgroups.8;9;17 Most recent and largest studies reported a lower mortality after AMI among South Asians, which is only in agreement with the non-significantly lower 5-year mortality among South Asians in our study.5 Varying results concerning ethnic inequalities in mortality between countries may be related to differences in baseline mortality rate of the majority population, and variation in risk profiles of migrants from the same country of origin living in different countries.20 Our study showed a worse prognosis on all outcomes among Chinese migrants compared with ethnic Dutch. Previously this was also shown in Canada and the UK, but for short- term mortality only.4;9 This finding needs further investigation in future research.

Prognosis after the first AMI hospitalisation by country of birth After a first AMI hospitalisation 28-day mortality was higher in Surinamese, Antillean, Indonesian, Chinese, and South-Asian migrants than in ethnic Dutch. Firstly, this might be explained by a higher degree of disease severity, comorbidity and concomitant cardiovascular risk factors. Although adjustment for disease type (STEMI/non-STEMI) and comorbidity did not influence results, we were unable to adjust for extent of disease (number of diseased vessels) and concomitant cardiovascular risk factors. Some migrant groups, for example South Asians, might present more often with the more severe triple vessel and left main stem disease.21 Ethnic minority populations may also suffer from more cardiovascular risk factors, particularly diabetes.22 Secondly, a longer time delay between symptom onset and treatment among migrants may decrease survival.23 More atypical symptoms could partially drive this time delay. Thirdly, revascularisation procedures may be less often performed among migrant groups, decreasing survival, although this was not observed in the most recent UK study anymore.4 5-year mortality and AMI readmission were mostly higher among migrants than among ethnic Dutch. This is probably highly dependent on differences in receiving and adhering to secondary preventive measures. Previous literature showed a lower use of cardiac rehabilitation among ethnic minorities, and difficulties in adhering to lifestyle changes and medication therapy.24-26 Among Surinamese in the Netherlands, for example, poor blood-pressure control has been reported.27 Low compliance among migrant groups may be caused by several factors, such as

151

Chapter 4.1 linguistic and cultural barriers, low health literacy, and differences in health beliefs. Difficulties in reaching mutual understanding between the migrant patient and the physician may also contribute to non-compliance to lifestyle modifications and drug use.28 CHF readmission was higher among Surinamese, Moroccan, Turkish, Antillean, and Chinese migrant groups compared with the ethnic Dutch population. CHF is often treated in outpatient care and only severe cases need a hospitalisation. A higher AMI severity among migrants, for example due to lack of timely reperfusion, could have driven the higher CHF readmission.23 Moreover, elderly migrants are less willing to be institutionalised. Institutionalised persons with a CHF exacerbation are more likely treated within the facility they stay, whereas those who live independently need a hospitalisation in an earlier phase.

Prognosis after the first CHF hospitalisation by country of birth After a first CHF hospitalisation 28-day mortality was mostly similar or non-significantly higher in migrant groups than in the ethnic Dutch population, except for the lower rate among Moroccan and Turkish groups. This is probably not the result of a younger age distribution, since results remained similar in the young (results not shown). More research is necessary to elucidate underlying mechanisms of these lower rates. 5-year mortality after CHF was mostly similar between migrants and the ethnic Dutch, except for the higher mortality among Surinamese, Chinese, and South Asians. Surinamese and South-Asian migrants have a high prevalence of both diabetes and hypertension, which are associated with worse outcomes in the presence of CHF.1 5-year mortality after CHF is also highly dependent on differences in receiving and adhering to secondary preventive measures, similar as for 5-year mortality after AMI as described above. The higher CHF readmission after a first CHF hospitalisation among many migrant groups may reflect difficulties in controlling heart failure symptoms by means of medication/lifestyle change. Also adhering to a diet low in sodium is difficult in most migrant groups, as food plays an essential part in social gatherings in many cultures. Again, the lower number of institutionalised elderly migrants may drive the higher readmission rates.

Considerations Strengths of our study are the large sample size which enabled us to include a wide variety of migrant groups (to our knowledge prognosis in Moroccans is never studied before) and restrict to first generation ethnic minorities only, the nationwide nature of the data, and the inclusion of the outcome measures mortality and readmission both. Inevitably our study has some limitations. First, the medical history data available ranged from three years to 16 years, which may have led to misclassifying recurrent events as first events in the earlier years. However, the risk on a recurrent

152

Ethnic inequalities in prognosis after acute myocardial infarction and congestive heart failure event is highest in the first year after the initial event, and drops dramatically after that.29 Furthermore, most migrants came to the Netherlands between 1940 and 1980, and the more recent migrants are mostly too young to suffer from AMI or CHF. Therefore, the distribution of AMI and CHF events between migrants and ethnic Dutch over time is fairly similar, and the risk on differential misclassification of recurrent events as first events between groups unlikely. Second, the analyses were only performed in the hospitalised subjects. Inclusion of pre-hospital deaths would have given a more complete picture, but only for 28-day mortality. Third, while the validity of AMI registration in the HDR is good, the validity of the ICD-9 codes for CHF is somewhat less adequate with a sensitivity of 43% and a positive predictive value of 80%.30 Although this means we missed more than half of the CHF patients, 80% of those in the HDR were correctly registered. This is a reasonable percentage and did probably not markedly influence results, also because differential misclassification between migrants and ethnic Dutch is implausible. Fourth, the analyses were adjusted for neighbourhood income as proxy for SES, which may have led to some non-differential misclassification. Fifth, the classification of the various groups was based on the country of birth, which is probably an unreliable proxy measure of ethnicity for the Surinamese, because of the mixed background of this group. Future studies should make efforts in distinguishing the South-Asian and the African Surinamese. Sixth, non-unique persons in the PR (with respect to the variables date of birth, gender, and four digits of the postal code) were excluded in our cohorts, who were more prevalent among migrant groups. Since uniqueness was not related to the outcomes under study, exclusion of PR non-unique persons will only have led to precision loss, not to biased results. Finally, throughout this paper we have not only highlighted the statistically significant results, but also the non-statistically significant once, that are in our opinion of clinical relevance. We acknowledge the possibility that these results are chance findings.

Conclusion Our study showed that prognosis after a first AMI hospitalisation was worse among most migrant groups compared with the ethnic Dutch population. Inequalities in prognosis after a first CHF hospitalisation were more diverse, and most adverse outcomes were seen among Surinamese, Chinese, and South Asians, and for CHF readmission. Underlying mechanisms of our findings are probably multifactorial. More extensive cohort studies containing a broad range of clinical, social, lifestyle, and health care variables are needed to disentangle the causes of found results, so that appropriate public health and clinical interventions can be developed to reduce found inequalities.

153

Chapter 4.1

REFERENCES 1. Bui AL, Horwich TB, Fonarow GC. Epidemiology and risk profile of heart failure. Nat Rev Cardiol 2011;8(1):30-41. 2. Hellermann JP, Goraya TY, Jacobsen SJ, Weston SA, Reeder GS, Gersh BJ, et al. Incidence of heart failure after myocardial infarction: is it changing over time? Am J Epidemiol 2003;157(12):1101-7. 3. Westert GP, Lagoe RJ, Keskimaki I, Leyland A, Murphy M. An international study of hospital readmissions and related utilization in Europe and the USA. Health Policy 2002;61(3):269-78. 4. Bansal N, Fischbacher CM, Bhopal RS, Brown H, Steiner MF, Capewell S. Myocardial infarction incidence and survival by ethnic group: Scottish Health and Ethnicity Linkage retrospective cohort study. BMJ Open 2013;3(9):e003415. 5. Justin M, Zaman S, Philipson P, Chen R, Farag A, Shipley M, et al. South Asians and coronary disease: is there discordance between effects on incidence and prognosis. Heart 2013;99(729):736. 6. Echols MR, Mahaffey KW, Banerjee A, Pieper KS, Stebbins A, Lansky A, et al. Racial differences among high-risk patients presenting with non-ST-segment elevation acute coronary syndromes (results from the SYNERGY trial). Am J Cardiol 2007;99(3):315-21. 7. Thomas KL, Honeycutt E, Shaw LK, Peterson ED. Racial differences in long-term survival among patients with coronary artery disease. Am Heart J 2010;160(4):744-51. 8. Albarak J, Nijjar AP, Aymong E, Wang H, Quan H, Khan NA. Outcomes in young South Asian Canadians after acute myocardial infarction. Can J Cardiol 2012;28(2):178-83. 9. Khan NA, Grubisic M, Hemmelgarn B, Humphries K, King KM, Quan H. Outcomes after acute myocardial infarction in South Asian, Chinese, and white patients. Circulation 2010;122(16):1570-7. 10. Nijjar AP, Wang H, Dasgupta K, Rabi DM, Quan H, Khan NA. Outcomes in a diabetic population of South Asians and whites following hospitalization for acute myocardial infarction: a retrospective cohort study. Cardiovasc Diabetol 2010;9:4. 11. Blackledge HM, Newton J, Squire IB. Prognosis for South Asian and white patients newly admitted to hospital with heart failure in the United Kingdom: historical cohort study. BMJ 2003;327(7414):526- 31. 12. Gordon HS, Nowlin PR, Maynard D, Berbaum ML, Deswal A. Mortality after hospitalization for heart failure in blacks compared to whites. Am J Cardiol 2010;105(5):694-700. 13. Kaul P, McAlister FA, Ezekowitz JA, Grover VK, Quan H. Ethnic differences in 1-year mortality among patients hospitalised with heart failure. Heart 2011;97(13):1048-53. 14. Newton JD, Blackledge HM, Squire IB. Ethnicity and variation in prognosis for patients newly hospitalised for heart failure: a matched historical cohort study. Heart 2005;91(12):1545-50. 15. Sosin MD, Bhatia GS, Zarifis J, Davis RC, Lip GY. An 8-year follow-up study of acute admissions with heart failure in a multiethnic population. Eur J Heart Fail 2004;6(5):669-72. 16. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 17. Joynt KE, Orav EJ, Jha AK. Thirty-day readmission rates for Medicare beneficiaries by race and site of care. JAMA 2011;305(7):675-81. 18. Ament P, Kessels W. Regionaal Inkomensonderzoek: uitgebreide onderzoeksbeschrijving. Voorburg: Centraal Bureau voor de Statistiek (CBS) 2008. 19. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56(3):221-9.

154

Ethnic inequalities in prognosis after acute myocardial infarction and congestive heart failure

20. Agyemang C, Kunst A, Bhopal R, Zaninotto P, Unwin N, Nazroo J, et al. A cross-national comparative study of blood pressure and hypertension between English and Dutch South-Asian- and African-origin populations: the role of national context. Am J Hypertens 2010;23(6):639-48. 21. Trevelyan J, Needham EW, Halim M, Singh H, Been M, Shiu MF, et al. Evaluation of patient characteristics and utilisation of invasive cardiac procedures in a UK ethnic population with unstable angina pectoris. Int J Cardiol 2001;77(2-3):275-80. 22. El Fakiri F., Bruijnzeels MA, Foets MM, Hoes AW. Different distribution of cardiovascular risk factors according to ethnicity: a study in a high risk population. J Immigr Minor Health 2008;10(6):559-65. 23. Kendall H, Marley A, Patel JV, Khan JM, Blann AD, Lip GY, et al. Hospital delay in South Asian patients with acute ST-elevation myocardial infarction in the UK. Eur J Prev Cardiol 2013;20(5):737-42. 24. Cortes O, Arthur HM. Determinants of referral to cardiac rehabilitation programs in patients with coronary artery disease: a systematic review. Am Heart J 2006;151(2):249-56. 25. Oosterberg EH, Deville W, Brewster LM, Agyemang C, van den MM. [Chronic disease in ethnic minorities: tools for patient-centred care in diabetes, hypertension and COPD]. Ned Tijdschr Geneeskd 2013;157(16):A5669. 26. Hempler NF, Diderichsen F, Larsen FB, Ladelund S, Jorgensen T. Do immigrants from Turkey, Pakistan and Yugoslavia receive adequate medical treatment with beta-blockers and statins after acute myocardial infarction compared with Danish-born residents? A register-based follow-up study. Eur J Clin Pharmacol 2010;66(7):735-42. 27. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7. 28. van Wieringen JC, Harmsen JA, Bruijnzeels MA. Intercultural communication in general practice. Eur J Public Health 2002;12(1):63-8. 29. Osler M, Rostgaard K, Sorensen TI, Madsen M. The effect of recurrent events on register-based estimates of level and trends in incidence of acute myocardial infarction. J Clin Epidemiol 1999;52(7):595-600. 30. Merry AH, Boer JM, Schouten LJ, Feskens EJ, Verschuren WM, Gorgels AP, et al. Validity of coronary heart diseases and heart failure based on hospital discharge and mortality data in the Netherlands using the cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009;24(5):237-47.

155

CHAPTER 4.2

Socioeconomic inequalities in short-term mortality after acute myocardial infarction

van Oeffelen AAM, Agyemang C, Bots ML, Stronks K, Koopman C, van Rossem L, Vaartjes I. The relation between socioeconomic status and short-term mortality after acute myocardial infarction persists in the elderly: results from a nationwide study.

Eur J Epidemiol 2012;27(8):605-13. Chapter 4.2

ABSTRACT Objective We assessed whether the previously observed relationship between socioeconomic status (SES) and short-term mortality (pre-hospital mortality and 28-day case-fatality) after a first acute myocardial infarction (AMI) in persons <75 years, are also observed in the elderly (i.e. ≥75 years), and whether these relationships vary by sex.

Methods A nationwide register-based cohort study was conducted. Between 1 January 1998 and 31 December 2007, 76,351 first AMI patients were identified, of whom 60,498 (79.2%) were hospitalised. Logistic regression analyses were performed to measure SES differences in pre-hospital mortality after a first AMI event and 28-day case-fatality after a first AMI hospitalisation. All analyses were stratified by sex and age group (<55, 55-64, 65-74, 75-84, ≥85 years), and adjusted for age, ethnic origin, marital status, and degree of urbanisation.

Results There was an inverse relation between SES and pre-hospital mortality in both sexes. There was also an inverse relation between SES and 28-day case-fatality after an AMI hospitalisation, but only in men. Compared with elderly men in the highest SES group, elderly men in the lowest SES group had a higher pre-hospital mortality in both 75-84 year-olds (OR=1.26; 95% CI: 1.09-1.47) and ≥85 year- olds (OR=1.26; 1.00-1.58), and a higher 28-day case-fatality in both 75-84 year-olds (OR=1.26; 1.06- 1.50) and ≥85 year-olds (OR=1.36; 0.99-1.85). Compared with elderly women in the highest SES group, elderly women in the lowest SES group had a higher pre-hospital mortality in ≥85 year-olds (OR=1.20; 0.99-1.46).

Conclusions In men there were SES inequalities in both pre-hospital mortality and case-fatality after a first AMI event; in women these SES inequalities were only shown for pre-hospital mortality. The inequalities persist in the elderly (≥75 years of age). Clinicians and policymakers need to be more vigilant on the population with a low SES background, including the elderly.

158

Socioeconomic inequalities in short-term mortality after acute myocardial infarction

INTRODUCTION

In the last decades the mortality after a cardiovascular event has steadily declined in Western societies,1,2 though a socioeconomic status (SES) gradient still persists.3 Acute myocardial infarction (AMI) is one of the major cardiovascular diseases with a SES gradient in short-term mortality as well as long-term mortality.4 Short-term mortality after an AMI comprises pre-hospital mortality and case-fatality. Pre-hospital mortality is defined as dying after a first AMI attack before hospitalisation; case-fatality is defined as dying within 28 days after the first AMI hospitalisation. Pre-hospital mortality has often shown to be stronger related to SES than case-fatality.5-8 This may be due to the clear relationship between SES and an unfavourable risk factor profile (e.g. smoking, unhealthy diet, inactivity, stress),9 which has more influence on pre-hospital mortality than on case-fatality.(10) Furthermore, time delay between the AMI event and hospitalisation is more prevalent in low SES subjects, which could contribute to the SES gradient in pre-hospital mortality.11 Case-fatality is also probably less subjected to SES differences, because the Netherlands is an equity-oriented country with a well developed social system. Every Dutch citizen is obligatory insured with a health insurance company. For citizens with a low income, the government contributes financially. Emergency hospital care is funded by the insurance company without additional costs. Thus, the availability of emergency hospital care should not vary by SES. Therefore, we expect the SES gradient for case- fatality to be smaller than for pre-hospital mortality. However, earlier studies in countries with similar health care systems still have shown a relation between low SES and increased case-fatality after an AMI.8,12 Studies on age- and sex-specific relations between SES and short-term AMI mortality are often restricted to patients younger than 75 years of age, probably due to a lack of large datasets that are needed for this kind of stratification.6-8,13-17 In the few studies that assessed this relation among patients over 75 years, it diminished in the elderly.13-15 This could be explained by ‘selective survival’.18 Selective survival prevents the sicker individuals in low-income groups to reach high ages, whereas mortality in the high-income groups is postponed to higher ages. This results in comparable mortality risks in the elderly low- and high-income groups. With an increasing ageing population in Europe, it is pertinent to gain more insight in SES inequalities in health in the elderly population. We therefore assessed whether the previously observed relationships between household SES and short-term mortality in the younger age groups are also observed in the elderly (i.e. ≥75 years), and whether these relationships vary by sex.

METHODS

Cohort enrolment Data were obtained from Dutch national registers; e.g. Hospital Discharge Register (HDR), Dutch Population Register (PR), Cause of Death Register (CDR), and Regional Income Survey (RIS). The

159

Chapter 4.2 registers were used to obtain the following information: the HDR for AMI hospitalisations and comorbidities, the PR for demographic factors, the CDR for fatal AMI events, and the RIS for income data used as SES indicator. The registers have been described in detail previously.19,20 By linking previous registers two cohorts were built. First, a cohort of all Dutch subjects with a first (non-fatal or fatal) AMI event (ICD-9 code 410 and ICD-10 code I21) between 1 January 1998 and 31 December 2007 was constructed. Subjects with a previous hospitalisation for AMI from 1995 onwards were excluded. This resulted in a cohort of 260,920 first AMI patients. Second, a cohort of all Dutch patients with a first AMI hospitalisation (ICD-9 code 410) between 1 January 1998 and 31 December 2007 was constructed. Subjects with a previous AMI hospitalisation from 1995 onwards were excluded. This resulted in a cohort of 199,096 subjects with a first AMI hospitalisation. The first cohort included all subjects with a first AMI event and was used to study the pre- hospital mortality, defined as death after a first AMI event before hospitalisation for this event. The second cohort only included subjects with a first hospitalisation for an AMI event and was used to study the 28-day case-fatality, defined as death within 28 days after the first AMI hospitalisation.

Exposure measures Socioeconomic status Income data were obtained from the RIS.21 The RIS started in 1994, when a representative sample of 1.9 million Dutch citizens was selected. Every year, the sample was corrected for emigration and mortality on one hand, and immigration and birth on the other hand. All persons belonging to the households of the sample population (about 5 million) were included in the RIS. SES was defined as the standardised disposable income on household level (adjusted for number of household members) in the year preceding the AMI.

Comorbidity The presence and the extent of comorbidity were determined with the Charlson index score, which proved to be a reliable and valid method to measure comorbidity in clinical research.22 The Charlson index score was constructed using 17 discharge diagnosis of previous admissions, which have been selected and weighted on the basis of the strength of their relation with mortality. The sum of weights represents the Charlson index score.23

Outcome measures Pre-hospital mortality, defined as dying after a first AMI event without hospitalisation, was determined among all subjects with a first AMI during the study period 1998-2007. Case-fatality,

160

Socioeconomic inequalities in short-term mortality after acute myocardial infarction defined as dying within 28 days after a first AMI hospitalisation, was determined among subjects with a first AMI hospitalisation only.

Data analysis To take the effect of inflation and small differences in definition of disposable income over the years into account, the income of the year-specific AMI subjects was divided into quintiles. Subsequently, all year-specific income quintiles were combined in one variable. Baseline characteristics were calculated for every SES quintile of first AMI subjects. Absolute mortality risks were calculated for every SES quintile, stratified by sex and age (<55, 55-64, 65-74, 75-84, ≥85 years). To correct for an unequal age distribution over quintiles, absolute mortality rates were standardised to the age distribution of the study population. Odds Ratio’s (OR) with accompanying 95% CI, expressing the relation between SES and both short-term mortality outcomes, were calculated using multivariate logistic regression models. These analyses were performed in two ways: first with SES quintile dummy variables, with the highest income quintile (quintile 1) as reference, and second, using SES quintiles as a continuous variable to assess the trends across SES quintiles. Similar models were used for analyses in men and women separately. The same approach was used for analyses in the various age strata (<55, 55-64, 65-74, 75-84, ≥85 years). Adjustments were made for potential confounding variables (age, sex, ethnic origin, marital status, degree of urbanisation and Charlson index score). Data were analysed with SPSS software, version 14.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation Netherlands.24

RESULTS

Baseline characteristics Of the 260,920 subjects with a first AMI event, 76,351 had income data available in the year preceding their AMI and were included in the study. Patients included in our study were more often male (68.4% vs. 61.6%), more often married or living together (73.4% vs. 62.2%), more often living in rural areas (38.4% vs. 36.1%), had less often comorbidities (14.9% vs. 16.5%), and were younger (66.6 yr vs. 70.1 yr) than those not included in our study. Compared with first AMI subjects in the high SES group, first AMI subjects in the low SES group were older, more often living alone, more often female, more often of non-native Dutch origin, more commonly living in strong urban areas, and had more often comorbidities (Table 1).

161

Chapter 4.2

c

59.4 83.8 61.6 19.1 23.5 19.4 21.4 16.6 17.4 69.2 9,508 15,267 SES quintile SES5 quintile

c

64.0 89.6 74.3 17.1 24.1 20.6 21.9 16.2 18.1 70.5 15,274 13,315 in the Netherlands between 1998 and 1998 and between Netherlands the in SES quintile 4

event

c

69.6 87.7 76.3 16.9 24.8 21.4 21.9 14.9 15.1 66.0 15,272 16,429 SES quintile 3

c

72.6 87.5 77.3 15.6 24.0 21.2 23.6 15.6 12.4 63.6 15,274 20,749

SES quintile 2

c

76.1 88.0 77.6 15.2 23.1 22.2 24.1 15.5 11.3 63.6

15,264 31,616

SES quintile 1

). Very urban=>2000, urban=1001-2000, urban/rural=501-1000, rural=251-500, very rural=<251 rural=<251 very rural=251-500, urban/rural=501-1000, urban=1001-2000, urban=>2000, Very ).

2 infarction subjects with income data available in year prior to the AMI to the prior year in available data income with subjects infarction

68.4 87.3 73.4 16.8 23.9 20.9 22.6 15.8 14.9 66.6 Total 76,351 18,323

Rural Rural

Urban Urban

b Very rural rural Very Very urban

Urban/rural

a

disposable standardised Characteristics of first acute myocardial myocardial acute of first Characteristics son index score >0 % >0 score son index l

subjects of Number Male % % Dutch Ethnic % together or living Married % urbanisation of Degree Char years in event AMI at Meanage Mean euros in income household Both parents born in the Netherlands the born in parents Both Population density (number of residents per km per of residents (number density Population

SES quintile 1 is the highest SES group, SES quintile 5 is the lowest SES group lowest the is 5 SES SES quintile group, highest the is 1 SES quintile

1 Table 2007 a b c

162

Socioeconomic inequalities in short-term mortality after acute myocardial infarction

Pre-hospital mortality Of all subjects who suffered a first AMI, 15,853 (20.8%) subjects died outside the hospital (60.5% men and 39.5% women). Although the absolute number of total first AMI events was substantially lower in ≥75-year olds compared to those <75 (24,719 and 51,632 respectively), the absolute number of pre-hospital mortality was higher (8,655 and 7,203 respectively) (Table 2). This makes the pre-hospital mortality risk in AMI subjects of ≥75-year old about 2.5 times as high as this risk in AMI patients <75 years. Results of logistic regression analyses showed an inverse relation between SES and pre-hospital mortality, restricted to those in the lowest SES quintile (men: OR=1.24; 1.15-1.33, women: OR=1.26; 1.14-1.39). The results did not meaningfully change after correcting for degree of urbanisation, ethnic origin, marital status and comorbidity. After age stratification, the relations in the younger age categories (<75 years of age) persisted in the elderly (75-84 year-old men: OR=1.26:1.09-1.4, ≥85 year-old men: OR=1.26; 1.00-1.58, ≥85 year-old women: OR=1.20: 0.99-1.46) (Table 3).

Case-fatality Of all subjects who suffered a first AMI, 60,498 (79.2%) were hospitalised (70.5% men and 29.5% women); 9,656 (16.0%) of them died within 28 days (62.5% men and 37.5% women). Although the absolute number of total AMI hospitalisations was substantially lower in ≥75-year olds compared with those <75 years (16,071 and 44,427 respectively), the absolute number of case-fatality was higher (5,227 and 4,423 respectively) (Table 4). This makes the case-fatality risk in AMI patients of ≥75 year old more than 3 times as high as this risk in AMI patients <75 year old. Results of logistic regression analyses showed a gradual increase in case-fatality risk with decreasing SES, but only in men (lowest vs. highest SES quintile: OR=1.28; 1.17-1.41). The results did not meaningfully change after correcting for degree of urbanisation, ethnic origin, marital status and comorbidity. After age stratification, a consistent inverse relation in men of 55 years of age and above was shown, persisting in the older age categories (75-84 years: OR=1.26; 1.06-1.50, ≥85 years: OR=1.36; 0.99- 1.85). In women, no clear SES gradients in case-fatality were found (Table 5).

