Disability pension and mortality in individuals with specific somatic and mental disorders: examining differences between refugees and Swedish-born individuals

Background More than half a million refugees have arrived to Sweden during the last decade. The aim was to investigate differences between refugees and Swedish-born individuals regarding different specific somatic and mental disorders, and subsequent disability pension and mortality. Methods All refugees (n=239 742) and Swedish-born individuals (n=4 133 898), aged 19–60 years, resident in Sweden on 31st of December in 2009 were included in this population-based prospective cohort study. Data from six nationwide Swedish registers were linked by the unique anonymised identification number. HRs with 95% CIs were computed for disability pension and mortality 2010–2013 by Cox regression models. Results Compared with their Swedish-born counterparts with mental or somatic diagnoses, refugees with these diagnoses had a higher risk of subsequent disability pension and a lower risk of mortality. Highest estimates for disability pension were seen for refugees with neoplasm (HR: 1.72; 95% CI: 1.56 to 1.91), musculoskeletal disorders (HR: 1.57; 95% CI: 1.47 to 1.67), diseases of the circulatory system (HR: 1.33; 95% CI: 1.22 to 1.45), depressive disorders (HR: 1.31; 95% CI: 1.21 to 1.41) and diabetes mellitus (HR: 1.30; 95% CI: 1.15 to 1.47). The risk of mortality was lowest for refugees with regard to bipolar disorders (HR: 0.37; 95% CI: 0.16 to 0.82), post-traumatic stress disorder (HR: 0.37; 95% CI: 0.25 to 0.54) and least pronounced in regard to neoplasm (HR: 0.69; 95% CI: 0.61 to 0.77) compared with Swedish-born with similar disorders. Conclusion Refugees have a generally higher risk of disability pension compared with Swedish-born with specific somatic and mental disorders. Despite this, refugees with all specific disorders have lower risk estimates of mortality, probably due to a healthy selection. The higher risk of disability pension might therefore be due to other causes besides poor health.


INTRODUCTION
Sweden has, like several other European countries, become a multicultural society due to increasing global migration. Of note, more than half a million refugees arrived to Sweden during the last decade, and many of them have experienced healthimpacting traumatic life events and bereavement in their home countries and during the flight. 1 This development puts refugees' health issues on the receiving societies' public health agenda. In the 1980s and 1990s, most of the refugees originated from Iran, Iraq, Turkey and former Yugoslavia; and from the beginning of the 2000s, also many refugees from northeast Africa arrived to Sweden. 2 Most refugees arrived to Sweden by themselves, but about 11% of all refugees during the period from 1980 to 2007 arrived to Sweden as resettled refugees on a quota negotiated by the Swedish government and the United Nations High Commissioner for Refugees. 2 A higher prevalence of somatic diseases, for example circulatory, respiratory and digestive disorders, compared with the host population, has been found among migrant subgroups. [3][4][5][6] Moreover, the prevalence of mental disorders is high among refugees. 7 This combined vulnerability may start a downward spiral of worsening health, leading to socioeconomic hardship after arrival in the host country. 8 In addition, poor accessibility to healthcare, less propensity and fear of seeking healthcare due to structural, cultural and lingual barriers among refugees might, on the one hand, worsen the course of the illness, and on the other hand add to the difficulties of finding a position at the labour market. [9][10][11][12] The link between premature exit from the labour market due to health reasons, that is, disability pension and subsequent premature death, is well described in the literature. 13 14 Still, these two outcomes might be differently driven by adverse health, poor socioeconomic conditions and negative health behaviour. 15 16 As most other studies, both generally and from our research group, have a focus on mental disorders, 7 12 17-19 studies that also elucidate the role of specific somatic disorders in the pathways to mortality and disability pension among refugees are warranted. There are to the best of our knowledge very few longitudinal studies related to specific mental disorders and no studies that are related to specific somatic disorders and subsequent adverse social and health outcomes, measured as disability pension and mortality among a refugee population.
Many studies have paradoxically found that immigrants in general have a mortality advantage compared with individuals in the host population. [20][21][22][23] Given that immigrants form a heterogeneous group, within-group differences in mortality on May 23, 2022 by guest. Protected by copyright. Original research risk might be expected. Refugees, in particular, might be at higher risk for adverse health and social outcomes due to the high likelihood of traumatic and life-threatening experiences, and the lack of access to healthcare during the often long journeys to Western host countries. The aim was to study the associations between commonly occurring somatic and mental disorders and subsequent disability pension and mortality among refugees, and to investigate if there were differences in such associations between refugees and Swedish-born individuals.

