Material and methods
Approximately 1700 questionnaires were distributed to RAM speaking Arabic or Dari and who participated in the mandatory public integration support programme in the Scania region of Sweden. The study population was recruited by inviting the participation of all eligible adult migrants who, between February 2015 and February 2016, were taking part in the compulsory civic and health information programme as part of the introduction plan for all RAM in Sweden [17]. In total, 681 questionnaires were returned, resulting in a response rate of approximately 39.5%. The vast majority of the study’s participants were granted their residence permit as asylum seekers or as part of family reunification; for the purpose of this research, they are considered as refugees. They have all fled from war-zones such as Syria, Iraq and Afghanistan. They were housed either in accommodations supplied by the government or in apartments arranged by themselves. Data collection was assessed by a self-administered questionnaire that included questions about health, sleep, level of education, well-being, accommodation type, social relations, work and access to health care.
Dependent variable
Mental health was assessed by the General Health Questionnaire (GHQ-12) scale [18], which is a combination of 12 questions regarding a person’s well-being and sleep. The questions ask for information about sleep, stress, concentration, decision-making, tension, problem-solving, sadness and depression. These questions were assessed together and responses were grouped into two categories: good metal health and poor mental health. If three or more of the twelve items indicated poor psychological health, the respondent’s GHQ-12 was defined as poor.
Independent variables
Crowded living was considered as the main exposure derived from the following questions: How many persons live in your house? How many rooms do you have? If more than two persons were living in the same bedroom, the housing was considered as crowded. This variable was based on a report from Statistics Sweden [19] and this way of calculating crowded living has been done in other Swedish reports as well [20].
Housing condition responses were categorised as indicating good living conditions if the person reported (a) living in a rental flat without a creditor or (b) owning an apartment. Unstable housing conditions were defined as (a) living in a rental flat with a creditor, (b) having a sub-let apartment, (c) living with friends or family or (d) living in a government accommodation facility.
Educational level was based on years of schooling, divided into low educational level (9 years or less), medium educational level (10–12 years of school) and high educational level (more than 12 years).
Age was measured as a continuous variable; but for descriptive statistics, it was divided into five categories: aged 18–34, 35–44, 45–54, 55–64 and 65–80.
Gender was divided into either male or female.
Statistical analysis
Descriptive statistics were calculated as frequencies and percentages. Logistic regression was used to analyse the association between the dependent variable mental health and the independent variable crowded living by calculating odds ratios, P-values and 95% confidence intervals. Multiple logistic regression was used to adjust the estimated odds ratios for the influence of confounding factors such as educational level, age and housing condition. The choice of variables in the adjustment model was based on previous research [21]. Statistical analyses were performed by SPSS version 22.