89 Psychiatric disorders in suicide and suicide attempters

This chapter reviews common psychiatric disorders and conditions which appear to be major risk factors for all types of suicidality, both non-fatal and fatal, among children and adolescents. These psychiatric conditions include personality disorders, conduct disorder, affective disorder, bipolar disorder, anxiety disorder, obsessive–compulsive disorder, schizophrenia, substance abuse, and eating disorders. Each psychiatric condition is described and discussed in terms of its unique features that are associated with suicide behaviours and its risk for suicide behaviour.

It appears that almost any diagnosable psychiatric disorder may increase the risk for suicide.

Psychiatric illnesses in adolescence are especially dangerous when they occur in conjunction with other risk factors for suicide (see Chapters 38 and 39 in Part 6 of the book) and when more than one illness is present (comorbidity). This chapter will deal with the common psychiatric disorders found in clinical practice that are associated with suicidal behaviour. Different treatments of suicidal young people are described in detail in Chapters 91, 92 and 93 in this part (Part 12) of the book.

Borderline personality disorder (BPD) is traditionally associated with non-fatal suicide attempts but there is increasing evidence that fatal suicide is also common in these patients. Intentional self-damaging acts and suicide attempts are the ‘behavioural specialty’ of these patients (Gunderson 1984). Although ‘affective instability’ is said to be one of the critical symptoms of this disorder, many seem to have a chronic underlying depression and most of the adolescent BPD patients who require psychiatric help meet criteria for an affective disorder, usually major depression. In addition, many suffer from chronically stable depression, hopelessness, worthlessness, guilt and helplessness.

Another group of suicide related symptoms are those associated with anger. Many of these patients are very angry and even violent; others are fearful of losing control of their anger and are unable to express their aggressive feelings. Other frequent comorbid conditions that increase the likelihood of suicide are conduct disorder, bulimia and substance abuse. An additional comorbid condition of considerable interest is dissociative disorder and the common origin of this combination often seems to appear after incest or continuous non-injurious (in the physical sense) sexual abuse. Some authors report even seeing multiple personality disorder developing in these patients (Pope et al. 2006), although our group has never seen such a case.

About 9 per cent of patients with BPD eventually kill themselves (Stone 1989). In a series of BPD inpatients followed for 10–23 years after discharge, patients exhibiting all eight DSM III criteria for BPD at the index admission had a suicide rate of 36 per cent, compared to a suicide rate of 7 per cent in people who had 5–7 of the criteria. Patients with BPD also have high repetition rates of suicidal behaviour (Brodsky et al. 1997). Individuals with BPD who show impulsive and aggressive characteristics combined with over-sensitivity to minor life events are at risk for suicidal behaviour. This sensitivity often leads to angry and anxious reactions with secondary depression. These subjects tend to have suffered childhood physical and sexual abuse and use defenses such as regression, splitting, dissociation and displacement. There is often a history of alcohol or substance abuse and there appears to be a connection to an underlying disturbance of serotonin metabolism which is genetic in origin (Apter and Ofek 2001).

Our own experience with this personality constellation and suicide has been based on psychological autopsies of young soldiers doing their military service in the Israel Defence Force (Apter et al. 1993a) Many of these suicides seemed to be very different from the patients seen in the adolescent unit or in the emergency room and in fact, the vast majority had never been in contact with a mental health professional. However, once alerted to this maladaptive personality style, there seemed to be resemblances to the life patterns of some of our adolescent patients, especially those with more serious suicide attempts.

Psychological post-mortem studies of soldiers in the Israeli Defence Force (IDF) have shown some features that differ from similar studies conducted in Europe and the USA (Apter et al. 1993a; King and Apter 1996). Strong narcissistic and perfectionist patterns were a feature of the lives of about a quarter of the suicides, while 40 per cent showed schizoid and avoidant traits in their personalities. The IDF is for many Israeli youths a chance to prove their worth—much like a prestigious college or university career for European or American students. For many the military is a second chance to redeem earlier shortcomings or to confirm a sense of a competent identity.

