56 Cognitive treatment of suicidal adults

The progress of cognitive psychotherapy is accounted for by a systematic use of phenomenology, theory, laboratory research, and clinical studies. Effect studies of problem-solving, interpersonal therapy (which has many traits in common with cognitive psychotherapy), treatment of depression for suicide prevention and a cognitive psychotherapy method especially for treatment of suicide attempters are reviewed. The use of metaphors opens new possibilities. Step by step the researchers approach the suicidal individual's own formulations about their suicidality, developing the language of suicidality. Generally increased problem-solving capacity, more intensive outreach activities, an invitation to the patient to participate more actively in the analysis of their own problems and efforts to ameliorate the feelings of shame and guilt are all efforts to deal with painful interpersonal problems. A growing amount of evidence links these cognitions, emotions and behaviours to attachment problems in early life.

The suicidal person is slowly emerging out of the realm of taboo and the strong prejudices against open communication in which they were trapped for centuries. The suicidal person is emerging as a person in their own right, speaking independently and subsequently leading to genuine understanding of the suicidal process. As part of this progress, cognitive psychotherapy has contributed with the objective of recognizing the patient and the therapist as two researchers of equal value trying to solve a pressing problem, important for the survival of the client. One of the prime reasons for the rapid progress and excellent treatment results in cognitive psychotherapy is the use of the normalcy theory of emotion and emotional problems (rather than a theory of pathology). Like anxiety and despondency, suicidal ideation is considered a normal phenomenon with survival value, as well as the potential risk for deterioration into destructive pathology. Such a normalization makes it easier for the suicidal person to accept and talk about suicidal thoughts. Furthermore, the advancement of cognitive psychotherapy is due in part to the systematic use of phenomenology, theory, laboratory research, and outcome studies designed to develop more effective treatments. The aim is to create more specific models in the context of Empirically Grounded Clinical Interventions (EGCI), an example of the scientist practitioner model (Salkovskis 2002). The strategy involves detailed clinical observation, theoretically driven and clinically relevant experimental studies and attention to treatment outcome. The model has been formulated as a way of understanding the rapid evolution of cognitive psychotherapy. However, it can fruitfully be used for the purpose of understanding the clinical process of cognitive therapy through collaborative empiricism, which implies a close collaboration between the clinician and the patient, setting up goals, evaluating outcomes, and adjusting their strategies together with regard to whether the patient is improving as expected or desired. It has been a challenge to use these principles in order to understand, develop and evaluate treatment strategies within the broad and complex area of suicidal behaviour. The outline provided in this chapter will show the long journey of developing treatment strategies, filled with disappointments along with several promising results.

In 1998, Hawton et al. conducted a meta-analysis of twenty randomized controlled trials (RCTs). They reviewed the effectiveness of psychosocial and drug treatments of a total of 2452 patients who had deliberately harmed themselves (self-poisoning or self-injury) divided into 10 groups of treatment strategies. The inclusion criteria were self-poisoning or self-injury shortly before the trial; participants were randomly assigned to treatment and control groups, with repeat of deliberate self-harm as the primary outcome criterion. Most of the trials used standard care as the comparison. The results were generally disappointing. The synthesis of results from the meta-analysis revealed no clinically significant improvements, either in the general sample or in the specific subgroups. The authors concluded that there was insufficient evidence on which to base firm recommendations about the most effective forms of treatment for patients who deliberately harmed themselves. Some promising results were, however, found for problem-solving therapy (Hawton et al. 1987; Salkovskis et al. 1990; McLeavey et al. 1994), depot flupenthixol (Montgomery et al. 1979), dialectic behaviour therapy (DBT) (Linehan et al. 1991; Linehan 1993) and for the provision of an emergency contact card in addition to standard care. There was no evidence that antidepressants were generally effective for these patients. The authors recommended that researchers design studies in which the statistical power is assessed in advance, probably leading to larger samples, and that they maximize the treatment effects through assertive reminders for poorly compliant patients.

