Comparison with previous investigations
Our initial analysis of the total cohort showed that almost all groups had an elevated risk of all the examined diagnoses, disability pension receipt and mortality. Adjustment for age, socioeconomic factors and time since first immigration did not markedly influence the outcomes. A decrease in adjusted HRs compared with crude HRs was found in a study on immigrants in Sweden . It was suggested that the higher unadjusted values can be interpreted as resulting from the immigrants’ experiences and situation in Sweden rather than their experiences before migration. Similar results from other studies have been reported, and one theory is that there is a selection of healthy persons who have the strength and inclination to migrate and settle in a new country, the so-called “healthy migrant effect” [31, 32]. Our findings illustrate the necessity of taking variables such as geographic origin, different diagnoses, employment status, and time since immigration into account, in order to achieve a more nuanced knowledge about immigrants’ health status.
After stratification into geographical origin and employment status, a more diversified risk pattern emerged. Only Nordic immigrants showed higher risks for all studied outcomes. An easier migration process with a more predictable outcome does not require potentially healthy migrant selection. Nordic migrants of both sexes more often occupy blue collar jobs than any other group, and may therefore be exposed to more health-damaging working conditions. They are also accustomed to similar social welfare in their native countries, which makes the Swedish systems more transparent and thereby more accessible for this group. These interpretations may partly explain the increased HRs for the Nordic immigrants, but the results for this group require further investigation.
The examined population was divided into two groups regarding employment status (employed/self-employed and unemployed). The relative wellbeing of unemployed men, except the Nordic group, is puzzling. It may, however, indicate that the higher HRs for the active workforce are caused by their working conditions. It is generally assumed that unemployment increases the risk of poor health, especially psychiatric problems. This study rather points in a diverging direction, since unemployed men presented equal or lower risk of hospitalization for psychiatric disorders than Swedish unemployed men, while all but the EU15+ group among unemployed women presented with increased risk of psychiatric disorders compared with Swedish unemployed women. The possible explanation that immigrants are less likely to seek professional treatment when experiencing mental disorders is partly contradicted by the outcome for employed men and women of all origins. A possible explanation for the elevated risk among the active workforce, not explored in this study, could be the existence of harassment and discrimination towards immigrants in the workplace.
There are studies indicating that immigrants do not seek medical treatment as often as natives because of cultural differences, distrust in Western medicine, or discrimination. This might partially explain low ratios for immigrants compared with Swedes. However, the studied disorders are severe, and normally lead to hospitalization (except musculoskeletal disorders). There are no economic barriers for healthcare utilization among legal immigrants in Sweden, since it is free of charge to everybody regardless of income or other selection criteria.
Most groups displayed HRs >1. A study on Swedish residents was consistent with these findings, and also showed that many immigrant groups had a high prevalence of risk factors for coronary diseases compared with native Swedes, such as smoking, inactivity and obesity . A recent Swedish study on inhabitants in Malmö (n=114 917, 15.2% born outside Sweden) confirmed our findings of increased risk of hospitalization for heart failure among immigrants from Finland and Eastern Europe .
In studies on Finnish male twin pairs where one twin had moved to Sweden, cardiovascular functions were better in the twin in Sweden compared with the one who stayed in Finland [17, 18]. Sweden has a lower heart disorder incidence than Finland and the results may indicate different lifestyle factors influencing heart disorders.
Another study investigated the mortality from heart disorders in different immigrant groups in Sweden . Finnish women showed a significantly higher risk of death from circulatory disease and ischemic heart disorder, as did Eastern European women. These findings are supported by our study. The same researchers also compared differences in coronary heart morbidity and mortality between immigrants to Sweden and residents in their countries of origin [36, 37]. Differences in both directions were found among immigrants from different origins.
For psychiatric disorders the pattern was similar, with high HRs for almost all groups, except surprisingly low risks for unemployed men. In a cross-sectional study an association between immigrants’ low social status and mental illness was found, after controlling for education, employment status, income, economic security, and social support . Studies in Norway found that the level of psychological distress was significantly higher in immigrants from low and middle income countries than in native Norwegians and immigrants from high income countries. The differences between the groups were explained by negative life events, poorer somatic health, difficult economic situation, and lack of social support in poorer migrants. The post-migration situation seemed more important for psychological distress than pre-migration experiences [39, 40]. These findings are partly contradicted by the outcomes for the Nordic group and unemployed men in our study.
Slightly elevated HRs, especially for Nordic immigrants, were found. We did not find any studies on immigrants versus natives regarding musculoskeletal diseases. From other descriptive studies we know that the incidence of osteoarthritis of the hip differs among different ethnic groups, probably due to hereditary factors .
Nordic and non-European immigrants showed HRs >1 for hospitalization for lung disorders. The only comparative studies found between immigrants and natives for lung disorders were attributable to cancer, infectious diseases or earlier exposure to dust, which were not examined in this study.
The eligibility rules for disability pensions are equal for all subjects legally residing in Sweden, natives and migrants alike. The disability pension is related to the individual’s work capacity, and thereby directly linked to health status. This study showed an incongruous relationship between low risk of mortality, moderate risk of hospitalization and significantly increased risk of receipt of a disability pension among immigrants. However, most disorders leading to a disability pension do not call for hospitalization, and a study of all diagnoses resulting in a disability pension would be of interest.
