Study design

Between January 2011 and December 2015, baseline HELIUS data were collected among Amsterdam residents of Dutch, Surinamese, Ghanaian, Turkish and Moroccan ethnic origin. More information on the migration backgrounds of the Dutch ethnic minority groups included in HELIUS can be found in table 1. Data were collected through a questionnaire (or interview) and a physical examination that included the collection of biological samples. All participants provided written informed consent.

Recruitment

People in the age range of 18–70 years were randomly sampled, stratified by ethnic origin, through the municipality register of Amsterdam. This register contains data on country of birth of citizens and of their parents, thus allowing for sampling based on the widely accepted Dutch standard indicator for ethnic origin. This country of birth indicator of ethnicity has the advantage of being objective and stable over time, and cross-validation studies showed a high correlation between the country of birth indicator and self-identified ethnic group indicator among Turkish, Moroccan and Surinamese people in the Netherlands.12 More specifically, a person was defined as of non-Dutch ethnic origin if he/she fulfilled one of two criteria: (1) he/she was born outside the Netherlands and has at least one parent born outside the Netherlands (first generation) or (2) he/she was born in the Netherlands but both parents were born outside the Netherlands (second generation). For the Dutch sample, we invited people who were born in the Netherlands and whose parents were born in the Netherlands. A limitation of the country of birth indicator for ethnicity is that people who are born in the same country might have a different ethnic background, which in the Dutch context is applicable to the Surinamese population (see table 1). Therefore, after data collection, participants of Surinamese ethnic origin were further classified according to self-reported ethnic origin (obtained by questionnaire) into ‘African’, ‘South-Asian’ or ‘other’.12

Selected individuals of Dutch, Surinamese, Ghanaian, Turkish and Moroccan ethnic origin received a written invitation combined with written information and a response card (in Dutch, and also in English for Ghanaian participants, in Turkish for Turkish participants and Moroccan Arabic for Moroccan participants). After a positive response, subjects received a confirmation letter of an appointment for a physical examination and a digital or paper version of the questionnaire (depending on the preference of the subject) to fill out at home. Questionnaires were also available in English for Ghanaian participants and in Turkish for Turkish participants. Participants who were unable to complete the questionnaire themselves were offered assistance from a trained ethnically matched same-sex interviewer, speaking their preferred language. Non-Dutch persons who did not respond to the written invitation letter were visited at home by an ethnically matched interviewer to provide additional information if needed (eg, due to language or reading problems) and to assist in filling out the questionnaire in case the subject was willing to participate in the study. At the physical examination, participants were asked whether they had family members aged 18–70 years old (parents, siblings, children and partner) living in Amsterdam who would also be willing to participate. If so, a maximum of three of these family members were also invited to participate. This multigenerational design enables us to study both family relations as well as different migration generations.

Response and participation rates

Of those invited (n=91 609), 28 individuals were deceased before we invited them, and 1562 appeared not to live at the address or recently moved outside of Amsterdam, leading to 90 019 eligible persons for response analyses.

Of the 90 019 persons invited, 35 322 (39%) responded to our written invitation (figure 1). Of those who responded, 20 445 (58%) agreed to participate. Of the 90 019 persons invited, 54 697 (61%) did not respond to our written invitation. We were able to visit 19 307 (35%) of those non-responders at home. Of the 19 307 visited at home, 4344 (23%) agreed to participate, whereas 10 286 (53 %) refused to participate. We were unable to contact 4677 (24%), even after five visits (both during day time and in the evening). Therefore, overall, we were able to contact and get a response from 55% (49 952/90 019) of those invited, either by written invitation or after a home visit by an interviewer. Of those, in total 50% (24 789/49 952) agreed to participate, which we define as the ‘participation rate’. There were modest ethnic variations in response and participation rates as shown in figure 1. The most frequently mentioned reasons for not participating were ‘no time’, ‘not interested’ or ‘having health problems’. Also, 1217 persons who initially agreed to participate repeatedly did not show up at their appointment or could not be reached to arrange an appointment and, for these reasons, did not participate.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow chart of recruitment for the HELIUS study. Participation rate: percentage of participants of those contacted; response rate: percentage of participants of those invited.

