Mental Health of Refugees in Austria and Moderating Effects of Stressors and Resilience Factors

Abstract

Given the exposure to stressors in their home countries, during migration and after arrival, refugees are vulnerable to mental health problems. Their access to adequate health care and other social infrastructures, however, is hampered. This reduces, in addition to other factors, their ability to take part in social and economic activities. We examine the prevalence of mental disorders among the refugee population that arrived in Austria mainly between 2013 and 2018, drawing on data from a refugee survey. We found a high share of refugees (32%) to have moderate or severe mental health problems. When investigating the effects of stressors on the mental health situation, we found a positive association with experienced discrimination in Austria and the fear for partners and children left behind, and a negative correlation with proficiency in the German language, being employed (including volunteer work), having more supportive relationships, and satisfaction with the housing situation.

1. Introduction

Mental health problems are a serious barrier for the integration of immigrants into host societies (Aroian et al. 1998). Refugees in particular are exposed to various risk factors for mental health problems before, during, and after migration. Pre-migration risk factors for mental distress are persecution, exposure to potentially traumatic events in person or as a witness and exposure to, or involvement in, armed conflicts. Many refugees have to face economic hardships including having their basic needs not met. Peri-migration risk factors are exposure to physical harm and life-threatening conditions during migration. Separation from family members and from support networks are additional stressors. Post-migration risk factors are manifold and include uncertainty about the asylum application, unmet health needs, fear for family members left behind, lack of close relationships, lack of social integration including difficulties in entering the labour market, recognition of degrees, loss of social status, and difficult socio-economic conditions including unsatisfactory housing conditions (Priebe et al. 2016; Giacco and Priebe 2018; Löbel 2020). These particular risk factors for the mental distress of refugees overlap with the general acculturative stress experienced by immigrants relating to loss, unfamiliarity with the tasks of daily living, unfamiliar (or very limited) occupational options, language barriers, discrimination, and feeling marginalized in the new surroundings or social structure (e.g., Aroian et al. 1998).

It can be assumed that the population group of refugees resettled in Europe in the years following 2013 shows high prevalence rates of mental disorders since they originated from war-affected countries like Afghanistan, Syria, and Iraq. This is supported by Lindert et al. (2018), who highlight that refugee groups from countries with intense human rights violations show higher rates of mental health problems. In the case of Germany, the prevalence of mental health disorders is reported to be significantly higher for refugees, in general, as compared to the host population (Nesterko et al. 2020; Frank et al. 2017). This is also the case for refugees from Ukraine who arrived most recently in Germany (Buchcik et al. 2023).

Austria was an important receiver of asylum seekers in Europe during the last decade. In the period 2014–2018, around 197,000 asylum applications were filed in Austria (Eurostat 2019), and roughly 109,000 individuals were officially granted asylum (including subsidiary protection and protection on humanitarian grounds). Between 2014 and 2018, six out of ten asylum applicants in Austria had Syrian, Afghan, or Iraq nationality. The number of asylum seekers arriving in Austria in 2019 and 2020 was comparably low (less than 28,000), but substantially increased again in 2022, with a peak amounting to 112,000, and declined to less than 26,000 in 2024 (not including displaced persons arriving from Ukraine). Overall, 460,000 individuals were applying for asylum in Austria over the period 2014–2024, a country with currently 9.2 million inhabitants (BMI 2019, 2025). In this paper, we examine the level of mental distress and thus the likelihood of mental disorders of adult refugees from Afghanistan, Iraq, and Syria who arrived mainly between 2013 and 2018 in Austria. We hypothesize that (a) mental distress levels among refugees are high and exceed those of the average Austrian population and (b) that supportive social connections and a stable environment (as experienced through employment, housing, and social bonds) have a moderating effect, while stressors like family members left behind in the country of origin have a negative effect on distress levels. To test these hypotheses, we draw on primary data from a refugee survey conducted in Austria between December 2017 and April 2018 (FIMAS+INTEGRATION). We describe the level of mental distress perceived and examine the mediating effects of stressors and supportive factors for resilience by applying multivariate regression analysis.

2. Literature Review

In the literature on moderators of mental distress of refugees, the actual resident status of forced migrants was found to have a significant impact on the prevalence of trauma-related mental health disorders: For Switzerland, Heeren et al. (2014) found that asylum seekers suffered more often from post-traumatic stress disorder (PTSD) than recognized refugees. The rate of depression among asylum seekers was nearly twice as high compared to that of migrants with permanent residency claims (and hence access to the labour market, language courses, and other activities). Lack of security and fear of deportation, connected to economic instability, substandard housing, and feelings of guilt and shame for having survived, tend to be the most relevant risk factors for developing mental health disorders, which can be exacerbated by a lengthy asylum process (Laban et al. 2005; Hajak et al. 2021; Leiler et al. 2019). The latter was relevant in the case of Austria for many refugees who had arrived in the years following 2014. Due to capacity constraints in the asylum administration, refugees had to wait for considerable periods until cases were decided by public authorities. While decisions in the regular asylum procedure should be issued within 6 months, in 2017, for example, the average duration was 16.5 months. During the asylum application procedure, refugees could work only to a limited extent, e.g., in seasonal jobs in agriculture and tourism. As a result, many of the refugees experienced a longer period of insecurity about their right of residence and inactivity in Austria (The Expert Council for Integration 2018).

