87 Psychopathology and risk factors for suicide in the young: Theoretical and empirical

In the first section risk factors along the categories of pathology, emotional states, personality traits, cognitive deficits, deficits in self-regulation, stressors and facilitators and inhibitors of suicidal behaviour are classified. The second section consists of a presentation of several theoretical perspectives of self-destructive behaviour in adolescents. The third section provides a review of theoretical and empirical models that try to track the interactions between the various risk factors leading to suicidal behaviour. The last section focuses on different pathways, based on different major dynamics, showing that there is more than one way to approach suicidal behaviour in adolescents.

Many risk factors have been implicated in suicidal behaviour. In order to gain some clarity in this web of multiple risk factors, a meaningful organization of the existing data is needed. In this chapter we will rely on Orbach's (1997) taxonomy of risk factors in order to facilitate in the organization of data related to suicidal behaviour in adolescents.

Research on suicidal adolescents suggests that psychopathology is very common within this population. Post-mortem studies show that more than 90 per cent of the adolescents who commit suicide have at least one major psychiatric disorder: this is especially prominent among older adolescents who commit suicide (Gould et al. 2003). Furthermore, a leading cause for hospitalization among adolescents is suicidal behaviour. Follow-up studies on former adolescent patients show that 7.1 per cent of the inpatient males committed suicide within a 6-year follow-up (Pelkonen et al. 1996).

When considering the bearing of psychiatric illness on adolescent suicidal behaviour, two mediating variables must be accounted for: gender and comorbidity. Male and female adolescents were found to be affected differently by the different risk factors (Andrews and Lewinsohn 1992). This gender-based differentiation was recently demonstrated by Fennig et al. (2005), who found that while antisocial behaviour and depression are predicting factors for male attempters, types of defence mechanism and destructiveness are predictive for female attempters. Comorbidity is another mediating variable when assessing suicide risk in adolescents with psychiatric disorders. Beautrais et al. (1996) suggest that the likelihood for suicide by a person who suffers from one psychiatric disorder is 17.4 compared to the odds for a person with no psychiatric diagnosis. The probability for a person who suffers from two diagnoses of psychiatric disorder is 89.7. However, Houston (2004) found no such effect.

Psychiatric disorders and comorbidities are very common in adolescent suicide attempters and are described by Apter et al. in Chapter 85.

Depressed mood can appear with or without a diagnosis of major depression. Depressed mood in and of itself is one of the most critical risk factors for suicidal behaviours among young adults. Wetzler et al. (1996) examined severe attempters, non-severe attempters, suicide ideators, and non-suicidal adolescents. They found that depressed mood is fundamentally associated with all forms of suicidality. Similarly, Spirito et al. (2003) found that the baseline for depressed mood was the most strongly related factor to future suicidal ideation and attempts. An additional study found that when depressed mood is controlled for, other factors may become non-significant (Wichstrom and Rossow 2002).

While a strong association has been established between hopelessness and suicide in adults (Beck et al. 1985), such an association is less clear-cut in adolescents. Multiple studies report a strong association between hopelessness and suicidal behaviour in adolescents (Horesh et al. 2003; Thompson et al. 2005), yet a recent study of Turkish adolescents found that hopelessness did not predict suicide risk within their sample (Sayar and Bozkir 2004). This seeming contradiction can be resolved considering the recent findings of Eposito et al. (2003), indicating that hopelessness may be more critical for multiple suicide attempters than for single suicide attempters. There is also some lack of clarity regarding the role of hopelessness in male versus female suicide attempters. One study found that hopelessness added significantly to the prediction of suicide risk scores in female juvenile detainees but did not add to the prediction of suicide risk scores in males (Sanislow et al. 2003). In a study of victims of sexual abuse who were suicidal, hopelessness was more strongly related to suicidal behaviour in male victims of sexual abuse, while depressive symptoms were more critical in the case of suicidal behaviour in female victims (Bergen et al. 2003). Some studies found that hopelessness increases suicide risk in youngsters who internalize anger but not in those who externalize anger, and that hopelessness plays a more significant role for older adolescents than for younger ones (Barbe et al. 2005). Other studies, such as that of Goldston et al. (2001), report that when depression is controlled for hopelessness it is no longer associated to suicidality.

