Study design and population
We established an open retrospective cohort of 1,328,397 people born 1984–1988 followed from January 1st 2002 to December 31st 2011. The population was divided into three groups. The first was persons born in Sweden to two Swedish-born parents (n = 1,267,938; 95.4%). The second and third group had arrived in Sweden between 2002 and 2011 with a permanent resident permit and were either unaccompanied refugee minors (n = 6133; 0.4%) or accompanied minors (n = 54,326; 4.0%).
To permit valid comparisons between unaccompanied refugee minors and accompanied minors, we restricted the sample to minors arriving from the same countries of origin. The accompanied minors were included regardless of if they or their parents were considered refugees or not. We excluded people without an official residence permit in Sweden—that is, undocumented migrants or people with an official asylum decision pending. Due to missing or incorrect data, we excluded 1125 persons (0.08%). In the Cox-regression analysis, we excluded additional 1957 (0.14%) persons due to incorrect data on survival times.
We extracted data from a large, longitudinal database of linked national registers, known as Psychiatry Sweden including all people officially resident in Sweden after 1 January 1932, linked via a unique personal identity number and anonymized by Statistics Sweden for research purposes. We obtained relevant outcome, exposure, and covariate data from this linkage. From Statistics Sweden we obtained: the register of the total population (RTB), to identify cohort participants and obtain basic demographic data (birth date, sex, country of birth); the multi-generation register to link participants to their parents for identification of the Swedish-born population; the register on immigration and emigration data (STATIV) to obtain migration and refugee data, including the unaccompanied refugee minors. From the National Board of Health and Welfare, we used the National Patient Register for data on psychiatry (inpatient care since 1973 and outpatient care since 2006) the Prescribed drug registry for data on prescribed and purchased drugs (since July 2005) the Cause of Death Register for causes of death (since 1951).
Psychiatric care use was studied as a combined variable (called any psychiatric care) and defined as receipt of the first time use of any psychiatric services during the study period and coded as binary. Psychiatric care use was also studied in separated: inpatient care (use of hospital psychiatric services), outpatient care (specialist care in community psychiatric services) and prescribed and purchased psychotropic medication including psychotropic prescriptions were categorised into Antidepressants (Anatomic Therapeutic Chemical [ATC] code N06A), ADHD medications (ATC N06BA01 to NO6BA04, N06BA09), Antipsychotics (ATC N05A), Anxiolytics (ATC N05A) and Sedatives/Hypnotics (ATC N05C). The type of care is not diagnose specific, however, inpatient care is care given for patients presenting with severe mental health and in need of comprehensive support and this type of care is nowadays relatively rare. The prescriptions include those from general practitioners and specialists in outpatient settings, but not inpatient settings (0.2% of the population), as the latter is not recorded in the Swedish Prescribed Drug Register. All children in Sweden, including both unaccompanied refugee minors and accompanied minors are entitled to the same healthcare as Swedish-born minors. Sweden has a universal health insurance system with small out-of-pocket costs. During the time of the study, patients paid a maximum 2200 SEK (c. £220) per year for prescribed psychotropic drugs and the rest was paid by state subsidies .
The second outcome, for the migrant group only, was time between immigration to first psychiatric care, measured in years.
The primary exposure was minor status. Unaccompanied refugee minors were compared with accompanied minors and Swedish-born minors.
We included age, as a continuous variable, and sex as potential confounder.
Demographic variables were compared using Chi-square tests. We reported basic descriptive statistics (Table 1). We fitted time-dependent Cox models to estimate hazard ratios (HR) for in- and outpatient care and psychotropic medication with 95% confidence intervals (95% CIs) (Table 2). Follow-up time used time in the cohort as underlying timescale adjusted for age and sex. Multivariate logistic regression was used to estimate odd ratios for specific psychotropic outcomes adjusted for age and sex (Table 3). Time to psychiatric care was compared using T tests. Statistical analysis was performed using SPSS and SAS.
Sensitivity test of age
There are discussions of whether the age of unaccompanied refugee minors who immigrate close to age 18, is correct . To test if misclassification of age among those close to 18 years old could influence the results we made a sensitive test. This was done by creating a subset excluding everybody aged 17 at immigration and running the analysis in this subset.
This research has ethical approval as part of Psychiatry Sweden “Psykisk ohälsa, psykiatrisk sjukdom: förekomst och etiologi,” approved by the Stockholm Regional Ethical Review Board (Number 2010/1185-31/5).