Abstract
This study aimed to estimate the prevalence of mental disorder in offenders convicted of serious violence, examine their social and clinical characteristics, and compare them with patients convicted of homicide. We examined a national clinical survey of all people convicted of serious violence in England and Wales in 2004. Mental disorder was measured by contact with mental health services within 12 months of the offense. Of the 5,966 serious violent offenders, 293 (5%) had been in recent contact with mental health services. Personality disorder (63, 22%) and schizophrenia (55, 19%) were the most common diagnoses. Most had previous convictions for violence (168, 61%). Seventy-two (25%) patients were at high risk of violence and 34 (49%) were not subject to the Care Programme Approach. Compared with serious violence offenders, homicide offenders were more likely to have been patients (293, 5% vs. 65, 10%; p < .01). We conclude that patients were responsible for a small proportion of serious violent offenses; however, high-risk patients require closer supervision, and regular inquiry about changing delusional beliefs, thoughts of violence, and weapon carriage.
Introduction
There were an estimated 1.9 million acts of violence reported in the Crime Survey for England and Wales in 2012 (Office for National Statistics, 2013). The U.K. Government’s action plan for reducing serious violent crime identified key targets: the reduction of gun crime and gang-related violence, knife crime, and sexual and domestic violence. The government also identified individual risk factors associated with violent behavior such as alcohol abuse and exposure to violence at an early age, for example, through domestic violence (Home Office, 2008). The link between criminal offending and mental illness has long been established. Previous research has shown that people with a major mental disorder have a small but increased risk of being violent (Arseneault, Moffitt, Caspi, Taylor, & Silva, 2000; Brennan, Grekin, & Vanman, 2000; Kooyman, Dean, Harvey, & Walsh, 2007; Lindqvist & Allebeck, 1990; Link & Stueve, 1994; Tiihonen, Isohanni, Rasanen, Koiranen, & Moring, 1997; Wessely, Castle, Douglas, & Taylor, 1994). The risk of violence has also been shown to increase with comorbid substance abuse (Fazel, Langstrom, Hjern, Grann, & Lichtenstein, 2009). Numerous studies have reported violence within a patient sample (Link, Andrews, & Cullen, 1992; Monahan & Applebaum, 2000; Steadman et al., 1998) or mental illness in violent prisoners (Eronen, Hakola, & Tiihonen, 1996; Taylor & Gunn, 1984; Teplin, 1990).
In the few studies that have investigated the prevalence of violence by people with mental disorder, attributable risk was shown to be between 3% and 5% (Swanson, 1994). It is important to accurately establish the prevalence of mental illness in a community sample of violent offenders to reliably inform public perception and reduce stigma. Previous research has highlighted the importance of managing patients considered at high risk of violent offending through the use of the Care Programme Approach (CPA; Appleby et al., 2006; Swinson, Flynn, Kapur, Appleby, & Shaw, 2010). CPA was introduced by the Department of Health in 1991 to provide specialist mental health services with an effective structure to care for patients with complex needs, such as risk of harm to self or others, increased disengagement, comorbidity, and those requiring multi-agency support. It involves (a) assessment of need, (b) the formation of a care plan, (c) the appointment of a care coordinator, and (d) regular multidisciplinary review (Department of Health, 1999).
The definition of a mental disorder in the literature can be limiting, with several studies reporting either a combined group of mental disorders, or specific diagnosis, particularly schizophrenia/psychosis (Elbogen, Van Dorn, Swanson, Swartz, & Monahan, 2006; Fazel et al., 2009; Kooyman et al., 2007; Silver, Felson, & Vaneseltine, 2008). A further limitation as highlighted by Vinkers, de Beurs, Barendregt, Rinnie, and Hoek (2011) has been the definition of “violent offending,” which is often over inclusive (i.e., any physical assault) or too specific (i.e., just homicide).
The theory suggesting that mental illness is a more significant feature in homicide than serious violence (Silver et al., 2008) has also been considered in this study. Based on attribution theory, it is hypothesized that the more deviant and abhorrent the act, the stronger the association with mental illness. Silver et al. (2008) tested the hypotheses by examining mental illness in those committing homicide and sexual offenses, compared with other physical assaults. They found homicide was more common than physical assault in those with minor mental health problems, but no relationship was found with severe mental illness. In light of this evidence, the early identification of mental health patients at risk of offending is essential to prevent behaviors escalating into serious violence.
