Population

The population consisted of parent-accompanied asylum-seeking children 4–16 years of age living in the asylum centres managed by the Danish Red Cross at the time of data collection, October–December, 2006. Of a total of 260 children, responses were obtained for 246 individuals (95%): 239 teacher-responded questionnaires (92%) supplemented by self-reports from 88 of the 11–16-year-olds (79%).

Outcome measures

To assess asylum-seeking children's mental health, the extended version of the Strengths and Difficulties Questionnaire (SDQ) was used within a cross-sectional study design. The questionnaire is available at a web address [22] in 64 languages. It is a screening instrument for child psychiatric disorder which can detect children with psychiatric disorders with reasonable efficiency in community samples [23] as well as in vulnerable groups [24]. Furthermore, the SDQ has been successfully used in both developed and developing countries [25], which provides evidence for the cross-cultural robustness of the questionnaire. The respondents can be parents and teachers of 4–16-year-old children as well as 11–16-year-old children themselves. It is found that a combination of responses from different respondents improves the prediction of psychiatric disorder of the questionnaire [23].

The SDQ consists of 25 assertions about the child's behaviour and it can be allocated to 5 subscales with 5 items each: emotional symptoms, hyperactivity, conduct problems, peer-problems, and pro-social behaviour [22]. The Total Difficulties Score can be calculated by aggregating 20 of the assertions which can each be scored 0, 1, or 2 points in accordance with the child's behaviour. The SDQ also contains an impact supplement that asks about the burden of problems as perceived by the respondent, and the Impact Score is estimated on the basis of the degree of severity of the burdens [26]. As a measure of the asylum-seeking children's mental health, we used the overall SDQ-prediction of mental difficulties, which is calculated by a computerised algorithm based on a combination of the subscales: emotional symptoms, hyperactivity, and conduct problems as well as the Impact Score from multi-respondents. The technical details of the algorithm used are specified in [27]. The SDQ-measure can be seen as an indicator of psychopathology warranting an ICD-10 psychiatric diagnosis [22, 23]. Depending on the generated mental difficulties-sum, the children are conventionally placed in three categories: psychopathology 'unlikely,' 'possible' and 'probable'(cut-offs, 80 and 90 percentiles) [22, 23] which complies with the official SDQ manual [22, 23]. The SDQ and its conventional interpretation are detailed in [22, 26].

Data collection

Teachers of the 4–16-year-old children and the 11–16-year-old children themselves were included as respondents. The responding teachers had substantial contact with each of the children to whom they were giving scores. Parents were not chosen as respondents due to logistic problems and funding limitations.

With the parents' written permission, given after personal information to each family by the first author (SSN) to secure that the parents understood the aim of the study, the 11–16-year-olds filled in the SDQ singly in a separate room at the school. The author (SSN) was present in the room to inform the children about the investigation and to assist any child who had difficulties in understanding. The children could choose to answer the SDQ in Danish or in their mother tongue. The vast majority chose to answer the SDQ in Danish, and only eight children who had recently arrived needed, and got, an interpreter. The children who were not at school on the day of data collection and those who attended a community school were given the questionnaire during a home visit.

For the 23 of the 11–16-year-olds not taking part in the study the reasons were: untraceable, moved within or departed from the country (N = 10), parents' refusal (N = 5), rarely at home at the asylum centres (N = 5), and children's own refusal (N = 4).

Subsequently, data were linked to selected background variables which were: sex, age, nationality, family size, number of parents in Denmark, length of stay in the asylum centres, number of relocations within the asylum system, and administrative phase of the asylum application (1 = unravelling of asylum case, 2 = asylum case work and 3 = asylum denied. Yet, phase 3 also comprises asylum seekers who have obtained asylum and are awaiting placement to a municipality) [20]. These variables were found by searching the internal asylum database of the Danish Red Cross as well as the Immigrant Information Portal of the Danish Immigration Service.

Statistical methods

Based on the humanitarian announcements from the Danish Red Cross advocating a maximum of one year of stay within the asylum system [28], and political announcements from the Minister of Refugee, Immigration and Integration Affairs regarding a maximum of three relocations within the asylum system [29], we categorized length of stay as 0–12 months versus 13+, and number of relocations as 0–3 versus 4+ relocations. Supplementary analyses were based on the actual covariates as a check but did not provide added insight (not shown). Regarding the overall outcome measure, the categories psychopathology unlikely and possible were taken as one category versus psychopathology probable. To assess the association between organisational factors and mental health, binary logistic regression analyses were carried out using backwards elimination. The analyses were performed in SPSS version 15.0.

Ethical considerations

Under Danish law, only investigations which include human biological material must be reported to and approved by the Danish National Committee on Biomedical Research Ethics; this does not apply to questionnaire surveys, interview studies and register research surveys [30]. This study required permission from the Danish Data Protection Agency only [31]. This permission was granted. Families received a written introduction to the study comprising a general description of the study including the aims and methods. It was underlined that the children were anonymous and that the investigation would have no influence on the children's asylum case. The letter was translated into the mother tongues of the asylum seekers, as were the consent forms signed by parents. For the analysis, an anonymised database was created.

Strengths and limitations of the study design

The strengths of this study lie in the relatively high number of study subjects and the low non-response rate, which is rare when dealing with this vulnerable and fluctuating population. Additionally, the use of a validated and widely used screening instrument must be considered a strength. The most important limitations of the present study include the cross-sectional design and the fact that the children were not subject to a clinical investigation.

