2.1. THE NEED FOR A VIOLENCE AND AGGRESSION GUIDELINE
The need for a guideline focused on the short-term management of violence and aggression in mental health, health and community settings arises because violence and aggression are relatively common and have serious consequences in such settings (Bourn et al., 2003; Flood et al., 2008). The prevention and management of violence and aggression are complex tasks, because their manifestation will depend on a combination of intrinsic and extrinsic factors as well as the setting and context in which it occurs.
The intrinsic factors are a combination of personality characteristics, current intense mental distress and problems in dealing with anger. The extrinsic factors are more varied, including the physical and social settings where violence and aggression occur, the attitudes of those whose behaviour is violent or aggressive, characteristics of the victims, the experience and training of health and social professionals, and the perceived risk of danger to others. Understanding how such variable contextual factors interact with historical behaviour in the aetiology of violence and aggression is important in informing evidence-based approaches to the prevention of violence and aggression that would otherwise emerge, and also in the management of violence and aggression that has already occurred or is still in progress (Dack et al., 2013). In preparing this guideline, the GDG was also aware of a number of preconceptions regarding the perceived relative and absolute dangerousness of certain groups of service users, particularly those with severe mental illness such as psychotic disorders (Walsh et al., 2002). It is therefore particularly important to distinguish from the outset between the ‘problem’ of violence and aggression, and the care of those often-distressed individuals who may exhibit violent or aggressive behaviour.
In the NHS there are currently several general policies that are difficult to integrate because of variability in the contexts within which violence and aggression may emerge. While the management of violence and aggression is a core component of criminal justice systems, it has not generally been at the heart of systems for health and social care, which have instead tended to emphasise ‘zero tolerance’ approaches (Bourn et al., 2003). This approach is anomalous because the impact of violence and aggression in mental health, health and community settings is significant and diverse, adversely affecting the health and safety of service users, carers and staff (NICE, 2005). Critically, the management of violence and aggression may itself be hazardous to those exhibiting violent or aggressive behaviour and accentuate risks to their health and safety (Nissen et al., 2013).
The consequences of violence and aggression in mental health, health and community settings are not confined to the immediate environment, but have an impact on the wider health and social care economy (for example, costs of secure care for service users), and the economy in general (for example, sickness absence for staff) (Flood et al., 2008). Incidences of violence and aggression may also affect the perception by staff of services and service users in a manner that has a strong negative impact on the overall experience of care (De Benedictis et al., 2011).
If imminent violence is anticipated, its overt manifestations maybe avoided and non-restrictive interventions suffice. However, complete avoidance of violence is impossible and so a graded set of preferably evidence-based interventions is needed to prevent minor violence from escalating into major violence. For recommendations about interventions, NICE guidelines rely primarily on the results of randomised controlled trials (RCTs) in providing the underpinning evidence. However, because of the risks associated with severe violence, it is often not possible to carry out RCTs and, although there have been significant developments in this field since the previous guideline was published in 2005, it is likely that many recommendations will be based on expert opinion of the GDG.
2.2. DEFINITIONS OF VIOLENCE AND AGGRESSION
For the purposes of this guideline, violence and aggression refer to a range of behaviours or actions that can result in harm, hurt or injury to another person, regardless of whether the violence or aggression is behaviourally or verbally expressed, physical harm is sustained or the intention is clear.
Definitions of violence and aggression usually include some combination of the following elements: an expression of energy that may be goal directed; an immoral, repulsive and inappropriate behaviour; the intention to harm, damage or hurt another person physically or psychologically; the intention to dominate others; the experience and expression of anger; defensive and protective behaviour; verbal abuse, derogatory talk, threats or non-verbal gestures expressing the same; the instrumental use of such threats to acquire some desired goal; damage to objects or the environment, from vandalism through to smashing of windows, furniture and so on; attempting to or successfully physically injuring or killing another person with or without the use of weapons, or forcing another to capitulate to or acquiesce in undesirable actions or situations through the use of force; and inappropriate, unwanted or rejected sexual display or contact.
The number of definitions in circulation are so great that they have been combined into a rating scale to measure the Perception of Aggression (Jansen et al., 1997) as held by different people. Factor analysis of this scale, based on 32 definitions of aggression, shows that the concept comprises 2 fundamental elements: a positive perception emphasising healthy, normal protective aggression; and a negative perception of aggression as undesirable and dysfunctional.
Another way to approach the definition is to inspect the contents of the most well-used research instruments and scales that have been used to measure these behaviours. The Overt Aggression Scale (OAS) (Yudofsky et al., 1986) and its derivatives (Sorgi et al., 1991) are used to record aggressive incidents, and include: verbal aggression ranging from angry, loud shouts and noises through to clear threats; physical aggression against objects, ranging from door slamming and making a mess through to fire setting and throwing objects dangerously; and physical aggression against other people, from threatening gestures through to attacking another person causing severe physical injury. However, the OAS and many other such scales include self-harm and suicide attempts as aggressive behaviours against the self. The Social Dysfunction and Aggression Scale (Wistedt et al., 1990) is used to assess the total level of aggression retrospectively, and while including verbal aggression, aggression towards objects and others, it also incorporates irritability, lack of cooperation, discontentment, provocative behaviour and self-harm. Because there is a separate guideline on self-harm (NICE, 2004), this is excluded from the definition of violence and aggression used in this guideline.
