Abstract :
[en] Considering the individual and socio-economic burden of early school leaving (ESL), the subject has gained primary concern among school professionals, researchers and political stakeholders who aim to gather the phenomenon of ESL and take action. In Luxembourg, since 2003, the ministry of education has systematically traced ESL and implemented targeted school-level interventions that certainly contributed to an overall decline of the national dropout rate. Nonetheless the proportion of young people who remain without any specific occupation after their dropout remained static, whereas the proportion of those who had resumed their education and then dropped out again even increased. These alarming observations challenge the efficiency of exclusively school-based measures and stretch the urge for considering a more holistic and preventive approach in reducing ESL. Furthermore, a significant proportion of adolescents develop a first episode of psychiatric disorder during their transition to adulthood, with incidence rates increasing by age. In this context, a primary objective of the present study was to explore, as a national novelty, prevalence rates of psychiatric disorders among early school leavers registered for the school year 2010/2011. Furthermore, the association between mental health variables as predictor of interest and early school leaving was explored by an incident-cases nationwide case-control study. Selected socio-demographic, school-related and family variables were included in the analyses as they are factors likely to interfere with both educational attainment and mental health. Cases were defined as students who had left the national educational system at some point during the school year 2010/2011 or at year-end but without having graduated. Controls were regularly enrolled students matched on age, gender, school, type of training and educational level. The presented research was jointly initiated by the University of Liège and the Centre for Health Studies of CRP-Santé in Luxembourg. Furthermore, it was part of the so-called PPSM « Prévention et Promotion de la Santé Mentale » mission, an « Interreg IV-A » project aimed at exploring and promoting mental health in the trans-border zone of Saarland (Germany), Lorraine (France), Wallonia (Belgium) and Luxembourg. Funding was granted under the AFR scheme of the « Fonds National de la Recherche ». The study protocol was fully approved by the national research ethics committee (« Comité National d’Ethique de Recherche ») and notified to the national commission for data protection (« Commission Nationale pour la Protection des Données »). Data collection was based on 200 individual interviews (100 in each group) and targeted socio-demographic variables, school experiences, psychiatric syndromes and disorder diagnoses according to DSM IV, problematic internet use and family variables.The present dissertation is divided into 5 main parts including an introduction, scope of the research, a description of the method and the results before finishing with the discussion and conclusion part. These 5 parts are again sub-divided into 15 chapters. The first part (introduction) starts with a brief general overview in chapter 1 illustrating the interregional and the national context of the study. Chapter 2 then describes the organization of the national school system whereas chapter 3 focuses on the core topic by providing a definition of ESL as well as main epidemiological results including national and international dropout statistics and identified school-related risk factors. In addition to school-related determinants, chapter 4 develops the association between mental health and educational attainment by providing an overview of relevant scientific results targeting psychosocial risk factors of ESL. The content of chapter 4 is expanded in a systematic literature review published with BMC Psychiatry.In the light of the previously exposed scientific contributions, part 2 illustrates the scope of the research project. Chapter 5 exposes hypotheses and objectives that had guided this work before raising expected contributions to scientific knowledge and practical implications. The methodology adopted to achieve the aforementioned objectives is presented in part 3 which is sub-divided into two chapters. Chapter 6 defines the case-control design, recruitment and field work procedures, data collection, quality control measures and return of results, whereas statistical analyses and evaluation of missing data are depicted in chapter 7. The detailed study protocol of this thesis had been published in BMC Public Health (cf. appendix 1).The following part consists of 4 different chapters dedicated to presenting the results of this research. Chapter 8 describes the inclusion of early school leavers into the study and evaluates their representativeness compared to the source population composed of all nationally registered school dropouts for the school year 2010/2011. Comparative analyses were run on age, gender, previous ESL, type and level of education. As they did not reveal any significant differences, participants were considered to be representative of the source population. In the same way, regarding the same characteristics, there were no significant differences between participants and non-participants. Chapter 9 illustrates the study sample in terms of socio-demographic background, school experiences, mental health and family functioning and allows to select potential predictor variables, according to a significance level of 20%, to include into subsequent multivariable analyses. Differences between cases and matched controls regarding these variables were explored with univariate conditional logistic regressions. Regarding socio-demographic variables, the decease of a parent was the only variable significantly associated with ESL. At school-level, absenteeism and being late for school were confirmed as potential predictors of ESL. Furthermore, early school leavers had significantly less positive relationships with teachers. They perceived them as less democratic, less cheerful and also less interested in the student as a person. Physical health variables did not significantly differ between cases and controls whereas mental health had a significant impact on ESL. More precisely, early school leavers reported more symptoms of withdrawn, anxiety or depression, somatic complaints without any medical reason, thought problems and aggressive behaviour. Depending on the type of syndrome, the probability of ESL increased by 6 to 20% when the syndrome score increased by one point. Diagnoses for any psychiatric disorder were highly prevalent in both groups with anxiety disorders being the most frequent. As for anxiety, the frequency of mood disorders was similar between cases and controls, but the presence of suicidal intentions was significantly associated with ESL. The same was observed for a diagnosis of antisocial personality disorder. Both groups did not differ on so-called addictive behaviours (alcohol, illicit drugs and problematic internet use). Regarding family functioning, task fulfilment, communication, connectedness, affective expression and shared values were perceived more problematic among early school leavers.Chapter 10 gives an overview of estimated prevalence rates for psychiatric disorders among early school leavers with antisocial personality disorder (19%), alcohol abuse and dependence (14% and 11% respectively), major depressive disorder (13%), agoraphobia (13%) and substance dependence (10%) being predominant diagnoses. Differences between genders were assessed with Chi Square or Fisher’s exact test and observed for the prevalence of panic disorder, agoraphobia, substance abuse and antisocial personality disorder thus confirming the hypothesis of so-called internalizing disorders being more prevalent in women, whereas men more often presented so-called externalizing disorders.After univariate conditional logistic regressions presented in chapter 9, chapter 11 provides results from multivariable logistic regressions with ESL as the outcome of interest and selected predictor variables that were below the 20% significance level in univariate analyses. The models revealed that the likelihood of dropping out of school increased with symptoms of anxiety and depression, absenteeism, poor self-rated school performance and less perceived peer support. An article presenting the results of this work is currently in preparation. The fifth part of the thesis discusses in 4 chapters the main findings in the light of previous literature and in terms of the selected matching criteria. Practical implications and guidelines for prevention strategies are drawn from the final results. The views expressed in this thesis are those of the author and do not necessarily represent those of the Ministry of Education or the Centre for Health Studies.
