Suicidal and violent or homicidal behaviour

About 30% of violent individuals have a history of self-destructive behaviour and 10–20% of suicidal persons have a history of violence to others (Reference Plutchik, van Praag, van Praag, Plutchik and ApterPlutchik & van Praag, 1990). Clinical studies have also confirmed in general a positive association between suicide or suicidal behaviour and violence to others (e.g, Reference Convit, Jaeger and LinConvit et al, 1988). Although suicidal and violent behaviour are both influenced by multiple factors, the common thread linking the two may be an underlying propensity for elevated impulsivity, affective lability and behavioural disinhibition (Reference Nock, Marzuk, Hawton and van HeeringenNock & Marzuk, 2000) and it is postulated that this could be mediated through impaired serotonin (5-HT) metabolism. A range of other social factors have also been recorded as being associated with both suicidal and violent behaviour, including a history of violence in the family, family conflict, access to weapons and male gender. Early Freudian psychoanalytic evaluation of the link between suicide and violence to others reflected a similar perspective: suicide was seen as aggressive and homicidal urges inwardly directed towards the self (Reference PerelbergPerelberg, 1999). Reduction in suicides during periods of war has also been noted.

The main mental disorders associated with increased risk of violence or homicide are schizophrenia, substance misuse and personality disorder. Suicide risk is elevated in all mental disorders except learning disability and dementia, but especially in depressive illness, schizophrenia, substance misuse and personality disorder (Reference Harris and BarracloughHarris & Barraclough, 1997). The potential, however, for depression or dysphoria to be associated with anger towards others should not be underestimated: people with depression do commit homicide (Reference MalmquistMalmquist, 1995).

Perhaps the most overt link between suicide and homicide is illustrated by the phenomenon of murder followed by suicide. Most commonly, the perpetrator is male and the victim female: a man killing his wife or lover and, less frequently, his children, with the context usually involving jealousy or threatened separation. Another type is that of an elderly male, declining in physical and mental health, who kills his wife and then himself, the motive here being more ‘altruistic’, the perception being that the killing relieves both parties of a burden. Occasionally, a perpetrator will kill several people outside of his family and then turn the gun on himself, although such mass homicides/suicide are more common in the USA than in Britain. The clinical literature also contains descriptions of patients in psychiatric hospitals who have killed other patients, having suicidal ideation either before or after the homicide, or both (e.g. Reference Modestin and BokerModestin & Boker, 1985).

Beyond the clinical sphere, mass suicides have occurred within some religious cults, although in a number of these cases, some of the cult members have been murdered or coerced into killing themselves. An example of contiguous homicides and suicide, where the deaths of the victims and that of the perpetrator may occur at the same time, is the suicide bomber, as illustrated by Japanese kamikaze pilots towards the end of the Second World War. More recently, this phenomenon has also been associated with Islamic fundamentalists in the Middle East and, on 11 September 2001, in the USA. Here, the act, as perceived by the perpetrators and their affiliates, is described as martyrdom. None the less, the elements of suicide are present, i.e. intent to die and with his/her death as the outcome, albeit along with the deaths of others (Reference LaqueurLaqueur, 2001). The phenomenon of suicide bombing may not, however, be construed as underpinned exclusively by religious nationalism. The Tamil Tigers in Sri Lanka have also used such methods extensively without the background of orthodox religious ideology.

Since the early 1990s, increasing public concern has been expressed in Britain regarding homicides by those with mental illness. Although the proportion of homicides committed by this group has not increased, and may actually have decreased (Reference Taylor and GunnTaylor & Gunn, 1999), public apprehension has not abated and, indeed, there may be an intrinsic fear of insanity engrained in the public psyche. It is worth noting, however, that the Royal College of Psychiatrists’ approach to unnatural deaths by psychiatric patients encompasses both suicide and homicide (Reference Amos, Appleby and ShawAmos et al, 1997). The National Confidential Inquiry into Suicide and Homicide by People with Mental Illness has recently published its 5-year report covering England and Wales, Scotland and Northern Ireland (Department of Health, 2001).

