During the past few years, there has been a substantial increase in migration globally.1 In 2017, there were 258 million migrants worldwide, including 25.9 million refugees and asylum seekers.1 Sweden is one of the largest European recipients of refugees, with more than 160 000 individuals seeking asylum in 2015 alone.2 Studies have shown that refugees have an increased risk of poor mental health outcomes, including depression, anxiety and post-traumatic stress disorder (PTSD).Reference Bogic, Njoku and Priebe3,Reference Tinghög, Malm, Arwidson, Sigvardsdotter, Lundin and Saboonchi4 Given the high levels of exposure to traumatic life events both before, during and after the migration process,Reference Ferrada-Noli, Åsberg, Ormstad, Lundin and Sundbom5 it has further been hypothesised that refugees also have an elevated risk of suicide attempt and suicide. However, we could only identify one study that examined this association.Reference Norredam, Olsbjerg, Petersen, Laursen and Krasnik6 This study reported lower rates of suicideReference Norredam, Olsbjerg, Petersen, Laursen and Krasnik6 in refugees when compared with Danish-born individuals. Studies on migrants not discriminating between refugees and non-refugee populations, on the other hand, have shown lower rates of suicide and higher rates of suicide attempt.Reference Bursztein Lipsicas, Mäkinen, Apter, De Leo, Kerkhof and Lönnqvist7,Reference Di Thiene, Alexanderson, Tinghög, La Torre and Mittendorfer-Rutz8 Mental disorders represent one of the most important risk factors for suicide and suicide attempt.Reference Hawton and van Heeringen9 Studies have reported geographical and cultural differences with respect to the prevalence of mental disorders in suicide victims. In China, for example, a much lower proportion of people who die by suicide seem to have psychiatric disorders.Reference Law and Liu10 Still, to what extent mental disorders affect suicide and suicide attempt in refugees remains unknown.
The current register-based study aimed to examine the association between specific mental disorders and suicide attempt and suicide in refugees, and to investigate if the risk differs when compared with Swedish-born individuals.
Method
Study population
The study population was defined as all individuals aged 16–64 years, residing in Sweden on 31 December 2004 (n = 5 750 669). Only those with complete information on their reason for settlement in Sweden were included (n = 5 525 815). Further, as we compared refugees with Swedish-born individuals, non-refugee immigrants (n = 442 368) were excluded. Thus, the final study population included 5 083 447 individuals, where 196 757 were refugees.
We used the unique (de-identified) Swedish personal identity numberReference Ludvigsson, Otterblad-Olausson, Pettersson and Ekbom11 to link information from several population-based registers. The Longitudinal Integration Database for Health Insurance and Labor Market Studies (https://www.scb.se/LISA/) contains data from the labour market and educational and social sectors. The Longitudinal Database for Integration Studies (STATIV) register holds migration-related information, including reasons for settlement (e.g. refugee status). The National Patient Register (NPR) (https://www.socialstyrelsen.se/patientregistret) includes information on in-patient care since 1987, and for specialised out-patient care since 2001. Diagnoses in NPR are coded according to the ICD-10.12 The Cause of Death Register (https://www.socialstyrelsen.se/statistik-och-data/register/alla-register/dodsorsaksregistret/) comprises information on all deaths of Swedish residents since 1952.
Measures
Mental disorders
Mental disorders were defined as having a main diagnosis during psychiatric in-patient or specialised out-patient care, as recorded in the NPR during 2000–2004. More specifically, the following disorders were studied (ICD-10 codes): substance misuse disorders (F10–19), schizophrenia/non-affective psychotic disorders (F20–29), bipolar disorder (F30–31), depressive disorders (F32–34), anxiety disorders (F40–42), stress-related mental disorders (F43), PTSD (F43.1) studied separately, behavioural disorders (F50–59, F90–99) and personality disorders (F60–69).
Suicide attempt and suicide
The study population was prospectively followed up from 1 January 2005 until 31 December 2013, with respect to suicide attempt and suicide. Suicide attempt was defined as having received at least one diagnosis (ICD-10 codes X60–84 or Y10–34) in in-patient care, obtained from the NPR, during follow-up. Suicide was defined by the presence of ICD-10 codes X60–84 (suicide) or Y10–34 (undetermined intent) as the underlying cause of death in the Cause of Death Register. Inclusion of undetermined intent in the measure of suicide and suicide attempt reduces underreporting and spatial and secular trends in detecting and classifying cases of suicide attempt and suicide when intent was indeterminable.Reference Linsley, Schapira and Kelly13
Refugee status
In this study, a refugee was defined as an individual receiving a residence permit in Sweden as a refugee (according to the Geneva Convention of Refugees14), or an individual who has been granted a residence permit because they are ‘in need of protection’ or on ‘humanitarian grounds’.
