Research report Longitudinal risk profiling for suicidal thoughts and behaviours in a community cohort using decision trees Philip J. Batterham a,n , Helen Christensen b a Centre for Mental Health Research, The Australian National University, Canberra, ACT 0200, Australia b Black Dog Institute, The University of New South Wales, Sydney, Australia article info Article history: Received 24 February 2012 Received in revised form 14 May 2012 Accepted 14 May 2012 Available online 26 July 2012 Keywords: Suicide ideation Suicide attempts Risk factors Decision tree models Suicide risk assessment abstract Background: While associations between specific risk factors and subsequent suicidal thoughts or behaviours have been widely examined, there is limited understanding of the interplay between risk factors in the development of suicide risk. This study used a decision tree approach to develop individual models of suicide risk and identify the risk factors for suicidality that are important for different subpopulations. Methods: In a population cohort of 6656 Australian adults, the study examined whether measures of mental health, physical health, personality, substance use, social support, social stressors and back- ground characteristics were associated with suicidal ideation and suicidal behaviours after four-year follow-up. Results: Previous suicidality, anxiety symptoms, depression symptoms, neuroticism and rumination were the strongest predictors of suicidal ideation and behaviour after four years. However, divergent factors were predictive of suicidal thoughts and behaviours across the spectrum of mental health. In particular, substance use was only associated with suicidal thoughts and behaviours in those with moderate levels of anxiety or depression. Limitations: Most of the measurements were based on self-report. Further research is required to assess whether changes in risk factors lead to changes in suicidality. Conclusions: Examining suicide risk factors using decision trees is a promising approach for developing individualised assessments of suicide risk and tailored intervention programs. & 2012 Elsevier B.V. All rights reserved. 1. Introduction Suicide risk has been widely researched by examining the association of specific risk factors with subsequent outcomes of ideation, plans, attempts and completed suicide. For example, extensive research has examined the value of major depression as a predictor of suicidality, which encompasses both suicidal thoughts and suicidal behaviours (e.g., Moller, 2006; Oquendo et al., 2006; Rihmer, 2001). Although depression is strongly predictive of suicidality, the positive predictive value of depres- sion as a risk factor for suicide is low, as only a small proportion people with depression develop suicidal behaviours. The same is true of anxiety, rumination, hopelessness and even suicidal ideation—each of these factors is strongly associated yet not sufficient for the development of suicidal behaviours. Rather than investigate the roles of individual factors, an alternative approach involves examining the interplay between risk factors in the development of suicide risk. Such an approach may lead to the development of individualized models to predict suicide risk and provide clearer definition of the types of risk factors for suicidality that might be important for different groups of people. However, it may be necessary to apply new methodologies to this field of research in order to create such models. In the domain of cardiovascular disease, there has been consider- able research aimed at predicting cardiovascular disease based on a constellation of risk factors. To establish individual risk profiles for cardiovascular disease, epidemiological research has been coupled with evidence from intervention studies which demonstrate that reducing factors such as smoking, blood pressure and lipids will reduce the risk of disease and stroke. Using a decision tree approach, risk assessment charts have been developed, together with guidelines to enable clinicians to predict risk for their patients (Jackson, 2000). Suicide risk may not be as readily identifiable as cardiovascular risk, and there has been limited research on whether interventions that address modifiable risk factors for suicide reduce its incidence. Nevertheless, the use of novel approaches such as decision trees may inform clinical decision making in the area of suicide prevention and identify at-risk groups that have previously been overlooked. Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/jad Journal of Affective Disorders 0165-0327/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jad.2012.05.021 n Corresponding author. Tel.: þ61 2 61251031; fax: þ61 2 61250733. E-mail address: philip.batterham@anu.edu.au (P.J. Batterham). Journal of Affective Disorders 142 (2012) 306–314