An accumulation of evidence over the past decade points to differences in morbidity and mortality of ethnic minorities compared with the majority population in many countries.^1–3 Descriptive epidemiological data on injury incidence and mortality consistently show increased rates among indigenous people,⁴ such as Indians and Alaska natives in the USA,⁵ Aboriginals in Australia,⁶ and Maori in New Zealand.⁷ In addition, occasional evidence suggests that migrant populations, especially children, also experience a higher injury risks.^8,9,10 Increased risks for ethnic minority groups are often ascribed to socioeconomic factors or inequalities in medical care.¹¹ In many instances, however, there is no clear cut explanation. Few studies include national level data, or cover all age groups, sexes, and ethnic minority groups.

About 10% of the population in the Netherlands is of non-Western origin with the largest representation of Turkish, Moroccan, Surinamese, and Antillean/Aruban immigrant groups. Ethnic minorities in the Netherlands more often live in low income urbanised areas, which are known to have higher injury mortality rates.¹² Previous Dutch studies found ethnic variations in mortality, including mortality from external causes.^2,10 Children of foreign descent are particularly vulnerable to pedestrian injuries and drowning compared with the native Dutch children.¹⁰ Ethnic variations in injury incidence and mortality beyond childhood have hardly been studied, although an increased risk of homicides among adult immigrants has been reported.^2,13 Limited to specific age groups, type of injuries, and selected determinants, previous Dutch studies thus failed to provide a comprehensive picture of ethnic variations in injury mortality and its major risk factors.

This study is the first to investigate and measure the magnitude of ethnic differences in mortality from a broad range of injuries among different age groups and sexes in the Netherlands. We also analyse the part that area income and urbanisation play in this association and we estimate which part of injury mortality in the Netherlands would have been avoided if all ethnic groups would have similar injury mortality rates as the native Dutch population.

METHODS

Numbers of deaths and population at risk for the period 1995–2000 were obtained from the cause of death register and the municipal population register that includes all inhabitants of the Netherlands with a legal status. These registers were linked using a personal identification number. An open cohort design was used: people were allowed to enter or leave the study (because of birth, death, migration, or administrative corrections) throughout the study period. All deaths that occurred abroad were excluded from the analysis. The data included information on the cause of death, sex, age, marital status, area income (household equivalent income of the neighbourhoods classified into deciles),¹⁴ and urbanisation degree (address density per square kilometre classified into five categories).

Country of birth of the person and both parents was used to measure ethnicity, according to the definition used by Statistics Netherlands.¹⁵ If at least one parent was born abroad, the person was considered to be of non-Dutch origin. In mixed ethnic families, the country of birth of the mother prevailed. We compared deaths rates of the four largest ethnic minority groups residing in the Netherlands (Turkish, Moroccan, Surinamese, and Antillean/Aruban) with native Dutch.

All injuries were classified into three main categories with a further distinction in specific causes of death: traffic related injuries (car driver and passenger, pedestrian, cyclist, motorcycle driver, other), non-traffic injuries (drowning, poisoning, fire and scalds, fall, other), and intentional injuries (suicide, homicide, undetermined event) (table 2). All causes of death were coded according to the International Classification of Diseases (ICD), 9th revision for the year 1995 and 10th revision for the period 1996–2000. Although there is some variability in the codes between ICD revisions, the changes were judged small enough to permit comparability over time.

Relative risks that compared injury mortality rates of ethnic minorities with that of the native Dutch population were calculated using Poisson regression (Stata software, version 7). Relative risks were adjusted for five-year age groups and sex. We performed an additional adjustment to estimate the contribution of area income and urbanisation on ethnic differences in injury related mortality.

The population attributable risk (PAR) was calculated to assess the reduction in cause specific injury mortality rates that would occur if minority ethnic groups experienced the injury mortality rates of the native Dutch population. To calculate PAR, we derived cause specific relative risks for all minority groups combined from the Poisson regression analysis.¹⁶

RESULTS

The native Dutch population contributed the most person time and the largest number of injury related deaths to the analysis (table 1). Turkish, Moroccans, and Surinamese groups were about equally large and Antilleans/Arubans were about three times smaller. All ethnic minorities were more likely to live in poorer and more urbanised areas than the native Dutch population.