Study population
All refugees and Swedish-born individuals between 19 and 60 years of age, who were resident in Sweden on 31st of December 2009, were included in the study base (n=4 376 055). Refugees with no data on reason for settlement to Sweden (n=307) or contradicting information on reasons of migration (n=2108) were excluded. The final population comprised of 4 373 640 individuals, of which 239 742 were refugees and 4 133 898 were Swedish-born individuals. In analyses regarding disability pension as outcome measure, individuals with baseline disability pension (n=314 922) were excluded. Refugees were in this study defined as: (1) individuals who received a residence permit in Sweden according to the rules in the United Nations Geneva Convention of Refugees from year 1951, (2) individuals who received a residence permit in Sweden due to humanitarian reasons, or (3) individuals who, due to need of protection, including resettled refugees, received a residence permit in Sweden. A sensitivity analysis, which compared each of these groups separately with Swedish-born individuals, was performed and showed similar results.

Register data
Data from different nationwide Swedish registers were linked by the unique anonymised identification number, from the following agencies:

Outcomes
The cohort was followed 2010-2013 with regard to: (1) disability pension and (2) mortality. In Sweden, all residents between 19 and 64 years of age can be granted disability pension if their work capacity is permanently or in the long term reduced due to disease or injury. Basic levels of disability pension can be granted without previous income from work. As opposed to the eligibility criteria for unemployment benefits, the requirements for being granted disability pension include work incapacity related to a disease. For this reason, the decision to receive disability pension is based on medical assessments and is often (not always) preceded by long periods of sickness absence. While being granted sickness absence requires income from work, basic levels of disability pension can be granted without previous income from work. The regulation regarding disability pension was tightened in 2008, and only individuals 30 years of age and above with an expected life-long disability could be granted permanent disability pension in Sweden. However, young adults (19-29 years of age) can still be granted temporary disability pension if their work capacity is reduced or compulsory school education not completed.

Covariates
The covariates included in the analyses were (

Statistics
First, after checking that the assumption of proportional hazards was met, crude and multivariate Cox regression models were used. HRs with 95% CIs were computed in relation to disability pension and mortality among Swedish-born individuals and refugees with and without somatic/mental disorders, with Swedish-born individuals without any somatic/mental disorders as the reference group. Thereafter, crude and multivariate adjusted HRs for diagnosis specific somatic/mental disorders (at ICD-10 group level) were calculated for refugees compared with Swedish-born with the same diagnoses. With regard to analyses of general somatic/mental disorders, the analyses were adjusted in three steps: model 1, sociodemographic factors; model 2, in addition adjusted for work-related factors, that is, income from work and unemployment; and model 3, in addition adjusted on May 23, 2022 by guest. Protected by copyright.
http://jech.bmj.com/ for health-related factors, that is, sickness absence and occurrence of a main or a secondary diagnosis of a mental/somatic disorder between 2006 and 2009. In addition, sex-stratified analyses were performed regarding pooled data of somatic and mental disorders. With regard to analyses on specific somatic/ mental disorders, the analyses were only adjusted in one step, as in model 3 above. All analyses were performed with SAS statistical software, V.9.4.