Potential confounders
A range of potential confounders, measured in 2004, were considered. Age and gender were included in the analyses. We controlled for education, family situation and type of residential area (please see Table 1 for information on categorisation of the confounders). History of suicide attempt was defined as at least one admission to hospital for suicide attempt between 1987 and 2005. Adjustments were made for somatic multimorbidities, defined as in-patient or specialised out-patient care in 2000–2004 with a main diagnosis for diseases of the musculoskeletal system and connective tissue (ICD-10 codes M00–99), neoplasms (ICD-10 codes C00–97, D10–48), diseases of the circulatory system (ICD-10 codes I00–99) and other somatic disorders (rest of the ICD-10 codes with exception of F00–99). Individuals could be included in more than one group. Long-term sickness absence (>90 net days per year), disability pension and long-term unemployment (>180 days per year) were used as indicators of labour market marginalisation (LMM).
Cohort characteristics stratified by mental disorder, in Swedish-born individuals and refugees aged 16–64 years, residing in Sweden in 2004
Data displayed as absolute numbers and column percentages.
a. In 2004.
b. In 2000–2004.
Statistical analyses
Statistical analyses were conducted with SAS version 9.4 for Windows (SAS Institute Inc., Cary, NC) Crude and multivariate analyses were performed with Cox regression models of time to suicide attempt and suicide, respectively. We assessed person-years at risk by totalling the years that the individuals were living in Sweden during the follow-up period. The entry date was defined as 1 January 2005, and the exit date as the date of in-patient care for suicide attempt (outcome suicide attempt), date of suicide or any other cause, date of emigration or the end of follow-up (31 December 2013), which ever came first. In the main analysis, three regression models were analysed: model 1, adjusted for age and gender; model 2, with further adjustments for education, family situation, type of residential area and LMM; and model 3, with the addition of history of suicide attempt and somatic multimorbidity. The reference group comprised Swedish-born individuals with no mental disorder. Additionally, as shown in the Supplementary Material available at https://doi.org/10.1192/bjp.2019.215, we reran the Cox regression analyses stratified by refugees and Swedish-born individuals (i.e. with two separate reference groups).
Sensitivity analyses
We conducted a sensitivity analysis, in which individuals who were granted residence permits because they were ‘in need of protection’ and on ‘humanitarian grounds’ were excluded from the refugee category. Additionally, we conducted separate analyses in which we excluded cases coded as undetermined intent.
Ethical approval
Ethical approval for this study was obtained from the Regional Ethical Review Board, Karolinska Institutet, Stockholm (review number 2016/1533-32). The ethical review board approved the study and waived the requirement that informed consent of research subjects should be obtained. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Results
Cohort characteristics of the study population, stratified by mental disorder in refugees and Swedish-born individuals, are presented in Table 1. A total of 4% of Swedish-born individuals were treated at least once in 2000–2004 with a mental disorder as a main diagnosis, of which the most common diagnoses were substance misuse (30%) and depressive disorders (26%) (Supplementary Table 1). In refugees, 6% were treated for mental disorders, where the most common types were stress-related disorders (31%), followed by depressive disorders (30%).
In general, among those with no mental disorder, the age- and gender-adjusted hazard ratio for suicide attempt was slightly higher for refugees compared with Swedish-born individuals (Table 2). However, this was not observed in the multivariable-adjusted models (models 2 and 3). For all diagnostic groups, the risk for suicide attempt was lower for refugees in individuals with mental disorders, with the exception of bipolar disorder (multivariable-adjusted hazard ratio for refugees 6.28, 95% CI 4.05–9.75; multivariable-adjusted hazard ratio for Swedish-born individuals 5.82, 95% CI 5.36–6.33), schizophrenia/non-affective psychotic disorders (multivariable-adjusted hazard ratio for refugees 4.19, 95% CI 3.39–5.19; multivariable-adjusted hazard ratio for Swedish-born individuals 4.81, 95% CI 4.51–5.13) and personality disorders (multivariable-adjusted hazard ratio for refugees 6.34, 95% CI 4.74–8.48; multivariable-adjusted hazard ratio for Swedish-born individuals 6.60; 95% CI 6.18–7.05). When sociodemographic and LMM factors were added to the initial model, all hazard ratios decreased markedly (Table 2, model 2), partly because of education and family situation (data not shown). When additional adjustments were made for history of suicide attempt and somatic multimorbidity (Table 2, model 3), the risk estimates decreased further. The highest estimates for suicide attempt in individuals with a mental disorder were observed in substance misuse (multivariable-adjusted hazard ratio for refugees 7.40, 95% CI 6.35–8.63; multivariable-adjusted hazard ratio for Swedish-born individuals 9.26; 95% CI 8.95–9.57).
