Dépistage des troubles de la personnalité avec la version française du Personality Diagnostic Questionnaire-4+ dans une population psychiatrique : une étude préliminaire

Abstract

BackgroundWe choose to examine the Personality Diagnostic Questionnaire-4+ (PDQ-4+), considered as one of the most used and representative measure for personality disorders as they are referenced in the DSM-IV. The PDQ-4+ is a self-report questionnaire assessing the ten personality disorders and the two additional personality disorders derived from the DSM-IV. Participants must answer by true or false to the 99 items, each item represents a personality disorder’ criteria. Few researches have pointed out the psychometric properties of the original version of the PDQ-4+. Validating studies on the French version have been rarely conducted and only among students and teenagers. Clinical samples have been laid aside. ObjectiveConsidering the importance of the assessment of personality disorders, the aim of the present study was to evaluate the main psychometric properties of the French version of the PDQ-4+ as a screening tool for borderline, antisocial, schizotypal, schizoid, obsessional-compulsive, narcissistic, histrionic, paranoid, dependent and avoidant personality disorders, in a psychiatric population. Materials and methodsA sample of 137 French psychiatric patients completed the French versions of the Structured Interview for DSM-IV Personality disorder-IV (SIDP-IV) and the PDQ-4+. Convergent validity, sensibility, specificity, positive and negative predictive values, and internal consistency have been evaluated. ResultsThe level of agreement between the PDQ-4+ and the SIDP-IV was very low for the majority of personality disorders and low for the dependent personality disorder. The sensitivity was high for antisocial and borderline personality disorders and moderate or low for the other. Only dependent (0.44) and borderline (0.39) personality disorders had a moderate predictive positive value, associated with a high negative predictive value (0.82 and 0.79). The other personality disorders had a low positive predictive value (ranged from 0.00 to 0.28) and a high negative predictive value (ranged from 0.76 to 1.00). The specificity was high for five disorders and moderate for the other. Internal consistency of the PDQ-4+ and its subscales was unsatisfying. Correlation coefficients between each PDQ-4+ subscales were low or moderate. ConclusionsAs other self-report measure, the PDQ-4+ screened more personality disorders than the clinical interview. Besides, even if the two diagnostic methods evaluated more borderline and avoidant personality disorders, level of agreements for each personality disorder were generally low, in concordance with previous studies. Only dependent personality disorder emerged with low agreement, moderate specificity and positive predictive value, and high sensitivity and negative predictive value. The influence of the cut-off scores on these results could be raised. The internal consistency for each personality disorders was unsatisfying and lower than other French and international validating studies. The psychometric properties of the French version of the PDQ-4+ were overall unsatisfying. This tool allows a large screening of personality disorders and thus only appears as a useful method to screen a global personality disturbance.