This research is part of an "Interreg IV-A" project aimed at analyzing and promoting mental health in the trans-border zone of Saarland (Germany), Lorraine (France), Wallonia (Belgium) and Luxembourg (PPSM "Prévention et Promotion de la Santé Mentale"). The PPSM project consists of promoting mental health by developing knowledge on risk and resource factors, implementing campaigns to sensitise for overall well-being and establishing an international knowledge network of good practice in mental health. In Luxembourg, the evaluation and promotion of mental well-being will focus on adolescents and young adults who dropped out of school and who may be more at risk to have mental health issues.
Although school dropout is not a recent topic, yet it has never been as discussed than in today's knowledge based society. The eclecticism of definitions and studies concerning early school leaving well reflect the complexity of the subject. In this article, the terms "school dropout" and "early school leaving" will be used interchangeably.
Definitions account for national specificities such as the educational system policies and the duration of compulsory education. The Organisation for Economic Co-operation and Development (OECD) and EUROSTAT have developed indicators that allow to compare national dropout rates, however conclusions should consider limitations due to the national specificities. Both indicators define school dropout as the "termination of schooling below upper secondary qualification and the absence of any educational program or training". They slightly differ on the age span, which spreads from 18 to 24 years for EUROSTAT and only from 20 to 24 years for the OECD.
In Luxembourg, since 2003, the Ministry for Education ("Ministère de l'Eduction Nationale et de la Formation Professionnelle" MENFP) has been a third source providing information on secondary school dropout. In contrast to the OECD and EUROSTAT, the ministry records dropouts of all ages, reported by the national school institutions. For the scope of this research project, we will focus on the data released by the MENFP because it is based on actual figures.
Between 2003 and 2006, the national authorities had to face dropout rates of more than 15%. As a result, they introduced accompanying and preventive measures to reduce linguistic and cultural risk factors that may impede educational engagement and vocational decisions. Since then, the dropout rate has overall declined. According to the latest figures, the national dropout rate, for the academic year 2008/2009, was set at 9%  and hence conceded with the European benchmark of no more than 10%. However the proportion of early school leavers who remain without any specific occupation is still alarming.
When questioned on their rationale for leaving school, students reported 1) difficulties in vocational guidance, 2) academic failure or the anticipation of failure, 3) a lack of incentive to pursue any formal education, 4) the lack of an apprenticeship, 5) personal reasons, or 6) health issues . In the light of these statements, dropout prevention needs to consider what may underlie the reported difficulties, by examining factors relating to the student. Insight into the correlates that can influence alienation from school, and thus dropout, will be provided by exploring individual factors such as mental health, family life and socioeconomic situation.
Adolescence can be an unsettling period of life, marked by physical, psychological and social changes. It is known that up to 50% of lifetime mental disorders have their onset during adolescence  and that prevalence rates of at least one DSM IV disorder range from 13.7%  to 22.2% . Such disorders include substance use, disruptive behaviour disorders, anxiety and mood disorders.
Nevertheless, in Europe, the exploration of early school leaving in terms of mental health is yet to mature, and only a few recent studies can be referenced. In France, Huerre and Leroy  presented a holistic analysis of school dropout, Brandibas  considered truancy to be associated with different types of anxiety, whereas Legleye  reported an association between dropout and the progression to daily cannabis use. Jonsson  pointed out that Swedish adolescents suffering from depression were less likely than their non-depressed peers to have graduated from higher education.
In North American and Canadian research literature, several studies explored the impact of mental health factors on academic attainment. They covered a whole spectrum of psychiatric disorders, and concluded that school dropout is not necessarily associated with motivational or institutional factors, but with serious social and cognitive impairments aroused by mental illness. Meldrum  confirmed the above hypothesis by revealing that up to 15% of Canadian students interrupted their studies for reasons of mental health. A recent study of a US national sample confirmed that 12 out of 17 psychiatric disorders were associated with subsequent failure to complete secondary education by the age of 18 . After controlling for potential confounders, the authors established that bipolar and conduct disorders were most consistently related to early school leaving. In fact, several studies concluded that there was a significant association between school dropout, Attention Deficit (Hyperactivity) Disorder/AD(H)D  and disruptive behaviour disorders, also known as externalizing disorders [13, 14]. For Kessler , the association with externalizing disorders was significant for males only, whereas so-called internalizing disorders were the most important psychiatric determinants of school dropout for females. Internalizing disorders include anxiety (specific phobia, social phobia, generalized anxiety disorder, panic disorder with or without agoraphobia, obsessive compulsive disorder and posttraumatic stress disorder) and mood disorders (major depressive disorder, bipolar disorder, dysthymia). In 2008, Fletcher  confirmed these findings by detecting a significant negative correlation between depression and high school attainment for females only; whereas Jonsson  did not observe any difference in gender.
In addition to the disorder categories mentioned above, substance abuse and dependence have significantly correlated with school dropout [13, 17, 18]. In their literature review, Townsend et al. analyzed 46 studies on the correlation between substance use and school dropout, and concluded that alcohol, tobacco and cannabis use had an impact on educational attainment and vice versa .
Regarding the consensus for a multifactorial approach to psychiatric disorders, we consider it crucial to investigate for risk factors and protective factors derived from the individual and social domains as well as interactions among those factors. Socioeconomic status (SES) was reported to have an impact on both mental health  and educational attainment . Indeed, among students with a low SES, 44% of school dropout could be attributed to psychiatric disorder, whereas the proportion increased to 61% for students with a high SES .
In the same way, family variables correlated with both mental health [23, 24] and education [25, 26]. In 1996, Bernstein and Borchardt analyzed different types of family composition in school refusal. They found out that communication and role behaviour seemed to be problematic in mother-only households . According to Bernstein, difficulties in role behaviour may reflect symptoms of separation anxiety in children, who refuse to leave home because they worry about or care for a physically or emotionally vulnerable parent .
In the light of previous research, we hypothesize that school dropouts have significantly more mental health problems than their peers who go to school. They are also more likely to report a poor socioeconomic background, negative school experiences (repeating classes, disciplinary measures, absenteeism, problematic relations with authority and peers) and difficulties in family functioning (task fulfilment, role performance, communication, affective expression, connectedness, control behaviour, values and standards).
The first objective of this research is to assess the type and prevalence of psychiatric disorders among early school leavers in Luxembourg and to compare the findings with those by a control group matched for gender, age, school and educational grade. If there is an association between mental illness and school dropout in Luxembourg, is it significant for only one psychiatric disorder or more (comorbid disorders)?
As a second objective, we aim to identify specific protective and risk factors derived from the individual and family domains by including school experiences, sociodemographic background and family variables (task fulfilment, role performance, communication, affective expression, connectedness, control behaviour and values and standards) in the analysis. These thematic analyses will establish which experiences are specific to school dropouts with mental health problems, which are specific to school dropouts without mental health problems, and which to students enrolled in regular secondary education.