Associations between specific mental disordersa and subsequent admission to hospital for suicide attempt in Swedish-born individuals and refugees aged 16–64 years, residing in Sweden in 2004
Data are displayed as hazard ratios with 95% confidence intervals. PTSD, post-traumatic stress disorder.
a. Mental disorders diagnosed in specialised healthcare in the period 2000–2004.
b. Model 1: adjusted for age and gender.
c. Model 2: adjusted for age, gender, education, family situation, type of residential area, unemployment, sickness absence and disability pension.
d. Model 3: model 2 with additional adjustments for history of suicide attempt and somatic multimorbidity.
Table 3 presents hazard ratios with 95% confidence intervals for suicide by refugee status and type of mental disorder. Swedish-born individuals without a mental disorder constituted the reference group. In the absence of a mental disorder, refugees had a significantly lower risk of suicide compared with the reference group when adjustments were made for age and gender (hazard ratio 0.60, 95% CI 0.51–0.71). In general, the suicide rates and hazard ratios in individuals with a mental disorder was lower in refugees (rate of 79.5 per 100 000 person-years) compared with the Swedish-born population (rate of 165.3 per 100 000 person-years) (Table 3). This pattern was somewhat different for the different diagnostic groups. For example, the suicide risks for refugees with a mental disorder were comparable with the suicide risk for Swedish-born individuals with a mental disorder for schizophrenia (multivariable-adjusted hazard ratio for refugees 4.88, 95% CI 3.25–7.32; multivariable-adjusted hazard ratio for Swedish-born individuals 4.99, 95% CI 4.42–5.63) and for personality disorder (multivariable-adjusted hazard ratio for refugees 6.57, 95% CI 3.51–12.28; multivariable-adjusted hazard ratio for Swedish-born individuals 5.56, 95% CI 4.74–6.52). The suicide risk for refugees with a mental disorder was lower than that for Swedish-born individuals with a mental disorder for substance misuse, depressive disorders, anxiety disorders including PTSD, and behavioural disorders (Table 3). The hazard ratios decreased considerably after adjustments were made for sociodemographics and LMM factors (Table 3, model 2). In the fully adjusted model (model 3), where we also included factors related to comorbidity, all hazard ratios decreased further markedly, as a large part of the association was explained by history of suicide attempt.
Associations between specific mental disordersa and subsequent suicide in Swedish-born individuals and refugees aged 16–64 years, residing in Sweden in 2004
Data are displayed as hazard ratios with 95% confidence intervals. PTSD, post-traumatic stress disorder.
a. Mental disorders diagnosed in specialised healthcare in the period 2000–2004.
b. Model 1: adjusted for age and gender.
c. Model 2: adjusted for age, gender, education, family situation, type of residential area, unemployment, sickness absence and disability pension.
d. Model 3: model 2 with additional adjustments for history of suicide attempt and somatic multimorbidity.
e. For ethical reasons (the risk of identification of individuals), if the number of suicides is fewer than ten, it is not reported.
Within-group analyses of refugees (Supplementary Tables 2 and 3) revealed that, when compared with refugees with no mental disorders, refugees with any as well as specific mental disorders had significantly elevated risk for suicide attempt and suicide.
Sensitivity analyses
In the sensitivity analysis, where we excluded individuals who were granted a residence permit on ‘humanitarian grounds’ or because they were ‘in need of protection’, results were similar to the main analyses. Finally, separate analyses for suicides and deaths coded as undetermined intent were conducted. These analyses generated similar estimates (data not shown).
Discussion
Key results
The present study examined the associations between mental disorder and suicide attempt and suicide, using a large cohort of 4 886 690 Swedish-born individuals and 196 757 refugees in Sweden. Our findings demonstrate that the prevalence of mental disorder was slightly higher in refugees than in Swedish-born individuals. Overall, the rates and risk for suicide attempt in individuals with a mental disorder were lower in refugees compared with the Swedish-born population; a pattern that was also true for most specific disorders. Similar patterns were seen for suicide.