2.3. INCIDENCE AND PREVALENCE OF VIOLENCE AND AGGRESSION IN DIFFERENT SETTINGS
Violence and aggression present a serious problem within the NHS to both service users and staff. Exposure to aggression in the healthcare workplace is common, constituting 25% of all workplace violence (Iennaco et al., 2013). In 2014, 14% of NHS staff reported having experienced physical violence from service users, their relatives or other members of the public in the previous 12 months, reduced from 15% in 2013. This figure was higher in staff in ambulance trusts (31%) and mental health trusts (17%) (NHS England, 2014).
More than 60,000 physical assaults were annually reported against NHS staff across the UK (NHS Protect, 2013), with the absolute rate steadily increasing since 2010– 2011 (57,830) and 2011–2012 (59,744). Of these assaults, 43,699 were in mental health or learning disability settings, 1628 involved primary care staff and 16,475 were targeted at acute hospital staff. More than 25% occurred in hospitals managed by acute trusts, including emergency departments (NHS Protect, 2013).
While some figures are collected and national audits conducted across different settings, the main focus has tended to be on inpatient psychiatric settings and emergency departments. Information from primary care settings, for example, is relatively scarce; 1 review found only 14 of 113 studies referred to violence in community settings.
In terms of the inpatient literature, 1 review (Bowers et al., 2011b) of 424 international studies reported that the overall incidence of violence by service users in inpatient psychiatric hospitals was 32.4%. Violent incidents across forensic settings were found to be consistently higher. The review team concluded that forensic inpatients were responsible for a higher proportion of violent incidents; but, given that acute wards admit a far higher number of people over time, on balance the risk of violence is actually greater in acute environments.
With regard to forensic settings, 2137 incidents involving 42.9% of service users were reported by a recent survey of a large independent secure care facility; this rate was greater in medium-secure as opposed to low-secure services (Dickens et al., 2013) Staff were the victim of assaults over twice as frequently compared with service users; however, if service users were the target, incidents were more likely to result in an actual injury. In a high-secure setting, Uppal and McMurran (2009) reported 3565 violent incidents over a 16-month period in just under 400 service users. Staff and service users were equally likely to be the target. In both surveys, a small proportion of service users was responsible for a disproportionate number of incidents.
Emergency department staff were also reported to have experienced a high exposure to aggression, particularly verbal aggression (Gates et al., 2006; Winstanley & Whittington, 2004). In long-term and older people's settings the figures for aggression were also found to be higher than general medical and surgical wards (Chapman et al., 2009).
Stathopoulou (2007) suggests that workplace violence affects every country and every healthcare setting. According to international data, nearly 4% of the total employee population has reported that they have experienced physical violence. The possibility of nurses being exposed to violence is 3 times higher than that of any other professional group (International Labor Office, 2002). This was reflected in a National Audit of Violence in the UK, which reported that 44% of clinical staff overall and 72% of nursing staff had been, or experienced feeling, unsafe at work (Royal College of Psychiatrists, 2007). The rates of psychiatrists being or feeling unsafe are reportedly lower than for nurses (Bowers et al., 2011c).
In light of these figures it is important to identify the causative factors that may contribute to these including care failures. This guideline aims to reduce such figures by suggesting best practice and preventative measures.
2.4. THE RELATIONSHIP BETWEEN MENTAL HEALTH PROBLEMS AND VIOLENCE AND AGGRESSION
Despite public perception that mental health problems (in particular, severe mental illness such as bipolar disorder and schizophrenia) and violence are associated (see section 2.5) the research evidence to support such a relationship is mixed, and most people with a mental health problem are not only never violent but are also more likely to be victims rather than perpetrators of crime (Pettit et al., 2013). However, a small proportion are and consensus has emerged among researchers that there is a consistent, albeit modest, positive association between mental health problems and violence. The extent to which mental health problems contribute to violent behaviour and the relative importance of psychiatric morbidity compared with other risk factors and service-related failings remain areas of controversial ongoing research.
To address the question of whether there is a link between mental health problems and violence, different research designs have been employed, including cross-sectional studies that investigate the prevalence of violence in those with mental health problems and, conversely, rates of mental health problems in those who have committed acts of violence (for example, offenders). While such studies have described a link between mental health problems and violence (Shaw et al., 2006), they are prone to selection bias as they tend to sample individuals detained in criminal justice or psychiatric settings. Some studies have been flawed by their lack of attention to potential confounding factors, such as psychosocial factors, comorbidity, substance misuse and so on. Prospective epidemiological studies of community samples following individuals for extended periods of time to identify those who will become violent and/or develop a mental health problem avoid some of these issues. However, other challenges in the interpretation of findings remain, for example the use of different methods to assess rates of violence, such as self-report, official criminal records and so on, each posing risks of misrepresenting the true prevalence of violence.
Until the 1980s, there was a general view that mental health problems and violence were unrelated; that is, that those with a mental health problem are no more likely to be violent than healthy individuals, and that the criminogenic factors relevant to violence risk are the same in people with a mental health problem as in healthy individuals (Häfner & Böker, 1973). Several large-scale studies in the 1980s and 1990s have resulted in a reappraisal and modification of this view.
