Article Text
Abstract
Background The aim was to elucidate if the risk of labour market marginalisation (LMM), measured as long-term unemployment, long-term sickness absence, disability pension and a combined measure of these three measures, differed between refugees and non-refugee migrants with different regions of birth compared with native Swedes.
Methods All non-pensioned individuals aged 19–60 years who were resident in Sweden on 31 December 2009 were included (n=4 441 813, whereof 216 930 refugees). HRs with 95% CIs were computed by Cox regression models with competing risks and time-dependent covariates with a follow-up period of 2010–2013.
Results Refugees had in general a doubled risk (HR: 2.0, 95% CI 1.9 to 2.0) and non-refugee migrants had 70% increased risk (HR: 1.7, 95% CI 1.7 to 1.7) of the combined measure of LMM compared with native Swedes. Refugees from Somalia (HR: 2.7, 95% CI 2.6 to 2.8) and Syria (HR: 2.5, 95% CI 2.5 to 2.6) had especially high risk estimates of LMM, mostly due to high risk estimates of long-term unemployment (HR: 3.4, 95% CI 3.3 to 3.5 and HR: 3.2, 95% CI 3.1 to 3.2). African (HR: 0.7, 95% CI 0.6 to 0.7) and Asian (HR: 1.0, 95% CI 1.0 to 1.1) refugees had relatively low risk estimates of long-term sickness absence compared with other refugee groups. Refugees from Europe had the highest risk estimates of disability pension (HR: 1.9, 95% CI 1.8 to 2.0) compared with native Swedes.
Conclusion Refugees had in general a higher risk of all measures of LMM compared with native Swedes. There were, however, large differences in risk estimates of LMM between subgroups of refugees and with regard to type of LMM. Actions addressing differences between subgroups of refugees is therefore crucial in order to ensure that refugees can obtain as well as retain a position on the labour market.
- disability pension
- sick leave
- unemployment
- immigrants
- labour-market marginalisation
- refugees
Statistics from Altmetric.com
Background
Due to different ongoing wars and conflicts around the world, Europe has during the last 5 years experienced the largest wave of refugees since the Second World War.1 Since many of the refugees are suffering from common mental disorders, that is, depression, anxiety and stress disorders like post-traumatic stress disorder (PTSD), they might face difficulties for social integration in terms of labour market integration in the host countries.2 3 Mental disorders often originate from the premigration and migration phases but may be sustained and worsened also during the postmigration phase due to, for example, labour market marginalisation (LMM).4 Moreover, many refugees have a low socioeconomic status when they arrive to the host country, which can in turn deteriorate their health and work ability. Refugees also have a higher prevalence of somatic disorders compared with non-refugee migrants.5 6 This combined vulnerability may lead to a downward spiral of worsening health and socioeconomic hardship after arrival to the host country and may lead to hardship in finding a long-standing position on the labour market.7
LMM can be defined as severe problems in obtaining and retaining a job and has during the last decades been a growing and serious public health challenge as well as an economic crucial issue in Europe.8 Migrants have generally been reported to have lower attachment to the labour market compared with the native population.9 10 One of the very few studies with regard to LMM especially among refugees concluded a high risk of unemployment among refugees with PTSD.11 As marginalisation might deteriorate refugees’ health further, and thereby imply a higher risk of permanent LMM, studies elucidating the risk of LMM among refugees is highly warranted.12–15 Moreover, as previous studies report that migrants and native Swedes have different patterns with regard to having benefits from social insurance, measures of both unemployment and measures based on medical assessments, that is, sickness absence and disability pension, should be used.10 This is of importance in order not to overestimate or underestimate the extent of LMM among refugees.
Aim
The aim was to elucidate if the risk of LMM differed between refugees and non-refugee migrants (regarding their region of birth) compared with native Swedes. LMM was measured as long-term unemployment, long-term sickness absence, disability pension and a combined measure of all three mentioned outcomes.