These high self-expectations and hopes may also have made it difficult for these young men to acknowledge or bear even marginal difficulties or personal limitations that emerged during their subsequent active duty: any shortcoming was seen by them as devastating.

These features were often complicated by strong isolative traits, which seemed to be common to many of the suicide victims. In most cases these seemed to be lifelong personality patterns which were not related to stress or periods of depression. Their parents, teachers and friends remembered them as being ‘very isolated’ children. Their superior officers often termed them as being ‘very private’ people.

This combination of traits had several very dangerous consequences. Many of these young people seemed to have felt an overwhelming need to make a good impression on their pre-induction assessments, which are made on all Israeli teenagers at the age of 16. A good score in these assessments gives the opportunity to be assigned to a prestigious battle unit—a highly valued attainment in the eyes of Israeli society.

This is exemplified by another finding that is unusual in most studies of suicide. These subjects had much higher physical fitness ratings than the average Israeli soldier, probably reflecting their minimizing of non-specific and subjective physical symptoms, such as backache and flat feet and/or intensive training to reach high levels of fitness before conscription. Once these young men encountered perceived difficulty, they felt shame related to their unrealistically high standards, which was combined in many cases with an isolative style that prevented them from turning to peers, officers or clinicians for help or support.

As a result, even minor setbacks (to the external observer) could rapidly spiral into disaster as burgeoning anxious preoccupation, depressive rumination and withdrawal further interfered with the recruit's ability to perform at the high levels he demanded of himself or to reach out to others, triggering a vicious cycle of isolative decompensation, with suicide as the only way out.

King and Apter (1996, p.112)

Those recruits who used achievement as a kind of pseudo-mastery to substitute for a lack of real interpersonal closeness seemed especially vulnerable to this kind of catastrophic decompensation.

Some basic rules and values are essential in treating the borderline suicidal adolescent. Probably most crucial is a stable therapeutic framework with consistent and reliable care. Next, behaviour and feelings are used as the principal mode of communication. The therapist is active and uses high-energy confrontation and care (so-called therapeutic pressing). The central message is always ‘doing something with the patient, not something to the patient’. This way patients feel somewhat in control, which might keep them in treatment. Reflecting their splitting mechanism, these patients alternatively feel inferior and omnipotent, angry at others and self-destructive, sensitive to rejection but usually provoking it. Flexibility in approach, but firmness in basic values, with creativity and readiness to step away from the rules to get out of frustrating ‘no way out’ situations, is essential. Many of these patients cannot tolerate feeling better, as this means that the therapist is successful. These and similar frustrative situations cause countertransference problems, with a potential loss of professional objectivity; constant supervision and a support network are therefore necessary.

At least initially, the therapy is supplemented by as much structuring of the patient's life as needed. This may range from directing behaviours, to day-hospital programmes, to hospitalization. Complications often require additional structuring such as phone calls or extra sessions. Structuring, advice, and logic are not expected to generate personality change, only to temporarily improve behaviour control.

Patients who are dysphoric and highly sensitive to social approval (i.e. who are high in reward dependence and harm avoidance) are most likely to improve on selective serotonin reuptake inhibitors (SSRIs). In contrast, those who are highly fearful but not socially dependent are most likely to improve on noradrenergic uptake inhibitors such as desipramine (norpramine). Children with attention-deficit/hyperactivity disorder (who are high in novelty-seeking) are efficiently treated with drugs that increase dopamine release and inhibit its reuptake, such as methylphenidate.

Aggressive impulsive behaviour is a major risk factor for suicidal behaviour in adolescence. A major concern is the high rate of suicidal behaviours among juvenile delinquents, especially in those who are incarcerated in remand homes or in prisons. Unfortunately, in many countries reform schools often do not have the facilities for adequate mental health treatment, while psychiatric units cannot cope with the violence and aggression displayed by these youths. Many of the risk factors for suicide are also risk factors for conduct disorder. These include broken homes, physical and sexual abuse as children, personal and familial alcoholism and substance abuse, unemployment and poverty and access to firearms.