In one of the few studies identified, immigrants in Sweden had, in general, a higher risk of receiving a disability pension than natives . Another study found that receipt of a disability pension differed according to country of birth. For example, it was three times more common among Finns and over five times more frequent among Greeks than among native Swedes. Possible causes for the differences could not be identified, although education, country of birth and marital status appeared to be important factors . Another study examined the risk factors for receipt of a disability pension in a population-based cohort of individuals on long-term sick leave. It found the most important factors to be higher age, low income, previous sick leave, unemployment and non-Swedish origin (odds ratio (OR) 1.3 for men, and 1.5 for women) .
A multilevel analysis of the entire population aged 40–64 years in 2003 in Malmö, Sweden, found that persons originating from middle income countries (World Bank Classification of Country Economies) had the highest likelihood of receiving a disability pension . A study on immigrants in Norway, contradicted our result. It followed up a large health survey, performed in 2000–2001, for receipt of disability pensions 4 years later. The age- and gender-adjusted OR was 2.3. However, when adjusting for occupation, working conditions and income, the OR was reduced to 0.9 .
A study on immigrants in the Netherlands observed that consulting physicians differentiated between native and immigrant patients. The ethnic categorization by the physicians triggered an interpretation of the immigrant patients’ behavior in cultural terms, which reduced the physicians’ capacities to adapt their consulting activities to the needs of the immigrants. The result was that many immigrants remained work-incapacitated for longer periods, with higher risk of ending up in the disability pension program .
There has been an interest for some decades in mortality studies of immigrant populations. In 1984, Marmot reported the results of a systematic review of mortality among immigrant groups in England and Wales. Mortality rates in 1970–1972 were compared with rates in the immigrants’ countries of origin. All-cause male mortality was lower in immigrants from Italy, the Caribbean, and Poland than in the countries of origin, suggesting a selective effect among migrants. The opposite pattern applied for immigrants from Ireland, suggesting that, because of the short geographical distance and cultural similarities, social and health disadvantages may be a stimulus to migration . If so, this could partly explain the increased mortality risk for the Nordic immigrants in our study.
A Swedish register study 1991–1998 reported that all-cause mortality was lower for immigrants to Sweden compared with all-cause mortality in their country of birth , which may indicate a healthy migrant effect, or/and that migrants acquire improved living and health conditions in Sweden. A study of mortality patterns among Canadian immigrants, refugees and non-refugees, 1980–1998, found that immigrants presented lower all-cause mortality compared with the general Canadian population (standardized mortality ratio between 0.34 and 0.58). Mortality rates differed by region of birth and were higher among refugees than other immigrants .
The low mortality risk for all groups in this study, except the Nordic group, whether employed or unemployed, is striking. This might be explained by the “healthy migrant effect”, i.e., select persons with a more favorable life prognosis. Another proposed explanation is that the lower mortality rate of migrants is a denominator effect, caused by immigrants leaving the country without reporting to the authorities, hence their deaths are not recorded [50, 51].
Studies on migrants’ health changes over time often refer to acculturation (e.g., occupational status, language skills, social and cultural integration) as an active factor causing differences between immigrant groups and between immigrants and the native population. Although the definition of the concept is much discussed [52, 53], it has been instrumental in explaining the more favorable health status of minorities compared with natives , as well as the opposite , while other studies found more heterogeneous outcomes . If acculturation is a factor in this study, it tends to agree with the latter, since there are contradictory outcomes, such as the favorable health outcomes for unemployed non-Nordic men (Table 5), and the improved health over time for most male groups, contrasted by the higher risks for most female groups with a longer period of residence (Tables 7, 8).
This study covers documented immigrants and Swedish-born persons aged 28 to 47 in 1990, who were followed in registers from 1991 to 2008. We chose the age of 28 at the lower end of the age spectrum as the baseline in the year 1990 for two reasons: the Swedish social security system has different sets of rules for persons with reduced work capacity younger and older than 28, and it is also a reasonable age to be established in the labor market. The upper end of the age spectrum was 47, as these persons would reach 65, the common age for retirement, in 2008. The studied disorders can, of course, appear later in life, but we do not believe that the distribution among groups of different geographical origins would change markedly.
The strengths of the study are that it included all immigrants matching the selection criteria, and that the Swedish registers are of a high quality, based on personal identification numbers connected to the country of birth, which enables all registered residents in Sweden to be followed in separate as well as in linked registers. Since the aim of the study was to investigate health outcomes for immigrants compared with native Swedes, we decided to adjust for socioeconomic status, measured as education level and income. This study can only offer a limited contribution to the understanding of health, disability pension receipt and mortality among immigrants compared with native Swedes, since the data do not allow us to identify all confounders of the outcomes, e.g., lifestyle, medical history, and co-morbidity not resulting in hospitalization. Some data, such as education level, are self-reported, which can result in underestimation as well as overestimation of the effects of education. We must also point out that we examined only a selection of the four main ICD diagnostic groups. The chosen diagnoses were, however, those known to be potentially work-related. Other studies have shown that migrants have elevated mortality risks and hence higher risks for other disorders, e.g., stroke, diabetes, infectious diseases and certain forms of cancer .
For this study, geographical areas were merged and the outcomes could differ from one country to another within each of the four groups, and would not be detected here.
Employment status and income were registered in 1990. Many individuals have certainly changed their socioeconomic status during follow-up. This is especially probable for young persons who were unemployed in 1990. Changes in occupational status, exposure patterns and socioeconomic position since 1990 could therefore have had an impact on health that is not revealed here. The rough division into white and blue collar work in 1990 is approximate and more detailed information on occupation can give more precise information about the health status of the population. It must also be mentioned that references to previous studies suffer from inconsistencies, since crucial concepts, such as “immigrant” and “native”, are defined differently in different studies.