The study has been undertaken as part of the National Confidential Inquiry Into Suicide and Homicide by People With Mental Illness (NCISH). The NCISH is a large-scale epidemiological study that provides a rich source of clinical data using a unique national database (Shaw et al., 1999). The methodology used by the NCISH was applied to a cohort of serious violent offenders. The aims of this study were to estimate the prevalence of mental disorder in offenders convicted of serious violence; to undertake an examination of the social, offense, and clinical characteristics of patients who committed serious violence, particularly those considered high risk; and to compare patients who committed serious violence with patients who were known to have committed homicide. The rational for these aims was to update the literature by reporting the prevalence of serious violence committed by people with mental illness in a current national population sample. We also wanted to present the diagnosis, clinical care, management, and treatment received by violent patients under the care of mental health services, and to determine whether patients committing serious violence were clinically different from those who committed a homicide.
Method
Design
The NCISH was established by the U.K. Department of Health in 1992 following a number of high-profile homicides by mental health patients, and criticism of the provision of community care for patients with severe mental illness. The main aims of the NCISH are to establish the rate of mental disorder in patients in contact with mental health services who died by suicide or had been convicted of homicide, and to recommend measures to reduce the risk of suicide and homicide by people under the care of mental health services. In this study, the NCISH methodology was applied to a national population of offenders convicted of serious violent offenses in England and Wales, and as such, this study was an extension of the NCISH’s ongoing core research into suicide and homicide. Our data collection process is outlined in Figure 1.

Data collection process.
Sample
Comprehensive national sample of serious violent offenders
We were notified by the Home Office Offender Index of all convictions for serious violent offenses in England and Wales between January 1, 2004, and December 31, 2004, with a total of 6,474 offenses committed by 5,966 individuals. Demographic information and details of the offender and offense were available for all perpetrators. The authors selected four offenses defined as “specified violent offenses” under Schedule 15 (section 244) of the Criminal Justice Act 2003, chapter 44 (HMSO, 2003). We specifically focused on Offenses Against the Person, which are considered to represent a potential “near miss homicide”; these were attempted murder, malicious wounding “and other like offenses” (Section 20), wounding or other act endangering life (Section 18), and threats or conspiracy to murder. Our contacts within the Police Service advised the inclusion of “threats to kill,” as this offense has long been considered an indicator of future violence, and this has also been previously reported in the literature (MacDonald, 1963; Planansky & Jihnson, 1977). Vinkers et al. (2011) in a Dutch study of over 20,000 pre-trial forensic psychiatric reports between 2000 and 2007 found threats of homicide and homicide attempts to be significantly associated with mental disorders such as personality disorders and psychotic disorders. It has been observed in an Australian study of adult forensic patients who had threatened to kill another person that of the 144 patients in their sample, over 20% had committed an assault including 1 homicide within 12 months of making the threat (Warren, Mullen, & Ogloff, 2011).
Greater Manchester Police provided a detailed history of previous offending on all the serious violence perpetrators in the study. The data were extracted from the Police National Computer.
Identification of mental health service contact
Identifiable information on the 5,966 perpetrators who had been convicted of a serious violent offense in 2004 was sent to all hospitals and mental health service providers in England and Wales, within the perpetrators’ district of residence and adjacent districts. We were notified by our contacts within the hospital of the date of the person’s last contact with mental health services (if any), and the name of the supervising clinician. The cases with mental health contacts will be henceforth referred to as “patients.”
Collection of clinical data
Following the confirmation of contact with mental health services, a questionnaire was sent to the supervising clinician. The questionnaire consisted of 10 sections containing questions on priority patients, demographic information, psychiatric and forensic history, details of the index incident, details of other violence incidents, psychiatric in-patient violence, outpatient community patients, treatment and compliance, last contact with mental health services, clinicians’ views on prevention, and a final section for additional information. This questionnaire required clinicians to extract factual information from case notes, and is not based on opinion. The questionnaire was completed by the supervising clinician at the time of the incident in conjunction with the multidisciplinary team, based on their personal knowledge of the patient and from a review of the patients’ case notes. We consider this to be the most reliable source of information regarding patient care and diagnoses. Diagnoses were made by the clinicians who treated the patients in accordance with the International Classification of Disease–10th Revision (World Health Organization, 1992). The questionnaire used in this study was a variant of the NCISH’s homicide questionnaire. A sample is available on the NCISH website (www.bbmh.manchester.ac.uk/cmhr/research/centreforsuicideprevention/nci/FAQs/Inquirymethodology/Coresuihomimethod/).
