BJPsych Advances
The management of violence in general psychiatry
Sophie E. Davison


There is increasing concern about the level of violence within mental healthcare settings. In this article I review what is known on this subject, discuss the relationship between mental disorder and violence and summarise the different ways to prevent and manage violence. When planning strategies to prevent violence in such settings it is important to consider not only patient risk factors but also risk factors in the environment. Staff need to have all the possible techniques for managing violent behaviour available to them in order to weigh up the risks and benefits for any specific patient in any particular situation.

Despite public concern about violence perpetrated by those with mental disorders, the number of homicides in the UK carried out by people with mental disorders has remained constant over the past 38 years as the total number of homicides has increased (Taylor & Gunn, 1999). There is no evidence that the killing of strangers has increased. Evidence does suggest that rates of all types of violence in mentally disordered offenders have increased since the 1970s, but this is matched by the rise in violence in community controls (Mullen, 1997). There are high rates of mental disorder among prisoners, and people with schizophrenia are more likely than controls to be convicted of any offence, including violence (Mullen, 1997). This does not necessarily mean that mental disorder is itself associated with offending. It could mean that people with mental disorders are more likely to be caught or convicted.

The most informative way to look at the association between violence and mental disorder is to consider community studies of self-reported violence, as many violent acts are never reported and even fewer lead to convictions. The findings of the three main studies in this area – the Epidemiologic Catchment Area survey (Swanson et al, 1990), the MacArthur study (Steadman et al, 1998 ) and Link & Stueve (1994) – are shown in Box 1.

Box 1

The relationship between mental disorder and violence

Epidemiological studies show that:

  • people with mental disorders are more likely to be violent than community controls

  • substance misuse greatly increases the risk of violence in people with mental disorders and community controls

  • gender, age, past violence and socio-economic status have a much greater effect on risk of violence than the presence of mental disorder

  • comorbid personality disorder independently increases the risk of violence

  • the increased risk of violence is mediated in part by active psychotic symptoms

  • ‘threat/control override symptoms’, i.e. persecutory delusions, delusions of control and passivity phenomena, seem particularly important

  • the vast majority of people with mental disorder are not violent

Violence in in-patient settings

The risk of a particular violent incident happening at a particular time and place depends on the combination of the characteristics and current state of the perpetrator, the set of circumstances at the time, victim availability and the characteristics of that victim. Thus, factors other than the perpetrator should be taken into consideration when planning the prevention and/or management of violence. Indeed, many service users believe that external factors are more important precipitants of their violent behaviour than are internal ones (National Institute for Clinical Excellence, 2004).

The literature describing in-patient violence has been reviewed extensively by the Royal College of Psychiatrists in its clinical practice guidelines on the management of imminent violence (Royal College of Psychiatrists, 1998). That publication has been updated and expanded upon in draft guidelines from the National Institute for Clinical Excellence (NICE; 2004) on the short-term management of violent (disturbed) behaviour.

It is difficult to generalise from the many studies because the clinical settings, patient populations and definitions of violence vary. Most studies are observational, without control groups. However, the factors observed to be associated with in-patient violence can be divided into patient factors, situational/environmental factors and victim factors. There is some evidence that assault rates in hospitals reflect the level of violence in the population they serve (Walker & Caplan, 1993).

Patient risk factors

Factors that emerge consistently as placing mentally disordered people at risk of becoming violent as in-patients are being young, having a history of violence and being compulsorily admitted. Findings with regard to gender have been inconsistent. Some studies find that women in a hospital setting are involved in more violent incidents than their male counterparts but that men are much more likely to cause injury. No consistent findings have been reported with regard to ethnicity. Schizophrenia is the most commonly reported diagnosis among violent psychiatric in-patients, but is also the commonest diagnosis in in-patient psychiatric settings. As in the community, the risk of violence is highest in the acute phase and substance misuse is a significant risk factor.

The majority of psychiatric patients are not violent and a small minority account for a disproportionately high number of incidents. Violence that causes serious injury is generally rare. Owing to the type of patient that they contain, violence rates are highest in settings such as psychiatric intensive care units, forensic units and locked wards.