163

Chapter 4.2

a

20.7 20.1 19.2 19.4 20.6 24.2 18.4 17.7 17.2 17.1 18.8 21.2 26.0 25.4 23.8 24.6 24.5 30.4 ASR

(16.1) (22.8)

Total N (%) 9,776 (21.8) 9,061 (22.7) 3,654 (22.9) 4,184 (20.5) 4,591 5,498 (25.5) 6,206 (34.1) 76,351 (20.8) 15,264 (17.7) 15,274 (16.9) 15,272 (18.7) 15,274 (23.1) 15,267 (27.4) 52,218 (18.4) 11,610 11.090 (15.5) 10,681 (16.9) 24,133 (26.0)

N (%)

≥85 years 935 (46.8) 505 (45.7) 394 (42.6) 473 (41.2) 746 (43.4) 839 (52.4) 565 (52.2) 541 (49.9) 617 (51.2) 973 (46.9)

7,326 (50.0) 1,070 (49.2) 1,090 (46.9) 1,719 (45.4) 2,512 (55.9) 2,957 (45.9) 4,369 (52.7) 1,673 (57.7)

7)

84 years N (%)

- 912 (30.6) 75 2,483 (28.1) 2,616 (26.9) 3,450 (26.8) 4,902 (27.7) 3,942 (33.0) 1,571 (26.7) 1,578 (26.4) 2,094 (26.1) 2,865 (28.5) 1.959 (32.2) 7,326 (29.5) 1,038 (27.7) 1,356 (27.8) 2,037 (26.7) 1,983 (33.8) 17,393 (28. 10,067 (28.1)

group

(21.0)

74 years N (%) - 744 (16.9) (18.0) 982 65 3,138 (18.0) 3,548 (18.0) 4,114 (17.7) 4,495 (20.2) 3,114 (22.4) 2,394 (18.3) 2,566 (18.0) 2,910 (17.9) 3,025 1,942 (22.4) 5,572 (18.8) 1,204 (17.4) 1,470 (18.5) 1,172 (22.4) 18,409 (19.2) 12,837 (19.4)

2007 1998 and between Netherlands the in risk mortality hospital -

(16.2)

64 years N (%) - standardisedto age the distribution of the total study population 721 (8.3) 721 649 (10.6) 707 (12.2) 555 (14.6) 730 (18.9) 55 4,426 (12.1) 3,888 (10.6) 3,201 (12.6) 2,153 (14.1) 2,860 3,777 (12.3) 3,167 (11.1) 2,494 (12.7) 1,598 (14.0) 2,130 (15.2) 3,362 (12.9) 16,528 (12.8) 13,166 (12.8) sex group and pre

- age

-

SES quintile 5 is the lowest income lowest the is 5 group, SES quintile

N (%) 784 (8.5) (7.1) 902 707 (8.2) <55 years <55 463 (11.2) 648 (12.7) 4,147 (9.3) 4,287 (9.0) 3,417 (8.4) 2,005 (9.1) 3,363 (9.4) 3,385 (9.5) 2,710 (8.4) 1,542 (8.5) 3,504 (9.2)

16,695 (9.3) 2,839 (11.1) 13,191 (9.3) 2,191 (10.6) hospital mortality rate: -

ile 1 Total Total Total

First AMI subjects in every SES every in subjects AMI First

standardised pre standardised SES Quintile SES Quintile

- SES Quint SES2 Quintile SES3 Quintile SES4 Quintile SES5 Quintile SES1 Quintile SES2 Quintile SES3 Quintile SES4 Quintile SES5 Quintile 2 SES Quintile SES3 Quintile SES4 Quintile SES5 Quintile ge Total Men Women A Table 2 Table a income highest the is 1 SES quintile

164

Socioeconomic inequalities in short-term mortality after acute myocardial infarction

a

1.10) 1.10) 1.02) 1.16) 1.11) 1.20) 1.58) 1.19) 1.22) 1.01) 1.46)

1.41)*

------

-

≥85 1.00 1.00 1.00 0.005 <0.001 <0.001 OR (95% CI) OR (95% CI) 0.92 (0.77 0.92 (0.78 0.88 (0.75 0.88 (0.67 0.86 (0.66 0.95 (0.76 1.26 (1.00 0.94 (0.74 0.97 (0.77 0.82 (0.66 1.20 (0.99 1.22 (1.05

1.07) 1.06) 1.09) 1.17) 1.14) 1.27) 1.05) 1.06) 1.31)

0.97)* 1.34)* 1.47)*

------

- - 84

- 1.00 1.00 1.00 0.001 75

<0.001 <0.001 OR (95% CI) (95% CI) OR

0.94 (0.83 0.95 (0.84 0.98 (0.88 0.99 (0.85 0.99 (0.85 1.10 (0.96 0.87 (0.71 0.88 (0.73 1.10 (0.93 1.20 (1.08 1.26 (1.09 0.82 (0.69

1.14) 1.09) 1.14) 1.10) 1.37) 1.28) 1.37)

1.29)* 1.39)* 1.35)* 1.41)* 1.64)*

------74

-

1.00 1.00 1.00 0.001 0.109 65 <0.001

(95% CI) OR 1.01 (0.89 0.97 (0.86 0.99 (0.85 0.96 (0.83 1.06 (0.83 1.00 (0.78 1.09 (0.86 index. In unstratified analyses also adjusted for sexfor adjusted also analyses unstratified In index. 1.14 (1.02 1.23 (1.08 1.17 (1.02 1.21 (1.04 1.29 (1.02

0.98) 1.16) 1.34) 1.02) 1.15) 1.29) 1.08) 1.62) 1.98)

1.48)* 1.37)* 2.49)*

------64

- 1.00 1.00 1.00 0.009 0.000 55 group <0.001 (1.01

(95% CI) OR 0.85 (0.74 1.01 (0.88 1.15 (0.98 0.88 (0.76 0.99 (0.85 1.09 (0.91 0.75 (0.52 1.15 (0.82 1.40 (0.99 1.29 (1.12 1.18 1.81 (1.31

first AMI event in the Netherlands between 1998 and 2007, specific for age and sex (odds ratio (OR)) ratio sex (odds and age for 1998 2007, specific and between Netherlands in the AMI event first

1.13) 1.08) 1.19) 1.19) 1.10) 1.13) 1.37) 1.17) 1.35) 1.93)

1.41)* 2.09)*

------

- -

<55 1.00 1.00 1.00 0.023 0.191 0.007 (0.77 OR (95% CI) (95% CI) OR 0.97 (0.84 0.92 (0.78 0.99 (0.82 1.01 (0.86 0.92 0.91 (0.73 1.14 (0.95 0.82 (0.57 0.94 (0.65 1.31 (0.89 1.20 (1.03 1.48 (1.04

hospital mortality after a

- SES quintile 5 is the lowest income lowest the is 5 group, SES quintile

1.00) 1.00) 1.07) 1.03) 1.02) 1.16) 1.00) 1.04) 1.00)

1.33)* 1.33)* 1.39)*

------

1.00 1.00 1.00 Total

<0.001 <0.001 <0.001 OR (95% CI) (95% CI) OR 0.94 (0.88 0.94 (0.89 1.01 (0.95 0.96 (0.89 0.95 (0.88 1.08 (1.00 0.90 (0.80 0.93 (0.84 0.90 (0.81

1.25 (1.18 1.24 (1.15 1.26 (1.14

b b b

P trend P trend P trend

uintile 1

Relation between SES and pre Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile Quintile 1 Quintile Quintile 1 Quintile

en SES SES SES SES SES SES SES SES SES SES SES SES SES SES SES Q SES Total M Women Reference Adjusted age,for ethnic origin, marital status, degree urbanisation,of andCharlson comorbidity

* p<0.05 income highest the is 1 SES quintile 3 Table a b

165

Chapter 4.2

a

16.0 14.9 15.9 15.6 15.8 17.5 14.2 13.0 14.1 13.7 14.3 16.0 20.2 19.6 20.2 20.1 19.5 21.5 ASR

Total N (%)

9,409 (11.2) 8,923 (12.2) 8,735 (13.4) 7,935 (17.7) 7,626 (17.4) 2,687 (17.5) 3,178 (17.6) 3,366 (18.6) 4,167 (21.3) 4,472 (24.1) 60,498 (16.0) 12,096 (12.6) 12,101 (13.6) 12,101 (14.9) 12,102 (18.9) 12,098 (19.9) 42,628 (14.2) 17,870 (20.3)

(43.8) ≥85 N (%) 526 (43.7) 469 (45.8) (44.6) 542 914 (42.7) 265 210 (50.0) 264 (44.3) 415 (44.1) 445 (50.3) 261 (43.7) 259 (42.5) 278 (45.0) 499 (41.5) 770 (44.4)

3,666 (44.8) 1,215 (46.6) 1,599 (46.6) 2,067 (43.4)

84

- N (%) N (%) 75 615 (25.7) 714 (27.9) 891 (27.3) 1,730 (27.1) 1,809 (28.7) 2,318 (26.7) 3,560 (28.9) 2,988 (31.9) 7,237 (29.4) 1,115 (27.9) 1,095 (29.3) 1,427 (26.3) 2,077 (29.5) 1,523 (33.2) 5,168 (28.3) 1,483 (28.1) 1,465 (30.5) 12,405 (28.9)

group

2007 1998 and between Netherlands in the risk fatality

- 74

- N (%) 65 587 (17.5) 756 (16.8) 942 (15.5)

2,462 (15.0) 2,761 (16.0) 3,265 (15.7) 3,669 (16.2) 2,714 (19.2) 1,875 (14.2) 2,005 (15.8) 2,323 (15.8) 2,432 (16.9) 1,710 (19.7) 4,526 (16.4) 1,237 (15.0) 1,004 (18.2) 14,871 (16.4) 10,345 (16.4)

distributionof total the study population

sex group and case and group sex

-

64 age

- - N (%) 55 558 (9.5) 515 (9.7) 627 (9.9) 630 (10.3) 598 (10.4) 3,781 (7.3) 3,322 (8.1) 2,789 (8.8) 2,000 (9.1) 2,520 (9.0) 3,223 (6.9) 2,692 (7.5) 2,191 (8.4) 1,485 (8.9) 1,893 (8.8) 14,412 (8.3) 11,484 (7.9) 2,928 (10.0)

income lowest the is 5 SES quintile

group,

(5.1)

<55 N (%) 666 (6.2) 819 (7.1) 657 (7.8) 433 (6.5) 606 (7.4) 3,597 (4.9) 3,740 (5.3) 3,187 (5.6) 1.959 (4.6) 2,661 2,391 (4.6) 2,921 (4.8) 2,530 (5.0) 1,526 (4.1) 2,055 (4.5) 3,181 (7.0) 15,144 (5.2) 11,693 (4.7)

age the to standardised rate: -fatality

ile 1 Total Total Total

First hospitalised AMI patients in every SES Quint 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile

standardised case standardised

- SES SES SES SES SES SES SES SES SES SES SES SES SES SES SES ge Total Men Women ES quintile 1 is the highest income highest the is 1 ES quintile A

4 Table a S

166

Socioeconomic inequalities in short-term mortality after acute myocardial infarction

a

1.42) 1.35) 1.23) 1.41) 1.88) 1.51) 1.47) 1.85) 1.35) 1.48) 1.25) 1.35)

------

≥85 1.00 1.00 1.00 0.481 0.216 0.901 (0.67 OR (95% CI) OR (95% CI)

1.10 (0.86 1.05 (0.83 0.99 (0.79 1.15 (0.93 1.30 (0.90 1.06 (0.75 1.07 (0.78 1.36 (0.99 0.95 1.05 (0.74 0.92 (0.68 1.02 (0.76

1.26) 1.14) 1.25) 1.30) 1.11) 1.28) 1.42) 1.38) 1.39) 1.52)

1.41)* 1.50)*

------

- - 84

- 1.00 1.00 1.00 0.002 0.003 0.400 75 OR (95% CI) (95% CI) OR

1.09 (0.94 0.99 (0.86 1.10 (0.96 1.08 (0.89 0.93 (0.78 1.09 (0.93 1.12 (0.87 1.09 (0.86 1.12 (0.90 1.23 (0.99 1.24 (1.08 1.26 (1.06 justed for sex

1.25) 1.19) 1.23) 1.34) 1.31) 1.41) 1.25) 1.11) 1.04) 1.29)

1.47)* 1.68)*

------74

-

1.00 1.00 1.00 0.009 0.003 0.290 65

(95% CI) OR 1.08 (0.93 1.03 (0.89 1.06 (0.92 1.13 (0.94 1.10 (0.93 1.19 (1.00 0.94 (0.70 0.84 (0.64 0.80 (0.61 0.98 (0.75 1.27 (1.09 1.40 (1.17

1.29) 1.38) 1.42) 1.37) 1.31) 1.44) 1.55) 1.48) 1.59) 1.55) 1.46) 1.42)

Netherlands between 1998 and 2007, specific for age and sex (odds ratio (OR)) ratio (odds sex and age for 2007, specific 1998 and between Netherlands

------

64

- 1.00 1.00 1.00 0.489 0.293 0.963 55 group

(95% CI) OR 1.08 (0.91 1.15 (0.96 1.16 (0.95 1.13 (0.94 1.07 (0.88 1.18 (0.96 1.23 (0.89 1.19 (0.97 1.09 (0.74 1.05 (0.71 0.97 (0.64 0.96 (0.65

1.34) 1.46) 1.27) 1.35) 1.35) 1.46) 1.29) 1.30) 1.82) 2.02) 1.83) 2.12)

------

<55 1.00 1.00 1.00 0.565 0.742 0.668

(95% CI) OR 1.09 (0.88 1.17 (0.95 0.98 (0.75 1.07 (0.85 1.06 (0.83 1.13 (0.88 0.94 (0.69 0.98 (0.75 1.20 (0.79 1.31 (0.86 1.11 (0.67 1.35 (0.87

fatality after a first AMI hospitalisationinthe -

1.14) 1.21) 1.17) 1.21) 1.17) 1.11) 1.26)

1.18)* 1.17)* 1.32)* 1.24)* 1.41)*

------SES quintile 5 is the lowest income lowest the is 5 group, SES quintile - - - - -

1.00 1.00 1.00 Total 0.172 <0.001 <0.001 (1.01 OR (95% CI) (95% CI) OR 1.06 (0.98 1.10 (1.00 1.06 (0.97 1.05 (0.91 1.02 (0.89 0.98 (0.86 1.11 (0.98 1.09 (1.01 1.09 1.22 (1.13 1.13 (1.03 1.28 (1.17

b b b

P trend P trend P trend

Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile

Quintile 1 Quintile 1 Quintile 1 Quintile

Relation between SES between case and Relation

SES SES SES SES SES SES SES SES SES SES SES SES

SES SES SES Total Men Women Reference

ad analysesalso unstratified In Index. andCharlson urbanisation, of degree status, marital origin, age, ethnic for Adjusted * p<0.05 income highest the is 1 SES quintile Table 5 Table a b

167

Chapter 4.2

DISCUSSION

Previous studies already showed a SES gradient in short-term mortality after AMI in those below 75 years of age. We expand this evidence by showing that this relationship persists in 75 year olds and beyond. Most previous studies excluded AMI patients ≥75 years of age. Those who included them often found less pronounced relations in the elderly.13-15 The lack of a socioeconomic gradient among the elderly is often explained by ‘selective survival’,18 which prevents the sicker individuals in low-income groups to reach high ages. Subsequently, elderly low-income subjects are healthier than their younger counterparts, and SES gradients diminish with age. In our study, the elderly low SES AMI subjects still had a higher short-term mortality compared with the elderly high SES AMI subjects. So even in case of selective survival, relations did not disappear. Besides selective survival, misclassification of SES may have influenced the diminishing relation with age in previous studies. The SES indicators in those studies were on neighbourhood level instead of individual or household level. Neighbourhood level SES is more sensitive to misclassification and underestimation of results, especially in elderly living in institutional care where neighborhood SES is probably less similar to individual SES. The results in our study are based on household level SES, which is a more reliable indicator of a person’s true SES and consequently less prone to misclassification.25 As in many European countries the Dutch population is increasingly ageing, which is partly caused by an increase in life expectancy. Moreover, the baby boomers born close after the Second World War are reaching the age of 65 by now and within ten years they will belong to the group of ≥75 year olds. Our study showed that short-term mortality after a first AMI is about three times as high in ≥75 year olds compared to <75 year olds. Diminishing SES differentials in short-term mortality in the elderly population might thus be a viable step in improving population health, and consequently in reducing health care costs. Interventions intended to promote healthy lifestyle should not only focus on direct change but also on maintenance of change. Elderly persons must be included in these interventions, because some lifestyle changes, like smoking cessation,26 still have beneficial health effects at an older age. Additionally, physicians need to be more vigilant on the elderly population from a low SES background, for example regarding therapy education and compliance.

Pre-hospital mortality Our results showed an increased risk of dying immediately after a first AMI event in patients with the lowest SES, which was of the same magnitude in men and women. The relations were only present when comparing the two most deviating SES quintiles, while previous studies reported graded relations over SES groups.7,8,17 This implies that in the Netherlands only the least wealthy group has a disadvantage with respect to pre-hospital mortality risk after AMI. There are several possible explanations for this, including more unfavourable risk factor profiles9 and seeking medical

168

Socioeconomic inequalities in short-term mortality after acute myocardial infarction care too late in low SES subjects.11 Also larger and more severe infarcts in low SES groups27 and differences in medical care prior to the AMI could have influenced pre-hospital mortality risk adversely.

Case-fatality We expected that the SES gradients in case-fatality would be less pronounced compared with the SES gradients in pre-hospital mortality. This was only correct for women, where no clear relation with case-fatality was found. Our results still showed an inverse relation between SES and case- fatality in men, which was of a similar magnitude as the inverse relation with pre-hospital mortality. This is in line with some previous studies using income on an individual level as SES indicator.8,17,28 The factors involved in an increased pre-hospital mortality risk in low SES groups mentioned before (unfavourable risk factor profiles, seeking medical care too late, and larger and more severe infarcts) can proceed after hospitalisation in survivors, and increase the case-fatality risk. More comorbidity (e.g. diabetes)29 and differences in aggressiveness of treatment after AMI30 are other possible explanations of the SES gradient in case-fatality among men. A SES gradient in case-fatality was found in men but not in women. Salomaa et al presented some possible explanations for this sex difference in Finland, which has a similar health care system as the Netherlands.8 Their research showed a delay from onset of symptoms to medical presence in men with a low SES compared with men with a high SES. In women, no SES difference was found. This time delay in hospitalisation could postpone necessary treatment in low SES men, which may increase case-fatality. They also found that angiography within 28 days after hospitalisation was significantly more often performed in men with high SES than in men with low SES. This was not the case in women. Delay in angiographic procedures could mask the severity of the AMI and might withhold the patient from necessary treatments. This implies that even in an equity-oriented country with a well developed social system, SES gradients in health care exist.

Limitations There are some limitations that should be mentioned. Firstly, although there has been corrected for co-existing diseases, some of them (e.g. diabetes) have been underreported in the HDR. Since persons with a low SES have more comorbidity, the relations reported in this study may have been overestimated. Secondly, the fact that we had no information regarding previous admissions before 1995 might have resulted in some ‘first-time’ AMI events that were actually recurrent events. Recurrence of AMI is non-significantly more common in low SES groups than in high SES groups. Therefore, the inclusion of recurrent events might have led to overestimation of mortality risks (as recurrent events are usually more severe) especially in the low SES groups. Thirdly, we were not able to go into depth concerning underlying mechanisms of our findings, due to the absence of

169

Chapter 4.2 information regarding risk factors, event severity, procedures and medication use. Finally, AMI subjects included in our study (subjects with income data available) had more favourable characteristics compared with AMI subjects not included in our study, which are in general related to better health and a higher income. Including all first AMI patients of the entire Dutch population would probably lead to a larger spread in income range, resulting in more pronounced relations.

Strengths By using national registers we were able to build a cohort of 260,920 first AMI subjects over a ten year time-span including all age ranges. Previously it has been investigated that the overall quality of Dutch national registers is high.31 ICD-9 code 410 and ICD-10 code I21 were used to identify patients with an acute myocardial infarction in the Hospital Discharge Register and the Cause of Death Register respectively. The validity of these ICD-codes was previously assessed, resulting in a sensitivity of 84% and a positive predictive value of 97%. This indicates that most patients coded with ICD-9 code 410 and ICD-10 code I21 actually experienced an AMI event, and are thus correctly coded.32 Furthermore, we had the opportunity to link the Regional Income Survey providing income data on household level. Most other studies used surrogate indicators of SES on neighbourhood level, which inevitably leads to some misclassification and may bias the results towards the null.25 Additionally, the high number of AMI subjects in our study gave us the opportunity to stratify by sex and age, while keeping enough power for the analyses. Unlike most previous studies we included elderly patients (≥75 years of age). With the ageing of the Western population, the number of persons at risk for short-term mortality after a first AMI grows simultaneously. This makes it important to include the elderly population when studying this matter. Because pre-hospital mortality and case-fatality partly differ in underlying mechanisms, we studied both outcome measures separately. This large register-based study includes persons of all ages, a SES indicator on household level, stratified analyses by sex and age, and distincts between pre-hospital mortality and case-fatality. As such this study is, apart from very relevant for clinicians and policy makers, unique and expands existing evidence.

Conclusion In men there were SES inequalities in both pre-hospital mortality and case-fatality after a first AMI event, in women these SES inequalities were only shown for pre-hospital mortality. The inequalities persist in the elderly. Interventions should focus on healthy lifestyle promotion and maintenance over the life course in low SES groups, which should not be restricted to the young. Clinicians working in primary as well as secondary care need to be more vigilant on the population with a low SES background, including the elderly.

170

Socioeconomic inequalities in short-term mortality after acute myocardial infarction

REFERENCES 1. Australian Institute of Health and Welfare 2010. Cardiovascular disease mortality: trends at different ages. Cardiovascular series no. 31. Cat. no. 47. Canberra: AIHW 2011. 2. Vaartjes I, O'Flaherty M, Grobbee DE, Bots ML, Capewell S. Coronary heart disease mortality trends in the Netherlands 1972-2007. Heart 2011;97(7):569-73. 3. Harper S, Lynch J, Smith GD. Social determinants and the decline of cardiovascular diseases: understanding the links. Annu Rev Public Health 2011;2132:39-69. 4. Alboni P, Amadei A, Scarfo S, Bettiol K, Ippolito F, Baggioni G. In industrialized nations, a low socioeconomic status represents an independent predictor of mortality in patients with acute myocardial infarction. Ital Heart J 2003;4(8):551-8. 5. Cesana G, Ferrario M, Gigante S, Sega R, Toso C, Achilli F. Socio-occupational differences in acute myocardial infarction case-fatality and coronary care in a northern Italian population. Int J Epidemiol 2001;Suppl 1:S53-S58. 6. Morrison C, Woodward M, Leslie W, Tunstall-Pedoe H. Effect of socioeconomic group on incidence of, management of, and survival after myocardial infarction and coronary death: analysis of community coronary event register. BMJ 1997;22314(7080):541-6. 7. Rosvall M, Gerward S, Engstrom G, Hedblad B. Income and short-term case fatality after myocardial infarction in the whole middle-aged population of Malmo, Sweden. Eur J Public Health 2008;18(5):533-8. 8. Salomaa V, Miettinen H, Niemela M, Ketonen M, Mahonen M, Immonen-Raiha P, et al. Relation of socioeconomic position to the case fatality, prognosis and treatment of myocardial infarction events; the FINMONICA MI Register Study. J Epidemiol Community Health 2001;55(7):475-82. 9. 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 Pt 1):2172-9. 10. Wannamethee G, Whincup PH, Shaper AG, Walker M, MacFarlane PW. Factors determining case fatality in myocardial infarction "who dies in a heart attack"? Br Heart J 1995;74(3):324-31. 11. Ghali JK, Cooper RS, Kowatly I, Liao Y. Delay between onset of chest pain and arrival to the coronary care unit among minority and disadvantaged patients. J Natl Med Assoc 1993;85(3):180-4. 12. Alter DA, Naylor CD, Austin P, Tu JV. Effects of socioeconomic status on access to invasive cardiac procedures and on mortality after acute myocardial infarction. N Engl J Med 1999;28341(18):1359-67. 13. Davies CA, Leyland AH. Trends and inequalities in short-term acute myocardial infarction case fatality in Scotland, 1988-2004. Popul Health Metr 2010;8:33. 14. Gerward S, Tyden P, Hansen O, Engstrom G, Janzon L, Hedblad B. Survival rate 28 days after hospital admission with first myocardial infarction. Inverse relationship with socio-economic circumstances. J Intern Med 2006;259(2):164-72. 15. MacIntyre K, Stewart S, Chalmers J, Pell J, Finlayson A, Boyd J, et al. Relation between socioeconomic deprivation and death from a first myocardial infarction in Scotland: population based analysis. BMJ 2001;12322(7295):1152-3. 16. 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. 17. Salomaa V, Niemela M, Miettinen H, Ketonen M, Immonen-Raiha P, Koskinen S, et al. Relationship of socioeconomic status to the incidence and prehospital, 28-day, and 1-year mortality rates of acute coronary events in the FINMONICA myocardial infarction register study. Circulation 2000;25101(16):1913-8.

171

Chapter 4.2

18. Merlo J, Gerdtham UG, Lynch J, Beckman A, Norlund A, Lithman T. Social inequalities in health- do they diminish with age? Revisiting the question in Sweden 1999. Int J Equity Health 2003;112(1):2. 19. Agyemang C, Vaartjes I, Bots ML, van Valkengoed IG, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 20. Koek HL, de Bruin A, Gast F, Gevers E, Kardaun JW, Reitsma JB, et al. Short- and long-term prognosis after acute myocardial infarction in men versus women. Am J Cardiol 2006;1598(8):993-9. 21. Ament P, Kessels W. Regionaal Inkomensonderzoek: uitgebreide onderzoeksbeschrijving. Voorburg: Centraal Bureau voor de Statistiek (CBS) 2008. 22. Groot de V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56(3):221-9. 23. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57(12):1288-94. 24. Reitsma JB, Kardaun JW, Gevers E, de Bruin A, van der Wal J, Bonsel GJ. [Possibilities for anonymous follow-up studies of patients in Dutch national medical registrations using the Municipal Population Register: a pilot study]. Ned Tijdschr Geneeskd 2003;15147(46):2286-90. 25. McLoone P, Ellaway A. Postcodes don't indicate individuals' social class. BMJ 1999;9319(7215):1003-4. 26. Appel DW, Aldrich TK. Smoking cessation in the elderly. Clin Geriatr Med 2003;19(1):77-100. 27. Barakat K, Stevenson S, Wilkinson P, Suliman A, Ranjadayalan K, Timmis AD. Socioeconomic differentials in recurrent ischaemia and mortality after acute myocardial infarction. Heart 2001;85(4):390-4. 28. Rosvall M, Chaix B, Lynch J, Lindstrom 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;8:44. 29. Miettinen H, Lehto S, Salomaa V, Mahonen M, Niemela M, Haffner SM, et al. Impact of diabetes on mortality after the first myocardial infarction. The FINMONICA Myocardial Infarction Register Study Group. Diabetes Care 1998;21(1):69-75. 30. Quatromoni J, Jones R. Inequalities in socio-economic status and invasive procedures for coronary heart disease: a comparison between the USA and the UK. Int J Clin Pract 2008;62(12):1910-9. 31. Mackenbach JP, van Duyne WM. [Registration and coding of various causes of death in the Netherlands and other EEC countries]. Ned Tijdschr Geneeskd 1984;7128(1):13-8. 32. Merry AH, Boer JM, Schouten LJ, Feskens EJ, Verschuren WM, Gorgels AP, et al. Validity of coronary heart diseases and heart failure based on hospital discharge and mortality data in the Netherlands using the cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009;24(5):237-47.

172

CHAPTER 5

INEQUALITIES IN CARDIOVASCULAR HEALTH CARE USE

CHAPTER 5.1

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care

van Oeffelen AAM, Vaartjes I, Stronks K, Bots ML, Rutten FH, Agyemang C, Spiering W. Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care.

Manuscript draft. Chapter 5.1

ABSTRACT Objective The aim of our study was to investigate ethnic inequalities in blood pressure and lipid lowering drug dispense and quitting rates in primary care.

Methods Persons aged ≥30 years or over without a prior cardiovascular event were selected from a database of the Achmea health insurance company during 2006-2010. The ethnic Dutch population and first generation ethnic minority groups (henceforth, migrants) from Suriname, Morocco, Turkey, Netherlands Antilles, Indonesia, China, and South Asia were included (n=210,728). Logistic regression analyses were used to identify ethnic inequalities in blood pressure and lipid lowering drug dispense. Ethnic inequalities in quitting rates were explored using Cox proportional hazard regression analyses.

Results Most migrant groups more often collected a prescription for blood pressure lowering drugs (except for Indonesians and Chinese) and lipid lowering drugs (except for Chinese) than ethnic Dutch. Among those on blood pressure lowering drugs, migrants less often dispensed beta-blockers; and among those on lipid lowering drugs, migrants more often dispensed simvastatin than ethnic Dutch. Migrants had a significantly higher risk on quitting drug treatment (except for Indonesians) than ethnic Dutch. For blood pressure lowering drug quitting, the Hazard Ratio ranged from 1.32 in Chinese to 2.17 in Turkish migrants, and for lipid lowering drug quitting, from 1.51 in Chinese to 2.22 in Turkish migrants.

Conclusion Cardiovascular drug dispense was higher in migrants than in ethnic Dutch, but migrants were more likely to quit. Research is necessary to disentangle the causes of the high cardiovascular drug quitting rates, so that care givers can make efforts to improve drug continuation among migrants.

