Background
Suicide is a major public health concern, ranking as the 10th leading cause of death in the United States, and accounts for over 42,773 deaths in the US each year [
1]. Suicide is a complex behavior reflecting a multi-level interplay of biological, psychological, social, and material factors such as access to means. Although the predictive value of long-term risk factors including biomarkers (e.g., serotonin levels), demographic factors (sex, age), and clinical factors (diagnosis of mental illness, prior suicide attempt) has been well established [
2,
3], a recent meta-analysis of 50 years of research on risk factors for suicidal thoughts and behaviors (STB) highlights the consistently small effect sizes of such risk factors and that no risk factor category or subcategory is substantially stronger than any other [
4]. Likewise, reliable predictors of acute suicide risk have also remained elusive. Recently, however, a small but growing body of evidence suggests the existence of an acute suicide crisis syndrome (SCS) [
5,
6], characterized by the experience of entrapment accompanied by intense negative affect, loss of cognitive control, hyper-arousal, and social withdrawal, which may precipitate the transition from chronic suicidal ideation (SI) to acute suicide attempt (SA) [
7‐
12]. Our previous work on a series of instruments aiming to define the SCS has shown several clinical factors, proposed as criterion symptoms of a clinical syndrome that characterizes the suicidal crisis, are indeed related to each other and short-term risk for suicidal behavior. These include the constructs of entrapment, ruminative flooding (a cognitive control symptom), panic-dissociation and fear of dying (hyper-arousal symptoms), and emotional pain (an intense negative affect symptom) [
10‐
13] and encompass the affective, cognitive, and somatic aspects of the hypothesized SCS. Notably, the Suicide Crisis Inventory (SCI) has demonstrated strong internal consistency, and significant predictive as well as incremental predictive validity over standard suicide predictors (SI, depression, state and trait anxiety) for short-term post-discharge suicidal behavior [
12].
The first construct, entrapment, has recently been proposed as a key psychological element of several models of suicidal behavior, notably the Arrested Flight/Cry of Pain model by Williams and Pollock [
14] and O’Connor’s Integrated Motivational-Volitional Theory (IMV) [
15]. According to these models, entrapment, defined as a felt urgency to escape from an unbearable situation from which there is no perceived escape [
16], could be a core psychological mechanism in causal pathways to suicide. In agreement with this hypothesis, a strong positive association was found between perceptions of entrapment and suicidality in a range of subacute outpatient populations [
17‐
23]. O’Connor’s group also demonstrated that entrapment added incremental predictive validity for suicidal behaviors over depression, hopelessness, SI, and the frequency of previous suicide attempts [
24]. Similar to their findings, in our group’s latest study of the SCS, entrapment was the strongest individual predictor of suicidal behavior within the 2 months following discharge from an inpatient unit [
12].
Surprisingly, although entrapment is a core concept of many models of suicide and of the hypothesized SCS, its role as a possible mediator of the relationship between other acute risk factors and SI/suicidal behavior has never been examined in
acutely suicidal patients. The IMV model [
15] posits that ruminations increase the likelihood of stressful experiences leading to perceptions of entrapment, which in turn may trigger SI. Alternatively, entrapment may lead to a cognitive search for escape routes resulting in rumination, thus triggering SI. Uncovering these relationships may not only help clarify theoretical models of suicidal behavior, but also establish important targets for therapeutic interventions in acutely suicidal individuals.
The second hypothesized component of the SCS, ruminative flooding, is characterized by uncontrollable perseverative thinking involving continual thoughts about the causes, meaning, and consequences of one’s negative mood. Ruminative flooding differs from simple ruminations in that it is experienced as an uncontrollable and overwhelming profusion of negative thoughts, often associated with somatic symptoms inside the head, such as headaches or head pressure [
10]. Similar constructs have been associated with elevated suicide risk by other research teams [
25,
26]. The third construct, panic-dissociation, describes somatic symptoms commonly associated with a panic-like dissociative state, and involves somatic experiences of unfamiliar sensations felt all over the body, and derealization [
11,
13]. Similar constructs have also been described by other suicide researchers [
27,
28]. The fourth construct, fear of dying, describes morbid cognitions during panic, which may mediate the transition from latent SI to active SI and suicide attempt in some depressed subjects [
29‐
34].
Finally, the fifth hypothesized component of the SCS is emotional pain, a mixture of intense and painfully felt negative emotions such as guilt, shame, hopelessness, disgrace, rage, and defeat, which arises when the essential needs to love, to have control, to protect one’s self image, to avoid shame, guilt, and humiliation, or to feel secure are frustrated [
35]. Emotional pain is similar to entrapment in that it is strongly correlated with but distinct from anxiety and depression, and can be so intense that the individual seeks to escape by – suicide [
35‐
39]. However, those in emotional pain may lack the desperation caused by the perception that all escape routes are blocked, whereas this desperation and subsequent escape motivation are intrinsic components of entrapment.
