Most child homicides are the result of the physical abuse of children,1 while others are associated with severe mental illness, anger arising from the breakdown of relationships, and a range of less common factors.2,3 In the 15 years between 1987 and 2001, 437 Australian children aged less than 15 years were victims of homicide, accounting for 1.5% of all child deaths and 9% of all homicides. The rate of homicide among infants aged less than 1 year is higher than rates among older children and adults. Australian children aged 0–4 years are about 50% more likely, and those aged 5–15 years are twice as likely, to die by homicide than children in the same age groups in the United Kingdom.4
Child homicide refers to the deaths from another’s deliberate actions of children aged under 18 years, while infant homicide refers to homicide of children aged less than 1 year. Several systems have been proposed to classify child homicide. The New South Wales Child Death Review Team (CDRT) has developed a simple classification: (i) deaths arising from non-accidental injury; (ii) deaths caused by parents affected by mental illness; (iii) deaths arising from family breakdown; and (iv) killings of teenagers.5,6 Other recognised classifications include the categories of infanticide (usually defined as the killing of an infant by a mentally ill mother), mercy killings, homicides associated with sexual assault, child homicide–suicide, child killings incidental to adult crimes, and the rare cases of children killing other children.7,8
The NSW Bureau of Crime Statistics and Research (BOCSAR) reports 159 homicides of people under the age of 18 in the 15 years from 1 January 1991 to 31 December 2005, but does not provide information about the circumstances of the deaths (Fiona Cotsell, Statistical Services Manager, BOCSAR, personal communication). We obtained legal documents describing 120 child homicides in this period from an examination of 3372 court documents located on LexisNexis9 and Australasian Legal Information Institute10 databases using the terms “murder OR manslaughter OR homicide AND child OR infant OR son OR daughter OR adolescent OR teenage OR teenager OR 1–17 years of age OR 1–17 years old OR 1–18 months of age”. Manual searches of published lists of offenders located a further 10 cases.11-15
Searches of the Sydney Morning Herald (SMH) archives (searched through the Dow Jones Factiva website http://factiva.com/) located media reports of 142 child homicides, including 122 of those located by earlier searches. The SMH cases included two coroner’s cases, seven homicide–suicides, and six unsolved homicides for which no court documents would be expected. The SMH reports led to the discovery of legal documents for a further 15 cases. Legal documents could not be located for three cases identified by the SMH searches that were the subject of legal proceedings, including one case subject to a suppression order and one homicide in the course of a police operation.
The cases were classified on the basis of the apparent motivation and the circumstances of the deaths. As the CDRT categories did not adequately describe all the cases, we used categories adapted from other studies.2,3,7,8 The term “fatal child abuse” was used in preference to “non-accidental death”; and “retaliatory killing” replaced “family breakdown”. Homicide–suicide cases were included as retaliatory killings. Additional categories were “fatal sexual assault” and “other homicide” (which included children who were the incidental victims of crime, homicides committed by children, a police killing, an altruistic killing and several cases not easily classified).
The term “homicide during psychotic illness” was used in preference to the CDRT category of “homicide caused by mental illness”, as the term mental illness can refer to a range of disorders that may not result in involuntary treatment under current mental health law in NSW and that do not form part of a legal defence such as “substantial impairment” or “not guilty by reason of mental illness”. We used a narrower definition of homicides occurring as a result of psychotic illness. This was possible because we were able to obtain psychiatric reports in all cases, which provided detailed information about the circumstances and reasons for each homicide in this category.16 Finally, we examined the homicides of children aged under 1 year.
Approval to perform the study was obtained from the St Vincent’s Hospital Human Research Ethics Committee.
In the 15 years from 1991 to 2005, BOCSAR reported a total of 1463 homicides, including 159 in which the victims were aged less than 18 years. Our search of legal and other documents found six more cases, giving a total of 165 child homicides.