In addition, mood disorders are often present in children who have some degree of irritability and aggressive behaviour. It may be quite difficult to make the differential diagnosis between major depressive disorder, bipolar disorder and conduct disorder. There is also a substantial comorbidity between conduct disorders and affective disorders in adolescence. Again, many factors predisposing to depression also predispose to conduct disorder including family conflict, negative life events, level of affiliation with delinquent peers and parental involvement.

Examining the Center's for Disease Control and Prevention (CDC) Youth Risk Behaviour Survey (YRBS) data for 3054 Massachusetts high school students, Woods et al. (1997) found lifetime suicide attempts were significantly associated with physical fights in the past year, regular tobacco use, lack of seatbelt use, gun carrying, substance use before last sexual activity, and lifetime drug use. An analysis of the national cross-sectional survey data on 11,631 high school students found significant and substantial correlations between suicide attempts and gun carrying, multiple sexual partners, condom non-use, fighting resulting in injury, driving while intoxicated, and cocaine use (Sosin et al. 1995). Orpinas et al. (1995) examined the YRBS data for 2075 Texas high school students and found that weapon-carrying and fighting resulting in injury were associated with suicidal ideation, as well as with alcohol use, number of sexual partners, and low academic performance.

Adolescent suicide attempters are also at ongoing risk for injury and death from motor vehicle accidents, substance abuse, homicide, etc. Prospective studies demonstrate a shared set of social, developmental, and psychopathological risk factors predicting completed or attempted suicide and unintentional injury or death (Fergusson & Lynskey 1995; Neeleman et al. 1998).

Risk factors and prognostic implications associated with adolescent suicidal behaviour differ with gender. In their psychological autopsy study of adolescent suicides, Shaffer et al. (1996) found that for boys, completed suicide was associated with major depression, substance abuse, and/or antisocial behaviour; in contrast, for girls, major depression and antisocial behaviour, but not substance abuse were associated with increased risk of suicide completion.

Dryfoos (1990) has summarized the extensive overlap in both the prevalence and risk factors for delinquency, substance use, teen pregnancy, and school failure, including the frequent commonalities of poverty, low resistance to deviant peers, insufficient bonding and communication with parens, and insufficient, harsh or inconsistent parental discipline or monitoring.

In conclusion, it seems as if there may be a type of suicidal behaviour characterized by low fatality suicidality related to a wide variety of other kinds of impulsive behaviours, as seen in adolescents with conduct disorders. The psychology of this form of suicidality is related to defects in impulse control similar to those seen in the borderline personality organization as described by Kernberg (1975) and may be related to defects in serotonin metabolism.

The very violent, suicidal and drug abusing adolescent usually requires some form of institution—if possible manned both by social services and mental health personnel. Interventions usually include working with families and schools and often with the police, especially where community policing is available. These young people often do not respond to traditional mental health approaches. However, when they are suicidal there is often quite extensive psychiatric comorbidity which should be treated. Theoretically, anti-aggressive medication (‘serenics’) should be useful for both internally and externally directed aggression but in practice this is usually not the case. Some clinicians favour the use of mood-stabilizing medications such as carbemazepine and sodium valproate.

There is significant agreement on criteria for hospitalization of patients with conduct disorder (CD) (Lock and Strauss 1994), but level of care decision-making continues to be complex and unsupported by empirical data. The psychiatric professional should choose the least restrictive level of intervention that fulfils both the short- and long-term needs of the patient. Imminent risk to self or others, such as suicidal, self-injurious, homicidal, or aggressive behaviour, or imminent deterioration in medical status, remain clear indications for the need for hospitalization (American Academy of Child and Adolescent Psychiatry 1996, 1990).

