Controlled Evaluation of a Computerized Anger-reduction Treatment for Suicide Prevention

Problematic anger is often found among returning military veterans, especially those exposed to combat. Anger control problems are associated with difficulties with reintegration. One study of Iraq-Afghanistan combat veterans who used VA medical services (N = 754) found that difficulty controlling anger was the most commonly reported problem experienced since homecoming, occurring among 57% of the sample (Sayer et al., 2010). Problematic anger and hostility have also been linked to risk for suicide (Daniel et al., 2009; Lehnert, Overholser, \& Spirito, 1994; Maiuro, O'Sullivan, Michael, \& Vitaliano, 1989). From the perspective of the interpersonal-psychological theory of suicide (Joiner, 2005), individuals with problematic anger may be at increased risk for suicide because they engage in impulsive behaviors (e.g., aggression, substance abuse) intended to manage anger. Such behaviors may lead to exposure to painful and provocative events, which increase their acquired capability for engaging in suicidal behavior. Additionally, anger problems often lead to social isolation, which may contribute to greater perceived burdensomeness and a failed sense of belonging, established risk factors for suicide (Van Orden et al., 2010).

Anger prone individuals are likely to hold the hostile attribution bias, which reflects the tendency to interpret the ambiguous actions of others as reflecting hostile intent (Wilkowski \& Robinson, 2008). Recently, researchers have developed computerized interpretation bias modification (IBM) protocols intended to efficiently reduce interpretive biases. These treatments have shown efficacy in reducing anxiety and depression (Beard \& Amir, 2008; Holmes, Lang, \& Shah, 2009; Mathews et al., 2007). We have conducted studies using similar procedures that focus on hostile interpretation biases and found our program to successfully reduce anger and hostility.

To evaluate the efficacy of the IBM protocol we have developed in reducing anger and suicide risk, we will conduct a three-arm randomized controlled trial over the internet. Military veterans and non-veteran community participants reporting elevated trait anger (N = 120) will be randomized to one of three conditions: 1) IBM; 2) progressive muscle relaxation (PMR); or 3) healthy education videos (control). Each condition will consist of eight 15-minute treatment sessions. Participants will complete two sessions per week for four weeks and will be administered assessments at pre- and post-treatment. IBM and PMR conditions will also complete 3-month and 6-month follow-up assessments. To ensure an adequate test of the effects of this intervention on suicide risk, we will oversample for individuals with current suicidal ideation. We hypothesize that: 1) IBM will lead to greater reductions in anger than PMR and control conditions in a sample with problematic anger; 2) IBM will lead to greater reductions in suicidal ideation, perceived burdensomeness, and thwarted belongingness than PMR and control conditions; 3) efficacy of the IBM condition in reducing suicidal ideation will be mediated by reductions in perceived burdensomeness and thwarted belongingness; 4) group differences between IBM and PMR will be maintained at the 3-month and 6-month follow-up assessments.