Excessive anxiety, especially trait anxiety, is an emotional characteristic of suicidal adolescents (Fennig et al. 2005). De WilDe et al. (1993) found that both hospitalized and non-hospitalized adolescent attempters experienced significantly more state and trait anxiety compared to non-attempters. Later, Goldston et al. (1999) distinguished trait anxiety, and not state anxiety, as associated with suicidal behaviour in adolescents.

Suicidal adolescents seem to experience more anger, more hostility, and more irritability than their non-suicidal counterparts (Penn et al. 2003). With regard to anger, some studies show that internalized anger is more critical to suicidal behaviour than externalized anger (Cautin et al. 2001). Other studies point to externalized anger as the more critical of the two types as far as suicidal behaviour is concerned (Zlotnick et al. 2004).

Feelings that are related to interpersonal relationships were also found to be associated with adolescent suicidal behaviour. Savarimuthu (2002) analysed audio-taped suicide notes of suicidal adolescents, following the path of expressed social emotions moment by moment. The investigators established that shame can be a devastating experience to the self, possibly leading to suicidal behaviour (see also Loraas 1997). Guilt feelings were found to be characteristic of suicidal young adults (Haliburn 2000). Suicidal youngsters experience far more inappropriate guilt than non-suicidal youngsters (Catalina-Zamora and Mardomingo-Sanz 2000). Suicidal adolescents also express a strong sense of loneliness (Batigun 2005). Guertin et al. (2001) found that sense of loneliness increases the odds of self-mutilation among suicidal adolescents almost sixfold.

Shneidman (1993) introduced the concept of unbearable mental pain, or as he terms it ‘psychache’, (Orbach et al. 2003) as the immediate reason for suicidal behaviour. Shneidman's concept, psychache, refers to a generalized emotional state that is different from any specific negative emotion. In recent studies, mental pain appears as a distinguished characteristic of the emotional state of suicidal adolescents (Orbach and Iohan 2005). Evren and colleagues (2001) found that 66.7 per cent of a given suicidal sample self-reported the reason underlying their suicidal attempt was ‘to get away from boredom and pain’ (Haliburn 2000).

Some suicidal adolescents are characterized by no specific emotional state of anxiety, anger, or hopelessness, but rather by rapid shifts in temperament. Such rapid shifts are related to difficulties in emotional regulation, and were found to be implicated in suicidal behaviour by some researchers (e.g. Miller et al. 2000)

Personality traits in and of themselves do not cause suicide. The same trait can be adaptive or non-adaptive depending on the situational demands. However, when some traits interact with other risk factors it may increase the risk of suicide by intensifying the suicidal crisis.

Impulsivity, aggression, and the tendency to act out in the face of frustration and interpersonal conflict are some of the most frequent personality traits found in suicidal adolescents (Fennig et al. 2005; Horesh et al. 2003). Eliason (2001) has found that impulsivity was the best discriminator between attempters and non-attempters and that among young impulsive individuals there is a very short span time between the suicidal thought and the attempt, thus exhibiting a difficulty controlling anger. Suicidal adolescents are often described as negativistic and as rejecting outside help for their emotional problems (Deane et al. 2001; Orbach 1997).

Other personality characteristics of suicidal adolescents are ambitiousness and perfectionism. An ambitious youngster who finds it difficult to compromise between high aspirations and the limitations of reality may choose to escape reality by taking their own life. Such a youngster may perceive compromise or failure as less bearable than death. Perfectionism is particularly critical in the development of hopelessness, although this is attenuated after controlling for depressive cognition (Donaldson et al. 2000). Yet socially prescribed perfectionism was found to be a primary factor in predicting the wish to die for suicide attempts among adolescents (Boerges et al. 1998).