RESULTS Characteristics
There was a higher proportion of men among refugees (58%) compared with Swedish-born individuals (51%) (table 1). Most of the refugees arrived from former Yugoslavia (29%), Iraq (20%), Iran (11%) and other countries in Asia (12%). Refugees had in general a lower educational level and lived to a higher extent in big cities compared with Swedish-born individuals. A substantially lower proportion of refugees (51%) had income from work compared with Swedish-born individuals (80%). Refugees had also a higher prevalence of long-term unemployment, rather equal prevalence of long-term sickness absence, as well as a higher prevalence of both mental (except for bipolar

Disability pension Somatic disorders
In the crude model, refugees with somatic disorders had almost seven times higher risk for disability pension compared with Swedish-born individuals without somatic disorders (table 2). The risk estimates were also two times as high compared with Swedish-born individuals with somatic disorders. In the final model, refugees with somatic disorders (HR: 2.7; 95% CI: 2.5 to 2.8) had just slightly higher risk estimates of disability pension compared with Swedish-born individuals with somatic disorders (HR: 2.0; 95% CI: 2.0 to 2.1). No sex differences in these associations were found. When the analyses were stratified by specific somatic disorders, the crude model showed that refugees with all subgroups of specific somatic disorders had a significantly higher risk for disability pension compared with Swedish-born individuals with these disorders (table 3). In the full model, however, there was a substantially higher risk for refugees compared with Swedishborn individuals only for those with neoplasm, diseases of the musculoskeletal system, diabetes mellitus, diseases of the circulatory systems and within the group with other somatic disorders (HR range: 1.3-1.7). Foremost work-related and health-related factors altered the risk estimates of disability pension.

Mental disorders
Refugees with mental disorders had in the crude model a 19-fold higher risk for disability pension compared with Swedish-born individuals without any mental disorder (table 4). The risk estimates for refugees with mental disorders were still higher compared with Swedish-born individuals with mental disorders in the full model (HR: 8.0; 95% CI: 7.7 to 8.4). The risk estimates for disability pension (compared with Swedish-born women/men without mental disorders) were slightly lower among refugee women with mental disorders (HR: 7.3; 95% CI: 6.8 to 7.8) than among refugee men with mental disorders (HR: 8.8; 95% CI: 8.3 to 9.4, figures not shown).
Stratified analyses on specific mental disorders showed that refugees had a higher risk for disability pension compared with Swedish-born individuals with all specific disorders (table 5). In the full model, however, there was a substantially higher risk for refugees compared with Swedish-born individuals only for those with depressive disorders (HR: 1.3; 95% CI: 1.2 to 1.4).

Somatic disorders
In the crude model, refugees with somatic disorders had an around 60% higher risk of mortality compared with Swedishborn individuals without somatic disorders (table 2). Refugees with somatic disorders had, however, a slightly lower risk estimate of mortality compared with Swedish-born individuals with somatic disorders. In the full model, refugees with somatic disorders had lower risk estimates for mortality (HR: 0.9; 95% CI: 0.9 to 1.0) compared with Swedish-born individuals with somatic disorders (HR: 1.7; 95% CI: 1.7 to 1.8), and noticeably lower risk than Swedish-born individuals without somatic disorders (reference group). No significant sex differences in these associations were found.
With regard to specific somatic disorders, refugees had, compared with Swedish-born individuals with similar disorders, a significantly lower risk of mortality in all the subgroups in the full model (HR range: 0.5-0.7; table 3). The estimates closest to those of the Swedish-born individuals were for neoplasms.

Mental disorders
In the crude model, refugees with mental disorders had a more than doubled risk of mortality compared with Swedish-born individuals without mental disorders, but a markedly lower risk estimate of mortality compared with Swedish-born individuals with mental disorders (table 4). Also in the full model, refugees with mental disorders had considerably lower risk estimates for mortality (HR: 1.1; 95% CI: 1.0 to 1.2) compared with Swedishborn individuals with mental disorders (HR: 2.2; 95% CI: 2.2 to 2.3), and hence an equal risk compared with Swedish-born individuals without mental disorders (reference group). The risk estimates for mortality (compared with Swedish-born women/men without mental disorders) were lower among refugee women with mental disorders (HR: 0.8; 95% CI: 0.7 to 0.9) than among refugee men with mental disorders (HR: 1.3; 95% CI: 1.2 to 1.4, figures not shown).