In line with and partly exceeding the minimum reception standards outlined by the Council of the European Union (2003), health insurance is granted for basic services during the asylum process, legally mandated in the Austrian General Social Insurance Law 2004. This includes access to public hospitals, psychological and psychiatric treatment, and medications (Kohlenberger et al. 2019). Access is organized via a substitute voucher, which overall makes it more bureaucratic for asylum seekers to seek and receive treatment than for recognized refugees and residents, who can do so via an electronic health insurance card (“e-card”).

Given the risk factors refugees are exposed to before, during, and post-migration, high prevalence rates of mental disorders are expected in comparison to the host population. Empirical evidence regarding PTSD exists (e.g., Priebe et al. 2016). According to studies in Germany, the prevalence is reported to be significantly higher for refugees compared to the host population (Nesterko et al. 2020; Frank et al. 2017; Georgiadou et al. 2018).

Systematic reviews on the prevalence of mental health disorders among refugees reveal a substantial variability in outcomes (Priebe et al. 2016; Bogic et al. 2015; Fazel et al. 2005), particularly when the specific, most common disorders (depression, general anxiety disorder, and PTSD) are investigated. In studies applying population-based sampling methodology, estimates on the prevalence of PTSD vary between 16.4 per cent and 54.9 per cent (Bozorgmehr et al. 2016). Variability in outcomes is due to the fact that refugee populations and their mental distress levels diverge between diverse countries of origin, on the one hand, and that reception conditions are different in host countries, on the other hand.

More significantly, the differences in sample selection and methodological approaches result in strong variations in prevalence rates. Studies using a more representative sampling methodology were shown to result in lower prevalence rates than research applying convenience sampling (Bozorgmehr et al. 2016). In the case of PTSD, the estimated mean of prevalence for the total refugee population is 9 per cent in Europe, which is much higher than the 1–3 per cent for the host population (Priebe et al. 2016). For depression and general anxiety disorder, however, results are more diverse. While Priebe et al. (2016) found no systematic differences between total refugee populations and host country populations regarding short-term resettled people, Turrini et al. (2017) claimed in their review that prevalence rates of depression and anxiety disorders are as high as rates of PTSD.

Reasons for the lack of differences in the cases of depression and general anxiety disorder found in research applying population-based samples can be manifold. Refugees, and especially those with mental distress, are a much more difficult group to survey in a representative way due to issues such as accessibility, mobility, trust, and concerns about the stigma of health problems (Enticott et al. 2017). Cultural differences in revealing information on personal mental distress (Kohrt et al. 2014) and in recognizing mental health concerns (Weigl and Gaiswinkler 2019) do matter for refugee–host population differences in prevalence rates and may moreover constitute a barrier for access to care. Thus, while selection bias overstates the prevalence rate of mental health problems in the case of convenience-based sampling (survey samples of refugees living in camps, visitors of specific health, or supporting institutions), studies using population-based samples may understate disorder prevalence in the refugee populations resettled in Europe. This is in line with Bogic et al. (2015), whose systematic review reported higher prevalence rates for the long-term resettled refugee population in comparison to the host population, not only for PTSD but also for depression and anxiety disorders.

Somatization symptoms (medically unexplained physical symptoms) are also more common among refugees. Comorbidity of somatic symptoms (i.e., physical symptoms or illness accompanying mental health problems) is high with PTSD (Liedl and Knaevelsrud 2008) but also occurs with other mental disorders (Priebe et al. 2016; Lolk et al. 2016). Prevalent somatization symptoms include pain, for example, in the back, heart, and muscles and feelings of physical weakness (Morina et al. 2018).

Resilience, i.e., the capability of people to adapt to psychological distress and (potentially) traumatic events (Bonanno 2004; Yehuda 2004; McLaughlin et al. 2009) helps refugees not only to overcome mental health problems triggered by harsh conditions before resettlement but also to cope with post-migration stressors. Psychological resilience of refugees can be improved by external conditions (available social support) but is also affected by individual factors such as education, religion, and belief systems (Siriwardhana et al. 2014) and relevant coping strategies (Renner et al. 2020).

Hence, while refugees’ mental health has garnered academic attention in various contexts and from a diverse set of disciplinary perspectives, including critical border and migration studies, a research gap and, even more noticeably, a policy gap remain (Khoury and Kohlenberger 2025). Refugees’ mental health needs continue to be poorly understood and seldom adequately addressed by health care providers, which perpetuates inequalities in health and leads to adverse permanent resettlement outcomes. This can be explained, partly, by language and cultural barriers, as discussions on mental well-being in refugees’ countries of origin tend to be more heavily stigmatized and may lead to under- or misdiagnosis (Mohammadzadeh et al. 2020; Gearing et al. 2015). This especially applies to the group of forced migrants studied in the present paper, recently arrived Syrian and Afghan refugees in Austria, for whom high levels of both depression and anxiety disorder have been found (Kohlenberger et al. 2019). In this article, based on one of the largest cross-sectional refugee surveys in the Austrian context, we aim to unearth some of the stressors and resilient factors, knowledge of which may inform more adequate reception and integration policies.

