|
M ental I llnessC oncernsA ll | The HCR20
It emerges from a detailed review of the cases that at least two are marginal for inclusion in the sample, and both are of White ethnic origin. If these cases were excluded, the distribution of ethnic origin would be further skewed towards over-representation of minorities. HCR20 Data - Historical Items This section of the HCR20 consists of 10 items in the history that are likely to be related to an increased risk of violence. These items are: H1: Previous Violence H2: Young Age at First Violent Incident H3: Relationship Instability H4: Employment Problems H5: Substance Use Problems H6: Major Mental Illness H7: Psychopathy H8: Early Maladjustment H9: Personality Disorder H10: Prior Supervision Failure It should be noted that the sample selection criteria ensure that all cases have H1 and H6. The criteria are listed here to demonstrate that it is not unreasonable 5 that a clinical team should have this sort of information available, particularly when there is a known history of violence. Table 2: Number of HCR20 Historical items present for each patient in the total sample (N = 25)
HCR20 Data - Clinical Items The five Clinical items are: C1: Lack of Insight C2: Negative Attitudes C3: Active Symptoms of Major Mental Illness C4: Impulsivity C5: Unresponsive to Treatment Again, it is reasonable to suppose that this sort of information should be available to the team. 6 Table 3: Number of HCR20 Clinical items present for each patient in the total sample (N = 25)
HCR20 Data - Risk Management Items The five Risk Management items are: R1: Plans Lack Feasibility R2: Exposure to Destabilisers R3: Lack of Personal Support R4: Non-compliance with Remediation Attempts R5: Stress R1 perhaps requires further explanation. It refers to the need for an agreed, realistic care plan. One of the best illustrations of a positive score on this item emerges from the Ritchie report on the care and treatment of Christopher Clunis. He was discharged on several occasions with no proper care plan in place, and no reasonable expectation that he would be able to cope with life. This item alone does not, of course, mean that violence is inevitable but it is an indication that the situation is unstable and a crisis will arise fairly soon. It is more difficult to complete these items in retrospect, than for the Historical and Clinical items. My point of reference was the last contact with the clinical team. At that time, were there feasible plans, was their current or likely exposure to destabilisers or stress, was there any personal support, and was the patient 7 compliant and with whatever treatment was offered? One hopes that these are the sorts of questions likely to be considered as part of a care planning meeting under CPA. Table 4: Number of HCR20 Risk Management items present for each patient in the total sample (N = 25)
Discussion of Preliminary Analysis Overall Numbers The first point to make is that the number of cases is small, despite the fact that the study period spans seven years. The average is less than four cases per annum. Of course there is more to this issue than simple numbers but this figure must be seen in the context of other homicides. The overall homicide rate for England and Wales from 1995 to 1997 ranged from 584 to 650 cases per year, and by 2001/2002 it was close to 800 cases per annum. In other words, the cases with which we are concerned here account for less than half of one percent of all homicides in England and Wales. It is important in this context to say more about case identification. The methodology of the CISH has been described elsewhere and should ensure that cases are not missed. The only added step in the current study was determination of whether or not there had been previous violence. The figures would have been biased if there had been unreasonable exclusion of cases on the grounds that there was no previous violence. My tendency was to include any record of previous violence. For example, in one case the only history of previous violence was a conviction for assaulting the police. The consultant concerned had assessed the patient at the time and concluded there had been only minor 8 attempts to resist when the police had arrested the patient for possession of cannabis. Despite this comment, the case stayed in the sample because there was a conviction for assault. The excluded cases had no history of violence as an adult in any available reports or records. Ethnic Origin There is an over-representation of ethnic minorities relative to the general population. The marginal cases described above were both White British, so if they had been excluded the sample would have been more skewed. The numbers are too small to draw any conclusions, but the data are obviously relevant to ongoing concern about possible over-representation of ethnic minorities among detained patients. HCR20 Data The Historical items are important to the present study because they can be assessed reasonably accurately in retrospect. They concern aspects of the history that can be determined from the records, so if the information can be obtained now it could presumably have been obtained at the time. Ten of the patients scored on all ten items of the History scale, and 20 patients (80% of the sample) had seven or more items present. The HCR20 is not an actuarial scale, and adding up the scores is not necessarily a good indication of overall risk of violence, but it is difficult to dismiss the presence in one patient of all or practically all of the historical factors that are known