176

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care

INTRODUCTION

Longevity has substantially improved by the treatment of cardiovascular risk factors and diseases with available cardiovascular drugs. Individuals at high risk for cardiovascular disease (CVD) need to take blood pressure and lipid lowering drugs. Drug therapy in combination with lifestyle modification are key to prevent or delay CVD in those not yet affected (henceforth, the asymptomatic population). To be effective, it is important that drugs are taken as prescribed. Underuse and quitting reduce the beneficial effect of these drugs and place a burden on health care.1 In the Netherlands, similar to other Western countries, ethnic minority groups have in general a somewhat more unbeneficial cardiovascular risk factor profile than the ethnic Dutch population.2-6 Therefore, we could expect a higher cardiovascular drug dispense among asymptomatic ethnic minorities compared with ethnic Dutch. To the best of our knowledge, this has not been investigated yet. Also, ethnic minority groups differ genetically, and it has been suggested that some subclasses of cardiovascular drugs have differential effects among ethnic groups, i.e. ACE- inhibitors and beta-blockers seem less effective in African Americans than in White Americans.7 To our knowledge, such a comparison is lacking for Europe. Our group recently showed that several ethnic minority groups (e.g. Surinamese, Turkish and South Asians) have a higher risk on acute myocardial infarction (AMI) compared with the ethnic Dutch population.8 Amongst other reasons, inadequate use or early quitting of preventive cardiovascular drugs may be part of this observation. Studies concerning ethnic inequalities in cardiovascular drug quitting rates in the asymptomatic population are scarce and limited to the USA. These studies showed a higher quitting rate among African Americans compared with White Americans.9;10 Underlying factors may include difficulties in access to care, paying out-of-pocket prescription costs, and more adverse side effects or beliefs about side effects.1;11;12 Generalisation of these results to the European situation is difficult due to differences in type of health care system and the more recent migration history of ethnic minority groups. The aim of our study was twofold. The first aim was to describe differences in blood pressure and lipid lowering drug dispense between first generation ethnic minority groups (henceforth, migrants) and ethnic Dutch. The second aim was to investigate differences in quitting rates of blood pressure and lipid lowering drugs between migrants and ethnic Dutch.

METHODS

Data sources Data were extracted from the Achmea Health Database (AHD). In the Netherlands all inhabitants are by law obliged to have medical insurance coverage. The Achmea health insurance company is the main health insurance company in the central part of the Netherlands. It provides health care

177

Chapter 5.1 coverage for 1.3 million Dutch residents. In the AHD payments for the provision of all dispensed drugs to insured patients are recorded. The AHD can be considered representative for the entire Dutch population. Detailed information about the AHD has been described previously.13 Because of a transition in the registration procedure in 2005, we restricted our dataset to the period 2006 to 2011. We used a subset of the AHD, consisting of data from all insured Hindustani Surinamese, Moroccan, and Turkish minority groups aged ≥30 years (n=162,484). Their ethnicity was based on nationality and surnames, as described previously.14 In short, first generation ethnic minorities were selected by their nationality. To select the second and third generation, the surnames of the selected first generation ethnic minorities were matched with the remaining of the database and visually controlled on origin of the name. We added a representative sample of the remaining insured group aged ≥30 years (n=194,993), consisting of ethnic Dutch and ethnic minorities other than Hindustani Surinamese, Moroccan, and Turkish. The total cohort consisted of 357,477 adult persons. Approval from the AHD research committee was obtained prior to access of the data. Figure 1 presents a flow chart to our final cohort. The inclusion of persons was limited to the years 2006 to 2010 to enable linkage of population characteristics from nationwide registers (which were not yet available for the year 2011), as described previously.8 Excluding the year 2011 left a cohort of 337,988 persons. We linked our dataset with the national Hospital Discharge Register (HDR) to obtain information regarding previous cardiovascular (CVD) hospitalisations and comorbidities. Only those without a CVD event in the HDR (data available from 1995 to 2010) and also in our dataset from the AHD (data from 2006 to 2010) were included, because we wanted to confine our question on drug dispense and quitting in primary prevention, that is in adults without CVD. This was the case in 254,315 persons. The dataset was also linked with the Population Register (PR) and the Regional Income Survey (RIS) to obtain information regarding country of birth, marital status and socioeconomic status (SES) of these persons.

Determinants Ethnic background Migrant groups were identified using the country of birth and the country of birth of the parents, as recorded in the PR. A person belonged to a migrant group if he/she was born abroad and at least one of the parents was born abroad.15 Second generation ethnic minority groups were excluded. For this study, ethnic Dutch and migrants born in Suriname, Morocco, Turkey, Netherlands Antilles, Indonesia, China, and South Asia were included. The Surinamese population in the Netherlands is ethnically diverse, and mainly consists of people from Hindustani South-Asian descent and from West-African descent. We disaggregated the Hindustani Surinamese from the non-Hindustani Surinamese by using Hindustani surnames available from the Achmea Health Database and the SUNSET study.2;14 The South-Asian migrant group consisted of migrants from India, Pakistan and Sri

178

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care

Other N=22,115

N=290 N=75

South Asian Chinese

N=345 Indonesian

N=1,966 South Asian

N=208

Antillean N=918 Chinese

N=23,726 N=23,726

Turkish N=4,149

N=1,996

dispense drug lowering Lipid Indonesian

N=4,978 Moroccan

N=194,993

2010

- 2011 -

N=2,357

Antillean

Hindustani N=415

-

Surinamese N=2,031 Random sample of the remaining group remaining the of sample Random N=11,235 Hindustani Surinamese Non Ethnic Dutch

3 million

Turkish N=35,632 N=232,843 N=210,728

N≈1. N=254,315

Study population Study population

N=408

N=164

AGIS Health Database 2006 Database AGIS Health Chinese South Asian

generation ethnic minority groups based on country of birth of country on based groups minority ethnic generation Asymptomatic population 2006 population Asymptomatic

N=44,443 Moroccan

N=629

Indonesian

minorities based on surname on based minorities

N=544

Antillean Hindustani Hindustani -

N=3,792 Surinamese Surinamese Ethnic Dutch and first first and Dutch Ethnic

Non

N=46,670 N=162,484 Turkish N=7,202

pressure lowering drug dispense dispense drug lowering pressure

N=15,353 Hindustani Hindustani Surinamese Surinamese Blood Blood

N=7,985 Moroccan

Flow inclusion chart of procedure

Hindustani N=104,271 N=971 - Ethnic Dutch Ethnic Hindustani Surinamese, Moroccan, and Turkish Turkish and Moroccan, Surinamese, Hindustani Surinamese N=3,993 N=24,783 Hindustani Surinamese Non Ethnic Dutch Figure 1 1 Figure

179

Chapter 5.1

Lanka. A patient was considered ethnic Dutch when both parents were born in the Netherlands. The final cohort comprised 210,728 adult persons.

Neighbourhood socioeconomic status Socioeconomic status (SES) was based on income data available from tax authorities registered in the RIS. The RIS is described in detail previously.16 In short, the RIS started in 1994, when a representative sample of 1.9 million Dutch citizens was selected. Every year, the sample was corrected for emigration and mortality on one hand, and immigration and birth on the other hand. All persons belonging to the households of the sample population (about one third of the Dutch population) were included in the RIS. Within each neighborhood, the mean disposable income of the residents with income data available was calculated and subsequently assigned to all residents living in that neighborhood. For this study, the neighborhood income during 2006 was assigned. Subsequently, neighborhood income was divided into SES tertiles, with the first tertile representing the lowest income group.

Comorbidity Presence and extent of comorbidity were determined with the Charlson index score, based on discharge diagnosis in the Hospital Discharge Register from 1995 to 2010.17 The Charlson index score ranges from zero to six (cut-off value), with zero representing no comorbidity. This score proved to be a reliable and valid method to measure comorbidity in clinical research.18

Outcomes Cardiovascular drug dispense Blood pressure lowering drug dispense was defined as having collected at least one prescription of beta-blockers, ACE-inhibitors, angiotensin-II-antagonists, calcium-channel blockers or diuretics during the study period 2006-2010. Lipid lowering drug dispense was defined as having collected at least one prescription of statins or other lipid lowering drugs during the study period. Drugs were identified using the Anatomical Therapeutic Chemical Classification System (ATC codes), as presented in Table 1.

Cardiovascular drug quitting rates Among those with at least one collected blood pressure lowering (n=46,679 (22.2%)) or lipid lowering drug prescription (n=23,726 (11.3%)) between 2006 and 2010, quitting rates were investigated until 31 December 2011. When the defined daily dose (DDD) plus a ‘grace period’ of 90 days expired before a new prescription was dispensed, we assumed the patient had quitted taking the drug.19 The DDD is the average maintenance dose per day for a drug used for its main indication

180

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care in adults. Persons were followed from the date of first drug dispense to the date of quitting, and were censored in case of death, end of Achmea insurance, or the end of the study period on 31 December 2011, whichever came first.

Table 1 Categories for each CVD drug type and accompanying Anatomical Therapeutic Chemical (ATC) codes Type of drug Drug category ATC classification Blood pressure lowering drugs Diuretics C03A, C03B, C03C Beta-blockers C07 Calcium-channel blockers C08 ACE-inhibitors C09A, C09B Angiotensin-II-antagonists C09C, C09D Lipid lowering drugs Statins C10AA Other lipid lowering drugs C10AB, C10AC, C10AD, C10AX

Data analysis Patient characteristics were analysed within the ethnic Dutch population and migrant groups separately, using cross tables and frequency tables. Logistic regression analyses were used to calculate differences in blood pressure and lipid lowering drug dispense between migrants and ethnic Dutch. Among those who collected at least one prescription of blood pressure or lipid lowering drugs, differences in type of first dispensed drug between migrants and ethnic Dutch were investigated using logistic regression analyses. Cox proportional hazard regression analyses were used to calculate differences in cardiovascular drug quitting rates between migrants and ethnic Dutch. Sensitivity analyses were carried out using a ‘grace period’ of 120 days and 180 days instead of 90 days, to explore whether results were robust. For logistic regression and Cox proportional hazard regression analyses, two consecutive models were built. In model 1 we adjusted for age, sex, degree of urbanisation, and marital status. In model 2 we additionally adjusted for neighbourhood SES and comorbidity. We used SPSS software, version 20.0 (SPSS Inc, Chicago, Illinois, USA).

RESULTS

Population characteristics Compared with ethnic Dutch, migrant groups were younger (except for Indonesians), and lived in more urbanised low SES areas (Table 2). Among ethnic Dutch, 23.8% collected a blood pressure lowering drug, whereas in migrants it ranged between 17.9% in Chinese and 31.5% in Indonesians. Among ethnic Dutch, 10.8% collected a lipid lowering drug, whereas in migrants it ranged between 8.2% in Chinese and 17.3% in Indonesians. Beta-blockers were the most commonly dispensed blood pressure lowering drugs, except among Surinamese, Antillean, and Indonesian migrants in whom diuretics were more often dispensed. Simvastatin was without an exception the most commonly used lipid lowering drug.

181

Chapter 5.1

46)

-

f f

- - 8.9 1.6 6.6 5.5 2.0 2.3 4.3 8.3 3.9 65.3 77.0 12.4 66.1 60.6 27.3 12.1 20.8 14.8 10.6

Asian 1,966 South 39 (32

53)

-

f

- 6.6 1.3 6.2 3.2 1.1 3.5 3.9 7.0 8.2 6.2 1.6 918 45.2 57.0 20.4 14.7 56.0 38.9 41.1 20.0 17.9

Chinese 41 (33

68)

-

2.3 8.1 5.8 4.3 3.3 4.8 0.5 10.1 26.6 19.7 10.9 30.2 29.7 37.9 40.5 46.3 14.4 40.1 31.5 17.3 12.1 1,996 56 (49 Indonesian

51)

-

f

- 3.0 0.6 6.6 3.5 2.2 2.8 8.0 8.8 7.0 1.8

48.0 56.0 23.3 17.2 21.3 10.2 52.0 32.8 15.3 23.1 2,357 Antillean 42 (33

46)

-

5.8 0.2 9.7 7.6 3.6 2.3 1.9 4.8 9.7 8.1 3.2 0.3 52.3 59.3 20.0 14.6 77.9 67.9 22.4 20.2 11.6 35,632 Turkish

38 (32 995 until 2010 500, very rural=<251 -

48)

-

3.3 0.3 6.3 4.5 1.5 1.6 4.1 5.5 8.6 2.4 0.2 54.4 72.6 13.7 10.2 74.4 68.2 20.4 11.3 18.0 11.2 44,443 Moroccan 38 (32 1000, rural=251

-

2010; SES: socioeconomic status socioeconomic SES: 2010;

- 51)

-

between 2006and 2010

2.3 0.3 7.4 4.4 2.5 3.4 7.8 8.2 2.5 0.3 45.3 69.3 14.1 14.0 23.1 13.0 57.7 27.2 15.1 25.6 10.9 3,792 Hindustani -

43 (35 Surinamese Non

2000, urban/rural=501

-

50)

-

1.9 0.4 6.8 4.8 2.6 4.1 7.7 2.8 0.3 46.2 70.0 14.2 13.4 26.1 12.0 57.7 27.4 14.9 26.0 13.2 10.2 15,353 42 (35 Hindustani Surinamese

umber of cases was less than ten than less was cases of umber

50)

-

0 years of age insured with Achmea 2.0 0.4 7.0 4.7 2.6 4.0 7.7 9.8 2.7 0.3 46.0 69.9 14.2 13.5 25.5 12.2 57.7 27.3 15.0 25.9 12.8 Total 19,148 42 (35 Surinamese ery urban=>2000, urban=1001

). V

59) 2

-

drugs dispensed

8.0 7.8 3.9 2.4 2.0 7.7 6.9 3.5 0.4 44.4 23.5 23.7 23.8 21.1 47.1 13.3 29.4 35.1 35.4 23.8 10.8 Dutch Ethnic 104,271 46 (36

hospitalisation for CVD or a PCI/CABG procedure during 1995

d % Total Total Rural residents km per Urban

blockers Diuretics c -blockers -

inhibitors Very rural pressure lowering b - - Very urban antagonists Simvastatin Urban/rural lowering drugs dispensed

- Other statins e

II Beta ACE - lowering drugs channel

- ) a

SES %

Tertile Tertile (low1 income) Tertile (high3 income) Calcium Other lipid Angiotensin Tertile Tertile (medium2 income)

pressure lowering drugs

Population characteristics of asymptomatic persons ≥3 prescription of blood t Persons Male % Median age (IQR Degree of urbanisation % Married/living together % Charlson index score > % 0 Neighbourhood Blood loweringLipid drugs % Interquartile range Charlson comorbitiy index > 0: at least one hospitalisation for a diagnosis included in the Charlson comorbidity index from 1 Population density (number of Firs First prescription of lipid Notgiven in line with theDutch dataprotection guidelineas the n

Table 2 a b c d e f a without those persons: Asymptomatic

182

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care

Cardiovascular drug dispense After adjustment for confounders, most migrant groups had a significantly higher blood pressure lowering drug dispense compared with ethnic Dutch (Table 3), ranging from a Hazard Ratio (HR) with accompanying 95% confidence interval (CI) of 1.22 (1.17-1.26) in Moroccans to 1.88 (1.80-1.97) in Hindustani Surinamese. Only Indonesians and Chinese had a similar blood pressure lowering drug dispense as ethnic Dutch. Also lipid lowering drug dispense was significantly higher among migrants, with HR (95% CI) ranging from 1.16 (1.02-1.31) in Indonesians to 2.70 (2.37-3.08) in South Asians. Only Chinese had a similar lipid lowering drug dispense as ethnic Dutch. Additional adjustment for neighbourhood SES and comorbidity did not influence results. Among migrants who collected blood pressure lowering drugs at least once, the first dispensed prescription was less often a beta-blocker than among ethnic Dutch (table 4). Dispense of ACE-inhibitors and angiotensin-II-antagonists was higher or at least similar in migrants than in ethnic Dutch, whereas dispense of calcium-channel blockers was higher among all migrant groups. Among those who collected lipid lowering drugs at least once, the first dispensed prescription was more often simvastatin in migrants than in ethnic Dutch (table 5).

Cardiovascular drug quitting rates Among ethnic Dutch, 21.7% stopped collecting their blood pressure lowering drugs during the study period. For lipid lowering drugs this was 25.3% (Table 6). Among migrant groups these percentages were higher, ranging between 24.3% and 40.7% for blood pressure lowering drugs, and between 27.0% and 41.7% for lipid lowering drugs. After adjustment for confounders (model 1), most migrant groups more often quitted their cardiovascular drug therapy than ethnic Dutch. Only in Indonesian migrants cardiovascular drug quitting rates were comparable to ethnic Dutch. Results remained in general similar when analyses were performed separately in the subgroups of blood pressure and lipid lowering drugs (results not shown). Additional adjustment for neighbourhood SES and comorbidity did not influence results.

183

Chapter 5.1

Table 3 Difference between migrants and ethnic Dutch in cardiovascular drug dispense N (% of population) OR (95% CI) Model 1c Model 2d Blood pressure lowering drugsa Ethnic Dutch 24,783 (23.8) 1.00 1.00 Total Surinamese 4,964 (25.9) 1.91 (1.83-1.99)* 1.89 (1.82-1.97)* Hindustani Surinamese 3,993 (26.0) 1.90 (1.82-1.99)* 1.88 (1.80-1.97)* Non-Hindustani Surinamese 971 (25.6) 1.72 (1.59-1.87)* 1.71 (1.58-1.86)* Moroccans 7,985 (18.0) 1.22 (1.17-1.26)* 1.22 (1.18-1.27)* Turkish 7,202 (20.2) 1.53 (1.47-1.58)* 1.50 (1.45-1.56)* Antillean 544 (23.1) 1.69 (1.52-1.87)* 1.69 (1.52-1.87)* Indonesian 629 (31.5) 0.91 (0.82-1.01) 0.91 (0.82-1.01) Chinese 164 (17.9) 0.93 (0.77-1.12) 0.94 (0.79-1.13) South Asian 408 (20.8) 1.96 (1.74-2.20)* 1.94 (1.73-2.18)* Lipid lowering drugsb Ethnic Dutch 11,235 (10.8) 1.00 1.00 Total Surinamese 2,446 (12.8) 2.00 (1.90-2.11)* 1.96 (1.86-2.07)* Hindustani Surinamese 2,031 (13.2) 1.98 (1.87-2.10)* 1.94 (1.83-2.05)* Non-Hindustani Surinamese 415 (10.9) 1.48 (1.33-1.66)* 1.45 (1.30-1.62)* Moroccans 4,978 (11.2) 1.51 (1.45-1.58)* 1.48 (1.42-1.55)* Turkish 4,149 (11.6) 1.76 (1.68-1.84)* 1.70 (1.62-1.78)* Antillean 208 (8.8) 1.38 (1.19-1.60)* 1.36 (1.17-1.58)* Indonesian 345 (17.3) 1.16 (1.02-1.31)* 1.15 (1.02-1.31)* Chinese 75 (8.2) 0.89 (0.70-1.15) 0.90 (0.70-1.15) South Asian 290 (14.8) 2.70 (2.37-3.08)* 2.64 (2.31-3.01)* a Beta-blockers, ACE-inhibitors, angiotensin-II-antagonists, calcium-channel blockers, diuretics b Statins and other lipid lowering drugs c Adjusted for age, sex, degree of urbanisation, and marital status d Adjusted for age, sex, degree of urbanisation, marital status, SES, and Charlson comorbidity index * p<0.05 Reference: ethnic Dutch

184

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care

b ugs at 1.28) 1.38) 1.41) 1.11) 1.36) 1.27)* 0.89)* 1.34)* 1.80)* blood

- - - - -

- - - -

1.00 (0.69 OR (95% CI) OR

1.13 (1.00 1.11 (0.89 1.05 (0.78 0.58 (0.31 0.97 1.13 (1.01 0.80 (0.72 1.22 (1.11 1.43 (1.13 antagonists - II - pressure lowering dr

a

Angiotensin N (%) 96 (9.9) 51 (9.4) 10 (6.1) 40 (9.8) 491 (9.9) 395 (9.9) 646 (8.1) 86 (13.7) 836 (11.6) 2,542 (10.3)

b b 1.15) 1.16) 1.26) 1.06) 1.14) 0.89)* 0.93)* 1.67)* 0.98)* 1.35) 1.22) 1.41) 1.61)

- - - - - 1.33)* 1.35)* 1.89)* 1.26)* 2.03)*

------

the of total populationwho collected any of type 1.00 1.00 (0.81 among those who collectedblood (1.63

OR (95% CI) OR OR (95% CI) OR

1.07 (0.99 1.07 (0.98 1.09 (0.95 0.89 (0.75 0.89 (0.70 0.83 (0.78 0.87 1.39 (1.16 0.67 (0.46 1.13 (0.95 0.96 (0.75 1.15 (0.93 1.07 (0.71 1.22 (1.11 1.23 (1.12 1.76 1.17 (1.08 1.61 (1.28

inhibitors - Diuretics

ACE

a a

(20.8) N (%) N (%) 36 (22.0) 85 82 (15.1) 29 (17.7) 297 (30.6) 189 (34.7) (32.1) 202 900 (18.1) 732 (18.3) (17.3) 168 115 (18.3) 108 (26.5) 8,043 (32.5) 1,482 (29.9) 1,185 (29.7) 1,821 (22.8) 1,704 (23.7) 4,058 (16.4) 2,006 (25.1) 1,296 (18.0)

pressure lowering drug dispensed

b

b 1.21) 1.64) 1.83) 1.32) 2.16)* 2.25)* 2.05)* 1.33)* 2.00)* 3.99)*

- - - 0.62)* 0.60)* 0.76)* 0.85)* 0.93)* 0.75)* 0.95)* 0.84)*

------

------1.00 1.00 (95% CI) (1.81

percentage the with first, drugs lowering pressure OR (95% CI) OR OR 1.09 (0.98 1.26 (0.97 1.33 (0.96 0.95 (0.68 1.95 (1.76 2.02 1.68 (1.38 1.20 (1.08 1.53 (1.18 2.69 (1.82 0.57 (0.53 0.55 (0.51 0.65 (0.56 0.80 (0.75 0.87 (0.82 0.62 (0.51 0.79 (0.66 0.67 (0.54

blockers - channel blockers channel -

Beta

a a ic Dutchin type of first blood status marital , and (26.3) Calcium N (%) N (%) 708 (8.9) 666 (9.2) 67 (12.3) 65 (10.3) 32 (19.5) 45 (11.0) 57 (34.8) 760 (15.3) 631 (15.8) 129 (13.3)

(28.9) 281 155 (28.5) 161 (25.6) 130 (31.9) 2,037 (8.2) 8,103 (32.7) 1,331 (26.8) 1,050 2,804 (35.1) 2,700 (37.5)

Difference between migrants and ethn Hindustani Surinamese Hindustani Surinamese Hindustani

- - Dutch Ethnic Surinamese Total Surinamese Hindustani Non Moroccans Turkish Antillean Indonesian Chinese South Asian Dutch Ethnic Surinamese Total Surinamese Hindustani Non Moroccans Turkish Antillean Indonesian Chinese South Asian Number persons of who collectedthat specific type blood of Adjusted age,for sex, degree urbanisation of

<0.05 * p <0.05 Table 4 Table least once a b Reference: ethnic Dutch once least at drugs lowering pressure

185

Chapter 5.1

Table 5 Difference between migrants and ethnic Dutch in type of first lipid lowering drug dispensed among those who collected lipid lowering drugs at least once N (%)a OR (95% CI)b Simvastatin Ethnic Dutch 5,070 (20.5) 1.00 Total Surinamese 1,285 (25.9) 1.82 (1.62-2.03)* Hindustani Surinamese 1,057 (26.5) 1.86 (1.65-2.10)* Non-Hindustani Surinamese 228 (23.5) 1.67 (1.33-2.11)* Moroccans 2,182 (27.3) 1.87 (1.71-2.05)* Turkish 1,665 (23.1) 1.30 (1.19-1.42)* Antillean 125 (23.0) 2.17 (1.54-3.04)* Indonesian 170 (27.0) 1.28 (1.02-1.62)* Chinese 32 (19.5) 1.81 (1.06-3.08)* South Asian 114 (27.9) 1.49 (1.14-1.95)* Other statins Ethnic Dutch 2,504 (10.1) 1.00 Total Surinamese 373 (7.5) 0.59 (0.52-0.66)* Hindustani Surinamese 303 (7.6) 0.58 (0.51-0.65)* Non-Hindustani Surinamese 70 (7.2) 0.62 (0.49-0.79)* Moroccans 726 (9.1) 0.57 (0.52-0.62)* Turkish 737 (10.2) 0.82 (0.75-0.90)* Antillean 32 (5.9) 0.54 (0.38-0.76)* Indonesian 67 (10.7) 0.80 (0.63-1.01) Chinese 10 (6.1) 0.51 (0.29-0.91)* South Asian 59 (14.5) 0.74 (0.57-0.98)* Other lipid lowering drugs Ethnic Dutch 214 (0.9) 1.00 Total Surinamese 29 (0.6) 0.50 (0.36-0.68)* Hindustani Surinamese 25 (0.6) 0.47 (0.33-0.66)* Non-Hindustani Surinamese 4 (0.4) 0.63 (0.34-1.16) Moroccans 56 (0.7) 0.49 (0.38-0.63)* Turkish 45 (0.6) 0.60 (0.47-0.77)* Antillean 1 (0.2) 0.11 (0.02-0.78)* Indonesian 6 (1.0) 0.80 (0.41-1.58) Chinese 3 (1.8) 1.14 (0.36-3.67) South Asian 5 (1.2) 0.43 (0.19-0.99)* a Number of persons who collected that specific type of lipid lowering drugs first, with the percentage of the total population who collected any type of lipid lowering drugs at least once b Adjusted for age, sex, degree of urbanisation, and marital status * p<0.05 Reference: ethnic Dutch

186

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care

Table 6 Difference between migrants and ethnic Dutch in cardiovascular drug quitting rates, using a 90 days ‘grace period’ N (% of population) HR (95% CI) Model 1c Model 2d Blood pressure lowering drugsa Ethnic Dutch 5,366 (21.7) 1.00 1.00 Total Surinamese 1,830 (36.9) 1.61 (1.52-1.72)* 1.62 (1.52-1.73)* Hindustani Surinamese 1,474 (36.9) 1.61 (1.51-1.72)* 1.62 (1.52-1.73)* Non-Hindustani Surinamese 356 (36.7) 1.58 (1.41-1.77)* 1.60 (1.42-1.79)* Moroccans 2,960 (37.1) 1.84 (1.74-1.94)* 1.86 (1.75-1.97)* Turkish 2,930 (40.7) 2.17 (2.05-2.29)* 2.20 (2.08-2.33)* Antillean 180 (33.1) 1.44 (1.23-1.68)* 1.45 (1.24-1.69)* Indonesian 153 (24.3) 1.09 (0.92-1.28) 1.10 (0.93-1.29) Chinese 49 (29.9) 1.32 (0.99-1.76) 1.34 (1.00-1.78) South Asian 147 (36.0) 1.69 (1.43-2.01)* 1.73 (1.46-2.05)* Lipid lowering drugsb Ethnic Dutch 2,842 (25.3) 1.00 1.00 Total Surinamese 933 (38.1) 1.76 (1.61-1.91)* 1.75 (1.60-1.90)* Hindustani Surinamese 773 (38.1) 1.77 (1.61-1.93)* 1.75 (1.60-1.92)* Non-Hindustani Surinamese 160 (38.6) 1.72 (1.46-2.03)* 1.70 (1.44-2.01)* Moroccans 1,987 (39.9) 1.94 (1.80-2.08)* 1.92 (1.78-2.07)* Turkish 1,705 (41.1) 2.22 (2.06-2.38)* 2.18 (2.02-2.35)* Antillean 67 (32.2) 1.49 (1.16-1.92)* 1.47 (1.14-1.90)* Indonesian 93 (27.0) 1.01 (0.82-1.25) 1.01 (0.82-1.25) Chinese 23 (30.7) 1.51 (1.00-2.28) 1.51 (1.00-2.28) South Asian 121 (41.7) 1.95 (1.61-2.36)* 1.93 (1.59-2.34) a Beta blockers, ACE-inhibitors, angiotensin-II-antagonists, calcium channel blockers, and diuretics b Statins and other lipid lowering drugs c Adjusted for age, sex, degree of urbanisation, and marital status d Adjusted for age, sex, degree of urbanisation, marital status, SES, and Charlson comorbidity index * p<0.05 Censoring in case of death, end of Achmea insurance, or end of study period on 31 December 2011 Quitting: not collecting a new prescription after the total number of DDD’s plus a ‘grace period’ of 90 days expired Reference: ethnic Dutch

187

Chapter 5.1

DISCUSSION

In this register-based study, we demonstrated that migrants without a prior cardiovascular event more often collected a prescription for blood pressure and lipid lowering drugs than ethnic Dutch. Among migrants, the first dispensed blood pressure lowering drug was less often a beta-blocker than among ethnic Dutch, whereas it was more often or just as often an ACE-inhibitor, angiotensin- II-antagonist, or calcium-channel-blocker. Among those who collected at least one prescription for lipid lowering drugs, migrants more often collected simvastatin than ethnic Dutch. Interestingly, migrants had a substantially higher risk on quitting cardiovascular drug treatment during the study period (except for Indonesians) compared with ethnic Dutch.