Major depressive disorder is reported to be the most common mood disorder. It may manifest as a single episode or as recurrent episodes. The course may be somewhat protracted—up to 2 years or longer—in those with the single-episode form. Whereas the prognosis for recovery from an acute episode is good for most patients with major depressive disorder, three out of four patients experience recurrences throughout life, with varying degrees of residual symptoms between episodes. Bipolar disorders consist of at least one hypomanic, manic, or mixed episode. Mixed episodes represent a simultaneous mixture of depressive and manic or hypomanic manifestations. Although a minority of patients experience only manic episodes, most bipolar disorder patients experience episodes of both polarity (Wasserman 2006).

Among teenagers, both attempted and completed suicide is, in the great majority of cases, preceded by depressive symptoms. Depressed young people who attempt suicide often come from broken families and have had one or more relatives who have committed or attempted suicide. They have also, relatively often, run away from home and thus been brought up without favourable role models. Physical and mental abuse, as well as sexual assault, is also more common in this group. Young people who have attempted suicide often have ongoing problems at school and also difficulties in achieving workable relationships with their peers, compared with young people who are depressed and have not attempted suicide. Abuse of alcohol and drugs, impulsive behaviour and asocial behaviour are additional risk factors for attempted and completed suicide among depressed young people. Owing to the high incidence of depression among young people who have attempted suicide, it is important to make a diagnosis and provide adequate treatment at an early stage. Studies show that depressive disturbances are more common among children and young people than has previously been believed. Unfortunately, many young people with depression are not identified, partly because their depressive symptoms are often atypical and partly because adults do not readily recognize depressive symptoms in the young, owing to their wish to see their children as happy and healthy.

Major depression is most easily diagnosed when it appears acutely in a previously healthy child and in these cases the symptoms closely resemble those seen in adults. In many children, however, the onset is insidious and the child may show many other difficulties such as attention deficit disorder or separation anxiety disorder before becoming depressed.

Mood disorders tend to be chronic when they start at an early age and the children come from families where there is a high incidence of mood disorders and alcohol abuse.

In some cases the depressed adolescent may also be psychotic and have hallucinations and delusions which are usually mood congruent. When the psychotic themes are related to suicide, such as in command hallucinations or delusions of guilt, the risk for suicide is very high.

Although antidepressant medications are widely used in the treatment of depression and suicidality, the effect of medication treatment on suicidal behaviour remains unclear. Although the evidenced-based literature is gloomy, most outpatient departments provide a multidimensional treatment approach involving clinical (psychological and pharmacological) interventions, common sense and supportive therapy. This approach has proven sufficient in most cases.

Electroconvulsive therapy may also be indicated in suicidal psychotic depression in adolescence.

Bipolar disorder was once thought to occur only rarely in youth. However, approximately 20 per cent of all bipolar patients have their first episode during adolescence, with a peak age of onset between 15 and 19 years of age. Developmental variations in presentation, symptomatic overlap with other disorders, and lack of clinician awareness have all led to under-diagnosis or misdiagnosis in children and adolescents. Therefore, clinicians need to be aware of some of the unique clinical characteristics associated with the early-onset form. Similarly, it is important to recognize the various phases and patterns of episodes associated with bipolar disorder. Youth may first present with either manic or depressive episodes. Twenty to thirty per cent of youth with major depression go on to have manic episodes (Weller et al. 2002).

Adolescents with bipolar disorder are at increased risk for completed suicide. Twenty percent of adolescents with bipolar disorder made at least one medically significant suicide attempt (Weller et al. 2002). Lithium has been found to effectively reduce suicide rates during long-term treatment of patients with bipolar disorders (Baldessarini et al. 2002). Data on the efficacy of anticonvulsant mood stabilizers in reducing suicide risk are sparse (Yerevanian et al. 2003).

