Objective: To examine several potential predictive screening tools for emergency department assessment of risk of self-harm.
Methods:Adult patients presenting with self-harm or suicidal ideation were enrolled at 2 emergency departments at large teaching hospitals in Edmonton, Alberta. Patients completed a brief interview assessing demographics and psychiatric history and several questionnaires (the Beck Hopelessness Scale, the Barrett Impulsiveness Scale [BIS], and the Brief Symptom Inventory [BSI]) and drug and alcohol abuse screens (Drug Abuse Screening Test and the Cut down, Annoyed, Guilt, Eye-opener [commonly referred to as CAGE] Questionnaire). At 3 months, patients were followed up via telephone and electronic health records to ascertain self-harm outcome. Questionnaires and their subscales were assessed using logistic regression. Receiver operating characteristic (ROC) analysis was performed on the results.
Results:Among the 157 patients enrolled, 49% were women and 36% (of the total) were aged 18 to 29 years. Several of the subscales of the BSI and BIS as well as the DAST-10 were significant predictors of self-harm (P < 0.05). ROC analysis showed that none of the scales in isolation were very strong predictors. Hierarchical regression analysis that combined the significant scales with clinical risk factors that were significantly related to self-harm (that is, age, education level, history of self-harm, and whether they presented with self-harm or only suicidal ideation) indicated that the BIS and DAST-10 questionnaires each added significantly to the predictive ability of a model with these risk factors.Conclusions: While many of the questionnaires and their related constructs are related to future self-harm, none of them are particularly strong and their diagnostic usefulness is limited.
W W WObjectif : Examiner plusieurs instruments de dépistage prédictifs potentiels pour l'évaluation au service d'urgence du risque d'automutilation.
Résultats :Parmi les 157 patients inscrits, 49 % étaient des femmes et 36 % (du total) étaient âgés de 18 à 29 ans. Plusieurs des sous-échelles du BSI et de la BIS ainsi que du DAST-10 étaient des prédicteurs significatifs de l'automutilation (P < 0,05). L'analyse ROC a indiqué qu'aucune des échelles en isolation n'était un prédicteur très fort. L'analyse de régression hiérarchique qui combinait les échelles significatives avec les facteurs de risque cliniques significativement liés à l'automutilation (c'est-à-dire, l'âge, le niveau d'instruction, les antécédents d'automutilation, et s'ils présentaient de l'automutilation ou seulement une idéation suicidaire) indiquait que les questionnaires BIS et DAST-10 ajoutaient chacun significativement à la capacité prédictive d'un modèle avec ces facteurs de risque.