Discussion of main findings Cardiovascular drug dispense To the best of our knowledge, this is the first study on ethnic inequalities in cardiovascular drug dispense in asymptomatic adults (those without CVD). Our data, showing a higher cardiovascular drug dispense among most migrant groups compared with ethnic Dutch, are compatible with the view that many migrant groups have a more unbeneficial cardiovascular risk factor profile than ethnic Dutch.2;4;5;20 Because we were unable to collect information on the actual ‘need’ for cardiovascular drugs (i.e. the indication), we could not investigate whether the higher dispense covered the need. Our study shows, given the apparent indication for use of a drug, differences in type of first dispensed blood pressure and lipid lowering drugs. Among migrants, beta-blockers were consistently less often dispensed than among ethnic Dutch, whereas especially calcium-channel blockers were more often dispensed. Variation in the metabolism and side-effects from subtypes of blood pressure lowering drugs between ethnic groups could underlie these differences. However, we were unable to collect such information. Dutch guidelines do not make a difference between ethnic groups regarding preferences for certain cardiovascular drugs, with the exception for those from African descent; in them, calcium-channel blockers and diuretics are preferred over other types of blood pressure lowering drugs.21 Africans are known not to react as well on beta-blockers, ACE-inhibitors and angiotensin-II-receptor blockers as Whites, and they experience more side effects.22 In our study, dispense of calcium-channel blockers among non-Hindustani Surinamese and Antillean migrants (both groups are in majority from African descent) was higher than among ethnic Dutch, which is in line with the Dutch guidelines. Why most other migrant groups also had a higher dispense of calcium-channel blockers remains unclear. ACE-inhibitors were more often dispensed among migrants born in Suriname, Morocco, Turkey, and South Asia. This may be related to the higher prevalence of diabetes in these migrant groups, and thus a higher risk on developing proteinuria.3;4;23 In patients with diabetes and proteinuria, ACE-inhibitors (but also angiotensin-II-

188

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care antagonists) are the preferred blood-pressure lowering drugs.24 For lipid lowering medication, simvastatin dispense was higher among migrants than among ethnic Dutch. In the Netherlands, simvastatin is the preferred lipid lowering drug.21 It is the cheapest statin type and when given in low-to-moderate doses, it does not differ in adverse side effects from other statins.25 Prescription of other more potent lipid lowering drugs is allowed given appropriate arguments why simvastatin is not adequate (e.g. desired achievable level, side effects). To the best of our knowledge there is no scientific evidence available concerning differences in need for more potent types of statins between ethnic groups, and the lower simvastatin use within the ethnic Dutch population remains to be elucidated.

Cardiovascular drug quitting rates Previous studies reported higher cardiovascular drug quitting rates among African Americans compared with White Americans.9;10 Our study builds on this evidence by showing a higher quitting rate with blood-pressure lowering and lipid lowering drugs in the majority of migrant groups in a European setting. Although drug dispense was higher, migrants probably faced more barriers which made them quit more often. First, migrants may have language difficulties impairing comprehension of medical care instructions concerning usage and importance.11 There are indications that it is more difficult to reach mutual understanding during a GP consultation among migrant groups than among ethnic Dutch, resulting in non-compliance to drug use.26 Second, migrants may experience more side effects, or are more worried about potential side effects. For example, ACE-inhibitors are more likely to cause angioedema in Africans, and cough in East-Asian and African groups, than in the general population.27 Of Surinamese men in the Netherlands we know that quitting their blood pressure lowering drug therapy is more common, because of worries about negative effects on their sexual performance.28 Third, health illiteracy and cultural differences may contribute to the higher CVD drug quitting rates among migrants.11;28;29 For example, migrants may interpret the lack of symptoms as a sign that no further drug therapy is needed. Finally, preference for alternative treatment may influence the decision to stop medication.28 The high quitting rate among migrants may partially underlie the higher AMI incidence among some of the major migrant groups (especially the Surinamese) compared with ethnic Dutch, that our group recently showed.8 In contrast to the other migrant groups, Indonesian individuals had similar cardiovascular drug quitting rates as ethnic Dutch. Underlying factors of this observation are unclear.

Considerations Our study has several strengths. The register-based design of the study resulted in a large dataset with over 200,000 subjects, which could be followed for a period of maximally six years. Information regarding drug use was obtained from a health insurance company, in which registration of

189

Chapter 5.1 dispensed drugs is extensively controlled because of financial reimbursement.13 We had the opportunity to link these data with several nationwide registers which allowed us to identify a wide range of migrant groups based on the country of birth. Data on cardiovascular history was available for minimally 11 and maximally 16 years. Therefore, we are almost certain we have dealt with an asymptomatic population (persons without CVD). Inevitably our study has some limitations. Firstly, cardiovascular drug dispense is highly dependent on the persons’ cardiovascular risk factor profile. Unfortunately, we did not possess such data and could not adjust analyses for it. Consequently, we cannot draw any conclusions on whether the drug dispense covered the need. Secondly, quitting was defined as not collecting a new prescription 90 days after the number of DDD’s had expired. The DDD is the average maintenance dose per day for a drug used for its main indication in adults. However, there are inter-personal differences in dosage of cardiovascular drugs. Some persons could have been identified as quitters, while they had a much lower maintenance dose and did not stop their medication at all. However, results of sensitivity analyses in which a period of 120 days and 180 days after the number of DDD’s expired was used remained similar, which indicates that data were robust. Although the use of registers has several benefits (such as the absence of recall-bias and overestimation of adherence), we were unable to determine whether the patient actually consumed the dispensed drugs.

Conclusion Cardiovascular drug dispense was higher in migrants than in ethnic Dutch, but migrants were more likely to quit. Research is necessary to disentangle the causes of the high cardiovascular drug quitting rates, so that care givers can make efforts to improve drug continuation among migrants.

190

Ethnic inequalities in cardiovascular drug dispense and quitting rates in primary care

REFERENCES 1. Kronish IM, Ye S. Adherence to cardiovascular medications: lessons learned and future directions. Prog Cardiovasc Dis 2013;55(6):590-600.

2. Agyemang C, Bindraban N, Mairuha G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7.

3. Agyemang C, van Valkengoed I, van den Born BJ, Bhopal R, Stronks K. Heterogeneity in seks differences in the metabolic syndrome in Dutch white, Surinamese African and South Asian populations. Diabet Med 2012;29(9):1159-64.

4. Ujcic-Voortman JK, Schram MT, Jacobs-van der Bruggen MA, Verhoeff AP, Baan CA. Diabetes prevalence and risk factors among ethnic minorities. Eur J Public Health 2009;19(5):511-5.

5. Ujcic-Voortman JK, Bos G, Baan CA, Uitenbroek DG, Verhoeff AP, Seidell JC. Ethnic differences in total and HDL cholesterol among Turkish, Moroccan and Dutch ethnic groups living in Amsterdam, The Netherlands. BMC Public Health 2010;10:740.

6. Ujcic-Voortman JK, Baan CA, Seidell JC, Verhoeff AP. Obesity and cardiovascular disease risk among Turkish and Moroccan migrant groups in Europe: a systematic review. Obes Rev 2012;13(1):2-16.

7. Brewster LM, van Montfrans GA, Kleijnen J. Systematic review: antihypertensive drug therapy in black patients. Ann Intern Med 2004;141(8):614-27.

8. van Oeffelen AA, Vaartjes I, Stronks K, Bots ML, Agyemang C. Incidence of acute myocardial infarction in first and second generation minority groups: Does the second generation converge towards the majority population? Int J Cardiol 2013;168(6):5422-9.

9. Charles H, Good CB, Hanusa BH, Chang CC, Whittle J. Racial differences in adherence to cardiac medications. J Natl Med Assoc 2003;95(1):17-27.

10. Lewey J, Shrank WH, Bowry AD, Kilabuk E, Brennan TA, Choudhry NK. Gender and racial disparities in adherence to statin therapy: a meta-analyses. Am Heart J 2013;165(5):665-78.

11. Gazmaraian JA, Kripalani S, Miller MJ, Echt KV, Ren J, Rask K. Factors associated with medication refill adherence in cardiovascular-related diseases: a focus on health literacy. J Gen Inern Med 2006;21(12):1215-21.

12. Kaplan RC, Bhalodkar NC, Brown EJ Jr., White J, Brown DL. Race, ethnicity, and sociocultural characteristics predict noncompliance with lipid-lowering medications. Prev Med 2004;39(6):1249-55.

13. Smeets HM, de Wit NJ, Hoes AW. Routine health insurance data for scientific research: potential and limitations of the Agis Health Database. J Clin Epidemiol 2011;64(4):424-30.

14. Boelman L, Smeets HM, Knol MJ, Braam AW, Geerlings MI, de Wit NJ. Psychotropic drug use in patients with various chronic somatic diseases. Eur J Psychiat 2012;26(4):236-47.

15. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of ‘country of birth’ for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69.

16. Ament P, Kessels W. Regionaal Inkomensonderzoek: uitgebreide onderzoeksbeschrijving. Voorburg: Centraal Bureau voor de Statistiek (CBS) 2008.

17. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57(12):1288-94.

191

Chapter 5.1

18. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. A critical review of available methods. J Clin Epidemiol 2003;56(3):221-9.

19. Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiol Drug Saf 2006;15(8):565-74.

20. Chaturvedi N. Ethnic differences in cardiovascular disease. Heart 2003;89(6):681-6.

21. www.nhg.org/standaarden/samenvatting/cardiovasculair-risicomanagement.nl

22. El Desoky ES, Derendorf H, Klotz U. Variability in response to cardiovascular drugs. Curr Clin Pharmacol 2006;1(1):35-46.

23. Barnett AH, Dixon AN, Bellary S, Hanif MW, O’hare JP, Raymond NT, et al. Type 2 diabetes and cardiovascular risk in the UK South Asian community. Diabetologia 2006;49(10):2234-46.

24. Lv J, Perkovic V, Foote CV, Craig ME, Craig JC, Strippoli GF. Antihypertensive agents for preventing diabetetic kidney disease. Cochrane Database Syst Rev 2012;12:CD004136.

25. Backes JM, Howard PA, Ruisinger JF, Moriarty PM. Does simvastatin cause more myotoxicity compared with other statins? Ann Parmacother 2009;43(12):2012-20.

26. van Wieringen JC, Harmsen JA, Bruijnzeels MA. Intercultural communication in general practice. Eur J Public Health 2002;12(1):63-8.

27. McDowell SE, Coleman JJ, Ferner RE. Systematic review and meta-analysis of ethnic differences in risks of adverse reactions to drugs used in cardiovascular medicine. BMJ 2006;332(7551):1177-81.

28. Beune EJ, Haafkens JA, Agyemang C, Schuster JS, Willems DL. How Ghanaian, African Surinamese and Dutch patients perceive and manage antihypertensive drug treatment: a qualitative study. J Hypertens 2008;26(4):648-56.

29. Beune EJ, Haafkens JA, Schuster JS, Bindels PJ. ‘Under pressure’: how Ghanaian, African-Surinamese and Dutch patients explain hypertension. J Hum Hypertens 2006;20(12):946-55.

192

CHAPTER 5.2

Ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction

van Oeffelen AAM, Rittersma S, Vaartjes I, Stronks K, Bots ML, Agyemang C. Are there ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction?

Submitted. Chapter 5.2

ABSTRACT Objective Previously we observed ethnic inequalities in prognosis after a first cardiovascular event. This might be due to differences in revascularisation rate between ethnic minority groups and ethnic Dutch. Therefore, we investigated inequalities in revascularisation rate after occurrence of an ST-elevation myocardial infarction (STEMI) between first generation ethnic minority groups (henceforth, migrants) and ethnic Dutch.

Methods All STEMI events between 2006 and 2011 were identified in a subset of the Achmea Health Database, which records medical care to persons insured at the Achmea health insurance company, a major health insurance company in the central part of the Netherlands. Ethnic Dutch and migrants from Suriname (Hindustani Surinamese and non-Hindustani Surinamese), Morocco, and Turkey were selected (n=1,765). Multivariable Cox proportional hazards regression analyses were used to identify ethnic inequalities in revascularisation rate (percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG)) after a STEMI event.

Results On average, 73.2% of STEMI events were followed by a revascularisation procedure. After adjustment for confounders (age, sex, degree of urbanisation) no significant differences in revascularisation rate were found between the ethnic Dutch population and Hindustani Surinamese (Hazard Ratio (HR): 1.04; 95% confidence interval (CI): 0.85-1.27), non-Hindustani Surinamese (HR: 0.98; 95% CI: 0.63-1.51), Moroccan (HR: 0.94; 95% CI: 0.77-1.14) and Turkish migrants (HR: 1.04; 95% CI: 0.88-1.24). Additional adjustment for comorbidity and neighborhood income did not change our findings.

Conclusion Our study suggests no ethnic inequalities in revascularisation rate after a STEMI event. This finding is in agreement with the universally accessible health care system in the Netherlands.

194

Ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction

INTRODUCTION

An ST-elevation myocardial infarction (STEMI) is characterised by an acute total occlusion of one or more coronary arteries, resulting in necrosis of the myocardium. In case of a STEMI, as diagnosed on the ECG, treatment is immediately indicated. The occlusion can be resolved using medication therapy (e.g. fibrinolysis) or a revascularisation procedure, such as percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG). In the past decade more evidence became available emphasizing the beneficial effect of revascularisation procedures in STEMI patients.1;2 Recent guidelines therefore recommend such a procedure over medication therapy after a STEMI event, as it improves prognosis when performed within an adequately short time frame (preferably within 90 minutes after first medical contact).3-5 Guidelines do not mention contra- indications for performing a revascularisation procedure. In 2004, the European Heart Survey reported that only 61% of all STEMI patients received a revascularisation procedure, suggesting a marked underuse.6 A previous study of our group showed that mortality after a first hospitalisation for cardiovascular disease (CVD) event was higher among ethnic minority groups compared with the ethnic Dutch population.7 We hypothesised that ethnic inequalities in cardiac revascularisation procedures might partially explain these findings, since this was also observed previously in USA studies. African Americans were less likely to receive a revascularisation procedure after an acute myocardial infarction (AMI) than their White American counterparts, which explained their higher mortality into some extent.8-11 An important underlying factor was the lack of health insurance and the inability to pay for such an expensive procedure, especially among the African-American population.12 In Europe, literature concerning ethnic inequalities in cardiac revascularisation procedures is scarce. Many European countries have a health care system with universal access to acute in-hospital care which aims at minimizing health care inequalities. Though, in the United Kingdom (where revascularisation procedures are performed without costs for the patient) a lower revascularisation procedure rate among African and South-Asian minorities compared with the general population was still reported, even after adjustment for the actual need.13-15 However, more recent data do not show ethnic inequalities in revascularisation procedure rate anymore.16 Evidence for other European countries, such as the Netherlands, is lacking. Given the lack of information on this issue, and the possibility that differences in revascularisation rate between ethnic groups may explain ethnic inequalities in prognosis after CVD, we set out to study differences in revascularisation procedure rate (PCI and CABG) after a STEMI event between first generation ethnic minority groups (henceforth, migrants) and ethnic Dutch in the Netherlands.

195

Chapter 5.2

METHODS

Data sources Data were extracted from the Achmea Health Database (AHD). In the Netherlands all inhabitants are by law obliged to have medical insurance coverage. The Achmea health insurance company is the main health insurance company in the central part of the Netherlands as it provides health care coverage for 1.3 million Dutch residents, of which 20% belongs to an ethnic minority group. The AHD records payments for the provision of all medical care to insured patients. Although the AHD is not completely representative for the entire Dutch population, it does represent the urbanised areas of the Netherlands. Detailed information about the AHD is described previously.17 Because of a transition in type of registration procedure in 2005, our dataset was restricted to the period 2006 to 2011. Due to privacy issues we only received a subset of the AHD. The subset consisted of data from all insured Hindustani Surinamese, Moroccan, and Turkish ethnic minority groups of ≥30 years of age (n=162,484), whose ethnicity was based on nationality and surnames, as described previously.18 In short, first generation minorities were selected by their nationality. To select the second and third generation, the surnames of the selected first generation minorities were matched with the remaining of the database and visually controlled on origin of the name. Furthermore, our subset consisted of a representative sample of the remaining insured group aged ≥30 years (n=194,993), consisting of ethnic Dutch and ethnic minorities other than Hindustani Surinamese, Moroccan, and Turkish. This resulted in a cohort of 357,477 insured persons. Approval from the AHD research committee was obtained prior to accessing the data. From the AHD cohort, all inpatient and outpatient STEMI events were selected. Subsequently, for every patient only the first event within 30 days was retained. For each patient we determined whether they received a PCI or CABG procedure. They were followed, and censored in case of death, end of Achmea insurance, or the end of the study period at 31 December 2011, whichever came first. The dataset was linked with the Population Register (PR), Hospital Discharge Register (HDR), and the Regional Income Survey (RIS) to obtain information regarding country of birth, comorbidity, and neighbourhood income (as indicator for socioeconomic status).

Determinants Ethnic background Migrant groups were identified using the country of birth and the country of birth of the parents, as recorded in the PR. A patient belonged to a migrant group if he/she was born abroad and at least one of the parents was born abroad.19 Migrants from Suriname, Morocco, and Turkey were selected. Other migrant groups were too small to include. The Surinamese population in the Netherlands is ethnically diverse, and mainly consists of people from Hindustani South-Asian descent and from West-African descent. We disaggregated the Hindustani Surinamese from the non-Hindustani

196

Ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction

Surinamese by using Hindustani surnames available from the Achmea Health Database and the Amsterdam SUNSET study.18;20 A patient was considered ethnic Dutch when both parents were born in the Netherlands. This left a cohort of 1,765 STEMI events.

Degree of urbanisation The degree of urbanisation was based on the population density of the city where the insured person lived on 1 January 2006, extracted from the AHD. Five categories were constructed: very urban, urban, urban/rural, rural, and very rural.

Neighbourhood socioeconomic status Socioeconomic status (SES) was based on income data registered in the RIS.21 The RIS started in 1994, when a representative sample of 1.9 million Dutch citizens was selected. Every year, the sample was corrected for emigration and mortality on one hand, and immigration and birth on the other hand. For all persons belonging to the households of the sample population (about one third of the Dutch population) the household income was available. Within each neighbourhood, the mean of all registered household incomes was calculated and subsequently assigned to all residents living in that neighbourhood. For this study, the neighbourhood income of the patients during 2006 was assigned. Subsequently, neighbourhood income was divided into SES tertiles, with the first tertile representing the lowest income group.

Comorbidity Presence and extent of comorbidity were determined with the Charlson index score,22 based on discharge diagnosis in the Hospital Discharge Register from 1995 to the date of the AMI event. The Charlson index ranges from zero to six (cut-off value), with zero representing no comorbidity. It proved to be a reliable and valid method to measure comorbidity in clinical research.23

Data analysis Patient characteristics were analysed within the ethnic Dutch population and the migrant groups separately, using cross tables and frequency tables. Cox proportional hazard regression analyses were used to calculate the difference in revascularisation rate after the STEMI event between migrant groups and ethnic Dutch (reference). Three consecutive models were built. Model one included only confounders (age, sex, degree of urbanisation). Model two and three also included comorbidity and neighbourhood SES, to investigate whether these factors could explain found relations. We used SPSS software, version 20.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation Netherlands.

197

Chapter 5.2

RESULTS

Table 1 presents the patient characteristics of the study population, comprising 1,765 STEMI events, of which 36% belonged to a migrant group.

75) -

7.2 5.0 73.2 69.7 66.7 45.8 28.2 26.0 43.1 18.5 18.3 15.1 63.9 Total 1,765 395,839 64 (53

65)

-

g

- 7.6 7.6 9.7 7.6 237 80.6 78.9 81.0 72.2 18.1 58.2 18.1 15.6 64.6 45,675 Turkish 54 (47

70)

ten -

g g

- - 6.2 6.2 9.4 192 71.9 69.8 83.9 71.9 14.6 13.5 77.6 10.9 23.4 57,413 Moroccan 62 (52

66) -

-

g g g g g g g

------

28 78.6 75.0 75.0 71.4 75.0 71.4 Non 7,543 56 (45 Hindustani Hindustani

Surinamese

65)

-

g g

- - 9.4 9.4 9.9 171 13.5 13.5 78.4 74.3 74.3 55.6 29.2 15.2 76.0 69.6 37,009 56 (48 Hindustani Hindustani Surinamese Surinamese

2011 2006 and between Database Health Achmea the in age

level

65)

-

g g

- - 9.0 199 27.6 14.6 10.6 13.1 78.4 74.4 74.4 57.8 75.9 69.8 Total Total 44,552 56 (48 Surinamese Surinamese ): very urban=>2000, urban=1001-2000, urban/rural=501-1000, rural=251-500, very rural=<251 rural=<251 very rural=251-500, urban/rural=501-1000, urban=1001-2000, urban=>2000, very ): 2

79) -

6.9 6.9 7.6

71.0 67.0 59.5 33.9 32.6 33.5 28.3 21.3 21.3 21.5 69.6 1,137 248,199

69 (57 Ethnic Dutch Ethnic

a b e

PCI

d Rural Rural

f Urban Urban CABG

)

rural Very c Very urban Urban/rural Urban/rural

%

days at risk at days - Characteristics of persons with a STEMI event ≥30 years of of years ≥30 STEMIa event with persons of Characteristics Tertile 1(low income) Tertile 3 (high income)

Tertile 2(medium income) events STEMI Person Procedures % type Procedure (IQR Median age Men % SES Neighbourhood % % urbanisation of Degree % 0 > index Charlson Socioeconomic status basedon income on neighbourhood Population density (number of residents per per km of residents (number density Population Percutaneous coronary intervention Coronary artery bypass grafting Interquartile range

than less were cases as the guideline protection data Dutch the with in line not given cases of number Actual STEMI the from 1995 until index comorbidity Charlson the in included diagnosis a for hospitalisation one at 0: least > index comorbidity Charlson event Table 1 Table a b c d e f g

198

Ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction

Within the ethnic Dutch population, 71.0% of STEMI events were followed by a revascularisation procedure. Within the migrant groups this percentage was higher (ranging from 71.9% among Moroccans to 80.6% among Turkish migrants). Migrants were more often men, younger, and lived in more urbanised low SES neighbourhoods. Comorbidity was about equal between migrants and ethnic Dutch, except for Moroccans who had substantially less comorbidity. Within all groups, PCI was the most commonly performed revascularisation procedure. After adjustment for the confounders age, sex, and degree of urbanisation (model 1) there were no significant differences in revascularisation procedure rate between migrant groups and the ethnic Dutch population (Table 2). Adding comorbidity (model 2) and neighbourhood SES (model 3) did not markedly influence results.

Table 2 Difference in revascularisation procedure rate after a STEMI event between migrants and the ethnic Dutch population ≥30 years of age Procedure (%) HR (95% CI) Model 1a Model 2b Model 3c Ethnic Dutch 807 (71.0) 1.00 1.00 1.00 Total Surinamese 156 (78.4) 1.04 (0.86-1.25) 1.04 (0.86-1.26) 1.03 (0.85-1.24) Hindustani Surinamese 134 (78.4) 1.04 (0.85-1.27) 1.04 (0.85-1.27) 1.03 (0.84-1.26) Non-Hindustani Surinamese 22 (78.6) 0.98 (0.63-1.51) 0.98 (0.64-1.52) 0.97 (0.62-1.49) Moroccan 138 (71.9) 0.94 (0.77-1.14) 0.89 (0.73-1.09) 0.87 (0.71-1.07) Turkish 191 (80.6) 1.04 (0.88-1.24) 1.04 (0.87-1.23) 1.01 (0.85-1.21) a Adjusted for age, sex, and degree of urbanisation b Adjusted for age, sex, degree of urbanisation, and Charlson Index c Adjusted for age, sex, degree of urbanisation, Charlson comorbidity index, and neighbourhood socioeconomic status * p-value <0.05 Reference: ethnic Dutch

DISCUSSION

Our study shows no ethnic inequalities in revascularisation rate after an ST-elevation myocardial infarction in the Netherlands.

Discussion of main findings Our results are in contrast with the overwhelming literature from the USA, showing considerable lower revascularisation rates among African Americans compared with White Americans after a coronary event.8;10;11 A major difference between the USA and the Netherlands is the universal access to health care implemented in Dutch society, in which everyone is obliged by law to have health insurance. In the Netherlands, persons with a low income are being supported by the government through health care benefits, and PCI and CABG are available without additional costs. In the United States, still more than 16% of the population is uninsured. This percentage is higher in

199

Chapter 5.2

African Americans (20.8%) than in non-Hispanic Whites (11.7%).12 The inability of the uninsured to pay for PCI and CABG procedures may partially underlie the lower procedure rate in African Americans. Nevertheless, lack of health insurance cannot fully explain the ethnic inequalities, since Cram et al recently reported a lower revascularisation rate after AMI among African Americans independent of health insurance.24 Also a study that was executed in Medicare beneficiaries only, still revealed lower revascularisation rates among African Americans.8 However, in most studies executed in the USA, not only STEMI events were included, but also non-ST-elevation myocardial infarctions (non-STEMI). The choice to perform a revascularisation procedure after a non-STEMI is much more dependent on clinical symptoms than after a STEMI event.25 The underlying indication, based on clinical symptoms, may differ across ethnic groups. Indeed, some studies reported that the lower revascularisation procedure rate among African Americans was partially explained by a lower coronary artery disease burden and severity.26-28 Yet, one study which included African Americans and Whites with a similar disease severity still showed a lower revascularisation rate among African Americans.10 More reluctance in African Americans than in Whites to accept a recommendation for surgery or PCI may additionally underlie the lower procedure use.29 The absent ethnic inequalities in revascularisation procedure rate in our study not only suggests equal provision of care by physicians, but also equal uptake of care by patients. Until now, European studies on ethnic inequalities in revascularisation rate after a coronary event were limited to the UK.13-15 Just as in the Netherlands, the UK has free access to acute in-hospital care. However, studies executed in the 2000s still found lower rates of revascularisation procedures after a coronary event among African and South-Asian minorities. A major drawback in two of these studies was that they did not adjust for need for a revascularisation procedure. This is an important factor in the decision to perform a revascularisation procedure after a non-ST-elevation cardiac event.4 Although one study also found ethnic inequalities among those deemed appropriate for revascularisation, this study dates back from 2002 and is therefore considered outdated given the current guidelines.13 The most recent UK study from 2013 did not find ethnic inequalities in revascularisation rate after AMI anymore.16 Previously, our research group found a higher mortality rate after a first hospitalisation for a cardiovascular event among ethnic minority groups compared with the ethnic Dutch population.7 Results from our present study suggest that this cannot be explained by differences in revascularisation procedure rate between ethnic groups. However, we did not have any information regarding the time span between symptom onset and the revascularisation procedure. International literature suggests that this time span might be longer among ethnic minority groups than among the ‘majority population’.30 Since ‘time is muscle’, a longer time delay can substantially undermine the beneficial effect of a revascularisation procedure on prognosis.31 Ethnic inequalities in the time

200

Ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction period between onset of symptoms and the revascularisation procedure has to be investigated in future studies.

