Introduction

Suicide is a major public health problem worldwide and its prevention is a priority in many countries. It is the leading cause of unnatural death in most countries [1]. In the USA, for every completed suicide, there were about 10 people treated in emergency departments for self-inflicted injuries in 2013 and 10–20 suicide attempts are estimated for every completed suicide [2, 3]. The consequences of completed suicide or suicide attempts in terms of quality of life in suicide attempters, grieving in surviving loved ones, and economic consequences of suicidal behaviors (SBs) have been the focus of numerous studies.

History of attempted suicide is the most important predictive risk factor for completed suicide [4]. About 25 % of suicide attempters have exhibited SBs in the previous year [5]. Among subjects who died due to completed suicide, more than 90 % in the Western and 60 % in the Eastern world were diagnosed with a psychiatric disorder. However, there are a percentage of subjects who did not fit any of the psychiatric disorders recognized today. Oquendo and Baca reported that researchers and clinicians have been quite disappointed in their objective to decrease suicide rates using primary, secondary, and tertiary prevention strategies as normal predictors simply do not work well, especially in identifying short-term risk [6]. In spite of several discussions [6], suicidal behavior is still not recognized as a diagnostic entity in the DSM 5 [7]. Similarly to other psychiatric disorders, SB includes specific characteristics, in this case, violence of the method used, medical consequences, and degree of planning involved. Moreover, possible cultural differences and functional consequences of the suicidal act should also be taken into account. The Guidelines for Making Changes to DSM-5 [8] takes Robins and Guze’s concept of validators [9] and modified the classification from that proposed earlier by Kendler as antecedent validators, concurrent validators, and predictive validators. Furthermore, these guidelines indicate that, in order to validate SB as a disorder in the DSM-5, it must meet the three types of validators: antecedent validators (four categories), concurrent validators (two categories), and predictive validators (three categories).

The objectives of this review were to review “psychological factors” related to SB in order to unearth the best scientific evidence regarding one of the three categories of concurrent validators, i.e., cognitive, emotional, temperament, and personality correlates (unrelated to the diagnostic criteria).

Method

The authors conducted a search on PubMed with the following keywords: suicidal behavior [title], suicide attempt* [title], suicide [title] and either temperamental [title], emotion* [title], impulsiv* [title], aggress* [title], hostil* [title], hopeless* [title], cogni* [title], neurocogni* [title], personality [title], and empt* [title] within the last 5 years (from 2012 to 2016). A selection of studies best fitted to our objective was performed. The authors also focused on a set of studies focusing on the role of emptiness as a correlate of SB. Moreover, reference lists from original studies were also screened.

Results and Discussion

Cognitive Correlates

In the last few years, there has been considerable advances in understanding the neurocognitive basis of SB. Neurocognitive distortions may be seen as vulnerability factors (and therefore therapeutic goals) as well as potential endophenotypes of SB [10•]. The most important findings were described.

Cognitive inflexibility can be defined as the tendency to maintain an idea despite evidence proving the contrary, i.e., the tendency to put forward problem-solving solutions that were previously proved inefficient. Neuringer described patients with a history of suicidal behavior who showed extreme cognitive inflexibility [11, 12]. In fact, this feature was pointed out as a factor dramatically increasing the tendency to suicidal ideas and suicide attempts, because it has been associated with greater brooding rumination [13, 14]. Furthermore, under certain circumstances, suicidal people come to the conclusion that suicide is the only way out to escape their situation. Numerous studies, following those of Neuringer, support the relationship between cognitive inflexibility and SB and the different suicide etiopathogenic models that were later added [13, 15–18]. However, some studies also challenge this correlation [19, 20]. As a matter of fact, several studies did suggest that this cognitive inflexibility could be more of a state than a trait, since in patients who clinically recover this cognitive, inflexibility disappears [16, 21, 22].

