Domestic homicide is defined as the killing of a person age 16 or older by a family member or a current or former partner. This accounts for a substantial proportion of homicides in high-income countries, particularly for female victims (
1). Approximately one-third to half of female homicide victims are killed by a current or ex-partner in the United States and United Kingdom, and 5%−11% are killed by another family member (
2,
3). Among male homicide victims, 2%−6% are killed by a current or former partner, whereas 3%−5% are killed by a family member (
2,
3).
Studies in high-income countries suggest that between 8% and 23% of all homicide perpetrators are mentally ill at the time of offense (
4–
8). In England and Wales the prevalence of mental illness at the time of homicide has been reported to be 10% (
9). However, it is not clear whether perpetrators of adult domestic homicide have a similar prevalence of mental illness (
10–
13). Research has been limited by inadequate assessment of mental illness, the omission of perpetrators with diminished responsibility and unfit-to-plead or insanity verdicts, the exclusion of ex-partner homicides, and small samples (
11,
12,
14). Most studies of adult family homicide have been conducted with even smaller samples (<50) and in psychiatric settings, introducing bias toward associating the offense with serious mental illness. These studies suggest that a high proportion of parricidal offenders (offenders whose victim is a parent) had a diagnosis of a psychotic illness and had prior contact with psychiatric services (
15–
21).
Using data from the National Confidential Inquiry Into Suicide and Homicide by People With Mental Illness (National Confidential Inquiry [NCI]), a data set that includes data for all perpetrators convicted of homicide in England and Wales (murder, manslaughter, infanticide, and unfit-to-plead or insanity verdicts), we can report on prevalence of mental illness with domestic homicide, including among both ex- and current partners. We aimed, first, to examine the prevalence of current and lifetime mental illness and contact with mental health services among convicted perpetrators of intimate partner and adult family homicide and second, to investigate the distribution of characteristics associated with intimate partner and adult family homicide by offenders with and without symptoms of mental illness at the time of offense. We hypothesized that the prevalence of symptoms of mental illness at the time of the homicide would be 10%.
Methods
The NCI database obtains information on all convicted cases of homicides in England and Wales (
22). This study used NCI data from convicted perpetrators of intimate partner and adult family homicide in England and Wales from January 1, 1997, to December 31, 2008. The data set does not include cases of homicide-suicide.
Data collection for the NCI has three stages (
9). First, the Home Office routinely collates data about people convicted of homicide in England and Wales. The NCI was notified of all homicide cases within the study period. Demographic data and details of the offense, verdict, and sentence (“disposal”) were provided by the Home Office Statistics Unit, and data on previous offenses were obtained from the Police National Computer. In England and Wales, a verdict of murder requires proof that there was intent to kill or cause grievous bodily harm, and it carries a mandatory life sentence. A verdict of manslaughter can be brought if it is determined that there was no intention to kill or if the defendant is deemed not responsible for his or her actions. A verdict of manslaughter on the grounds of diminished responsibility is allowed if the perpetrator experienced an “abnormality of mind” at the time of killing. Not guilty by reason of insanity is a rarely used defense (
23). A judge may declare a defendant unfit to plead if he or she is unable to comprehend the course of criminal proceedings; in a majority of such cases the defendant is expected to respond to medical treatment, with the trial taking place at a later date. Infanticide refers to the killing of a child less than one year old by his or her mother; these cases are not included in this study.
Second, the NCI requests for all perpetrators the psychiatric reports prepared by psychiatrists for court. The reports provide information on demographic characteristics, alcohol and substance misuse, history of violence, and mental state at the time of offense. Psychiatric diagnoses are classified according to
International Classification of Diseases, 10th Revision diagnostic criteria (
24). The proportion of homicide perpetrators with psychiatric reports has decreased annually since 2001, when psychiatric assessments for homicide perpetrators ceased to be mandatory. Although it is possible that perpetrators with mental illness may have been missed, recent evidence from the NCI data set shows an increase in the number of perpetrators with schizophrenia and other delusional disorders and of perpetrators who had symptoms of mental illness at the time of offense (
25). Perpetrators without reports (those who did not warrant a psychiatric assessment) are considered not to have serious mental illnesses (
26).