It is a well-known fact that suicidal people have a lot of problems or, seen from another angle, that they are psychosocially deprived. It is self evident that an increase in problem-solving capacity may first reduce the burden of unsolved problems and, second, reduce suicidal feelings. An experimental study (Schotte and Clum 1982) supported this idea theoretically. When comparing university students with and without suicidal ideation, the authors found that during life-stress, poor problem-solving subjects had significantly higher suicidal intent than other students. There is also considerable consensus regarding how to teach patients such strategies (Hawton et al. 1989; Heard 2000; Salkovskis 2001; Hawton and James 2005; Reinecke 2006). The patient can usually be motivated for the treatment when they understand how their low problem-solving capacity causes repeated failures to reach subjectively desirable goals, thus causing disappointment and psychological distress. A strategy involving a reduction of problems into small manageable steps is designed through problem listing, prioritizing the most important problem, brainstorming on possible solutions, selection of the most appropriate way to tackle the problems, identifying the first steps as well as psychological obstacles and monitoring the whole process. Every difficulty met by the patient during the process is reframed as a learning opportunity. The patient is also trained in generalizing their advances to new situations. This procedure thus introduces rational thinking to the client, helping them to separate themself from the problems. However, Williams (2001) noted that ‘unsurprisingly’, patients with less severe problems tend to show the largest response to problem-solving therapy.

In a new meta-analysis of five of the studies with problem-solving strategies in the study mentioned above, and one new study (Townsend et al. 2001), it was demonstrated that even if a group analysis of reduced repetition rate did not show significant improvement, the secondary outcome criteria significantly improved, in comparison with treatment as usual. They found improvements not only of their problems (OR = 2.31; 95 per cent CI 1.29 to 4.13) but also of their scores for depression (standardized mean difference = −0.36; 95 per cent CI −0.61 to −0.11) and hopelessness (weighted mean difference = −3.2; 95 per cent CI −4.0 to −2.41). Salkovskis et al. (1990) also found evidence of an effect on the rates of repetition over the 6 months after treatment. The relationship between problem-solving deficits and suicidality appear to be complex. Experimental studies have further elucidated this phenomenon. The major burden of problems is attributed to personal conflicts. The skills to solve such problems can be measured by the Means–Ends Problem-Solving Procedure (MEPS) (Platt et al. 1975). Using this method, Mitchell and Madigan (1984) were able to show that people who were depressed scored lower than those who were not depressed. Schotte et al. (1990) used the same method in a short-term, longitudinal study of 36 patients hospitalized for ‘suicide observation’, 39 per cent of whom were admitted for a suicide attempt and 22 per cent for reported past attempts. The researchers found marked improvements within one week in depressive symptoms: the number of patients with Beck Depression Inventory (BDI) scores over 21 decreased from 78 to 11 per cent, and also in states of anxiety and hopelessness. The changes were associated with improvements in interpersonal problem-solving skills. Such deficits thus seem to be concomitant with other symptoms (states of vulnerability) rather than a relative stable cause of depression, hopelessness and suicide intent (vulnerability traits).

Williams et al. (2005) also used the MEPS instrument in a study of three groups of participants (N = 34); one with no history of depression, one with previous history of depression without suicidality, and one with a previous depression combined with suicidal ideation and behaviour. Low mood was brought about among the participants through sad music and the reading of sentences with sad content. In the last group, the suicidal thoughts of the patients were activated and their capacity to solve interpersonal problems decreased. These results support the hypothesis of low problem-solving deficit as a state as opposed to a trait phenomenon. It also points out that suicidal ideas may be present without concomitant depression.

In a European multi-centre study (N = 386 medically treated deliberately self-harming [DSH] patients) McAuliffe et al. (2006) found that passivity and avoidance of problems (coupled with low self-esteem) was the most frequent factor associated with repetition of deliberate self-harm. They concluded that intensive therapeutic input and follow-up are required.