Problem-solving or coping is the ability to identify and analyze elements in a situation of conflict and choose the most effective solution in a given context. Patients with poor problem-solving skills exhibit a greater tendency to suicidal ideation and behavior [23–25]. Adolescence is a vital period characterized by the appearance of many changes (anatomical, psychological, and social interaction alterations) that need to be faced. It has been proven that during adolescence, subjects with poorer problem-solving skills bear the highest risk of suicidal ideation [26], although depression and hopelessness should be also considered [27]. Similarly, old age is a crucial period of changes requiring significant adaptations, and in the elderly population, poor problem-solving skills were associated with suicidal attempts [28]. Several studies put forward the idea that problem-solving skill training could improve depression [27, 29] and SB [24], although results were inconclusive [30, 31•].

Rumination is the compulsively focused attention on the symptoms of one’s distress and the possible meanings, causes, and consequences in a repetitive and passive manner, instead of implementing an active problem-solving attitude. In the systematic review by Morrison and O’Connor’s, 10 out of the 11 reports evaluated did support the association between rumination and suicidality [32] with current suicidal ideation [33], suicidal ideation over time [34], and suicide attempts [33, 35]. Closely related to hopelessness and depression (well-known suicide factors), rumination was also correlated with other suicide-related neurocognition aspects [36, 37], i.e., cognitive inflexibility and poor problem-solving [13]. Similarly, the authors suggested the existence of an indirect path that went from cognitive inflexibility to brooding rumination, from rumination to hopelessness, and from hopelessness to suicidal ideation [13].

Cognitive control regulates the information processed and our behaviors in order to conform to environmental needs [38]. Thought suppression is the attempt to willingly avoid the onset of unwanted thoughts, especially when they are associated to unpleasant emotions. This strategy, apart from being ineffective at times, could also be paradoxical, promoting on the contrary the onset of those unwanted thoughts [39]. Apart from being associated to several mental disorders, thought suppression was linked to ideation and suicidal attempts [40–43].

Decision making is the process of choosing one option out of different available possibilities. It is related to suicidal attempts in adolescents [44, 45], adults [46], and elderly subjects [47] and has been linked to suicidal attempts in patients with unipolar and bipolar depression [48]. Furthermore, impaired decision-making skills were evidenced in most violent suicide attempts [46, 49]. Decision-making impairment was observed in euthymic patients who committed suicidal attempts vs. euthymic patients who did not [50]. This suggests a steady cognitive trait of suicidal vulnerability [48]. Furthermore, healthy (without previous suicidal attempts) first-degree relatives of suicide victims showed poorer results in decision-making tasks [51], suggesting the existence of a cognitive endophenotype which is the inherited trait associated with suicidal vulnerability. Subjects with suicidal behaviors present disadvantageous decision-making skills, which reflect alterations in value-based/motivational/reward-learning processes [52]. It has been proposed that suicidal behavior may rely on a combination of deficits affecting value-based processes (such as an increased risk-taking propensity) and deficits affecting cognitive control [10•]. Current concepts regarding the influence of decision-making impairment in suicidal behavior suggest that in the presence of psychological pain, related to negative life events particularly those threatening social integration, the person will select options associated with an immediate reward (pain relief), in spite of the risk (death) [53].

Subjects with previous suicide attempts present a different autobiographical memory compared to the controls: they lack the ability to retrieve specific autobiographical memory, and instead, they produce significant over-general autobiographical memories [54, 55]. It has been suggested that ineffective problem-solving in suicide attempters could be caused by the difficulty in recalling specific autobiographical memories: the latter seemingly work like a database offering possible solutions to cope with problems [56]. A more general autobiographical memory was reported in different suicidal samples: depressed and non-depressed subjects [57], women with borderline personality disorders [58], adolescents [59], older adults [60], and subjects with schizophrenia [61]. Interestingly, Taylor et al. reported in subjects with schizophrenia spectrum disorders that the ability to recall specific autobiographical memories was related to greater suicidality, and conversely, they suggested that non-specific memory may have adaptive qualities in these individuals inhibiting the recall of certain distressing memories [62].