Finally, of all 90 019 invited persons, baseline data were obtained from 24 789 participants (28%), which we define as the ‘response rate’. The response rate also showed some variation across ethnic groups (figure 1). The response among family members was somewhat higher (~40%) than among index persons (~25%), but in the end, only 12.9% of all participants were recruited as family members (and part of them were also already selected in the random samples of index persons). Of the 24 789 participants, 23 942 participants completed the questionnaire, 23 012 completed the physical examination including the collection of biological samples and 22 165 participants completed both. Table 2 presents the demographic characteristics at baseline, stratified by the six major ethnic groups. As expected, it shows large variations in both age and educational levels (self-reported highest level of education attained, either in the Netherlands or in the country of origin), within and across the ethnic groups. There were gender differences in educational level in some of the ethnic groups. Among Ghanaian, Turkish and Moroccan participants, the proportion of low education was higher in women than in men, whereas the proportion of medium-low education was lower (data not shown).

Differences between participants, non-participants and those not contacted

Among all ethnic groups, women were more likely to participate than men, as evidenced by the higher percentage of women among participants as compared with that of the total random samples (all invited) (table 3). Among those not contacted, the percentage of women was particularly low (except among Surinamese).

On average, those who participated were slightly older than those who refused to participate (except for the Dutch) or were not contacted. Those not contacted were the youngest among all ethnic groups. Figure 2 (upper panel) shows the response rates by age groups and ethnicity. The highest response was in the age group 45–54 years among the Ghanaian, Turkish and Moroccan groups, whereas in the Dutch and Surinamese groups, the highest response was among those aged 55–64 years. The lower response rates in the Ghanaian, Turkish and Moroccan groups of older age might be due to language difficulties experienced, in particular, in these older groups. Among those contacted, the percentage of those agreeing to participate (participation rate) also differed slightly by age group across all ethnic groups (figure 2, lower panel).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Response rate (upper panel) and participation rate (lower panel) by 5-year age groups and ethnicity. Participation rate: percentage of participants of those contacted; response rate: percentage of participants of those invited (see also figure 1).

To explore to what extent the response was related to socioeconomic status (SES), we constructed two SES indicators based on registry data at the level of the six-digit postcodes maintained by the Department of Research and Statistics of the Municipality of Amsterdam: (1) the average property value of dwellings and (2) the percentage of residents living on a minimum income. This six-digit postcode area is the smallest geographical unit available. On average, these units are 50 by 50 m in size and include 10–20 households. Table 3 shows that both SES indicators are more favourable (ie, higher property value and lower percentage of residents living on minimum income) among participants as compared with non-participants. However, the differences are relatively small, particularly when compared with differences in these SES indicators across ethnic groups. The differences in SES indicators between the ethnic groups are in line with the ethnic differences in individual-level measured educational level as measured among participants (table 2), showing the highest educational level among Dutch (highest percentage of medium-high plus high education) and the lowest among Ghanaians (highest percentage of low plus medium-low education).

Measurements of risk factors and health outcomes

Table 4 provides an overview of the variables that were measured either by questionnaire or during the physical examination (more details on measurements are available from the authors on request). Risk factors include general factors such as physical, behavioural, psychosocial and biological factors, as well as ethnicity-specific characteristics such as culture, migration history, ethnic identity, socioeconomic factors and discrimination. In addition, more extensive measures were measured not in the total study population but in subsamples only. For example, dietary intake was estimated by an additional extensive food frequency questionnaire in a subsample of about 5200 participants.13 Table 5 lists the biological samples that were collected (and, for those who gave permission, also stored) and the laboratory measurements that are already available. Faeces microbiome data will be available from a subsample of about 6000 participants; this will provide a unique cohort to extend research on the role of gut microbiota composition in the development of diabetes, cardiovascular diseases and other health outcomes.14 15