3. Data and Methodology

The FIMAS+INTEGRATION survey was carried out by the International Centre for Migration Policy Development (ICMPD) in cooperation with the University of Graz and the Vienna Institute for International Economic Studies (wiiw). FIMAS+INTEGRATION is the second wave of a longitudinal refugee survey in Austria. The first wave (FIMAS) was carried out between August 2016 and May 2017. Interviews of FIMAS+INTEGRATION were conducted between December 2017 and April 2018 via different survey modes. In a first step, 413 computer-assisted self-administered and personal interviews (CASIs) were carried out at various relevant establishments (public employment services, counselling and training centres for migrants, etc.) in Vienna, Salzburg, Graz, Linz, and Innsbruck (i.e., in Austria’s capital and other Austrian provincial capitals). Second, refugees were invited via mail and e-mail in order to expand the coverage of the survey. Therefore, an invitation to the online version of the survey was sent to addresses provided by the public employment service of Austria (the AMS). This second part of the survey was based on a stratified random sample of refugees registered in the AMS database. In total, 1635 refugees participated in the survey.

The participating refugees were informed at the beginning of the interviews about the aims of this study, the voluntary nature of participation, and the guarantee of anonymity in the data analysis. Informed consent was obtained from all interviewees. Ethical approval for the data collection procedure and the questionnaire was granted by the Ethics Committee of the University of Graz, Austria. Many of the measures applied in the FIMAS+INTEGRATION survey were already applied before in the IAB-BAMF-SOEP refugee survey conducted in Germany (Brücker et al. 2016) or other refugee surveys. In order to circumvent, as far as possible, language barriers and comprehension challenges, the interviews were carried out in the preferred language of the interviewees, i.e., Arabic, Farsi-Dari, or German. Extensive pre-tests (including a pilot with 25 test respondents) took place before the actual interviews. This allowed us to identify and adapt the items that proved difficult to comprehend and to make the questionnaire manageable for the interviewees.

As intended, the majority of the participants originated from Syria, Iraq, and Afghanistan. A small number of other refugees—mostly from Iran—and persons, who were still in the asylum application process, also entered the sample at the interview stage. Apart from being a refugee originating from one of the four countries mentioned above, the only additional inclusion criterion was to belong to the age group 15–60.

Apart from demographic and household characteristics, the questionnaire focused specifically on labour market issues and additionally covered various spheres of life of the refugees from which we selected those that are associated with mental health, like social and cultural integration, qualifications, physical health and well-being, migration experiences, and housing. Further, a number of possible stressors and resilience factors that could influence the level of mental strain of refugees were also captured in the survey. Demographic characteristics comprised age (in the 15–60 years bracket), gender, and detailed information on the household structure. Thus, we know not only whether a person lived in partnership and the number of children but also if the latter or further family members lived in the same household at the time of the interview.

Information on physical health included self-rated health and health problems; in particular, the level of physical pain experienced in the past four weeks was reported. Since physical health problems can result in psychological strain, we included this information as a control variable. The subjective impression of physical pain was measured on a 6-item Likert scale from “no pain” (0) to “very strong pain” (5).

For reasons of research ethics and in order to prevent potential re-traumatisation of respondents, information on potentially traumatic events before and during migration that could have had an influence on their mental health at the time of the interview was not captured. In order to examine the effect of the migration experience on respondents, we can only use the information on the duration of their journey (captured in months), assuming that longer periods increase mental distress levels. Such a link between the length of the journey and migrants’ higher distress levels was found, e.g., by Jankovic-Rankovic et al. (2022), Poole et al. (2018), and Strømme et al. (2020). It is assumed that distress levels remain elevated after arriving in the host country. Once refugees have found a safe place to reside, we expect stress levels to decline gradually. Therefore, we calculated the time elapsed since the arrival in Austria.

The majority of the refugees interviewed had been granted asylum or the status of subsidiary protection. However, some persons were still in the asylum application procedure, which is captured by a binary variable. In the period 2014–2018, about half of the asylum seekers received a positive decision on their asylum application (e.g., 51% in 2017), while in the years before that, the share was somewhat lower (The Expert Council for Integration 2018). Thus, we expect refugees to experience the asylum application procedure as a state of limbo and, therefore, a very stressful period. We included a dummy variable for respondents still in the asylum procedure.

Our data covers information on acculturative stress. On the one hand, experienced discrimination was asked for and measured on a 5-item Likert scale from “never” to “very often”. On the other hand, the inability to communicate in the host country’s language, which hinders refugees in finding their way in the society, was also reported. We used the mean value of the information from the questions: “How well do you understand German” and “… speak German”. In both cases, a 5-item Likert scale from “not at all” to “like my mother tongue” was applied.

Close relationships with relatives and friends are known to help people cope with stressful episodes in their lives. Refugees were asked if they knew someone in Austria who would help them in different circumstances. One relevant question was whether they had a person with whom they could talk about personal problems. This information was used as a binary variable.