Considerations Ethnic inequalities in revascularisation rate after AMI is rarely studied in countries with universal access to health care. We had the opportunity to use data from the Achmea health database, in which registration of procedures is extensively controlled for the reason of financial reimbursement.17 We used surname algorithms in combination with country of origin to disaggregate the Hindustani Surinamese from other Surinamese, who have very different cardiovascular disease profiles. Furthermore, our analyses were based on those with a STEMI event only. The choice to perform a revascularisation procedure after a STEMI is more straightforward than after a non-STEMI, and not dependent on other clinical factors (such as coronary artery disease burden and severity). The absence of ethnic inequalities in revascularisation procedure rate after STEMI as reported in this study therefore truly reflects equity in acute in-hospital health care use for individuals with STEMI. Our study has some limitations. Although registration of procedures in the Achmea Health Database is extensively controlled, this is a very rough check to secure that the registered procedures are in agreement with the registered diagnosis. There are no detailed validation studies available that have investigated the sensitivity and the positive predictive value of registered diagnosis and procedures. Furthermore, the revascularisation rate of 73% seems rather low, since there are no clear contra-indications for performing a revascularisation procedure among STEMI patients. However, this percentage is higher as the previously reported 61% in Europe.6 More research is necessary to understand the underlying factors of this percentage. For now, we do not assume these factors to differ between ethnic groups, and therefore we believe it would not have affected our results concerning ethnic inequalities in revascularisation procedure rate. Finally, in the Achmea Health Database, persons living in urban areas of the Netherlands are overrepresented. However, correction for degree of urbanisation did not change relations, which indicates that results are representative for the general Dutch population.

Conclusion Our findings show no ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction in the Netherlands, suggesting equity in acute in-hospital care. The previously observed higher mortality rate after cardiovascular disease among ethnic minority groups in the Netherlands are therefore unlikely to be explained by differences in revascularisation procedures.

201

Chapter 5.2

REFERENCES 1. Boersma E. Does time matter? A pooled analysis of randomized clinical trials comparing primary percutaneous coronary intervention and in-hospital fibrinolysis in acute myocardial infarction patients. Eur Heart J 2006;27(7):779-88. 2. Keeley EC, Boura JA, Grines CL. Primary angioplasty versus intravenous thrombolytic therapy for acute myocardial infarction: a quantitative review of 23 randomised trials. Lancet 2003;361(9351):13-20. 3. Chan MY, Sun JL, Newby LK, Shaw LK, Lin M, Peterson ED, et al. Long-term mortality of patients undergoing cardiac catheterization for ST-elevation and non-ST-elevation myocardial infarction. Circulation 2009;119(24):3110-7. 4. Steg PG, James SK, Atar D, Badano LP, Blomstrom-Lundqvist C, Borger MA, et al. ESC Guidelines for the management of acute myocardial infarction in patients presenting with ST-segment elevation. Eur Heart J 2012;33(20):2569-619. 5. Stenestrand U, Wallentin L. Early revascularization and 1-year survival in 14-day survivors of acute myocardial infarction: a prospective cohort study. Lancet 2002;359:1805-11. 6. Mandelzweig L, Battler A, Boyko V, Bueno H, Danchin N, Filippatos G, et al. The second Euro Heart Survey on acute coronary syndromes: Characteristics, treatment, and outcome of patients with ACS in Europe and the Mediterranean Basin in 2004. Eur Heart J 2006;27(19):2285-93. 7. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 8. Freund KM, Jacobs AK, Pechacek JA, White HF, Ash AS. Disparities by race, ethnicity, and sex in treating acute coronary syndromes. J Womens Health (Larchmt) 2012;21(2):126-32. 9. Iribarren C, Tolstykh I, Somkin CP, Ackerson LM, Brown TT, Scheffler R, et al. Sex and racial/ethnic disparities in outcomes after acute myocardial infarction: a cohort study among members of a large integrated health care delivery system in northern California. Arch Intern Med 2005;165(18):2105-13. 10. Thomas KL, Honeycutt E, Shaw LK, Peterson ED. Racial differences in long-term survival among patients with coronary artery disease. Am Heart J 2010;160(4):744-51. 11. Ting HH, Roe MT, Gersh BJ, Spertus JA, Rumsfeld JS, Ou FS, et al. Factors associated with off-label use of drug-eluting stents in patients with ST-elevation myocardial infarction. Am J Cardiol 2008;101(3):286-92. 12. US Census Bureau. Income, Poverty and Health Insurance Coverage in the United States: 2010. 13. Feder G, Crook AM, Magee P, Banerjee S, Timmis AD, Hemingway H. Ethnic differences in invasive management of coronary disease: prospective cohort study of patients undergoing angiography. BMJ 2002;324(7336):511-6. 14. Mindell J, Klodawski E, Fitzpatrick J. Using routine data to measure ethnic differentials in access to coronary revascularization. J Public Health (Oxf) 2008;30(1):45-53. 15. Trevelyan J, Needham EW, Halim M, Singh H, Been M, Shiu MF, et al. Evaluation of patient characteristics and utilisation of invasive cardiac procedures in a UK ethnic population with unstable angina pectoris. Int J Cardiol 2001;77(2-3):275-80. 16. Bansal N, Fischbacher CM, Bhopal RS, Brown H, Steiner MF, Capewell S. Myocardial infarction incidence and survival by ethnic group: Scottish Health and Ethnicity Linkage retrospective cohort study. BMJ Open 2013;3(9):e003415. 17. Smeets HM, de Wit NJ, Hoes AW. Routine health insurance data for scientific research: potential and limitations of the Agis Health Database. J Clin Epidemiol 2011;64(4):424-30. 18. Boelman L, Smeets HM, Knol MJ, Braam AW, Geerlings MI, de Wit NJ. Psychotropic drug use in patients with various chronic somatic diseases. Eur J Psychiat 2012;26(4):236-47.

202

Ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction

19. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69. 20. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7. 21. Ament P, Kessels W. Regionaal Inkomensonderzoek: uitgebreide onderzoeksbeschrijving. Voorburg: Centraal Bureau voor de Statistiek (CBS) 2008. 22. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57(12):1288-94. 23. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. A critical review of available methods. J Clin Epidemiol 2003;56(3):221-9. 24. Cram P, Bayman L, Popescu I, Vaughan-Sarrazin MS. Racial disparities in revascularization rates among patients with similar insurance coverage. J Natl Med Assoc 2009;101(11):1132-9. 25. Wijns W, Kolh P, Danchin N, Di MC, Falk V, Folliguet T, et al. Guidelines on myocardial revascularization. Eur Heart J 2010;31(20):2501-55. 26. Echols MR, Mahaffey KW, Banerjee A, Pieper KS, Stebbins A, Lansky A, et al. Racial differences among high-risk patients presenting with non-ST-segment elevation acute coronary syndromes (results from the SYNERGY trial). Am J Cardiol 2007;99(3):315-21. 27. Peniston RL, Lu DY, Papademetriou V, Fletcher RD. Severity of coronary artery disease in black and white male veterans and likelihood of revascularization. Am Heart J 2000;139(5):840-7. 28. Whittle J, Kressin NR, Peterson ED, Orner MB, Glickman M, Mazzella M, et al. Racial differences in prevalence of coronary obstructions among men with positive nuclear imaging studies. J Am Coll Cardiol 2006;47(10):2034-41. 29. Sedlis SP, Fisher VJ, Tice D, Esposito R, Madmon L, Steinberg EH. Racial differences in performance of invasive cardiac procedures in a Department of Veterans Affairs Medical Center. J Clin Epidemiol 1997;50(8):899-901. 30. Kendall H, Marley A, Patel JV, Khan JM, Blann AD, Lip GY, et al. Hospital delay in South Asian patients with acute ST-elevation myocardial infarction in the UK. Eur J Prev Cardiol 2013;20(5):737-42. 31. De Luca G, Suryapranata H, Ottervanger JP, Antman EM. Time delay to treatment and mortality in primary angioplasty for acute myocardial infarction: every minute of delay counts. Circulation 2004;109(10):1223-5.

203

CHAPTER 5.3

Ethnic inequalities in cardiovascular drug dispense and quitting rates after a first acute myocardial infarction

van Oeffelen AAM, Agyemang C, Stronks K, Bots ML, Vaartjes I. Ethnic inequalities in cardiovascular drug dispense and quitting rates after a first acute myocardial infarction: an analysis based on health insurance data.

Manuscript draft. Chapter 5.3

ABSTRACT Objective The aim of our study was to investigate ethnic inequalities in cardiovascular (CVD) drug dispense and quitting rates after a first acute myocardial infarction (AMI) event.

Methods Based on health insurance data from 2006 to 2010, 1,885 first AMI patients were included (1,268 ethnic Dutch, 161 Hindustani Surinamese, 31 non-Hindustani Surinamese, 189 Moroccan, and 236 Turkish migrants). Cox proportional hazard regression analyses were used to investigate ethnic differences in CVD drug dispenses (blood pressure lowering drugs, lipid lowering drugs, and antithrombotic drugs) within 6 months after the AMI event, and to investigate ethnic differences in cardiovascular drug quitting rates until 31 December 2011, among those on cardiovascular drug therapy.

Results There were no substantial ethnic inequalities in cardiovascular drug dispense within 6 months after the AMI event. However, over time migrants were more likely to quit their drug therapy than ethnic Dutch. This was especially the case among the Moroccan (blood pressure lowering drugs: Hazard Ratio: 2.06, 95% confidence interval (CI):1.38-3.06; lipid lowering drugs: HR: 1.34, 95% CI: 0.86-2.10; antithrombotic drugs: HR: 1.97, 95% CI: 1.26-3.06) and Turkish AMI patients (blood pressure lowering drugs: HR: 1.94, 95% CI: 1.38-2.73; lipid lowering drugs: HR: 1.26, 95% CI: 0.88-1.82; antithrombotic drugs: HR: 1.78, 95% CI: 1.18-2.67).

Conclusion Our data suggest that migrants are just as likely as ethnic Dutch to collect cardiovascular drugs at the pharmacy within 6 months after an AMI event, but over time are more likely to quit their therapy.

206

Ethnic inequalities in cardiovascular drug dispense and quitting rates after a first acute myocardial infarction

INTRODUCTION

After patients have been diagnosed with an acute myocardial infarction (AMI), Dutch guidelines recommend the prescription of a range of cardiovascular drugs, mainly blood pressure lowering drugs, lipid lowering drugs, and antithrombotic drugs.1 To be effective, it is important that those drugs are taken as prescribed. Most patients have to use them during the rest of their life. Underuse or too early termination of cardiovascular drugs increases risk on a recurrent event and mortality.2 Previous studies reported ethnic inequalities in cardiovascular drug use and quitting rates after a CVD event.3-6 Although results are not fully consistent, they often point towards a lower drug dispense and lower adherence after an AMI among ethnic minority groups. Underlying factors for this adverse pattern may include language barriers, cultural barriers, and (beliefs about) more adverse side effects.2;7;8 Previously, our group reported ethnic inequalities in prognosis after cardiovascular disease (CVD) in the Netherlands.9 Lower cardiovascular drug dispenses and higher quitting rates among ethnic minorities may partially explain these adverse findings, but as far as we are aware this has not been studied yet. Therefore, the first aim of our study was to explore differences in blood pressure lowering, lipid lowering and antithrombotic drug dispense after a first AMI event between first generation ethnic minority groups (henceforth, migrants) and the ethnic Dutch population. The second aim was to explore differences in cardiovascular drug quitting rates between migrants and ethnic Dutch.

METHODS

Data sources Data were extracted from the Achmea Health Database (AHD). The AHD records payments for the provision of all medical care and dispensed drugs to patients insured with the Achmea health insurance company, the main health insurance company in the central part of the Netherlands. Detailed information about the AHD is described elsewhere.10 We received data from all insured Hindustani-Surinamese, Moroccan, and Turkish ethnic minority groups (n=162,484), whose ethnicity was based on nationality and surnames as described previously.11 Furthermore, we received a representative sample of the remaining group (n=194,993). Approval from the AHD research committee was obtained prior to accessing the data. For the present study, only those insured with Achmea between 1 July 2006 and 30 June 2010 were included to enable linkage with Dutch nationwide registers (Hospital Discharge Register (HDR), Population Register (PR) and Regional Income Survey (RIS)). Subsequently, only those with a first registered AMI event in the AHD, and who did not have a previous AMI hospitalisation in the HDR from 1995 onwards, were selected (n=2,057).

207

Chapter 5.3

Determinants Ethnic background was based on the country of birth and the county of birth of the parents, obtained from the PR. A person belonged to a migrant group if he/she was born abroad and at least one of the parents was born abroad.12 Doing so, migrants from Suriname (n=192), Morocco (n=189), and Turkey (n=236) were identified. Subsequently, Hindustani Surinamese migrants were identified from the total Surinamese group using Hindustani surnames available from the AHD and the Amsterdam SUNSET study.11;13 The remaining Surinamese were categorised as Non-Hindustani Surinamese who are predominantly of West-African origin. A person was ethnic Dutch when both parents were born in the Netherlands (n=1,268). Socioeconomic status (SES) was based on disposable income on household level registered in the RIS, available for one third of the Dutch population.14 Within each neighborhood, the mean disposable income on household level in 2006 was calculated and subsequently assigned to all residents living in that neighborhood. Neighbourhood income was divided into SES tertiles, with the first tertile representing the lowest income group. Presence and extent of comorbidity were determined with the Charlson index score,15 based on discharge diagnosis in the Hospital Discharge Register from 1995 to 2010. The Charlson index score has been proven to be a reliable and valid method to measure comorbidity in clinical research.16

Outcomes Blood pressure lowering drug dispense was defined as collecting at least one prescription of beta- blockers, ACE-inhibitors, angiotensin-II-antagonists, calcium-channel blockers or diuretics at the pharmacy within 6 months after the AMI event. Lipid lowering drug dispense was defined as collecting at least one prescription of statins or other lipid lowering drugs at the pharmacy within 6 months after the AMI event. Antithrombotic drug dispense was defined as collecting at least one prescription of vitamin K antagonists, heparin, thienopyridines, or aspirin at the pharmacy within 6 months after the AMI event. Drugs were identified using the Anatomical Therapeutic Chemical Classification System (ATC codes), as presented in Table 1. Persons were followed from the date of the AMI event to the date of the first drug dispense, and were censored in case of death or the end of Achmea insurance. Among those who collected their blood pressure lowering drugs (n=1,869), lipid lowering drugs (n=1,743), or antithrombotic drugs (n=1,881) at the pharmacy within 6 months after the AMI event, quitting rates were investigated. When the number of defined daily doses (DDD) collected from the pharmacy plus an extra period of 90 days expired before a new prescription was collected, we assumed the patient quitted treatment.17 The DDD is the average maintenance dose per day for a drug used for its main indication in adults. Persons were followed from the date of first drug

208

Ethnic inequalities in cardiovascular drug dispense and quitting rates after a first acute myocardial infarction dispense after AMI to the date of quitting treatment, and were censored in case of death, end of Achmea insurance, or the end of the study period on 31 December 2011, whichever came first.

Table 1 Cardiovascular drugs and accompanying Anatomical Therapeutic Chemical (ATC) codes Type of drug Drug category ATC classification Blood pressure lowering drugs Diuretics C03A, C03B, C03C Beta-blockers C07 Calcium-channel blockers C08 ACE-inhibitors C09A, C09B Angiotensin-II-antagonists C09C, C09D

Lipid lowering drugs Statins C10AA Other lipid lowering drugs C10AB, C10AC, C10AD, C10AX

Antithrombotic drugs Vitamin K antagonists B01AA Heparin B01AB Thienopyridines B01AC04/05/22/24 Aspirin B01AC06/07/08 N02BA01/15

Data analysis Cox proportional hazard regression analyses were used to calculate ethnic inequalities in blood pressure lowering drug dispense, lipid lowering drug dispense, and antithrombotic drug dispense within 6 months after the AMI event, and to calculate ethnic inequalities in quitting rates of these drugs. Three models were created: model 1 was adjusted for age, sex, degree of urbanisation, and marital status; model 2 was additionally adjusted for comorbidity; model 3 was additionally adjusted for comorbidity and neighbourhood SES. We used SPSS software, version 20.0 (SPSS Inc, Chicago, Illinois, USA). All analyses were performed in accordance with privacy legislation Netherlands.

RESULTS

Patient characteristics Compared with the ethnic Dutch population, migrant groups were more often men, younger, and lived in more urbanised low SES areas (Table 2). Comorbidity was about equal between migrants and ethnic Dutch, except among Moroccans who had substantially less comorbidity (21.2% had a comorbidity compared with 77.1% in the ethnic Dutch group). Among ethnic Dutch, 88.2% collected a blood pressure lowering drug prescription, 79.9% a lipid lowering drug prescription, and 89.0% an antithrombotic drug prescription within 6 months after the AMI event. Among migrant groups these percentages were somewhat higher.

209

Chapter 5.3

66)

-

e - 6.8 6.8 236 78.0 58.9 18.6 15.7 75.4 73.3 70.8 19.1 10.2 Turkish

53 (45

70)

-

e e - - 7.9 7.9 189 82.0 79.4 12.2 85.2 21.2 66.1 20.6 13.2 Moroccan 66 (56

66)

-

e e e e e e ------31 74.2 77.4 29.0 67.7 67.7 Hindustani - 59 (49

Surinamese Non

64)

-

e e - - ten 9.9 9.9 9.9 161 11.2 11.2 71.4 77.0 31.7 72.7 57.1 32.9 between 1 July 2006 and 30 June 2010 June 30 2006 and 1 July between

56 (47 Hindustani Hindustani Surinamese

65)

-

e e - - 9.9 9.9 192 71.9 77.1 10.9 10.4 31.2 71.9 58.9 31.2 level

56 (47

Total Surinamese Total

in the Achmea Health Database Health Achmea in the

): very urban=>2000, urban=1001-2000, urban/rural=501-1000, rural=251-500, very rural=<251 rural=<251 very rural=251-500, urban/rural=501-1000, urban=1001-2000, urban=>2000, very ): 2

80)

-

7.6 7.6 56.2 26.9 21.4 24.5 19.6 53.9 77.1 36.1 32.6 31.3 1,268 71 (59 Dutch Ethnic

b

d

Rural

c Urban Urban

)

Very rural a Very urban Urban/rural Urban/rural

rbanisation % rbanisation % 0 > ndex

Patient characteristics of patients of first AMI characteristics Patient

Tertile 1(lowest income) Tertile 3(highest income) ot given in line with the Dutch data protection guideline as the cases were less were than less the as cases guideline data the protection Dutch line with in given ot Tertile 2(medium income) patients AMI Male % (IQR Median age u of Degree % together Married/living i Charlson SES Neighbourhood % Socioeconomic status basedon income on neighbourhood Interquartile range Population density (number of residents per km per of residents (number density Population N

Charlson comorbidity index > 0: at least one hospitalisation for a diagnosis included in the Charlson comorbidity index from 1995 until the STEMI event the index from 1995 until comorbidity Charlson the in included diagnosis a for hospitalisation one least at 0: > index comorbidity Charlson d 2 Table a b c e

210

Ethnic inequalities in cardiovascular drug dispense and quitting rates after a first acute myocardial infarction

Cardiovascular drug dispense After adjustment for age, sex, degree of urbanisation, and marital status no significant differences in cardiovascular drug dispense were observed between migrants and the ethnic Dutch population (Table 3). After additional adjustment for comorbidity, Moroccans had a statistically significantly lower cardiovascular drug dispense compared with their ethnic Dutch counterparts. Further adjustment for neighbourhood SES did not alter results.

Table 3 Difference between migrants and ethnic Dutch of ≥30 years of age in cardiovascular drug dispense 6 months after the first AMI event N (%) HR (95% CI) Model 1d Model 2e Model 3f Blood pressure lowering drugsa Ethnic Dutch 1,119 (88.2) 1.00 1.00 1.00 Total Surinamese 176 (91.7) 1.14 (0.95-1.36) 1.14 (0.95-1.36) 1.15 (0.96-1.38) Hindustani Surinamese 148 (91.9) 1.14 (0.94-1.38) 1.14 (0.94-1.38) 1.14 (0.94-1.39) Non-Hindustani Surinamese 28 (90.3) 1.10 (0.75-1.62) 1.09 (0.74-1.60) 1.09 (0.74-1.60) Moroccans 173 (91.5) 0.91 (0.76-1.09) 0.79 (0.66-0.96)* 0.79 (0.66-0.96)* Turkish 225 (95.3) 1.02 (0.87-1.20) 1.00 (0.85-1.17) 0.99 (0.84-1.17) Lipid lowering drugsb Ethnic Dutch 1,013 (79.9) 1.00 1.00 1.00 Total Surinamese 172 (89.6) 1.09 (0.91-1.31) 1.08 (0.90-1.29) 1.09 (0.90-1.31) Hindustani Surinamese 144 (89.4) 1.07 (0.88-1.30) 1.06 (0.87-1.29) 1.07 (0.88-1.30) Non-Hindustani Surinamese 28 (90.3) 1.11 (0.75-1.63) 1.10 (0.75-1.61) 1.11 (0.75-1.64) Moroccans 166 (87.8) 0.92 (0.77-1.11) 0.80 (0.66-0.97)* 0.81 (0.67-0.98)* Turkish 220 (93.2) 1.00 (0.85-1.18) 0.98 (0.84-1.16) 0.99 (0.84-1.17) Anti-thrombotic drugsc Ethnic Dutch 1,128 (89.0) 1.00 1.00 1.00 Total Surinamese 176 (91.7) 1.04 (0.87-1.25) 1.04 (0.87-1.25) 1.05 (0.88-1.26) Hindustani Surinamese 147 (91.3) 1.02 (0.84-1.23) 1.02 (0.84-1.23) 1.02 (0.84-1.24) Non-Hindustani Surinamese 29 (93.5) 1.14 (0.78-1.66) 1.11 (0.76-1.62) 1.12 (0.77-1.65) Moroccans 173 (91.5) 0.92 (0.77-1.10) 0.81 (0.67-0.97)* 0.82 (0.67-0.99)* Turkish 228 (96.6) 1.02 (0.87-1.19) 0.99 (0.85-1.16) 1.00 (0.85-1.17) a Beta blokers, ACE-inhibitors, angiotensin-II-antagonists, calcium channel blockers, diuretics b Statins and other lipid lowering drugs c Vitamin K antagonists, heparin, thienopyridines, aspirin d Adjusted for age, sex, degree of urbanisation, and marital status e Adjusted for age, sex, degree of urbanisation, marital status, and Charlson comorbidity index f Adjusted for age, sex, degree of urbanisation, marital status, Charlson comorbidity index, and SES Censoring in case of death, end of Achmea insurance, or end of study period at 31 December 2011

Cardiovascular drug quitting rates Among ethnic Dutch, 26.2% of those who dispensed a blood pressure lowering drug prescription at a pharmacy within 6 months after the AMI event stopped collecting their blood pressure lowering drugs during the study period (Table 4). For lipid lowering drugs and antithrombotic drugs, these

211

Chapter 5.3 percentages were 23.8% and 24.9% respectively. Among migrants, these percentages were markedly higher than among ethnic Dutch. After adjustment for age, sex, degree of urbanisation, and marital status, migrant groups often had a higher quitting rate than ethnic Dutch. Statistically significant higher quitting rates were reported for blood pressure lowering drugs and antithrombotic drugs in the Moroccan and Turkish groups. After adjustment for comorbidity, the higher quitting rate for blood pressure lowering drugs among Moroccans reduced to some extent, but the significant difference remained. This higher rate among Moroccans was therefore partially explained by their very low prevalence of comorbidity. Comorbidity did not affect results in the other groups. Further adjustment for neighbourhood SES did not influence results at all. Importantly, the quitting rate for statins was higher among all migrant groups compared with ethnic Dutch, although the differences were not statistically significant.

Table 4 Difference between migrant and ethnic Dutch AMI patients ≥ 30 years of age in cardiovascular drug quitting rates N (%) HR (95% CI) Model 1d Model 2e Model 3f Blood pressure lowering drugsa Ethnic Dutch 293 (26.2) 1.00 1.00 1.00 Total Surinamese 68 (38.6) 1.10 (0.74-1.65) 1.11 (0.74-1.66) 1.13 (0.75-1.70) Hindustani Surinamese 58 (39.2) 1.08 (0.70-1.66) 1.08 (0.70-1.66) 1.09 (0.70-1.68) Non-Hindustani Surinamese 10 (35.7) 1.28 (0.56-2.94) 1.34 (0.58-3.08) 1.36 (0.59-3.14) Moroccans 82 (47.4) 2.06 (1.38-3.06)* 1.67 (1.10-2.54)* 1.71 (1.12-2.61)* Turkish 113 (50.2) 1.94 (1.38-2.73)* 1.93 (1.37-2.70)* 1.93 (1.36-2.73)* Lipid lowering drugsb Ethnic Dutch 241 (23.8) 1.00 1.00 1.00 Total Surinamese 62 (36.0) 1.28 (0.86-1.91) 1.28 (0.86-1.91) 1.26 (0.84-1.88) Hindustani Surinamese 48 (33.3) 1.25 (0.82-1.91) 1.25 (0.82-1.91) 1.22 (0.80-1.87) Non-Hindustani Surinamese 14 (50.0) 1.58 (0.68-3.64) 1.58 (0.69-3.65) 1.52 (0.66-3.51) Moroccans 53 (31.9) 1.34 (0.86-2.10) 1.27 (0.79-2.01) 1.21 (0.75-1.93) Turkish 77 (35.0) 1.26 (0.88-1.82) 1.27 (0.88-1.83) 1.21 (0.83-1.77) Anti-thrombotic drugsc Ethnic Dutch 281 (24.9) 1.00 1.00 1.00 Total Surinamese 54 (30.7) 1.41 (0.91-2.19) 1.41 (0.90-2.19) 1.44 (0.92-2.25) Hindustani Surinamese 44 (29.9) 1.48 (0.93-2.35) 1.48 (0.93-2.35) 1.53 (0.96-2.44) Non-Hindustani Surinamese 10 (34.5) 1.17 (0.42-3.25) 1.17 (0.42-3.25) 1.18 (0.43-3.28) Moroccans 64 (37.0) 1.97 (1.26-3.06)* 1.95 (1.23-3.09)* 1.95 (1.22-3.12)* Turkish 80 (35.1) 1.78 (1.18-2.67)* 1.78 (1.18-2.67)* 1.79 (1.18-2.72)* a Beta blokers, ACE-inhibitors, angiotensin-II-antagonists, calcium channel blockers, diuretics b Statins and other lipid lowering drugs c Vitamin K antagonists, heparin, thienopyridines, aspirin d Adjusted for age, sex, degree of urbanisation, and marital status e Adjusted for age, sex, degree of urbanisation, marital status, and Charlson comorbidity index f Adjusted for age, sex, degree of urbanisation, marital status, Charlson comorbidity index, and SES Censoring in case of death, end of Achmea insurance, or end of study period at 31 December 2011

212

Ethnic inequalities in cardiovascular drug dispense and quitting rates after a first acute myocardial infarction

DISCUSSION Our study, using data from a large health insurance company between 2006 and 2011, showed that about 90% of persons who had suffered a first acute myocardial infarction, collected a prescription for cardiovascular drugs (blood pressure lowering drugs, lipid lowering drugs, and antithrombotic drugs) within 6 months after their event. Migrants were just as likely as their ethnic Dutch counterparts to collect these drugs. Among those who had collected at least one prescription, 25% to 50% quitted their drug therapy during the study period. Migrants were more likely to quit than ethnic Dutch. This was especially evident among Turkish and Moroccan migrants for blood pressure lowering and antithrombotic drugs, and to a lesser extent among Surinamese migrants. The higher mortality after a first CVD event among ethnic minority groups that our group previously reported, may therefore be partially explained by their higher cardiovascular drug quitting rates. These findings are of relevance for care givers (e.g. general practitioners and cardiologists), as they may develop targeted approaches to improve drug continuation in all AMI patients, especially migrant groups. Further research into the specific barriers among migrants (e.g. cultural barriers, language barriers) that may inhibit them to adhere to prescribed drug therapy is mandatory and will enhance the understanding of the problem and help in finding solutions.