An audit of the NCISH’s case ascertainment method and inter-rater reliability of the questionnaire was previously undertaken across 16 hospitals. The identification of a mental health contact ranged from 98% to 100%. Agreement on the validity of key questions (including diagnoses) ranged from 90% to 100% (Windfuhr et al., 2008).
Serious Violence Sample
Between January 1, 2004, and December 31, 2004, in England and Wales, 307 offenders were known to have been in contact with mental health services 12 months prior to the offense. This figure of 307 is probably an underestimation as it is likely that some service contacts were not identified, particularly for those individuals who had been under services in one locality and moved areas without making contact with services in their new district of residence. Of the 307 patients, data were not obtained on four cases due to missing case notes. Participation was not mandatory and a further 10 clinicians did not wish to be involved; therefore, we had a response rate of 95%.
NCISH Homicide Sample
The NCISH focuses on two core areas of research: suicide and homicide. To date, the NCISH has collated data on 95,000 suicides and 10,000 homicides over 17 years. A sample of homicide cases from the NCISH’s national clinical survey was used as a comparator with people convicted of serious violence for the same time period. The data were extracted from the NCISH homicide database on all homicide perpetrators convicted between January 1, 2004, and December 31, 2004, in England and Wales. Of the 655 convicted homicide perpetrators, 65 were confirmed as having been in contact with mental health services (i.e., mental health patients) within 12 months of the offense. Clinical data were available on all of these patients, with a questionnaire response rate of 100%.
The NCISH clinical survey method of data collection was used for both the serious violence sample and the comparison homicide group. These 65 homicide patients were used as a comparison group only and were not included in the main “serious violence” analysis.
Measures
Previous Violence
We defined a history of previous violence as convictions for homicide, attempted murder, grievous bodily harm, actual bodily harm, threats to kill, and other violent acts such as common assault. These are listed as violent offenses under the Offenses Against the Person Act 1861.
High Risk
The study distinguishes a group of patients who posed a significant risk to others, referred to as “high-risk patients.” The definition is based on the Department of Health’s white paper “Reforming the Mental Health Act—Summary” (Department of Health, 2000) that refers to high-risk patients with mental illness who had been “detained under civil powers, and others who are remanded or convicted offenders.” Our definition of high-risk patients included those with severe mental illness, a history of violence, or previous detention under mental health legislation, but does not include patients with severe personality disorder. Information on previous detention was extracted from the clinical questionnaire and convictions for previous violence from antecedent data via the Police National Computer. The rationale for the definition of high risk was based on potential dangerousness of patients based on illness, previous violence, and involuntary detention.
Severe Mental Illness
Severe mental illness is an accepted clinical term. In this study, we have adopted a broad definition based on diagnoses of schizophrenia and other delusional disorders and affective disorders. We have not assessed the duration of illness or level of dysfunction in this definition.
Data Analysis
To address the first aim of the study, general population and patient rates were calculated using mid-year population estimates (aged 10 and over) as a denominator. These data were obtained from the Office for National Statistics. In addition, descriptive statistics were used to describe the proportion of serious violence in England and Wales committed by mental health patients. The second aim was addressed by using descriptive statistics to examine the demographic, behavioral, offending, and clinical characteristics of these patients. For the third aim, inferential statistics were used to test for differences between patients convicted of serious violence and patients who had been convicted of homicide. The findings are presented in tables as proportions with 95% confidence intervals (CIs). Chi-square was used to test associations between the independent variables (demographic, behavioral, offending, and clinical characteristics) and the dependent variable (serious violence or homicide outcome). Logistic regression analyses were also conducted to test for association. Odds ratios (ORs) were calculated and the proportions are presented with 95% CIs.
As this study mostly reported descriptive statistical analysis, we did not undertake sensitivity analysis to measure the effect of missing data on these findings. However, we found the proportion of missing data to be acceptable. Analysis showed a median of 2% missing data for demographic/behavioral and offense variables and 6% for clinical data. If an item of information was not available or not known for a particular case, the case was removed from the analysis of that item; the denominator in all estimates was the number of valid cases. Data were analyzed using STATA Version 11 (StataCorp, 2009).