Having contact with people outside of the inner family circle fosters integration of individuals into society. In order to measure the extent of the social network of a person, the refugees were asked the following: “How many people do you know in Austria whom you feel close to?”. Thus, the question covers only the inner circle of well-known persons and not random acquaintances. A cut-off point of 50 persons was imposed on this variable.

Information on educational attainment was classified according to ISCED 2011. However, the information was originally supplied according to the national schooling system in the country of origin. Conversion into the ISCED classification was, particularly in the case of Afghanistan, ambiguous.

Economic as well as social integration into the host society is fostered by employment. We included a binary variable for whether the respondent was working or not, irrespective of whether it was paid employment or voluntary work. Furthermore, detailed information on the housing situation (type and size) was collected. In order to have a measure for housing, we chose the simple subjective information on satisfaction with the housing situation, which was an 11-item Likert scale from “very unsatisfied” (0) to “very satisfied” (10).

3.1. Measuring Mental Distress

In order to measure the level of mental distress, we applied the Kessler 10 scale (K10). The K10 scale is a simple measure of non-specific psychological distress and therefore does not focus on a specific mental health diagnosis like depression or anxiety disorder (Kessler et al. 2002). It has been applied among other large-scale surveys in the case of the US National Health Interview Survey and the Australian National Survey for Mental Health and Wellbeing and has been ascertained as an efficient screening instrument outperforming other scales (Furukawa et al. 2003). Good psychometric properties have been obtained in several countries and language versions (Carrà et al. 2011; Sampasa-Kanyinga et al. 2018; Thelin et al. 2017), also when screening refugee populations (Fassaert et al. 2009; Sulaiman-Hill and Thompson 2010). The German translation of the K10 questionnaire has been tested in Austria with outpatients and students. It has been identified as a reliable screening and research instrument, based on internal consistency and convergent validity in relation to other instruments (Giesinger et al. 2008). The K10 scale comprises ten questions about mental distress symptoms experienced during the previous four weeks, for example, “In the past 4 weeks; about how often did you feel so nervous that nothing could calm you down?” or “… about how often did you feel that everything was an effort?”. Answers are based on a 5-item Likert scale from “none of the time” (1) to “all of the time” (5). Scores of the 10 items are then summed up, yielding a minimum total score of 10 and a maximum score of 50. A score of 20–24 is interpreted as being at risk of a mild form of mental distress, 25–29 as a risk of having a moderate mental health problem, and 30 and above as a risk of a severe mental health problem at the time of the interview. The distinction between moderate and severe mental health problems follows the Global Assessment of Functioning (GAF) Scale (American Psychiatric Association (APA) 2000): Moderate forms of mental health problems are evident when persons show symptoms like occasional panic attacks or flat affect (i.e., strong reduction in emotional expressiveness) and circumstantial speech. Severe mental health problems are evident when at least serious symptoms (e.g., suicidal ideation, severe obsessional rituals) or any serious impairments in social or occupational functioning (e.g., no friends, unable to keep a job) are present.

3.2. Multivariate Estimation Approach

We study determinants of mental distress by estimating the following specification:

$${y_i=\beta }_0+{\mathit{\beta }\mathit{x}}_{\mathit{i}}+{ϵ}_i.$$

(1)

In multivariate analyses, we provide models with two different dependent variables, a linear and a dichotomous one. In regression models [1] and [3], $y_{\mathrm{i}}$ is the value on the Kessler scale, ranging from 10 to 50. In order to rule out the effects of heteroscedasticity, we estimate by applying robust standard errors. In regression models [2] and [4], $y_i$ is a dummy variable that indicates whether individuals experience levels of distress that indicate moderate or severe mental health problems. In these cases, psychotherapeutic treatment is recommended or necessary. ${\mathit{x}}_{\mathit{i}}$ is the vector of explanatory variables and ${ϵ}_i$ is the error term.

The set of explanatory variables includes age groups, gender, and educational attainment level, level of physical pain, duration of migration, and time of stay in Austria in months, with dummies to indicate whether the respondent is still in the asylum application procedure, lives with a partner in a household, with children, or if he/she has a partner left behind in the home country or another foreign country, or children not living in the household. Further dummies indicate if the respondent has someone to talk to about personal problems or if he/she is working (paid or unpaid/voluntary work). The scores of the Likert scales were applied for information on proficiency in the German language, experienced discrimination, and housing satisfaction. In addition, the country of origin, the province of residence in Austria, and the mode of the interview were used as control variables.

4. Empirical Results

In the FIMAS+INTEGRATION survey, 1518 out of the 1635 interviewed refugees provided enough information to calculate a composite score at the K10 scale. In this remaining sample, about 79 per cent of the respondents are men (see

Table 1. Descriptive results of the FIMAS+INTEGRATION refugee sample.

