The Predictive Validity of the Beck Depression Inventory Suicide Item

OBJECTIVE: The current study examines the predictive validity of the Beck Depression Inventory (BDI) suicide item for death by suicide and suicide attempts. METHOD: The study included 2 samples: (1) 5,200 psychiatric outpatients who were evaluated between 1975 and 1995 and followed prospectively for up to 20 years (all psychiatric diagnoses based on DSM-III and DSM-III-R), and (2) 119 patients who, between 2000 and 2004, participated in a randomized controlled trial of outpatient Cognitive Therapy for Suicide Prevention after a suicide attempt and were followed for 18 months (all psychiatric diagnoses based on DSM-IV-TR). All patients completed structured diagnostic interviews, as well as the BDI and Scale for Suicide Ideation. RESULTS: Cox regression models demonstrated that the BDI suicide item significantly predicted both deaths by suicide (Wald χ(2)1 = 35.67; P < .001 N = 5,200) and repeat suicide attempts (Wald χ(2)1 = 8.82; P < .01 N = 119), with each successive rating on the item conferring greater risk. Using receiver operating characteristic (ROC) curves, optimal cutoff scores of 1 and above for suicide and 2 and above for suicide attempts were identified as providing the best balance between sensitivity and specificity. CONCLUSIONS: The BDI suicide item is associated with both risk of repeat suicide attempts and death by suicide. The use of the item as a brief, efficient screen for suicide risk in routine clinical care is recommended. Clinicians would then conduct a comprehensive suicide risk assessment in response to a positive screen. Future research examining the item's performance in other at-risk groups (ie, older adults, adolescents, inpatients, etc) is encouraged.