Original research
With regard to specific mental disorders, refugees had, compared with Swedish-born individuals with similar disorders, as also shown regarding the somatic disorders, a markedly lower risk of mortality during the follow-up in all the subgroups (HR range: 0.4-0.5; table 5).

Main findings
Refugees had to a higher extent both mental (except for bipolar disorder) and somatic (except for neoplasm) morbidity compared with Swedish-born individuals. Additionally, refugees with specific somatic/mental disorders, especially neoplasms, diseases in the musculoskeletal system and depressive disorders had higher risk estimates of disability pension compared with Swedish-born individuals with similar disorders. Refugees with all subgroups of specific somatic and mental disorders had lower, mainly around half the risk of mortality compared with Swedishborn individuals with similar disorders.

Disability pension
We found a considerably higher risk of disability pension among refugees with specific disorders compared with Swedish-born individuals with the same disorders. With regard to somatic disorders, this is also in consensus with previous scientific knowledge on immigrants in general. 26 27 Especially among immigrants from low-income countries, this has been explained by poor working conditions, for example, work in unhealthy environments or irregular working hours. [26][27][28] In this study, refugees were found to have a higher prevalence of most subgroups of somatic/mental disorders compared with Swedish-born individuals. When comparing refugees with mental/somatic disorders with Swedish-born with similar disorders, refugees might have a higher work incapacity related to these disorders than Swedish-born individuals as they might have sought healthcare at a later stage due to the known restrictions in access to care for refugees. 12 Moreover, more pronounced stigma of mental disorders among refugees and lack of competence in transcultural medicine of the treating physician might result in incorrect diagnostics and subsequently treatment and prognosis. As an example, musculoskeletal disorders among refugees might, to a higher extent, reflect somatisation or symptoms of mental disorders. 17 29-32 The analyses also revealed that comorbid mental disorders explained a part of the higher risk for disability pension among refugees with somatic disorders compared with Swedish-born individuals with somatic disorders. Comorbid disorders might therefore give refugees a higher rate of work incapacity compared with Swedish-born individuals. In order to better understand this phenomenon, studies that assess the interaction of specific mental and somatic comorbidities in the pathways to disability pension in refugees are highly warranted.
Other reasons for why refugees may encounter obstacles in establishing themselves on the labour market, which in turn may eventually lead to disability pension, may be low employability, due to existing structural barriers such as discrimination and lack of skills with regard to the demands on the labour market in the host country. 33 Also, low health literacy and lower helpseeking behaviour among refugees, resulting in a non-optimal treatment, may be explanations to the higher risk of disability pension among refugees. 10 34 As low educational level is more prevalent among refugees, low employability leading to marginalisation at the labour market might be another explanation to higher levels of disability pension among refugees. 2 In addition, the design of the social security system where sickness absence cannot be granted without previous work may promote a higher level of disability pension among refugees, as also those without gainful employment are eligible to such pension. Although many factors may interact and start a downward spiral of deteriorating health and social marginalisation, which eventually might lead to premature exit from the labour market, also specific somatic and mental disorders may be key drivers in this regard. This study is one of the first to investigate the association between specific somatic and mental disorders and subsequent disability pension among refugees. Findings do not, however, differ considerably compared with earlier studies on immigrants in general. 27 35 This might suggest that refugee-specific risk factors like traumatic life experiences do not contribute considerably to the disabling process of becoming work incapacitated. Still, the pathways to disability pension in refugees seem to be effected by the poor labour market attachment in this group, as controlling for information on income from work had a strong effect on the risk estimates. This effect might to some extent be diagnosis specific as the estimates related to some diagnoses regarding subsequent disability pension yielded low or non-significant differences between refugees and Swedish-born individuals.