to be associated with an increased risk of violence. Similarly, the findings for the Clinical and Risk scales were also high for many of the patients, although it must be borne in mind that these figures are less reliable when derived from records after the event. This reservation need not be overstated. Items such as non-compliance, lack of response to treatment, or negative attitudes are often easy to determine from records. Also, most errors arising from retrospective scoring will understate the problem. For example, lack of insight or the presence of symptoms may have been overlooked in the records, whereas a systematic inquiry at the time may have revealed them. The first point to make is that these ratings are at the top end of the spectrum of risk. The HCR20 is not an actuarial instrument, so a high score cannot be directly equated to high risk. Even so, these cases are impressive in having all, or nearly 9 all, the factors that tend to predict violence risk. This suggests that homicides by the severely mentally ill are far from random events that could happen in any patient. Many of these cases were extraordinary patients before the homicide, when looked at in terms of violence risk indicators. This conclusion is based on the assumption that most patients will not score at such high levels. The use of the HCR20 is at an early stage in the UK but early results from Prof Shelagh Hodgins and Prof Tom Fahy at the Institute of Psychiatry suggest that forensic patients generally score highly on the H items of the HCR20, whereas many patients within general services score highly on C or R items (personal communication). Most of the patients in the present sample resemble forensic patients in terms of their HCR ratings, although very few were under the care of a forensic service. Discussion and Recommendations Structured Clinical Assessment of Violence Risk Methodological Issues A structured violence risk assessment instrument is a useful and informative way of looking at homicides by psychiatric patients. It was possible to complete the instrument in all cases, although the quality would undoubtedly have been higher had it been possible to carry out the assessment before the event, with he addition of data from an interview. This limitation does not negate the findings. The nature of the HCR20 is that it encourages a search for risk factors. If there is more information, it may be possible to identify more risk factors, but the additional information is unlikely to negate factors that are already identifiable. In other words, the scores presented here may have been higher if the assessment had been done at the time, but they are unlikely to have been lower. The value of the HCR20 in the present exercise, is that it gets round some of the problems relating to hindsight. A common criticism of Homicide Inquiries is that they look at clinical practice in a biased or distorted way because the “future” has already happened. There are many things in life we would do differently if we had the privilege of knowing how things would turn out, but this insight has no practical value. The HCR20 is different because it looks at the patient rather than at clinical decisions, and the criteria for each item have a degree of objectivity. This is particularly so for the Historical items, on which I have placed most reliance. For example, characteristics such as “Previous Violence” (H1), “Young Age at First Violent Incident” (H2), “Relationship Problems” (H3), “Employment Problems” (H4) or “Substance Use Problems” (H5) are not likely to have been altered by hindsight. The only way in which hindsight is likely to elevate ratings is if the homicide leads to the collection of data that was not available before. It is one of the cruel ironies of a homicide inquiry that hundreds of thousands of pounds are expended on the collection of detailed information about a patient, when the service looking after him did not have the resources to gather data from so many sources. By contrast, the HCR20 relies on a limited set of variables and the data required ought to be available to any team looking after a patient with a history of violence. For example, the Clinical Items are: Lack of Insight; Negative Attitudes; Active Symptoms of Major Mental Illness; Impulsivity; and Unresponsiveness to Treatment. If the team do not have the information available to estimate these basic features of a case, they are not in a position to address risk. This is the sort of information that should be available to a team, rather than the detailed historical information collected by an Inquiry. 58 I was not able to complete the second part of the HCR20, which involves the construction of possible scenarios of violence and the development of a risk management plan. This would normally be undertaken at a CPA meeting and the problem of hindsight makes it an artificial exercise after the event. Prospective studies confirm the usefulness of this approach, and a paper by Dolan and Doyle, reporting positive findings, has been accepted for publication. The value of structured clinical assessment of violence risk Many of these Inquiries, and others before them, called for the introduction of better risk assessment procedures. Most Trusts have some form of clinical risk assessment in place but most do not have a true, structured clinical assessment such as the HCR20. The most basic approach involves ticking boxes relating to a patient's history and may offer a false sense of reassurance. An effective system requires systematic consideration of past history, present