Suicides in locked wards in general psychiatry

Psychiatric intensive care units

Not all psychiatric intensive care units (PICUs) in general psychiatry are locked wards but they accept patients whose level of disturbance, involving aggression or self-harm, cannot be adequately contained on open wards. Patients in such units tend to show high levels of aggression to self and to others. A high turnover of patients is also a feature, and indeed necessary, to maintain their function. The main issues related to the risk of self-harm in a PICU are as follows (after (Reference Pereira, Lipsedge, Beer, Pereira and PatonPereira & Lipsedge 2001):

• Level of suicidal intent The cognitive element common to many suicides is probably ambivalence rather than undiluted intent. Many give warnings of their thoughts of killing themselves prior to doing so. The absence, however, of prior warnings does not eliminate the potential for suicide, as some patients may prefer not to disclose their intentions. In a few cases, patients experience auditory hallucinations commanding them to kill themselves and not to tell others about it.

• Previous history of deliberate self-harm A previous episode of deliberate self-harm is a major risk factor for completed suicide, but most patients with such a history eventually die of natural causes.

• Clinical risk factors Diagnoses of depressive illness, schizophrenia, substance misuse or personality disorder raise the risk of suicide, as does a positive family history of suicide.

• Malignant alienation Staff treating patients who show persistent challenging behaviour may find that, over time, their ability to maintain a therapeutic relationship is impaired. This may lead to distancing or avoidance of the patient, a form of negative countertransference (Reference Watts and MorganWatts & Morgan, 1994). Malignant alienation was not, however, found to be a risk factor for suicide in a later study (Reference Powell, Geddes and HawtonPowell et al, 2000).

• Demographic and social factors Risk factors for suicide include being male, of older age (but there have been increases in suicide in young men over the past few decades), divorce, bereavement and adverse employment status.

• Physical aspects of the unit Access to potential points from which a ligature could be attached, electrical wiring, glass from a breakable window or heights from which to jump should all be eliminated and exit from the PICU needs to be controlled.

Medium secure units

The primary role of medium secure units is the assessment and treatment of patients with mental disorders who pose a degree of risk to others beyond that which is safely manageable in a general psychiatric unit, but does not reach a level of risk which would necessitate location in high-security conditions, as pertain in a special hospital. Most patients in medium secure units are male and the commonest diagnosis is schizophrenia or a related psychosis. Most are referred from the criminal justice system, but a minority are transferred from general psychiatric units following difficulties of management, and some are transferred from special hospitals after a degree of improvement. Several studies have found an overrepresentation of male patients of African–Caribbean origin, the vast majority of whom are diagnosed as having a schizophrenic illness (Reference Mohan, Murray and TaylorMohan et al, 1997).

Few studies of medium secure units have referred to suicides within the unit or, indeed, to levels of deliberate self-harm, the main focus having been on violence to others. Non-fatal deliberate self-harm is usually more associated with female patients, although one study in a medium secure unit found that 45% of male patients had such a history, a level similar to that found in female patients (Reference White, Leggett and BeechWhite et al, 1999). On the other hand, in a study in a medium secure unit by Reference Torpy and HallTorpy & Hall (1993), only 19 incidents of self-harm were recorded during a 2-year period, compared with 563 acts of physical aggression and 257 of verbal aggression.

Only two studies on medium security have considered suicide in more depth. Reference HambridgeHambridge (1994), in a paper on the treatment of mentally disordered offenders who had been convicted of homicide, refers to the need always to recognise the potential for suicide in this group. Where such a potential exists, the pace of treatment must be consistent with that with which the patient can cope and, where necessary, restriction of freedom must be negotiated or imposed to prevent serious self-harm. Some patients, especially when gaining insight and where their victim was a close family member, may suffer guilt that triggers a suicidal crisis. Hambridge is also the only writer on suicide in medium security who notes the potentially protective role of harnessing religious beliefs in the minimisation of the risk of suicide. Rates of suicide are markedly less in societies with a predominantly Islamic faith, and some protection may also, to a lesser extent, be afforded by Catholicism.

In a survey of suicides in medium secure units in Britain, Reference JamesJames (1996) recorded that 9 out of 21 units approached reported a death by suicide, a total of 13 such deaths being noted. Of these, 11 were by hanging, 1 by self-strangulation and 1 by drowning. James noted that patients in medium secure units often have multiple risk factors for suicide although, in fact, suicide in medium security is rare. The most common diagnosis for the suicides was schizophrenia, followed by affective disorder.

