Le modèle alternatif des troubles de la personnalité du DSM-5 se propose d’évaluer les troubles de la personnalité en fonction du niveau de fonctionnement (critère A) et des dimensions pathologiques (critères B). L’objectif de cet article est de mettre en évidence les seuils de dépistage en population non clinique de la traduction française de deux outils brefs, la LPFS-BF de Hutsebaut et Bach et le PID 5-BF de Kruger et al.

Ces questionnaires et les items de l’évaluation symptomatique transversale de niveau 1 du DSM-5 mesurant le risque de trouble de la personnalité ont été proposés à un échantillon de 433 personnes issues de la population générale. Les seuils pour lesquels ces questionnaires dépistaient les mieux des personnes à risques de troubles de la personnalité modérés et sévères ont été déterminés à partir des indicateurs classiques des courbes ROC (sensibilité, spécificité, indice de Youden, faux positifs, faux négatifs, etc.).

Un score de 24 à la LPFS-BF présente les caractéristiques suffisantes pour évaluer les critères A du MATP. Sur la base de ces résultats, une nouvelle analyse ROC montre que le PID5-BF permettrait d’évaluer les critères B.

Les résultats sont discutés au regard de la composition de l’échantillon et des données de la littérature. L’examen des qualités du PID 5-BF à évaluer les critères B du MATP nécessiterait une recherche ultérieure prenant en compte l’évaluation des différents troubles de la personnalité et non pas la seule présence d’un trouble de la personnalité.

The aim of this work was to study whether the French versions of the brief tools available to clinicians within the framework of the Alternative Model of Personality Disorders (AMPD) can account for the risks of personality disorders in the general population. Tools are available to accurately investigate either the Level of Personality Functioning (LPF) or the Pathological Personality Dimensions (PPD) which in turn allow the validation of the relevance of the AMPD for its criteria A and B. As these tools, such as Morey's Level of Personality Functioning Scale Self Rated (LPFS-SR) for Criteria A or the Personality Inventory for DSM-5 (PID5) by Krueger et al. for Criteria B, are lengthy, the question arises as to the use of the short tools derived from them.

Data was collected from a sample of 433 people recruited on a volunteer basis with a complete protocol. The sample was predominantly female (83% female, 16% male, 2 people who did not wish to report their gender) and rather young (67% were 18–24 years old). The short version, the LPFS- BF of Hutsbaut et al., which we used in this work allows, as confirmed by several works, to consider on the basis of 12 items the global level of personality functioning. In order to assess the pathological dimensions of personality (PPD), we chose the short version of the Personality Inventory for DSM 5 (PID 5 BF) by Krueger et al. and used its validated French translation that satisfies the factor composition of the original version: Negative Affectivity, Antagonism, Detachment, Disinhibition and Psychoticism. To assess the intensity of personality disorders we used the dedicated subscale (Items 19 and 20) that the DSM 5 proposes in its Cross-Cutting Symptoms Measures of Level 1, in its French translation. A score higher than 2 was our Gold Standard when we tested the metric capacity of the two questionnaires to evaluate the A Criteria and then the B Criteria of the AMPD.

The overall results (Table 1) show levels that place the group in a non-clinical level. In terms of the severity of personality disorders it can be seen that 27 % are at risk of personality disorder (PDs>2). Comparing these two sub-groups (Table 1), we observed significant differences for all the factors studied, pointing towards a higher score for people at risk of PDs. A logistic regression analysis of the evaluation of persons at risk lead us to find that gender and age do not have a significant influence (p=0.225 and p=0.065 respectively) in a valid model (chi square=157, df=4, p<0.001) including the overall score on the LPFS (z=5.76, p<0.001) and the PID 5 (z=2.26, p<0.001). The Area Under the Curve (AUC=0.859) of this translation (Table 3) is consistent with the original version (AUC=0.84). It has metrological qualities (Sn=73.91%, Sp=85.33%, LR+=5.1, LR−=0.3005) that allowed us to use a threshold of 24 as a discriminant of a risk of moderate or severe personality disorder. In addition, if we followed the AMPD and considered the threshold of 24 on the LFPS-BF to be a risk score for personality disorder, we could see (Fig. 2) that the scores on the PID 5 BF fairly well reflected the expected pattern with a large AUC (0.901). According to the AMPD, the cut-points for the dimensions that would evoke the presence of criteria B in the case of the presence of criterion A (LPFS-BF>24) could be either a score greater than 2 for Negative Affectivity, a score greater than 0.8 for Detachment, Antagonism and Disinhibition, or a score greater than 1.2 for Psychoticism (Table 4).

The translation of the LPFS-BF that we used in this work has sufficient qualities to assess situations at risk of personality disorders when higher than 24. Its consistency was good (=0.84), and its factor composition in two factors (Self and Interpersonal Relations) was equivalent to the original version. The use of PID5-BF could therefore be used as a complement to the screening of AMPD A criteria, with a 25 for cut-point. The evaluation of the AMPD B criteria with the PID5-BF seemed relevant in view of our results; each of the subscales seemed to be able to correctly evaluate (AUC) persons with an LPFS-BF score at risk. However, the risk thresholds need to be confirmed in further work because of the essential role that the dimensions play in the diagnosis of types of personality disorders.