The Epidemiological Catchment Area study (Swanson, 1994) comprised a community sample of over 17,000 participants in 5 large US cities, though only about 7000 subjects contributed to the data on violence. Individuals were asked to report any acts of violence they had committed within the previous year and in their lifetime. The study found a lifetime prevalence of violence in the non-psychiatric population of 7.3%. In those with schizophrenia or major affective disorders this rate was more than doubled at 16.1%, but in those with substance-use disorders it rose further to 35%, and those with a substance-use disorder and comorbid mental health problem had a lifetime prevalence of violence of 43.6%. Several early Scandinavian birth cohort studies (Hodgins, 1992) have identified a higher likelihood of having committed a violent crime in those with severe mental illness compared with those with no such diagnosis. A recent longitudinal Swedish study linking national registers of hospital admissions and criminal convictions over 33 years found that individuals with schizophrenia and bipolar disorder were more likely to commit violent acts than matched controls. In the period 1973–2006, 8.5% of individuals with schizophrenia without a substance-use disorder and 5.1% of the matched control group were convicted of at least 1 violent crime; for bipolar disorder these figures were 4.9% and 3.4%, respectively. However, those with dual diagnoses showed rates of 27.6% and 21.3% of violent offending for people with schizophrenia and bipolar disorder, respectively.
One of the most influential studies to disentangle some of the complex relationships between mental health problems and other risk factors for violence, in particular substance misuse, has been the MacArthur Violence Risk Assessment Study (Steadman et al., 1998). This follow-up study of over 1000 people discharged from psychiatric care used self-report triangulated with information from carers and criminal records to assess violence rates. The study found no significant difference between the prevalence of violence in patients and others living in the same neighbourhood when only taking those with no substance misuse into account. Substance misuse raised the rates of violence in people with mental health problems as well as healthy individuals, but disproportionately so in the patient group. Elbogen and Johnson (2009) also argued that a mental health problem on its own does not increase violence risk. They evaluated data on about 35,000 individuals who were part of the US National Epidemiological Survey on Alcohol and Related Conditions. Participants were interviewed in 2 waves in 2001–2003 and 2004–2005 to identify factors that predicted violence in the time between interviews. The researchers found that the incidence of violence was slightly higher in those with a mental health problem but significant only in those with a comorbid substance-use disorder. The researchers concluded that historical, dispositional and contextual factors were more important in determining the risk of future violence than a mental health problem. However, a later re-analysis of these data (van Dorn et al., 2012), using different statistical methods and diagnostic categories found that those with severe mental illness were significantly more likely to be violent than those with no illness, regardless of substance misuse.
More recently a number of meta-analyses have been conducted in an attempt to systematically re-assess the evidence and explore the reasons for variations in findings (Douglas et al., 2009; Fazel et al., 2009; Fazel et al., 2010). These studies, drawing on a large number of primary studies (204 and 20 for schizophrenia, and 9 for bipolar disorder, respectively), concluded that schizophrenia, other psychoses and bipolar disorder are associated with violence. However, large variations were identified with odds ratios between 1 and 7 for schizophrenia in males and between 4 and 27 for females. For bipolar disorder the odds ratio estimates ranged from 2 to 9. However, for both disorders a comorbid substance-use disorder increased odds ratios up to 3-fold. For bipolar disorder the significant relationship with violence disappeared when controlling for substance misuse whereas for schizophrenia the relationship weakened but remained, although in those with a history of substance misuse, schizophrenia did not contribute any additional risk compared with substance misuse alone.
Determining which symptoms of mental health problems drive the increased risk of violence requires further exploration. In the early 1990s, researchers first identified a set of symptoms called threat/control-override symptoms, which seemed to be linked to this risk (Link & Stueve, 1994). Threat/control-override symptoms are delusional symptoms that cause the person to feel severely threatened and believe that external forces override their self-control. Further studies of the relationship between threat/control-override symptoms and violence revealed conflicted findings with some but not all studies confirming a relationship. In an attempt to disentangle this issue further, Stompe and colleagues (2006) examined a sample of 119 offenders with schizophrenia found to be not guilty by reason of insanity and a matched sample of non-offending service users with schizophrenia (n = 105). While they found no significant difference in the prevalence of threat/control-override symptoms between the 2 groups overall, when only taking severe violence into account threat/control-override symptoms were found to be associated with this form of violence. It seems, therefore, that the relationship between threat/control-override symptoms and violence is not straightforward and that more research is needed to explore the concept further. In the meantime, clinicians are advised to conduct a comprehensive mental state examination as part of their risk assessment, including threat/control-override symptoms.
In summary, a mental health problem on its own appears to be only a modest predictive factor for violence while other factors, most significantly substance misuse, are more relevant in predicting risk. Because of the low base rates of mental health problems, its actual contribution to violence in the general population is small and the vast majority of violence is carried out by those without a mental health problem.
2.5. SOCIAL ATTITUDES TOWARDS VIOLENCE AND AGGRESSION
There has long been an association in the mind of the public between mental health problems and violence (Monahan, 1992), often bound up with moral and judgmental attitudes, whereby people who have a mental health problem are viewed as being irrational, unpredictable and dangerous and presenting with an increased risk of violence (Blumenthal & Lavender, 2000; Butler & Drakeford, 2003; Petch, 2001).