Anxiety has been identified as an important risk factor for suicidal behaviour in adults. A follow-up study of patients with major affective disorder (Fawcett et al. 1990) found that anxiety symptoms were strongly related to completed suicide within one year of assessment. There have also been studies that indicate that anxiety disorders are associated with an increased risk of suicidal behaviour (Allgulander and Lavori 1991; Mannuzza et al. 1992; Massion et al. 1993). Studies with adolescents have shown mixed results. Taylor and Stansfeld (1984) found that when compared to psychiatric outpatients, suicide attempters exhibited higher levels of anxiety (38 per cent vs 22 per cent); however, the difference was not significant. Another study (Kosky et al. 1986) reported that depressed suicidal ideators (of whom 39 per cent had attempted suicide) manifested high levels of anxiety (76.4 per cent), but these levels were not significantly different from those of depressed non-suicidal adolescents. Andrews and Lewinsohn (1992) found in a large community sample of adolescents a significant association between anxiety disorders and suicide attempts in males, but not in females. Most research on anxiety as a risk factor for suicidal behaviour has focused on the measurement of state anxiety. This may not be a fruitful method if state anxiety is significantly reduced following a suicide attempt. Ideally, risk factors used for predictive purposes should be stable (Hawton 1987). As such, research on the relationship between anxiety and suicidal behaviour might benefit from focusing on the measurement of anxiety as a trait rather than as a state. In fact, Apter and his colleagues (Apter et al. 1990, 1993b) found that adult psychiatric inpatient suicide attempters had significantly higher levels of trait anxiety than inpatient non-attempters, whereas state anxiety did not discriminate between the two groups. Moreover, trait anxiety was highly associated with a self-report scale of suicide risk. Another study (Oei et al. 1990) found that adult depressed patients with suicidal ideation had significantly higher levels of state and trait anxiety than depressed patients with no suicidal ideation. A study of Dutch adolescents (De WilDe et al. 1993) found that suicide attempters (half of them psychiatric patients and half high school students) exhibited significantly higher levels of state and trait anxiety than non-depressed non-attempters (high school students).

There are no studies on treatment of suicidal behaviour in anxiety disorders and the treatment should be based on the same considerations as those for depressive disorders, with CBT probably being the most efficacious psychotherapeutic method for the basic anxiety disorder.

Obsessive–compulsive disorder (OCD) is a well-described disorder in childhood and adolescence occurring in about 3 per cent of 16-year-olds (Apter et al. 1997). Although effective methods of treatment are available, a substantial minority of patients do not respond to therapy and have a relatively poor prognosis. A minority will require psychiatric hospitalization (Jenike 1990). The factors contributing to treatment failure include: comorbidity with personality disorders, tic disorders, conduct disorders and oppositional disorders, dysfunctional families and the presence of depression and suicidal behaviour (Foa 1979; March and Mulle 1998). OCD is a common comorbid condition with affective disorder. About 10 per cent of unipolar depressive individuals and 21 per cent of those suffering of bipolar disorders, show comorbid OCD (Chen and Dilsaver 1995). Perugi et al. (1997) found 15.7 per cent of subjects with OCD have episodes of a bipolar disorder. It appears that depression is also common among adolescents with OCD (Valleni-Basile et al. 1994) and that depression in combination with OCD may lead to an increased risk for suicidal behaviour (Valleni-Basile et al. 1994; Chen and Dilsaver 1995). Non-fatal suicidal behaviour occurs in about 15 per cent of individuals suffering from OCD (Angst 1993; Hollander et al. 1996), principally in those showing comorbidity with antisocial behaviour or borderline personality disorder.

Adolescent inpatients with OCD show high levels of depression, suicidality, anxiety, impulsivity, violence and aggression. The fact that all the hospitalized adolescents, irrespective of their diagnosis, showed similar high scores on the BDI tend to support the notion that these depressive symptoms are non-specific and represent a reaction to illness and hospitalization.

