We found total avoidable mortality to be slightly elevated for all migrant groups combined compared to the native Dutch population. Cause specific examination showed a higher risk of death among migrants from infectious and several chronic conditions and lower risk of death from malignant conditions. Ethnicity specific investigation showed that the Surinamese and Antillean groups had higher risks of death and Turkish and Moroccan groups had generally lower risks of death from 'avoidable' conditions compared to the native Dutch population. Control for demographic and socioeconomic factors explained a substantial part of ethnic differences in 'avoidable' mortality. Recent immigrants had higher risks of death from infectious diseases, but lower risk from suicides compared to those who resided longer than 15 years in the Netherlands.
Some potential limitations of the data should be considered. First, the power of the study was too limited to allow examination of all causes of death for each ethnic group separately. Second, there is a possibility of an insufficient adjustment for socioeconomic status (SES) since an ecological measure of SES based on income matched on postcode was used. It is likely that further adjustment for SES would provide additional explanation of the higher mortality for some causes in ethnic groups [24
]. Third, the definition of ethnicity is based on available information on country of birth of the subject and both parents. Even though this definition is largely applied in the Netherlands, it does not take into account factors such as ethnic identity, culture, language or ancestry. As a result, it was impossible with our data to describe mortality differences within the four broad migrants groups distinguished in our study. Finally, two selection effects, 'the healthy migrant effect' and 'the unhealthy remigration effect', may have influenced the observed results. Recent studies, however, showed that they fail to explain differences in mortality between ethnic groups in Europe [19
Our selection of causes of death was based on the recent work of Tobias and Jackson, and it aimed to focus on conditions that are primarily avoidable through secondary and tertiary prevention. Despite our effort to prepare a selection in a consistent way, some choices had to be made. One example is our decision to include suicide, which is based on recent evidence on the effectiveness of mental health care services to prevent a considerable part of suicides[23
]. We also included ischemic heart disease (IHD) and stroke, although the contribution of non-medical factors (smoking, nutrition) to the prevention of death from IHD and stroke is large. This decision was based on the advancement in medicine that may have made the healthcare system an important determinant in shaping the patterns of IHD and stroke mortality. In absolute terms, the role of the healthcare system in preventing death from IHD and stroke is higher than for many other conditions combined. Important for the present paper is to note that any modification that may be made to our selection of causes of death, would probably not change the general conclusion that the relative level of mortality greatly varies according to 'avoidable' death, with overall levels being close to the Dutch average.
For diabetes mellitus and leukemia, our standard age interval of 0 to 74 years may be too high, as death at ages of 50 years and over becomes less 'avoidable'. The increased age-limit for diabetes and leukemia, thus, to some extent, overestimates the number of 'avoidable' deaths from diabetes and leukemia. However, it might equally overestimate the mortality risk for both the native Dutch and migrant populations. Our paper focuses on the difference in risk of death from Diabetes between native Dutch and migrant populations. We re-calculated this difference in relative risks of death from diabetes and leukemia for reduced age-limits. We found that this does not substantially change our results and still supports the conclusion that migrant populations have a significantly higher risk of death from diabetes. More specifically, the RR for diabetes in the age-group 0–49 was equal to 3.13, while in the age-group 0–74 this RR = 3.45 (Table ). Similar results were found for leukemia (RR for age-group 0–44 = 0.90 CI: 0.68–1.21 vs. RR for age-group 0–74 = 0.81 CI: 0.60–1.15).
Additional care should be taken when interpreting the role of the healthcare system. Mortality levels are influenced by a series of factors and activities of which health care is only a part. One of the largest effects on ethnic variation in mortality may be produced by variation in incidence of the selected diseases [28
]. Unfortunately, we did not have the incidence data that would be needed to perform additional adjustment for ethnic differences in incidence of infectious diseases. Furthermore, some of avoidable death could be the late consequence of inadequate care in the earlier stages of the disease before arrival to the Netherlands. Despite the problems with the validity and interpretation of the results, our overview could help identify some potential shortcomings in the healthcare system and justify further investigations in particular areas.
