Search | Search by Center | Search by Source | Keywords in Title
Ilgen M, Tiet Q, Byrnes H, Moos R. Risk of suicide attempt in substance use disorder patients. Paper presented at: Research Society on Alcoholism Annual Scientific Meeting; 2005 Jun 1; Santa Barbara, CA.
Introduction: Suicide risk is higher among people with substance use disorders (SUD) than in the general population. Only a few studies have attempted to empirically test models for predicting suicide and these have generally yielded low sensitivity and/or specificity for identifying suicidal patients. Given the difficulties in identifying the risk of suicide, or suicide attempt, clinicians and researchers could benefit from the development of a decision tree that is grounded in empirical evidence. Methods: SUD treatment providers within the Veterans Administration (VA) healthcare system assessed all patients at entry into treatment using the Addiction Severity Index (ASI) as part of a nationwide evaluation of the process and outcomes of treatment. Clinical staff interviewed and obtained baseline ASI assessments from a total of 34,251 patients in 150 VA facilities nationwide during fiscal year 2000. The present study focused on those individuals who reported suicidal thoughts during this baseline interview. A Receiver Operating Characteristic (ROC) analysis was used to identify relationships between items on the ASI that maximally discriminated between patients with and those without a suicide attempt at baseline. Results: Out of the 5,671 individuals who reported suicidal ideation, 1,031 (18%) reported a suicide attempt within 30 days of the administration of the ASI at baseline. A history of a prior suicide attempt maximally discriminated between patients with and without a baseline suicide attempt. However, the relationship between lifetime suicidality and risk of baseline suicide attempt was not uniform for all participants. Specifically, the relationship between past suicidality and recent suicide attempt was moderated by recent alcohol intoxication, cocaine use, hallucinations, and exposure to violence, past suicidal ideation, and employment history. An empirically derived decision tree is presented to aid in understanding the higher-order interactions detected by the ROC analysis. Conclusions: The present results indicate that investigating the interrelationships between predictors of suicidality can greatly improve the sensitivity for detection of a recent suicide attempt. Because information from the ASI is readily available to SUD researchers and treatment professionals, the findings are more meaningful and easily applicable to SUD research and treatment practices.