Filicide: the Australian story

Abstract A filicide death, meaning the killing of a child by their parent or equivalent guardian, is a tragic event. Sadly, a UK study suggests Australia has the fourth highest rate of filicide among similar developed nations. Since Australian research studies on the incidence of filicide, or indeed on any other aspect of the problem, are limited, it is impossible to know if this finding is correct or not. However, in the last several years more research on filicide has emerged in Australia and by reviewing the recent research in detail, this article develops an integrated analysis of Australian filicide research and contributes to the knowledge bank on Australian filicide that can be used by professionals undertaking practice and research in intervention and prevention. Analysis of the studies shows one child dies at the hands of a parent every fortnight and that this number has not changed for many years. The analysis identifies the profiles of victims and perpetrators. The constellation of circumstances and stressors associated with each of the parental perpetrator groups is discussed, including the perpetrators’ contact with, and mostly unsuccessful use of, services. Based on the analysis, a way forward to prevention is proposed.


Introduction
Australia has been depicted as a nation with a high incidence of filicide (Pritchard et al., 2013). Since neither the Commonwealth of Australia nor its states and territories publish regular data on the incidence and circumstances of such deaths, it is difficult to know if this is an accurate representation of the problem. Furthermore, for some decades, research more widely on filicide in Australia has been sparse, with few studies focused explicitly on investigating the nature, causes and extent of the problem.

Purpose
However, in very recent years, some Australian research has been published shedding light on the problem. By examining this research, this article will produce an integrated analysis of the most recent knowledge on filicide, with a view to producing the first Australian research-based knowledge bank for professionals working to overcome filicide, at the individual client and community-wide level.
The research to be reviewed is composed of studies from two major groups, the first group being large-scale studies, like the 10 year plus retrospective studies of filicide deaths in NSW (Butler & Buxton, 2013), Victoria (Brown et al., 2014) and Australia (Brown et al., 2019), and the second group being smaller scale studies examining sub-categories of filicide deaths, like neonaticide (De Bortoli et al., 2013a), maternal infanticide (De Bortoli et al., 2013b) and familicide (Johnson, 2005(Johnson, , 2008 and issues such as the role of parental separation in filicide (Brown et al., 2014;Kirkwood, 2012) and case studies of perpetrators' characteristics, backgrounds and personality (Eriksson et al., 2016;Eriksson et al., 2019).
The article will pursue three questions as a way of building the knowledge bank. The questions are: • What is the incidence of filicide deaths in Australia and how does this incidence compare with other similar countries? • What are the characteristics of filicide victims and perpetrators in Australia and how do these compare with those from other countries? • What are the implications for prevention that arise from the answers to these questions?

Definitions
Before answering the three questions above, the meaning of terms used in filicide research needs to be clarified.

The meaning of filicide
In this article, the term filicide is used to mean the intentional or unintentional killing of a child by a biological or adopting parent, a legal guardian or a step-parent; it includes child homicide when the perpetrator is a parent or equivalent, fatal child abuse by a parent or equivalent and the non-accidental killing of a child by a parent or equivalent.

Sub-categories of filicide
In this article, all sub-categories of filicide will be included in the term filicide. These sub-categories are Neonaticidethe killing of a child within the first 24 hours of birth, Serial neonaticidethe repeated killing of a newborn, Infanticidethe killing of child within the first 1-2 years of life, Familicidethe killing of all children and other family members, such as the other parent, grandparent or other family member and Murder/suicidethe killing of a child and the suicide of the perpetrator.
However, as research studies use differing combinations of these sub-categories, it is recognised that comparability of studies has some difficulties, especially between countries.

The meaning of child
In this article, the term 'child' means a child of any age, with children being divided up into those under 18, as is usual in Australian filicide studies, and over 18, which is unusual, but pursued in the national study.
Question 1: what is the incidence of filicide in Australia?
It is difficult to establish incidence in Australia as data are hard to obtain and not always accurate. No one state or federal government or non-government body reports the numbers of such deaths annually, even though data do exist.