An advantage of problem-solving therapy for suicidal patients is that it is brief, and therefore cheap. The same idea, along with a broader foundation in cognitive theory leading to an multitude of intervention strategies, was used in the PROMACT study, a multi-centre study of 480 patients with recurrent episodes of deliberate self-harm but with no substance use disorder (Tyrer et al. 2003a, b). The patients were randomly assigned to a Manual-Assisted Cognitive Behavioural Therapy (MACT), or to treatment as usual. The patients were sent a 70-page booklet by mail and offered up to 7 sessions of cognitive behaviour therapy. There were no differences in baseline characteristics. The follow-up period was 12 months. The results from the above study were also disappointing. There were no significant differences in either repetition rates of deliberate self-harm, or on secondary outcome criteria such as clinical diagnosis, risk of parasuicide, self-rated anxiety, depressive symptoms or other parameters. The figures however were more favourable at 12 months than at baseline in both groups. A cost-effectiveness evaluation indicated the superiority of MACT over treatment as usual (Byford et al. 2003). The negative results in the previous study may be due to several weaknesses that will be discussed below (Hawton and Sinclair 2003; Arensman et al. 2004 including a reply by Tyrer et al. 2004).

First, patients who did not show up to the first appointment for treatment were offered only one second appointment with no further contact or home visit. Several other studies of patients with deliberate self-harm showing significant differences between experimental and control groups have used a more active approach to the patients (Salkovskis 1990; Guthrie et al. 2001; Brown et al. 2005). This is also true for a few studies with home visits by a nurse or mental health worker but without psychotherapy (Welu 1977; van Heeringen 1995). Second, the internal drop-out from the therapy was high. Only 60 per cent of the MACT group used both the 70-page booklet and the sessions, 2 per cent did not use the booklet and 38 per cent did not show up at all for the sessions. Third, the group differences were reduced as the MACT therapists were not well-trained cognitive therapists but ordinary hospital staff with short training. On the other hand, the treatment as usual included psychological treatment in various degrees, such as problem-solving approaches and dynamic psychotherapy.

Interpersonal problems may be perceived as a major cause of suicidal problems. As a treatment tradition especially focusing on such problems, interpersonal therapy was primarily used in the treatment of depression and focused on the relationship between the onset of depressive symptoms and the patients' current interpersonal problems (Klerman and Weissman 1989). Such manual-based, focused and active interventions are similar to cognitive therapy, which is clearly demonstrated in the reference lists. It thus seems adequate to discuss it here.

A randomized controlled study (Guthrie et al. 2001) offering interpersonal therapy to patients after deliberate self-poisoning (N = 119) yielded interesting results. The therapy in the intervention group (N = 58) was based on a model by Hobson (1985), and manualized by Shapiro and Startup (1990). The therapy was offered within one week, delivered by nurse therapists in the patient's home, and proceeded for a four week duration. Its focus was to identify and help resolve interpersonal difficulties which caused or exacerbated psychological distress. The control group (N = 61) was offered treatment as usual mostly, by their GPs, that did not include psychological therapy. The two groups were similar in terms of baseline characteristics with the exception of marital status. At the 6 months follow-up the self-reported subsequent episodes of self-harm were lower in the intervention group (9 vs 28 per cent, p = 0.009), and the differences in scores on the Beck scale for suicidal ideation were significantly lower, after a correction for marital status (p = 0.027). The patients' satisfaction with the therapy was significantly greater in the intervention group. The good results were obtained in spite of the fact that only 35 patients participated in all 4 sessions, and 50 patients in more than 2 sessions.

The strengths of the study discussed by the authors are the RCT design and the use of a method that previously has shown good results in treatment of other disorders (Guthrie et al. 1991; Shapiro et al. 1995). The good results may partly be due to the rapid intervention in the participants' home, the focus on interpersonal problems eliciting psychological distress, and the clear differences between intervention and control group. The limitation was that the rate of self-harm at follow-up was based on self-reports, data which is usually weaker than hospital admissions. On the other hand the broader range of self-harm in self-reports includes events that are often neglected in other studies.