Fifty-one women were responsible, or jointly responsible, for 53 deaths and 100 men were responsible, or jointly responsible, for 106 deaths. Six assailants of teenagers were not identified. There was more than one offender in 11 cases, more than one victim in 15 cases, and an adult (usually a parent) was also killed in 14 cases (Box 1 and Box 2).
Twenty-six of 151 offenders (17.2%) committed homicide during a psychotic illness (Box 2). The symptom most commonly associated with lethal assault was a persecutory delusional belief concerning the child, usually arising from auditory hallucinations. The main delusions were that the child presented a supernatural threat or was somehow caught up in a conspiracy that placed the patient in danger, or that the patient was saving the child or others from a worse fate. Two offenders reported the belief that the child was already dead and had been replaced by an imposter (Capgras syndrome), and one described being commanded by the voice of God.
Patients in first-episode psychosis (FEP) were over-represented, consistent with findings of other studies showing an increased risk of homicide before treatment.16,17 The 15 patients who had not received treatment were almost all mothers, over the age of 29, and many were from non-English-speaking backgrounds. Patients in FEP had shown signs of mental illness for an average of 6 months and had acute psychotic symptoms for an average of 6 weeks before the homicide. Three patients, including two diagnosed with postpartum psychosis, killed their children soon after the acute onset of psychotic symptoms. Most of the offenders in FEP had had contact with some form of health service in the 2 weeks before the homicide, and several others had been seen by a mental health professional at some stage in the past.
Fatal child abuse was the most common cause of death (59 of 165 victims). After excluding teenage homicides, the mean ages of both offenders and victims were significantly younger than those of the offenders and victims in other child homicides (offenders, 25.0 years v 30.5 years; t = − 3.84; P < 0.001; and victims, 1.5 years v 6.7 years; t = − 7.3; P < 0.001). The offender was often the child’s mother or de-facto partner. In several cases, both the mother and partner were charged over the death. Substance misuse, unstable accommodation, unemployment and past criminal convictions were often features of these homicides. This group also included five cases in which the cause of death was an overdose of methadone administered by a person who was responsible for the child’s care (usually to sedate the child), and five deaths initially attributed to sudden infant death syndrome.
There were 30 homicides that seemed to have been motivated by some form of retaliation. Ten men and seven women were responsible for 30 such deaths. The motivation for homicide–suicide is a matter for speculation, but retaliatory killing seemed to be the most likely explanation for all of the seven homicide–suicides in this group. Retaliatory killings were more likely to result in multiple homicides, and in eight cases a parent was also murdered. In most cases, a non-custodial parent was responsible.
Nine men killed a child after a sexual assault. None of the offenders were related to their victims, and several were not thought to have had prior contact with the victim. Five victims were aged less than 10 years, including two aged less than 5 years. Four victims were boys and one was a teenage sex worker. In most of the cases, the homicide was thought to have occurred to conceal the sexual assault.
Almost all the teenage homicides occurred in public places. Most were associated in some way with substance misuse. Six teenagers were killed after relationship breakdowns, five in the course of altercations not associated with antisocial groups, and five others probably associated with gang violence. All but one of the offenders were male, several of the victims were female, mostly ex-girlfriends. Typically, the offenders were slightly older than the victims. In some cases, the reason for the dispute leading to the homicide appeared to be trivial.
Ten of the 21 child-homicide victims in the “other” group were incidental victims of violent crimes, such as robberies or killings associated with the illegal drug trade. The remaining cases included homicides in which the motives were not established, two killings of toddlers by teenagers, four murders by very disturbed individuals (two in company and two alone) in which the killing of a young person might have been an end in itself, and one teenage victim of a mass killing in a public place.
Men killed 21 of the 37 infants. Women killed 16 infants: four of these women were mentally ill mothers; one did not want her child; one committed a homicide–suicide and one committed a retaliatory killing. Two babies died after the administration of methadone.