Further, low self-esteem and a negative self-concept are significantly related to adolescent suicidal behaviour (Martin et al. 2005). By looking for adolescents with low self-esteem, researchers were able to distinguish young suicide ideators from suicide attempters (Merwin and Ellis 2004).

Finally, identity confusion, a lack of self-cohesiveness and self-integration, low self-complexity, lack of differentiation between self and parents, and discrepancies between actual self, normative self, and the ideal self, represent structural aspects of personality that are related to adolescent suicidal behaviour. These structural aspects of personality hamper the ability to regulate and cope, thereby increasing self-negativity and suicidal risk (Orbach et al. 1998; Brunstein-Klomek et al. 2005).

Self-regulation is conceptualized as the ability to control a range of internal systems that include affect regulation, modulation of anger, inhibition of self-destructive behaviour, minimization of negative ruminations, and self-soothing. It has been repeatedly found that suicidal youngsters encounter difficulties in regulation of negative emotions, negative cognitions, and impulsive behaviours (e.g. Esposito et al. 2003; Zlotnick et al. 1997). Negative attributional style is another self-regulation deficit that influences suicidal behaviour. Negative attributional style includes attributing positive outcomes to external forces of change and negative outcomes to the self (Schwartz et al. 2000; Rotheram-Borus 1988), negative self-appraisal, and negative appraisal of one's own degree of controllability of stressful events.

Fritsch and colleagues (2000) found that suicidal adolescents, as compared to non-suicidal adolescents, use more forceful and less conforming regulation strategies. Piquet and Wagner (2003) have categorized the cognitive and regulative patterns of suicidal youngsters into two systems and subsequently into four subtypes of coping strategies:

It was found that, relative to the comparison group, the adolescent suicide attempters made fewer approach-effortful responses and less avoidance-effortful coping responses, both considered the more adaptive coping strategies of the four.

Problem-solving deficits are a distinct cognitive characteristic of suicidal youngsters, expressed in their tendency to produce more problematic alternatives and fewer effective alternatives in interpersonal problem-solving tasks. Metacognitively, suicidal adolescents exhibit more pessimistic appraisal of their ability to solve problems. In addition, their problem-solving abilities are compromised by their inability to produce specific autobiographic memories (see review by Specker and Hawton 2005). Suicidal adolescents are also inclined to focus on problematic aspects of a stressful situation, yet at the same time they resort more to wishful thinking strategies when confronted with a problem (Goldston et al. 2001). In problem situations, suicidal adolescents prefer drastic solutions and dependence on others (Orbach et al. 1990). A history of repetitive failures seems to condition suicidal youngsters to perceive problems as inherently unsolvable (Orbach et al. 1999) and as out of their control (Wilson et al. 1995). Suicidal youth also have a cognitive style of an automatic production of negative thoughts (Nock and Kazdin 2002). Events and situations are automatically evaluated negatively (Kienhorst et al. 1992). Negative attributions are assigned to oneself, others, and to the future (Rudd 2000).

A variety of life events have been found to be related to suicidal behaviour, including bereavement, breakdown of close relationships, interpersonal conflicts, financial difficulties, legal setbacks, or disciplinary problems (Fergusson et al. 2000). Different life events have a different impact at different ages. Pertaining to interpersonal conflicts, parent–child conflicts constitute a greater risk factor for early adolescence, whereas romantic difficulties constitute a greater suicide risk factor during later adolescence (Groholt et al. 1998).