Environmental risk factors

The environment is very important, as it can be manipulated to reduce the risk of violence. Three groups of environmental factors seem particularly influential: the physical facilities provided for patients, visitors and staff; the experience, training, supervision and numbers of staff; and the policies in place to manage the clinical environment (Royal College of Psychiatrists, 1998). Individual factors found to increase the risk of violence are shown in Box 2.

Box 2

Environmental risk factors that increase the risk of violence

  • Lack of structured activity (there are fewer violent incidents in occupational and other therapy areas)

  • High use of temporary staff

  • Low levels of staff–patient interaction

  • Poor staffing levels

  • Poorly defined staffing roles

  • Unpredictable ward programmes

  • Lack of privacy

  • Overcrowding

  • Poor physical facilities

  • Availability of weapons

(Royal College of Psychiatrists, 1998)

High-morale wards – those with experienced trained staff and good leadership – report lower levels of violence. Issues that do not appear to have been addressed adequately in the literature but seem to be relevant are: optimal ward layout; optimal patient numbers; ideal staffing ratio; ideal proportion of staff with professional mental health training; optimal ward observation policy; optimal diagnostic mix of patients; the role of substance misuse policies; and the role of prosecution policies. There is probably no single ‘answer’ to these issues and each will need to be tailored to the setting and profile of the patients being cared for.

Victim risk factors

It is more difficult to build a coherent picture about the victim factors that increase the risk of violence. There is no consensus in the literature as to whether staff or patients are assaulted more often. Of the different disciplines, nurses are the most likely to be assaulted. Some units report that staff without professional mental health training are more likely to be assaulted, whereas others report that the most senior experienced staff are more likely to be assaulted because it is they who are involved in restraint. It is notable that physical aggression towards staff appears to be more likely to result from restraint than from direct assault. In one study a small group of staff were found to have been repeatedly assaulted. It is not clear whether this is because they were working in a high-risk environment or they had particular personal attributes. Women are more likely to be assaulted by women and men by men. Certain staff attributes have been reported as increasing the risk of being assaulted, namely rigid, authoritarian and custodial attitudes and a lack of respect towards patients.

Prevention of violence

The main aim of managing violence is to prevent it where possible.

Patient factors in prevention

Staff in all settings need to be adequately trained in clinical risk assessment so that they can determine whether individuals present a risk and, if they do, what may happen, to whom and in which circumstances. This allows a coherent management plan to be implemented that reduces the likelihood of that set of circumstances occurring. One approach is to screen for risk using a simple checklist and carry out a more detailed risk assessment if an individual appears to present a potential risk on the basis of the screen.

Risk assessment involves taking a thorough history that includes the patient’s personal history, past and present mental state, substance misuse and social functioning and looks at risk factors for violence. Particular attention needs to be paid to any past acts of violence, looking for the circumstances in which they arose, any early warning signs and any effective interventions. Risk is dynamic and must be regularly reassessed. A good risk assessment requires access to good-quality information from a range of sources. Thus, patient notes should be available at all times, including out of hours, and there should be robust inter-agency information-sharing protocols. Structured clinical judgement tools such as the 20-item Historical, Clinical and Risk Management (HCR–20) scale (Webster et al, 1997) can be very useful in clinical practice for assessing patients deemed at risk, as they provide a systematic and comprehensive clinical risk assessment on which to base a risk management plan. Research suggests that such instruments may increase predictive accuracy compared with clinical judgement alone (National Institute for Clinical Excellence, 2004).

The most effective preventive measures are treating the patient’s mental disorder effectively and tackling substance misuse, both at patient and ward level. Comorbid personality disorder and cognitive deficits should also be treated.

Environmental factors in prevention

A comprehensive list of recommended measures to alter the environment to reduce the risk of violence can be found in the Royal College of Psychiatrists’ clinical practice guidelines and the draft NICE guidelines (Royal College of Psychiatrists, 1998; National Institute for Clinical Excellence, 2004). These are summarised in Box 3.