213

Chapter 5.3

REFERENCES 1. www.nhg.org/standaarden/samenvatting/cardiovasculair-risicomanagement.nl. 2. Kronish IM, Ye S. Adherence to cardiovascular medications: lessons learned and future directions. Prog Cardiovasc Dis 2013;55(6):590-600. 3. Hempler NF, Diderichsen F, Larsen FB, Ladelund S, Jorgensen T. Do immigrants from Turkey, Pakistan and Yugoslavia receive adequate medical treatment with beta-blockers and statins after acute myocardial infarction compared with Danish-born residents? A register-based follow-up study. Eur J Clin Pharmacol 2010;66(7):735-42. 4. Lai EJ, Grubisic M, Palepu A, Quan H, King KM, Khan NA. Cardiac medication prescribing and adherence after acute myocardial infarction in Chinese and South Asian Canadian patients. BMC Cardiovasc Disord 2011;11:56. 5. Ringback WG, Ericsson O, Lofroth E, Rosen M. Equal access to treatment? Population-based follow-up of drugs dispensed to patients after acute myocardial infarction in Sweden. Eur J Clin Pharmacol 2008;64(4):417-24. 6. Zhang Y, Baik SH, Chang CC, Kaplan CM, Lave JR. Disability, race/ethnicity, and medication adherence among Medicare myocardial infarction survivors. Am Heart J 2012;164(3):425-33. 7. Gazmararian JA, Kripalani S, Miller MJ, Echt KV, Ren J, Rask K. Factors associated with medication refill adherence in cardiovascular-related diseases: a focus on health literacy. J Gen Intern Med 2006;21(12):1215-21. 8. Kaplan RC, Bhalodkar NC, Brown EJ, Jr., White J, Brown DL. Race, ethnicity, and sociocultural characteristics predict noncompliance with lipid-lowering medications. Prev Med 2004;39(6):1249-55. 9. Agyemang C, Vaartjes I, Bots ML, van Valkengoed I, de Munter JS, de Bruin A, et al. Risk of death after first admission for cardiovascular diseases by country of birth in The Netherlands: a nationwide record-linked retrospective cohort study. Heart 2009;95(9):747-53. 10. Smeets HM, de Wit NJ, Hoes AW. Routine health insurance data for scientific research: potential and limitations of the Agis Health Database. J Clin Epidemiol 2011;64(4):424-30. 11. Boelman L, Smeets HM, Knol MJ, Braam AW, Geerlings MI, de Wit NJ. Psychotropic drug use in patients with various chronic somatic diseases. Eur J Psychiat 2012;26(4):236-47. 12. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69. 13. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7. 14. Ament P, Kessels W. Regionaal Inkomensonderzoek: uitgebreide onderzoeksbeschrijving. Voorburg: Centraal Bureau voor de Statistiek (CBS) 2008. 15. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57(12):1288-94. 16. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56(3):221-9. 17. Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiol Drug Saf 2006;15(8):565-74.

214

CHAPTER 6

GENERAL DISCUSSION

General discussion

In this thesis, inequalities in incidence and prognosis of cardiovascular disease (mainly acute myocardial infarction (AMI) and stroke) and health care use between ethnic minority groups and the ethnic Dutch population were investigated by using several nationwide registers provided by Statistics Netherlands, and data provided by the Achmea health insurance company. Evidence on ethnic inequalities in cardiovascular disease (CVD) was produced for several stages in the cardiovascular pathway (Figure 1). The cardiovascular pathway describes the process from the development of cardiovascular risk factors towards a cardiovascular event (incidence), and subsequently towards stabilisation, a recurrent event, or death (prognosis). If numbers allowed, results were generated for Surinamese (mainly from South-Asian and West-African origin), Moroccan, Turkish, Antillean, Indonesian, Chinese, and South-Asian ethnic minority groups. Furthermore, the role of socioeconomic status (SES) on several outcomes was examined, since SES and ethnic background are closely related.1 It has been proposed that ethnic inequalities in CVD are to some extent socioeconomic in nature, as ethnic minority groups have in general a lower SES than the ‘majority population’ in the host country, and low SES is related to a higher cardiovascular disease risk.2-4

KEY FINDINGS The key findings of this thesis are described in the section below and presented in table 1.

Ethnic inequalities in the incidence of cardiovascular disease Acute myocardial infarction - There were clear differences in AMI incidence between ethnic minority groups and the ethnic Dutch population. Lower rates were observed among Moroccans and Chinese, whereas higher rates were observed among Surinamese, Turkish, and Indonesian men, and South Asians. - Ethnic inequalities seen in first generation ethnic minorities often converged towards the ethnic Dutch population in second generation ethnic minorities. - The sex disparity in AMI incidence was more pronounced among non-Western ethnic minority groups than among ethnic Dutch. - The higher AMI incidence in Surinamese and Turkish ethnic minorities decreased with age and became absent in those older than 70 years of age. - There was a significant decline in AMI incidence among all ethnic minority groups except among Turkish women and Moroccans. In most recent data, Moroccans even showed an increase in AMI incidence over time. - Incidence of AMI was higher among medium- and low-income groups than among high- income groups regardless of ethnic background.

217

Chapter 6

Stroke - There were clear differences in stroke incidence between ethnic minority groups and ethnic Dutch. For all stroke subtypes, lower rates were observed among Moroccans, whereas higher rates were observed among Surinamese. Among other ethnic minority groups, ethnic inequalities greatly depended on stroke subtype and/or sex. - Ethnic inequalities in total stroke incidence remained stable independent of socioeconomic group. - Incidence of total stroke was higher among medium- and low-income groups than among high-income groups regardless of ethnic background.

Ethnic and socioeconomic inequalities in prognosis after cardiovascular disease - The majority of ethnic minority groups had a higher mortality and readmission rate after a first AMI hospitalisation than ethnic Dutch. - Ethnic inequalities in prognosis after a first CHF hospitalisation were more diverse and depended on country of birth. - AMI patients belonging to the lowest socioeconomic group had a higher short-term mortality compared with AMI patients belonging to the highest socioeconomic group.

Ethnic inequalities in cardiovascular health care use - In the primary care setting, most ethnic minority groups on cardiovascular drug therapy had higher quitting rates than their ethnic Dutch counterparts. - There were no ethnic inequalities in revascularisation rate after an ST-elevation myocardial infarction. - After a first AMI, ethnic minority groups were just as likely to collect cardiovascular drugs at the pharmacy as their ethnic Dutch counterparts, but they were more likely to quit their drug therapy over time.

METHODOLOGICAL CONSIDERATIONS Nationwide registers The studies concerning ethnic inequalities in CVD incidence and prognosis presented in this thesis (Chapter 2, 3, and 4) were conducted by linking several nationwide registers, which were available from 1995 until 2007/2010. The Hospital Discharge Register (HDR) was used to identify hospital admissions for cardiovascular events and to identify diagnosis included in the Charlson comorbidity index (measure for comorbidity). The Population Register (PR) was used to obtain demographic data (such as birth date, sex, country of birth, parental country of birth, degree of urbanisation and marital status). The Cause of Death Register (CDR) was used to identify out-of-hospital

218

General discussion cardiovascular deaths. Finally, the Regional Income Survey (RIS) was used to identify income as a proxy for SES.

Table 1 Differences in cardiovascular incidence, prognosis, and health care use between ethnic minority groups and the ethnic Dutch population Surinamese Moroccan Turkish Antillean Indonesian Chinese South Asian INCIDENCE OF CVD AMI ↑ ↓ M↑/W- = = ↓ ↑ Total stroke ↑ ↓ = M↑/W- = ↓ NA ICH ↑ ↓ ↑ M↑/W- ↑ ↑ NA SAH ↑ ↓ M-/W↓ M-/W↑ M- /W↑ ↓ NA IS ↑ ↓ M-/W↓ M↑/W- = ↓ NA PROGNOSIS AFTER AMI Short-term mortality ↑ ↓ = ↑ ↑ ↑ ↑ Long-term mortality ↑ ↑ ↑ ↑ = ↑ ↓ AMI readmission ↑ ↑ ↑ ↑ ↑ ↑ ↑ CHF readmission ↑ ↑ ↑ ↑ = ↑ = PROGNOSIS AFTER CHF Short-term mortality = ↓ ↓ ↓ = ↑ ↑ Long-term mortality ↑ ↓ = = = ↑ ↑ CHF readmission ↑ ↑ ↑ = = ↑ ↑ CVD HEALTH CARE USE IN PRIMARY CARE Quitting BPL drugs ↑ ↑ ↑ ↑ = ↑ ↑ Quitting LL drugs ↑ ↑ ↑ ↑ = ↑ ↑ CVD HEALTH CARE USE AFTER AMI Revasc. STEMI = = = NA NA NA NA BPL drug dispense ↑ = = NA NA NA NA LL drug dispense = = = NA NA NA NA AT drug dispense = = = NA NA NA NA Quitting BPL drugs ↑ ↑ ↑ NA NA NA NA Quitting LL drugs ↑ ↑ ↑ NA NA NA NA Quitting AT drugs ↑ ↑ ↑ NA NA NA NA ↑ Higher compared with ethnic Dutch (Hazard Ratio/Odds Ratio: >1.10) = Similar compared with ethnic Dutch (Hazard Ratio/Odds Ratio: 0.90-1.10) ↓Lower compared with ethnic Dutch (Hazard Ratio/Odds Ratio: <0.90) NA: Not assessed in this thesis M: Men; W: Women AMI: acute myocardial infarction; ICH: intracerebral haemorrhage; SAH: subarachnoid haemorrhage; IS: ischemic stroke; CHF: congestive heart failure; BPL drugs: blood pressure lowering drugs; LL drugs: lipid lowering drugs; AT drugs: antithrombotic drugs; revasc.: revascularisation procedure.

Strengths Nationwide registers are an inexpensive and easily accessible source of information. The nationwide coverage can deliver valuable information regarding all kinds of subpopulations. With respect to ethnic health research, nationwide registers have a number of advantages. Although about 20% of the Dutch population belongs to an ethnic minority group, individual groups are still small. The nationwide coverage provides the opportunity to study individual ethnic minority groups with sufficient power. Inclusion of ethnic minorities in cohort studies is often challenging and results in

219

Chapter 6 inadequate number of persons, partially due to the relatively low participation rate.5 By using registers, we were not only able to distinguish the individual ethnic minority groups, but also to stratify results by age, sex, disease subtype, and/or generational status. Furthermore, the risk on selection bias was greatly reduced by using nationwide registers. In traditional cohort studies, non- response associated with factors relevant to the study outcomes (for example due to language barriers), and oversampling of ethnic minorities to assure sufficient power, may induce selection bias.6 Although many investigators see ICD-coding of causes for hospitalisation and death as inferior, there is sufficient evidence to suggest that the coding in the Netherlands with respect to CVD is at high level. Positive predictive values (indicating the percentage of cases that was correctly registered) of registration in the Dutch HDR have shown to be 97% for AMI, 95% for subarachnoid haemorrhage, and 80% for congestive heart failure.7;8 Reliability of cause of death registration in the Dutch CDR was also high (89% for AMI, 79% for cerebrovascular disease and 76% for heart failure).9

Limitations A major limitation of nationwide registers is the lack of information concerning cardiovascular risk factors, disease severity and social determinants, which could have given insight in the underlying mechanisms of found ethnic inequalities. In this thesis, we could only correct for the explanatory variables ‘comorbidity’ and ‘neighbourhood SES’, since these variables were available for all persons in the nationwide registers. In all studies of this thesis (except for Chapter 2.4) analyses were corrected for comorbidity. Presence and extent of comorbidity were determined with the Charlson comorbidity index score,10 which is based on 17 discharge diagnosis in the HDR. The Charlson comorbidity index score has proven to be a reliable and valid method to measure comorbidity in clinical research.11 However, it does not incorporate comorbidity that did not lead to a hospitalisation, such as diabetes and hypertension. Since these conditions are often more common among ethnic minorities, comorbidity may have been somewhat underestimated among ethnic minorities resulting in a suboptimal adjustment for comorbidity.12;13 In all studies of this thesis, except for Chapter 2.4, 2.5, 3.2 and 4.2, analyses were corrected for neighbourhood SES, based on income data provided by tax authorities registered in the RIS.14 The RIS started in 1994, when a representative sample of 1.9 million Dutch citizens was selected. For all persons belonging to the households of the sample population (about one third of the Dutch population) household income was available. To enable inclusion of all residents, neighbourhood income was used as SES indicator. It was calculated by assigning the mean of all registered household incomes within a neighbourhood to all residents living in that neighbourhood. Neighborhood income may be a less reliable SES indicator than individual or household SES leading

220

General discussion to misclassification.15 On the other hand, previous research also shows that although individual/household SES is stronger related to health than area based indicators, these area based indicators are still useful in assessing inequalities in health.16 The studies on socioeconomic inequalities in CVD incidence among ethnic groups in this thesis (Chapter 2.5 and 3.2) were performed using income on household level as SES indicator. Individual/household income has proven to be a sensitive SES indicator for studying socioeconomic inequalities in CVD incidence.4 Ethnic background was based on the country of birth of the person and its parents according to the definition of Statistics Netherlands.18 The advantages are its objective and stable character allowing uniform identification of groups. Furthermore, it can be relatively easily applied and can distinguish first and second generation ethnic minority groups. On the other hand, country of birth is an unreliable measure of ethnicity when the population of the country of origin consists of different ethnic groups, such as the Surinamese (who are mainly from South-Asian or West-African descent). In some studies of this thesis we were able to extract the South-Asian Surinamese by using surname algorithms (Chapter 5). Unfortunately, our data did not allow us to capture the multidimensional character of ethnicity, since we had no information regarding geographic origin of the ancestors, language spoken, self-identification, culture and migration history.17;18 In the Netherlands, however, country of birth classification highly correlates with self-identification of ethnicity (often used in UK studies), especially among the elderly (>95% correlation).18

Achmea Health Database The studies concerning health care use presented in this thesis (Chapter 5) were conducted in the Achmea Health Database (AHD). The Achmea health insurance company is the main health insurance company in the central part of the Netherlands as it provides health care coverage for 1.3 million Dutch residents. The AHD records payments for the provision of all medical care to patients insured with the Achmea health insurance company.19 Therefore, it is a valuable source of health care data that can be used to explain ethnic inequalities in cardiovascular disease incidence and prognosis. The AHD gave us the opportunity to study ethnic inequalities in cardiovascular drug use and quitting rates by using Anatomical Therapeutic Chemical Classification System (ATC) codes. Furthermore, it enabled us to study ethnic inequalities in revascularisation rate after an AMI using Diagnosis Treatment Combinations (DTC). Registration of ATC and DTC codes in the AHD is extensively controlled for the reason of financial reimbursement, improving validity of results. For DTC codes, however, the validation encompasses a very rough check to secure that registered procedures are in agreement with the registered diagnosis. Misclassification in the AHD remains possible. Until now, there are no detailed validation studies available that have investigated the sensitivity and the positive predictive value of registered diagnosis and treatments in the AHD (such

221

Chapter 6 as revascularisation procedures). Since there are no indications of differential misclassification between ethnic groups, this unlikely influenced the results in this thesis.

REFLECTIONS ON THE FINDINGS IN THIS THESIS This thesis demonstrates clear ethnic inequalities in the incidence of and prognosis after cardiovascular disease. In this section we will reflect on possible causes of the observed inequalities, with a focus on the role of health care use and SES.

What is the role of primary health care use in ethnic inequalities in CVD incidence? This thesis showed that especially the Surinamese, Turkish, and South Asians were at higher CVD risk (AMI and stroke) than the ethnic Dutch population. On the other hand, Moroccan and Chinese groups were often at lower risk. To what extent can these ethnic inequalities in CVD incidence be attributed to inequalities in the use of primary health care, more specifically cardiovascular risk management? In this thesis we tried to answer this question by studying ethnic inequalities in cardiovascular drug dispense and quitting rates in the primary care setting. Results showed that among those without a previous cardiovascular event, cardiovascular drug dispense (blood pressure lowering drugs and lipid lowering drugs) was higher among most ethnic minority groups (except for Chinese) and was most pronounced in Surinamese and South Asians. Unfortunately, we did not have information on risk factor profiles and actual ‘need’ for cardiovascular drug use. Therefore, we could not draw any conclusions about ethnic inequities in cardiovascular drug dispense. However, previous studies show a worse CVD risk factor profile among most ethnic minority groups (Table 2), which suggests a higher need for cardiovascular drugs. The higher cardiovascular drug dispense found in our study is in line with these previous findings. Future research has to elucidate whether the higher cardiovascular drug dispense covers the higher need, and into what extent it explains ethnic inequalities in CVD incidence. This thesis also shows that among those on cardiovascular drug therapy, the majority of ethnic minority groups were more likely to quit than ethnic Dutch. Ethnic minority groups may encounter several barriers to adhere to medication and other primary preventive measures. For example, linguistic and cultural barriers could alter the GP-patient communication and lead to misunderstanding and misinterpretation of advice.31;32 There are indications that it is more difficult to reach mutual understanding during a GP consultation among ethnic minority groups than among ethnic Dutch. Consultations without mutual understanding more often result in non-compliance to lifestyle modifications and drug use.33 Furthermore, more side effects or beliefs about side-effects, lack of understanding, and difference in health beliefs (preference for alternative medicine) may all contribute to the higher cardiovascular drug quitting rates among ethnic minorities.34-36 Since higher quitting rates were not only present among those with a high CVD incidence (Surinamese, Turkish,

222

General discussion

South Asians), but also in Moroccans who have a very low CVD incidence, it is certainly not a sole explanation of ethnic inequalities in CVD incidence. Nevertheless, it may adversely affect CVD incidence among all ethnic minority groups. Future research has to elucidate to what extent the high CVD drug quitting rates explain the high CVD incidence among some ethnic minority groups.

Table 2 Distribution of risk factors in ethnic minority groups as compared with the majority population based on previous studies20-30 Moroccana Chineseb Turkisha Surinamesea South Asianc Diabetes ↑ NA ↑ ↑ ↑ Hypertension ↓ ↑ ↓ ↑ ↑ Hyperlipidemia ↓ ↓ ↓ ↓ ↑ Decreased HDL cholesterol ↑ = ↑ ↑ ↑ Obesity ↑ ↓ ↑ ↑ ↓d Smoking ↓ ↓ M↑/W- M↑/W↓ Me /W↓ Physical inactivity M-/W↑ NA ↑ ↑ ↑ a Compared with ethnic Dutch b Compared with the Canadian or UK population c Compared with mainly the Canadian or UK population d BMI is low, but waist-to-hip ratio is high, increasing cardiovascular risk e Depends on specific South-Asian population ↑ higher prevalence compared with majority population ↓ lower prevalence compared with majority population = No difference in prevalence compared with majority population

Other primary health care factors not studied in this thesis could have affected ethnic inequalities in CVD incidence. Previous research in the Netherlands reported a higher GP consultancy among ethnic minority groups than among ethnic Dutch, suggesting that ethnic minorities do not experience difficulties in finding primary care.37 However, difficulties in reaching mutual understanding during a GP consultation among ethnic minority groups may give rise to multiple visits for the same complaint and lack of compliance to lifestyle modification advice.33 Future research should elucidate whether the lack of mutual understanding during GP visits influences the effectiveness of cardiovascular risk management among ethnic minorities.

What is the role of socioeconomic status in ethnic inequalities in CVD incidence? The majority of ethnic minority groups investigated in this thesis had, on average, a lower socioeconomic status than the ethnic Dutch population. Since SES has repeatedly been related to a higher cardiovascular risk, we hypothesised that SES would be an important factor in our studies.2-4

Is SES a mediator in the relation between ethnicity and CVD incidence? When SES is a mediator, SES can explain ethnic inequalities in CVD incidence to some extent. In Chapter 2.1 to 2.4 and in Chapter 3.1 we adjusted our analyses for neighbourhood SES (based on

223

Chapter 6 income on neighbourhood level). Surprisingly, results did not change at all after adjusting for it, indicating that neighbourhood SES is no explanation of found ethnic inequalities in AMI and stroke incidence. This could be due to differential misclassification of our neighbourhood SES indicator, as suggested in our methodological considerations above. A SES indicator on individual or household level may reflect actual SES more accurately, and could have explained ethnic inequalities in CVD incidence into some extent. Some studies indeed found attenuated relations after correcting for several individual SES indicators (education, occupational status, income, housing conditions) whereas this was not observed among others.38-46 The absent mediating role of SES in these studies could be due to the observation that a low SES among ethnic minorities, in contrast to the general population, is not necessarily related to a higher disease risk.44-47

Is SES an effect modifier in the relation between ethnicity and CVD incidence? When SES is an effect modifier, the relation between ethnicity and CVD incidence differs between socioeconomic groups. In recent decades there has been a decline in CVD incidence. This decline has affected persons in high SES groups in high income countries first, as proposed by the diffusion theory.48 This has resulted in a SES gradient in CVD incidence, in which low SES groups have a higher CVD risk than high SES groups. Among ethnic minority groups originating from low income countries, this SES gradient may still be absent as reported by several studies in the 1970s, 80s, and 90s.44-47 The general expectation is that the SES gradient in CVD will eventually emerge in ethnic minority groups and converge towards the SES gradient seen in the host populations. When this occurs, ethnic inequalities in CVD incidence will become equal over all SES groups. In this thesis, we analysed data from a more recent period (1998-2007/2010) and showed that the SES gradient (based on household income) in AMI and stroke incidence among ethnic minority groups was almost similar to the ones seen among ethnic Dutch (Chapter 2.5 and 3.2). This indicates that the SES gradient among ethnic minority groups has indeed converged towards the ethnic Dutch population as proposed by the diffusion theory. Individuals belonging to a low socioeconomic group have a comparatively high CVD incidence, independent of ethnic background. As a consequence, ethnic inequalities in stroke incidence did not substantially differ between SES groups (Chapter 3.2). For example among Surinamese men, stroke incidence was significantly higher in the low-, medium-, and high-income groups compared with their ethnic Dutch counterparts. In Moroccans, stroke incidence was significantly lower than in ethnic Dutch, irrespective of SES group. On the other hand, there was some kind of effect modification present in the relation between ethnicity and AMI incidence (Chapter 2.5). For example, among Surinamese women only those in the medium-income group had a higher AMI incidence, and among Turkish men only the low- and medium-income groups had a higher AMI incidence compared with ethnic Dutch. This indicates that although the SES gradient in AMI incidence has converged in a large extent towards the ethnic Dutch, there are still

224

General discussion some discrepancies that cause differences in ethnic inequalities over SES groups. In conclusion, SES was sometimes an effect modifier in the relation between ethnicity and AMI incidence, but not in the relation between ethnicity and stroke incidence.

What is the role of specialised health care use in ethnic inequalities in AMI prognosis? Ethnic inequalities in specialised health care use could only be investigated in the ethnic Dutch population and among Surinamese, Moroccan and Turkish minorities, due to low numbers in the other ethnic minority groups. This thesis showed a higher mortality and readmission rate after AMI among these ethnic minority groups compared with ethnic Dutch. In this paragraph we will try to elucidate the role of acute in-hospital care and cardiovascular drug therapy in these ethnic inequalities. Previous literature shows that timely revascularisation after an AMI event improves prognosis drastically and is therefore highly recommended.49;50 This thesis shows no clear ethnic inequalities in performing a revascularisation procedure after an ST-elevation myocardial infarction in the Netherlands. Consequently, the generally worse prognosis after a first AMI hospitalisation among ethnic minorities is unlikely caused by differences in acute in-hospital care. These results underscore the universally accessibility of the health care system in the Netherlands, established to provide optimal and equal care to everyone in society. This is in contrast to the USA, where patients without health insurance have to pay for their procedure themselves. Accordingly, extensive literature reports a much lower revascularisation rate among African-American AMI patients than among their White counterparts.51-55 It should be noted that in this thesis we did not have information on time delay between symptom onset and the revascularisation procedure. In other countries, time delay among ethnic minority groups have been reported (for example among South Asians) and this could importantly undermine the beneficial effect of the procedure on prognosis.56;57 Future research should elucidate the role of time delay on ethnic inequalities in prognosis after AMI in the Netherlands. After an AMI event, a cascade of cardiovascular drugs is prescribed to control cardiovascular risk factors and prevent cardiovascular readmission and death. This thesis showed that Surinamese, Moroccan, and Turkish minorities are just as likely to collect blood pressure lowering, lipid lowering, and antithrombotic drugs at the pharmacy as the ethnic Dutch population, reflecting an adequate secondary prevention practice by cardiologists. However, ethnic minorities were more likely to quit their drug therapy than their ethnic Dutch counterparts, especially the Moroccan and Turkish patients. These results suggest that the worse prognosis after AMI among Moroccan and Turkish minorities (and in a lesser extent among Surinamese) may be affected by their lower drug continuation. We were unable to investigate whether the dispensed drugs were actually consumed and used properly, which could weaken their protective effect and further

225

Chapter 6 worsen prognosis. Also internationally, higher quitting rates of cardiovascular drugs after an AMI event among ethnic minority groups have been reported.58;59 Just as in the primary care setting, high cardiovascular drug quitting rates could be caused by several factors, such as linguistic and cultural barriers, more side effects or beliefs about side effects, lack of understanding, and difference in health beliefs (preference for alternative medicine).31-36 Other health care factors related to specialised care not studied in this thesis could have affected ethnic inequalities in prognosis after AMI. A major secondary preventive strategy is referral to cardiac rehabilitation. In the Netherlands, everyone with a cardiovascular event has an indication for cardiac rehabilitation (CR). CR is a multifaceted and multidisciplinary approach for secondary prevention of CVD. Core components include exercise training, nutritional counseling, stress management, smoking cessation therapies, mental health counseling, and education. A Dutch study from 2012 showed that less than one third of the patients with an acute coronary syndrome followed CR.60 In the USA and the UK, studies reported an even lower percentage among ethnic minority groups such as the African Americans and South Asians, implying that ethnic minorities face more barriers in accessing CR services.61;62 Underlying factors may include lack of understanding of the causes and prevention of heart disease (for example due to language barriers), negative experiences of cardiac care, religious reasons and cultural norms (for example regarding physical activity and sport among women), fatalistic health beliefs (‘it is up to God’), and practical considerations (poor health, lack of time, transport issues).32;61;63;64 Future research has to elucidate into what extent inadequate CR affected the worse prognosis after AMI among ethnic minorities in the Netherlands.