Beyond the general pool of life events, certain specific life events have consistently been found to influence the presence of suicidal behaviour in young adults. Sexual abuse and physical abuse, for example, were found strongly associated with suicidal behaviour. Sexual abuse is statistically predictive of suicidal behaviour even after controlling for depression, hopelessness, and family dysfunction. Girls who report distress about their experience of sexual abuse have a threefold increased risk of suicidal thoughts and plans compared to non-abused girls. Similarly, male adolescents who are highly distressed about their experience of sexual abuse have a tenfold increased risk for suicide attempts compared to non-abused adolescent males. However, the relationship between sexual abuse and attempted suicide also varies along gender lines: sexually abused males face a significantly higher risk of suicide (55 per cent) than sexually abused adolescent females (29 per cent) (Martin et al. 2004). Similar findings were found with regard to physical abuse (Johnson et al. 2001).

Interpersonal stressors within the family context have also been implicated in youth suicidal behaviour. These stressors include conflicts, rejections, harsh demands and expectations, faulty communication, ‘scapegoating’, family dysfunction, negative attachment, and lowered parental responsiveness (Cetin 2001; Orbach 1989; Wagner et al. 2003). However, there is some evidence that family effects might be mediated by the adolescent's psychopathology (Wagner et al. 2003).

Failure, especially academic failure, also constitutes a critical life stressor associated to suicidal behaviour among youngsters. Poor academic performance (compared to above average) is associated with a fivefold increased likelihood of a suicide attempt and has long-term predictive value of suicidality (Richardson et al. 2005).

Losses that are early, recent, or multiple are empirically associated with suicidal behaviour. Losses can take on the form of death, separation, or parental divorce (Liu and Tein 2005). In a recent study, Orbach and Iohan (2007) studied experiences of loss among psychiatric suicide attempters, non-attempters, and controls. The types of losses studied included the loss of a close person (e.g. parent), material loss (e.g. loss of a job), mental loss (e.g. loss of faith), and physical losses (e.g. loss of good health). Compared to the non-attempters and the control group, the suicidal group reported more mental loss, physical losses, and loss of a close person. The number of losses was found to be significantly related to suicidal behaviour. The relationship between type of loss and number of loss and suicidal behaviour were sustained even after controlling for depression.

Liu and Tein (2005) found that an accumulation of such negative life events, regardless of the type of event, is a critical factor in suicidal behaviour. They report that 4–6 events have an odds ratio of 1.40; 7–9 events have an odds ratio of 2.02, and 9 and up events have an odds ratio of 3.73 (see also Roberts et al. 1998).

Facilitators and inhibitors (protective) factors can increase or decrease the probability of acting out suicidal impulses. These factors are not considered direct causes for suicide, rather they determine whether the suicidal person will act on the already existing suicidal tendencies, ideation, or wishes.

Attraction to death and distorted beliefs about death, e.g. perceiving death as an improved state of life (Orbach 1994), bodily dissociation (numbness, detachment, high sensation threshold) and negative attitudes towards the body (Orbach et al. 1997, 2006) have been identified as facilitators of suicidal behaviour. In accordance with these findings, attraction to death and bodily dissociation make it easier for the suicidal youngster to choose death and to carry out an aggressive act against their own body.

Other facilitators include exposure to the suicidal acts of others, specifically when the possibility of vicariousness is more tangible. One obvious example is being made aware of the suicide of a friend (Stack 1996) or of a close relative (Gallo and Pfeffer 2003). However, a less apparent, yet no less potent form of vicarious exposure to suicide can occur through media reports of a suicidal act committed by an individual. The magnitude of suicides committed as a result of media facilitation increases in proportion to the amount, duration, and prominence of media coverage on suicide (Pirkis and Blood 2001). Suicide coverage may increase suicidal behaviour in several ways. Repeated exposure of a given suicidal act may promote identification with the attempter as well as with the method of attempt (Schmidtke and Schaller 2000; Pirkis and Blood 2001). Repetitive reporting of suicidal acts may also portray suicide as normative behaviour (Schmidtke and Schaller 2000). Furthermore, presenting suicide as a feature story has been hypothesized to promote the idealization of suicide and to engender the wish to receive attention (Orbach 1997).