Box 3

A calming environment

The following factors can reduce the risk of violence among psychiatric in-patient units:

  • a pleasant environment in which there is no overcrowding

  • a predictable ward routine

  • a good range of meaningful activities

  • well-defined staffing roles

  • good staffing levels

  • privacy and dignity without compromising observation of the ward

(Royal College of Psychiatrists, 1998; National Institute for Clinical Excellence, 2004)

The enquiry into the death of David Bennett highlighted other issues to be tackled (Norfolk, Suffolk and Cambridgeshire Strategic Health Authority, 2003). These are reflected in the NICE guidelines, which state that training that highlights awareness of racial, cultural, religious, gender and special needs issues also helps mitigate against violent behaviour. The guidelines also recommend that each service should have a local policy on alarms that are easily accessible, regularly checked and always responded to.

Victim factors in prevention

Some patients attract assault because of their own behaviour. Staff should pre-empt problems by identifying such individuals and encouraging them to modify their behaviour.

Staff should be trained and supervised to ensure that they are not drawn into showing disrespect and custodial authoritarian attitudes towards patients.

Management of imminent violence

Despite the best-laid plans, violent incidents do sometimes occur. In considering which of the different methods to use to manage a violent incident it is important to weigh up the risks and benefits in that particular situation. The draft NICE guidelines suggest that service users’ preferences be taken into account in the form of an advance directive.

Non-coercive methods

If at all possible, non-coercive methods should be used to manage violent behaviour. The aim is to engage the patient in calming down and dealing with anger or frustration in a non-violent way. Techniques include de-escalation, time out, increased observation and support, and offering medication with consent, if indicated.


De-escalation, or talking down, involves the use of psychosocial techniques aimed at calming disturbed behaviour and redirecting the patient to a calmer personal space (Dix, 2001). The successful use of de-escalation techniques requires training and a sophisticated understanding of aggression and its management. There are a number of differing theoretical approaches with no gold standard (National Institute for Clinical Excellence, 2004). All emphasise the following three basic components for de-escalation in an in-patient setting: assessment of the immediate situation; verbal and non-verbal communication designed to facilitate cooperation; and problem-solving tactics (Dix, 2001).

Dix has suggested that situational analysis is a useful basis for assessing the situation. This views the aggressive incident as an interactive process and involves trying to understand what has caused the patient to become aroused, why the patient thinks the situation has arisen, the emotional response evoked and the behavioural result.

Developing good communication skills requires staff to be aware of and monitor their own non-verbal and verbal behaviour, for example body posture, eye contact, tone of voice, use of clear language, being at the same height as the patient, proximity to the patient, and avoiding reassuring touching of the patient, which may be experienced as provocative (Dix, 2001; National Institute for Clinical Excellence, 2004). A member of staff facing a violent patient should present him- or herself as someone who can solve the problem, as a listener and not a restricter. A number of different tactics can then be used to try to work collaboratively to help the patient seek alternative solutions to their perceived problem.

Time out

Time out differs from seclusion (see Geographical restraint, opposite) in that the patient voluntarily moves out of the aggressive situation to a less stimulating environment.


The primary aim of observation should be to engage positively with the patient (National Institute for Clinical Excellence, 2004).


If the more collaborative approaches fail or the situation is acutely dangerous, staff have to take immediate action to make themselves, others and the patient safe. This usually involves restraining the patient in some way, i.e. constraining their movement so they are unable to act violently. Restraining methods can broadly be separated into geographical restraint (moving the patient to a quieter place, a more secure ward or seclusion), physical restraint and chemical restraint (rapid tranquillisation). All should only be used as a last resort. Each of these interventions has rare but potentially fatal complications. The risks are increased by the high physical morbidity of psychiatric patients. However, avoiding these interventions altogether is not an option if serious injury to others is to be prevented. Box 4 lists ways of reducing the associated risks.

Box 4

Ways of reducing the physical risks associated with physical restraint, rapid tranquillisation and seclusion

  • Have the full range of options available in order to weigh up the risks for that patient

  • Have enough staff properly trained in the techniques used

  • Ensure that all staff involved in physical restraint, seclusion or rapid tranquillisation are trained to a minimum of intermediate life support (National Institute for Clinical Excellence, 2004)

  • Have fast (within 3 min) access to modern life support equipment (automatic defibrillators) and emergency drugs (National Institute for Clinical Excellence, 2004)

  • Have a professional immediately available, qualified to prescribe and administer emergency drugs

  • Have policies about the use of the different interventions

  • Have proper mechanisms for monitoring their use and reviewing adverse incidents

Geographical restraint

Geographical restraint essentially involves moving the patient to an environment where they can more safely be managed. This might be a more secure setting such as a psychiatric intensive care unit or even a forensic unit, a less stimulating part of the ward or a seclusion room.