		Subgroups, %	K-10 Score, Mean Value	Moderate or Severe Mental Health Problems, %
Total population			20.9	31.6
Gender	Men	79.0	20.5	30.7
	Women	21.0	22.3	35.1
Age group	15–24	33.1	21.5	34.9
	25–34	39.8	20.7	32.5
	35–44	18.7	19.5	24.3
	45–60	8.4	20.8	27.7
Country of origin	Afghanistan	27.1	20.5	28.1
	Iraq	13.7	22.6	40.7
	Syria	57.8	20.5	31.0
	Iran	1.5	23.5	40.9
Year of arrival	Before 2013	13.4	20.7	30.6
	2013	4.2	20.9	32.3
	2014	16.2	20.1	28.0
	2015	48.7	21.0	33.0
	2016–2018	17.5	21.1	31.4

Source: FIMAS+INTEGRATION, own calculations.

below). Younger age groups between 15 and 34 years dominate. Close to 60 per cent of the refugees are from Syria, 27 per cent from Afghanistan, a further 14 per cent from Iraq, and only 1.5 per cent are from Iran. Those who arrived in 2015 account for about half of the sample; the rest of the sample is divided equally between the arrival cohorts before 2014, in 2014, and 2016–2018. In order to test the internal consistency of the K10 scale with the refugee population covered in our survey, we calculated Cronbach’s alpha (Cronbach 1951), which indicated very strong reliability with α = 0.93.

Descriptive results show that 20 per cent of the refugees in our sample state a level of mental distress that suggests a severe mental health problem at the time of the interview (see Figure 1). A further 11 per cent are likely to have moderate mental health problems, while another 15 per cent state to be slightly stressed. The remaining refugees, i.e., 54 per cent of the respondents, report having no or only negligible symptoms of mental distress and can be counted as unstressed. Women are, on average, more at risk (35%) of having moderate or severe, i.e., clinically relevant problems, as compared to men (31%). In the general population in Austria and other European countries, women display a significantly higher risk than men for certain mental health diseases (Kerkenaar et al. 2013).

Younger age groups (15–24 years: 35%; 25–34 years: 33%) show significantly higher risk levels compared to middle- and older-aged refugees (35–44 years: 24%; 45–60 years: 28%). In our sample, refugees from Iran are the group most affected by moderate or severe problems (41%), followed by refugees from Iraq (41%), Syria (31%), and Afghanistan (28%). Refugees who immigrated in 2015 show slightly higher risk levels than those who arrived in the years before and thereafter.

The observed shares of refugees with mental health problems may seem high but are in line with the results of refugee surveys performed in other countries, where the Kessler 10 scale was applied to measure mental distress, like in the case of Australia (De Maio et al. 2017).

Regression Results

In the multivariate framework, various stressors and protective factors were significantly associated with mental distress. In the following section, we will discuss the results of regression model [1] in

Table 2. Regression results.