Working memory deficits were reported in past suicide attempters with mood disorders [63, 64], showing differences between high- and low-lethality suicide attempts [20]. In psychotic disorders, the tendency was that subjects with previous SB performed better in working memory tasks [65]. Several studies observed that abnormalities in the working memory of subjects with previous SB disappeared after adjusting for motor impulsivity [66] or depressive symptoms in subjects with schizophrenia [67].

Reduced language fluency was reported in depressed suicide attempters [68, 69]. Later research did not report differences between attempters and non-attempters in depressive [70] or bipolar [71] samples. As observed in other cognitive domains, including language fluency tasks, suicide attempters with schizophrenia or schizoaffective disorder exhibited higher cognitive functioning than did non-attempters [65].

The evidence available still does not allow the elaboration of a cognitive suicidal model. The lack of consistent results can be explained by taking into account the variety of samples used, size of sample (sometimes too small), patients’ diagnosis, treatment, clinical state, time since their last suicide attempt, or characteristics of the attempt. Notwithstanding, there has been significant progress in our understanding of the neurocognitive factors predisposing patients to SB. Richard-Devantoy et al., in a recent meta-analysis, reviewed vulnerability neuropsychological markers for SB in patients with mood disorders, and they found that the most decisive alterations are those shown in Iowa Gambling task tests, categorical verbal fluency test, and Stroop test [10•].

Temperament, Emotional, and Personality Correlates

Currently, the stress-diathesis model is considered as the gold standard for understanding SB. In this model, the core diathesis factors are impulsiveness, aggression, and hopelessness. Thus, individuals characterized by high impulsiveness, aggression, or hopelessness were more likely to act on suicidal ideation when suffering from psychiatric illness (the stressor) [72]. But, this model, based on 347 inpatients aged 14 to 72 years old, mostly Caucasian subjects recruited in the USA, cannot be generalized to psychiatric patients worldwide [73]. Indeed, we recently stressed that most SB research has been conducted in areas of the world where the risk of suicide is at its lowest. And, this is probably hampering our capacity to build predictive and preventive models that can be extrapolated to a worldwide population [74]. For instance, in a study comparing depressed suicide attempters and controls from Madrid (Spain) and New York City (USA), the authors reported that New York citizens—either attempters or controls—appeared to be more aggressive than did their Spanish counterparts, whereas impulsiveness appeared to be a more stable personality trait across populations [75]. In this part of the present review, we will comment on recent findings regarding impulsiveness, aggressiveness, and hopelessness, as well as stressing the emergence of emptiness as a particularly relevant correlate of SB.

Emotions and temperament features (i.e., hopelessness, emptiness, hostility, impulsiveness, aggressiveness, anger) may indicate a risk for both mood disorders and SB, but their relationship remains unclear [76]. Indeed, impulsive aggressiveness and hopelessness are among risk factors for suicide attempts (for reviews, see [77] and [78]). All these features are diathesis factors that might increase the risk of SB by mediating the impact of Axis I disorders, particularly in patients diagnosed with major depressive disorder (MDD) [79]. For instance, Keilp et al. reported that aggressiveness was the distinguishing feature of suicide attempts, at least in patients with MDD [80]. Furthermore, in the classic study by Soloff and colleagues [81] comparing 81 inpatients with cluster B personality disorders (BPDs)—including 49 patients with BPD and MDD—to 77 inpatients with MDD, both hopelessness and impulsive aggression predicted a risk of suicide. In 2008, Brent and Melhem suggested that aggression could predict SB, and that impulsiveness and aggression might explain the familial transmission of SB [82]. Furthermore, other authors showed that BPDs were also linked to a family history of suicide attempts [83]. For instance, impulsive aggressiveness might explain the increased risk of repeated suicide attempts, including violent suicide attempts, in antisocial individuals [84, 85]. In this part of the review, we will consider three relevant concepts to SB risk: (1) hostility, impulsiveness, aggressiveness, anger, and rage; (2) emotional (affective) dysregulation; and (3) hopelessness and emptiness.