Cross-cultural validity of measurements

Whenever possible, we used standard validated questionnaires. For example, the Patient Health Questionnaire-9 (PHQ-9) was used to measure depressive symptoms,16 the 12-Item short form survey (SF-12) was used to measure quality of life17 and the short questionnaire to assess health-enhancing physical activity (SQUASH) was used to assess physical activity.18 However, because validity studies for these instruments were performed in the Dutch general population or in only a few ethnic groups in the Netherlands only,19 it was unknown whether they had similar validity and reliability for all the ethnic groups included in HELIUS. Validity studies within the HELIUS study indicated that the PHQ-9 20 and SF-12 indeed measure the same concepts in all ethnic groups and that there are no systematic differences in reporting between the groups. Also, the validity of self-reported physical activity with the SQUASH was similar across ethnic groups. However, consistent with findings in the literature, we observed low agreement between self-reported physical activity and objectively measured physical activity by accelerometer and heart rate monitor. In addition, low test–retest reliability was found for the SQUASH in all groups, implying no valid basis for the comparison of physical activity between the different ethnic groups.21

Ethnic inequalities in health

The prevalence of important cardiovascular disease risk factors varies largely between the ethnic groups. Figure 3A (upper panel) shows that, while the prevalence of diabetes in the Dutch group remains below 5%, the prevalence of diabetes ranges from 10% to 12% in participants of African Surinamese, Ghanaian, Turkish and Moroccan origin and is particularly high among the South-Asian Surinamese (20%). Ethnic minority groups also have a 1.3 (Moroccans) to 3.6 (Ghanaians) times higher prevalence of hypertension as compared with the Dutch groups, and hypertension prevalence is particularly high in the two groups of African origin (Ghanaians and African Surinamese).22 The higher prevalence of hypertension among the Turkish and Moroccan groups compared with the Dutch group suggests that risk patterns may be changing unfavourably over time because, just over a decade ago, these groups had a lower prevalence of hypertension compared with the Dutch.23 A large variation is also seen in depressive symptoms, with particularly high prevalence rates of depressed mood among the South-Asian Surinamese (19%), Turkish (23%) and Moroccan (21%) participants (figure 3B) (middle panel).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

The prevalence of diabetes (upper panel), depressive mood (middle panel) and weight status (lower panel) by ethnicity. Diabetes is defined by self-reported diagnosis of diabetes, fasting glucose ≥7.0 mmol/L and/or use of glucose lowering medication; depressed mood defined as a PHQ-9 sum score ≥10; obesity defined as a BMI ≥30 kg/m²; overweight defined as a BMI 25–30 kg/m²; normal weight defined as a BMI <25 kg/m². Afr, African; BMI, body mass index; PHQ-9, patient health questionnaire-9; SA, South-Asian.

Overweight is one of the most important causes of increased cardiovascular risk and is also associated with depression.24 As expected, large differences in overweight and obesity were found between the ethnic groups (figure 3C) (lower panel). While 60% of the Dutch population is considered to have a normal weight based on their body mass index, this percentage is only about 30% in the ethnic minority groups. Particularly in the Ghanaian and Turkish groups, the prevalence of normal weight is very low, and the prevalence of obesity (extreme overweight) is extremely high, that is, up to 35% as compared with 10% in the Dutch. However, the ethnic inequalities in both hypertension and diabetes are not explained by differences in overweight, or other ‘traditional’ risk factors such as fat distribution, educational level or health behaviours.22 25 Ethnic inequalities were not only observed for cardiovascular risk factors, but also for early markers of cardiovascular disease or markers of end-organ damage, such as arterial wave reflection and chronic kidney disease26 27; again, this could not be attributed to traditional cardiovascular disease risk factors. Our results suggest ethnic inequalities in the aetiology of these diseases and emphasise the importance of further research on the underlying causes of ethnic differences in cardiovascular health.