Psychiatric intensive care units have an important role to play in the management of acutely disturbed patients (Beer et al, 2001).

The code of practice governing the Mental Heath Act 1983 (Department of Health, 1999) defines seclusion as the forcible confinement of a patient alone in a room for the protection of others from serious harm. Because it involves a severe restriction of a patient’s liberty, there are strict guidelines to be followed. The code of practice states clearly that seclusion must only be used as a last resort and must never be used as part of a treatment plan or as a punishment. It must not be used as an alternative to having adequate staffing levels and highly trained staff.

The main risks associated with seclusion are suicide and, for patients medicated before being moved, the risks associated with rapid tranquillisation. These are best dealt with by having clear protocols for the observation and physical monitoring of patients in seclusion.

The Royal College of Psychiatrists’ clinical practice guidelines hypothesise that seclusion may be unnecessary if restraint is properly applied. However, in the absence of robust research it seems premature to ban the use of one of the ways of dealing with acute violence, especially as some studies show that most staff injuries occur during physical restraint.

Furthermore, both physical restraint, especially if prolonged, and medication carry significant physical risks. Therefore seclusion might be preferable in some circumstances, for example to avoid prolonged restraint or for an over-aroused patient who is already being treated with high doses of medication or for whom restraint brings back memories of past abuse.

Obviously, for individuals who cannot cope with being alone, seclusion might not be the best option as it may increase the likelihood of suicidal behaviour.

For further reading on the issues surrounding seclusion see Beer et al(2001).

Physical restraint

In the UK, when discussing physical restraint we are usually referring to trained staff using recognised techniques of physical contact to hold a patient and restrict their movement, thus preventing them from causing injury. Physical restraint should be used with consideration for the self-respect, dignity, privacy, cultural and special needs of the patient (Royal College of Psychiatrists, 1998). Like seclusion, it should only be used as a last resort. Staff restraining patients must be properly trained. This is to prevent injury to themselves and, particularly, to the patient: physical restraint can sometimes be associated with sudden death (O’Halloran & Frank, 2000; Norfolk, Suffolk and Cambridgeshire Strategic Health Authority, 2003). The exact mechanisms are unclear but it seems that asphyxia, especially if an individual is restrained in a prone position, plays a large role. Other factors increasing the risk are heart disease, over-arousal, struggling, obesity and drug effects, all of which are particularly relevant in a psychiatric population (O’Halloran & Frank, 2000). Box 5 lists issues that should be covered in staff training.

Box 5

Key points in physical restraint training

  • Avoid pressure to neck, thorax, abdomen, back and pelvic area

  • Prop prone patients up so they can breathe more easily

  • Make one team member responsible for ensuring that airway and breathing are not compromised

  • Restrain patients for the shortest period possible (this will depend on access to alternatives such as seclusion and rapid tranquillisation)

Some physical restraint techniques involve the deliberate use of pain. The NICE guidelines state that such techniques should be used only in exceptional circumstances.

A core training module for physical interventions is being developed by the National Institute of Mental Health for England.

(National Institute for Clinical Excellence, 2004)

Mechanical restraint

Mechanical restraints are no longer used in the UK, apart from in exceptional circumstances (Gordon et al, 1999). This appears to be because of a cultural view that they are not ethically acceptable. Other countries, for example the USA, do use mechanical restraints in the prevention of suicide and violence.

Chemical restraint (rapid tranquillisation)

Ideally, staff should obtain the patient’s consent to receive any medication that might be appropriate to reduce their level of arousal and prevent violence. However, in some circumstances it is necessary to give rapid tranquillisation. The aim of this is to sedate the patient sufficiently to reduce their immediate suffering and minimise the risk of violence; it is not to treat the underlying condition (Taylor et al, 2005). The patient should be able to respond to the spoken word throughout the period of tranquillisation (National Institute for Clinical Excellence, 2004).