Methods
Study design
The study is a prospective cohort study where the study population was defined on 31 December 2009. The study population consisted of all individuals, aged 19–60 years, who were resident in Sweden on 31 December 2009 (n=5 092 354). Individuals with ongoing disability pension during 2009 (n=372 046), missing data on reason for settlement to Sweden (n=262 579) or contradicting reasons to migration (n=15 916) were excluded, and the final study population consisted of 4 441 813 individuals, whereof 216 930 (4.9%) were refugees. The cohort was followed 2010–2013 with regard to following measures of LMM: (1) (time to) long-term unemployment, defined as >180 annual days registered as full-time unemployed at the Swedish Public Employment Service during any of the years during follow-up. This cut-off was chosen as most individuals are entitled to extra support by the Public Employment Service after 180 days of unemployment; (2) (time to) long-term sickness absence, defined as the first spell of >90 net days during the follow-up period. Net days means that part-time sickness absence is recalculated into whole days, for example, 2 days on half sickness absence equals one net day of sickness absence. Ninety days of sickness absence is equal to the first assessment in the rehabilitation scheme of the Social Insurance Agency in Sweden; (3) (time to) granting of disability pension; and (4) (time to) a combined measure of LMM, defined as sum of all measures above, and measured in days, in order to get a measure of the total burden of LMM.
Migration status and categorisation of region of birth
Migrants were categorised into refugees, that is, those who came to Sweden due to humanitarian reasons or in need of protection, and non-refugee migrants consisting of, for example, students, family reunion and labour migrants. They were divided with regard to region of birth: (1) Africa (Eritrea, Ethiopia, Somalia and other countries in Africa), (2) Asia (Afghanistan, Iran, Iraq, Syria and other countries in Asia), (3) South America (Chile and other countries in South America), (4) Western countries consisting of European countries outside EU-25 (former Yugoslavia and other countries in Europe outside EU-25, which consists of all current countries in the European Union, except for Bulgaria, Romania and Croatia), (5) Western countries, that is, Nordic countries (Finland, Denmark, Norway and Iceland), EU-25 and North America/Oceania. Native Swedes were defined as all individuals who are born in Sweden.
Variables
Data on long-term unemployment were obtained from Statistics Sweden’s longitudinal integration database for health insurance and labour market studies (LISA) and data on sickness absence (date, grade and duration) and disability pension (date grade and duration) were obtained from the data base Microdata for analysis of social security (Midas) hosted by the Swedish Social Insurance Agency. Reason for settlement in Sweden 1997–2013 was obtained from Statistics Sweden’s longitudinal database for integration studies (STATIV).
Covariates in the analyses were: (1) sociodemographic/socioeconomic factors: sex, age, educational level (low [0–9 years of education], medium [>9–12 years in education] and high [>12 years in education]), type of living area (big cities [Stockholm, Gothenburg and Malmö] and medium-sized cities [cities with >90 000 inhabitants within 30 km distance from the centre of the city]), family situation (married/cohabiting and living together without children, married/cohabiting and living together with children, single without children living at home [including children up to 20 years living with parents]) and in analyses on migrant subgroups, also duration of residence in Sweden (0–5 years, 6–10 years and >10 years). All sociodemographic/socioeconomic variables were measured on 31 December 2009 and were obtained from the LISA database; (2) work-related factors: labour market attachment (income from work and no income from work) and unemployment (no days, 1–179 days and ≥180 days) were obtained from the LISA database. Sickness absence (no days, 1–89 days and ≥90 days) were obtained from the Midas database. All work-related factors were measured during 2009; (3) health-related factors: a record of a main or a side diagnosis from inpatient or specialised outpatient healthcare due to somatic disorders during 2009–2012 (International Classification of Disorders, version 10 [ICD-10]: A00-E99, G00-T99) and a record of a main or side diagnosis from inpatient or specialised outpatient healthcare due to mental disorders 2009–2012 (ICD-10: F00–F99) were obtained from the National Patient Register. Prescription of anxiolytics, sedatives and antidepressants 2009–2012, coded according to the Anatomical Therapeutic Chemical Classification codes N05B, N05C and N06A, respectively, were obtained from the National Pharmaceutical Register. Information on cause and date of death were obtained from the Cause of Death Register. All health-related factors were obtained from databases hosted by the National Board of Health and Welfare.