Within HELIUS, we also examined some risk factors that are less often investigated. For example, hand grip strength (a marker of muscle function) was found to be strongly related with diabetes, and large differences in hand grip strength were found between the ethnic groups (highest in Dutch and lowest in South-Asian Surinamese).28 In addition, short sleep duration (<7 hours per night) was more prevalent in the ethnic minority groups as compared with the Dutch,29 and short sleep duration was related to overweight, diabetes and hypertension.30 However, despite these strong relations, both short sleep duration and hand grip strength only marginally contributed to the large ethnic inequalities in cardiovascular risk factors.28 30 Chronic stress is also suggested to contribute to cardiovascular risk, and an important chronic stress factor specific for ethnic minority groups is the level of perceived ethnic discrimination (PED).31 In HELIUS, a positive association was found between PED and the clustering of traditional cardiovascular risk factors (metabolic syndrome) in some ethnic groups, with PED contributing about 5%–7% to the metabolic syndrome in Surinamese and Moroccans.32 PED was also related to depressive symptoms. In ethnic minority groups, PED is reported to account for about 25% of depressive symptoms.33

Human papillomavirus (HPV) is an easily transmittable sexually transmitted infection, and persistent infections with high-risk HPV (hrHPV) types cause nearly all cases of cervical cancer.34 Higher incidences of cervical cancer (the fourth most common female cancer globally) have been observed among ethnic minority women as compared with women of Dutch origin.35 In HELIUS, the seroprevalence of hrHPV seroprevalence in blood, and the prevalence of vaginal high-risk hrHPV infection, differed between the ethnic groups.36–38 South-Asian Surinamese, Moroccan and Turkish women had an overall lower seroprevalence of hrHPV (22%, 14% and 15%, respectively) as compared with Dutch women (30%).38 Vaginal hrHPV prevalence was highest in the Dutch (42%) followed by the African Surinamese (32%), Turkish (29%), Ghanaian (26%), Moroccan (26%) and South Asian Surinamese women (18%).36 When adjusting for sexual risk behaviour, the odds to be hrHPV positive were similar for all ethnic groups. However, because of the higher incidence of cervical cancer in ethnic minority groups compared with the Dutch,35 we expected to observe higher (sero)prevalences of hrHPV in ethnic minority women than in Dutch women. Future studies should explore other factors potentially responsible for the higher incidence of cervical cancer in these ethnic minorities (eg, differences in clearance/persistence of hrHPV infection or differences in participation in cervical cancer screening programmes).

Prevention and healthcare

Results of the earlier Surinamese in The Netherlands: Study on Ethnicity and Health (SUNSET) study in 2001–2003 showed that, at that time, Surinamese participants were less often aware of their high blood pressure and, therefore, were also less often treated, as compared with Dutch participants.39 This suggested a reduced access or quality of healthcare among Surinamese as compared with the Dutch, which could contribute to ethnic disparities in health. Our recent HELIUS results show that, as compared with the Dutch, the Surinamese groups (and the other ethnic minority groups) are currently even more often aware of their high blood pressure and are more often treated for hypertension.22 These results suggest an improvement in hypertension management among ethnic minorities in Amsterdam in the last decade. However, despite medical treatment, ethnic minority groups suffering from hypertension still have lower rates of adequately controlled blood pressure levels.22 A similar pattern is observed for diabetes: although similar or even higher awareness and treatment levels are observed in ethnic minority groups compared with the Dutch, poor glycaemic control is still highly prevalent.25 As poor blood pressure control and poor glycaemic control are serious risk factors for cardiovascular diseases, it is important to investigate the causes of this poor control among the ethnic minority groups. One important factor to consider is the use of healthcare outside the Netherlands. In HELIUS, we observed that some ethnic minority groups use healthcare in their country of origin more often than other ethnic minority groups, ranging from 4% among Surinamese to over 20% among Turkish ethnic groups.40 We need to evaluate to what extent this influences their healthcare use in the Netherlands and their health.

Of all participants included in HELIUS, although 5% reported a history of cardiovascular disease (myocardial infarction, stroke and/or revascularisation), many of them did not have the risk factors under control; that is, 33% was still smoking, 76% was overweight, 55% did not achieve the norm for physical activity, 31% had uncontrolled blood pressure and 58% still had increased low-density lipoprotein (LDL)-cholesterol levels.41 These results show that risk factor control in secondary prevention is poor in the majority of individuals with cardiovascular diseases. The results suggest that secondary prevention should perhaps focus on different risk factors, depending on the ethnic origin of the individual patient.