Regarding hostility, impulsiveness, and aggressiveness, all three features are powerfully associated [80, 86, 87] with an increased risk of SB. For instance, in a sample of violent adult men and women (n = 235; 193 agreed to participate) referred to court because of spousal abuse, those affected by BPD as measured by the Self-Harm Inventory were more likely to be suicide attempters [88]. In another study on 68 BPD outpatients, suicide attempters were more likely to display higher aggression scores than were non-attempters [89•]. BPD patients displaying suicide attempts were characterized by higher levels of aggression, particularly self-directed aggression, than were BPD patients without suicide attempts.

Affective or emotional dysregulation is a construct composed of avoidance of abandonment, inappropriate anger, and affective instability [90]. Some authors [89•] suggested that patients showing this type of affective dysregulation might be more likely to display self-aggression (i.e., suicide attempts) when facing negative emotions such as guilt, anger, or self-hatred [91]. For instance, in one study, BPD patients who attempted suicide were more likely to display impaired impulse control and showed higher scores of aggressiveness vs. BPD non-attempters [92, 93]. In another interesting study based on the NESARC survey, the authors reported that the main contributor of the relationship between BPD and suicide attempt could largely be explained by affective dysregulation [94].

Furthermore, recently, Ansell et al. [95] compared 431 individuals with PDs who completed a 10-year follow-up to a sample of individuals diagnosed with MDD. The authors replicated the results of Bolton and Robinson [96] who showed that BPD in itself explained the increased risk of suicide attempts over time. They suggested that their findings probably reflected some personality endophenotypes—i.e., emotional dysregulation, impulsive aggression—underlying multiple PDs [83, 84, 97]. For instance, they found that the main PD explaining multiple suicide attempts in their sample was Narcissistic PD. As they suggested, emotional dysregulation associated with failure to meet inflated expectations might result in interpersonal hostile or aggressive behaviors [98]. Furthermore, we also reported that frequent repeaters (individuals with ≥5 lifetime suicide attempts) exhibited higher levels of trait anger (OR = 2.82, 95 % CI = 1.18–6.75) and lower levels of Anger Expression-Out (OR = 0.17, 95%CI = 0.06–0.47) than did individuals with <5 lifetime suicide attempts [99].

Regarding hopelessness and emptiness, whereas the literature on the relationship between hopelessness and SB is quite extensive, there is surprisingly little empirical research on the relationship between emptiness and suicidal behavior [100]. Emptiness may be defined as a human condition related to a sense of boredom, apathy, and social alienation, although there are many ways classify it depending on the point of view (personality, existential state, emotion, or a symptom) [100]. Some authors recently suggested that self-conscious emotions might hide a range of personality pathologies and mediate the increased risk of SB [95]. For instance, shame proneness was related to self-harm and SB [101, 102]. High levels of hopelessness are a strong predictor of suicide attempt [103, 104] and were also associated with an increased risk for repetitive SB [81, 105, 106]. In a recent study, however, Stringer et al. [107], after adjusting for depression, did not find a statistically significant association between hopelessness and recurrent suicide attempts.

Finally, in a recent review and a series of methodologically different studies, we underlined the critical role of emptiness in suicide attempters, particularly for major repeaters (see above) [108•,109–112]. For instance, in one study aimed at identifying the most discriminative items for SB assessment, we found that emptiness was the most relevant personality factor to discriminate between suicide attempters, psychiatric patients, and healthy controls [111]. In a more recent study, we demonstrated that emptiness was a more relevant pathway than BPD is to prevent repetitive suicide attempts [108]. We recently speculated that emptiness might play an important role in the development of SB addiction [110, 112].

An important percentage of personality disorders were identified within both suicide attempters and completers [113]. Also, personality traits were studied and proposed as an intermediate phenotype in SB in order to facilitate the study of genetics [114•, 115]. Personality traits differ according to SB subpopulations, such as repeated attempters [116], older depressed attempters [117], or completers [118]. These distinctions may help refine the different needs of patients in prevention programs.

