Ethnic disparities in estimated cardiovascular disease risk in Amsterdam, the Netherlands

Ethnicity was defined according to the country of birth of the participant as well as that of his/her parents [15]. A participant was considered as of non-Dutch ethnic origin if he/she was born abroad and had at least one parent born abroad (first generation); or he/she was born in the Netherlands but both his/her parents were born abroad (second generation) [15]. Surinamese subgroups were classified according to self-reported ethnic origin. Approximately 80% are either of Creole (African) origin or of Hindustani (South Asian) origin.

CVD risk was estimated via the SCORE algorithm based on total cholesterol/high-density lipoprotein cholesterol ratio (TC/HDL) for low-risk countries [16]. This algorithm estimates the ten-year risk of mortality due to coronary heart disease or stroke based on gender, age, blood pressure, TC/HDL and smoking status (yes/no) among those who are 40–65 years of age without diabetes or prior CVD [16]. In addition, we estimated the total cardiovascular risk (mortality plus morbidity) using the Dutch SCORE algorithm (dSCORE), which is derived from the same algorithm, but includes an age-dependent conversion factor between 3 and 5, with a higher conversion factor among younger people, to estimate the ten-year risk of fatal plus non-fatal CVD [5]. The dSCORE algorithm also adds fifteen years to the calendar age among those with diabetes and includes a specific algorithm to estimate CVD risk for participants above the age of 65. In addition, we also estimated total CVD risk using the age-specific conversion factors described by Jorstad et al. [17], which are based on the European Prospective Investigation into Cancer and Nutrition cohort (eSCORE).

Smoking status and prior CVD were assessed by questionnaire. Blood pressure was measured in duplicate using a validated automated digital blood pressure device (WatchBP Home; Microlife AG) on the left arm in a seated position after the person had been seated for at least 5 min. Fasting blood samples were drawn, and glucose and TC/HDL were determined. Participants were considered to have diabetes if they reported a diabetes diagnosis, use of glucose-lowering medication, or in case of a fasting glucose equal or above 7.0 mmol/l [18].

We excluded participants with a Javanese Surinamese (n = 233), ‘other Surinamese’ (n = 267) or unknown/another ethnicity (n = 48). Furthermore, we excluded participants based on missing data regarding cardiovascular risk factors (i. e. age, sex, smoking status, blood pressure, lipid profile or diabetes status) (n = 372) and due to not being eligible for CVD risk estimation using the SCORE algorithm based on age below 40 (n = 7581), age above 65 (n = 868), the presence of diabetes (n = 1,911) or prior myocardial infarction and/or stroke (n = 1,756). We also excluded one participant with unlikely high cholesterol values. This resulted in a study population of 9,128 participants.

Analyses were stratified by gender. We used multiple linear regression analyses to determine ethnic differences in estimated CVD risk, adjusted for age. Because of non-linear associations between age and estimated CVD risk, age was stratified into five-year age categories. Next, we determined ethnic differences in CVD risk factors using logistic regression for dichotomous, and linear regression for continuous CVD risk factors. Similar analyses were performed using the dSCORE and the eSCORE as outcome variable in the same study sample.

For sensitivity analyses, we excluded participants using antihypertensive and/or lipid-lowering medication (n = 1,639) and repeated all analyses. Furthermore, we repeated the dSCORE analyses including participants who are not eligible for CVD risk estimation using the SCORE algorithm, but who are eligible for CVD risk estimation using the dSCORE (in accordance to the current clinical practice in Dutch primary care). We therefore included participants aged 25–55 with diabetes (n = 863) and participants without diabetes above the age of 65 (n = 499).

Fatal CVD risk as estimated by SCORE is higher among South Asian Surinamese men and women, lower among Ghanaian men and similar among all other ethnic groups relative to the native Dutch group. When fatal plus non-fatal CVD risk is estimated using traditional SCORE CVD risk factors plus diabetes, all ethnic minority groups show higher estimated CVD risk relative to the native Dutch group, except for Ghanaian men who show a similar CVD risk. Ethnic groups differ in whether they show a dyslipidaemic, hypertensive or combined CVD risk profile relative to the participants of Dutch origin.

Due to the large multi-ethnic sample size, we were able to determine ethnic differences in estimated CVD risk, stratified for gender. Furthermore, this study compared ethnic differences in both fatal CVD risk and fatal plus non-fatal CVD risk.

A potential weakness is the application of the SCORE algorithm to different ethnic groups, as SCORE is validated mainly among ethnic majority populations [1, 19]. The SCORE algorithm may underestimate or overestimate CVD risk among ethnic minority groups [1]. Moreover, recent studies have identified CVD risk factors which may substantially increase CVD risk independently of traditional CVD risk factors (e. g. psoriasis, human immunodeficiency virus infection or lipoprotein (a)) [1, 20, 21]. These factors may not be distributed equally among ethnic groups [22, 23]. Thus, ethnic differences in true CVD risk may differ from the risk differences as estimated in our study.

As this study was conducted in Amsterdam, results may not be generalisable to ethnic minority populations residing in other cities or countries. Studies in other settings are needed to identify possible common patterns. For example, a study from Norway described a similar higher estimated CVD mortality risk among foreign-born participants relative to participants born in Norway [24].

In contrast to our results, an earlier nation-wide study showed lower CVD mortality among the Moroccan population relative to the population of Dutch origin [2, 3]. This discrepancy may be attributable to risk factors not included in SCORE, but may also reflect that, in this nation-wide study, the ethnic differences in CVD mortality were observed within older participants and were mainly driven by the lifetime exposures to risk factors of these participants.

Among Ghanaians, estimated CVD risk was relatively low compared with other ethnic groups despite a higher prevalence of systolic hypertension. This may be due to a low prevalence of smoking. Earlier nationwide analysis also showed a lower incidence of acute myocardial infarction among Ghanaians compared with the native Dutch [3]. However, estimated CVD risk among Ghanaians may underestimate true lifetime CVD risk, for example due to an early onset of hypertension and thus a longer risk exposure, which the SCORE algorithm does not take into account [25]. If so, current guidelines for antihypertensive treatment may not be optimal for the Ghanaian population.

Our final analyses included diabetes as a CVD risk factor by adding 15 years to the calendar age among those with diabetes, which may result in a great increase in estimated CVD risk [5]. Considering the relatively high prevalence of diabetes among ethnic minority groups, it is not surprising that considering diabetes resulted in greater ethnic disparities in estimated CVD risk. These disparities may be even greater when considering the relatively young onset of type 2 diabetes among ethnic minority groups and thus the relatively long exposure to hyperglycaemia among these groups [10].

This study found greater ethnic differences in CVD risk when CVD risk is estimated based on fatal plus non-fatal CVD instead of fatal CVD alone. This may suggest that ethnic disparities in CVD risk already occur at a young age, as conversion factors to estimate non-fatal CVD risk based on fatal CVD risk are higher at a young age. Accordingly, previous studies reported a younger onset of CVD risk and a higher prevalence or mean of CVD risk factors at a young age among ethnic minority groups relative to the host group [10, 26, 27].

Earlier studies from several countries have reported relatively low access to health care facilities among ethnic minority groups [28]. Therefore, these groups may benefit less from efforts to reduce CVD risk. However, as a result from the universally accessible health care system, ethnic differences in health care access may not be as pervasive in the Netherlands [29, 30]. Thus, in the Netherlands, we do not expect that access to health care facilities will contribute to ethnic disparities in the potential effect of CVD prevention strategies on CVD risk.