FUTURE DIRECTIONS For research The majority of studies in this thesis focused on the description of ethnic inequalities in incidence and prognosis of cardiovascular disease; only a few studies focused on ethnic inequalities in health care use. With respect to incidence, more research is necessary on ethnic inequalities in the effectiveness of GP consultation on cardiovascular risk management. Furthermore, efforts should be made to study cardiovascular drug use in the primary care setting while correcting for the underlying need (risk factor profile, indication) for these drugs. Linkage with other studies that have incorporated this information, such as the HELIUS study, can make it possible to draw conclusions on equity in cardiovascular drug use.65 Also expanding current health insurance databases (such as the AHD used in this thesis) with more clinical data, for example by including the indication for drug prescriptions, is desirable. Exploring switching patterns between different types of blood pressure lowering and lipid lowering drugs may indicate specific types of drugs that are least or best tolerated within specific ethnic minority groups, which can have a major impact on drug prescription

226

General discussion guidelines. With respect to prognosis, more research is necessary on ethnic inequalities in time delay between symptom onset and acute in-hospital care, and on ethnic inequalities in use of cardiac rehabilitation therapy and compliance. In this thesis, ethnic inequalities in CVD incidence and prognosis were studied separately from ethnic inequalities in health care use. Therefore, we can only make assumptions on whether health care use affected incidence and prognosis. In order to identify into what extent inequalities in health care use explained inequalities in CVD incidence and prognosis, these factors have to be studied simultaneously. The main part of this thesis consists of studies on ethnic inequalities in the incidence of acute myocardial infarction. However, linkage of nationwide registers gives an abundance of opportunities for further research on ethnic inequalities in incidence and prognosis of other types of CVD and other diseases. In this thesis we observed that the higher or lower AMI incidence in first generation ethnic minorities often converged towards the ethnic Dutch population in the second generation. For example, the significantly higher AMI incidence in first generation Surinamese became equal to the ethnic Dutch in second generation Surinamese. Since Surinamese minorities also have a very high stroke incidence, it would be interesting to investigate whether there is also a change over generations for stroke. In the Netherlands, the majority of Moroccan and Turkish minority groups at risk for CVD are belonging to the first generation. The second generation is still too young to develop CVD. In the future, it would be of interest to investigate the intergenerational change in CVD incidence and prognosis among Moroccan and Turkish groups, two of the major ethnic minority groups in the Netherlands and in Europe. Time trend analyses in this thesis showed some remarkable findings among Moroccans, who seem to lose their beneficial position with respect to AMI incidence over time. Future studies should continue monitoring trends for CVD incidence and prognosis among ethnic minority groups to identify high risk groups that may need extra attention in preventive strategies. Furthermore, efforts should be made to elucidate the role of health care use in the observed differences in ethnic inequalities over generations and time.

For practice The studies in this thesis are predominantly descriptive and do not immediately call for action in primary health care and clinical practice. Results do not give rise to ethnicity specific guidelines in first instance. However, some important findings need to be stressed. First, there are major differences in the incidence of CVD between ethnic minority groups and the ethnic Dutch population. Especially Surinamese minorities are at high cardiovascular risk and need extra attention in primary preventive strategies. Since the beneficial position among Moroccans with respect to AMI is rapidly fading, primary preventive strategies should also focus on Moroccans. Prognosis after AMI is often worse among ethnic minority groups, which should be kept in mind in secondary preventive care. Furthermore, GP’s and leading physicians should make efforts to improve cardiovascular drug

227

Chapter 6 continuation among ethnic minority groups. They have to be aware of possible barriers for adhering (such as (beliefs about) adverse drug reactions, inadequate health literacy, and cultural barriers), and actions must be taken to remove those barriers.

OVERALL CONCLUSION The studies described in this thesis have led to the following main conclusions: - There are ethnic inequalities in the incidence of cardiovascular disease. Especially Surinamese minorities have a substantially higher incidence of cardiovascular disease compared with their ethnic Dutch counterparts, whereas Moroccan minorities have a substantially lower incidence. However, Moroccan minorities seem to lose this beneficial position regarding cardiovascular disease rapidly over time. - Ethnic minority groups without cardiovascular disease have a higher quitting rate of prescribed cardiovascular drugs compared with the ethnic Dutch population without cardiovascular disease. This may partially underlie the higher cardiovascular disease incidence found among some ethnic minority groups. - Most ethnic minority groups have a worse prognosis after a first hospitalisation for acute myocardial infarction than the ethnic Dutch population. This is probably not the result of differences in revascularisation rate. The worse prognosis may be partially induced by a higher quitting rate with cardiovascular drugs to prevent a recurrent event among ethnic minority groups. It is of importance that health care professionals take the findings of this thesis into account when developing primary and secondary preventive strategies, and when treating ethnic minority groups in the primary as well as specialised care setting.

228

General discussion

Genes Environment factors factors

Cardiovascular risk factors

Unbeneficial cardiovascular risk profile and/or cardiovascular symptoms

Incidence Primary prevention Cardiovascular event

Chapter 5.1

Non-fatal, non-hospitalised event Hospitalisation Fatal event

Chapter 2 and 3 Chapter 2 and 3

Acute in-hospital care Death

Chapter 5.2 Chapter 4

Secondary prevention

Prognosis Chapter 5.3

Stabilisation Recurrent event Death

Chapter 4.1 Chapter 4.1

LEGEND Potential barriers in receiving adequate care among ethnic minority groups

Figure 1 Pathway of cardiovascular disease

229

Chapter 6

REFERENCES 1. Stronks K, Kunst AE. The complex interrelationship between ethnic and socio-economic inequalities in health. J Public Health (Oxf) 2009;31(3):324-5. 2. Addo J, Ayerbe L, Mohan KM, Crichton S, Sheldenkar A, Chen R, et al. Socioeconomic status and stroke: an updated review. Stroke 2012;43(4):1186-91. 3. Cox AM, McKevitt C, Rudd AG, Wolfe CD. Socioeconomic status and stroke. Lancet Neurol 2006;5(2):181-8. 4. Manrique-Garcia E, Sidorchuk A, Hallqvist J, Moradi T. Socioeconomic position and incidence of acute myocardial infarction: a meta-analysis. J Epidemiol Community Health 2011;65(4):301-9. 5. Ranganathan M, Bhopal R. Exclusion and inclusion of nonwhite ethnic minority groups in 72 North American and European cardiovascular cohort studies. PLoS Med 2006;3(3):e44. 6. Ujcic-Voortman J. Ethnic disparities in cardiovascular disease risk: the distribution of risk factors among Amsterdam residents with a Turkish and Moroccan ethnic background. Thesis 2011. 7. Merry AH, Boer JM, Schouten LJ, Feskens EJ, Verschuren WM, Gorgels AP, et al. Validity of coronary heart diseases and heart failure based on hospital discharge and mortality data in the Netherlands using the cardiovascular registry Maastricht cohort study. Eur J Epidemiol 2009;24(5):237-47. 8. Nieuwkamp DJ, Vaartjes I, Algra A, Rinkel GJ, Bots ML. Risk of cardiovascular events and death in the life after aneurysmal subarachnoid haemorrhage: a nationwide study. Int J Stroke 2012 [Epub ahead of print]. 9. Harteloh P, de Bruin K, Kardaun J. The reliability of cause-of-death coding in The Netherlands. Eur J Epidemiol 2010;25(8):531-8. 10. Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol 2004;57(12):1288-94. 11. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56(3):221-9. 12. Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005;23(11):1971-7. 13. Ujcic-Voortman JK, Schram MT, Jacobs-van der Bruggen MA, Verhoeff AP, Baan CA. Diabetes prevalence and risk factors among ethnic minorities. Eur J Public Health 2009;19(5):511-5. 14. Ament P, Kessels W. Regionaal Inkomensonderzoek: uitgebreide onderzoeksbeschrijving. Voorburg: Centraal Bureau voor de Statistiek (CBS) 2008. 15. McLoone P, Ellaway A. Postcodes don't indicate individuals' social class. BMJ 1999;319(7215):1003-4. 16. Pampalon R, Hamel D, Gamache P. A comparison of individual and area-based socio-economic data for monitoring social inequalities in health. Components of Statistics Canada Catalogue no. 82-003-X. Health Reports 2009. 17. Rafnsson SB, Bhopal RS. Large-scale epidemiological data on cardiovascular diseases and diabetes in migrant and ethnic minority groups in Europe. Eur J Public Health 2009;19(5):484-91. 18. Stronks K, Kulu-Glasgow I, Agyemang C. The utility of 'country of birth' for the classification of ethnic groups in health research: the Dutch experience. Ethn Health 2009;14(3):255-69. 19. Smeets HM, de Wit NJ, Hoes AW. Routine health insurance data for scientific research: potential and limitations of the Agis Health Database. J Clin Epidemiol 2011;64(4):424-30. 20. Agyemang C, Ujcic-Voortman J, Uitenbroek D, Foets M, Droomers M. Prevalence and management of hypertension among Turkish, Moroccan and native Dutch ethnic groups in Amsterdam, the Netherlands: The Amsterdam Health Monitor Survey. J Hypertens 2006;24(11):2169-76.

230

General discussion

21. Agyemang C, van Valkengoed I, van den Born BJ, Bhopal R, Stronks K. Heterogeneity in sex differences in the metabolic syndrome in Dutch white, Surinamese African and South Asian populations. Diabet Med 2012;29(9):1159-64. 22. Anand SS, Yusuf S, Vuksan V, Devanesen S, Teo KK, Montague PA, et al. Differences in risk factors, atherosclerosis, and cardiovascular disease between ethnic groups in Canada: the Study of Health Assessment and Risk in Ethnic groups (SHARE). Lancet 2000;356(9226):279-84. 23. Cornelisse-Vermaat JR, van den Brink HM. Ethnic differences in lifestyle and overweight in the Netherlands. Obesity (Silver Spring) 2007;15(2):483-93. 24. de Munter JS, van V, I, Agyemang C, Kunst AE, Stronks K. Large ethnic variations in recommended physical activity according to activity domains in amsterdam, the netherlands. Int J Behav Nutr Phys Act 2010;7:85. 25. Dijkshoorn H, Uitenbroek DG, Middelkoop BJ. [Prevalence of diabetes mellitus and cardiovascular disease among immigrants from Turkey and Morocco and the indigenous Dutch population]. Ned Tijdschr Geneeskd 2003;147(28):1362-6. 26. Harland JO, Unwin N, Bhopal RS, White M, Watson B, Laker M, et al. Low levels of cardiovascular risk factors and coronary heart disease in a UK Chinese population. J Epidemiol Community Health 1997;51(6):636-42. 27. Hosper K, Nierkens V, Nicolaou M, Stronks K. Behavioural risk factors in two generations of non- Western migrants: do trends converge towards the host population? Eur J Epidemiol 2007;22(3):163- 72. 28. Tierney S, Deaton C, Mamas M. Heart failure among South Asians: a narrative review of risk, nature, outcomes and management. Heart Fail Rev 2013;18(2):197-206. 29. Ujcic-Voortman JK, Bos G, Baan CA, Uitenbroek DG, Verhoeff AP, Seidell JC. Ethnic differences in total and HDL cholesterol among Turkish, Moroccan and Dutch ethnic groups living in Amsterdam, the Netherlands. BMC Public Health 2010;10:740. 30. Ujcic-Voortman JK, Bos G, Baan CA, Verhoeff AP, Seidell JC. Obesity and body fat distribution: ethnic differences and the role of socio-economic status. Obes Facts 2011;4(1):53-60. 31. Harmsen JA, Bernsen RM, Bruijnzeels MA, Meeuwesen L. Patients' evaluation of quality of care in general practice: what are the cultural and linguistic barriers? Patient Educ Couns 2008;72(1):155-62. 32. Safeer RS, Cooke CE, Keenan J. The impact of health literacy on cardiovascular disease. Vasc Health Risk Manag 2006;2(4):457-64. 33. van Wieringen JC, Harmsen JA, Bruijnzeels MA. Intercultural communication in general practice. Eur J Public Health 2002;12(1):63-8. 34. Beune EJ, Haafkens JA, Schuster JS, Bindels PJ. 'Under pressure': How Ghanaian, African-Surinamese and Dutch patients explain hypertension. J Hum Hypertens 2006;20(12):946-55. 35. Beune EJ, Haafkens JA, Agyemang C, Schuster JS, Willems DL. How Ghanaian, African-Surinamese and Dutch patients perceive and manage antihypertensive drug treatment: a qualitative study. J Hypertens 2008;26(4):648-56. 36. Gazmararian JA, Kripalani S, Miller MJ, Echt KV, Ren J, Rask K. Factors associated with medication refill adherence in cardiovascular-related diseases: a focus on health literacy. J Gen Intern Med 2006;21(12):1215-21. 37. Uiters E, Deville WL, Foets M, Groenewegen PP. Use of health care services by ethnic minorities in The Netherlands: do patterns differ? Eur J Public Health 2006;16(4):388-93. 38. Bhopal RS, Bansal N, Fischbacher C, Brown H, Capewell S. Ethnic variations in chest pain and angina in men and women: Scottish Ethnicity and Health Linkage Study of 4.65 million people. Eur J Cardiovasc Prev Rehabil 2012;19(6):1250-7.

231

Chapter 6

39. Borne Y, Engstrom G, Essen B, Sundquist J, Hedblad B. Country of birth and risk of hospitalization due to heart failure: a Swedish population-based cohort study. Eur J Epidemiol 2011;26(4):275-83. 40. Gadd M, Johansson SE, Sundquist J, Wandell P. Morbidity in cardiovascular diseases in immigrants in Sweden. J Intern Med 2003;254(3):236-43. 41. Hedlund E, Lange A, Hammar N. Acute myocardial infarction incidence in immigrants to Sweden. Country of birth, time since immigration, and time trends over 20 years. Eur J Epidemiol 2007;22(8):493-503. 42. Hempler NF, Larsen FB, Nielsen SS, Diderichsen F, Andreasen AH, Jorgensen T. A registry-based follow- up study, comparing the incidence of cardiovascular disease in native Danes and immigrants born in Turkey, Pakistan and the former Yugoslavia: do social inequalities play a role? BMC Public Health 2011;11:662. 43. Khan FA, Zia E, Janzon L, Engstrom G. Incidence of stroke and stroke subtypes in Malmo, Sweden, 1990-2000: marked differences between groups defined by birth country. Stroke 2004;35(9):2054-8. 44. McKeigue PM, Marmot MG. Mortality from coronary heart disease in Asian communities in London. BMJ 1988;297(6653):903. 45. McKeigue PM, Miller GJ, Marmot MG. Coronary heart disease in south Asians overseas: a review. J Clin Epidemiol 1989;42(7):597-609. 46. Williams R, Wright W, Hunt K. Social class and health: the puzzling counter-example of British South Asians. Soc Sci Med 1998;47(9):1277-88. 47. Bos V, Kunst AE, Garssen J, Mackenbach JP. Socioeconomic inequalities in mortality within ethnic groups in the Netherlands, 1995-2000. J Epidemiol Community Health 2005;59(4):329-35. 48. Mackenbach JP, Cavelaars AE, Kunst AE, Groenhof F. Socioeconomic inequalities in cardiovascular disease mortality; an international study. Eur Heart J 2000;21(14):1141-51. 49. Chan MY, Sun JL, Newby LK, Shaw LK, Lin M, Peterson ED, et al. Long-term mortality of patients undergoing cardiac catheterization for ST-elevation and non-ST-elevation myocardial infarction. Circulation 2009;119(24):3110-7. 50. Stenestrand U, Wallentin L. Early revascularization and 1-year survival in 14-day survivors of acute myocardial infarction: a prospective cohort study. Lancet 2002;359:1805-11. 51. Echols MR, Mahaffey KW, Banerjee A, Pieper KS, Stebbins A, Lansky A, et al. Racial differences among high-risk patients presenting with non-ST-segment elevation acute coronary syndromes (results from the SYNERGY trial). Am J Cardiol 2007;99(3):315-21. 52. Freund KM, Jacobs AK, Pechacek JA, White HF, Ash AS. Disparities by race, ethnicity, and sex in treating acute coronary syndromes. J Womens Health (Larchmt ) 2012;21(2):126-32. 53. Iribarren C, Tolstykh I, Somkin CP, Ackerson LM, Brown TT, Scheffler R, et al. Sex and racial/ethnic disparities in outcomes after acute myocardial infarction: a cohort study among members of a large integrated health care delivery system in northern California. Arch Intern Med 2005;165(18):2105-13. 54. Thomas KL, Honeycutt E, Shaw LK, Peterson ED. Racial differences in long-term survival among patients with coronary artery disease. Am Heart J 2010;160(4):744-51. 55. Ting HH, Roe MT, Gersh BJ, Spertus JA, Rumsfeld JS, Ou FS, et al. Factors associated with off-label use of drug-eluting stents in patients with ST-elevation myocardial infarction. Am J Cardiol 2008;101(3):286-92. 56. De Luca G., Suryapranata H, Ottervanger JP, Antman EM. Time delay to treatment and mortality in primary angioplasty for acute myocardial infarction: every minute of delay counts. Circulation 2004;109(10):1223-5. 57. Kendall H, Marley A, Patel JV, Khan JM, Blann AD, Lip GY, et al. Hospital delay in South Asian patients with acute ST-elevation myocardial infarction in the UK. Eur J Prev Cardiol 2013;20(5):737-42.

232

General discussion

58. Hempler NF, Diderichsen F, Larsen FB, Ladelund S, Jorgensen T. Do immigrants from Turkey, Pakistan and Yugoslavia receive adequate medical treatment with beta-blockers and statins after acute myocardial infarction compared with Danish-born residents? A register-based follow-up study. Eur J Clin Pharmacol 2010;66(7):735-42. 59. Lai EJ, Grubisic M, Palepu A, Quan H, King KM, Khan NA. Cardiac medication prescribing and adherence after acute myocardial infarction in Chinese and South Asian Canadian patients. BMC Cardiovasc Disord 2011;11:56. 60. Van Engen-Verheul M, De Vries H, Kemps H, Kraaijenhagen R, de Keizer N, Peek N. Cardiac rehabilitation uptake and its determinants in the Netherlands. Eur J Prev Cardiol 2013;20(2):349-56. 61. Chauhan U, Baker D, Lester H, Edwards R. Exploring uptake of cardiac rehabilitation in a minority ethnic population in England: a qualitative study. Eur J Cardiovasc Nurs 2010;9(1):68-74. 62. Cortes O, Arthur HM. Determinants of referral to cardiac rehabilitation programs in patients with coronary artery disease: a systematic review. Am Heart J 2006;151(2):249-56. 63. Tod AM, Wadsworth E, Asif S, Gerrish K. Cardiac rehabilitation: the needs of South Asian cardiac patients. Br J Nurs 2001;10(16):1028-33. 64. Tod AM, Lacey EA, McNeill F. 'I'm still waiting...': barriers to accessing cardiac rehabilitation services. J Adv Nurs 2002;40(4):421-31. 65. www.heliusstudie.nl.

233

CHAPTER 7

SUMMARY SAMENVATTING DANKWOORD CURRICULUM VITAE LIST OF PUBLICATIONS

Summary

SUMMARY

Cardiovascular disease (CVD) is the main contributor to morbidity and mortality and places therefore a huge burden on both patients and health care costs. There is growing evidence of differences in CVD incidence, prognosis, and health care use between ethnic minority groups and the majority population, but European studies on this topic remain limited. Seen the high and rising number of ethnic minority groups in European countries, the first aim of this thesis is to investigate ethnic inequalities in the incidence of and prognosis after CVD by linking several Dutch nationwide registers (Population Register, Hospital Discharge Register, Cause of Death Register, and the Regional Income Survey). The second aim of this thesis is to provide insight into ethnic inequalities in cardiovascular health care use in the primary and secondary care setting by using data from the Achmea health insurance company. In Chapter 1 the background, objectives and outline of this thesis are described.

Inequalities in incidence of acute myocardial infarction In Chapter 2 ethnic inequalities in the incidence of acute myocardial infarction (AMI) is elaborately investigated. Previously, differences in AMI incidence between ethnic minority groups and the majority population have been reported. Chapter 2.1 shows that the direction and extent of these differences depend on the country of origin, and often vary between ethnic minorities originating from the same geographical region. For example, among North-African and Mediterranean minorities, AMI incidence is higher in Turkish, but lower in Moroccans compared with ethnic Dutch. Most ethnic minority groups have a similar or lower AMI incidence compared with ethnic Dutch. This lower incidence is mostly present among first generation ethnic minorities (born abroad and at least one of the parents born abroad), and are less clear or absent among second generation ethnic minorities (born in the Netherlands with at least one of the parents born abroad). In contrast, ethnic minorities from the former Dutch colonies (Suriname, Indonesia, and Netherlands Antilles) have a similar or higher AMI incidence compared with ethnic Dutch, which changes beneficially over generations (the second generation has a similar or lower AMI incidence compared with ethnic Dutch). In additional analyses, no major differences are seen between men and women. Only among Turkish minorities, men have a higher AMI incidence than their ethnic Dutch counterparts, whereas no differences are observed among women. It is well known that the AMI incidence is much higher in men than in women; however the extent of this difference varies widely between countries. Chapter 2.2 confirms a higher AMI incidence in men than in women among ethnic Dutch as well as among all ethnic minority groups under study. Results show that sex disparities are more pronounced among most non-Western ethnic minority groups (especially those originating from Morocco, Turkey, and South Asia). The difference in sex disparity between ethnic groups are predominantly evident in the young (<55 years

237

Summary of age), and are mainly provoked by a higher AMI incidence in ethnic minority men compared with ethnic Dutch men. In previous decades the incidence and mortality of AMI steadily declined in Western countries such as the Netherlands. Chapter 2.3 reports that this beneficial tendency was also present in ethnic minorities between 1997 and 2007. Among ethnic minorities, AMI incidence declined between 4% and 12% in men and between 10% and 16% in women. The decline did not significantly differ from those seen among ethnic Dutch (men: -12%, women: -10%). Only in Turkish women and in Moroccans the incidence remained stable over time. Few studies have described absolute age- and sex-specific incidence rates of acute myocardial infarction among ethnic minority groups; age- and sex-specific time trends among these groups are even more limited. In the Netherlands such estimates are as yet unavailable. This information is important to monitor whether preventive strategies have gained effect, and may target specific groups that need extra attention in future preventive strategies. Chapter 2.4 shows that, regardless of ethnic background, absolute AMI incidence rates are higher in men than in women and increase with age. Incidence significantly declined over time among ethnic Dutch as well as among most ethnic minority groups under study. Only in Moroccan minorities, AMI incidence significantly increased between 2000-2004 and 2005-2010. Among ethnic minorities with a significantly higher AMI incidence (Surinamese men and women, Turkish men, and Indonesian men), the elevated incidence persisted over time. However, during both time periods the higher incidence was absent among those ≥70 years of age. In many industrialised countries, individuals with a low socioeconomic status (SES) have a higher risk on cardiovascular disease compared with individuals with a high SES. However, among ethnic minority groups, the relationship between SES and the risk of CVD remains uncertain. The general expectation is that the socioeconomic gradient in CVD will eventually converge towards the socioeconomic gradient seen in the host population. Chapter 2.5 demonstrates that, just as in ethnic Dutch, AMI incidence is higher in low-income groups than in high-income groups among all ethnic minority groups under study. Results are similar in men and women. Within all income groups, Surinamese men have a higher AMI incidence, whereas Moroccan men have a consistently lower AMI incidence compared with their ethnic Dutch counterparts.

Inequalities in incidence of stroke Chapter 3 reports on ethnic inequalities in the incidence of stroke. Worldwide, stroke is the second most common cause of mortality and the third most common cause of disability. While several European studies have demonstrated ethnic inequalities in stroke mortality, data on ethnic differences in stroke incidence are limited. Chapter 3.1 gives insight into ethnic differences in the incidence of stroke subtypes (intracerebral haemorrhage, subarachnoid haemorrhage, and ischemic

238

Summary stroke) in the Netherlands. Our data show mixed results with a higher incidence among Surinamese and a lower incidence among Moroccans compared with ethnic Dutch for all of the various stroke subtypes. Among the other ethnic minority groups, the incidence appears to depend on stroke type and sex. For example, among Chinese, the incidence of intracerebral haemorrhage is higher but the risk of ischemic stroke is lower compared with ethnic Dutch. Recently, a consistent pattern of a higher stroke incidence and mortality in low SES groups as compared with high SES groups was found. While SES inequalities in stroke has long been documented in European populations, less consistent results were reported among ethnic minority groups. Chapter 3.2 reveals that socioeconomic inequalities in stroke incidence are present regardless of ethnic background. Furthermore, compared with ethnic Dutch, the incidence of stroke is lower in Moroccans, but higher in Surinamese within all income groups.

Inequalities in prognosis after cardiovascular disease In Chapter 4 ethnic and socioeconomic inequalities in prognosis after a first cardiovascular event are examined. Evidence regarding differences in prognosis after AMI and congestive heart failure (CHF) between ethnic minority groups and the majority population is limited, inconsistent and mainly focusing on mortality only. Chapter 4.1 demonstrates that after the first AMI hospitalisation, mortality rates as well as readmission rates for AMI and CHF are higher in most ethnic minority groups compared with ethnic Dutch. After the first CHF hospitalisation, ethnic inequalities in prognosis are more diverse. Previous research showed a higher short-term mortality after AMI in low SES groups as compared with high SES groups. However, studies are often restricted to patients younger than 75 years of age. Chapter 4.2 assesses whether the SES gradient in mortality after AMI seen in younger patients persists in the elderly (≥75 years of age). Results show an inverse relation between SES and pre-hospital mortality in both men and women; and also between SES and case-fatality, but only in men. These inequalities persist in elderly AMI patients.

Inequalities in cardiovascular health care use In Chapter 5 ethnic inequalities in cardiovascular health care use in primary as well as secondary care are investigated. Chapter 5.1 indicates that most asymptomatic ethnic minority groups (those without CVD) are more likely to quit their cardiovascular drug therapy (blood pressure lowering and lipid lowering drugs) compared with the asymptomatic ethnic Dutch. This may partially contribute to the higher AMI and stroke incidence among some ethnic minority groups (especially Surinamese minorities) reported in Chapter 2 and 3 of this thesis. Chapter 4.1 of this thesis showed a worse prognosis after a first AMI hospitalisation among ethnic minority groups than among the ethnic Dutch population. USA studies suggest this may be

239

Summary caused by a lower revascularisation rate (percutaneous coronary intervention and coronary artery bypass grafting) among ethnic minority groups. However, there is limited evidence on this topic for European countries with a universal access to acute in-hospital care. Chapter 5.2 shows that there are no ethnic inequalities in revascularisation rate after an ST-elevation myocardial infarction between ethnic minority groups and the ethnic Dutch population, suggesting equity in acute in- hospital care. Another factor that may underlie the worse prognosis after AMI among ethnic minority groups is an inadequate cardiovascular drug use. Chapter 5.3 reveals that ethnic minorities are just as likely as ethnic Dutch to collect cardiovascular drugs (blood pressure lowering, lipid lowering, and antithrombotic drugs) at the pharmacy after an AMI event as the ethnic Dutch population, but over time are more likely to quit their therapy.