Availability of means is an important facilitator of suicidal behaviour, especially among impulsive suicides (Hawton et al. 2001). The presence of firearms within the home is a critical risk factor for suicide in adolescents (Brent et al. 1988; Grossman et al. 2005).

Moreover, social norms in and of themselves can facilitate suicide. In Domino and Takahashi's (1991) taxonomy study, Japanese medical students scored higher than their American counterparts on the Right to Die Scale. Differences in the scores were found related to suicide rates in the two countries.

Social support can serve as strong protective factors against suicide. Peer and family support were found to reduce various risk behaviours in youngsters who were sexually abused (Perkins and Jones 2004). Similarly, a sense of connectedness to parents or peer groups, as well as a sense of belonging to a positive school climate, was found to be strong protective factor against emotional distress and suicidal behaviour (Perkins and Jones 2004). Extra-curricular activities such as engaging in sports also contribute to suicide prevention in adolescents (Perkins and Jones 2004; Tomori and Zalar 2000).

Family cohesion in the form of mutual involvement, shared interests, and emotional support is another protective factor. This was found to be true in a longitudinal study of middle school students (McKeown et al. 1998). One way that family cohesion can serve as a protective factor is that it seems to imbue its members with a sense of responsibility to family, as well as to close ones in general (Kyle 2004). Family cohesion was also found to mitigate suicidal behaviour, depression, and general life stress (Rubenstein et al. 1998).

A sense of self-cohesion and a strong sense of identity have been found to protect against suicidal behaviour, especially under stressful conditions. Katzir (2005) evaluated the sense of self-cohesion and identity in Israeli 18-year-olds prior to enlisting in the army, and did a follow-up on them throughout their military service. He found that young soldiers who had a strong sense of self-cohesion and identity were more resilient and less suicidal even under very stressful conditions compared to their counterparts.

The concept of religiosity as a protective factor was first introduced by Durkheim (1897, 1951). The protective value of religion lies in that it offers cohesiveness and integration (Durkheim 1897, 1951), social support and sense of belongingness (Pescosolido and Georgianna 1989), commitment to a few core life-saving beliefs (Stack 1992), rules and customs (Greening and Stoppelbein 2002) and moral obligations (Kyle 2004). Subsequently, Greening and Stoppelbein (2002) studied intrinsic and extrinsic religiosity and orthodoxy and their respective relationships to suicidality, depression, and hopelessness among a very large population of white Christian adolescents in the United States. In this study, orthodoxy emerged as a strong protective factor against suicidality (see also Hilton et al. 2000; Gould et al. 2003).

One of the most widely accepted theories regarding the relationship between problem-solving deficits and suicidal behaviour is the stress–diathesis–hopelessness hypothesis (SDH) developed by Schotte and Clum (1987). According to Schotte and Clum, individuals with difficulties in divergent thinking are unable to develop efficient solutions while under stress. As a result of their inability to conceive of a rational solution, their efforts are often reduced to purely psychological reactions of helplessness and hopelessness, leading individuals to view suicide as the only solution. Several authors have recently suggested elaborating the SDH hypothesis so that it considers not only the leverage of the cognitive deficits in problem-solving, but also the equally important role of the under-evaluation of one's own ability to problem-solving. Dixon et al. (1994)  Rudd et al. (1994), and Yang and Clum (1996) have found that the under-evaluation of one's own ability to solve problems, rather than one's actual ability or performance, is a critical factor in the SDH process. This is also consistent with assertions that even highly intelligent individuals, who have the cognitive abilities to create solutions, may encounter problem-solving difficulties due to their lack of confidence (Shure 1997).

Clinical observers of suicidal youngsters (Sabbath 1969) report that these youngsters have often experienced strong rejections by their parents from very early on in life. Sabbath gives examples of commonplace phrases employed by the parents of suicidal adolescents, conveying ruthless messages implying that the adolescent would be better of dead. One mother, for example, was in the habit of telling her 15-year-old daughter to ‘drop dead.’ Another example is of a father who would often convey tell his daughter: ‘If you've got one rotten apple in the barrel you've got to get rid of it’ (Orbach 1988).