The dominant hypothesis in psychiatry concerning the origin of suicidality starts from the fact that mental disorders, especially depression, are important risk factors. In the classification of psychiatric disorders, the DSM-IV (American Psychiatric Association 1994), suicidal ideation and acts are presented as symptoms of depression. It is a clinical and scientific fact that suicidal thoughts often vanish after effective treatment of a depressive episode. Adequate treatment of depressive disorders is generally accepted as an effective method for suicide prevention.

Aaron Beck observed the importance of cognitions in the development of depression. He enriched the understanding of the field with many new mediating factors. Dysfunctional thinking, especially always interpreting experiences in a negative way (habitual negative thinking) may lead to a de-evaluation of the self and feelings of helplessness, hopelessness and perception of a black future. Other ‘thinking errors’ are for instance the claim to understand other persons' thoughts without asking them (mind-reading), taking everything too personally (personalization) and overgeneralizing the importance of negative, subjective experiences (Beck 1967; Clark et al. 1999). These factors may start vicious spirals leading to more and more severe depression. In experiments using a number of new scales it was possible to demonstrate, quantitatively, the interaction of such factors in the development of depression, and their importance for creating and maintaining suicidal thoughts and plans. Based on these and other assumptions, a wealth of cognitive techniques have been developed for the treatment of depression. In numerous studies those methods have been shown to be as effective as drugs in the treatment of depression and anxiety disorders. Cognitive psychotherapy is therefore now accepted as a powerful treatment tool.

Maladaptive thinking (dysfunctional assumptions and attitudes) was typically thought to be elicited by very persistent dysfunctional schemas, i.e. dysfunctional meaning-making structures of cognition, assumed to have a neuronal basis but modified by personal experiences. They could be activated by even small internal and external stimuli and then accepted as true reality. In the therapy situation they could be disclosed by a close observation of the automatic thoughts they give rise to. The dysfunctional assumptions and attitudes could be measured by the Dysfunctional Attitude Scale (Weissman and Beck 1978). If such attitudes were of importance for the relapse of depression, they would be persistent during healthy phases as well. However, Ingram et al. (1998), analysing many such studies, found that this was not the case. The hypothesis that persistent cognitive structures elicit mood changes through dysfunctional coping with external stimuli was then replaced by the counter-hypothesis that even small moments of sad mood could elicit memory biases in the direction of remembering more negative and less positive events. Dysfunctional thinking styles thus existed only under the influence of a sad mood. Nevertheless, they could result in vicious circles leading to a relapse into depression. This ‘differential activation hypothesis’ (Teasdale 1988) was supported in studies of experimentally induced sad mood. These findings highlight the importance of even brief moments of sad feelings for eliciting depression and suicidality. As a consequence, the training of attention or ‘mindfulness’ in order to understand and cope with moments of sad mood and negative memory bias, was shown to be useful in reducing the relapse frequency in depression. Such awareness could break the development toward rapidly recurring depressions elicited by even lesser strains (Williams et al. 2000; Segal et al. 2002).

Cognitive psychotherapy involves supporting the client with new concepts, new knowledge and new techniques, adapted to their actual personal needs, and immediately useful in their efforts to change their behaviour. Recently, a manual, specifically for suicide attempters, has been developed (Brown et al. 2002, 2006, Henriques et al. 2003; Berk et al. 2004) and tested in a randomized controlled trial (Brown et al. 2005). Ten sessions of cognitive therapy were given to 60 patients who hade attempted suicide and whose suicidal intention still persisted, according to the Suicidal Intent Scale. The intervention group was compared with 60 patients in a control group offered treatment as usual. There were no significant differences in demographic and psychiatric baseline variables. Among the participants 77 per cent had major depressive disorder, 68 per cent substance use disorder, and 85 per cent more than one diagnosis. The cumulative attrition rate at the 18 months follow-up was 25 per cent for the cognitive therapy group and 34 per cent for the treatment as usual group (p = 0.045). The therapists' adherence to the manual was controlled, and divergences were discussed.