	[1]	[2]	[3]	[4]
Regression Models	OLS	Logit	OLS	Logit
Dependent Variable	Kessler 10 Score	Medium or Severe Mental Health State	Kessler 10 Score	Medium or Severe Mental Health State
	(10–50)		(10–50)
Years of Arrival	All	All	2014–2018	2014–2018
Age groups (reference group: 35–44)
15–24 years	3.426 ***	1.049 ***	3.808 ***	1.103 ***
	(1.115)	(0.334)	(1.211)	(0.362)
25–34 years	2.377 ***	0.804 ***	2.287 ***	0.759 **
	(0.825)	(0.278)	(0.884)	(0.298)
45–60 years	0.843	0.186	0.906	0.030
	(1.081)	(0.378)	(1.204)	(0.443)
Women	1.446 *	0.223	1.154	0.287
	(0.850)	(0.241)	(0.945)	(0.278)
Physical pain (reference group: no pain)
Very slight	2.768 ***	0.594 **	2.719 ***	0.596 **
	(0.827)	(0.252)	(0.895)	(0.280)
Slight	3.017 ***	0.813 ***	3.002 ***	0.850 ***
	(0.910)	(0.267)	(0.985)	(0.286)
Moderate	4.930 ***	1.167 ***	5.410 ***	1.312 ***
	(0.897)	(0.248)	(0.942)	(0.266)
Strong	5.899 ***	1.028 **	5.940 ***	0.982 **
	(1.544)	(0.399)	(1.719)	(0.421)
Very strong	−1.321	−0.100	−5.051	−1.225
	(3.485)	(0.891)	(3.582)	(1.367)
Duration of migration in months	0.007	0.000	0.009	−0.000
	(0.011)	(0.004)	(0.015)	(0.005)
Months since arrival in Austria	0.012	0.004 *	0.014	0.011
	(0.008)	(0.002)	(0.038)	(0.012)
Asylum application pending	6.226 ***	1.664 ***	6.522 ***	1.927 ***
	(1.705)	(0.480)	(1.753)	(0.520)
Living with a partner in same household	0.598	0.220	0.441	0.275
	(0.801)	(0.251)	(0.856)	(0.275)
Having a partner who is living in	3.234 **	1.014 **	1.051	0.355
home country or other foreign country	(1.577)	(0.421)	(1.600)	(0.515)
Living with children in same household	−0.791	−0.240	−0.860	−0.347
	(0.921)	(0.270)	(0.990)	(0.297)
Having children who are not living	3.284 **	1.085 **	3.758 **	1.420 ***
in the same household	(1.616)	(0.468)	(1.733)	(0.525)
Living with further family members	−1.324 *	−0.297	−1.834 **	−0.452 *
	(0.774)	(0.242)	(0.808)	(0.263)
Someone to talk to	−1.198 *	−0.140	−1.168 *	−0.162
	(0.648)	(0.198)	(0.695)	(0.217)
Social network	−0.362 ***	−0.094 ***	−0.349 ***	−0.081 **
	(0.085)	(0.029)	(0.088)	(0.032)
Social network2	0.007 ***	0.002 ***	0.007 ***	0.001 **
	(0.002)	(0.001)	(0.002)	(0.001)
Education (reference group: ISCED 0)
ISCED 1	3.363 ***	1.272 ***	2.990 **	1.188 **
	(1.298)	(0.473)	(1.422)	(0.603)
ISCED 2	3.916 ***	1.361 ***	3.152 **	1.195 *
	(1.404)	(0.494)	(1.537)	(0.622)
ISCED 3–4	3.126 **	1.275 ***	3.300 **	1.357 **
	(1.255)	(0.466)	(1.373)	(0.587)
ISCED 6–8	3.878 ***	1.312 ***	4.166 ***	1.434 **
	(1.341)	(0.483)	(1.461)	(0.605)
Working	−2.004 ***	−0.397 *	−1.843 **	−0.325
	(0.654)	(0.203)	(0.717)	(0.224)
Language proficiency	−1.075 **	−0.325 **	−0.930 *	−0.338 *
	(0.499)	(0.161)	(0.545)	(0.176)
Experienced discrimination	1.427 ***	0.345 ***	1.552 ***	0.370 ***
	(0.280)	(0.079)	(0.302)	(0.088)
Housing satisfaction	−0.240 ***	−0.061 **	−0.204 **	−0.063 **
	(0.091)	(0.028)	(0.098)	(0.032)
Country of origin (reference group: Afghanistan)
Iraq	1.646	0.553	1.902	0.687
	(1.280)	(0.365)	(1.372)	(0.419)
Syria	1.318	0.386	1.274	0.515
	(0.983)	(0.307)	(1.022)	(0.348)
Iran	−0.195	0.088	1.053	0.521
	(3.023)	(1.028)	(3.523)	(1.222)
Mode of interview (reference group: CAWI—by e-mail)
CAWI—Panel	−2.530 ***	−0.852 **	−2.346 **	−0.732 **
	(0.928)	(0.353)	(1.038)	(0.373)
CASI/CAPI—at refugee facilities/centres	−2.015 **	−0.330	−2.082 **	−0.341
	(0.879)	(0.293)	(0.897)	(0.313)
CAWI—invitation by mail	0.504	−0.088	0.644	−0.091
	(1.361)	(0.348)	(1.406)	(0.368)
Constant	9.585 ***	−4.523 ***	9.751 ***	−4.948 ***
	(2.472)	(0.824)	(3.179)	(1.108)
Observations	794	792	691	687
R-squared	0.272	0.173	0.294	0.186
Province fixed effects	YES	YES	YES	YES

Note: Robust standard errors in parentheses; *** p < 0.01, ** p < 0.05, and * p < 0.1. Source: FIMAS+INTEGRATION, own calculations.

. As mentioned above already, when interpreting the results, we should be aware of the existence of endogeneity. Thus, in the case of a number of variables, strictly speaking, we cannot assume causal effects but just correlations.

Since the interviewed refugees did not answer all the questions in the various thematic fields, the regression includes 794 individuals at most.1 The signs of the coefficients of the standard explanatory variables are as expected; however, not all of those are significant. The level of mental distress turned out to be substantially higher for younger refugees. The age groups 15–24 and 25–34 show much higher levels of mental distress than the reference group aged 35–44 years. The result of higher distress for older respondents (45–60 years) is not significant (compared to the reference group).

In our descriptive analysis, women were found to be more mentally distressed than men. However, in our multivariate regression, this result is only significant at the 10 percent level—other moderators seem to have a stronger influence. When examining the effects of physical health problems, we can see that currently experienced physical pain is associated with the mental state of a person.

The only variable capturing some of the stress experienced when fleeing to Austria is the duration of migration from home to host country. However, in our regression analysis, it does not seem to explain differences in mental distress levels; other factors seem to be more important. One may expect refugees’ stress levels to decline after their arrival in Austria. However, our variable on the time passed since arrival in Austria (captured in months) does not reveal a clear relationship between distress levels and time of residence in the host country. As expected, the stress level of the small group of respondents (6% in the sample) who were still in the asylum application procedure was different. The uncertainty about their future results in a significantly higher risk for mental health problems.

Regarding family context, our results show that being in a partnership or having children does not significantly alter the level of distress. However, if family members have been left behind in the home or a foreign country, the worry about their safety significantly increases the risk of mental health problems. This is the case both for those who have a partner who does not live with them and for those who have children living elsewhere. If further close family members are living in the same household, people report a lower level of symptoms of distress.