While there may be certain characteristics of some people with a mental health problem that may increase the risk of violence or indeed self-harm, as has been outlined in Section 2.4 the association between mental health problems and violent or aggressive behaviour is not established. One key issue for the public debate is whether violence generated by people with a mental health problem is increasing, but according to the Avoidable Deaths report from the National Confidential Inquiry, homicides by current or recent service users peaked in 2006, and has fallen since that year (Appleby et al., 2006).
However, a perceived association between mental health problems and violence is nevertheless often reinforced by images in the media and other cultural representations. As an example, in September 2013 a leading UK supermarket chain advertised a Halloween ‘mental patient fancy dress costume’ with an image of a person in a bloodied suit holding a meat cleaver. Negative media attention caused the supermarket to withdraw this item. The key point from this example is how such an image could have been brought to mind by those creating and marketing such products. While there are a number of theories about this, ‘labelling’ and the ‘availability heuristic’ (the process whereby people assess the frequency or probability of an event by the ease with which instances or occurrences can be brought to mind (Tversky & Kahneman, 1974) are 2 mechanisms that can influence negative attitudes and responses towards people with a mental health problem.
Labelling theory in sociology proposes that labelling occurs when certain members of society interpret certain behaviours as deviant and then attach this label to individuals (Becker, 1963) as a means to identify and control such behaviour. Labelling theory examines who applies what label to whom, why it is applied and what the effects are. The consequences of someone being labelled as having a propensity to violence just because they have a mental health problem can be negative and far-reaching. Labelling results in people having fears engendered by their attributions towards a person, leading them to conclude that the person is highly likely to be violent with no other knowledge of them other than the diagnosis. This in turn will affect their attitudes to, and communications with, people with mental health problems.
The ‘availability heuristic’ (Middleton et al., 1999) affects our attributions towards a particular idea or group of people; in this case, reporting in the media that draws attention to violence and murders carried out by people with mental health problems (often in a gory and sensationalist way) results in the attribution of violent behaviour to those with a mental health problem. This discourse was played out in the case of Philip Simelane, who murdered a 16-year-old female stranger, Christina Edkins, on a bus. The headline in the Daily Mail on the 3 October 2013 was: ‘Why was schizophrenic who stabbed this girl to death on a bus not having treatment?’ The focus, as here, tends to be on the fact that the person had a mental health problem, implying the murder occurred because of the person's mental health problem; other factors that might have been considered if the person had committed the same offence without having a mental health problem do not appear relevant. The more dramatic and easy to visualise the reported event, the more likely it will be contained within such a heuristic, with menacing photographs of ‘perpetrators’ and ‘horror stories’ of what they have done. Because of this, for many people, the first thing that often comes to mind about those with a mental health problem is that they are highly likely to be violent. There is much less reporting of other aspects of having a mental health problem, or of people with a mental health problem being more likely to be a victim of violence than a perpetrator, as found by 1 large-scale study in the US (Choe et al., 2008).
What is necessary instead is for the reality of the risks to be recognised and taken into account by both the public and professionals in a considered and fair manner, for the sake of all involved.
2.6. PERSONAL CONSEQUENCES OF VIOLENCE AND AGGRESSION FOR THE INDIVIDUAL AND FOR OTHERS
The under-reporting of violence and aggression (Gates et al., 2006; Holmes et al., 2012; McLean et al., 1999), and the varied effects it may have on those subjected to violence and aggression, limits our understanding of the consequences for the individual. Research into the effects of violence at the individual level has largely been focused on staff. While this is not surprising (because, by and large, staff have conducted the research and published the findings), other areas are less well covered. Other consequences of violence are only spelt out obliquely by research, resulting in limited understanding of the consequences for the individual who is prone to behaving in a violent manner.
The earliest work concerning the effects on staff and others of violence from people with mental health problems was produced by the Department of Health and Social Security (1976) and the Confederation of Health Service Employees (1977). The issues raised were in relation to physical violence in inpatient psychiatric units, and the concerns of the Confederation of Health Service Employees were about how their members needed greater recognition for, and protection from, such violence. In social care work in the community, the effects of violence to staff came later in the 1980s (Brown et al., 1986).
Holmes and colleagues (2012) concluded that the consequences of workplace violence for individuals were far-reaching and included absenteeism related to illness, injury and disability, staff turnover, decreased productivity, decreased satisfaction at work, and decreased staff commitment to work.
Physical injury as a result of assault by a service user can be serious, including injuries such as head, back, facial and eye injuries, broken bones, sprains, cuts, grazes and scratches. A review of multiple previous research studies estimated that 26% of violent incidents resulted in mild, 11% in moderate and 6% in serious injuries (Bowers et al., 2011b). A similar review of the psychological impact of violence found by previous research reported that the 3 most common responses to injury were anger, fear and guilt (self-blame and shame) (Needham et al., 2005). The fear can generalise into avoidance of the service user who has been violent or aggressive (Needham et al., 2005), or all service users, and some victims report persistent ruminations and intrusive thoughts about the incident, with symptoms severe enough to be classified as post-traumatic stress disorder.