The suicides often occurred soon after transfer from prison, but suicides also occurred while prisoners were awaiting transfer to a medium secure unit. Staff commented that at the time of the suicide, the patient's mental state had seemed to be stable, such that suicide was unexpected. James considered that a balance is necessary between the need for restrictions to limit suicide risk and the need to prevent undue compromise in the overall quality of life for patients in a unit. Effective clinical assessment, a high staff-to-patient ratio, curtailment of materials that could be used as ligatures and environmental adjustments to remove hazards that could be used to suspend a ligature must all be effectively addressed. Constant supportive observation should be available, when appropriate, with intermittent extra observation when the degree of risk is still raised but is declining. Taking account of the highly challenging nature of some patients in medium secure units, malignant alienation is also relevant, a point reinforced by Reference WhittleWhittle (1997). Reference JamesJames (1996) rightly, therefore, emphasises the need for all staff, especially nursing staff, to have multi-disciplinary support in the implementation of care plans, as it is they who are with patients on an ongoing basis and who may feel most vulnerable in the event of an unexpected death.

Suicide in special hospitals

Although there is an overlap in the risk posed by patients in medium security compared with those in high security, the most dangerous patients are undoubtedly to be found in the special hospitals. These are mostly longer-established institutions with larger numbers of patients, most of whom have schizophrenia, but with sizeable minorities with personality disorder or learning disability. Comorbidity with a history of substance misuse is common. The male:female ratio in special hospitals is about 5:1, most wards being for either male or female patients. Risk factors for suicide in patients in special hospitals are probably even more pronounced than in medium secure units (Box 5).

My own unpublished research has indicated that between 1864, soon after Broadmoor opened, and 1933, i.e. the first 70 years of that hospital's history, there were only 21 suicides. In the period from 1934 to 2000, i.e. the next 66 years, there were 81 suicides, the peak decade being 1964–1973, when there were 23. Further research is required to determine not only the absolute numbers of suicides but also the suicide rates in the earlier and later periods. The overall death rate from all causes in Broadmoor in its early years is recorded as having been much lower than in general psychiatric hospitals, its suicide rate having been particularly low. It is notable that the alteration from a low rate of suicide in Broadmoor to a higher rate can be dated from the 1930s, a time when effective treatments for mental illness such as electroconvulsive therapy (ECT) and, subsequently, antidepressants, antipsychotics and psychological therapy, became available. This seeming paradox requires further historical research as the usual assumption is that treatment reduces the risk of suicide. None the less, the reality seems to have been that in Broadmoor, the oldest special hospital in England and Wales, the number of suicides was very much lower at a time when the only treatment available was that of the milieu itself.

More contemporary studies of suicide and deliberate self-harm in special hospitals have been undertaken, but as with medium security the issue has to an extent been eclipsed by the main focus on violence towards others.

Reference Swinton and HopkinsSwinton & Hopkins (1996) found a correlation in Ashworth Hospital between violence and self-injury in female patients with personality disorder but not in those with mental illness. Although not in itself a factor that determines admission to a special hospital, self-harming behaviour was seen by clinicians as an obstacle to transfer to conditions of lower security (Reference Maden, Curle and MeuxMaden et al, 1995). Absconding or escapes from special hospitals are rare, but in various studies absconding has not been associated with suicide in any patient, except in one case later in the year after return (Reference MooreMoore, 2000), where the absconding and suicide were seen as having a common element of psychological avoidance. My colleagues and I (Reference Gordon, Oyebode and MinneGordon et al, 1997) have studied factors pertaining to relationships between homicide or violent behaviour and suicide in special hospitals. Further unpublished enquiries carried out with colleagues found that, in the 30-year period between 1966 and 1995 inclusive, there were just under 200 deaths from all causes in Broadmoor, of which about 30% were by suicide or probable suicide. Almost all of the suicides were by hanging or, less commonly, self-strangulation.

Occasionally, clusters of suicides have been reported, as in general psychiatric hospitals, but the reasons for their starting, or indeed stopping, are unclear. However, the role of imitation or ‘contagion’ needs to be considered, especially if more than one suicide has occurred within a short period.

Suicides in prison

Several studies have shown high rates of mental disorder in both remand and sentenced prisoners (e.g. Reference Gunn, Maden and SwintonGunn et al, 1991). The main mental disorders in prisoners are personality disorder and substance misuse, as well as elevated rates of schizophrenia and related psychoses, all of which are associated with increased risk of suicide. There is also an excess of people with adjustment reactions, especially among young males in prison.