Data on incidence
The National Homicide Monitoring Program (NHMP), housed within the Australian Institute of Criminology, Canberra, collects data on all homicides, including filicides, annually. However, it does not publish it regularly, only in occasional studies (Mouzos & Rushforth, 2003). The NHMP counts only those deaths where a perpetrator is charged with the crime (Brown et al., 2019) and when the perpetrator is unknown, dies or is not charged, the death is not counted. However, it is acknowledged that there are some high-profile cases such as the Luke Batty murder, Farquharson murders and Freeman murder that have been important to the public narrative on filicide and were, in that sense, of significant impact.
More reliable and extensive data are collected in State and Territory Coroner's offices; each office holds a file on every filicide death, regardless of any criminal justice action. These data are not published annually in every state and in some states must be extracted in special studies, such as in the NSW and Victorian retrospective studies both covering the years 2000-2009 (Brown et al., 2014;Butler & Buxton, 2013).
No matter what data source is used, some deaths will remain hidden as De Bortoli et al., (2013a) point out; the births are not recorded and neither are the deaths; the births are recorded but the child and their death disappear (Brown et al., 2014); and in some cases the cause of death is wrongly assessed by the medical practitioners and not counted as a filicide (Packer, 2013).

Patterns of incidence
The findings on incidence, drawn from the three large-scale studies on state and national incidence, arouse considerable concern. The national study, covering 12 years from 2000, showed the yearly rate of deaths to be higher than in countries, like Canada, and that it had not fallen in that period, even though total homicide and intimate partner homicide rates had (Brown et al., 2019). The Victorian study showed the same. That study found the deaths of children known to child protection had declined whereas total filicide deaths had not, suggesting successful preventive action through the child protection service, but not elsewhere.
According to the national study, some 274 children under 18 years were killed by a parent or parent equivalent, giving an average of 23 children killed per year (Brown et al., 2019). However, the Victorian and NSW studies, both of which covered 10 of the same years, showed higher numbers in those same years, higher on average by 10%. The differences in numbers between the state and federal data sources probably arose from differences in definitions of what constitutes a death to be placed in the databases and from delays in moving data from state to federal levels. If the increased numbers in the two most populous states of NSW and Victoria hold true across all states and territories, the average number of children under 18 years of age killed annually rises to 25, to 1 child killed on average each fortnight.

Variations between states
Taken together, the three large-scale studies showed differences in incidence between the states and territories, no doubt reflecting the different populations, patterns of settlement, economic activities and social problems in each location. The national study showed the greatest number of deaths occurred in New South Wales (30% or 84 deaths), followed by Queensland (28% or 79 deaths) and Victoria (21% or 60 deaths). Queensland had the second highest victimisation rate per capita and the second highest numbers of deaths, well above the numbers to be expected from that population. South Australia, with the third highest victimisation rate, had a striking occurrence of mothers as perpetrators.

Deaths of indigenous children
The picture of the deaths of indigenous children was puzzling. On the one hand, the rate of deaths was higher than for non-indigenous children, with 10% of the victims under 18 being indigenous, a proportion around three times their numbers in the wider community. At the same time, the actual numbers were so low they may not be reliable. On the other hand, the rate of deaths was lower than the reported rates of the abuse of indigenous children or the reported rates of indigenous children placed in care (Australian Institute of Health and Welfare [AIHW], 2018; Australian Institute of Family Studies [AIFS], 2018), suggesting different dynamics were at play. Higher rates for the deaths of indigenous children have been reported in New Zealand (Family Violence Death Review Committee, 2017; Health Quality and Safety Commission New Zealand, 2017) and were higher than the rate in Australia.

International comparisons
The one study (Pritchard et al., 2013) comparing filicide rates across countries placed Australia fourth highest in incidence of filicide among the 17 countries on one measure and fifth highest on another. While inter-country comparisons may not be reliable, with different parameters for age of children, for categories of deaths, as well as the use of different types of data sources, it is of concern that Australia's incidence appears to be so high.
A later Canadian study by Dawson (2017Dawson ( , 2018, using the Canadian Homicide database which covers all filicide deaths from 1961 to 2011, confirmed the findings of the English study. It showed that the Canadian filicide rate was less by one-third than the Australian rate and that the Canadian death rate was declining, whereas the Australian rate was not.