Orbach (1986, 1988, 1989) suggests that suicidal tendencies in youth are directly linked to family situations and demands that pressure the child or adolescent to solve irresolvable problems. Some typical irresolvable problems are so because of the very nature of the problem (e.g., to excel beyond one's capability), a family problem that is disguised as a problem of the child (e.g. one parent exacerbating the child's problem and using it as leverage to keep the other parent within the family unit), limiting alternatives for solutions, and creating a new problem whenever the old one is resolved. In a recent empirical test of this theory, four experiential elements of facing irresolvable problems have emerged: feeling that the demands are unattainable (realistically so); a sense of an inextricable commitment to parental happiness; a sense that the youngster is required to behave in a problematic way; and giving up individuality for the sake of the parents (Orbach et al. 1999).

Richman (1978) found that suicidal adolescents were often the product of a family characterized by symbiosis without empathy. Symbiosis without empathy is a forced strangling bond without any expression of love and warmth. The parents demand total loyalty, yet are distant and estranged. As a result of these paradoxical conditions, the symbiotic family develops a massive generalized identity, with little distinction among the different members, and each family member is often ironically left with a feeling of isolation and loss of self. Another result of the family's extreme drive for unity is that intimate relationships outside the home become a complete taboo. Ultimately, the youngster is forced to choose between a complete break with the outside world or a complete separation from the family. Therefore, when adolescents from a symbiotic family experience failure outside the family unit, suicide often becomes a symbolic route for reunification and total fusion with there family. Thus, suicide in such youngsters is both an escape from, as well as a reunion with, the family.

This theory suggests that there is a basic difference between suicidal and non-suicidal individuals in terms of an early propensity for suicidal behaviour. One version of this theory was suggested by Mann et al. (1999), and it was expanded by Wasserman (2001). According to the propensity theory, the suicidal propensity (stress–diathesis) is rooted in genetic, biological, and biochemical deficits (e.g. genetic inheritance, low serotonin activity). This propensity, which can be intensified by early and prolonged stress (e.g. mental illness, long-standing relationship problems) creates a readiness to respond to life difficulties with hopelessness, suicidal ideation, and with the planning of suicide. The suicidal propensity also involves a tendency to act out impulsively and aggressively. The propensity becomes active when it interacts with self-regulation deficits and current stressful life triggers such as acute psychiatric illness, loss, and separation or narcissistic injury.

Boergers et al. (1998) studied male and female adolescents in a general hospital emergency room in terms of their self-reported reasons for their suicide attempt. The most frequently reported reason given for the suicide attempt was that they simply wanted to die, followed by the desire to be relieved from a terrible state of mind, to escape from an impossible situation, and to make people understand how desperate they feel. The rest of the reasons provided were of a more manipulative nature (e.g. to influence someone, to find out whether someone loved you). These findings were also confirmed by Haliburn (2000).

In order to understand the relationships and interactions between the many factors involved in suicidal behaviour of adolescents in a more coherent way, several theoretical models have been introduced (Orbach 2001; Beautrais 2003; Rudd 2000; Sandin et al. 1998; Yang and Clum 1996). The various risk factors can be categorized into several main categories. The following categorization is suggested in this review: biological factors, morbidity factors, background factors (e.g. age, gender), stressors and triggers, mediating and moderating factors such as cognitive deficits, emotional deficits (e.g. depressive mood), personality aspects (e.g. impulsivity), and facilitators and inhibitors (protective factors).