Most studies that examined the relationship between personality dimensions and SB were based on three psychobiological models of personality with the following instruments: Temperament and Character Inventory (TCI) [119], NEO Personality Inventory (NEO-PI) [120], and Eysenck Personality Questionnaire (EPQ) [121]. Several studies analyzed the correlations between the dimensions of these instruments [122–124].

Personality traits more commonly associated with SB include neuroticism and extraversion [117]. The first one, neuroticism, is defined as a stable personality trait that causes a predisposition to breakdown when stressed. The second one, extraversion, is a predisposition to greater social interactions with others. Clonninger pointed harm avoidance as the temperament trait conditioning the behavioral inhibition against relevant stimuli (conditioned signals for punishment, novelty, or frustrative non-reward), and novelty seeking would be the behavioral activation in response to novel stimuli (novelty, potential reward, or potential relief of monotony or punishment) [125].

TCI harm avoidance was related to the NEO-PI neuroticism domain [126] and EPQ neuroticism (and correlated negatively with EPQ extraversion) [122] while the TCI novelty-seeking was correlated to EPQ extraversion, and, to a lesser extent, to EPQ psychoticism [122] and NEO-PI extraversion domain [126]. In addition, between EPQ and NEO-PI, we also found correlations between both neuroticism domains [123].

Among EPQ studies, some fundamental studies were published by Indian researchers. Ponnudurai et al. [127] studied the personality in suicide attempts and reported that nearly half of the attempters obtained high score in extraversion, and almost all subjects had a low psychoticism score. Subsequent studies confirmed higher scores in extraversion and in neuroticism in suicide attempters [128, 129]. Another feature associated with suicide attempters was the correlation between neuroticism and impulsiveness, as some authors suggested that this interaction might trigger vulnerability in crisis situations [129]. In a study on a population of suicide attempters aged 75 and above, these patients scored higher on neuroticism and lower on extroversion compared to general population controls [130].

Among studies using the TCI, results commonly report that suicide attempters exhibit higher scores in harm avoidance (HA) and lower scores in self-directedness (SD) and cooperativeness (C), compared to controls [131–135], even in a young population [132]. Higher harm avoidance scores were also reported among suicide attempters compared to non-suicidal subjects with specific diagnoses, such as anorexia or bulimia [136, 137], bipolar disorders [138, 139], schizophrenia [140], substance abuse [141], and depression [142]. Other variables linked, albeit to a lesser extent, with suicide attempts were higher novelty seeking, self-transcendence [133, 138], and reward dependence [132, 139]. Also, in a further analysis of the dimensions of the personality model “Big Five” (involving neuroticism, conscientiousness, openness to experience, agreeableness, and extraversion) in more than 600 suicide attempts, it was observed that males with a history of attempted suicide were more prone to negative emotions while female attempters exhibited dimensions related to activity and self-regulation [128].

The NEO-PI has been used in SB in depressed subjects aged 50 and above (n = 110). Results indicated that those with a history of suicide attempts were less inclined to feel positive emotions and were more self-effacing [117].

Despite the above results, some authors found that there were some differences in personality traits that become less significant when depressive symptoms were taken into account [143]. This would go along the same trend underlining the influence of personality traits in the presence of psychiatric disorders [144]. More specifically, neuroticism was linked to anxiety disorders [145] and depression [146]. Personality traits may also contribute to the suicide risk mediating the presence of adverse events such as childhood abuse [147].

Few studies that assessed personality traits in suicides conducted psychological autopsies. There is an Australian study in suicide victims above the age of 35 with the NEO-PI where the authors found that suicide victims of both sexes (n = 259) vs. controls (n = 181) had higher mean scores in neuroticism and openness and lower scores on extraversion and agreeableness [148]. The authors stated that the control group may influence the results, since it were composed by sudden death subjects that may have “unhealthy lifestyles,” which may explain the higher scores in both neuroticism and openness compared to the control group. In another study that applied the Personality Assessment Schedule (PAS) [149] on completed suicide in individuals aged 60 and above, the authors identified personality “trait accentuation” in a third of the suicide victims (anankastic personality and/or anxious traits), both of these related to neuroticism [150].

