Introduction
The link between a diagnosis of posttraumatic stress disorder (PTSD) and suicidal ideation is well-established in the literature (Jakupcak et al.
2010; Panagioti et al.
2009,
2012; Richardson et al.
2012; Tarrier and Gregg
2004). Suicidal ideation in PTSD is associated with increased rates of subjective distress and disproportionally heightened rates of healthcare utilization (Chan et al.
2009; Bell and Nye
2007; Schnurr et al.
2000). Suicidal ideation is also a strong predictor of subsequent suicide attempts (Miranda et al.
2014; Simon et al.
2016). Thus, the identification of the best predictors of suicidal ideation in PTSD, which could be targeted by psychological suicide-prevention interventions, is of critical importance.
Based on previous developments, it has been emphasized that suicidal ideation in PTSD is potentially driven by both trans-diagnostic and PTSD specific features (Bolton et al.
2007; Johnson et al.
2008). To this end, important progress has been made towards understanding the role of generic (trans-diagnostic) factors that drive suicidal ideation in PTSD (Blake et al.
1995; Panagioti et al.
2012). In contrast, the role of PTSD-specific factors, such as distinct PTSD symptom clusters, in the development of suicidal ideation is less clear. Demonstrating that specific PTSD symptoms are strongly associated with suicidal ideation compared to others has the potential to facilitate the prompt identification of those individuals with PTSD who are at the highest risk for attempting suicide. So far, the studies that investigated the effects of specific PTSD symptoms on suicidal ideation have yielded inconsistent outcomes. Among veterans, two studies demonstrated that only the re-experiencing symptom cluster was significantly associated with suicidal ideation (Bell and Nye
2007) and acquired capacity for suicide (Bryan and Anestis
2011), whereas another study (Guerra et al.
2011) showed that suicidal ideation was uniquely positively associated with heightened levels of numbing symptoms. In addition, evidence was obtained that depression accounts for the impact of PTSD symptom clusters on suicidal ideation in veterans (Hellmuth et al.
2012). Among community research samples, Ben-Ya’acov and Amir (
2004) showed that suicidal risk was predicted by high levels of hyperarousal symptoms and low levels of avoidance/numbing symptoms, whereas another study showed that hyperarousal and avoidance/numbing symptom clusters were associated with suicidal ideation in child Earthquake survivors even after controlling for depression symptoms (Chan et al.
2016). However, there are two key methodological limitations within the extant literature regarding the association between particular PTSD symptoms and suicidal ideation and behaviors, which could explain the above discrepancies. First, these studies were based on different populations and the direct comparison of their results is difficult (some were based on veterans, others on clinical populations, or community-based populations). Second, all the previous studies employed cross-sectional designs to investigate the link between specific PTSD symptom clusters and suicidal ideation. The above limitations suggest that there is a need for further studies using a prospective methodology to investigate the effects of specific PTSD symptom clusters on suicidal ideation.
Furthermore, a related unresolved issue concerns whether or not PTSD-specific factors contribute to suicidal ideation over and above the effects of generic and trans-diagnostic psychological factors. A newly developed psychological perspective of suicide, the Schematic Appraisals Model of Suicide (SAMS; Johnson et al.
2008), provides a useful theoretical framework, which could explain the psychological mechanisms of suicidal ideation in PTSD. The SAMS model is a re-conceptualization of a highly influential model of suicide, the Cry of Pain Model of Suicide (Williams
1997; Williams et al.
2005). The advance of the SAMS model over its predecessor is that it has highlighted the importance of a personal cognitive appraisal system (e.g. an internal network that incorporates the individual’s subjective beliefs, judgments and attitudes towards the self, the world and the others) in the development of suicidal thoughts and behaviors (Panagioti et al.
2012). A category of negative appraisals, central to the SAMS, refers to defeat and entrapment, which are viewed as the proximal mechanism to suicidal ideation (Johnson et al.
2008). Consistent with this view, a number of recent cross-sectional studies has demonstrated that defeat/entrapment is strongly associated with suicidal ideation in a range of research populations including those with PTSD (Panagioti et al.
2012; Taylor et al.
2010,
2011). In addition to defeat/entrapment, another well-established generic predictor of suicidal ideation in PTSD is comorbid depressive symptoms (Bryan and Corso
2011; Clover et al.
2004; Menard
1995; Panagioti et al.
2009).