All ethnic minorities combined and the native Dutch population followed a similar general pattern with intentional injuries constituting the largest subcategory in absolute numbers of death (8630 for Dutch and 711 for ethnic minorities) and suicide being the most frequent cause of death (table 2). In both ethnic groups, for most causes men experienced a two to three times higher number of deaths from injuries than women.

We found total injury related mortality for the studied ethnic minorities combined to be significantly increased (RR = 1.29) compared with the native Dutch population (table 3). Among traffic injuries, only pedestrian accidents had higher relative risk (RR = 1.87), while relative risks for almost all non-traffic and intentional injuries were significantly increased for ethnic minorities. Mortality among cyclists and motorcycle drivers, in comparison, had a significantly lower risk for ethnic minorities compared with the native Dutch population.

Additional adjustment for urbanisation and area income produced several important effects. In some cases it explained a major part of the mortality difference: a reduction in relative risk of about 50% and more was seen in case of total injury related mortality (RR decreased from 1.29 to 1.11, when adjusted for all factors), the non-traffic injuries group (RR from 1.51 to 1.26), accidental fire/scalds (RR from 1.95 to 1.35), the intentional injuries group (RR from 1.57 to 1.11), homicides (RR from 6.41 to 3.24), and events of undetermined intent (RR from 2.11 to 1.22). In case of poisoning, full adjustment resulted in a complete disappearance of the higher relative risk for ethnic minorities (RR from 1.76 to 0.76). For some other conditions, however, the relative risk increased as a result of full adjustment, as in case of car driver and passenger related accidents (from RR = 0.86 to 1.37). Inequalities in neither drowning nor pedestrian accidents were influenced by urbanisation and area income.

Overall, a higher risk of injury related mortality for ethnic minorities was seen among men, but not among women (RR 1.17 compared with. 0.96, table 4). After control for age, area income, and urbanisation men from ethnic minority groups had an excess risk of death from car driver and passenger related accidents, pedestrian accidents, drowning, other non-traffic injuries, and homicides. Women from ethnic minorities only experienced a higher risk of death from pedestrian accidents, drowning, and homicide.

If ethnic minority groups would experience the same injury related death rates as the native Dutch population, the injury related mortality rate in the total Dutch population would remain almost constant (table 4). A reduction of 1.2% in injury mortality would occur among men and 0.3% increase in injury mortality would occur among women in the total Dutch population. Across all injuries the PAR varied from −5.2% to 15.4%. It was the highest for homicides and drowning in both men and women and took the highest negative values in accidents related to cyclists and motorcycle drivers.

There were large variations in the relative risks of death from injuries according to age group (table 5). The excess risk of total injury related mortality among ethnic minorities was the highest in childhood and young adult age group and gradually reduced with increasing age. Compared with the native Dutch population, children from ethnic minority groups were more vulnerable to non-traffic injuries, especially drowning (RR = 2.90) and accidental fire and scalds (RR = 3.66). Ethnic minorities aged 15–24, in contrast with the native Dutch group of the same age, experienced the largest inequalities in the risk of death from drowning (RR = 5.51) and intentional injuries, especially homicides (RR = 4.55). For the ethnic minority adult population we also found an ncreased risk of death in traffic related injuries, especially in the car driver and passenger and pedestrian death risks (RRs = 1.58 and 2.92, respectively). The excess mortality from drowning among migrants groups was high in all age groups, except the oldest. Homicide was high in all age groups except the youngest.

Not all ethnic groups had similar rates of injury related mortality. As table 6 shows, the Surinamese and Antillean population seemed at an increased risk of the total injury related mortality (RRs above 1.30), while the total injury mortality of Turkish and Moroccan groups did not significantly differ from the native Dutch population (RR = 0.91). Car driver and passenger, pedestrian, drowning, and homicide mortality remained increased among almost all ethnic minority groups. In contrast with Turkish and Moroccans, Antillean/Aruban and, especially, Surinamese populations had higher risks of death from almost all non-traffic injuries (most RRs above 1.20) and higher rates from suicides. Compared with other ethnic groups, Antillean/Aruban had an exceptionally high homicide rate (RR = 7.13).