Mortality
Despite higher prevalence of both somatic and mental morbidity, as well as high levels of disability pension, refugees with specific disorders had around half the risk of mortality during the follow-up period compared with Swedish-born individuals with the same disorders. Similar findings have been shown for immigrants in general and researchers have been puzzled by this paradox. 21 23 36 We could now replicate these findings for refugees with several specific somatic and mental disorders. Several possible explanations have been suggested, for example, a positive health selection of immigrants, that unhealthy immigrants remigrate (salmon bias), cultural features such as better food habits, but also pure flaws in the data collection. 21 23 A study tested all these hypotheses and concluded that the most plausible explanation is that the immigrant population seems to be highly selected due to the fact that many foreign-born with lethal disorders or vulnerabilities may already have died in their home country or during the flight or that they have not been able to make the journey due to health reasons. 23 In this study, despite of reports on high levels of exposure to traumatic life experiences among refugees, 37 we found a lower risk of mortality in refugees, which is in line with previous findings. However, asylum seekers do not seem to have a mortality advantage compared with the host population, 10 36 thus the health selection process seems to involve the whole process from living in the home country until residence permit in the host country is granted.
The risk estimates for specific disorders in relation to subsequent mortality did not differ strongly with diagnoses. Moreover, risk estimates were only marginally affected when controlling for the various covariates. This suggests that there are no major differences in the risk for and in the pathways to preterm mortality for specific diagnoses among refugees compared with Swedish-born individuals. Moreover, in contrast to reported findings on lower treatment rates for refugees, 9 10 38 our findings do not suggest major disadvantages for refugees in receiving adequate healthcare after a diagnosis.

Strengths and limitations
The prospective cohort design with several years of longitudinal observation and use of high-quality population-based Swedish nationwide register data 39 can be regarded as the main strengths of this study. Register data on the working-age population of Sweden was included from different sources and thereby we avoided recall and selection bias. Moreover, there was no loss to follow-up and all data on morbidity were based on physicianbased diagnoses-that is, it was not based on self-reports. We also had the opportunity to include a wide range of potential confounders derived from different registers leading to results with better validity.
Also limitations of this study should be mentioned. Our data on morbidity were based on specialised care, and data on primary healthcare visits were not available. For this reason, we most likely had a selection of cases with higher medical severity. This might have led to an overestimation of the risk for disability pension or mortality among persons with specific disorders. The quality of the Swedish NPR regarding the specific diagnoses has been shown to be good, 40 thus, misclassification due to exposure is unlikely.

CONCLUSIONS
Refugees have a higher prevalence of most somatic and mental disorders compared with Swedish-born, and refugees with specific somatic/mental disorders have a higher risk of disability pension, but a lower risk of mortality compared with Swedishborn with these somatic/mental disorders. Possible explanations of this discrepancy include the effect of a positive health selection among refugees who resettle in Sweden. The higher risk of disability pension among refugees may hence have other causes than worse health status among refugees.

What is already known on this subject
► Migrant subgroups have in general been reported to have a higher prevalence of somatic and mental disorders, as well as a higher risk of disability pension compared with Swedishborn individuals.

What this study adds
► We now know that refugees have a higher prevalence of several subgroups of somatic and mental disorders compared with Swedish-born and that refugees with somatic/mental disorders have a considerably higher risk of a permanent exit from the labour market, by granting of disability pension, compared with Swedish-born with these disorders. Despite this, refugees with somatic/mental disorders had a significantly lower risk of mortality during the followup period compared with Swedish-born individuals with similar disorders. The higher risk of disability pension among refugees may hence have other causes than worse health status among refugees, and efforts to include refugees into the labour market should be prioritised.
Contributors MH and EMR made substantial contributions to the conception and design of the study. MH did the analyses in collaboration with SR. MH and SR drafted the manuscript, and EMR and FS critically revised drafts of the manuscript on several occasions. All authors 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.

Patient consent for publication Not required.
Ethics approval This project was approved by the regional ethical review board in Stockholm which also waived the requirement that informed consent from research subjects should be obtained.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data may be obtained from a third party and are not publicly available. Data supporting this study cannot be shared.