In addition to good close relationships, the possibility to interact in the wider social sphere is also expected to reduce stress levels: Our social network variable shows that the larger the group of people in Austria that refugees feel close to and are appreciated as important to them, the lower their level of mental distress. The squared term of the variable indicates that the additional effect diminishes when the size of the social network increases. The variable on quality of relationships, i.e., having someone to talk to about personal problems, does not provide an additional significant explanation. One of the aspects of acculturative stress after having arrived in Austria is to be confronted with discrimination. The more frequent discrimination is experienced, the higher the level of distress for refugees.

Having a proper place to live has an important effect on individuals’ well-being. Accordingly, satisfaction with the housing situation in our data is significantly correlated with mental distress levels. Similarly, having the possibility to work is an important way to experience self-efficacy. Those who have a job or do voluntary work experience lower levels of mental distress.

Education is expected to be a protective factor for (mental) health, in general, since people with higher education levels experience higher levels of self-affirmation and pursue healthier lifestyles (Zajacova and Lawrence 2018). However, our results are vice versa. This could be an effect of the loss of status and the fear of non-recognition of qualifications when immigrating as a refugee, which might be more strongly perceived by highly educated people. Porter and Haslam (2005) found the same result in the literature on the mental health of refugees.

Language proficiency was significantly associated with mental health: the higher the language skills (subjective assessment of reading and writing German), the lower the conditional level of distress of refugees. This shows that acculturative stress can be reduced if being in contact with the host population is easier.

Finally, we applied controls for country of origin, province of residence, and modes of interview. The first two are not significantly associated with mental distress. The mode of the interview, however, seems to have an influence on how severe symptoms of stress are reported by interviewees. Refugees interviewed via computer-assisted self-administered interview (CASI) and those participating in a previous wave of the survey (CAWI-Panel) stated lower levels of mental distress.

In order to test the robustness of our results, we also estimated logit regression model [2], where we distinguish between those having no or a low level of distress, on the one hand, and those having a stress level that indicates a high probability of having a moderate or severe mental health problem, on the other hand. The results in Table 2 are very similar to those of regression model [1]. However, the significance vanishes in the case of the following variables: ‘Living with further family members in the same household’ and ‘someone to talk to about personal problems’. The significance level of the variable ‘Working’ is substantially reduced.

As a further test for the robustness of the results, we performed both regression models restricted to those individuals who arrived in Austria over the shorter time span 2014–2018. The results of models [3] and [4] are very similar to those of regression models [1] and [2]. However, the significance vanishes in the case of the variable ‘Having a partner, who is living in home country or other foreign country’ and is reduced in the case of ‘Language proficiency’ and ‘Working’. The latter results seem to be driven by the fact that the share of respondents who have higher language proficiency or are already working is obviously lower in the group of more recently arrived refugees. The significance of the variable ‘Living with further family members’, however, increases.

5. Discussion

This paper focuses on the mental health state of refugees who arrived in Austria mainly between 2013 and 2018, originating from Syria, Afghanistan, Iraq, and Iran. In order to measure mental distress, we applied the Kessler index, which captures the level of non-specific psychological distress and thus the likelihood of serious mental health problems. Based on survey data from 1518 individuals, we found that a high share of refugees (32%) state symptoms of moderate or severe mental health problems. The descriptive results show a higher prevalence in the cases of younger persons, women, refugees from Iran and Syria, and those individuals who arrived in Austria in 2015. However, in our multivariate analysis, only young refugees (15–34 years) show higher risk levels, while country of origin, gender, and length of stay in the host country become insignificant, when taking other mediating factors into account. Investigating the effects of stressors on the mental health situation, we found a positive association with experienced discrimination in Austria, physical pain, and the fear for partners and children left behind.

On the other hand, the results show a negative correlation for a couple of mitigating factors that foster resilience, i.e., proficiency in the German language, being employed (including volunteer work), having more supportive relationships, and satisfaction with the housing situation. Our study thus contributes to the literature on forced migrants’ resilience in terms of coping with mental distress caused by the experiences of conflict, forced migration, and resettlement (Earnest et al. 2015; Marsh 2012). As Ager and Strang (2008) show in their widely received model of refugee integration, mental and physical health is closely linked to other markers of (and means to) integration, including employment, housing, and education, leading to a sense of belonging. Our results corroborate the close entanglement of (mental) health outcomes with refugees’ social and economic integration, as strong social ties, good proficiency in the host country’s language, employment, and secure housing can act as resilience factors.

The high rates of mentally distressed persons found in our analysis are in line with other refugee surveys where the Kessler 10 scale was applied to measure mental distress. In Australia, for example, almost 2400 refugees were interviewed between 2013 and 2016 in the three waves of the longitudinal survey ‘Building a New Life in Australia’ (De Maio et al. 2017). Between 35 per cent and 46 per cent of the refugee population showed moderate or severe mental health problems, whereas the rate in the general Australian population is 7 per cent among men and 11 per cent among women. The data also showed a high correlation between the level of mental distress and experienced traumatic events. Post-migration risk factors were found to explain part of the mental health problems experienced by refugees (Chen et al. 2017).