The primary aim of the trial, was to decrease the risk of recurrent suicidal acts. Thoughts, images and basic assumptions, activated immediately before the attempt, were identified. Specific strategies to deal with such manifestations were worked out. Attention was also directed to coping with stressful situations and more general vulnerability factors, such as feelings of hopelessness, poor impulse control, low problem-solving capacity, non-compliance during the treatment and social isolation. Towards the end of treatment patients were helped to devise strategies to deal with acute suicidal ideation and intent. After 18 months, 13 (24.1 per cent) in the cognitive group, and 23 (41.6 per cent) in the control group made at least one more suicide attempt (p = 0.49). The probability of repeated suicide attempt was 50 per cent lower in the cognitive group (hazard ratio 0.51; 95 per cent CI, 0.26–0.997). Some secondary measures, the severity of self-reported depression and of hopelessness, were also reduced. Interestingly, suicidal ideation was not significantly reduced.

The strengths of this well-designed and conducted study is its specificity: targeting the suicidal mode as it expressed itself immediately before the act, the inclusion of patients with different psychiatric disorders, including a high percentage of substance use disorders, special study case managers following the patients and making contact when necessary, as well as the creation of action plans for avoiding further suicide attempts at the end of the therapy. The study showed that relatively brief interventions may give positive results, even if the result of the main criterion, a reduced frequency of recurrent attempts, was just on the border of significance. The therapeutic concentration on the acute suicidal episodes is the principal interesting aspect of the study above. Such episodes are usually time-limited, as it is often very time-limited. Thus there are important differences in the risk pattern between a baseline suicide risk and the acute episode (Brown et al. 2002; Berk et al. 2004; Brown et al. 2005; Rudd 2006), requiring a risk analysis in two steps. The suicidal episode is determined by strong rapid influences in a short time, accompanied by a loss of control. It seems likely that the probabilistic perspective is especially useful in terms of acute episodes, with its higher speed in the interaction between the person and the environment. In these respects, they are similar to accidents (Beskow 1983; Clarke and Lester 1989; Reason 2006).

It is popularly believed that suicidal ideation and acts are primarily a form of communication, a cry for help. On the contrary, Williams (2001) and Williams and Pollock (2000) suggested that suicidality primarily emerges from real suffering, eliciting a cry of pain, with communication only as a secondary effect. The inspiration for the model came from ethological research on the reactions of birds to losing a fight over territorial boundaries (Gilbert 1989). The defeated bird lowers its wings and head and retreats; it looks ‘depressed’ or ‘shamed’. Submission saves its life. When there is no place to identify a new territory, the ‘depression’ continues and the bird becomes easy prey to predators. In humans, theoretically, the same situation occurs when there is no chance for either flight or fight, and where submission is the only available reaction mode. The pain is caused by both feelings and images of being ‘defeated’, ‘closed in’ in a trap with ‘no way out’. It is then impossible to go on with life. These three components have been elaborated in theory and studied experimentally. Psychological measures and cognitive treatment strategies have been developed.

The feeling of being defeated can arise from both external and internal circumstances. Numerous situations in the family, at school and at work can both elicit and maintain feelings of worthlessness, uselessness and powerlessness, giving rise to feelings of being a loser. These feelings may, however, to a considerable extent, be a product of the person's own constructive processes (Ellis 2006; Neimeier and Winter 2006) and may be maintained by dysfunctional schemas, such as negative thinking. Moreover feelings of shame, humiliation and weakness act as a barrier to peers and family, leading to a downward spiral of increased loneliness and a sense of not belonging. Such feelings are also present in exhaustion reactions, with a wish to rest, to be alone and to be sheltered from further painful stimuli. Other types of distorted or biased thinking may add to the negative experience, resulting in feelings of extreme pain. Early in the process, where escape potential is threatened but not yet eliminated, escape attempts will be characterized by high levels of activity, anger and ‘protest’. Later in the process the person may feel totally closed in. These feelings may originate from a safety-seeking behaviour with avoidance of specific and concrete experiences, images and descriptions, that evoke strong feelings and a preference for interpreting them by use of general images and descriptions. It has been understood as the construction and use of over-generalized memories (OGM) (Williams and Broadbent 1986; Williams 2001; Williams and Pollock 2001). In these efforts the patients unfortunately also lose the positive feelings necessary for close human relations and for feelings of a meaningful life. Their life will be colourless, silent, numb and odourless. The patient says: ‘I am alone’, ‘I no longer feel alive’. These are the walls of the trap. The experience of a positive future is essential for the feeling of hope and the will to live. Asking patients and controls to describe events that might happen to them within one day, one week etc. up to ten years from now has shown that depressive/suicidal persons anticipate more difficulties, but especially fewer positive events in the future, and that these are correlated with a greater degree of hopelessness (MacLeod et al. 1993). A totally dark future informs us that there is no way out of the trap and the ongoing pain. The possibility of suicide may then be perceived as the least unattractive option.