However, in a follow-up study of 181 subjects with a history of suicide attempt, 11 individuals died by suicide. They exhibited higher scores in the extraversion dimension on the Karolinska Scales of Personality (KSP) [151]. Another way to study completed suicide is via ecological studies or cohort studies. Among the former, two studies examined suicide rates with personality traits in all the 50 states of the USA. In both studies, lower rates of neuroticism were associated with higher rates of suicide [152, 153]. In a Japanese study on nearly 30,000 people with a 19-year follow-up, it was noted that suicide rates and personality traits (EPQ) were differently correlated according to the time period before the economic crisis than during the economic crisis. Thus, while neuroticism in the period prior to the economic crisis did not significantly influence suicide rates, it did in fact influence suicide rates during the economic crisis [154]. These results of neuroticism seem to be contradictory to the non-ecological studies that may be, in part, due to the method that do not study subjects in particular but by population as a whole.

Differences Between Suicide Attempters and Suicide Completers

Some differences arise in the results obtained in attempters versus completers. These differences may be due to the existence of two different populations with differentiated intention to die and some shared overlapping risk factors [155, 156] whether they are two distinct populations (categories) or they form a spectrum (dimensional). Moreover, intentionality can be analyzed based on the results of the suicidal act (low-lethality; high-lethality; or violent, completed suicide), which may refine and characterize subpopulations of suicide attempters [157]. Those who completed suicide were more likely to be male and were typically older than those who survived a suicide attempt [158, 159]. In this sense, impulsiveness was reported to be higher in suicide attempters than in suicide completers [113] and higher in young than in older completers [160].

Mental Pain as Psychic Endophenotype of SB With Potential Biological Correlates

It has been proposed that SB is a way to escape from the unbearable suffering of mental or psychological pain [161, 162], identified as psychache by Shneidman [163]. There are different situations that favor the appearance of psychic pain, including the perception of social rejection, mental disorders, personality disorders, and cognitive factors [164–166]. The mental pain is also associated with suicidal ideation in non-clinical population [167, 168]. Data suggest that subjects with a history of suicide attempts have a decreased threshold to pain, regardless of depressive disorder [135]. Moreover, the intensity and frequency of suicidal ideation may correlate positively with psychological pain [169]. The importance of social factors of belongingness in SB has been previously remarked [170–173]. The social rejection has been the subject of numerous studies that state the similarities between physical and psychological pain in activation of brain areas [174] or implication of both opioid and cannabinoid systems [175]. These areas also seem to be related to the control of aggression [176]. Moreover, pain rejection favors greater sensitivity to rejection [177], which creates a vicious circle. However, mental pain may encourage social reconnection after the rejection, which would be a reparative function [164], similar to physical pain that helps to prevent the somatic injuries. According to this, there may be some population in which that reconnection is not effective, either due to the intensity of pain or other psychological factors (cognitive, temperamental, personality, mental diagnosis). Currently evidences suggest that the mental pain is related to a low-grade neuroinflammation, which is also related to SB (for a review see Courtet et al. [178•]). There are many studies that confirm the antisuicidal [179] and antidepressant [180] proprieties of ketamine, which may be related to both analgesic and anti-inflammatory actions [181]. For all these reasons, the psychological pain and neurobiological bases can be used for characterization and prevention as a common pathway in an important percentage of people with SB.

Conclusion

The presence of cognitive, temperament, personality, and emotional factors in relation to SB is common to multiple studies and may serve as a link between genetic and clinical research in suicidality. They also allow evidence-based profiling models that are essential to identify different subpopulations within SB patients. These populations range from subjects attempting suicide with little or low intentionality to those who die by suicide. This distinction between subpopulations is necessary in order to research new psychopharmacological or psychotherapeutic approaches and describe their effectiveness. The future of research, prevention, and treatment of SB should aim at integrating genetic, epigenetic, biological, cognitive, temperamental, personality, and clinical knowledge for a better understanding of all types of suicidal behaviors.