Statistics
Cox regression models with competing risks and time-dependent covariates were applied. HRs with 95% CIs were computed with native Swedes as reference category. By using a Cox regression model, we can assess time under risk and also take care of competing events, especially important with regard to disability pension, where persons on disability pension no longer are at risk of unemployment and sickness absence. The analyses compared refugees and non-refugee migrants from different countries/regions of birth with native Swedes by including them in the same analytic model: (1) refugees differentiated by region of origin versus native Swedes and (2) non-refugees differentiated by region of origin versus native Swedes. In order to elucidate differences between refugees with different origin, the analyses were stratified by specific countries with a high proportion of refugees in Sweden. Death, emigration and disability pension (in analyses with regard to long-term unemployment and long-term sickness absence) were considered as competing events, that is, an outcome that prevents the occurrence of another outcome, and were censored at the time of that event in the model. Somatic and mental disorders were modelled as time dependent dichotomised variables. All analyses were adjusted in steps for: (1) sociodemographic and work-related factors and (2) additionally health-related factors. Finally, separate analyses on duration of residence among refugees and non-refugee migrants were performed. All analyses were conducted by SAS Statistical Software version 9.4.
Results
Characteristics
There was a majority of men among native Swedes (52.1%) and refugees (58.7%) but not among non-refugee migrants (44.7%) (table 1). Migrants were, in general, younger compared with the native Swedish population. Both refugees (24.2%) and non-refugee migrants (18.1%) had low educational level to a higher extent compared with native Swedes (10.6%). Non-refugee migrants lived to a much higher extent in big cities (58.0%) compared with both refugee migrants (48.1%) and native Swedes (36.4%). A higher share of refugees (44.7%) and non-refugee migrants (50.1%) had no income from work during 2009 compared with native Swedes (15.8%). Refugees had during 2009 higher prevalence of both long-term sickness absence (1.2%) and long-term unemployment (10.4%) compared with both non-refugee migrants (unemployment: 0.8%, sickness: 6.8%) and native Swedes (unemployment: 1.1%, sickness: 2.4%). Refugees had to a slightly higher extent a record of inpatient or specialised outpatient healthcare due to both somatic (34.0%) and mental disorders (12.5%) compared with both non-refugee migrants (somatic: 28.0%, mental: 8.2%) and the native Swedish population (somatic: 28.5%, mental: 10.8%).
Long-term unemployment
Refugees had in the crude model an almost five times higher risk of long-term unemployment (HR: 4.86) and non-refugee migrants almost four times higher risk of long-term unemployment (HR: 3.71) compared with the native Swedish population (table 2). Refugees from Africa (HR: 7.40) and Asia (HR: 5.95) had especially high risk estimates, while refugees from South America (HR: 2.96) and former Yugoslavia (HR: 2.85) had a moderately increased risk compared with the native Swedish population. In the full model, the risk estimates for long-term unemployment among refugees were reduced by half (HR: 2.42). Educational level was most important in lowering the risk estimates in model 1. Also, most of the differences in risk estimates between refugees from different regions were eradicated. South American refugees had, however, compared with native Swedes, still the lowest risk estimate for long-term unemployment (HR: 1.77). The difference in risk estimates was now much smaller compared with refugees from Africa (HR: 3.00), Asia (HR: 2.71) and former Yugoslavia (HR: 1.80). Refugees from Somalia (HR: 3.41) and Syria (HR: 3.16) displayed the outmost highest risk estimates for long-term unemployment compared with the native Swedish population. There were, compared with the native Swedish population, no differences in risk estimates for long-term unemployment between refugees and non-refugee migrants from Africa or South America, respectively, while refugees from Asia and former Yugoslavia had slightly higher risk estimates for long-term unemployment compared with non-refugee migrants from the same countries. Refugees from Africa, Asia and Europe outside EU-25 (all current countries in the European union except for Bulgaria, Romania and Croatia) with shorter duration of residence in Sweden than 6 years had higher risk estimates of unemployment compared with refuges from the same countries with duration of residence longer than 10 years (data not shown).