The psychological model above has recently been tested (O'Connor 2003) in a cross-sectional comparison between 30 patients admitted for deliberate self-harm and 30 controls from the same emergency wards, matched for age, sex and marital status. The escape potential was a composite score of two items, assessing escapability and controllability of ‘the most stressful life event that they had experienced in the last six months’, and it was assessed on 5-point Likert scales. The defeat scale consisted of four items measuring the degree to which their most stressful recent life event led to feelings of defeat, rejection, loss and failure. The possibility of rescue was operationalized in terms of availability of social support, measured by an 18 item multidimensional instrument. The result gave support to ‘the cry for pain’ model. Furthermore, the feelings of entrapment in the parasuicidal group were associated with intrusive thinking more frequently than the controls, a characteristic trait in depression. The model can readily be used in psychotherapy and also be the starting point for further scientific development.

The normality perspective which has been so fruitful within cognitive psychotherapy implies that suicidal images may sometimes be functional, even life-saving. To accept suicide as a way out of the trap may ameliorate the pain and make further living possible.

‘During the assault I felt myself first dying and then dead. However, I was not afraid that they would kill me. At home one hour later I looked at myself in the mirror. There was only an empty shell looking at me, a living dead. In that moment I said: “I will fix this as long as I can. When (not if) I can't fix it any more, I will take my life.” I will never tell anyone.’

‘These thoughts were not the result of pondering the situation but emerged suddenly as a completed decision. These suicidal images however were only linked to carrying it through to a small degree. The thought of suicide was “my place of refuge”, comforting knowledge that there was a way out, a way I might take or escape. In the time immediately after, the suicidal images were “good” for me ….

I felt as if I would literally die of shame. I used the whole night in efforts to clean myself. Later the shame spread like a mist over my whole life. At first it lay there as a thick carpet but slowly it penetrated everything.’

Those lines summarize part of an interview with a 34-year-old woman made 14 years after she had been subject to violent sexual assaults (Vea 2006 and personal communication). She tells us about unexpected, sudden and strong emotions that threatened her mental life and about her intensive struggle to master them with rapid cognitions and firm decisions. She lost against her assailants and was consequently overwhelmed by shame and captured in a cage with no way out.

Interestingly, the idea of taking her life emerged immediately and spontaneously. It functioned as a refuge from the intolerable pain, thus underlining that suicidality sometimes may be an adaptive cognitive structure emerging instantly and that depressive symptoms may come later. Williams et al. (2006) have shown that in recurrent depressive disorder, suicidal ideas have a much higher consistency compared to the broad variations of other depressive symptoms. They are consistent with the relative ease of achieving changes in symptoms of depression measured by traditional scales compared to changes in cognitive functioning and self-esteem (Grawe 2007). The importance of traumatic experiences for the development of suicidality is now increasingly acknowledged (Meichenbaum 2006). It seems as if the first suicidal episode is often elicited by strong psychosocial strain, but that later episodes are much more easily elicited owing to an increase in vulnerability, just as is the case in depression.

Grasping the rapid images, thoughts, and behaviours of the patient requires an open collaboration on equal terms. However, this type of therapeutic alliance is still not in use everywhere, as noted by a working group of distinguished experts (Michel et al. 2002).