Although the OCD adolescents resemble their inpatient counterparts on almost all the psychopathological dimensions, they do seem to differ in regards to their suicidal behaviour. This difference is however complex and not easy to explain.

First, although on a scale of suicidality ranging from no suicidality through ideation threats, gestures and attempts, the OCD inpatients are equal to the non-OCD patients, they are significantly less likely to have made an actual attempt. Second and most striking is the relationship between depression and suicidal behaviour. In the patient controls there is an expected strong correlation between levels of depression and suicidal behaviour. In the OCD group there is in fact a non-significant trend in the opposite direction. Similarly the relationship between both trait and state anxiety and suicidal behaviour is highly significant in the patient controls but weak and non-significant in the OCD inpatients. The same findings apply to aggression, anger and violence. The findings for the non-OCD inpatients are in accord with what is generally reported in the literature (Apter et al. 1993b, 1995). The findings for the OCD patients seem to indicate that the correlates of suicidal behaviour in these adolescents are different. In fact, the only significant correlate with suicidal behaviour in the OCD subjects is the presence of antisocial behaviour (Apter et al. 2003).

Classic psychoanalytic theory predicted that the individual suffering from OCD would be protected from both depression and suicide. This appears to be incorrect as far as depression is concerned, but may be true with regard to suicidal behaviour. The mechanism that ‘protects’ the OCD patients from suicidal behaviour is unclear. It may be related to the ‘harm avoidance’ described by Nelson et al. (1996) or to some more specific defense or other mechanism.

There are no studies on suicidal behaviour in OCD and the treatment should be based on the same considerations as those for depressive disorders, with CBT probably being the most efficacious psychotherapeutic method for the basic OCD.

Schizophrenia is a common psychiatric disorder of adolescence (hence the term dementia praecox). Because schizophrenia is a serious disorder with ominous prognosis and social stigma, some clinicians are hesitant to make this diagnosis, even when there is sufficient evidence to do so. This potentially denies the child and family access to appropriate treatment, knowledge about the disorder, and specialized support services.

The differentiation between schizophrenia, psychotic depression or mania and schizoaffective disorder is not always easy in adolescence and many conceptual and nosological issues remain to be decided. The patient must then be followed longitudinally, with periodic diagnostic reassessments, to ensure accuracy. Patients and families should be educated about these diagnostic issues.

The depression in schizophrenia may be related to the fact that the young person feels that they are falling apart and becoming mentally ill and there is indeed evidence that suicidality and depression in these patients is related to good premorbid function, better insight, higher intelligence and preservation of cognitive function. Post-psychotic depression, and depressive states due to neuroleptic medications may also have a role to play in this dangerous condition.

Many schizophrenic patients are depressed and suicidal especially when they are young and have not been ill for a long time.

At least two-thirds of the suicides seen in persons with schizophrenia are related to depression, and only a small minority to the psychotic symptoms such as command hallucinations. The suicide is often shortly after discharge and thus may be related to lack of social support.

Finally, many adolescents with schizophrenia also abuse drugs and alcohol, thus increasing their risk for suicide. Sometimes the abuse is an attempt at self-medication. Anti-cholinergic medications given for the relief of extra-pyramidal symptoms (EPS) often give some adolescent patients a high to which they become addicted and those patients may simulate EPS in order to obtain these drugs. Child and adolescent onset of schizophrenia are often preceded by difficulties of attention and learning for which stimulant medications are given. Again, in the context of a developing schizophrenic condition there is a potential for abuse and drug-induced depression.