The service users consulted during the preparation of the Royal College of Psychiatrists’ clinical practice guidelines reported that they preferred medication to physical restraint or seclusion when they behaved violently.

All staff should be familiar (to the level of their involvement) with local protocols for rapid tranquillisation. More details about choice of medication, route of administration and procedures can be found in Taylor et al(2005), Beer et al(2001) and National Institute for Clinical Excellence (2004).

The most serious risks associated with rapid tranquillisation are: respiratory depression or arrest; cardiovascular complications and collapse; seizures; and dystonia. Good procedures need to be in place to monitor the patient’s physical condition after rapid tranquillisation. In particular, pulse oximeters must be available (National Institute for Clinical Excellence, 2004).

The skills needed by doctors prescribing rapid tranquillisation are summarised in Box 6.

Box 6

Skills of doctors prescribing rapid tranquillisation

The Royal College of Psychiatrists (1998) and National Institute for Clinical Excellence (2004) recommend that doctors who prescribe rapid tranquillisation should:

  • be familiar with the properties of benzodiazepines and their antagonists, antipsychotics, antimuscarinics and antihistamines

  • be able to assess the risks associated with rapid tranquillisation, particularly when the patient is highly aroused and may have been misusing drugs, be dehydrated or physically ill

  • understand the cardiovascular effects of the acute administration of the tranquillising drugs and the need to titrate the dose

  • recognise the importance of nursing in the recovery position

  • recognise the importance of monitoring pulse, blood pressure and respiration

  • be familiar and trained in the use of resuscitation equipment

  • undertake regular resuscitation training

  • understand the importance of maintaining an unobstructed airway

Multi-agency working

Sometimes mental health professionals need to engage the help of other agencies, in particular the police, in managing violence. Police assistance may be requested if staff cannot contain a particularly violent incident or the police may be contacted after the event, to discuss whether to prosecute the patient. If a patient becomes violent, it is important that all involved are very clear about their roles. It is helpful for mental health organisations to develop joint working protocols with their local police to clarify in advance what the police are and are not able or willing to provide in the way of assistance during a violent incident. There is a move nationally to encourage all trusts to develop such protocols.

Prosecution of psychiatric in-patients

Prosecution of in-patients has historically been very difficult because of reluctance on the part of the Crown Prosecution Service (CPS) to view it as in the public interest. However, prosecution is now more common owing to the advent of the National Health Service’s policy of ‘zero tolerance’ of violent behaviour, the aim of which is to reduce violence in all heathcare settings.

Prosecution is clearly not feasible or desirable in all cases of violence in mental healthcare settings. It can damage the therapeutic relationship and may not be necessary if the patient was very ill at the time and is now making a good recovery. However, in some circumstances it can be of value: it might, for example, change the patient’s legal status in a clinically helpful way; it can aid future risk assessment if offences are officially on record; and it helps some patients to start taking some responsibility for their behaviour. Sometimes the local CPS needs to be educated about these potential benefits. Prosecution also sends out the message that the safety of mental healthcare professionals is taken seriously by the wider society.

The prosecution process runs most smoothly where the healthcare organisation has developed a policy in conjunction with the local police and CPS to determine which incidents will be reported, what information will be exchanged and what response the organisation can expect.

Violence in community settings

Although most violence perpetrated by people with mental disorders (and, indeed, by those without) occurs within the domestic environment, the vast majority of the literature on the prevention and management of imminent violence in this group relates to in-patient settings. However, over recent decades more and more psychiatric care has been delivered in a community setting. The introduction of home treatment/crisis intervention teams has meant that increasingly people who are acutely ill are being managed at home. Violence encountered by community mental health teams, particularly during domiciliary visits, can be more difficult to manage, as the full range of interventions and a highly trained response team may not be available. Also, the patient’s risk and response to different situations may be less well known.

Staff should be trained in personal safety methods such as thorough risk assessment in order to anticipate potential violence, and basic breakaway and de-escalation techniques. Teams should develop clear policies on how to deal with issues such as alarms sounding in consultation rooms; who will respond to an incident; protocols to ensure the safety of home visits; home visiting in pairs if necessary; and protocols agreed with the local police, clarifying when they will assist with Mental Health Act 1983 assessments. Galloway (2002) has written in more detail about safety in the community in a previous issue of APT.

