Using Bayesian mixed-effects models to predict self-injurious thoughts in intensive longitudinal data

Abstract

Suicide is a leading cause of death in the United States and around the world. Despite decades of research aimed at improving the understanding of self-injurious thoughts and behaviors (SITBs), researchers are not currently able to reliably predict when someone is at high-risk for suicide. This might be due to methodological limitations of prior studies which relied on retrospective recall of distal SITB risk factors (e.g., temperamental characteristics, mental health diagnoses, etc.). Research focused on proximal risk-factors (e.g., momentary emotional states, environmental contexts, stressful events, etc.) is likely to lead to a better understanding of SITBs given their dynamic nature. The current study used ecological momentary assessment to examine proximal relations between negative emotions, coping behaviors, impulsivity, and real-time SITBs among youth (N = 60) at high-risk for suicide. Bayesian multi-level models revealed associations between negative emotions and suicidal thoughts. There was some evidence that specific emotions, as well as particular coping strategies, were weakly correlated with suicidal thoughts. Disengagement coping strategies strengthened the association between negative emotions and suicidal thoughts. There was no evidence that coping strategies or impulsivity attenuated the relation between negative emotions and thoughts of non-suicidal self-injury. Due to the large degree of heterogeneity present in the data, more research concentrated on understanding subject-level predictors of SITBs is needed to precisely predict when individuals are at elevated risk for suicide. This type of research could be crucial to improving behavioral treatments through the discovery of person-specific treatment targets.