Long-term sickness absence
Refugees had in the crude model a slightly higher risk for long-term sickness absence (HR: 1.12), while non-refugee migrants had a decreased risk of long-term sickness absence (HR: 0.75) compared with the native Swedish population (table 3). When comparing refugees with regard to region of birth, there were some differences. Refugees from Africa had lower risk (HR: 0.56), refugees from Asia had an equal risk (HR: 1.02) and refugees from South America (HR: 1.32) and former Yugoslavia (HR: 1.49) had higher risk for long-term sickness absence compared with the native Swedish population. In the full model, the risk estimates for long-term sickness absence were unchanged among refugees (HR: 1.12). Adjustment for all covariates resulted in a slight increase of the risk estimates among refugees from Africa (HR: 0.67) and Asia (HR: 1.04) but decreased among refugees from South America (HR: 1.07) and former Yugoslavia (HR: 1.39). With regard to specific countries of birth, refugees from Somalia (HR: 0.38) had an especially low risk, while refugees from former Yugoslavia (HR: 1.39) had the highest risk of long-term sickness absence compared with the native Swedish population. Non-refugee migrants had in the fully adjusted model a slightly lower risk of long-term sickness absence (HR: 0.94) compared with native Swedes, and only non-refugee migrants from Western countries had a significantly higher risk of long-term sickness absence compared with the native Swedish population. Refugees from Africa, Asia and Europe outside EU25 with shorter duration of residence in Sweden than 6 years had significantly lower risk estimates of sickness absence compared with refuges from the same countries with duration of residence over 10 years (data not shown).
Disability pension
In the crude model, refugees had a more than doubled risk of disability pension (HR: 2.29) compared with the native Swedish population (table 4). When comparing refugees from different regions, refugees from Africa (HR: 1.22) and South America (HR: 1.31) had just slightly higher risk, while refugees from Asia (HR: 2.12) and particularly former Yugoslavia (HR: 3.27) had rather high risk of disability pension compared with the native Swedish population. Non-refugee migrants (HR: 0.85) had a slightly lower risk of disability pension compared with native Swedes. In the full model, much of the increased risk among refugees was diminished (HR: 1.39). Mainly educational level contributed to the attenuation of the risk estimates in model 1. Refugees from former Yugoslavia (HR: 2.03) had the highest risk estimates of disability pension compared with the native Swedish population. Non-refugee migrants, from all studied regions, had lower risk of disability pension compared with the native Swedish population. Refugees from Africa, Asia and Europe outside EU25 with shorter duration of residence in Sweden than 6 years had a much lower risk of disability pension compared with refuges from the same countries with duration of residence over 10 years (data not shown).
Combined measure of LMM
Refugees had in the crude model three times higher risk of the combined measure of LMM (HR: 3.00) and non-refugee migrants had just over two times higher risk (HR: 2.16) compared with the native Swedish population (table 5). Refugees from Africa (HR: 3.90) and Asia (HR: 3.47) displayed a high risk for LMM, while refugees from former Yugoslavia (HR: 2.22) and South America (HR: 2.12) displayed moderately increased risk estimates for LMM compared with the native Swedish population. Refugees from Somalia (HR: 4.85) and Syria (HR: 4.34) had especially high risk of LMM compared with native Swedes. Non-refugee migrants from Western countries (HR: 1.69) had the lowest risk of LMM compared with native Swedes. In the full model, the risk estimates were decreased both among refugees (HR: 1.96) and non-refugee migrants (HR: 1.71). Educational level was most important in lowering the risk estimates in model 1. Refugees from Africa (HR: 2.38) and Asia (HR: 2.16) had still the highest risk of LMM, while refugees from former Yugoslavia (HR: 1.64) and South America (HR: 1.46) had moderately increased risk for LMM compared with the native Swedish population. Refugees from Somalia (HR: 2.70) displayed the highest risk for LMM among African countries, while Syrian refugees (HR: 2.54) displayed highest risk estimates for LMM among Asian countries compared with the native Swedish population. Refugees from Chile (HR: 1.41) displayed the outmost lowest risk for LMM compared with native Swedes. Overall, refugees from Africa, Asia and Europe outside EU25 with shorter duration of residence in Sweden than 6 years had significantly higher risk estimates of LMM compared with refuges from the same countries with duration of stay over 10 years (data not shown).
