Predictors of suicide-related behaviors during treatment following a first episode of psychosis: The contribution of baseline, past, and recent factors

Abstract

Background

Suicide-related behaviors (suicide attempts and suicides) are common in the early phase of psychotic disorders. Studies have examined risk factors among baseline and historical (i.e., past) variables, yet little is known about recent characteristics that increase suicide risk during treatment for first-episode psychosis (FEP). This study had two aims: first, to determine the relative importance of baseline, past, and recent variables to the prediction of suicide-related behaviors in patients with FEP; second, to identify recent characteristics that exert most influence on suicide risk levels and which could become foci of preventive interventions.

Methods

This was a case–control study of 180 patients from a cohort entering a specialist FEP service between 1/12/2002 and 30/11/2005. Data for 72 cases and 108 matched controls were obtained via medical record audit. Multivariate logistic regression models assessed the contribution of baseline, past, and recent domains. Suicide attempt or suicide during treatment was the outcome variable.

Results

The strongest risk factors for suicide-related behaviors were: baseline depressive symptoms, baseline suicidal ideation/intent, past negative events, past non-suicidal self-injurious behavior, recent negative events, recent depressive symptoms, and recent non-suicidal self-injurious behavior. However, when these were entered into a hierarchical logistic regression model, only recent non-suicidal self-injurious behavior (AOR = 72.96, p < 0.001), and recent negative events (AOR = 1.90, p = 0.003) remained significant predictors. The final model accurately classified 75.5% of cases and 89.2% of controls, and explained 72.0% of variance in the suicide attempt status.

Conclusions

Since recent negative events and recent non-suicidal self-injurious behavior were the strongest predictors of suicide-related behaviors during treatment for FEP, psychiatric services could consider incorporating psychosocial interventions addressing affect regulation, interpersonal effectiveness, stress management and problem solving, alongside case management and pharmacotherapy, to help to reduce the rates of suicide attempts and suicides in first-episode patients.

Introduction

Suicide-related behaviors (suicide attempts and suicides) are common among individuals who receive psychiatric treatment for first-episode psychosis (FEP), and pose a challenge to mental health services due to associated psychosocial and economic sequelae. Approximately 12.0% of patients attempt suicide within 24 months of treatment for FEP (Verdoux et al., 2001, Robinson et al., 2009, Fedyszyn et al., 2011), with over a quarter making multiple attempts (Fedyszyn et al., 2011). Up to 1.5% of first-episode patients die from suicide within 18 months of initial contact with psychiatric services (Verdoux et al., 2001, Nordentoft et al., 2002, Tarrier et al., 2006, Robinson et al., 2009).

Studies have examined a range of variables with a view to assisting in the prediction, and hence prevention, of suicide-related acts during treatment for FEP. The most consistently supported risk factors include: past suicidal ideation and suicide attempts; poorer premorbid functioning; past substance abuse/dependence; past history of depression; and depression, hopelessness, and low levels of life satisfaction at entry to treatment (Verdoux et al., 2001, Nordentoft et al., 2002, Bertelsen et al., 2007, Robinson et al., 2009, Bakst et al., 2010, Melle et al., 2010, Robinson et al., 2010). However, the findings have not yet translated into clinical practice and preventive interventions because these variables have low sensitivity and specificity; that is, they correlate well with suicide-related behaviors, but are poor predictors of short-term risk (Cassells et al., 2005).

Predictive properties of risk factors for suicide-related behaviors in first-episode patients could be enhanced by distinguishing between past and recent factors. This distinction, also known as the distal–proximal dimension of risk factors (Mościcki, 1995, Mościcki, 2001), has been used to study a variety of health events (e.g., Roy, 2003, Roy, 2005, Foley et al., 2006), supporting the validity of clarifying risk factors on the basis of their temporal relationship with the outcome. Specifically, past factors represent an underlying vulnerability to suicide-related behaviors conferred by historical and background characteristics which may be person-related and/or environmental (Mościcki, 1995, Mościcki, 2001). Exposure to past factors is necessary, albeit insufficient in and of itself; the timing of suicide-related behaviors is determined by recent risk factors which also often act as precipitants (Mościcki, 1995, Mościcki, 2001). Thus, without the influence of recent factors, the risk potential conveyed through past variables may never actualize into an event. Likewise, recent risk factors will not result in a suicide-related act unless they augment existing vulnerability conferred by past influences (Mościcki, 1995, Mościcki, 2001).

Previous studies of risk factors for suicide-related behaviors in first-episode patients have mostly examined past and baseline (i.e., assessed at intake) variables. However, currently, little is known about recent risk factors as well as the relative importance of baseline, past, and recent variables to the prediction of suicide-related acts. Furthermore, past variables by definition are not amenable to intervention, whereas recent factors are potentially modifiable. Thus, little progress can be made in the development of effective preventive strategies unless recent variables are better understood.

Accordingly, the present study sought to determine the predictive power of baseline, past, and recent variables. It was hypothesized that recent factors would be stronger predictors of suicide-related acts during initial treatment for FEP, compared to baseline and past variables. The second aim of the study was to identify the most influential recent factors which could become targets of preventive interventions.