state, and future placement and stressors. The Inquiries were certainly right to call for better risk assessment procedures. There was a sense in some of these cases that the team had no awareness of the level of risk they were dealing with. Even after the event, some commented that the risk had been low even when the indications of increased risk were obvious once a structured approach was followed. A lot of the debate about violence risk assessment focuses on the accuracy of prediction but these cases reveal a more basic problem. Teams are often unable to formulate, describe or communicate an assessment of violence risk. Such an assessment is multifaceted and ought to include consideration of the type of violence, the likely victims, exacerbating and alleviating factors, and the duration and immediacy of any risk. Instead, assessments are limited to “low” or “high” and they are often wrong even in their own simple terms. The first step in improving risk management is to develop an agreed way of describing risk. Until there is a common language, it is impossible to make much progress. Recommendation: Mental health teams need to develop a common way of formulating, describing and communicating violence risk. Recommendation: All mental health teams should have access to a structured clinical assessment of violence risk and should incorporate its findings into the care management of patients with a history of violence. 59 Compliance and Compulsory treatment in the community From the small number of cases that form the sample for this report, this one issue stands out above all others. Risk assessment alone is not enough and there has to be an effective means of managing that risk. Given the nature of the sample, which is defined by the presence of serious mental illness, it is inevitable that medication and compliance are major issues. Non-compliance was an identified problem in most of these cases. It is not surprising that discussions of compliance featured in most of the Inquiry reports, and there were a range of suggestions for addressing the problem. They include compliance therapy, motivational interviewing, better involvement of carers and relatives, use of restriction orders to deal with previous violent offending, and better monitoring to ensure compliance. All of these are good suggestions, and most services use these strategies and techniques to a greater or lesser degree. However, when the cases are considered together, these suggestions are not an adequate response to the problem. The sample was defined by the presence of serious mental illness and previous violence but many of these cases had a well-documented lack of insight into their mental illness, accompanied by negative and hostile attitudes to services, and to authority in general. Some had criminal or antisocial attitudes that were well entrenched before the onset of mental illness. Even with persuasion, many of these patients were never going to comply voluntarily with medication. The task of persuasion in such patients is likely to be time-consuming, difficult and unsuccessful, particularly because the aim is compliance with medication over many years. It is reasonable to expect a team to sustain a high level of input to ensure compliance over a brief period but it is not reasonable to expect that effort to be sustained indefinitely. There are also resource implications. Should teams divert most of their resources to problems they probably cannot solve? When we consider existing powers, it is easy but misleading to criticise individual decisions on detention. The 1983 Act is worded so as to allow the detention of practically any non-compliant patient with a mental illness, on the grounds of detention being necessary for the patient's health, but such detention appears pointless if it is just another turn of the revolving door, rather than the beginning of long-term, effective treatment. The histories of many of these patients are peppered with brief admissions, and it is unrealistic so suppose the answer is to be found in yet another admission. Schizophrenia is a chronic relapsing condition and it follows that effective compulsory treatment will need to be administered on a long-term basis. In many of these cases, services were struggling with difficult and dangerous patients, sometimes with well-established histories of violence and other antisocial 60 behaviour before the onset of mental illness. If they are to take on and manage such people, services must be given the legal powers to do so effectively. Added force is given to the case for compulsory powers by the fact that their absence has been mentioned so many times in homicide inquiries. Case #3 above is an ideal example. The Inquiry was thorough and identified only minor issues relating to the patient's care but came to the firm conclusion that the case could only have been managed safely if there had been the power to impose treatment in the community. It is a truism of risk management that risks become less acceptable once attention has been drawn to them. The Inquiry report on Case #3 was published in 1999 and, given the stated priority of reducing violence risk, urgent consideration should be given to introducing some form of compulsory treatment in the community. Case 3 suggests that any such change in the law would have wide ramifications. A new law could have been used in that case to good effect but the impact on other cases may have been even greater. Case 3 was an example of team doing everything possible within existing law to measure and manage violence risk. In