INTRODUCTION

This work deals with the comparative study of two standardised instruments, which can be used to diagnose personality disorders (PD): the SCID-II and the DIP. Each instrument used as a self-questionnaire followed by a semi-structured interview by the same clinician was applied to 21 patients suffering from PD. The DIP (DSM-IV and ICD-10 Personality), which is a recent instrument, consists of a self-questionnaire (DIP-Q) and a semi-structured interview (DIP-I), created by Bodlund and Ottosson. It makes it possible to evaluate PD from criteria based on the DSM-IV as well as the ICD-10. We translated it into French then evaluated it in comparison with another instrument, the Structured Clinical Interview for DSM-IV Axis II PD (SCID-II) whose validity was demonstrated by Bouvard.

METHOD

For the self-questionnaire (SCID-auto), we used CUNGI'S computerised version. The present version of the semi-structured interview SCID-E (French translation by Bouvard et al.) evaluates the 10 PD of the DSM-IV, the depressive personality and the passive-aggressive personality, included in the DSM-IV appendix B. The DIP-Q questionnaire is made up of 140 right/wrong items referring to the 10 PD of the DSM-IV and the eight disorders of the ICD-10. The DIP-I is the self-structured interview created by Ottosson et al. and it is built on the same pattern as the SCID-II. It provides diagnoses for all DSM-IV and/or ICD-10 PD as well as the schizotypic disorder. The DIP-I is usually preceded by a general "scan" interview in order to assess an existing personality disorder corresponding to Axis I of the DSM-IV or the ICD-10. In our study, we substituted a Mini International Neuropsychiatric Interview (MINI) questionnaire for this interview. Twenty-four patients suffering from one or several PD were chosen among ambulatory or out-patients by clinicians from the Saint-Etienne Psychiatric University Hospital Center. The diagnosis was not revealed to the examiner during the study. The subjects filled in the DIP-I and the SCID-II self-questionnaires. The answers to each test were first processed through a computer, then the patients were seen over the following weeks for the DIP-I and SCID-II semi-structured interviews. For both questionnaires, we only explored the diagnostic categories reaching pathological level (as was recommended by the authors). Considering the small number of patients involved, we used nonparametric tests: Wilcoxon test, Mac Nemar test and the Kappa.

RESULTS

As far as the self-questionnaire results are concerned, we noticed important differences for the schizoid and the schizotypic PD between the DIP-Q (ICD) and the DIP-Q (DSM). The most represented PDs are the paranoiac, borderline, avoiding and obsessional personalities. After the semi-structured interviews, it appears that only 30 to 50% of the diagnoses obtained through self-evaluation were confirmed (with the exception of the schizotypic personality and the antisocial personality for the SCID with perfect agreement between self and clinical evaluation). Globally, the agreement between diagnosis by self-evaluation and diagnosis by semi-structured interview is not very satisfactory. Finally, a cluster analysis of the results of the three semi-structured interviews put together reveals that five patients show at least one PD diagnosed in the three clusters, two have no diagnosis, six patients have one or several PDs in clusters B and C, three patients have some in clusters A and C, and five patients only have some in cluster C. Our results lead to several remarks: the size of our group is small, but it must be pointed out that the investigations for each patient took about three hours, which made it difficult for the patients to agree when the clinicians proposed the study; three patients originally included could not be evaluated because of suicidal behaviour. In their self-administered form, the SCID and the DSM version of the DIP-Q broadly diagnose a little more than three PDs per patient, whereas the ICD version of the DIP-Q diagnoses more than five. The administration of semi-structured interviews leads to an average of 1.3 diagnosis for the DIP-Q DSM-IV and 1.6 for the ICD against 1.9 PD for the SCID interview. These results correspond to the literature data. There are differences between the SCID and the DIP-I, as regards to the way they were used: the SCID-II makes it necessary to repeat the questions positively answered in the self-questionnaire, whereas the DIP-I explores all the criteria of the whole diagnosed PD, which may favour the inclusions. Concerning other instruments compared to the SCID-II in the international literature, our results with the DIP are globally satisfactory.

CONCLUSION

The results must be interpreted with some care, considering the small number of patients. Important discrepancies were noticed between the diagnoses obtained through self-evaluation and the semi-structured interview, mainly for the A and C personality clusters of the DSM-IV, showing the tests to be extremely sensitive, but not specific enough for detection. However, the agreement between both instruments referring to the DSM-IV is satisfactory. The main interest of our work was to make the first French translation of the DIP known and to compare it to another instrument, which has often been evaluated previously.