Mental disorders in refugees and Swedish-born individuals
Our results revealed slightly higher levels of mental disorders in refugees than in Swedish-born individuals, which has been reported by others.Reference Bogic, Njoku and Priebe3,Reference Tinghög, Malm, Arwidson, Sigvardsdotter, Lundin and Saboonchi4 Previous studies have shown that particularly PTSD and depressive disorders are common among refugees, which was confirmed in our study. The high level of exposure to traumatic life events in refugees has been pointed to as an explanation for the elevated psychiatric morbidity in this group.Reference Tinghög, Malm, Arwidson, Sigvardsdotter, Lundin and Saboonchi4,Reference Ferrada-Noli, Åsberg, Ormstad, Lundin and Sundbom5 Furthermore, post-migration stress, including acculturation problems and ethnical discrimination, may also contribute to adverse mental health.
Suicide attempt and suicide in refugees and Swedish-born individuals
Our findings showed that refugees in general had a lower risk of suicide attempt and suicide compared with Swedish-born individuals. To the best of our knowledge, our study is the first to examine these associations (i.e. both suicide attempt and suicide) in a nationwide setting. One study investigating differences in injury mortality in refugees, immigrants and Danish-born individuals reported lower rates of suicide in refugees, when compared with natives.Reference Norredam, Olsbjerg, Petersen, Laursen and Krasnik6 However this study focused solely on suicide death, and the lower risk was seen for males only.Reference Norredam, Olsbjerg, Petersen, Laursen and Krasnik6 Thus, there is a gap in the literature on studies examining these associations, particularly in refugees. There have been studies on immigrants (not discriminating between refugees and non-refugee migrants) and asylum seekers, where both higher and lower rates of suicide attempt and suicide have been reported.Reference Bursztein Lipsicas, Mäkinen, Apter, De Leo, Kerkhof and Lönnqvist7,Reference Di Thiene, Alexanderson, Tinghög, La Torre and Mittendorfer-Rutz8
There are several possible explanations for the lower suicide rates in refugees. Among others, cultural influences might explain differences in rates of suicide attempt and suicide.Reference Bursztein Lipsicas, Mäkinen, Apter, De Leo, Kerkhof and Lönnqvist7,Reference Bhui, McKenzie and Rasul15 Here, culture may contribute to determining the way refugees experience and express mental health problems, and reflect upon their needs for mental healthcare.Reference Lawrence, Oquendo and Stanley16 Moreover, culture may also influence a person's attitudes toward suicidal behaviour, such as how one reacts to suicidal thoughts.Reference Lawrence, Oquendo and Stanley16 Most of the refugees included in our study come from Muslim majority countries (data not shown), and studies have shown that cultural and religious factors and stigma attached to suicide attempt and suicide are associated with much lower suicide rates in Islamic nations, and especially in Muslim countries.Reference Lester17 Further studies on the potential influence of cultural features and religious aspects in the risk of suicide attempt and suicide in refugees are warranted.
Mental disorders and suicide attempt and suicide in refugees and Swedish-born individuals
Moreover, our study is the first to examine associations between mental disorders and suicide attempt and suicide in refugees. Studies investigating these associations in general populations regardless of migration status have demonstrated that individuals suffering from mental disorders have a considerable higher risk for suicide attempt and suicideReference Hawton and van Heeringen9,Reference Björkenstam, Björkenstam, Holm, Gerdin and Ekselius18,Reference Hawton, Sutton, Haw, Sinclair and Harriss19 compared with individuals without mental disorders.Reference Hawton, Sutton, Haw, Sinclair and Harriss19,Reference Nepon, Belik, Bolton and Sareen20 In our study, Swedish-born individuals with a mental disorder had a sixfold elevated risk for both suicide and suicide attempt compared with Swedish-born with no mental disorder, after adjustments for several important confounders.