Staff in the hospital
On any psychiatric ward a proportion of the staff's time is taken up with protecting service users from each other via the identification and protection of the vulnerable, general supervision of the environment and rapid response to any noise or cry for help, among other strategies. In addition, service users may also become involved in trying to defuse and deal with violence and aggression between service users, and between service users and staff. A proportion of the injuries that occur in staff happen during the breaking up of fights between service users, for example, but staff may also be assaulted unpredictably as service users respond to the symptoms they experience, or as a consequence of confrontations about leaving the ward, medical treatment or other issues (Nicholls et al., 2009). Staff also have to physically intervene to stop service users injuring themselves or trying to leave the ward, sometimes eliciting an aggressive response. Most assaults and aggression against staff – and by service users on other service users – are thankfully minor, but they can occasionally be severe. Every year several hundred injuries on staff are officially reported to the Health and Safety Executive by psychiatric hospitals as resulting in periods of sickness lasting 5 or more days. As a consequence of physical and/or psychological injuries, staff may leave psychiatry to work elsewhere. Verbal aggression to staff is extremely common and takes the form of abuse, shouting, threats, racism and generalised anger (Stewart & Bowers, 2013). Verbal aggression can have a profound psychological impact (Stone et al., 2010), affect performance and functioning (Uzun, 2003) and is the particular form of aggression that is associated with low staff morale (Bowers et al., 2009; Sprigg et al., 2007).
Staff in the community
Violence and aggression to staff in the community is less well documented and reported. While rates among NHS community teams are lower than those experienced by staff in hospital, the consequences are the same when assaults do occur. In England, since the early 1980s, 9 social work and social care staff have died as a result of violence from service users. The majority of those killed worked in mental health or child protection. Rates of assault experienced by staff working in supported accommodation run by a range of charities and private companies are unknown.
Personal consequences
Violent behaviour associated with a mental health problem is a criterion for admission to hospital, compulsory admission under the Mental Health Act (1983, amended 1995 and 2007) (HMSO, 1983; HMSO, 1995; HMSO, 2007), transfer or admission into more secure settings such as psychiatric intensive care or forensic services, and the use of severe containment methods such as manual restraint, rapid tranquillisation and seclusion. All things being equal, service users who exhibit violent behaviour will therefore experience more frequent admissions, more compulsory admissions, to greater security settings, for longer lengths of stay, with more restrictions on their liberty, greater coercion and higher doses of medication. Because violent behaviour is a criterion for exclusion from shared accommodation and social activities, service users who behave violently are likely to experience more accommodation instability and change, reduced social networks, social support and be more isolated. They may also have impaired access to mental health services in the community and, for safety reasons, home visits may be avoided and all appointments offered at clinics where the backup of other staff is available. Violent behaviour is therefore problematic for the person concerned and is likely to have a negative impact on their quality of life.
Relatives, carers and social networks
Where the risk of violence does exist, family members, carers and those in close contact with the individual concerned are most likely to be injured. Major injuries and deaths are rare, but the number of minor assaults is unknown as they may never be reported to the police or to anyone else. Living with a potentially violent person can lead to the family member or carer becoming severely stressed or developing a mental health problem. Alternatively, if the person concerned is living independently, relatives may withdraw, cease support or stop visiting if they are regularly faced with abusive and aggressive behaviour.
Other service users
People who share a ward with a potentially violent service user are also at risk of physical and psychological harm. Most aggression is directed at staff who are in positions of power, control access to desirable resources and discharge from the ward, and who may impose unwanted treatment. However, living in close proximity with others whose violence is unpredictable coupled with the service user's own psychiatric symptoms does place them at risk. Very occasionally that risk is severe and deaths have been reported. Minor assaults and injuries are regrettably more common, and approximately 20% of violent incidents on psychiatric wards are between service users (Daffern et al., 2006; Foster et al., 2007). The research literature tends to focus on consequences for staff in terms of physical injury and psychological distress, with service user outcomes seldom mentioned or studied. However, the consequences of an assault on people who already have a mental health problem may be considered to be negative, possibly hindering their recovery. It is known that inpatients are at times fearful and frightened of each other, leading to a range of avoidant behaviours that are employed to maintain distance from other service users who are regarded as having violent propensities (Quirk et al., 2004). Bullying between service users has also been reported (Ireland, 2006) as has sexual aggression. The move to single sex wards in UK psychiatry in recent years has been largely in response to a desire to protect female service users from unwanted or aggressive sexual advances from male service users (Department of Health, 2003). The consequences of unwanted sexual advances, harassment, bullying or assault are acknowledged as impeding the treatment and recovery of service users subjected to such behaviours, besides such incidents being extremely unpleasant in their own right.
Societal
Violent behaviour by people with a mental health problem is rare and only carried out by a small minority. However, it looms large in the public estimation (Bowen & Lovell, 2013; Thornicroft et al., 2007), adding to the stigma, fear and exclusion faced by this population. As such, the impact of violent behaviours is far bigger than the actual scope of the problem because it corrodes trust between people and makes it more difficult for the mentally ill to reveal their situation and to seek or obtain social support from others.
Dealing with the consequences
From the above discussion, it can be seen that violence and aggression have consequences for staff, service users and their families, carers and significant others, and the relationships between these people.