As the current study had a cross-sectional design, the observed associations between the included variables are not necessarily causal. Nonetheless, there is overwhelming evidence that the process of seeking asylum is both directly and indirectly a stressful and disturbing experience [3, 7–9, 12, 15, 35], and the longer the time spent within the asylum system the higher the risk of developing mental disorder among adults [7–9, 13]. Although, former studies on this subject have been carried out among a small number of children, they still show similar findings [3, 11]. In addition, evidence exists that families' stressful experiences and parental mental health – also during post-migration – have a harmful influence on the children [1, 3, 14, 36]. Thus, the causal interpretation of an effect of the length of stay at the asylum centres on the children's mental health is likely; however, the present estimated odds ratio values must be interpreted cautiously.

In quantifying the outcome measure of mental difficulties, there is a possibility of it being influenced by a number of uncertainties. Firstly, the children were not subject to an individual psychiatric or psychological investigation. In order to assess validity of our screening, it would have been relevant to let the children with high scores undergo an individual psychiatric investigation but because of anonymity that was not possible. Conversely, the SDQ is an internationally validated screening tool with a sensitivity of 63–85% and a specificity of 80–95% as well as positive and negative predictive values of 53–74% and 89–96%, respectively [23, 24]. Yet, there are several ways to handle the SDQ-scores and the way chosen makes some difference regarding the estimated outcomes [37]. On the background of the present study, we cannot estimate the occurrence of psychiatric disorder with certainty for which reason the strength of the association between the organisational factors and the mental health might be somewhat imprecise.

Secondly, the assistance of a researcher during the children's responses may have influenced the children's replies towards both under- and overrating of symptoms due to, for instance, reluctance to expose his or her own vulnerability or an interest in gaining sympathy. Thirdly, linguistic misunderstandings may have occurred in the children's replies to the questionnaires. However, this problem seems significantly reduced by the fact that the researcher took steps to ensure that the children understood each single question and answer category by dialogue and exemplification as well as employment of an interpreter in special cases. Furthermore, the problem was reduced by the use of different data sources. Nevertheless, the 11–16-year-old children reported a considerably higher number of symptoms relative to the teachers' replies. The children's high self-report could reflect an actual overrating arising from a wish to strengthen own asylum case by appearing mentally affected; however, it could also reflect that the 11–16-year-olds had severe mental problems that were far more common than their teachers realised. The same pattern has been seen in former investigations of children in the general population [38]. Potential language barriers limiting the teachers' observations were unlikely as the teachers completed the SDQs only for children who they knew well, implying that the children spoke reasonable Danish and had not arrived very recently. Fourthly, the use of two respondents in itself implies casting a wider net. At least with the standard cut-off scores used here, the result is that more children were seen, rightly or wrongly, as having mental difficulties. Hence, when combining teachers' and children's responses we may have overestimated the true prevalence of mental ill-health.

Factors not investigated

For a large part of the children, the measure of length of stay in the Danish asylum system represents a minimum time of risk since neither time spent during the journey to Denmark was recorded nor the time spent outside the formal asylum system. This would be the case if the child had gone into hiding for some periods or had been applying for asylum in other countries. Number of relocations embodies a minimum estimate for the same reasons. These periods have an impact on the absolute exposure of traumatic events, which can be a contributing factor to the high level of mental illness found among the children [1, 5]. The high level of mental illness is not necessarily a result of the duration of stay in Denmark, but the differences found in this study between the children who had been asylum-seeking one year or less and one year or more, respectively, are presumably true.

It is important to consider whether there might be any differences between the two groups of children (those who have stayed within the asylum system relatively briefly vs. longer) which could explain the differences in their mental health. This matter seems implausible as 1) country of origin was not found to be a significant confounder, 2) the children in question predominantly came from war-torn countries, both those who had been asylum-seeking for years (for instance children from former Yugoslavia) and those who had recently arrived (for instance from Chechnya), 3) the adjacent assumption that resourceful families who are less traumatized are more likely not to continue the asylum process is counteracted by the criteria for obtaining asylum in Denmark which are based on the UN Refugee Convention as well as other international conventions to which Denmark must acceded [19].

Although several factors related to conditions in Denmark were assessed in this study, many other factors were not addressed, such as parental mental health and number of traumatic experiences. Other studies have shown that refugee and asylum-seeking children develop mental illness in response to their parents being functionally impaired because of depression, anxiety, PTSD, or other mental problems relating to the stresses due to the asylum process [1, 3, 14, 36]. Thus, these unexplored variables may have helped to explain some of the associations found.

Generalizability

The overall response rate was 95%. Beyond that, a drop-out analysis showed no deviation of the various background variables between those participating and those not participating. The study population must therefore be assumed to be representative of the present children seeking asylum in Denmark. The existing asylum population in Denmark is characterised by long duration of stay and many relocations at the asylum centres.

Implication of asylum politics

Long duration of stay at asylum centres seems to have an adverse effect on the children's mental health. Even though some of the children might be traumatised when they arrive in Denmark, it appears that the time of stay in the asylum system may harm their mental health even more. These findings have implications for both the national and international asylum politics and they underline the importance of scientific research to support observations of health professionals and social caretakers within this field [6, 11]. To meet our ethical responsibility, we need to voice the words of the unheard by documenting the consequences of the current asylum politics on child mental health. Former studies suggest that a combination of parental, child, and environmental factors constitute the risk factors of child mental illness [1]. As the recipient countries have control of what conditions they offer children seeking asylum, they should seek to minimise the environmental risk factors. Children seeking asylum are among the most vulnerable in our societies, and it is critical that the asylum systems in Western host countries seek to protect children in accordance with the Convention of the Rights of the Child and other international rights documents.