Ethical Approval
We sought permission to extend the NCISH methodology to include serious violence from the Patient Information Advisory Group (now the National Information Governance Board for Health and Social Care). Authorization was obtained under Section 251 of the National Health Service Act 2006 (HMSO, 2006). This was granted in the interest of improving patient care, permitting confidential and identifiable information to be obtained without informed consent from a large cohort. Ethical approval was obtained on September 9, 2003. The study is also registered under the Data Protection Act 1998 (Parliament, UK, 1998).
Results
Serious Violence in the General Population
A total of 6,474 offenses were committed by 5,966 individuals, with a rate of 12.7 offenders per 100,000 population. The most common offense was malicious wounding (see Figure 2). The majority of perpetrators were male (5,480, 92%). The median age of offenders was 26 (range = 17-88), with nearly half unemployed at the time of the offense (2,749, 47%). Most of the perpetrators were born in the United Kingdom (5,264, 90%).

Number of serious violent offenses in England and Wales, 2004, by general population and patients.
Serious Violence Among Mental Health Patients
Prevalence of patients committing serious violence
Three-hundred and seven offenders were known to have been in contact with mental health services 12 months prior to the offense, with a 1-year rate of 0.66 per 100,000 population. The main findings were based on the 293 patients for whom clinical data were received. Patients committed a total of 354 offenses, the most common being malicious wounding (see Figure 2).
Demographic and behavioral characteristics
The demographic and behavioral characteristics of patients are presented in Table 1. Patients were predominantly male, unmarried, and unemployed, with a median age of 30. Both alcohol misuse and drug misuse were frequently reported. The majority had previous convictions, and the number of court appearances ranged from 1 to 213 (with a median of 8). Nearly two thirds had convictions for previous violence, including four patients who had previously been convicted of homicide. Half had previously received a custodial sentence.
Demographic, Behavioral, and Offense Characteristics of Patients Committing Serious Violence and Homicide in England and Wales, 2004.
Note. CI = confidence interval.
Offense characteristics
Figure 3 shows the number of offenses by primary diagnosis. Attempted murder was the least common violent offense committed by patients; all four of the patients convicted of this offense had severe mental illness.

Patients: Number of serious violent offenses by primary diagnosis.
Clinical characteristics
The clinical characteristics of offenders are presented in Table 2. The most common diagnosis was personality disorder. Severe mental illness (schizophrenia or affective disorder) was diagnosed in 98 cases (34.5%), in 23 (23%) of these patients, the onset of illness had been in the previous year. Of the 55 patients with schizophrenia (0.9% of all serious violent offenders), 47 (85%) had dual diagnosis (severe mental illness and drug and/or alcohol misuse and/or dependence). The 55 patients with schizophrenia committed 71 offenses, the most common being malicious wounding (29, 41%). Patients with personality disorders were most frequently convicted of threats to kill (52, 59% of 88 offenses), and malicious wounding was the most common violent offense committed by patients with affective disorders (21, 46% of 46 offenses; see Figure 3). A total of 78 (27%) patients had dual diagnosis.
Comparison of Clinical Characteristics of Patients Committing Serious Violence and Homicide in England and Wales, 2004.
Note. CPA = care programme approach; MH Act = Mental Health Act.
Most patients had their last contact with general psychiatry services (209, 72%); 7 (2%) were last seen by forensic services. Over a quarter of patients missed their last appointment with mental health services. A fifth refused treatment with medication within a month of the offense. The majority of patients (231, 80%) were not receiving care under the CPA at a level requiring regular multidisciplinary reviews. Clinicians stated that 34 (16%) of the violent assaults were preventable. Factors that could have made the violence significantly less likely included better patient compliance with medication (75, 32%), closer patient supervision (45, 19%), availability of dual diagnosis services (44, 18%), and better liaison between services (44, 18%).
High-risk patients
Our definition of high-risk patients included those with severe mental illness (schizophrenia or affective disorder) and a history of violence or previous detention under mental health legislation. Of the 72 (25%) high-risk patients, 13 (18%) were in-patients at the time of the offense, and 7 (10%) had been discharged from in-patient care within 3 months before the offense. Twenty (29%) had disengaged with services, and 22 (34%) were non-adherent with medication. A total of 57 (83%) had a history of alcohol and/or drug misuse, and 34 (49%) were not subject to enhanced CPA. Of those high-risk patients not under the CPA, 11 (32%) missed their last appointment, and 6 (21%) were non-adherent with their medication in the month before the offense. Most were under the care of general adult services (89%).