The consequences of violence and aggression cannot be dealt with unless incidents are reported and unless those reporting them feel they will benefit from doing so. Staff working in health and social care may not report incidents because they believe that they will not be dealt with sympathetically and are worried that they will be viewed negatively by colleagues and managers (Holmes et al., 2012).
Harris and Leather (2011) found in their research with social work and social care staff that as exposure to service user violence increased, so did reporting of stress symptoms and reduction in job satisfaction. Harris and Leather also found that fear or feeling vulnerable was an important consequence of exposure to violence and aggression; the same consequences of fear and feeling vulnerable can also occur in service users.
Ilkiw-Lavalle and Grenyer (2003), in a study on differences between service user and staff perceptions of aggression in mental health units, found that staff often perceived service users' illness as the cause of aggression, while service users perceived illness, interpersonal and environmental factors as having equal responsibility for their aggression. Such attributions from staff are important in how they will respond to incidents, and this will therefore affect their need for support post-incident in order for them to deal effectively and fairly with the consequences for themselves, service users, staff and others (Paterson et al., 2014).
Shapland and colleagues (1985) found that there were special considerations for victims of violence at work. Where staff could depend on supportive work colleagues and managers, and were employed by an organisation that proactively offered support, staff were more able to overcome the negative effects of violence at work.
The need for support depends on several factors:
The nature of the emotional and/or physical effects on the individual victim.
How the victim's views about the nature and causes of the violence might affect their approaches to that service user, and possibly other service users.
The individuals' experiences of support in dealing with the consequences.
Service users also have a need for agencies and staff groups to recognise that they, too, are affected and take measures to make them be and feel safe (
Holmes et al 2012).
2.7. CURRENT MANAGEMENT OF VIOLENCE AND AGGRESSION IN THE NHS
Given the risks posed by violent behaviour in mental health, health and community settings, all trusts have policies for its prevention and management. These policies can be wide ranging and are often directed at other primary goals, but also have secondary beneficial impacts on reduction of violent incident rates, reductions in their severity when they do occur and amelioration of their outcomes. For example, prompt and effective psychiatric treatment resolves acute symptoms and, because symptoms can be linked to violent behaviour, this constitutes one way that incidents are reduced. Within forensic settings, specific psychotherapies may be available to help people reduce their own capacity to act in a violent way. Buildings and wards are sometimes designed with the possibility of violent behaviour in mind, so in many areas, and especially in forensic or psychiatric intensive care settings, buildings are made out of stronger materials. Doors and furniture may be more robustly constructed, windows are fitted with stronger or safety glass, and living areas are designed in a way to maximise observation and supervision so that violent incidents can be quickly identified and responded to. Service users are searched for weapons on admission to hospital, and a number of items that could be used as weapons are banned from being brought onto the wards. As an aid to observation, closed-circuit television (CCTV) may be fitted in public areas and a variety of alarm systems may be fitted, from wall-mounted buttons to personal alarms for staff that quickly identify where an incident is taking place. These measures are accompanied by policies dictating their use and procedures as to who responds and takes control. In most psychiatric hospitals, if weapons are involved or the situation is beyond the capacity of staff to manage, the police may be called to manage the situation.
Within psychiatric hospitals, the main professional group that manages violent incidents (and who are most likely to be victims) are mental health nurses and healthcare assistants. The basic training of mental health nurses includes instruction on the causes of aggression, good communication skills and non-confrontational practice. During their training, nurses learn how to quickly establish and strengthen good relationships with service users, and these act as a safeguard against violence to staff, or aid in the de-escalation and management of agitated and violent behaviour. De-escalation or defusion refers to talking with an angry or agitated service user in such a way that violence is averted and the person regains a sense of calm and self-control. Most potential occurrences of violence are averted in this way, especially when there is some warning that they are about to occur, such as raised voices and abusive language. Of course some instances of violent attack occur suddenly and apparently ‘out of the blue’, and these are more difficult to prevent. All NHS psychiatric services provide additional training to their staff, especially those working in inpatient areas, in the prevention and management of violence. Such training typically (but not always) consists of 5 days with subsequent annual refresher courses, contains instruction on de-escalation, breakaway techniques and manual restraint, and is provided by an in-house training team. Where such training is commissioned from external private providers, a plethora of courses exists with different content. In-house courses are often linked to private providers via ‘train the trainer’ schemes. There are no detailed national guidelines on the content of violence management courses or on the specific physical techniques that are taught, and there are no standards, quality control processes or accreditation procedures for the courses concerned, whether provided in house or by external providers.
If an actively violent service user cannot be verbally calmed and is judged likely to imminently assault another, they will be manually restrained by suitably trained nurses and healthcare assistants. Such manual restraint is aimed at securely holding the person so that they cannot strike out or hurt others, so that they are not injured themselves and so that attempts to verbally engage with them can continue. Such holds can be slowly released when the person is emotionally calmed and can negotiate about their behaviour. If a state of calm cannot be immediately achieved, sedating medication may be offered by mouth or given by injection without the person's consent (rapid tranquillisation). If these efforts fail, the service user may be secluded in a specially constructed room, although not all hospitals have these. Additionally or alternatively, as the person becomes calmer, they may be asked to stay away from other service users by remaining in their own bedroom or other area (but without the door being locked), or be placed on some form of special psychiatric observation to facilitate early intervention if the violent behaviour seems likely to recur. Further changes to the person's regular medication regime may occur following a violent incident in an effort to prevent recurrence. Debriefing of the staff team and service user involved may also occur in an effort to learn from the incident and plan, so as to prevent the chance of a repetition. All these procedures are variously guided by a trust's policies and training provision for staff.