several other cases, there was a failure to address known risks of violence, deriving in part from a sense that nothing could be done about it. By focussing on individual decisions not to detain a patient at a particular time, Inquiries sometimes miss the wider perspective. A sense of fatalism and futility pervades the management of some of these cases. The prospects for compliance are correctly assessed as poor and the illness is severe, chronic and relapsing, so a decision on a brief admission becomes relatively unimportant. It offers no solution to the long-term problems. Of course, that changes when a homicide occurs, but this is with the wisdom of hindsight. It is unrealistic to expect that teams will be able to know precisely when to intervene to prevent such events when they are managing high, long-term risks of violence in uncooperative patients. The lack of such powers discredits the whole enterprise of risk management. There is understandable scepticism about the value of measuring a risk when one lacks the means to do anything about it. Also the focus on compulsory admission, because that is the only form of compulsion permitted, reduces the question to one of admission v. non-admission. If there were compulsory powers in the community, the focus would move away from an obsession with admission, to the more pressing question of safe management in the community. Recommendation: There is a need for legal powers to allow compulsory treatment in the community of patients with a serious mental illness and a history 61 of violence and non-compliance. Drug and Alcohol Misuse: Dual Diagnosis In all but two of the sample cases there were known problems of drug and/or alcohol misuse. The case descriptions above show how often intoxication was involved in the homicide itself. This is not surprising, in that intoxication is a well-recognised risk factor for violence. In community surveys of violence and mental disorder, substance misuse is reliably identified as having a stronger association with violence than does mental illness it its own right, and the combination of mental illness and substance misuse problems usually has the highest risk. Whilst there is no doubting the importance of substance misuse, the association is complex and does not lend itself to simple solutions. There is a vast criminological literature on alcohol/drugs and violence, and the only simple message to emerge is that the links are strong but complicated. Apart from the direct effects of intoxication and the social decline that may result from spending money on substances rather than on life's necessities, there is a social dimension. Heavy drinking or drug use often takes place in a social environment where violence is not only acceptable but a preferred and respected method of resolving conflict. As cases 15 and 24 illustrate, the mentally ill are not isolated from that world. Leaving aside the complex questions in these cases concerning the possible effects of (treated) schizophrenia in leading to paranoid thoughts, impulsive aggression and violence, both concern violence that originated in the pub and involved the active participation and encouragement of friends or family in whom there was no question of mental illness. Cases such as these show the complexity of the issues facing mental health teams attempting to address the risk of violence. It is obviously right for mental health teams to respect the family and cultural background and beliefs of patients, but there is a dilemma when that background includes heavy drinking and use of illicit drugs, in a subculture that condones violence in many situations. The cases summarized above show a range of responses to the problem, with different degrees of tolerance to the use of cannabis or alcohol. That is not unreasonable, given the complexity of the problem, but there was no sense of any systematic assessment or decision-making process. The common response of Inquiries to substance misuse problems is to recommend the greater involvement of substance misuse services, or the setting up of specialist Dual Diagnosis teams. It is hard to disagree with this advice but it has far-reaching implications, given the frequency of the problem. The immediate question is therefore one of determining priorities. 62 In this respect it is useful to look at the way in which substance misuse figures in risk assessment schemes such as the HCR20. It is taken into account alongside other risk factors, both as a Historical indicator of risk, and as a Risk Management item, “Exposure to Destabilisers”. It may also be scored under the Risk Management heading of “Stress” if substance abuse contributes directly or indirectly to the stresses facing a patient in the community. These include the social and financial stresses that may flow from substance use. In this way, the problem of substance misuse is assessed within the wider context of the patient's life and the overall violence risk. Substance misuse in the context of serious mental illness and violence may greatly increase risk, yet there is little sign within these cases of it being considered as an indication for use of compulsory powers. The 1983 Act does not allow detention for substance use alone but it is an important indicator of the nature and extent of a mental illness, and the associated risks to others, so it should be included in any assessment for possible detention. Recommendation: Substance misuse problems in patients with severe mental illness and a risk of violence should be assessed and managed within a structured clinical risk