Discussion
Main findings
In general, refugees had higher risk estimates of all measures of LMM compared with native Swedes. These higher risks remained also after adjustment for sociodemographic, work-related and medical factors, even if the differences generally were attenuated. Refugees from Africa and Asia had the highest risk estimates, while refugees from Europe and South America had relatively low risk estimates for the combined measure of LMM. Refugees from Africa and Asia, especially refugees from Somalia and Syria, had high risk estimates of unemployment. Refugees from Europe (mainly from former Yugoslavia) had rather high risk estimates of both long-term sickness absence and disability pension in relation to risk estimates in other refugee groups. Refugees from Chile had rather low risk estimates of all studied measures of LMM. Sociodemographic factors, mainly educational level, explained a large share of the differences between refugees and native Swedes for most outcome measures.
The risk of long-term unemployment was, compared with native Swedes, especially high among refugees from Africa and Asia. Sociodemographic and work-related factors played, however, an important role in explaining differences in long-term unemployment between refugees and the native Swedish population. Both Asian and African refugees had in the final model still an almost three times higher risk of unemployment compared with the native Swedish population and reasons for that may be, for example, mismatch on the labour market due to problems with approval of education from the birth country, language problems but also discrimination.10 16 Moreover, business cycles seem to be of importance for the likelihood to be established on the labour market in the host country. If migrants arrive in Sweden during economic prosperous times, their chances to be established permanently on the labour market will increase.16 This might apply to whole cohorts of refugees from one country or region. There was a large inflow of refugees from Africa during the late 2000s, which was coinciding with the global financial crisis. During the first years of the 1990s, many refugees came to Sweden from former Yugoslavia, which in turn coincided with a large financial crisis in Sweden. Therefore, refugees from different countries might have different initial problems to have gainful work, which might affect the subsequent position on the labour market. Also duration of residence might play a crucial role, as most refugees from Chile came during the 1970s; they have been here for nearly 40 years, while most of the African refugees came during the last years of the 2000s. We have seen that duration of residence was of importance mostly among refugees from Africa, Asia and Europe outside EU25. This has been reported also by previous studies.17
Refugees had, compared with native Swedes, a generally increased risk of all outcome measures of LMM, but there were large differences between regions/countries. Many factors other than illness, such as educational level, eligibility for social security benefits, health status, employability and so on, seem to predict the patterns of LMM.10 14 In order to have benefits from the sickness insurance, a certificate from a physician is required. A study from Norway reported that Somalian refugees often rely on family, friends and their ethnic/religious community, rather than seeking healthcare.12 This may apply also to refugees from other countries and might reduce their likelihood of receiving a certificate from a physician. Moreover, the high risk of unemployment among African refugees likely makes them less eligible for sickness absence, as those benefits require income from previous work.18 Moreover, the higher risk estimates of LMM among refugees might be a consequence of hardships before and during the migration process, which may lead to higher prevalence of mental disorders. The increased risk estimates of LMM might also be due to that family reunion migrants and labour migrants have an established network to a higher extent when arriving to the host country, giving them an advantage compared with refugees.19 There were, however, some exceptions and refugees from Africa had about the same risk estimates for both unemployment and disability pension as non-refugee migrants from the same continent. Reasons for this might, for example, be that family reunion migrants have been seen to have higher rates of sickness compared with labour migrants, and this may eradicate some of the differences between refugees and other migrants.17 Ethnical discrimination and other forms of obstacles on the labour market may have similar impact on both refugees and non-refugee migrants and may hence prevent that migrants in general get a permanent position on the labour market.20 The differences in risk estimates of LMM between refugees, non-refugee migrants and natives might also be due to the healthy migrant effect, that is, that only individuals with good physical and mental health, often in young age, have the strength to leave the home country and start over again in a new country.21 However, even if these migrants may be healthier compared with the general population in their home countries, they are typically less healthy compared with the host populations in most Western countries.21 Reasons for this discrepancy likely include premigration factors such as poor disease prevention in the birth country and higher prevalence of some unhealthy behaviour that might lead to poor health. As an example, smoking is more than doubled among some migrant groups compared with the native Swedish population and has been shown to increase the risk of both physical and mental ill health.15 22 These factors increase the risk of having sickness absence and disability pension. Postmigration factors such as low socioeconomic status in the host country as well as discrimination may be explanations to worse health among migrants compared with the host population.10 As reported by previous studies, medical factors seem, however, to be of rather low importance in explaining differences in unemployment between migrants/refugees and the host population.10