It is important to note that the nature and extent to which violence and aggression is experienced in the NHS varies greatly with the setting. The experience and hence the management of such incidents will differ between community and hospital environments. The interface with non-NHS agencies (such as the police, the courts and social services) has a role to play, and these links are well developed in some settings. Within the NHS hospital setting, there are particular areas that are better developed (by virtue of their philosophy of care, skills mix and clinical experience) to therapeutically manage acute or sustained risk of violence and aggression in the context of mental or physical health problem. These include emergency departments linked to general medical hospitals, psychiatric intensive care units within the acute inpatient psychiatric care pathway and forensic psychiatric inpatient facilities.
2.8. PREDICTING THE RISK OF VIOLENCE AND AGGRESSION AND THE CULTURE OF THE NHS
The prediction of the risk of violence and aggression by service users in mental health, heath and community settings is challenging in a number of ways. The key challenges include the lack of definition of what is being predicted, over what time-frame and in which context. Intuitively, the clinical tools required to predict imminent or short-term violence and aggression would be different by some degree to those utilised in the prediction of medium to longer-term violence or aggression. Furthermore, the heterogeneity in clinical populations where violence and aggression is exhibited seriously hinders the reliability and validity of specific clinical tools; there is no broad clinical assessment tool that can be applied in all circumstances where violence and aggression needs to be predicted.
Clinicians in the healthcare system have a duty to protect service users (both as potential perpetrators of violence and aggression, and as the victims of such acts), to protect healthcare and other professionals (which includes the attending clinician's personal safety), and to protect the wider public. Such duties are explicit in most professional codes of practice and are most apparent in the codes that regulate the practice of medical doctors and nursing staff.
In this guidance, the prediction of violence and aggression relates to that which is felt to be imminent or occurring in the very short-term; that is, within minutes or up to 72 hours. The fundamentals of predicting the risk of violence and aggression are driven by the best available psychiatric assessment of the person. Comprehensive assessment, which includes a psychiatric history, a mental state examination and an assessment of physical health, leading to clinical and risk formulations, will usually be difficult to achieve in acute clinical scenarios, and much of the clinical and risk information may not be readily available at the outset.
The assessment is an iterative and dynamic process that should lead to responsive changes in the clinical and risk management plan. Particular significance is attached to a past history of violence and aggression because past behaviour is a guide to future presentation. The impact of mental health problems, physical health problems, personality disorders, substance-use disorders, social impairment and cultural factors should be considered within the health or social care framework so as to understand the aetiology of the person's violent or aggressive presentation.
The approach described in the preceding paragraph is essentially that of unstructured clinical assessment. Although it suffers with low reliability, it is operator dependent and the reliability and validity are likely to be improved when it is used by more experienced and skilled clinicians. There is some evidence to support the notion that in the case of predicting inpatient aggression in acutely unwell service users, short-term clinical assessment can be useful (McNiel & Binder, 1991; McNiel & Binder, 1995).
There are 2 other types of violence-related risk assessment: actuarial risk assessments and structured clinical judgements.
Actuarial risk assessments use quantifiable predictor variables based on empirical research (often derived of an actual patient dataset, which ultimately limits their generalisability); they aim to provide a quantifiable value to the outcome in question. For the purposes of this discussion, the outcome in question would be the probability of violence or aggression occurring in the short-term.
Structured clinical judgements are an amalgam of the clinical assessment approach and the actuarial approach. Risk factors derived from a broad literature review are rated by the assessor using multiple sources of clinical information.
A number of violence-related risk assessment tools are currently available and some are in general use in specified clinical settings. These include:
Current clinical wisdom is that many of the available risk assessment instruments that predict future violence are broadly similar in their somewhat moderate predictive efficacies (Yang et al., 2010). The risk assessment tools listed above cover a wide variety of clinical settings, and most progress has probably been made in the area of forensic psychiatry. The majority of the risk assessment tools focus on medium- to long-term risk. A few have some emerging evidence base for their applicability to the prediction of violence and aggression in the short term and in non-forensic settings.
Any method that is to predict violence and aggression in the healthcare setting needs to look further than just patient-related factors. Patient-related factors are often well covered in clinical assessments and in violence-related risk assessment tools. Other areas requiring consideration include: staff-related factors (staff experience and training, role clarity); service-related factors (staff-patient ratios; the physical fabric of the ward, the philosophy of care and the ‘atmosphere’ of the clinical setting, multidisciplinary and multiagency input); and organisational factors (the culture of the organisation shaping the engagement philosophy between service users and staff). These non-patient-related factors are just a few examples, but they serve to illustrate the multitude of factors that can potentially shape the expression of violence and aggression. The knowledge and understanding of such factors by staff in more secure settings, such as PICU or forensic psychiatric services, is well described by the model of relational security (Department of Health, 2010). In terms of prediction, with its aim to better manage and reduce violence and aggression, these areas are probably of equal relevance to the direct patient-related factors.