Different models may use different systems of categorization. For example in some models depression is conceptualized as a mediating factor (Thompson et al. 2005) while in others, depression is conceived of as a more independent morbidity factor (Wasserman 2001). Some of the factors can appear in more than one category. For example, in Beautrais' (2003) model, life stress can appear first as an independent factor and later as a mediating factor. While some theoretical models make use of many of the factors found to be involved in suicidal behaviour, others make use of very few. Most of the theoretical models suggested emphasize interactional rather than linear models. The general trend of interaction that emerges within the various models is described as a flow from biological factors to morbidity factors, background factors, mediating factors, moderators, and finally facilitators and inhibitors.

Below, three examples of empirical examinations of theoretical models of suicidal behaviour in adolescents are described. Lewinsohn et al. (1996) examined the contribution of several independent variables on suicidal behaviour in adolescents (thoughts about death, death wishes, suicidal ideation, suicide planning, less serious suicide attempts, more serious suicide attempts, multiple suicide attempts). The independent variables included psychopathology (depression, anxiety, disruptive behaviour, and substance abuse), physical illness (number of sick days, number of doctors visits, physical symptoms), background/personal history factors (parental divorce/separation, death of a parent, teenage mother, moving away from home, suicide attempt by friend, death of a relative, poor social support, conflict with parents, everyday problems), interpersonal problems (involvement in arguments or fights, break-up with a friend, emotional over-dependency, and emotional estrangement); mediating variables (negative cognitions attribution style, self-esteem and coping skills). Lewinsohn et al. (1996) found that each factor (psychopathology, physical illness, environment and interpersonal) constituted a distinguished pathway to suicide with the first three showing a direct influence on suicidal behaviour. At the same time, however, all four categories were also found to have an indirect contribution mediated by the faulty cognitions.

Orbach and Iohan (2005) found a different configuration of interaction. They studied personality characteristics (negative emotional regulation, tolerance for mental pain, gender, dependency, perfectionism, and self-criticism), environmental stress (perceived stress, number of various types of loss), negative experiential aspects (mental pain), and depressive symptoms. Personality variables such as negative emotional regulation, self-criticism and perfectionism had a direct impact on suicidal behaviour (ideation, tendencies, and attempts). These variables, as well as a tendency for dependence, were found to have an additional indirect impact as well, mediated first by perceived pressure and number of losses and then by mental pain and depressive symptoms. In contrast to (Lewinsohn's et al. 1996) model, Orbach and Iohan's model separated between personality aspects (e.g. perfectionism) and more subjective aspects (e.g. mental pain). However, similar to the Lewinsohn et al. (1996) model, multiple interactions were found among the independent and mediating factors.

Rather than following the interactional flow of different factors, Roberts et al. (1998) examined the accumulative impact of different contributing factors. Among other factors, they studied lifetime suicide attempts, age, gender, socio-economic status, depression, loneliness, life stress, fatalism, pessimism, and self-esteem. They computed the odd ratios of each variable for suicidal behaviour and found depression, lifetime suicide attempts, and life stress to have the highest odds ratios respectively. They also found that the odd ratios for having one of the 6 factors examined was 3.48, and for 6 factors the odd ratios increased to 67. Other empirical models have discovered different factor structures and interactions for male and female adolescents (Lewinsohn et al. 2001).

Achenbach (1991) has distinguished between externalizing pathologies and internalizing pathologies in youth. Externalizing pathologies include such symptoms as delinquent behaviour, aggressive behaviour, impulsivity, oppositional behaviour, and hyperactivity. Internalizing pathologies include withdrawal, somatic complaints, anxiety, depression, inhibition, and being self-demanding. These two types of pathologies can be linked to different pathways of suicidal behaviour in adolescents independent of negative life events. Similar findings were reported by Vermerien et al. (2002).

Orbach (1997) has also distinguished between different pathways to suicide in general, positing three clusters of suicidal behaviour. The depressive perfectionist cluster (the internalizing cluster) is hypothesized to be mediated by severe negative emotions. The aggressive impulsive cluster (externalizing cluster) is hypothesized to be mediated by deficits in impulse control. The disintegrating cluster characterized by panic, severe anxiety and psychiatric pathology is hypothesized to be mediated by a severe loss of control.