We identified 165 child homicides, six more than were recorded by BOCSAR. Similar differences are found in other homicide statistics because of differences in inclusion criteria and methods of data collection.18 Our study confirmed the findings of previous research that most child-homicide offenders are men and a minority of offenders were affected by severe mental illness at the time of the homicide.7,19,20
Fatal child abuse was the most common reason for child homicide, accounting for 36% of deaths. Measures to reduce the rate of physical abuse of children would therefore have the greatest potential to reduce child homicide in NSW. Fatal child abuse declined to very low levels after corporal punishment of children was outlawed in Sweden,21 where there were 103 child homicides in the 15 years after 1987, despite Sweden having both a larger population and a higher rate of total homicide2 than NSW. Following the Swedish example, most countries in the European Union have adopted a total ban on corporal punishment of children after a Council of Europe 2004 resolution.22 The only English-speaking country to ban corporal punishment in the home has been New Zealand, which did so in 2007, partly in response to a high rate of child homicides.23 A complete ban on all forms of corporal punishment means that some parents may have to be taught other ways to control their children. Parent training programs could meet this need and may also be directly helpful in reducing child abuse.24 The CDRT has reported that at least 60% of child homicides occurred in families that had had recent contact with health services, the police or the Department of Community Services.6 Hence, there appear to be opportunities for early intervention through parent training and education, offered together with other forms of assistance.
Twenty-seven of the homicides were committed during a psychotic illness, mostly in the first episode of psychosis. Earlier recognition and treatment of the first episode of psychosis could reduce the rate of this form of homicide.17 Child homicide by patients with established mental illness is rare, but health workers should always consider the safety of children under the care of acutely mentally ill patients, especially if the patient reports symptoms involving the children.
Some homicide–suicides and retaliatory killings involving children occur without warning, but child welfare agencies and the police should be notified of domestic violence and of any specific threats to the mother or children. Some cases might be prevented by improved arrangements for non-custodial parents.
Teenage homicide is usually perpetrated by other young people, and is similar to some adult homicides, occurring in public places or after relationship breakdowns. Measures to reduce intoxication and the possession of guns and knives in public places could prevent some teenage homicides.
Twenty-two per cent of the homicides involved victims aged less than a year old. Our study confirms that most infants are killed by men.21,22 Moreover, most infant homicides are committed by people who do not have a severe mental illness.19,20 Most cases of infant homicide are not, therefore, a result of infanticide, which is defined in law as the killing of a child during a mental illness arising from the effects of childbirth. Hence, any attempt to reduce infanticide will need to target young, socially disadvantaged parents as well as the early treatment of postpartum psychosis.
1 Sociodemographic and homicide characteristics associated with 165 child homicides in New South Wales during 1991–2005
Homicide during psychotic illness |
|||||||||||||||
Method of homicide (by victim) |
|||||||||||||||
Suffocated, strangled or drowned |
|||||||||||||||
2 Comparison of the sociodemographic and diagnostic characteristics of 26 child-homicide offenders who had a psychotic illness at the time of the offence
Received 14 May 2008, accepted 14 October 2008
Abstract
Objective: To examine the circumstances of homicides of children in New South Wales from 1991 to 2005.
Design and setting: Retrospective analysis of all identified child homicides in NSW from 1991 to 2005, based on data on offenders and victims obtained from crime statistics, documents located by systematic searches of legal databases and media reports, and medicolegal reports of offenders who committed child homicides during psychotic illness.
Main outcome measures: Demographic characteristics of homicides and a history of prior psychiatric treatment among offenders with psychosis.
Results: We located documents describing 165 homicides by 157 offenders. Fifty-nine deaths were a consequence of child abuse, including those of five children who died from methadone overdoses. Both the offenders and the victims in fatal child abuse were significantly younger than in other forms of child homicide. The courts found that 27 child homicides had been committed by 26 offenders during the acute phase of psychotic illness, and 15 of these offenders had never been treated with antipsychotic medication.
Conclusions: Earlier identification and treatment of psychotic illness in mothers, and changes in the way methadone is provided to opiate-dependent parents, might result in a small overall reduction in the number of child deaths. More lives could be saved by measures that reduce the incidence of child abuse, including the prohibition of corporal punishment of children.