Setting and study population
The data for this cross-sectional study was collected in the three largest cities in Sweden, Gothenburg, Stockholm and Malmö, during the period 2014–2016. Included in the study were persons over 18 years of age, defined as UMs because of any of the following: a) having applied for asylum and a residence permit but the application having been rejected and the decision having gained legal force; b) persons from outside the EU having overstayed in Sweden after their visa has expired; or c) persons having moved to Sweden without applying for a visa. Excluded from the study were migrants from EU countries who had overstayed their visa.
Sampling of the study population
Because of the fear UM commonly have of authorities, especially the fear of being disclosed and deported, this study had to use unconventional methods to make contact with and recruit respondents. We therefore aimed at creating a convenience sample. The researchers contacted representatives of various informal networks that provide support to UMs in Stockholm, Gothenburg and Malmö to make contact with UMs. The network around the Rosengrenska clinic in Gothenburg, Doctors of the World in Stockholm and the Red Cross in Stockholm and Malmö, organisations that provide informal health care and financial support to UMs, were particularly important in providing contacts with informants, but churches and other supporting networks and organisations were also contacted. These networks provide a variety of services to UMs, serve as meeting places for social support and/or provide clothes, food and health services. The majority of respondents in our study did not visit the clinic due to physical or mental problems, but because of social activities and food distribution. Within these networks, persons that held a position of trust among the migrants, often a nurse or social worker, asked potential respondents if the researchers could contact them, informed them about the study and asked them to take part. When researchers/interviewers informed the UM about the study, a translator was present so that correct information could be given in the respondent’s own language. An information brochure was handed out, presenting the study in nine languages. In cases where no translator was available at the clinic, a later meeting was set up. Both authorised translators and volunteer translators were used at the clinics for this information. The written and oral information clearly stated that the researchers were not in any way associated with any authorities, that anonymity was strict and that participation in the study would not affect their asylum cases. When an UM accepted being interviewed, an agreement was made on time and place. Most interviews were done in churches and voluntary organisations. It was crucial that the respondent felt secure and safe with the time and setting. Psychological counselling was offered if the respondent needed it after the interview. About five respondents felt such a need. The final sample consisted of 104 UMs: 54 men and 50 women. In this paper, we present findings from those respondents who answered all questions on at least one of the three psychiatric scales, resulting in 88 individuals, 48 men and 40 women.
An extensive questionnaire concerning living conditions and health was developed in the project in collaboration with a reference group consisting of four persons who were, or recently had been, in an undocumented situation. Items used in other studies in the field of migration and/or health were used together with items developed specifically to meet the aims of this study. The questionnaire consisted of 112 multiple-choice questions concerning the following topics: background, family and children, access to food, living conditions, education, work and income, maternal care, physical and mental health, access to care and experiences of care, medicine, barriers to care, social support and social networks, traumatic events in Sweden, experiences of violence in Sweden, strategies when undocumented, views on returning to the country of origin. Open-ended questions were also used. After piloting of the questionnaire, the following sections were withdrawn from the questionnaire since we realised that it was too long and potentially exhausting, and were explored in qualitative interviews instead: child health and living conditions among children, social support and social networks, traumatic events in Sweden, and strategies when undocumented.
In addition to the questionnaire, three psychiatric instruments were included in the study to screen for the respondent’s mental health.
Beck’s Depression Inventory (BDI-II) is a scale that has been widely used and has been validated in many populations and cultural groups. This instrument consists of 21-item multiple choice questions based on DSM-IV criteria for depressive disorders. The questions concern the past two weeks, and responses are rated on a 4-point Likert scale, ranging from 0 (not at all) to 3 (severely). The 21 items are combined into a single sum score, ranging from 0 to 63. A sum score of 0–13 is considered to be a minimal score of depressive symptoms, 14–19 is considered a mild, 20–28 a moderate, and a sum score of 29–63 a severe range of depressive symptoms .
Beck’s Anxiety Inventory (BAI) is a similar 21-item multiple choice instrument that assesses anxiety symptoms with a focus on somatic symptoms of anxiety that have occurred in the last week . Responses are rated on a 4-point Likert scale and range from 0 (not at all) to 3 (severely). The BAI is developed to measure anxiety that is relatively independent of depression. The total score ranges from 0 to 63. The following interpretation of scores is suggested: 0–9, normal or no anxiety; 10–18, mild to moderate; 19–29, moderate to severe; and 30–63, severe anxiety .
PCL- 5 – is a screening tool to measure post-traumatic stress disorder (PTSD). It measures symptoms according to DSM-5 during the past month. Total scores range 0–80, and suggested cut off for a PTSD diagnosis is 38 . The scale consists of an introductory text that gives examples of several stressful experiences, or likewise that may have happened to the respondent. If so, five questions about the worst event follow. If there are several traumatic events, the respondent is asked to think about the worst when answering the following 20 multiple-choice questions concerning how this stressful event has bothered him/her in the past month. In the study, the respondents did not have to describe what type of event they had experienced, just answer yes/no regarding whether they had experienced anything as in the introduction to the questionnaire: ‘a very stressful experience involving actual or threatened death, serious injury, or sexual violence. It could be something that happened to you directly, something you witnessed, or something you learned happened to a close family member or close friend. Some examples are a serious accident; fire; disaster such as a hurricane, tornado, or earthquake; physical or sexual attack or abuse; war; homicide; or suicide’. The US version 2013 was used  as well as the translated Swedish 2014 version [33, 34] depending on the preference of the interviewee.
To test the reliability of the scales in this specific population we calculated Chronbachs alpha scores; 0,904 for BAI, 0,901 for BDI and 0,893 for the PCL.
The study was approved by the Swedish Central Ethical Review Board (ref. no. Ö 25–2013). Ethical considerations with particular respect for the vulnerable situation of undocumented people were in focus in the study. The ethical principles of informed consent and the possibility of withdrawing consent at any time without any disadvantages for the reporting provider were clearly articulated and repeated. Since many people have experience of traumatic situations, the interview could arouse memories that were difficult to talk about. We therefore placed great emphasis on only using interviewers in the project who had good psychosocial skills and knowledge to handle conversations and meetings with people in vulnerable situations and had information on how to refer the respondent to a counsellor or health staff if it was needed.
Training of fieldworkers and data collection
Fifteen persons were trained to be interviewers. They had a bachelor’s degree or higher in social science, psychology or medicine. The training took approximately three hours and consisted of information about the study, knowledge about how to use the iPad/tablet for data collection where the questionnaire was placed and how to use the psychiatric instruments. Ethical aspects of performing interviews and interview techniques was also presented. After this the interviewers had support from the principal investigator (HA) and project coordinator (LA) who also performed interviews. Most interviews were conducted with the assistance of authorised translators.
Descriptive analyses with chi-2 analyses were performed with the dichotomised outcomes of the three scales according to the cut-off points defined in the validity studies. Outcomes were analysed by gender as well as overall. The depression and anxiety scales, but not the PCL-5 scale, were found to have a fairly normal distribution, and the variance of these scales was therefore explored in linear regression models. The models included various potential determinants of mental health problems: gender, age, time living as a UM, unstable housing situation, food shortage and having been persecuted or exposed to war in the country of origin.