Research suggests that staff safety in interview rooms in all mental healthcare settings remains inadequate in many situations (Sipos et al, 2003). Box 7 shows recommended safety features for interview rooms.

Box 7

Essentials for interview room safety

  • Easily accessible, functioning alarm systems

  • Clear, unobstructed exits

  • Doors that open outwards, cannot be locked from the inside and allow easy access from the outside in the event of an emergency

  • Location close to staff areas

  • Removal of all potential weapons (these are a particular risk if the room has a dual function)

  • An unobstructed viewing window

  • A furniture layout that minimises violence

(Osborn & Tang, 2001; Galloway, 2002)

The management of chronically assaultive behaviour

General strategies

So far I have discussed the prevention of violence and the management of acute violent incidents. There is, however, a small minority of patients who remain chronically assaultive. Individual incidents should be managed as described above. In addition, a management strategy is needed to try to reduce the overall assaultive behaviour of these individuals.

In the first instance their diagnosis and treatment should be reviewed and their mental disorder effectively treated. Sometimes, partially treated psychosis is mistakenly relabelled personality disorder because the disturbed behaviour continues after the obvious acute symptoms have started to resolve. The contribution of comorbid substance misuse, personality disorder and cognitive deficits should be addressed. As with acute violence, the contribution of environmental factors should be assessed. It is very helpful to gain a psychological understanding of the individual’s chronically disturbed behaviour in order to formulate strategies to manage it. Psychological management along cognitive–behavioural lines and treatments aimed at improving engagement can all be effective (for further details see McKenzie, 2001).

Pharmacological intervention

There is emerging research evidence, largely from uncontrolled trials, that clozapine reduces persistent aggression in schizophrenia and that the reduction in hostility and aggression may be independent of its antipsychotic effect (for reviews see Glazer & Dickson, 1998; Buckley, 1999; Volavka & Citrome, 1999). There is little robust research looking at the effect of typical antipsychotics on aggression. There is some evidence from case studies and uncontrolled trials that mood stabilisers, in particular carbamazepine might be useful as an adjunct in assaultive patients with schizophrenia (Citrome & Volavka, 2000). There is no good evidence for the use of benzodiazepines or high-dose antipsychotics in the treatment of chronic assaultive behaviour. Studies of patients with brain-injuries have found beta blockers to be useful in reducing aggression, but their usefulness in individuals with functional mental disorder is less clear (Citrome & Volavka, 2000).


The number and impact of violent incidents in mental healthcare settings can be reduced by the appropriate, therapeutic and effective use of the full range of interventions. This can only occur if adequate numbers of professionals are properly trained in the different techniques and organisations have robust systems for auditing and monitoring the prevention and management of violence.


  1. The following are consistently identified risk factors for violence in people with mental disorder:

    1. substance misuse

    2. young age

    3. male gender

    4. past violence

    5. active psychotic symptoms.

  2. The following have been associated with an increased risk of in-patient violence:

    1. overcrowding

    2. high-morale wards

    3. a predictable ward programme

    4. lack of privacy

    5. staff with authoritarian attitudes.

  3. When managing acute violence:

    1. de-escalation techniques require no training

    2. non-coercive techniques should always be considered first

    3. physical restraint of an individual in the prone position carries risks

    4. time out does not require patient consent

    5. the aim of rapid tranquillisation is to treat the psychosis.

  4. The following may be useful in the management of chronic violent behaviour in people with a psychotic illness:

    1. management along cognitive–behavioural lines

    2. clozapine

    3. mood stabilizers

    4. benzodiazepines

    5. treatment of comorbid disorders.

  5. In relation to the management of acute violence:

    1. prosecution has no role to play

    2. all patients should be prosecuted

    3. the police should never be called to assist in containing a situation

    4. it is helpful to have agreed joint working protocols with the police

    5. it is helpful to develop with the Crown Prosecution Service a policy in relation to prosecution.


MCQ answers


  • A more detailed review of guidelines relating to restraint and rapid tranquillisation will appear in the next issue of APT (Macpherson et al, 2005). Ed.


View Abstract