The aim of this study was to examine whether the three PTSD symptom clusters predict suicidal ideation prospectively (13–15 months later). To ensure that the effects of the specific PTSD symptom clusters on suicidal ideation are not an artifact of the effects of other key generic predictors of suicidal ideation, we accounted for baseline suicidal ideation, perceptions of defeat and entrapment and severity of depressive symptoms in the analyses. The SAMS argues that perceptions of defeat and entrapment should be best represented as a single construct, because they are conceptually synonymous and originate from the same cognitive mechanism (a negatively biased appraisal system; Johnson et al.
2008; Taylor et al.
2011). This suggestion has also been confirmed by a study that employed an explanatory factor analysis in a student population (Taylor et al.
2009). Therefore, consistent with our previous work (Panagioti et al.
2012; Schnurr et al.
2000; Taylor et al.
2010a,
b), defeat and entrapment in this study was conceptualized as a single construct (referred to as defeat/entrapment hereafter).
Discussion
The aim of this study was to examine the effects of distinct PTSD symptom clusters on suicidal ideation using a prospective design. The outcomes of the analyses showed that the hyperarousal symptom cluster was the only PTSD symptom cluster that predicted suicidal ideation 13–15 months later. This study advances the extant literature regarding the link between the distinct PTSD symptom clusters and suicidal ideation in two ways. The first key advance of this study is that it tested prospectively the effects of the three specific PTSD symptom clusters on suicidal ideation. Hyperarousal was the only PTSD symptom cluster that predicted suicidal ideation 13 to 15 months later. Ben’Yacov and Amir (
2004) also showed that only the hyperarousal, but not the other two symptom clusters, predicted suicide risk in a community male sample. This finding suggests that hyperarousal symptom cluster is the most critical PTSD symptom cluster in relation to suicidal ideation and emphasizes the need for continuously monitoring and therapeutically targeting hyperarousal symptoms to prevent suicidal ideation in people with PTSD symptoms.
The second key advance of this study was that it demonstrated that the hyperarousal symptom cluster prospectively predicted suicidal ideation even after accounting for the effects of three key trans-diagnostic predictors of suicidal ideation in PTSD including perceptions of defeat/entrapment, severity of comorbid depressive symptoms and past suicidal ideation. Although the findings of this study are in agreement with the existing theoretical (SAMS and Cry of Pain model) and research evidence, which has shown that perceptions of defeat/entrapment and comorbid depression are core psychological processes implicated in suicidal ideation (as indicated by the significant regression effects; Johnson et al.
2008; Panagioti et al.
2014; Williams et al.
2005), it further suggests that the contribution of the severity of hyperarousal symptom cluster on suicidal ideation is equally important to that of the above generic factors and therefore there are also PTSD-specific factors that are implicated in the process of suicide thinking. It should be noted, however, that hyperarousal symptoms such as physiological arousal, anxiety, insomnia and poor concentration are considered core elements of the PTSD experience, but they could also be seen as trans-diagnostic features because they are also present in other mood and anxiety disorders. Theories such as the diathesis-stress model and the positive feedback model for suicide emphasize the role of hyperarousal in the development of suicidal ideation and have received empirical support in other research populations including people with substance use disorders and depression (Capron et al.
2016; Katz et al.
2011). These models could be useful frameworks for understanding and preventing suicidal ideation in PTSD. Another possibility is that other unmeasured factors such as perceptions of guilt, self-deprecation and self-consciousness, which are identity-based cognitions, best account for the experience of suicidal ideation in people with PTSD symptoms compared to perceptions of defeat/entrapment, which are mainly situational cognitions. Consistent with this hypothesis, a number of cross-sectional studies found that feelings of guilt, shame and self-deprecation mediated the effects of PTSD symptoms on suicidal ideation (Bryan et al.
2013; Carr et al.
2013). Prospective studies are needed to examine these potentially important mechanisms of suicidal ideation in PTSD. In particular, the investigation of the unique effects of several theory-driven trans-diagnostic factors and PTSD-specific factors has the potential to significantly advance the understanding of the main drivers of suicidal ideation in PTSD. This approach will allow us to draw more definitive conclusions of whether suicidal ideation in PTSD is primarily driven by trans-diagnostic, disorder-specific or comorbid factors or by an amalgam of these three factors (Bolton et al.
2007; Guerra et al.
2011; Jurisic and Marusic
2009; O’Connor et al.
2013; Schnurr et al.
2000; Williams
1997). With regards to clinical practice, these findings suggest that to facilitate the accurate identification and therapeutic management of suicide risk in PTSD, a multi-target strategy, which would focus simultaneously on the severity of perceptions of defeat/entrapment, the severity of the hyperarousal symptoms, and the severity depressive symptoms, is likely to be the most productive.