Severe or pervasive SI is consistently found to be a significant predictor of subsequent suicidal behaviors [
40]. Even passive ideation, such as a wish to die, has been identified as a risk factor for death by suicide [
41]. Although the relationship between SI, suicidal behavior, and suicide is not straightforward, and several other symptom dimensions may synergistically influence suicide risk [
20,
42], SI has a much higher base rate than suicides or suicide attempts and thus represents a scientifically useful and clinically meaningful modifiable element of suicide risk.
Given the evidence of the central role of entrapment in the suicidal process, the goal of the present report was to establish whether entrapment mediates the relationship between the other clinical factors of the SCS and SI in psychiatric inpatients recently hospitalized for suicide risk. As, the SCS is conceptualized as a clinical syndrome that activates SI, motivating the transition from SI to SA, in the context of the current study, severity of SI serves as a proxy for nearness to suicidal action. We therefore reanalyze the data from our published validation study of the SCI [
12] to examine the interrelationships among the SCI subscales and severity of suicidal ideation. We hypothesized that entrapment will mediate the relationships between SI and ruminative flooding, panic-dissociation, fear of dying, and emotional pain.
Discussion
The primary aim of the present study was to test the hypothesis that entrapment mediates the relationships between other constructs of the SCS (ruminative flooding, panic-dissociation, fear of dying, and emotional pain) and SI. To our knowledge, this is the first study to show that entrapment plays a central role in the relationship between the symptoms of the SCS and SI in acutely suicidal individuals. Further, ruminative flooding, panic-dissociation, and fear of dying did not significantly mediate the effect of entrapment on SI, indicating that entrapment fully mediated the relationships between each of these constructs and SI. However, entrapment was only a partial mediator of the relationship between emotional pain and SI, and emotional pain was a significant partial mediator of the effect of entrapment on SI. This suggests both common and independent effects of entrapment and emotional pain on SI.
These results are consistent with our previous work, which has documented the prominence of entrapment in several populations of acutely suicidal patients both in the psychiatric emergency department and among psychiatrically hospitalized patients [
11,
13]. Our data also support the central role of entrapment in the SCS, as assessed by the Suicide Crisis Inventory [
12].
Ruminations have been previously linked to SI and suicidal behavior [
25,
48,
49] although prior research has also pointed to an indirect rather than direct effect. Smith, Alloy [
50] demonstrated that hopelessness partially mediated the relationship between ruminations and SI, and fully mediated the association between ruminations and duration of suicidality. Notably, Teismann and Forkmann [
51] reported no direct relationship between ruminations and SI in two outpatient populations, as the link between the two was fully mediated by entrapment. Our results also show that the relationship between the ruminative flooding subscale of the SCI and SI is fully mediated by entrapment. This finding complements that by Teismann and Forkmann [
51], and suggests that as ruminations during a suicidal crisis become uncontrollable and overwhelming, the perception of being trapped may precipitate SI.
Similarly, although numerous studies have shown that the experience of somatic symptoms, [
11,
13,
27,
28,
52] and fear of dying [
29‐
34], particularly in the context of anxiety and panic, predict SI or suicidal behavior, our current study indicates that in the acute suicidal state there are no direct linkages between either somatic symptoms or fear of dying and SI. Our results indicate instead that the somatic symptoms of panic, dissociation, and fear of dying are indirectly linked with SI via a pathway of entrapment. Thus, as with ruminative flooding, experiencing a fear of dying and dissociative symptoms may increase or simply be correlated with the intensity of stressful experiences that result in a sense of entrapment, which in turn might trigger SI.