The problem of aggression and violence seems to be endemic in the healthcare sector. The background literature is equivocal and the prediction of violence and aggression is an area of ongoing debate and research. Good clinical teams will make ongoing clinical and risk assessments (with or without the benefit of a violence-related risk assessment tool), and have quite a low threshold when considering a service user to be at high risk of violence or aggression. The low threshold usually leads to the use of clinical measures to prevent or manage the behaviour in the least restrictive and most therapeutic manner possible. Therefore, one argument is that good clinical management should lead to false positive predictions of violence and aggression, where it is predicted that violent and aggressive behaviour will occur but it does not (Steinert, 2006). With this in mind, the very purpose of risk assessment can be brought into question: is the purpose to predict violence or to intervene to prevent violence? The 2 outcomes would seem to require different instruments; the latter would be based in more of a formulation approach to identify relevant factors that may incite violence in a particular service user, rather than estimate how likely that person is to be violent in the future. Clinicians may be well advised to consider a formulation-based approach that facilitates the prevention and management of aggression and violence, as opposed to an over-reliance on purely predictive methods.
2.9. ECONOMIC COSTS OF VIOLENCE AND AGGRESSION TO THE NHS
Due to the complex determinants and broad manifestations of violence and aggression, its full economic impact is difficult to measure and, to date, no formal attempt has been made to quantify this for the UK.
Violence and aggression in the context of mental health issues is associated with a range of negative consequences, which may be broadly grouped into costs to individuals and costs to the UK health service. Incidents of violence and aggression may result in physical pain, stress, loss of confidence and other psychological problems. These personal costs accrue to the individuals at the centre of the episode, to other staff and fellow service users.
The wider health and social care system incurs the costs associated with secure care for service users, staff absence, legal services, extra training costs, NHS trust liabilities, compensation, ill-health retirements, staff replacement costs, counselling, and a myriad of retention and recruitment issues.
Combining data from the NHS protect physical assault statistics with health body declarations of staff, NHS Protect (NHS Protect, 2009; NHS Protect, 2010; NHS Protect, 2011; NHS Protect, 2012; NHS Protect, 2013) reported that there were an average of 188 assaults per 1000 staff per year in mental health/learning disability trusts. There was a wide variation between the numbers of reported incidents in the different sectors with an average of 36 assaults per 1000 staff reported in the ambulance sector, 19 per 1000 staff reported in the acute sector and 16 per 1000 in the community care sector.
Furthermore, the same report suggested that incidents of assaults across all sectors may be increasing with 44.4 incidents per 1000 staff in 2008/09 rising to 53 incidents per 1000 in 2012/13. This trend has the opposite direction in mental health and learning disabilities trusts with incidents falling from 193.9 per 1000 to 188 per 1000 between the same periods. Apparent trends in this data should be interpreted with caution because changes in populations, service provision health body amalgamations and reporting culture may all affect published figures.
Another report from the Wales Audit Office (Colman et al., 2005) supports the finding of increased incidents of violence and aggression in mental health services. In 2003–04 in Wales, most ‘generic’ incidents of violence took place in mental health settings, with 1790 such incidents representing 22% of all violent incidents in the country during that period. Incidents of violence and aggression also varied according to service area within mental health services. Violent and aggressive incidents are the third biggest cause of workplace injuries in the health and social care sector, as reported to the Reporting of Injuries, Diseases and Dangerous Occurrences Regulations.
To estimate the healthcare costs associated with incidents of violence, Flood and colleagues (2008) collected 6 months' of incident data from a sample of 136 acute psychiatric wards in England and combined these with end-of-shift reports from nurses in 15 wards to estimate the resource use per violent event. The cost calculation only accounted for the payment of identified staff and medication costs and as such does not observe fixed costs such as specialised facilities. The outputs of this analysis are estimates for the mean cost of violent incidents for individual psychiatric wards and for England as a whole. According to these authors, the annual cost in England of physical assaults is £5.3 million (2013/2014 prices), of aggression to objects is £3.7 million and of verbal abuse is £11.5 million. The analysis also estimated the costs associated with various containment strategies. In dealing with incidents, the use of general ‘as required’ medication was estimated to cost £8.6 million annually, with intramuscular (IM) medication in particular costing a further £3.9 million. Furthermore, transferring care to psychiatric intensive care services was estimated to cost £1.1 million and seclusion £2.2 million per year. Intermittent observation was estimated to cost £49.3 million and constant special observation £38.5 million per year. Manual restraint was estimated to cost £6.1 million and time out £1.3 million per year.
In terms of individual psychiatric wards, the work of Flood and colleagues (2008) estimates that approximately £270,000 of nursing cost per ward per year is associated with the management of violence and aggression. That is, more than one-third of the estimated total nursing cost (£736,000) per ward per year is connected with managing violence and aggression.
Although the currently available estimates of the costs of violence and aggression suggest substantial impact, these estimates remain inherently conservative due to the difficulty of measuring system-wide costs associated with incidents of violence and aggression. That the true costs are likely to be even greater emphasises the need to ensure efficient use of health and social care resources to deal with incidents of violence and aggression in a manner that maximises safety, quality and value for service users, carers and society in general.