The risk of suicide in prison has been known for many years. However, a range of more recent studies has seemed to indicate a significant excess of suicides in prison compared with the general population (e.g. !Reference Towl, Towl, McHugh and JonesTowl, 1999). However, taking account of prisoners’ high rates of mental disorder and substance misuse, including injectable opioids, it is not clear whether rates of suicide in prison are higher than in matched population samples in the community (Reference GoreGore, 1999).

Box 6 shows the factors associated with suicides in prison. The National Confidential Inquiry into Suicide and Homicide by People with Mental Illness (Department of Health, 2001) found that 1% of all suicides in England and Wales were by prisoners who had been under National Health Service (NHS) mental health care during the previous 12 months.

A report on suicides in prisons in England and Wales by Her Majesty's Chief Inspector of Prisons (1999) noted that the rate of self-inflicted deaths in prisons more than doubled between 1982 and 1998, the rise being proportionately larger than that of the overall prison population during that period. However, taking account of the range of risk factors for suicide which are present in the prison population, it is not entirely clear whether the number of suicides in prison is in excess of that in a matched sample of the general population (Royal College of Psychiatrists, 2002). However, this is not to underplay the need for better prevention of suicide in prison. Both the Chief Inspector of Prisons’ report (1999) and the National Confidential Inquiry report (Department of Health, 2001) also note that offenders with mental illness are still, at times, being sent to prison rather than to hospital, which would be more appropriate.

Suicide in other mentally disordered offenders

Concern has been expressed about the risk of suicide by mentally disordered offenders on probation in the community (Reference Harding and CameronHarding & Cameron, 1999). Probation officers have expressed the view that it would be more helpful to them if psychiatrists adopted a broader definition of mental disorder beyond that solely of mental illness, thus helping with people with personality disorder, substance misuse and sexual disorders.

Suicide is also the most likely cause of death in those in police custody; the majority of such deaths are by hanging, although some die from undetected drug overdose prior to having been taken into custody (Reference Norfolk and CartwrightNorfolk & Cartwright, 1996).

Treatment of suicidality in secure facilities

Effective clinical assessment of suicide risk in prisoners and in patients in secure psychiatric facilities must take full account of the evaluation of the mental state and stress factors in the individual and in the environment, and the potential for access to means of suicide. Prisoners with mental disorder who are assessed as suicidal should be considered for urgent transfer to a psychiatric hospital with the appropriate level of security.

Although for whole populations the introduction of effective treatments for mental illness has not resulted in any reduction in suicide, it is essential, none the less, to afford adequate treatment on an individual level. Psychosis should be controlled with conventional antipsychotic medication, or where efficacy or side-effects preclude its use, atypical antipsychotics should be given. Some studies have shown clozapine to have a specific effect of reducing suicidality in those with schizophrenia (Reference Meltzer and OkayliMeltzer & Okayli, 1995). Depressed mood in patients with either depressive illness or schizophrenia may require adequate doses of antidepressants (tricyclics or selective serotonin reuptake inhibitors). In cases of high risk of suicide, there is evidence that ECT may be helpful and, indeed, it may well be that ECT is now underused as a result of its problematic public perception. As a treatment for depression, it may work rapidly and one study has shown that it carries a very low medico-legal risk of litigation (Reference SlawsonSlawson, 1992).

Patients with a history of substance misuse require an effective programme of relapse prevention. Those with personality disorder should have as equal a right to treatment as those with mental illness, albeit in a secure psychiatric facility with an appropriate remit. Provision of a combination of biological and psychological treatments may be beneficial in many cases of psychosis, depression and personality disorder.

Prevention of suicide in secure psychiatric facilities

The Mental Health Act (1983) Code of Practice (Department of Health & Welsh Office, 1999) outlines the need for patients to be protected from self-harm when the drive to do so is a result of mental disorder for which they are receiving care and treatment.