General discussion In Chapter 6, the general discussion, the key findings and some methodological considerations are described. Furthermore, the role of primary and secondary health care use and socioeconomic status in ethnic inequalities in incidence of and prognosis after cardiovascular disease is discussed. In addition, recommendations for future research and practice are given. The chapter ends with the following main conclusions: - There are ethnic inequalities in the incidence of cardiovascular disease. Especially Surinamese minorities have a substantially higher incidence of cardiovascular disease compared with their ethnic Dutch counterparts, whereas Moroccan minorities have a substantially lower incidence. However, Moroccan minorities seem to lose this beneficial position regarding cardiovascular disease rapidly over time. - Ethnic minority groups without cardiovascular disease have a higher quitting rate of prescribed cardiovascular drugs compared with the ethnic Dutch population without cardiovascular disease. This may partially underlie the higher cardiovascular disease incidence found among some ethnic minority groups. - Most ethnic minority groups have a worse prognosis after a first hospitalisation for acute myocardial infarction than the ethnic Dutch population. This is probably not the result of differences in revascularisation procedure rate. The worse prognosis may be partially induced by a higher quitting rate with cardiovascular drugs to prevent a recurrent event and mortality among ethnic minority groups. It is of importance that health care professionals take the findings of this thesis into account when developing primary and secondary preventive strategies, and when treating ethnic minority groups in the primary as well as specialised care setting.

240

Samenvatting

SAMENVATTING

Hart- en vaatziekten zijn de belangrijkste veroorzakers van ziekte en sterfte en daardoor een grote belasting voor zowel de patiënt als voor de gezondheidszorg. Er komt steeds meer bewijs voor de aanwezigheid van verschillen in incidentie, prognose en gebruik van gezondheidszorg met betrekking tot hart- en vaatziekten tussen etnische minderheidsgroepen en de autochtone populatie. Europese studies over dit onderwerp blijven echter beperkt. Aangezien een groot en groeiend deel van de Europese populatie uit etnische minderheidsgroepen bestaat, is het doel van dit proefschrift om etnische verschillen in incidentie van en prognose na hart- en vaatziekten te onderzoeken. Om dit te bereiken zijn diverse Nederlandse landelijke registraties met elkaar gekoppeld, zoals de gemeentelijke basisadministratie, de landelijke medische registratie, de doodsoorzakenstatistiek en het regionaal inkomensonderzoek. Daarnaast brengt dit proefschrift etnische verschillen in het gebruik van gezondheidszorg met betrekking tot hart- en vaatziekten in kaart door gebruik te maken van gegevens van de Achmea zorgverzekeringsmaatschappij. In Hoofdstuk 1 worden de achtergrond, doelstellingen en de opzet van dit proefschrift beschreven.

Verschillen in incidentie van een acuut hartinfarct In Hoofdstuk 2 worden etnische verschillen in de incidentie van een acuut hartinfarct uitgebreid bestudeerd. Voorgaande onderzoeken hebben verschillen gerapporteerd tussen etnische minderheidsgroepen en de autochtone bevolking. Hoofdstuk 2.1 laat zien dat de richting en mate van deze verschillen afhankelijk zijn van het herkomstland en vaak variëren tussen etnische minderheidsgroepen afkomstig van dezelfde geografische regio. Bijvoorbeeld onder etnische minderheidsgroepen van Noord-Afrikaanse en Mediterrane landen hebben Turken een hogere incidentie, maar Marokkanen juist een lager incidentie van een acuut hartinfarct dan Nederlanders van autochtone afkomst. De meeste etnische minderheidsgroepen hebben een gelijke of lagere incidentie van een acuut hartinfarct dan Nederlanders van autochtone afkomst. Deze lagere incidentie is voornamelijk te zien bij eerste generatie etnische minderheden (geboren in het buitenland met ten minste één van de ouders geboren in het buitenland) en is minder duidelijk of afwezig bij tweede generatie etnische minderheden (geboren in Nederland met ten minste één van de ouders geboren in het buitenland). Aan de andere kant hebben etnische minderheden afkomstig van de voormalig Nederlandse koloniën (Suriname, Indonesië, Nederlandse Antillen) een gelijke of hogere incidentie van een acuut hartinfarct dan Nederlanders van autochtone afkomst, welke positief veranderd over generaties (de tweede generatie heeft een gelijke of lagere incidentie dan Nederlanders van autochtone afkomst). Aanvullende analyses laten geen duidelijke verschillen zien tussen mannen en vrouwen. Alleen onder Turkse minderheden hebben mannen een hogere incidentie dan Nederlanders van autochtone afkomst, terwijl er geen verschillen geobserveerd worden bij vrouwen.

241

Samenvatting

Het is algemeen bekend dat de incidentie van een acuut hartinfarct veel hoger is bij mannen dan bij vrouwen, hoewel de mate van dit verschil enorm kan variëren tussen landen. Hoofdstuk 2.2 bevestigt een hogere incidentie van een acuut hartinfarct bij mannen dan bij vrouwen onder alle onderzochte etnische groepen. De resultaten laten zien dat de man-vrouw verschillen groter zijn bij de meeste niet-Westerse etnische minderheidsgroepen (vooral bij de groepen afkomstig uit Marokko, Turkije en Zuid-Azië). Dit grotere man-vrouw verschil is voornamelijk aanwezig bij de jongeren (<55 jaar oud) en wordt hoofdzakelijk gedreven door de hogere incidentie bij mannen behorend tot een etnische minderheid vergeleken met Nederlandse mannen van autochtone afkomst. In de afgelopen decennia is de incidentie van en sterfte door een acuut hartinfarct drastisch afgenomen in Westerse landen zoals Nederland. Hoofdstuk 2.3 rapporteert dat deze gunstige trend ook aanwezig was bij etnische minderheden tussen 1997 en 2007. Bij etnische minderheden daalden de incidentie van een acuut hartinfarct tussen 4% en 12% onder mannen en tussen 10% en 16% onder vrouwen. Deze afname verschilde niet significant met de afname onder Nederlanders van autochtone afkomst (mannen: -12%, vrouwen: -10%). Alleen bij Turkse vrouwen en bij Marokkanen was deze daling afwezig. Slechts enkele studies hebben de absolute leeftijd- en geslacht-specifieke incidentie van een acuut hartinfarct bij etnische minderheidsgroepen beschreven. Gegevens over leeftijd- en geslacht-specifieke trends over tijd bij deze groepen zijn nog beperkter. In Nederland zijn zulke schattingen tot nu toe afwezig. Deze informatie is belangrijk om te monitoren of preventieve maatregelen effect hebben gehad en om specifieke groepen te identificeren die extra aandacht nodig hebben bij toekomstige preventieve maatregelen. Hoofdstuk 2.4 laat zien dat, onafhankelijk van etnische groepering, de absolute incidentie van een acuut hartinfarct hoger is bij mannen dan bij vrouwen en toeneemt met de leeftijd. De incidentie neemt significant af met de tijd bij zowel Nederlanders van autochtone afkomst als bij de meeste etnische minderheden. Alleen bij Marokkanen is een significante toename te zien tussen 2000-2004 en 2005-2010. De significant hogere incidentie bij Surinaamse mannen en vrouwen, Turkse mannen en Indonesische mannen vergeleken met Nederlanders van autochtone afkomst in 2000-2004, bleef hoger in 2005-2010. Deze hogere incidentie was gedurende beide perioden echter niet waarneembaar bij degene van 70 jaar en ouder. In veel geïndustrialiseerde landen hebben personen met een lage sociaaleconomische status (SES) een hoger risico op het ontwikkelen van hart- en vaatziekten vergeleken met personen met een hoge SES. Bij etnische minderheidsgroepen is de relatie tussen SES en het risico op hart- en vaatziekten minder duidelijk. De algemene verwachting is dat de SES gradiënt in het risico op hart- en vaatziekten bij etnische minderheden uiteindelijk gelijk wordt aan de SES gradiënt in de autochtone populatie. Hoofdstuk 2.5 toont aan dat de incidentie van een acuut hartinfarct hoger is

242

Samenvatting bij groepen met een laag inkomen dan bij groepen met een hoog inkomen, onafhankelijk van etnische groepering. Deze resultaten verschillen niet tussen mannen en vrouwen. Binnen alle inkomensgroepen hebben Surinaamse mannen een hogere incidentie, terwijl Marokkanen een lagere incidentie hebben vergeleken met Nederlanders van autochtone afkomst.

Verschillen in incidentie van een beroerte Hoofdstuk 3 rapporteert over etnische verschillen in de incidentie van een beroerte. Wereldwijd is een beroerte de op een na grootste veroorzaker van sterfte en de op twee na grootste veroorzaker van lichamelijke gebreken. Verscheidene Europese studies hebben etnische verschillen in sterfte door beroerte aangetoond. Gegevens over etnische verschillen in de incidentie van beroerte zijn echter beperkt. Hoofdstuk 3.1 geeft inzicht in etnische verschillen in de incidentie van beroerte (intracerebrale bloedingen, subarachnoïdale bloedingen en ischemische beroerte) in Nederland. De resultaten zijn gevarieerd met een hogere incidentie bij Surinamers en een lagere incidentie bij Marokkanen vergeleken met Nederlanders van autochtone afkomst voor alle types beroerte. Bij de andere etnische minderheidsgroepen is het risico afhankelijk van het type beroerte en geslacht. Nederlanders van Chinese afkomst hebben bijvoorbeeld een hogere incidentie van intracerebrale bloedingen maar een lagere incidentie van ischemische beroerte dan Nederlanders van autochtone afkomst. Lage SES groepen hebben een hogere incidentie van en sterfte door beroerte dan hoge SES groepen. Deze SES gradiënt in het optreden van beroerte wordt al geruime tijd gerapporteerd in Europese landen, maar er is nog onduidelijkheid over deze SES verschillen onder etnische minderheidsgroepen. Hoofdstuk 3.2 laat zien dat de SES gradiënt in de incidentie van beroerte zowel te zien is bij Nederlanders van autochtone afkomst als bij etnische minderheidsgroepen. Verder laat dit hoofdstuk zien dat de incidentie van een beroerte binnen alle inkomensgroepen hoger is bij Surinamers, maar lager bij Marokkanen dan bij Nederlanders van autochtone afkomst.

Verschillen in prognose na hart- en vaatziekten In Hoofdstuk 4 worden etnische en sociaaleconomische verschillen in prognose na een eerste hart- en vaatziekte bestudeerd. Er is tot nu toe weinig onderzoek gedaan naar verschillen in prognose na een eerste ziekenhuisopname voor een acuut hartinfarct en hartfalen tussen etnische minderheidsgroepen en de autochtone populatie. Daarnaast is het aanwezige bewijs inconsistent en vooral gericht op sterfte. Hoofdstuk 4.1 toont aan dat na een eerste ziekenhuisopname voor een acuut hartinfarct, de kans op zowel sterfte als heropname hoger is bij de meeste etnische minderheidsgroepen vergeleken met Nederlanders van autochtone afkomst. Na de eerste ziekenhuisopname voor hartfalen zijn etnische verschillen in prognose meer divers.

243

Samenvatting

Voorgaand onderzoek laat een hoger risico op korte termijn sterfte na een acuut hartinfarct zien bij personen met een lage SES dan bij personen met een hoge SES. Deze onderzoeken hebben zich echter vaak beperkt tot personen jonger dan 75 jaar. In Hoofdstuk 4.2 hebben we gekeken of de SES gradiënt in korte termijn sterfte na een acuut hartinfarct zoals gerapporteerd bij jongere patiënten ook nog aanwezig is bij oudere patiënten (≥75 jaar). De resultaten laten een inverse relatie zien tussen SES en directe sterfte na een acuut hartinfarct (voorafgaand aan ziekenhuisopname) bij zowel mannen als vrouwen. Ook is deze inverse relatie aanwezig tussen SES en korte termijn sterfte na een ziekenhuisopname voor een acuut hartinfarct, maar alleen bij mannen. Deze SES verschillen zijn zowel bij de jongere als oudere patiënten te zien.

Verschillen in gebruik van gezondheidszorg met betrekking tot hart- en vaatziekten In Hoofdstuk 5 zijn etnische verschillen in gebruik van zowel de primaire als secundaire gezondheidszorg met betrekking tot hart- en vaatziekten onderzocht. Hoofdstuk 5.1 toont aan dat de meeste etnische minderheidsgroepen die nog nooit een hart- en vaatziekte hebben gehad vaker stoppen met het gebruik van de voorgeschreven medicatie ter preventie van hart- en vaatziekten (bloeddrukverlagende en cholesterolverlagende medicatie) dan Nederlanders van autochtone afkomst. Dit zou mogelijk kunnen bijdragen aan de hogere incidentie van een acuut hartinfarct en beroerte onder sommige etnische minderheidsgroepen (voornamelijk Surinamers) gerapporteerd in hoofdstuk 2 en 3 van dit proefschrift. Hoofdstuk 4.1 van dit proefschrift liet een slechtere prognose zien na een eerste ziekenhuisopname voor een acuut hartinfarct bij etnische minderheidsgroepen dan bij Nederlanders van autochtone afkomst. Studies uit de Verenigde Staten geven aan dat dit mogelijk veroorzaakt zou kunnen worden door een lager percentage revascularisatie procedures (dotterbehandelingen en coronaire omleidingen) dat wordt uitgevoerd bij etnische minderheidsgroepen. Voor Europese landen met een universele toegang tot acute ziekenhuiszorg is onderzoek over dit onderwerp beperkt. Hoofdstuk 5.2 laat zien dat er geen verschillen zijn in het uitvoeren van een revascularisatie procedure na een ST-elevatie hartinfarct tussen etnische minderheidsgroepen en Nederlanders van autochtone afkomst. Dit suggereert dat er in Nederland gelijkheid bestaat in acute ziekenhuiszorg. Een andere mogelijke oorzaak voor de slechtere prognose na een acuut hartinfarct bij de meeste etnische minderheidsgroepen is een inadequaat gebruik van medicatie tegen het opnieuw optreden van een hart- en vaatziekte. Hoofdstuk 5.3 laat zien dat etnische minderheden met een recent doorgemaakt acuut hartinfarct net zo vaak bloeddrukverlagende, cholesterolverlagende en antitrombotische medicatie bij de apotheek ophalen als Nederlanders van autochtone afkomst, maar dat zij vaker stoppen met het gebruik hiervan.

244

Samenvatting

Algemene discussie In Hoofdstuk 6, de algemene discussie, worden de belangrijkste bevindingen en een aantal methodologische overwegingen beschreven. Verder wordt de rol van primaire en secundaire gezondheidszorg en sociaaleconomische status bij het ontstaan van etnische verschillen in de incidentie van en prognose na hart- en vaatziekten bediscussieerd. Daarnaast worden aanbevelingen aangedragen voor verder onderzoek en voor de praktijk. Dit hoofdstuk eindigt met de volgende belangrijkste conclusies: - Er zijn etnische verschillen in de incidentie van hart- en vaatziekten. Met name personen van Surinaamse afkomst vertonen een sterk verhoogde incidentie van hart- en vaatziekten vergeleken met Nederlanders van autochtone afkomst. Daarentegen hebben personen van Marokkaanse afkomst een sterk verlaagde incidentie. Echter, personen van Marokkaanse afkomst lijken deze gunstige positie ten aanzien van het ontwikkelen van hart- en vaatziekten over de tijd snel te verliezen. - Etnische minderheidsgroepen die nog nooit een hart- en vaatziekte hebben gehad stoppen vaker met het gebruik van de voorgeschreven medicatie ter preventie van hart- en vaatziekten dan Nederlanders van autochtone afkomst. Dit is mogelijk één van de oorzaken van de hogere incidentie van hart- en vaatziekten onder sommige etnische minderheidsgroepen. - De meeste etnische minderheidsgroepen hebben een slechtere prognose na een eerste ziekenhuisopname voor een acuut hartinfarct dan Nederlanders van autochtone afkomst. Dit is waarschijnlijk niet het gevolg van verschillen in het uitvoeren van revascularisatie procedures. De slechtere prognose wordt mogelijk wel gedeeltelijk verklaard doordat etnische minderheden vaker stoppen met het ophalen van voorgeschreven medicatie om het opnieuw optreden van een hart- en vaatziekte te voorkomen. Het is van belang de bevindingen van dit proefschrift onder de aandacht te brengen van beleidsmakers en behandelaars zodat zij zich extra kunnen richten op de etnische groepen met een hoog risico op hart- en vaatziekten.

245

Dankwoord

DANKWOORD

Hierbij wil ik iedereen bedanken die gedurende de afgelopen drie jaar een bijdrage heeft geleverd aan de totstandkoming van dit proefschrift en/of interesse heeft getoond in de voortgang van mijn project. Een aantal personen wil ik in het bijzonder bedanken.

Prof.dr. M.L. Bots, beste Michiel. Je hebt me geleerd een zelfstandig onderzoeker te zijn en te vertrouwen op mijn eigen kunnen. Gedurende het promotietraject wist jij altijd de grote lijn te behouden. Je stimuleerde me mezelf niet te verliezen in de beperkingen van het onderzoek, maar juist de mogelijkheden in te zien en te benutten. Dit zal me de rest van mijn loopbaan bijblijven.

Prof.dr. K. Stronks, beste Karien. Ook al zagen we elkaar niet vaak, je wist altijd de vinger op de zere plek te leggen. Je korte maar krachtige feedback heeft verschillende artikelen in dit proefschrift naar een hoger niveau getild.

Dr. I. Vaartjes, beste Ilonca. Vanaf het begin van mijn promotietraject ben je een betrokken en gedreven begeleidster geweest. Je maakte altijd tijd voor me vrij. Op momenten dat ik het even niet meer zag zitten, wist jij structuur te geven en mij te motiveren om door te gaan. Zonder jou was hoofdstuk 5 er waarschijnlijk niet in deze uitgebreide vorm geweest!

Dr. C.O. Agyemang, dear Charles. Over the last few years you have been a kind and easily accessible supervisor. Thank you for your elaborate and very fast feedback on my manuscript drafts, and for sharing your extensive knowledge regarding ethnicity with me. I could always count on you!

De leden van de beoordelingscommissie, Prof.dr. Y. van der Graaf, Prof.dr. Th.J.M. Verheij, Prof.dr. L.J. Kapelle, Prof.dr. R.J.G. Peters en Prof.dr. M.L. Essink-Bot, ben ik zeer erkentelijk voor het lezen en beoordelen van mijn manuscript.

Alle artikelen in dit proefschrift zijn in zijn geheel of gedeeltelijk gebaseerd op data van het Centraal Bureau van de Statistiek (CBS). Dank aan de medewerkers van de microdataservices, die ervoor gezorgd hebben dat de data op een geordende manier aangeleverd werden, waardoor de koppeling en de daaropvolgende analyses zo soepel mogelijk konden worden uitgevoerd.

Hoofdstuk 5 in dit proefschrift is voornamelijk gebaseerd op data beschikbaar gesteld door de Achmea zorgverzekeringsmaatschappij. Dr. Hugo Smeets en Henk Evers ben ik zeer dankbaar voor de geïnvesteerde tijd en moeite die nodig was voor het aanleveren van deze grote databestanden.

246

Dankwoord

Diane, zonder jou was het niet haalbaar geweest om in de nog beschikbare tijd van de promotie hoofdstuk 4.1 af te ronden. Jij hebt ervoor gezorgd dat ik op een snelle manier een dataset kon bouwen om de prognose na hart- en vaatziekten te onderzoeken.

Alle co-auteurs bedankt voor de inzet en belangrijke feedback bij het schrijven van de artikelen. Als niet-medicus waardeer ik vooral de inbreng van de medisch specialisten enorm.

Medewerkers van de ICT, het epi-secretariaat en de receptie, jullie waren onmisbaar bij computerproblemen, het maken van afspraken en bij de verspreiding van mijn proefschrift.

Ik wil alle collega’s van het Julius Centrum bedanken, met name de kamergenoten van Stratenum 5.122, van Geuns verdieping 7 en de ‘CBS kamer’, voor de gezellige werkomgeving en de nuttige vakinhoudelijke gesprekken. Ook wil ik alle collega’s bedanken die mee zijn geweest naar de congressen in Edinburgh, Kopenhagen, San Diego, New Orleans en Rome. Het waren bijzondere ervaringen die ik niet snel zal vergeten.

Ilona, mijn kamergenoot gedurende het laatste jaar van mijn promotie. Beiden deden wij onderzoek op het gebied van etniciteit, een uitzondering bij het Julius Centrum. Het was fijn om met jou op dit ‘etniciteit eilandje’ te zitten en om samen over dit onderwerp te overleggen.

Papa en mama, jullie betrokkenheid, zorgzaamheid en steun zijn van grote waarde geweest in de afgelopen drie jaar. De bemoedigende en opbeurende woorden tijdens moeilijke perioden hebben veel voor mij betekend. Bedankt dat jullie er altijd voor mij zijn.

Linda, van kleins af aan sta jij altijd voor me klaar. Ik vind het fijn dat je mijn paranimf wilt zijn om me ook tijdens mijn verdediging bij te staan.

Lieve Robbert, wat ben ik blij met jou. Door jouw humor en relativeringsvermogen heb ik vooral de laatste loodjes van mijn promotietraject als minder zwaar ervaren. Bedankt voor alles.

247

Curriculum Vitae

CURRICULUM VITAE

Louise van Oeffelen was born on July 31th 1984 in Terneuzen, the Netherlands. In 2008 she obtained her bachelor degree in Nutrition and Dietetics at the Hogeschool van Arnhem en Nijmegen (HAN). In the same year she started her academic education by studying Nutrition and Health at the Wageningen University, with a specialisation in Epidemiology and Public Health. Her master thesis was performed at the Dutch National Institute of Public Health and the Environment (RIVM), where she investigated the relation between serum micronutrient concentrations and the development of asthma in children. After she graduated in 2010, she worked four months at the RIVM as a junior researcher on a project about the prevalence and consequences of chronic diseases in children. In February 2011 she started her PhD research described in this thesis at the Julius Center for Health Sciences and Primary Care of the University Medical Center Utrecht, in collaboration with the Department of Social Medicine of the Academic Medical Center in Amsterdam. She worked under supervision of Prof.dr. Michiel Bots, Prof.dr. Karien Stronks, Dr. Ilonca Vaartjes, and Dr. Charles Agyemang. She combined her PhD project with the post-master program Clinical Epidemiology at the Utrecht University, where she graduated in April 2014.

Louise van Oeffelen werd geboren op 31 juli 1984 in Terneuzen. In 2008 behaalde ze haar bachelor diploma Voeding en Diëtetiek aan de Hogeschool van Arnhem en Nijmegen (HAN). In datzelfde jaar begon ze haar master studie Voeding en Gezondheid aan de Universiteit van Wageningen, waarbij ze zich specialiseerde in Epidemiologie en Volksgezondheid. Ze deed haar afstudeerstage bij het Rijksinstituut voor Volksgezondheid en Milieu (RIVM), waar ze de relatie tussen serum micronutriënten en het ontwikkelen van astma bij kinderen onderzocht. Na haar afstuderen in 2010 werkte zij vier maanden bij het RIVM als junior onderzoeker op een project over de prevalentie en gevolgen van chronische aandoeningen bij kinderen. In februari 2011 startte ze met het promotieonderzoek beschreven in dit proefschrift bij het Julius Centrum voor Gezondheidswetenschappen en Eerstelijns Geneeskunde, in samenwerking met de afdeling Sociale Geneeskunde van het Academisch Medisch Centrum in Amsterdam. Ze werd begeleid door Prof.dr. Michiel Bots, Prof.dr. Karien Stronks, Dr. Ilonca Vaartjes en Dr. Charles Agyemang. Ze combineerde haar promotieonderzoek met de post-master Klinische Epidemiologie aan de Universiteit van Utrecht, waar ze afstudeerde in April 2014.

248

List of publications

LIST OF PUBLICATIONS van Oeffelen AAM, Rittersma S, Vaartjes I, Stronks K, Bots ML, Agyemang C. Are there ethnic inequalities in revascularisation procedure rate after an ST-elevation myocardial infarction? Submitted.

Agyemang C, van Oeffelen AAM, Norredam M, Kapelle LJ, Klijn CJM, Bots ML, Stronks K, Vaartjes I. Socioeconomic inequalities in stroke incidence among migrant groups: Analysis of nation-wide data. Submitted.

Agyemang C, van Oeffelen AAM, Norredam M, Kapelle LJ, Klijn CJM, Bots ML, Stronks K, Vaartjes I. Ethnic disparities in the incidence of ischemic stroke, intracerebral haemorrhage and subarachnoid haemorrhage in the Netherlands. Submitted. van Oeffelen AAM, Agyemang C, Stronks K, Bots ML, Vaartjes I. Prognosis after a first hospitalisation for acute myocardial infarction and congestive heart failure by country of birth. Heart ‘under revision’. van Oeffelen AAM, Agyemang C, Stronks K, Bots ML, Vaartjes I. Incidence of first acute myocardial infarction over time specific for age, sex, and country of birth. Neth J Med 2014;72(1):20-7.

Agyemang C, van Oeffelen AAM, Bots ML, Stronks K, Vaartjes I. Socioeconomic inequalities in acute myocardial infarction incidence in migrant groups: has the epidemic arrived? Analysis of nation-wide data. Heart 2014;100(3):239-46. van Oeffelen AAM, Vaartjes I, Stronks K, Bots ML, Agyemang C. Incidence of acute myocardial infarction in first and second generation minority groups: does the second generation converge towards the majority population? Int J Cardiol 2013;168(6):5422-9. van Oeffelen AAM, Vaartjes I, Stronks K, Bots ML, Agyemang C. Sex disparities in acute myocardial infarction incidence: Do ethnic minority groups differ from the majority population? Eur J Prev Cardiol 2013 [Epub ahead of print]. van Oeffelen AAM, Agyemang C, Koopman C, Stronks K, Bots ML, Vaartjes I. Downward trends in acute myocardial infarction incidence: how do migrants fare with the majority population? Results from a nationwide study. Eur J Prev Cardiol 2013 [Epub ahead of print].

249

List of publications

Koopman C, Bots ML, van Oeffelen AAM, van Dis I, Verschuren WM, Engelfriet PM, Capewell S, Vaartjes I. Population trends and inequalities in incidence and short-term outcome of acute myocardial infarction between 1998 and 2007. Int J Cardiol 2013;168(2):993-8. van Oeffelen AAM, Agyemang C, Bots ML, Stronks K, Koopman C, van Rossem L, Vaartjes I. The relation between socioeconomic status and short-term mortality after acute myocardial infarction persists in the elderly: results from a nationwide study. Eur J Epidemiol 2012;27(8):605-13.

Van Oeffelen AAM, Agyemang C, Bots ML, Stronks K, Koopman C, van Rossem L, Vaartjes I. Sociaaleconomische status en korte termijn sterfte na een acuut hartinfarct bij mannen en vrouwen van verschillende leeftijden. In: Koopman C, van Dis I, Visseren FLJ, Vaartjes I, Bots ML. Hart- en vaatziekten in Nederland 2012, cijfers over risicofactoren, ziekte en sterfte. Den Haag: Hartstichting, 2012.

Koopman C, van Oeffelen AAM, Bots ML, Engelfriet PM, Verschuren WM, van Rossem L, van Dis I, Capewell S, Vaartjes I. Neighbourhood socioeconomic inequalities in incidence of acute myocardial infarction: a cohort study quantifying age- and gender-specific differences in relative and absolute terms. BMC Public Health 2012;12:617. van Oeffelen AAM, Bekkers MB, Smit HA, Kerkhof M, Koppelman GH, Haveman-Nies A, van der A DL, Jansen EH, Wijga AH. Serum micronutrient concentrations and childhood asthma: the PIAMA birth cohort study. Pediatr Allergy Immunol 2011;22(8):784-93.

Wijga AH, Scholtens S, van Oeffelen AAM, Beckers M. Klachten en kwalen bij kinderen in Nederland: Omvang en gevolgen geïnventariseerd. RIVM rapport 260136001 2010. http://www.rivm.nl/bibliotheek/rapporten/260136001.pdf.

250