Serious Violent Patients Versus NCISH Homicide Patients
We compared the 293 patients convicted of serious violence with 65 patients convicted of homicide (see Tables 1 and 2). Schizophrenia was shown to be 3 times more likely in patients who committed homicide than in patients who committed serious violence (16, 2% vs. 55, 0.9%; ORs = 2.7). Patients also committed proportionally more homicides than serious violence (65, 10% vs. 293, 5%; p < .01). Patients who committed homicide were more likely to have missed their last appointment with mental health services, and their long-term risk of violence was more commonly estimated to be low or none by the clinicians responsible for their care. Serious violence patients were more frequently detained under the Mental Health Act. Serious violence patients with severe mental illness were also more likely to have dual diagnosis compared with homicide offenders with severe mental illness (78, 80%, vs. 14, 54%; p > .01). There were no significant differences in the proportion of patients with previous convictions for sexual offenses, criminal damage, or carrying offensive weapons. We found a borderline significant difference in the proportion of patients with previous violent convictions (p = .05), which was more common in patients who committed serious violence.
Discussion
Main Findings
We reported findings from a large population-based study examining perpetrators convicted of serious violence in England and Wales. We found that 5% of offenders were patients who had been in recent contact with mental health services. Most of the violence by mental health patients was committed by young males, who had a history of violence, alcohol abuse, and drug misuse, and had previously been imprisoned. Personality disorders and schizophrenia were the most common diagnoses. The commonest offense among those with personality disorder was threats to kill, whereas malicious wounding was more common in patients with severe mental illness.
It has previously been suggested that offenders who commit serious violence had an increased relative risk of committing homicide (Soothill, Francis, & Liu, 2008), and that there is a similarity in the etiology of people who commit serious violence and homicide (Gottfredson & Hirschi, 1990). If the criminological theories are carried, one would assume that the characteristics of serious violence offenders and homicide offenders would be similar, including rates of contact with mental health services. We found there to be comparable demographic and behavioral characteristics; however, there were some interesting differences. Strikingly, the proportion of patients in contact with mental health services was significantly lower in the serious violence sample. However, serious violence patients were more likely to have a history of previous violence and previous detention under the Mental Health Act compared with homicide patients, suggesting that a higher proportion had previously presented as a danger to themselves or others. Serious violence patients with severe mental illness were also significantly more likely to have comorbid substance misuse than homicide patients. Substance abuse comorbidity has been found to increase risk of violence in patients with schizophrenia (Fazel et al., 2009), and Volavka and Swanson (2011) suggested substance abuse treatment be incorporated into risk management plans of patients with severe mental illness.
How Do We Explain Higher Rates of Mental Disorder in Those Convicted of Homicide?
It has been asserted that perpetrators of homicide and those who commit serious acts of violence are exhibiting the same behavioral characteristics, the only difference being the “legal label” applied to a fatal or non-fatal outcome (Harries, 1990). Proponents of this theory suggest the reason why some acts of violence result in the loss of life could be associated with the lethality of the method used, the response of the police, and the timely receipt of medical attention for the victim. Conversely, an alternative hypothesis suggests that these offenses can be distinguished by the perpetrators’ motivation, whether they had the intent to kill or to injure (Kleck, 1991). Silver et al. (2008) tested the hypothesis that the more serious or heinous the crime, the more likely the perpetrator will have severe mental illness, but an association was only observed in those with minor mental disorder. Therefore, it could be argued that the higher proportion of mental disorder in homicide compared with other violence may be linked to the severity of symptoms in some perpetrators, which leads them to kill rather than injure, but this would only account for a minority of cases.