Finally, details of each perpetrator are forwarded to the National Health Service (NHS) and independent-sector providers of mental health services in the perpetrator’s district of residence. If hospital records show that the perpetrator had been in contact with mental health services, the supervising clinician is asked to complete a questionnaire that includes information on demographic characteristics, clinical history, aspects of care such as detention under the Mental Health Act (1983), Care Programme Approach (CPA) support (care coordination and regular multidisciplinary review for people who have been detained or have complex mental health needs) (
27), details of final contact with services, and views on the prevention of the offense. The return rate of questionnaires sent to consultant psychiatrists was 96%. An assessment of the accuracy of hospital checks showed that 97% of patients in contact with services in the previous year were detected (
28). Discrepancies between psychiatric diagnoses in the court report and mental health service questionnaire are resolved by consensus agreement by senior clinical members of the NCI team (
9).
The NCI refers to symptoms of psychosis and depression at the time of homicide as symptoms of mental illness at the time of offense. We use the term “intimate partner homicide” to refer to the homicide of a current or ex-partner or spouse age 16 or older and the term “adult family homicide” to refer to the homicide of a parent or of a sibling or child age 16 or older. There is a lack of evidence on whether the demographic, clinical, behavioral, or offense characteristics of these groups are sufficiently similar to be aggregated as “domestic homicide”; findings are therefore presented separately for intimate partner and adult family homicide.
Analysis
Descriptive statistics were calculated for the sociodemographic, behavioral, clinical, service use, and offense characteristics of perpetrators of intimate partner and adult family homicide. Characteristics were selected for analysis on the basis of evidence from the existing literature. Pearson’s chi square tests were used to compare sociodemographic, behavioral, and offense characteristics between perpetrators of intimate partner and adult family homicide and also between perpetrators with and without symptoms of mental illness at the time of offense. Fisher’s exact tests were used for small cell counts (<5). All analyses were conducted in Stata 11.0 (
29).
Ethics
The NCI was granted an exemption under Section 251 of the NHS Act 2006 and was authorized to use identifiable and confidential patient information without informed consent in the interest of improving patient care (
30,
31). Data were deidentified for analysis and are reported in an anonymous and aggregated format.
Results
Overview of domestic homicide case series, 1997–2008
The NCI was notified of 7,124 homicide convictions in England and Wales between January 1, 1997, and December 31, 2008. A total of 1,180 (17%) convictions were related to intimate partner homicide (958 current partners and 222 former partners), and 251 (4%) were related to adult family homicide (213 parents, 12 siblings, and 26 adult children). Psychiatric reports were obtained for 67% (N=791) of intimate partner and 75% (N=188) of adult family homicide perpetrators.
As shown in
Table 1, most perpetrators were male. The proportion of female victims was higher for intimate partner homicide (76%) than for adult family homicide (39%). The mean±SD age of perpetrators of adult family homicide (32.2±.8) was younger than that of perpetrators of intimate partner homicide (39.6±.4). A higher prevalence of unemployment was recorded for perpetrators of adult family homicide (55%) than for perpetrators of intimate partner homicide (39%).
Mental illness and use of mental health services
More than 90% (N=289) of perpetrators with symptoms of mental illness at the time of the index offense also had a lifetime diagnosis of a mental illness (91% [N=210] for perpetrators of intimate partner homicide and 93% [N=79] for perpetrators of adult family homicide). Thus a majority of those who had symptoms of mental illness at the time of offense had a lifetime history of mental illness (longstanding illness rather than a sudden onset of symptoms).
Intimate partner homicide
Mental illness at the time of homicide.
Of the 1,180 intimate partner homicide perpetrators, 20% had symptoms of mental illness at the time of offense: 7% experienced symptoms of psychosis, and 13% experienced symptoms of depression (
Table 2). Thirty percent (69 of 231) of perpetrators with symptoms of mental illness at the time of offense had been in contact with mental health services in the year before the homicide.
Lifetime mental illness.
One-third of perpetrators of intimate partner homicide had a lifetime diagnosis of a mental illness (N=378): 6% had received a primary diagnosis of schizophrenia and other delusional disorders, 17% of affective disorder, and 7% of personality disorder (
Table 2). Of those with a lifetime diagnosis of schizophrenia and other delusional disorders or affective or personality disorder, 44% (N=168) were not acutely ill at the time of homicide.