management plan. Proper consideration should always be given to the possibility of using the Mental Health Act in such patients. Recommendation: In patients subject to a restriction order there should always be consideration of setting conditions relating to abstinence from drugs or alcohol and the standard procedure should be immediate recall if that condition is breached. Setting Limits and Early Intervention To some extent, this general topic cuts across several other recommendations but it deserves emphasis. In several of these cases (e.g. Cases 8, 17, 19 and 20), including two conditionally discharged restricted patients (Case 13 and 25), there was a clear case for earlier intervention. The teams seemed to have no clear idea of where to draw the line when there was deteriorating behaviour, non-compliance or drug misuse. The restricted cases were particularly anomalous, with a history of known cannabis use for many years. It has become commonplace for CPA meetings to list signs of relapse or deterioration but the missing element seems to be a clear statement of when to intervene. It is pointless to identify early warning signs of relapse unless they lead to action. Several factors appear to have impinged on the management of cases where limits were a factor. There was a sense of familiarity leading to complacency. For example, Case 13 had a long history of drug misuse so it became acceptable. In 63 Case 17, one wonders if the forensic team had become desensitised to the indicators of violence risk that were all too apparent. Limits are important because they can make psychotic patients feel more secure when their own, internal boundaries are fragile. Also, if there is no attempt to set limits, it becomes impossible to access further information about the patient's mental state. For example, the suggestion to a patient that he should stop using drugs may lead into an exploration of his attitudes to drugs, his illness, and treatment, whereas no information is obtained if the behaviour is never challenged. The argument against early intervention would be that it is too intrusive and would lead to unnecessary treatment. This argument is weak when there is a history of violence, and structured risk assessment of violence risk would also guide teams in deciding which patients should be given priority. Recommendation: There should be early intervention when there are signs of deterioration or risky behaviours in patients with a history of violence. This principle should rarely if ever be ignored, even when a patient is well known to the service. If in doubt, the team should err on the side of caution. Recommendation: When dealing with patients with a history of violence and serious mental illness, CPA meetings should set clear, operational criteria for intervention. These criteria should be communicated to patients and carers so they have clear expectations. Forensic and Generic Services Forensic services were not directly involved in most of these cases at the time of the homicide and there were worrying aspects to those cases in which they did have responsibility. Of course, this is a highly selected sample and does not reflect the overall impact of forensic services in any way. Still, some general points can be made. There would be a case for greater forensic involvement in those patients with the highest violence risk. Professors Fahy and Hodgins have suggested that the presence of many risk factors in the history (a high H score on the HCR20) is a good indicator of a forensic patient, whereas it is not unusual to find elevated Clinical or Risk Management scores in general psychiatry patients who are acutely unwell. This may be an over-simplification but it is a good starting point. Most of the high H scores in this sample did not have input from forensic services. Case 2 was an exception, where there had been 64 a recent opinion from a forensic psychiatrist that greatly strengthened the position of the team at the Inquiry. Recommendation: The basis for referral of cases between general and forensic teams should be a structured clinical assessment of violence risk. Recommendation: A forensic opinion should be sought in respect of those patients with the highest level of violence risk. Recommendation: Forensic teams should manage patients in the community with a higher level of supervision than the general team can provide. They should intervene earlier when there is deterioration, because of the higher background risk of violence. Diagnosis and the Medical Model Diagnosis is important in risk management because the identification of a mental illness implies the availability of an effective treatment. However, diagnosis is less important in the assessment of violence risk and is correctly considered as just one of many factors. In many of these cases, a lot of time and effort was spent on the question of the correct diagnosis, whilst failing to recognise the violence risks involved in the case, irrespective of the precise diagnosis. It is often difficult and sometimes impossible to arrive at a definitive diagnosis in psychiatry, so the team needs to be able to assess and manage violence risk in conditions of diagnostic uncertainty. If there is an adequate risk management plan in place, based on a provisional diagnosis, a definitive label can safely wait. There were particular problems over the diagnosis of drug-induced psychosis, which seemed to be approached with an emphasis on voluntary intoxication, and insufficient recognition that it is a psychosis with the same potential for disaster as in schizophrenia. Also, as