Our results show that the relationship between emotional pain and SI was only partially mediated by entrapment, indicating a direct link between emotional pain and SI. Furthermore, the finding that the association between entrapment and SI was partially mediated by emotional pain suggests that these two constructs have both a direct and mutually mediatory relationship to SI. It may be that both of these constructs substantially entail the other. Namely, emotional pain may reliably produce the escape motivation that is necessary for entrapment, and the perception of entrapment may necessarily induce emotional pain. This understanding of the findings is consistent with Baumeister’s Escape Theory of Suicide [
53] which conceptualizes suicide as an escape from otherwise seemingly inescapable painful self-states. Likewise, Shneidman [
35] defines emotional pain or mental pain as a compilation of negative emotions (e.g., anxiety, helplessness, and despair) that reflects frustration with one’s inability to fulfill basic psychological needs. In Orbach and Mikulincer’s formulation, emotional pain includes loss of self-esteem, as well as feelings of failure, abandonment, emotional flooding, and emptiness and, notably, irreversibility [
36‐
39]. Thus, emotional pain as assessed in studies using the Orbach Mikulincer Mental Pain scale, appears to be substantially convergent with the entrapment subscale of the SCI. The emotional pain subscale of the SCI, on the other hand, comprises only items directly descriptive of an emotional experience described in terms of pain. Past research has indicated that emotional pain mediates the relationship between perfectionism and suicidality [
54], between hopelessness and SI [
55], and between non-suicidal self-injury and suicide risk [
56]. Shneidman [
35], and later Orbach [
36] further suggested that unbearable psychological pain may mediate the relationships between other relevant psychological factors and suicide. Similarly, in Klonsky & May’s recent Three Step Theory of suicide, the combination of pain and hopelessness drives the development of suicidal ideation; our findings for entrapment, which may be understood as a product of emotional pain and hopelessness, are thus also broadly consistent with the three-step theory [
57]. Orbach’s studies and the strong correlation we found between entrapment and emotional pain suggest that entrapment is an emotionally painful experience. At the same time, however, the aversive experience of emotional pain may drive an urgency for escape which itself is requisite for the experience of entrapment. Our finding in supplementary analyses that controlling for severity of depression in mediation models of entrapment and emotional pain reduced mediation effects to non-significance points to the variance common to all three constructs. Nonetheless, entrapment and emotional pain each show incremental prediction of SI with respect to depression severity, highlighting the pertinence of each of these constructs.
Thus, taken altogether, our data indicate that in the SCS, entrapment acts as a conduit between most of the associated symptoms and SI. This finding is highly consistent with recent theoretical models of suicide including the Cry of Pain Model [
14,
58] and the IMV Model of suicide [
15], as well as Baumeister’s Escape Theory of Suicide [
53] and Gilbert’s phenomenon of ‘arrested flight’ [
16] and Galynker’s Narrative Crisis model [
6]. Williams and Pollock [
14] posit that suicide results from a feeling of defeat in response to humiliation or rejection, which in turn leads to a perception of entrapment. When the latter is combined with a failure to find alternative ways to solve a problem, i.e. cognitive rigidity [
59], a suicidal person may see no exit out of his or her perceived entrapment other than suicide. One interpretation of the mediating role of entrapment is that perceptions of entrapment may constitute a final common pathway to SI. Therefore, therapeutic interventions aimed at reducing the perception of entrapment and creating actionable alternatives may eventually reduce suicide risk. In addition, clinical evaluation of entrapment, as illustrated in our previous reports [
11,
13] by explicitly asking whether a patient felt “trapped,” or that “there was no way out,” or that he or she “had no good options,” would be essential for a suicide risk assessment. Further, our data suggest that in addition to explicitly asking whether patients feel trapped in their current life situation, relationship, or within themselves, it is important to assess whether they have experienced fear of dying, felt unusual sensations in their bodies or pressure inside their heads, and lost the ability to control and stop their ruminative thoughts, as such experiences were robustly correlated with entrapment and may provide a path of approach to assessment of suicidal thoughts that the patient might not otherwise disclose.
Based on our current results, and our findings from the complementary study of prediction of suicidal behavior [
12], it can be hypothesized that while either unbearable emotional pain or entrapment may mediate the relationship between other symptoms of the SCS and SI, it is entrapment that may be the final mediator of the relationship of multiple psychological factors and suicidal action. A mediation analysis of prospective suicidal behavior using a larger sample size may provide support for this testable hypothesis. As the SCS is proposed to be a syndrome, the strong inter-correlations among the SCI subscales are certainly supportive of the SCS concept. Nonetheless, further study is needed to determine whether such a syndrome is more predictive of STB than simple severity of entrapment.
The results of this study need to be considered in the context of its limitations. This study represents a novel analysis of our previously published scale-validation data and thus all conclusions are tempered by the need for replication. Further, the study examines the relationship between the SCS constructs and SI rather than suicidal behavior or actual suicide, making the results less relevant to the urgent clinical task of suicide prevention. However, both active and passive SI have been shown to predict death by suicide marking it as a significant modifiable element of suicide risk [
40,
41]; thus, understanding the mediation effects of the relationships of psychiatric symptoms in acutely suicidal individuals with SI can improve our understanding of the suicidal process in general. Second, the cross-sectional design of the study limits any causal interpretation of our findings. Third, the current study utilized only a self-report measure of SI, and the data may be confounded if some participants were not forthcoming in revealing their suicidality. Finally, the study was conducted at a single site, and study findings need to be replicated at other locations.