Educational level was one of the most important variables in explaining differences between refugees and native Swedes. Low educational level generally increases the likelihood of having benefits from sickness insurance, and a high educational level in general leads to a lower risk of such benefits.23 Educational level explained much of the difference between refugees/non-refugee migrants and native Swedes for all outcome measures except for sickness absence. There is a considerable heterogeneity among refugees with different origin, where some groups have higher educational level than others. Among migrants from, for example, Chile and Iran, a rather high share of them has completed upper secondary education.13 This might be one explanation to why Chilean refugees have a lower risk of LMM compared with other refugees. However, Iranian refugees display rather high levels of LMM despite a relatively high educational level. Educational level might, however, also be connected to the ability and capacity to gather information about, for example, social insurance regulations, at least during the first time in the country, and hence the possibility to have benefits. We can see that migrants from, for example, Africa, where a relatively high proportion of individuals has just elementary education, had a very low risk of long-term sickness absence.13 Both high and low educational level might therefore paradoxically increase the propensity to have long-term sickness absence and to some extent also disability pension.10 24
Strengths and limitations
Strengths include the use of high-quality data from Swedish nationwide registers providing large study populations with practically no loss to follow-up.25–27 The inclusion of health-related factors as time-dependent variables makes the analyses more accurate. We can thereby consider variation in health and social measures also during follow-up. We have also considered competing events, as individuals having disability pension no longer are at risk for the other outcome measures of LMM. This strengthens the results. Moreover, previous LMM has been taken into account, as joblessness in itself may have detrimental effects on later labour market participation.28 29 Migrants were divided into groups with regard to both reason for settlement as well as for region of birth country, which has not been done previously in this research field.
This study had also some limitations. Somatic and mental disorders were defined by inpatient or specialised outpatient care, which mostly reflect medically more serious cases since individuals treated in primary healthcare could not be included due to lack of register data in national Swedish registers. Although the analyses were adjusted for previous LMM and labour market attachment, there might still be remaining residual confounding.29 30 In this study, we did not stratify by sex, and therefore there might be masked differences between men and women. We did, however, adjust the analyses for sex. The study will mainly be generalisable to countries with similar welfare regimes. We have, however, tried to increase the generalisability to other countries by introducing a combined measure of LMM. Finally, the use of large data sets will increase the chance of finding statistically significant differences between groups. Small significant differences may not, however, be clinically relevant.
Conclusions
Refugees had in general higher risk estimates of all measures of LMM compared with native Swedes. There were, however, large differences with regard to risk estimates of type of LMM and subgroups of refugees. African and Asian refugees had, for example, a particularly high risk of unemployment compared with native Swedes. Efforts should therefore be taken to ensure that refugees can increase their chances to get a long-standing position on the Swedish labour market.
What is already known on this subject
Migrants have in general a higher risk of labour market marginalisation compared with the native Swedish population.
The pattern with regard to benefits from the social insurance is, however, different among migrants compared with native Swedes.
What this study adds
Refugees are reported to have a higher risk estimates of labour market marginalisation compared with non-refugee migrants.
There are also significant differences in labour market marginalisation between refugees from different parts of the world. African and Asian refugees have, compared with the native Swedish population, the highest risk of labour market marginalisation. Efforts should be taken to ensure that vulnerable subgroups of refugees may increase their chances to get a position on the labour market.
References
Footnotes
Contributors MH and EMR conceived and designed the study, with support from MW, TN and FS. MH, MW and EMR were involved in the statistical analysis. MH drafted the manuscript. All authors gave input to drafts and approved the final manuscript.
Funding This study was funded by the Swedish Research Council for Health, Working Life and Welfare, grant number 2016-07194.
Competing interests None declared.
Ethics approval This project was evaluated and approved by the regional ethical review board in Stockholm, Sweden. The ethical committee approval number is 2007/762-31. The ethical review board approved the study and waived the requirement that informed consent of research subjects should be obtained.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The data that support the findings of this study are available from Statistics Sweden, Swedish Social Insurance Agency and The Swedish National Board of Health and Welfare. Restrictions apply to the availability of these data, which were used with ethical permission for the current study and therefore are not publicly available.
Patient consent for publication Not required.