many if not most patients with schizophrenia use drugs, usually to the detriment of their mental state, the distinction between schizophrenia and drug-induced psychosis is often impossible and of little practical importance. Personality disorder also presents difficulties, as it was too often seen as precluding the need for a full assessment of risk. Recommendation: Violence risk assessment should be undertaken early in a patients contact with services. It will always be subject to revision as new 65 information emerges but it should never depend on the presence of a specific diagnosis alone. Recommendation: The diagnosis of drug-induced psychosis should be discouraged or abandoned. Involvement of Carers This is a complex issue, as illustrated by those cases in which carers opposed necessary intervention or participated in antisocial behaviour that made violence more likely. Notwithstanding these difficulties, there was a worrying lack of involvement of carers in some cases, and in others their clearly expressed concerns about violence were ignored. This is a particularly serious failing given that carers or relatives are most exposed to violence risk. Recommendation: Concern about violence risk should be shared openly with patients and with carers whenever possible, supported by copies of care plans and other relevant documents. Recommendation: The assessment of violence risk should be thoroughly reviewed whenever the carer appears more worried than the team. The Inevitability of Violence I have focussed on ways in which care can be improved but the message from some of these cases is that serious violence cannot always be predicted or prevented. This is particularly the case when it is coincident with the mental illness rather than a consequence of it. The policy of social inclusion means that services now treat patients with a high background risk of violence, who would probably have been excluded twenty years ago. Documents such as “Personality Disorder: No Longer a Diagnosis of Exclusion” will inevitably lead to an increase in serious violence by patients, by making more violent people patients. When reviewing Inquiry reports one is confronted by the unfairness of some comments made with the benefit of hindsight, and the consequent damage to morale in general, as well as to the staff directly involved. Recommendation: The Department's strategy for managing serious violence by patients needs to recognise that it is sometimes inevitable and should offer support to staff as well as reassurance to the public. 66 Headline Recommendations Recommendation: Mental health teams need to develop a common way of formulating, describing and communicating violence risk. Recommendation: All mental health teams should have access to a structured clinical assessment of violence risk and should incorporate its findings into the care management of patients with a history of violence.
Recommendation: There is a need for legal powers to allow compulsory treatment in the community of patients with a serious mental illness and a history of violence and non-compliance.
Recommendation: Substance misuse problems in patients with severe mental illness and a risk of violence should be assessed and managed within a structured clinical risk management plan. Proper consideration should always be given to the possibility of using the Mental Health Act in such patients. Recommendation: In patients subject to a restriction order there should always be consideration of setting conditions relating to abstinence from drugs or alcohol and the standard procedure should be immediate recall if that condition is breached.
Recommendation: There should be early intervention when there are signs of deterioration or risky behaviours in patients with a history of violence. This principle should rarely if ever be ignored, even when a patient is well known to the service. If in doubt, the team should err on the side of caution. Recommendation: When dealing with patients with a history of violence and serious mental illness, CPA meetings should set clear, operational criteria for intervention. These criteria should be communicated to patients and carers so they have clear expectations. 67 Recommendation: The basis for referral of cases between general and forensic teams should be a structured clinical assessment of violence risk. Recommendation: A forensic opinion should be sought in respect of those patients with the highest level of violence risk. Recommendation: Forensic teams should manage patients in the community with a higher level of supervision than the general team can provide. They should intervene earlier when there is deterioration, because of the higher background risk of violence.
Recommendation: Violence risk assessment should be undertaken early in a patients contact with services. It will always be subject to revision as new information emerges but it should never depend on the presence of a specific diagnosis alone. Recommendation: The diagnosis of drug-induced psychosis should be discouraged or abandoned.
Recommendation: Concern about violence risk should be shared openly with patients and with carers whenever possible, supported by copies of care plans and other relevant documents. Recommendation: The assessment of violence risk should be thoroughly reviewed whenever the carer appears more worried than the team.
Recommendation: The Department's strategy for managing serious violence by patients needs to recognise that it is sometimes inevitable and should offer support to staff as well as reassurance to the public. 68 |