Box 7 lists the main parameters for the prevention of suicide in prisons and secure psychiatric facilities. Accurate clinical sensitivity to the patient's inner world is probably of paramount relevance in the approach to suicide risk (Reference Morgan, Buckley and NowersMorgan et al, 1998). Suppression of the active symptoms of mental illness, reduction of access to alcohol and drugs and treatment of substance misuse, supportive approaches to sources of stress and removal of the means of committing suicide should significantly reduce suicides in secure locations. In light of the complex relationship between suicide and violence towards others, any patient who is prone to violence towards others should be assessed for suicidality and, conversely, any patient who is suicidal should be assessed for risk of harm to others. Where a prisoner or patient is potentially suicidal, it is preferable to arrange constant rather than intermittent observation, as hanging or self-strangulation can cause death or irreversible brain damage within 5 minutes. The fact that suicides have been known to occur even during constant observation is not a reason to preclude its use. The efficiency of constant observation can be improved in a number of ways: for example, by ensuring that the patient does not hide his or her head under the bed clothes, by using nursing staff who know the patient and by limiting the duration of each nurse's observation period, to avoid fatigue. Some special hospitals still (controversially) use protective bedding and clothing that cannot be shredded or used as a ligature. In cases where the duration of suicidality is protracted, I feel that the use of such materials is justifiable, even if a little undignified. (It is very difficult to ensure the human rights of a patient if he or she is dead.)

In prisons, the use of strip cells has been much criticised and recently it has been phased out, with alternative strategies being advised. However, it is too early to know if these will be effective. Patients who are suicidal should not be secluded, according to the Code of Practice. In cases where there is a concurrent raised risk of serious violence to others, seclusion may be unavoidable, but it would have to be justified if subsequently challenged.

One of the few protective factors in suicide is religious commitment and, in some cases, this can be harnessed to reduce suicidality (although religion can also have its dangerous aspects, as noted earlier). I believe that in certain cases a chaplain or imam can play as much of a positive role in reducing suicidality as does a mental health team. Even where mental disorder is severe, the maintenance of hope is essential in the management of patients or prisoners, and loss of hope can foreshadow a suicide.

Postvention

The term ‘postvention’ was coined by Schneidman (Reference Jobes, Luoma, Hustead, Maris, Berman and SilvermanJobes et al, 2000) to refer to appropriate and helpful acts that may be used after an adverse event. The victims of a suicide include not only the prisoner or patient, but also those close to them – family or friends in the community, other prisoners or patients on the unit or in the hospital and the health professionals caring for them. A suicide is potentially traumatic to all of these people and support may be helpful in its aftermath. Family members may be vulnerable because of the effects of bereavement and shock at the manner of death, and their grief may encompass anger towards staff whom they may feel should have prevented the suicide. Patients on the ward should be sensitively informed and counselled, as should any other patients on other wards who were friendly with the dead patient. Owing to the occasional risk of ‘contagion’, mental state evaluation of patients on the dead person's ward and other relevant wards should be undertaken and high-risk patients identified. The police have to be informed in order that alternative causes of death, including homicide, can be ruled out. In the case of detained patients, the Mental Health Act Commission must be informed, and notification must be made to the Confidential Inquiry into Suicide and Homicide by People with Mental Illness. The hospital should also hold an internal inquiry and there should be an ongoing internal hospital audit of such adverse events (Reference RoseRose, 2000).

Concluding remarks

Risk factors for suicides in prisons and in secure psychiatric hospitals are extensive, and almost certainly far more suicides are prevented than actually succeed. But it is the completed suicides that become the focus of attention. Suicide has, however, been part of the human condition throughout history and its enduring nature testifies to the fact that, at times, the urge to destroy oneself overwhelms the instinct to continue with life. A paradox is that the rate of suicide did not decline during a century when the most effective treatments for mental illness were discovered, so clearly the phenomenon of suicide is affected by many parameters other than mental illness. None the less, attention to known risk factors, the use of vigorous treatment where appropriate, the removal of the means of suicide and ongoing audit of suicides may form a comprehensive basis for their reduction, both in the community and in secure psychiatric facilities.

Multiple choice questions

1. 1. Suicide is defined as:

   1. a attempting to harm oneself deliberately

   2. b killing oneself by accident

   3. c intentional self-killing

   4. d killing another person

   5. e deliberately taking risks that could be fatal.

2. 2. Homicide and suicide:

   1. a are completely unrelated behavioural phenomena

   2. b never occur together

   3. c are regarded by Freudians as both being types of homicide

3. 5. Suicides in prisons:

   1. a are not a matter for NHS involvement

   2. b may in some cases be triggered by stress factors

   3. c are mostly by hanging

   4. d are best dealt with by solitary confinement

   5. e may be reduced if other prisoners help in the management of prisoners who are suicidal.