Contact with mental health services.
Overall, 14% (N=164) of perpetrators of intimate partner homicide had been in contact with mental health services in the year before the homicide (
Table 2), 9% (N=108) had been in contact less than 13 weeks before the homicide, and 3% (N=14) less than one week before the homicide. Of the 164 perpetrators who had been in contact with mental health services in the year prior, 42% (N=69) had symptoms of mental illness at the time of offense (including 21% [N=34] with symptoms of psychosis and 21% [N=35] with symptoms of depression). Twenty-two percent (N=36) were receiving support under CPA at the time of offense, 38% (N=62) had missed their last contact with services, and 13% (N=22) were thought by the responding mental health professional to have been noncompliant with medication in the month before the homicide. Supervising clinicians responding to the NCI reported in only 13 cases that patients were thought to pose a moderate or high immediate risk of violence to others at last contact.
Over half (91 of 164, 55%) of the perpetrators who had been in contact with mental health services in the year before the homicide had ever been admitted as psychiatric inpatients; 34 perpetrators (21%) had been discharged from psychiatric inpatient care in the three months before the homicide.
Characteristics associated with intimate partner homicide.
When comparing the sociodemographic characteristics of perpetrators with and without mental illness at the time of offense, we found that perpetrators of intimate partner homicide who had symptoms of mental illness at the time of offense were more likely to be older (χ2=40.3, df=6, p<.001), male (χ2=11.3, df=1, p=.001), and employed (χ2=4.1, df=1, p=.043). No differences were identified with respect to racial-ethnic minority group, marital status, or living arrangement.
With regard to behavioral characteristics, perpetrators with symptoms of mental illness at the time of offense were less likely to have previous convictions (χ2=39.7, df=1, p<.001), including convictions for violence, threats of violence, criminal damage, and possession of weapons. Perpetrators with symptoms of mental illness were less likely to have ever abused alcohol (χ2=6.9, df=1, p=.009). The prevalence of a history of self-harm was also lower among perpetrators with symptoms of mental illness at the time of offense (χ2=5.2, df=1, p=.022).
Adult family homicide
Mental illness at the time of homicide.
A third (85 of 251) of perpetrators of adult family homicide had symptoms of mental illness at the time of offense; 27% experienced symptoms of psychosis, and 7% experienced symptoms of depression (
Table 2). Thirty-six percent (36 of 85) of perpetrators who had symptoms of mental illness at the time of offense had been in contact with mental health services in the year before the homicide.
Lifetime mental illness.
Forty-five percent (N=113) of adult family homicide perpetrators had ever been diagnosed as having a mental illness: 28% had received a primary diagnosis of schizophrenia and other delusional disorders, 8% had a diagnosed affective disorder, and 5% had a diagnosed personality disorder (
Table 2). Of those with a lifetime diagnosis of schizophrenia or other delusional disorder or an affective disorder or personality disorder, 30% (N=34) were not acutely ill at the time of homicide.
Contact with mental health services.
Overall, 23% (N=57) of adult family homicide perpetrators had been in contact with services in the year before the homicide (
Table 2), 15% (N=38) had been in contact less than 13 weeks before the homicide, and 6% (N=14) less than one week before the homicide. Of the 57 perpetrators who had been in contact with mental health services in the year before the homicide, 63% [N=36] had symptoms of mental illness at the time of offense (including 56% [N=32] with symptoms of psychosis and 7% [N=4] with symptoms of depression). Twenty-five percent (N=14) were receiving support under CPA, 39% (N=22) had missed their last contact with services, and 26% (N=15) were thought by the responding mental health professional to have been noncompliant with medication in the month before the homicide. Supervising clinicians responding to the NCI reported in only two cases that patients were thought to have posed moderate or high immediate risk of violence to others at last contact.
Sixty-three percent (36 of 57) of perpetrators who had been in contact with mental health services in the year before the homicide had ever been admitted as a psychiatric inpatient. Seven perpetrators (12%) had been discharged from inpatient care in the previous three months.
Characteristics associated with adult family homicide.