Our regression analysis shows significantly higher risk for younger age groups (15–34 years) of experiencing mental distress than middle- and older-aged respondents (35–60 years). As the theoretical model of Correa-Velez et al. (2010) on factors predicting well-being among young refugees shows, living with a family is a factor significantly associated with greater well-being among refugees, which is corroborated by our study. This constitutes a particular risk factor, individually as well as economically, since young refugees are at the beginning of their life and career in Austria. Periods of inactivity in early phases of labour market careers have long-lasting scarring effects (Marbach et al. 2018). Several limitations apply to our study. First, our sample is not fully representative of the refugee population that was addressed in our analysis. As described in the data section above, about a quarter of the individuals were selected by applying convenience sampling, while the rest were selected by random sampling of refugees who were registered with the Austrian Public Employment Service (AMS). Being registered at the AMS was the case for the majority of refugees in asylum status, but not all of the refugees were in asylum or had subsidiary protection status. Thus, although our data describes the refugee population analysed well, it is not a random sample, which causes several selectivity problems. We can thus expect the individuals in our dataset to be more economically and socially active than the average refugee population. Second, assuming that people with severe mental illness participate in surveys less often, again, our sample is biased towards more socially active persons. Our results thus underestimate the prevalence of mental health disorders. Third, our cross-sectional data and methods of regression analysis do not allow us to draw strict causal conclusions but only correlations. In the case of some mediating variables (e.g., family members left behind in the country of origin), the causal effect on mental distress is most likely. However, in the case of perceived ‘experienced discrimination’, higher distress levels can also result in selective attention and perception of discrimination. Thus, we cannot rule out reciprocal relationships between explaining and explained variables. Thus, the results of the analysis must be interpreted carefully. Only analysis based on, e.g., panel data would allow us to draw strict causal conclusions.

Fourth, our set of variables, which attempts to capture moderating effects, is almost completely restricted to post-migration risk factors. Due to ethical considerations, no questions on potentially traumatic events before or during migration were included, as they would require a clinical context.

A strength of our study lies in the identification of positive and negative factors moderating mental distress among refugees in one of the biggest European host countries of refugees from the Middle East, relative to population size. For resilient factors, we find social connections, good proficiency in the host country’s language, secure housing, and employment to show a negative correlation with mental distress, which provides further empirical evidence for the centrality of refugees’ social and economic integration into host communities to mitigate mental distress and prevent high follow-up costs in secondary and tertiary care. We thus contribute to a growing body of research on refugees’ mental health, which continues to be poorly understood and seldom adequately addressed by host community policies, thus perpetuating inequalities in health and impeding refugees’ integration into society.

6. Conclusions and Recommendations

The high prevalence of mental health problems found in our analysis is a serious barrier to the social and economic integration of refugees in Austria. According to the Nationale Akademie der Wissenschaften Leopoldina [German National Academy of Sciences Leopoldina] (2018), refugees’ mental health problems can, for example, lead to “a high drop-out rate in language courses or the fact that many refugees cannot actively shape their everyday lives” (Nationale Akademie der Wissenschaften Leopoldina [German National Academy of Sciences Leopoldina] 2018, p. 8, own translation). In addition, psychological strain can pose a greater risk of dissocial (auto)aggressive behaviour (Collier 2014). Parental mental health problems also have a significant positive association with internalizing and externalizing behaviour problems in children (van Ee et al. 2016). Thus, adverse parenting styles, including neglect and/or aggression, can also lead to cycles of violence, intensified by epigenetic processes (Ullmann et al. 2018). We derive several recommendations from our study: The extent of mental health problems found in the surveyed refugee population shows the need for expansion in psychotherapeutic and psychiatric capacities. Further, barriers to health care access have to be lowered for refugees in Austria, in line with a previous study on displaced persons in Austria (Kohlenberger et al. 2019). In particular, sufficient financing of easily accessible psychosocial services, taking into account refugees’ cultural backgrounds and specific needs (mother tongue therapists and interpreters), is necessary. Facilities and organizations offering such services exist, at least in most urban regions in Austria, for example, Hemayat, SINTEM Caritas, and others. Sustainable funding should be guaranteed since the services will be required on a long-term basis, as our regression analysis shows that the length of stay in Austria does not reduce the level of mental distress.

Regarding the higher risk for younger age groups, investments in tailored education and training measures are necessary, as are integrative leisure activities, in order to prevent creating a lost generation.

As refugees in the asylum application procedure show elevated levels of mental distress, a reduction in the length of the asylum application procedure helps reduce the occurrence of mental health problems later on, triggered by long periods of chronic stress. Early access to the labour market is also effective in reducing mental distress but is currently highly restricted for asylum seekers. In general, early labour market access not only encourages self-efficacy and reduces the costs of public support for refugees (in Austria, the so-called ‘Grundversorgung’) but also has positive effects on the later careers of refugees. Furthermore, the recognition of qualifications and the improvement of skills (via training, etc.) reduce the loss of human capital, support general well-being, and counteract the feeling of loss of status.

Family reunions and housing conditions that allow families to live together foster well-being and assist in adjusting to life in the host country. Language proficiency was found to be significantly related to stress levels. The financing of a sufficiently large number of language courses to facilitate interaction with the host population supports closer contact with the host society. There are a variety of routes to obtaining this closer contact: fostering inclusive housing, leisure activities, or training measures.