Using the death/suicide implicit association task to prospectively predict near-term suicidal behavior in high-risk veterans

Abstract

The 90-day period after a suicide attempt or hospitalization for suicidal behavior is a period of increased risk for psychiatric patients. However, predicting who among patients at suicide risk will engage in suicidal behavior in this window remains an elusive task. Neurocognitive task performance, such as performance on the Death/Suicide Implicit Association Task (D/S IAT), shows some promise for aiding in the prediction of suicidal behavior beyond risk factors routinely assessed in clinical practice, such as self-reported suicidal ideation. This study considered whether performance on the D/S IAT could prospectively predict suicidal behavior within a 90-day window. Sixty Veterans at high risk for suicide completed the D/S IAT at multiple timepoints over a one-year period. Each testing session was then coded according to whether the participant displayed suicidal behavior within 90 days following that testing session. As a secondary aim, drift diffusion modeling (DDM) was used to estimate latent cognitive processes mediating D/S IAT performance, including decisional efficiency. Two applications of the Generalized Estimating Equations (GEE) method were used to evaluate the incremental utility of (1) D/S IAT performance and (2) DDM-derived latent variables on predicting 90-day suicidal behavior over standard suicide risk factors including suicide attempt history, major depressive or bipolar disorder, and suicidal ideation at the time of testing. Age was also included in the models given the impact of age on reaction times and thus D/S IAT performance. The odds of 90-day suicidal behavior were significantly increased as implicit bias linking the self-concept to the concept of life as opposed to death weakened. The latent cognitive process of reduced decisional efficiency towards categorizing the concept of life relative to death as “like me” predicted 90-day suicidal behavior. D/S IAT performance may add to near-term suicidal behavior prediction. Reduced decisional efficiency is emerging as a general cognitive factor implicated in suicidal behavior.