When comparing the sociodemographic characteristics of adult family homicide perpetrators with and without symptoms of mental illness at the time of offense, we identified no differences in respect to sex, age, racial-ethnic minority status, marital status, or living arrangement. Perpetrators with symptoms of mental illness were, however, less likely to be employed (χ2=9.3, df=1, p=.002).
In regard to behavioral characteristics, perpetrators with symptoms of mental illness at the time of offense were less likely to have any previous convictions (χ2=11.1, df=1, p=.001) or convictions for violence or criminal damage. Perpetrators with symptoms of mental illness were also less likely to have ever abused alcohol (χ2=5.0, df=1, p=.026).
Discussion
Between 1997 and 2008, 20% of intimate partner homicide perpetrators and 34% of adult family homicide perpetrators in England and Wales had symptoms of mental illness at the time of offense, higher than has been reported in this setting among perpetrators of other types of homicide (
9). The prevalence of symptoms of mental illness was particularly high among adult family homicide perpetrators, with 27% experiencing psychotic symptoms. In contrast, perpetrators of intimate partner homicide who had symptoms of mental illness were more likely to have depressive symptoms at the time of offense. Indeed, our results show that perpetrators of intimate partner and adult family homicide differ substantially in terms of their sociodemographic, offense, and clinical characteristics.
Our results suggest, however, that policy makers are likely to face considerable challenges in identifying the risk of, or preventing, domestic homicide. Despite significant levels of mental illness at the time of offense and of lifetime mental illness, around a third had never been in contact with mental health services. Moreover, a majority of perpetrators who had a mental illness at the time of homicide had not been in contact with mental health services in the year before the homicide.
Among perpetrators who had used mental health services in the year before the homicide, factors that may be associated with increased risk included missing the last appointment with mental health services or noncompliance with medication, but numbers were small. Supervising clinicians reported that only 15 perpetrators of domestic homicide who had been in contact with services in the year before the homicide were thought to pose a moderate to high immediate risk of violence at last contact. Moreover, the prevalence of previous alcohol abuse and past violence (indicated by previous convictions for violent offenses)—known risk factors for domestic homicide (
11,
13,
32–
36)—was lower among perpetrators who had a mental illness at the time of homicide compared with those without one.
Although this study overcame several weaknesses of previous research into domestic homicide and mental illness (
12,
14), a number of limitations should be noted. Psychiatric reports were available for most cases and should have captured all cases of serious mental illness (that is, psychosis and depression). The prevalence of other mental disorders, such as personality disorder and substance use disorder, may, however, be underestimated. In addition, clinical information was based on clinical judgment rather than on standardized assessment. Furthermore, our analysis did not include cases of homicide-suicide; these cases have been reported elsewhere and were excluded because they were not convicted (
37). We were also unable to analyze the prevalence of a number of previously identified risk factors, including childhood abuse, domestic violence, cohabitation versus marriage, and gun ownership (
11,
12,
36). Shootings made up only 2% of domestic homicides in this sample, which is unsurprising in England and Wales, where gun ownership is low.
Conclusions
A significant minority of domestic homicide perpetrators had symptoms of mental illness at the time of the homicide, with adult family homicide more likely to be associated with psychosis. Most perpetrators were not in contact with mental health services in the year before the homicide and the prevalence of previously identified risk factors for domestic homicide was lower among perpetrators with symptoms of mental illness at the time of homicide compared with those without. Reduction of homicide risk could be achieved through health initiatives that encourage individuals with mental illnesses to access mental health services and that further develop interagency collaboration among mental health services, police, social services, and domestic violence services.
Acknowledgments and disclosures
The study was carried out as part of the National Confidential Inquiry Into Suicide and Homicide by People With Mental Illness, funded by the Healthcare Quality Partnership, and was sponsored by the University of Manchester. Prof. Howard is supported by the National Institute for Health Research (NIHR) South London and Maudsley NHS Foundation Trust Biomedical Research Centre–Mental Health. The views expressed in this article are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health. The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Dr. Oram had full access to all the data in the study and had final responsibility for the decision to submit this report for publication.
Prof. Howard reports grant support from Tommys the Baby Charity supported by a grant from Johnson and Johnson Corporate Social Responsibility fund for a project on antipsychotics and pregnancy. The other authors report no competing interests.