As Wagner and Hustead (2002) perceive it, the pathways to youth suicidal behaviour are paved upon child–family relationships. One such pathway is the child-driven pathway wherein the suicidal behaviour is primarily enabled by the child's problems. This pathway is characterized by children who develop insecure or disrupted attachments towards the parents despite their parents' supportiveness and competence. Furthermore, the child's relationship with, and treatment of, their parents is distinctively negative and aggressive. This pathway involved children with high psychopathology to parents with low psychopathology. The child has a short history of suicidal behaviour, yet their suicide attempts are highly lethal and driven by an attempt to escape pain and do not implicate the family. The second pathway is the parent-driven pathway wherein the child's suicidal behaviour is primarily enabled by the parents. This pathway involves parents who are poorly competent, and who are insecure in their attachment to the suicidal child. Their treatment of and relationship with the suicidal child is marked by aggression and negativity. In this pathway the parents are highly psychopathological and the child is low in psychopathology. The child has a long history of suicidal behaviour, yet their attempts are low in lethality. The attempts are described as interpersonal and communication-based, and the family is implicated in precipitating the attempt. The third pathway posited by Wagner and Hustead is the reciprocal pathway characterized by poor parental competence and support, high parental and child psychopathology, mutual parent and child insecure attachments, long history of suicidal behaviour and parent and child mutual perceptions of negative–aggressive relationships. This pathway usually results in lower lethality of attempts that are described as interpersonal messages and as being precipitated by the family. The authors report finding strong empirical supports for the first and third pathways and somewhat weaker support for the second pathway.

From a different perspective, Blatt (1995) posits personality develops as a consequence of a complex interaction between two fundamental lines: (a) the development of the capacity to establish mature and satisfying interpersonal relationships and (b) the development of a realistic, positive, and integrated self-definition and identity. An overemphasized interpersonal relatedness may lead to an anaclitic (or dependent) depression, whereas overemphasized individuality and self-definition may result in self-critical (or introjective) depression. Anaclitic depression involves a deep longing to be loved and cared for. The overly individualized person is characterized by self-criticism, feelings of inferiority, and guilt. Each of these imbalances were found to be related to suicidal behaviour in different ways (see also Brunstien-Klomek et al. 2005; Fehon et al. 2000; Orbach and Iohn 2005). Both anaclytic depression and introjective depression were found to be implicated in suicidal behaviour.

Applying a cognitive approach, Dieserud et al. (2001) offer a two-path model of suicide attempt for all ages that is somewhat parallel to Blatt's two pathway model. The first pathway begins with low self-esteem, loneliness, and separation or divorce, advancing to depression, then hopelessness, suicide ideation, and finally a suicide attempt. The second pathway begins with low self-esteem and a low sense of self-efficacy, followed by negative self-appraisal of one's own problem-solving capacity, and poor interpersonal problem-solving skills, finally leading to suicide. This model emphasizes the importance of addressing both depression and hopelessness, as well as problem-solving deficits when working with suicide attempters.

There is an abundance of information on suicidal behaviour in general and on adolescent suicidal behaviour in particular. Unfortunately, the natural conclusion—that we have a good understanding and knowledge of this tragic phenomenon—is not fully accurate. One reason for this counterintuitive reality is that empirical findings have been transmitted through various and non-concurring theories and terminologies. It is our belief that the first step to furthering our knowledge of suicide is to promote a coherent organization of the data for the sake of advancing conceptual clarity. Further, the present review shows that there that there is more than one pathway or one dynamic of suicidal behaviour. It is evident that these should be taken in consideration when prevention programmes are planned. One size does not fit all and different prevention programmes should be tailored for each dynamic or pathway. Such efforts can help us to better define our future goals in the study of suicide and eventually lead to an improved effort in the prevention of suicidal behaviour in adolescents.