Soixante adolescents scolarisés volontaires ont été évalués par les versions françaises de la CES-D (Center for Epidemiological Studies – Depression Scale), du MINI (Mini International Neuropsychiatric Interview) et du SIDP (Structured Interview for DSM IV Personality). L'importance de la symptomatologie dépressive montrait que ces adolescents n'étaient pas représentatifs des adolescents scolarisés. La personnalité limite s'est montrée la plus fréquente (26,7 %). Les autres troubles de la personnalité sont apparus principalement comme une comorbidité du trouble limite. En l'absence de la personnalité limite, les autres troubles sont rares et plus rarement encore associés. La consistance interne de la personnalité limite, évaluée par le coefficient alpha de Cronbach (0,82), s'est montrée élevée et les analyses factorielles exploratoires puis confirmatoires ont retenu une structure monofactorielle, argument en faveur de la validité de ce diagnostic à l'adolescence. Toutefois, la comorbidité élevée avec la dépression majeure et les troubles de la personnalité des trois clusters apparaît compromettre la validité de construit de la personnalité limite du DSM IV à l'adolescence. La symptomatologie limite à l'adolescence paraît plus en adéquation avec un modèle dimensionnel qu'avec une classification typologique.

Objective – To examine the comorbidity of borderline personality disorder and other personality disorders in a nonclinical sample of high-school students. Method – 311 high-school students who completed the French version of the CES-D (Center for Epidemiological Studies – Depression Scale), were asked to participate to interviews evaluating personality functioning : 60 subjects (19 %) accepted to participate in the study. The mean CES-D score of these 60 subjects (16 boys, 44 girls, mean age = 17.7 1.7) was significantly higher than the mean score of the whole sample (23.9 10.4 versus 16.7 9.8). Thus the interviewed sample was not representative of the population of high-school students. Subjects were assessed using the major depressive episode module of the MINI (Mini International Neuropsychiatric Interview) and the SIDP IV (Structured Interview for DSM IV Personality). Inter-rater reliability was determined by comparing the independent ratings of interviewers and an experienced clinician on a random sample of 20 interviews. For DSM IV borderline personality disorder diagnosis, the Cohen's kappa coefficient was 0.85. For personality disorder criteria, kappa ranged from 0.6 to 1.0 (average kappa = 0.79). Results – Sixteen of these subjects (26,7 %, 4 males, 12 females) received a diagnosis of borderline personality disorder according to DSM IV criteria. The mean CES-D score of borderline subjects (30.6 10.2) was significantly higher than the mean score of nonborderline subjects (21.6 10.5). Of the 16 borderline subjects, 11 (75 %) received a diagnosis of major depressive disorder versus 14 (31 %) of the non borderline subjects. None of the other personality disorders approached the frequency of borderline personality disorder. The next most frequent diagnoses were depressive and dependent personality disorders which occurred in respectively in 16.6 % and 10 % of the 60 subjects. All the personality disorders occurred at higher rates in the group with borderline personality disorder with the exception of obsessive-compulsive personality disorder which was diagnosed only in nonborderline subjects. Of the 16 borderline subjects, 11 (68.7 %) met the criteria for another personality disorder which were depressive personality disorder (N = 5), paranoid personality disorder (N = 4), dependent personality disorder (N = 3), antisocial personality disorder (N = 2), histrionic personality disorder (N = 2), avoidant personality disorder (N = 2), negativistic personality disorder (N = 2), schizotypal personality disorder (N = 1), narcissistic personality disorder (N = 1), self-defeating personality disorder (N = 1). The optional diagnoses (self-defeating, depressive and negativistic personality disorders) accounted for 8 of 23 (34.7 %) cases of personality disorders diagnosed among borderline subjects. Among these 11 adolescents, 5 received 2 diagnoses of personality disorders (borderline and paranoid personality disorders, N = 1 ; borderline and dependent personality disorders, N = 1 ; borderline and depressive personality disorders, N = 3), 3 received 3 diagnoses (borderline, antisocial and histrionic personality disorders, N = 1 ; borderline, avoidant and negativistic personality disorders, N = 1 ; borderline, depressive and negativistic personality disorders, N = 1), 3 received 5 diagnoses (borderline, paranoid, histrionic, narcissistic and dependent personality disorders, N = 1 ; borderline, paranoid, dependent, avoidant and depressive personality disorders, N = 1 ; borderline, paranoid, schizotypal, antisocial and self-defeating personality disorders, N = 1). Among the 44 adolescents (12 boys, 32 girls) without borderline personality disorder, 10 (22.7 %) (3 boys, 7 girls) met the criteria for another personality disorder which were depressive personality disorder (N = 5) or cluster C disorders -obsessive-compulsive personality disorder (N = 4), dependent personality disorder (N = 2), avoidant personality disorder (N = 1) - with the exception of one diagnosis of histrionic personality disorder. Two subjects received 2 diagnoses (obsessive-compulsive and depressive personality disorder). The internal consistency of personality disorders criteria was assessed with Cronbach's alpha coefficient. Borderline personality disorder criteria had high internal consistency (0.82). The factor structure of borderline personality disorder criteria was studied with an exploratory factorial analysis which extracted three factors. The eigenvalues were 3.70, 1.06, and 1.01. Confirmatory factorial analyses were conducted. The correlated two-factor model and the three-factor model fit the data well but the correlation between factors was, however, judged too high, ranging from 0.70 to 0.78. The one-factor model proved to have a good fit (Goodness of Fit Index = 0.89, Comparative Fit Index = 0.90, Root Mean Square Residual = 0.07). As a previous study showed the frequency of two schizotypal personality disorder criteria (odd beliefs/magical thinking experiences and unusual perceptual experiences), an exploratory factorial analysis was performed on the combined set of criteria of borderline and schizotypal personality disorders. It yielded 2 factors : the first factor consisted of all the borderline personality disorder criteria, odd beliefs/magical thinking, and unusual perceptual experiences and could be called the borderline factor ; the second factor consisted of the paranoid and the social avoidance criteria and could be called the interpersonal hypersensitivity factor. A confirmatory factor analysis showed that this two-factor model provided a good fit to the data (GFI = 0.82, CFI = 0,91, RMSR = 0.10). The correlation between factors was weak (0.25). These results suggest that odd beliefs/magical thinking and unusual perceptual experiences are a component of borderline symptomatology in adolescents. Discussion – The high frequency of major depressive disorder and personality disorders in the interviewed sample may be due to the possibility that adolescents with psychological problems have used the interview as a way to obtain attention and support from a psychologist. The interviewed sample, which was characterized by a high intensity of depressive symptomatology and by a high frequency of borderline personality disorder, could thus be seen as intermediate between a clinical and a community sample. Our results may be more generalizable to an outpatients population of adolescents. This study found conflicting results about the construct validity of borderline personality disorder in adolescent. The high internal consistency and the one-factor structure of the borderline personality disorder criteria argue for their validity in adolescents. However, the high rates of comorbidity of borderline personality disorder with depression and other personality disorders, extended to clusters A, B and C and to optional diagnoses, suggest the lack of construct validity of either borderline personality or cluster B disorders in adolescents. Conclusion – Borderline symptomatology in adolescents appears more in adequacy with a dimensional model than with a typological classification. More studies are needed to assess and improve the construct validity of borderline personality disorder in adolescents.