DISCUSSION

Our results show that, compared with the native Dutch population, ethnic minorities combined had an increased injury mortality rate. Ethnic minorities experienced a higher risk of death from pedestrian accidents, drowning, poisoning, fire and scalds, and homicides. Mortality for cyclists and motorcycle drivers was significantly lower among ethnic minorities compared with the native Dutch. Adjustment for urbanisation and area income decreased the mortality risk difference for most non-traffic injuries, but showed the difference in risk for car driver and passenger accidents. Injury mortality among ethnic minorities was lower for cyclist accidents, motorcycle driver accidents, and suicides. Compared with the native Dutch inhabitants, Surinamese and Antillean/Aruban population had a higher risk of total injury mortality, while Turkish and Moroccans had increased risk only for selected conditions. Inequalities in injury mortality were the highest among children and young adults, but persisted in the age group above 50 years old.

In interpreting the results of our study, some potential limitations should be considered. Firstly, the power of the study was too limited to show with statistical significance possible ethnic differences for specific injuries, age-sex groups, and ethnic groups. Secondly, all deaths that occurred abroad were excluded from the analysis. We should therefore emphasise that our results are restricted to injury related deaths that occurred within the Netherlands. Injury related deaths abroad, which may be considerably higher among ethnic minority groups, require specific research, as they require specific approaches for injury prevention

Several factors might have played a part in explaining the ethnic differences in injury mortality found in our study.

Higher levels of car driver and passenger related accident mortality among minority groups could probably be explained by a higher injury risk per kilometre travelled instead of a higher rate of exposure. Ethnic minorities, who mostly live in urban areas, possess and use cars less often than native Dutch inhabitants. This may compensate for a potentially higher injury risk per kilometre travelled among ethnic minorities. This higher risk is suggested by the increase in relative risk (from 0.99 to 1.48) after adjustment for urbanisation and an excess risk among men (RR = 1.51), primary car users, but not among women (RR = 0.98).

A higher risk per kilometre travelled could be partly related to a more risky driving behaviour, such as speed limit excess and less common use of car safety equipment, as is frequently found among ethnic groups in other countries.^17,18 This is supported by the general more risk taking behaviour as seen from the higher criminality among ethnic minorities¹⁹ and also could be related to less frequent use of these protective measures in immigrants’ countries of origin. In contrast with the findings in the USA,²⁰ driving under the influence of alcohol is less likely among ethnic minority groups in the Netherlands as alcohol consumption is consistently lower in these groups²¹ because of religious reasons. Less frequent drink driving behaviour is an important consideration for future interventions.

Mortality during traffic accidents is also highly dependent on the age of the car. Newer car models are more often equipped with functional built in safety equipment such as multisided airbags, headrests, ABS systems, speed alerts, and other technologies that help to prevent the collision and to protect against death and severe injuries in case of a collision. Ethnic minorities might have fewer possibilities to procure newer car models because of their lower income, compared with Dutch.²²

Bicycles and motorcycles serve as important means of transportation for the Dutch population, especially in densely populated areas. The amount of kilometres ridden by cyclists grows every year. According to CBS data, there were 187 billion kilometres ridden by cyclists in 2000, which is an increase by 6% compared with 1995.²³ Bicycle and motorcycle use in countries where immigrant populations originate from is much less frequent than in the Netherlands. Ethnic minorities do not favour these types of transportation even after immigration to the Netherlands. This might explain their lower risk of cyclist and motorcyclist mortality, compared with the Dutch population.

Less frequent use of bicycles and motor vehicles makes immigrant groups, especially women, take more often the role of pedestrians. High pedestrian exposure among ethnic minority groups might explain their increased mortality, a phenomenon that has been seen also in other countries.^24–26 The observed higher risk among ethnic minority women compared with men (RR = 2.04 and 1.54 respectively) provides additional support to the above suggestion.

Drowning was more frequent among ethnic minorities of all ages up to 50, with the highest relative risk found among young adults aged 15–24 (RR = 5.51). The ability to swim seems an obvious explanation for the observed difference in drowning mortality risks. In Amsterdam only 56% of the 13–14 year old Turkish and Moroccan girls had a swimming certificate compared with over 95% of the native Dutch children.²⁷ Although swimming is a part of the school curriculum, not every school has appropriate facilities and lessons often have to be arranged and covered privately, which is not always an option for the ethnic minority families. Because higher risks of drowning are not only restricted to younger age groups, but persist at older ages, swimming educational programmes for ethnic minorities of all ages are warranted.