To put our findings in context, the prevalence of psychotic disorders in the general population of men aged 25 to 35 years in England was 0.6% in 2007 (National Centre for Social Research, 2009). Although not directly comparable, we found a higher proportion of serious violent and homicide offenders with schizophrenia in our general population sample (0.9% and 2%, respectively). Contact with mental health services before the offense was also more common among perpetrators of homicide than serious violence. The difference could be explained by a psychotic mental state at the time of the offense. In a study of homicide perpetrated by men with schizophrenia, Joyal, Putkonen, Paavola, and Tiihonen (2004) found that 60% were motivated by psychotic symptoms. Meehan et al. (2006) also observed that in homicides committed by people with schizophrenia (59% were experiencing delusions at the time of the offense), psychotic symptoms may place some patients at risk of committing violent acts due to a sense of personal threat and loss of control. Therefore, it is possible that there is a group of people with psychosis who when violent are more likely to have a lethal outcome, and this is related to specific delusions, intent, and the carriage of lethal weapons (Rodway et al., 2009). Evidence has shown that escalating psychotic symptomology (paranoia and command hallucinations) increases homicidal ideation and intent (Kaplan & Sadock, 1995; Schwartz, Petersen, & Skaggs, 2001). In addition, physical assaults motivated by psychosis and manifesting extreme violence are more likely to have a fatal outcome.
Limitations
The study describes a large national consecutive case series of offenders convicted of violence against the person who had been in recent contact with mental health services prior to the offense. The NCISH’s established methodology has proven validity and representativeness, with mental health contact identified in 98% to 100% of cases. The comprehensive clinical data set enabled an in-depth analysis of the care and treatment received by patients prior to the incident. The findings should be considered in the context of the methodological shortcomings. Reporting data on convicted offenders rather than those suspected or charged with an offense ensured that the individuals were found guilty; however, this may have led to an underestimation of violence by people with mental illness, as not all incidents are reported or prosecuted. Contact with mental health services as a measure of mental illness may also underestimate the extent of violent offenders who had a mental illness at the time of offense. This definition does not include patients whose care was managed by a general practitioner or primary care teams in the community. Unlike the NCISH homicide study, we did not obtain reports from psychiatric assessment used in court; therefore, a comparison of symptoms of mental illness at the time of the offense between serious violence and the NCISH homicide patients was not possible. Furthermore, we were unable to provide a detailed examination of personality disorder types, which would have been a valuable addition to the results. Finally, as this was largely a descriptive study, we were unable to test hypotheses or establish causal factors that may have led to the violent incident.
Implications for Service Provision
Our findings have shown that mental health patients were responsible for a small proportion of serious violence committed nationally. This is a key message that can be used to reduce stigma and challenge the public perception that people with mental illness are dangerous. Of the patient group who committed serious violent acts, our study highlights implications for mental health service management. We reported that over a third of patients who committed a serious act of violence had severe mental illness, a quarter were defined as high-risk, having severe mental illness and previous convictions for violence or detention under the Mental Health Act. However, our findings show that despite the high-risk status of these patients, CPA was under-used and disengagement from treatment and non-adherence with medication was not uncommon, a finding consistent with a study of homicide patients by Swinson et al. (2010). To reduce the risk of adverse incidents, care plans should be in place to manage risk and specify action if a crisis does occur. New guidance for the use of CPA was published by the Department of Health in 2008 (Department of Health, 2008). Whether this has led to fewer high-risk patients disengaging with services is unknown, but the evidence from this study suggests that better risk management for these patients is required.
The majority of the patients in this study had their last contact with general adult services rather than the forensic service. Future research into the use of risk assessment tools such as the Historical, Clinical, Risk Management–20 (HCR-20) in this setting may help to identify patients most at risk of committing serious acts of violence and inform future management plans. Staff need to ensure that high-risk patients maintain contact with services and that staff regularly inquire about changing delusional beliefs, thoughts of violence, and weapon carriage.
Footnotes
Acknowledgements
The study was carried out as part of the National Confidential Inquiry Into Suicide and Homicide by People With Mental Illness. We acknowledge the help of district directors of public health, health authority and trust contacts, and consultant psychiatrists for completing the questionnaires. We acknowledge the contribution made by members of the National Confidential Inquiry: Kelly Hadfield, James Burns, Professor Navneet Kapur, Dr. Kirsten Windfuhr, Dr. Alyson Williams, Dr. Isabelle. M. Hunt, Dr. David While, Alison Roscoe, Saied Ibrahim, Rebecca Lowe, Phil Stones, Julie Hall, and Huma Daud.
Declaration of Conflicting Interest
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Louis Appleby is the National Clinical Director for Health and Criminal Justice for England.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the Healthcare Quality Improvement Partnership (HQIP), Department of Health, UK.
