Clasificación de la conducta suicida utilizando cuestionarios psicométricos

Résumé

Background: In order to improve the prediction of suicide, many studies have sought to describe and specify those factors that are associated with increased risk. Yet, only a few studies have provided robust and highly predictive indices able to identify those subjects who are more likely to consummate suicide. The main objective of this thesis is to classify suicidal behavior by using pattern recognition techniques in psychometric questionnaires. In addition, it is intended to create a global classification scale of suicidal behavior by choosing the most predictive items of the scales used in this study. Methods: The cross-sectional study was conducted in a global sample of 2.256 patients, 47% of whom had a history of suicide and 44.4% had no previous history. Specific subsamples were obtained, and completed the different questionnaires (BIS-11, IPDE-SQ, SRRS, BG). The analysis was carried out by using innovative pattern recognition techniques (Boosting, Support vector machines, Elastic-Net or Fisher linear discriminant analysis). Results: The best scales in the identification of suicidal behaviour were the IPDE-SQ and the SRRS. Therefore, the items most closely related to suicidal behavior were those associated with personality disorders and potentially stressful life events. The feelings of emptiness, rage and anger, marital conflict and changing personal habits are particularly discriminative and helpful to identify subjects at risk for suicidal behaviors. The Global Classification Scale of Suicidal Behavior, developed by choosing the most discriminative items, obtained an accuracy of 85.4%, a sensitivity of 77.3%, a specificity of 89.8%, and a positive likelihood ratio of 7.57, which is an improvement in the classification and prediction of suicidal behaviors with respect to previous instruments. Conclusion: Currently, the prediction of suicide is a difficult task to achieve. Future studies should be aimed at creating new tools to improve the identification of subjects with increased suicidal risk. This thesis is just a first step in the validation of the Global Classification Scale. Future studies are needed to further explore the applicability of the Global Classification Scale in clinical practice.