Differences in risk of death from traffic and, especially, non-traffic injuries could be also related to differential mortality from occupational accidents. As ethnic minorities are more often employed in low skilled manual jobs where they are often exposed to occupational hazards, it is plausible that there are differences in mortality from occupational injuries between ethnic minorities and the native Dutch population. This suggestion is supported by the fact that we found an excess of risk of death from fire, scalds and non-traffic injuries especially among ethnic minority men.

The high homicide mortality risks for ethnic minorities found in our study are in line with the data from earlier Dutch publications.^2,28,29,31 Several factors could be involved. Firstly, homicide is related to criminal affairs,²⁸ and therefore, higher involvement in criminal acts might partly explain the above finding. According to the court registries, 31% of prosecutions in the year 2000 occurred among first generation immigrants who were directly involved in criminal acts (an 11 times higher rate than that of the Dutch population).³² Turkish people were the least involved in criminal acts, while Antilleans were the most involved.^30,32 Secondly, a lower socioeconomic position of the migrants, stress associated with work, discrimination, and culture shock were previously suggested to explain high homicide rates among ethnic minorities.^30,31 Thirdly, psychosocial disorders, especially frequent among Surinamese and Antilleans could add to the explanation of the increased homicide risks.³³ Fourthly, family attitudes and rearing practices could lead to a more aggressive behaviour. A study of adolescents in Rotterdam found that only 30% of Antilleans and 52% of Surinamese teenagers were growing up in a full (with two parents) family, whereas that was the case for 77% of the Dutch.³⁴ Additionally, ethnic minority teenagers, especially Antilleans, more often expressed an aggressive and violent behaviour, committed vandalism or theft, or possessed a weapon.³⁴

Compared with the native Dutch population, the risk of death from suicides is consistently lower among Turkish and Moroccan minorities. Cultural norms, higher religiousness, and strong and supportive family ties could have an important role in explaining the low risks. The situation is different for the young adults of Surinamese and Antillean descent, for who the suicide mortality risks are higher compared with the native Dutch population (RR = 1.58 and 1.38 respectively, data not shown). The latter could be partly associated with the higher rates of mental disorders in these populations.³³

We found an increased risk of death from events of undetermined intent among ethnic minorities (RR unadjusted = 2.11). This finding might be partly attributed to a slightly less accurate cause of death registration among ethnic minorities than among the native population. Differential misclassification of causes of death by ethnic origin could have resulted in an incorrect estimate of the difference in homicide and suicide mortality between ethnic minority groups and the native Dutch population. Apart from intentional injuries, we do not have reasons to suspect occurrence of differential misclassification among other causes of death.

Access to healthcare is an important factor for injury mortality, as prompt and high quality emergency services may prevent death. In the Netherlands no pronounced inequalities in access to health care facilities were found for ethnic minorities,^35,36 therefore, such inequalities are unlikely to play a part in explaining ethnic differences in injury mortality.

The results of this study point to areas that require priority attention from both researchers and policy makers. Activities should be aimed at prevention of car and pedestrian accidents, drowning, and homicides among ethnic minority groups most at risk. Up to 15% of homicide deaths and 10% of drowning would have been avoided in the general Dutch population if ethnic minorities had the same mortality rates for these causes as the native Dutch inhabitants. Additional research is needed to establish the specific determinants for the increased mortality among ethnic groups and to identify ways to effectively tackle them.

The study also shows opportunities for preventive activities that target the native Dutch population. A substantial percentage of lives of bicycle and motorcycle riders would have been saved among the native Dutch people if they had mortality rates similar to the ones of ethnic minorities. Although safety helmets were shown to protect cyclists against heavy injuries and death,^37–39 their use among native Dutch is still low,⁴⁰ thus warranting appropriate interventions.

The results of our analysis cannot be directly generalised to other countries because of differences in migrants groups and national context. None the less, similar patterns may be seen, for example, among the Turkish and Moroccan migrants living in Germany and France. In addition, some of the priority areas identified in our study, such as the high mortality from pedestrian accidents, poisoning, and fire, may be similar in other countries with a different configuration of migrant groups. Our study, therefore, needs replication in other European countries where injuries among migrant groups have not yet been described in detail. International cooperation and exchange of findings may contribute to explaining these inequalities and developing effective ways to reduce them.