‘The working group agreed that current mental health practice often does not take into account the subjective experience of patients at-tempting suicide, and that contemporary clinical assessments of suicidal behaviour are more clinician-centred than patient-centred. The group concluded that clinicians should strive for a shared understanding of the patient's suicidality, and that interviewers should be more aware of the suicidal patient's inner experience of mental pain and loss of self-respect.’

Michel et al. (2002, p. 424)

In order to do that the therapist must be prepared to receive and keep the pain of the patient and have a network for their own support.

The trap model summarizes the deep sense many suicidal people have of living in ultimate isolation or rejection. These feelings may be influenced by early acquired schemas. Bowlby's attachment theory maintains that dysfunctional inner ‘working models’ of the self and the environment develop as a function of the experiences the child has in their family, and may be the result of rejection by parental attitudes (Bowlby 1980). The importance of attachment in borderline states has been further developed by for example Fonagy et al. (1995) and in suicidal behaviour by Adam and colleagues (1996).

Studying therapy-refractory affective disorders with retrospective life-charts, Ehnvall et al. (2005, 2008) investigated if perception of themselves as being rejected/neglected by either parent in their childhood influenced the respective patients' depressive disorder and the number of suicide attempts. The subjects who felt unwanted had significantly more days of depressive illness, higher percentage of total days in illness over their lifetime and a higher number of illness days per episode. Female patients had a twofold greater chance of making at least one lifetime suicide attempt. Females reporting higher levels of rejection/neglect reported a greater number of lifetime suicide attempts. These findings could not be explained by mood-congruent recall.

From the adult perspective, there is a growing interest in how suicidal people express themselves (Beskow et al. 2005). This interest has led to observations of vicious circles between criticism, e.g. shame and guilt, and consolation, e.g. too much eating and drinking (Firestone 2006). Rudd et al. (2001) and Rudd (2006) noted these vicious cycles, and also demonstrated that some schemas are specific to suicidality and quite different from those routinely associated with depression, such as: ‘I'm worthless and don't deserve to live’ expressing the core belief of unlovability; ‘I can't fix this problem and should just die’ (helplessness); ‘I'd rather die than feel this way’ (poor distress tolerance) and ‘everyone would be better off if I were dead’ (perceived burdenness). A cascade of such themes can rapidly elicit a life-threatening situation for the patient. These verbal expressions may be seen as expressions of painfully frustrated basic needs of attachment, pleasure/avoidance of pain, self-enhancement and orientation/control (Grawe 2007). Probably these needs will best be provided for by a positive approach, opening the window for positive psychotherapy (Wingate et al. 2006).

During the development of cognitive psychotherapy, the understanding of suicidality has changed profoundly from primarily a symptom of depression (a pathological perspective) to the result of a complex interaction of many cognitive, emotional and behavioural structures (a normal and systemic perspective). The suicidal process may be seen as a cognitive process of its own, separated from, but interacting with, depressive processes. All those structures interact in a very complex way on different levels of personality organization. They may primarily be functional coping strategies, such as avoidance of experienced harmful situations in personal relations, but later they may change toward dysfunction with increasing difficulties in solving interpersonal problems. The pain before a self-destructive act is often expressed in a vivid inner debate concerned with the questions of belonging or not belonging and the meaning of one's own life, often sandwiched in with an elaboration and evaluation of different methods of self-destruction.

Cognitive therapy, increased problem-solving capacity, outreach activities and an invitation to the patient to take a more participative role in the analysis of their own problems (features which have been successful for patients with other mental disorders, for example schizophrenia), as well as efforts to ameliorate the feelings of shame and guilt so essential in the ‘cry for pain’ model, are all efforts to deal with painful interpersonal problems and how they can be solved. A growing amount of evidence links these cognitions, emotions and behaviours to attachment problems in early life. Looking at suicidality from the point of view of frustrated, basic needs, the efforts to treat suicidal people may at this time be perceived as preliminary, identified parts of a support system, which eventually will become almost as differentiated and complex as suicidality itself.