Recently Schwartz et al. (2006) found that depression can occur in some but not all adolescent patients with schizophrenia in the weeks following an acute psychotic episode. This depression can be quite severe and is associated with suicidal risk and actual suicidal behaviours. The schizophrenia-related depression in general shares some common features with the depression seen in major depressive episode (MDE) although in most cases it appears to be quantitatively less severe. In general, it seems that depression in schizophrenia can be distinguished from negative symptoms of schizophrenia. Finally, it appears that depression, hopelessness, and suicidality in schizophrenia are strongly related to the degree of insight and may be related to youth's sense that they are developing a mental illness. In fact, there is evidence that suicidality and depression in these patients is related to good premorbid function, better insight, higher intelligence and preservation of cognitive functions (De Hert and Peuskins 2000). Paradoxically, it is the patients with better insight and thus probably with better prognosis who are more likely to be depressed and suicidal (Schwartz et al. 2006). Post-psychotic depression and neuroleptic medications may also have a role in this dangerous condition. Suicide among this population often occurs shortly after discharge, and thus may be related to lack of social support. About 10–15 per cent of patients suffering from schizophrenia eventually commit suicide.

Adolescents with Psychoactive Substance Abuse Disorder (PSUD), especially males, are more likely to commit suicide with guns than are adolescents without PSUD. Adolescent suicide also seems to be related to more chronic PSUD in subjects who have not sought treatment.

For any age group, acute intoxication often precedes suicide attempts. Intoxication for the purpose of self-medication of anxiety and despondency, which often follows a crisis, may trigger suicide in an adolescent who feels shame, humiliation or frustration. It has been suggested that adolescents may use psychoactive substances to bolster their courage to carry out the suicide attempt or suicide. Intoxication may also lead to impaired judgement and decreased inhibition and thus may facilitate suicidal behaviour.

This topic of adolescent suicide and drug abuse has been extensively reviewed by Kaminer (1996). Many studies have reported an elevated suicide-risk ratio for adults diagnosed with PSUD (Kaminer 1996).

Although there has not been extensive research on this subject in adolescence it is well-known that conduct disorders and mood disorders are frequently comorbid with both substance abuse and suicidal behaviour.

The relationship between suicide, aggression and alcoholism may be especially relevant to people with type 2 alcoholism. These people are characterized by high novelty-seeking, low harm avoidance, and low reward dependence, and there alcoholism has an early onset and is characterized by a rapid course, severe psychiatric symptoms, fighting, arrests, poor prognosis and multiple suicide attempts. Surveys have found that suicidal thoughts were experienced by more than 25 per cent of college students aged 16–19 (Kaminer 1996). This supports a general non-specificity for adolescent suicidal thoughts. However, students with PSUD had more frequent and more severe suicidal thoughts than average, and also were more likely to have a prolonged desire to be dead. PSUD was also found to be associated with more severe medical seriousness of actual suicide attempts.

Studies of completed suicide in adolescents have shown that in Scandinavia, Canada and the USA, PSUD is more common among victims than in the general adolescent population. There is some evidence that alcohol and cocaine may be especially dangerous with regard to suicide, but this is yet to be validated (Kaminer 1996).

Impulsive rather than planned suicides by adolescents have been reported in large numbers. Many adolescents manifest suicidal behaviour after an acute crisis such as perceived rejection or interpersonal conflict, an acute disciplinary act, sexual assault, or immediate loss. Intoxication for the purpose of self-medication, which often follows a crisis, may trigger suicide in an adolescent who feels shame, humiliation or frustration.

There has also been recent recognition of the very definite increased risk for suicide in girls with eating disorders (Apter et al. 1995). The relationship between anorexia nervosa (AN) and depression is well documented. However, the suicide potential of these adolescents has been neglected in the literature, perhaps because these youngsters use denial to a large extent and because it was felt that starvation was a suicidal equivalent, obviating the need for a direct self-attack in these patients. Recently, however, it has been pointed out that suicide is not rare in AN and suicidal behaviour may be an important indication of poor prognosis for AN. Patton (1988) followed up 460 patients with eating disorders and found that the increased standard mortality rate in anorectic patients was mostly due to suicide, with death occurring up to 8 years after the initial assessment. These findings were similar to those of the Copenhagen Anorexia Follow-up study. Projective testing of anorectic patients also shows a preponderance of suicidal indicators (Gordon et al. 1984). Standard mortality ratio for death and suicide were 9/6 and 58.1, respectively (Herzog et al. 1994).