A number of limitations of this study warrant discussion. First, this study only accounted for the effects of baseline suicidal ideation and depressive symptoms in the analyses. Whether these findings persist over and above the effects of other important risk factors for suicidal ideation (i.e., hopelessness, no rescue, perfectionism, perceived burdensomeness and thwarted belongingness; Joiner and Van Orden
2008; O’Connor et al.
2010; Williams et al.
2005) need to be addressed by subsequent studies. Second, the sample of the study was relatively small. However, the analyses clearly demonstrated that the hyperarousal symptom cluster significantly predicted suicidal ideation 13 or 15 months later. Third, this study only examined the effects of distinct PTSD symptom clusters on suicidal ideation, but not on other suicidal behaviors, such as suicide attempts. More research is needed to examine whether the same pattern of findings is evident among PTSD individuals with a history of suicide attempts. Although suicidal ideation and suicide attempts often co-exist, contemporary theoretical and empirical evidence suggests that different psychological mechanisms might be implicated in the development of suicidal ideation and suicide attempts (Wyder and De Leo
2007). According to ‘the ideation to action’ framework, pain and hopelessness are the main factors that lead suicidal ideation, whereas acquired capacity for suicide is the strongest predictor of suicide attempts beyond the effects of suicidal ideation (Klonsky and May
2015). Fourth, a single 5-point scale, the second item of the SBQR-II measure, which has sound psychometric properties (Taylor et al.
2010), was used to measure suicidal ideation in this study. We chose this measure because it is brief, easy to complete, and fitted well with the time-line of the study. Although previous investigations have also utilized one-item measures to assess suicidal ideation (Casey et al.
2008; Fialko et al.
2006), the future use of more extensive measures of suicidal ideation, would be beneficial. Fifth, in this study, we included participants with a current diagnosis of PTSD and also participants who were diagnosed with PTSD in the past and were experiencing subthreshold PTSD symptoms at present. This decision was based on evidence that the experience of subthreshold symptoms of PTSD is associated with significant functional impairment and comparable rates of suicidal behaviors with a full current diagnosis of PTSD (McLaughlin et al.
2015; Zlotnick et al.
2002). It has also been suggested that restricting the research and clinical focus to people with a full PTSD diagnosis is likely to hamper treatment access for people with subthreshold but clinically significant PTSD symptoms (Greenberg et al.
2015; Grubaugh et al.
2005). As a sensitivity analysis, we repeated the analyses in the subsample of the 39 people who had a current diagnosis of PTSD at baseline and completed the follow-up assessment measures. We noted no changes in the results. However, future larger studies on people with a current diagnosis of PTSD would be useful to replicate these findings. Finally, the DSM-IV criteria were used in assigning a PTSD diagnosis and in classifying the PTSD symptoms into three distinct clusters. In DSM-5, a number of major revisions were introduced. The most relevant revision for the results of this study is the division of the PTSD symptoms into 4 clusters in DSM 5 as opposed to 3 clusters in DSM-IV and the addition of three new symptoms. The re-experiencing cluster remained the same and only one change was introduced to the hyperarousal cluster (one symptom assessing the presence of “reckless or destructive behavior” was added), which was found to be the most important predictor of suicidal ideation in this study. Given the minor changes on the arousal cluster and the commonalities of the newly introduced symptom of recklessness and destructive behavior with the concepts of impulsivity and self-harm (which are highly correlated with suicidal ideation), we assume that the effects of the arousal and the re-experiencing clusters on suicidal ideation would be the same if the DSM-5 criteria had been used. However, the avoidance/numbing symptom cluster was divided into two separate clusters, the avoidance cluster and negative alterations in cognition and mood cluster, and two new symptoms were introduced to the latter cluster. These extensive changes do not allow any assumptions about the effects of these two revised symptom clusters on suicidal ideation. We certainly recommend the replication of these findings by future studies using the DSM-5 framework.
Despite the limitations above, this is the first study that demonstrated, using a prospective design, that the hyperarousal is the only PTSD symptom cluster which prospectively predicts suicidal ideation. Overall, suicidal ideation in PTSD appears to be a multifaceted phenomenon which is likely to be driven by an amalgam of trans-diagnostic factors, including defeat/entrapment and comorbid factors, such as severity of depressive symptoms and the hyperarousal symptom cluster. Future larger studies are encouraged to confirm and expand these findings.