We can now further show that in the presence of a mental disorder, the risk of suicide attempt and suicide is lower in refugees than in Swedish-born individuals. These findings are interesting as they challenge the idea of lower rates of suicide attempt and suicide in refugees being mainly attributable to a positive health selection of refugees – the healthy migrant effect.Reference Kennedy, Kidd, McDonald and Biddle21 Moreover, our findings are counterintuitive as refugees with a mental disorder in specialised healthcare might even be hypothesised to suffer from more severe forms of mental disorders than Swedish-born individuals. This reasoning is based on clinical and scientific knowledge on differences in access to and acceptance of specialised healthcare of refugees compared with the population in the host country in countries of the Organisation for Economic Co-operation and Development (OECD).Reference Ekblad and Kastrup22,Reference Graetz, Rechel, Groot, Norredam and Pavlova23 Moreover, healthcare and treatment needs might not be met to the same extent for refugees as for natives.Reference Kohlenberger, Buber-Ennser, Rengs, Leitner and Landesmann24 Adequate care in immigrants and refugees might be hampered by language barriers, differences in the clinical manifestation and symptom course of the underlying disease, and consequently in its diagnostics, as well as the lack of competence in transcultural psychiatry and psychology in the healthcare settings of the host country.Reference Bhugra, Gupta, Schouler-Ocak, Graeff-Calliess, Deakin and Qureshi25
Potential protective factors in refugees
Despite these risk factors and common aetiological features of suicidal behaviour and suicide across groups with different ethnicity, our results suggest that refugees with mental disorders have protective factors regarding the risk of suicide attempt and suicide, which Swedish-born individuals with the same disorders might lack. One such protective factor might be the lower likelihood to act upon suicidal ideas among refugees, which in turn might be affected by the strong stigma associated with suicidal behaviour and suicide prevailing in many countries, from which refugees migrate to Sweden.Reference Lawrence, Oquendo and Stanley16 Such a culturally influenced stigma might prevent suicidal behaviour and suicide by strongly shaping the individual's attitudes toward ending one's own life. Other potential protective factors among refugees might be related to differences in personality and resilience factors as well as differences regarding social connectedness.Reference Shaw, Atkin, Bécares, Albor, Stafford and Kiernan26 In our cohort, refugees with mental disorders were considerably more often cohabiting with partners or children than Swedish-born individuals. We have controlled for family situation in the analyses, but this observation might still be a marker of a stronger social connectedness in refugees with mental disorders. Further studies on protective factors as well as risk factors for suicidal behaviour and suicide among refugees are warranted.
Specific mental disorders
The aforementioned pattern was somewhat different for the different diagnostic groups. Estimates for suicide attempt and suicide were equally high for refugees and Swedish-born individuals with schizophrenia, bipolar disorder and personality disorder. Previous studies have shown a higher risk for schizophrenia in refugees compared with the population in the host country.Reference Hollander, Dal, Lewis, Magnusson, Kirkbride and Dalman27 In our study, 13% of refugees had been treated for schizophrenia and non-affective psychotic disorders compared with 10% of Swedish-born individuals. These differences in patterns related to these three disorders might be related to the morbidity level (i.e. all three are severe mental disorders) and to the stigma associated with living with such a disorder as a refugee. In both refugees and Swedish-born individuals, substance misuse disorder entailed a particularly high risk for suicide. Especially alcohol misuse has been pointed out as a key risk factor for suicidal behaviour and suicide in the general population,Reference Hawton and van Heeringen9,Reference Reutfors, Brandt, Ekbom, Isacsson, Sparen and Osby28 and this also seems to hold true for refugees.
Strengths and limitations
This study has several strengths, including the longitudinal population-based design and use of national registers with high completeness and validity. The large cohort size allowed for detailed analyses of different types of mental disorders and the ability to adjust for important confounders. Nevertheless, there are limitations. Information on mental disorder and suicide attempt relied solely on register data, and thus we have likely captured the most severe cases. This may lead to misclassification of undiagnosed and/or untreated individuals as unexposed, especially in refugees where the prevalence of psychological distress is often very high.Reference Tinghög, Malm, Arwidson, Sigvardsdotter, Lundin and Saboonchi4 Potential differences in healthcare-seeking behaviour in Swedish-born individuals and refugees may also affect our findings (e.g. poor language skills may prevent refugees from seeking healthcare). Furthermore, because of cultural and religious stigma, underreporting of suicide attempt may be especially prevalent in refugees from Islamic or Muslim majority countries. On the other hand, cultural and religious aspects might represent protective factors for suicide attempt and suicide in refugees. Such information was not available for the present study but merits future investigations. Another limitation is that the refugee population also included those granted a residence permit because they were ‘in need of protection’ and on ‘humanitarian grounds’. It has been shown that the comorbidity in the latter group is considered to be higher than in other groups who obtain a residence permit in Sweden. This might have introduced a potential negative health selection in the refugee group in our study. However, in the sensitivity analyses where we included and excluded these individuals, there were no significant differences in the results.
In conclusion, this study is the first to provide evidence that, in individuals with a mental disorder, refugees have a lower risk for both suicide attempt and suicide compared with their Swedish-born counterparts; however, risk estimates differed by type of disorder. Nevertheless, compared with those with no mental disorder, both refugees and Swedish-born persons had an elevated risk for both suicide attempt and suicide. This knowledge should guide healthcare planners and providers who need to be aware that mental disorders are strong risk factors for suicide attempt and suicide in both refugees and Swedish-born individuals.
Supplementary material
Supplementary material is available online at https://doi.org/10.1192/bjp.2019.215.
eLetters
No eLetters have been published for this article.