The decreased risk of death from ischemic heart disease among Moroccans (RR = below 0.78) might be a reflection of the healthier lifestyle that Moroccan migrants lead as compared to the native Dutch population[19
]. Levels of tobacco consumption were much lower in first generation Moroccans, which is also testified by relatively low levels of lung cancer mortality. Similarly, lower levels of alcohol use and possibly a healthier traditional diet may have protected this migrant group from "western" common cardio-vascular diseases. Similar findings were reported earlier among immigrants in the Netherlands and Germany [29
]. Given current changes in diet and smoking [31
], a higher mortality may however be expected in the future and especially among second generation migrants.
Control for demographic and socioeconomic factors explained a substantial part of ethnic differences in avoidable mortality, sometimes completely abolishing the excess risk. A more comprehensive socioeconomic measure could have explained excess mortality even more substantially [24
]. This indicates that socio-economic factors are important in explaining ethnic differences in mortality in the Netherlands. Similar conclusions were reached earlier by other researchers [32
]. For a few "avoidable" causes of death, however, the situation is more complex, and adjustment for social factors only somewhat attenuated the considerably higher risks. We will discuss in more detail the possible explanations for these causes of death.
The higher risk of mortality from tuberculosis, hepatitis and chronic rheumatic heart disease among ethnic minorities in the Netherlands is likely to be the result of a higher exposure to infectious agents in the migrants' country of origin and, as a result, a higher incidence of these diseases among the migrants [34
]. The high mortality risk can be explained, at least in part, by ethnic differences in the incidence of infectious diseases. Additional factors contributing to the higher risks of death might be substandard housing, overcrowding and poor sanitation that migrants often experience [36
], partly ineffective screening programs[37
], and limited access to healthcare services in the first years after migration. Although generally access in the Netherlands was found to be quite adequate [18
], access in the first years after migration could be hampered due to financial barriers, unclear legal status and limited entitlements to healthcare, and low knowledge on the use of healthcare services. The elevated risk of death from infectious diseases among recent immigrants compared to 'older' immigrants also supports this suggestion.
The observed increased risks of death from diabetes among all four migrant groups is not a surprise and was described earlier in the Dutch literature[39
]. Genetic and behavioral factors were suggested to explain the differences, among them higher low birth weight prevalence [40
] and nutritional differences with higher intake of fat and carbohydrates [41
]. However, some features of the present healthcare system may play an additional role by functioning less adequately for migrant groups and, thus, increasing ethnic differences in health outcomes. These include: (a) lower rate of referrals to the specialists [42
] (b) somewhat less frequent use of primary healthcare facilities and poorer secondary prevention, especially among Surinamese [42
]; (c) difference in the relative importance of risk factors for prediction of outcomes [43
], which is not taken into account in current clinical guidelines [44
]; (d) less efficient communication between providers and patients of non-Dutch origin due to cultural differences in attitudes towards health and healthcare, and illiteracy or inadequate command of Dutch language [45
Elevated maternal mortality among migrant women is another point of concern. It may be related to fertility patterns (migrant women on the average give more often birth to children and, therefore, have a higher risk of maternal mortality per 100,000 person years), but also be related to medical services, such as reported substandard care [46
], delayed prenatal care, higher frequency of unassisted births [47
], and lower use of maternity home care [38
]. Underreporting of maternal [48
] and child [49
] mortality (the last found to be associated with ethnicity) might have hindered assessment of the full extent of the ethnic gap. Elevated maternal mortality is characteristic particularly to Turkish and Moroccan groups and is not elevated among Surinamese and Antilleans. The last observation could be attributed to on average a better integration into the local Dutch society, higher local language proficiency, and more advanced education level of Surinamese and Antilleans compared to Turkish and Moroccans [50