One can speculate on this association between depression, suicide and eating disorders. It is possible that for many girls weight loss is a form of self-medication for depression since in fact many healthy women feel much better when they lose weight. These good feelings are related in many ways to social approval, but may also result from the release of endorphins from damaged muscle tissue or from vomiting. In some cases we have seen depression resulting from weight gain as if the patient was suffering from withdrawal symptoms from her addiction to thinness. However, weight loss may of itself induce quite severe depression and suicidal ideation, even in volunteers and in normal dieters. Another very dangerous form of depression occurs in treatment-resistant cases where the constant battle against gaining weight on the one hand, and the constant social pressure to gain weight on the other becomes an intolerable burden. The diary of Ellen West, a famous anorectic patient who eventually killed herself, contained the following passages:

The most horrible thing about my life is that it is filled with continuous fear. Fear of eating but also fear of hunger and fear of fear itself. Only death can liberate me from this dread.

Since I am doing everything from the point of view of whether it makes me thin or fat, all things lose their real value. It has fallen over me like a beast and I am helpless against it.

Adolescents with bulimia nervosa are also highly prone to suicidal behaviours as part of an impulsive and unstable life style. Many show self-mutilation and cutting but often they also make serious suicide attempts which sometimes end in suicide. In our own series of former bulimic adolescent inpatients about 3.5 per cent died from suicide in a 15-year-long follow-up. This was about 300 times higher than the risk for other former female adolescent psychiatric inpatients. Recently, the term multi-impulsive bulimia has been coined to describe the increasingly more common association between bulimia, borderline or unstable personality disorder, substance abuse, depression and conduct disorder. Most patients with this comorbid constellation of disorders are women showing high risk for replication of suicide attempts and for fatal suicide.

Eating disorder patients who are suicidal should probably be hospitalized in most cases since the risk for completed suicide is great. Most will refuse treatment and legal commitment may be necessary. Many of these patients, however, are extremely resistant to therapy. In the chronic resistant patient the mainstays of usual treatment such as family therapy or cognitive behavioural therapy may be ineffectual. In the suicidal multi-impulsive bulimic individual there may be a place for dialectic behaviour therapy (DBT) (Linehan 1993). In anorectic patients weight gain and refeeding are essential and in many individuals this will restore mood and alleviate suicidal pain. In others, weight gain can cause tremendous disappointment and this may be an incentive for suicide.

Although there was an initial enthusiasm for the SSRI treatment of bulimia, recent results have been disappointing and in the suicidal patient with extensive psychiatric comorbidity these medications are usually not effective, although most clinicians usually give them a try. Antidepressant medications are usually unhelpful in the underweight depressed, suicidal anorectic, but they are of use in preventing relapse after weight gain has been restored. Our own experience (uncontrolled without any evidence base) is that long-term open-ended supportive therapy is usually the best way to keep these patients alive in the long run, and where a therapeutic relationship is achieved suicide can be obviated although the eating disorder usually persists to a greater or lesser degree.

Suicidal behaviour appears in many if not all psychiatric conditions of adolescence. Brent et al. (1993) found the following odds ratios (OR) for adolescent (completed) suicide risk factors: major depression (OR = 27); bipolar disorder (OR = 9); psychoactive substance use disorder (OR = 8.5) and conduct disorder (OR = 6).

It appears that suicidality cuts across nosological boundaries and thus may be regarded as an independent psychopathological dimension. This is supported by the work of Van Praag (1997) who found that biological markers for suicide were far more robust than biological markers for DSM-related nosological entities. In addition it may be that any serious debilitating mental illness leads to demoralization and hopelessness with resulting mental pain (Orbach et al. 2003). In any event, psychiatric illness remains the most well-defined risk factor for adolescent suicide and suicidality should be assessed in every young person suffering from a serious mental illness.