Explanations of incidence
Why is the Australian incidence high? The high incidence of indigenous children's deaths in relation to their numbers in the community might be a reason, but their actual numbers are so low they do not seem to account for this pattern. The characteristics and stressors of Australian perpetrators presented in the next section do not seem to be different in Australia than in Canada or the UK. Service absence or unavailability and lack of communitywide polices for preventive action may be an explanation. For example, the over-representation of mothers as perpetrators in South Australia may be related to an absence of services for mothers of babies and young children, and the high rate of filicide in Queensland may similarly relate to a lesser availability of health and welfare service provision. Several studies (Brown et al., 2014;Rodriguez Manriquez & Fernandez Arias, 2018) have found that filicide death rates are related to service availability.

Question 2: characteristics of victims and perpetrators
Answering the question as to the characteristics of victims and perpetrators is not as difficult as determining incidence. The recent Australian studies have produced substantial information on victims and perpetrators. Nevertheless, one gap in knowledge is the lack of research on the relationship between the victim and the parent(s) and any role that relationship plays in the children's deaths.

The characteristics of victims
Victims in all the three large-scale studies were young; in the national study, typical of all the others, the victim's median age was 2. Children under 1 were the most frequently killed and two-thirds of children killed were under 5 (Brown et al., 2019), replicating the age pattern reported for abused children in Australia (AIHW, 2018). As the children grew older, they became less vulnerable. However, deaths occurred in all age groups, even among children over 18; some 10 additional children were killed as adults, that is, being over 18, raising the annual death rate to 26 per year.
More male children were killed than female children: 158 male children and 125 female children. Two-thirds of the children killed were not at the time, nor ever had been, known to child protection services. This demonstrates the way in which these deaths stretched beyond the child abuse-linked child deaths framework being addressed by the child protection services.

International comparisons
Victims in Australian studies were similar to those in overseas studies, in terms of their age range and gender breakdown (Bourget et al., 2007;Dawson, 2018;Sidebotham et al., 2016). No Australian studies found children with a disability or chronic illness to be more at risk, but overseas studies have done so (Abdullah et al., 2019;Rupp, 2018). The study examining filicide in Ghana described how communities supported parents killing a child with a chronic illness and or disability (Abdullah et al., 2019). The NHMP does not seek information on a child's disability and this may be a reason the factor does not appear in Australian research.
Australia reports a lower incidence of neonaticide than some other countries, such as some African states (Abdullah et al., 2019;Mathews & Abrahams, 2018), some Asian states (Yasumi, 2018) and some European states (Klier et al., 2019). Serial neonaticide, where a mother kills one newborn child and then later another (Klier et al., 2019), has not been recorded in any of the Australian studies. However, there are cases in which a parent has killed her children at older ages (see, for instance, Kathleen Folbigg's conviction for killing her four children (Sas, 2019)).

Explanations for victimisation
Current explanations for filicide rest with the circumstances of the perpetrators and not the victims. Nevertheless, one of the obvious characteristics of the children killed suggests their powerlessness and their dependence must be part of the explanation for their victimisation. It is not clear why more male children are victims than female children, both in Australia and overseas. Absence of, or a poor, attachment between a child and a parent has been suggested in the child abuse literature (Sidebotham et al., 2016) as an explanation. The extensive problems, which the research shows perpetrator parents have, have been thought to undermine their parenting (Dobash & Dobash, 2018) and diminish their relationship with their children.

The characteristics of perpetrators
Australian research has studied perpetrators as a total population and found that perpetrators share a constellation of characteristics. However, when perpetrators are divided into parental role groups the frequency of each characteristic varies. Australian perpetrators belonged to one of five parental groups. These groups were mothers, fathers, mothers and fathers acting together, stepfathers, and mothers and stepfathers acting together. After the last study reviewed was completed, one aunt and one grandfather were identified as perpetrators (Australian Broadcasting Corporation [ABC], 2014; Paddenburg, 2018).

Parental role group in relation to children killed
The parental role group to which the perpetrators belonged affected the event. In the Australian studies, it influenced which children were killed, why and in what circumstances. The national study showed the distribution of victims among five groups of perpetrators (Brown et al., 2019) as follows: Mothers killed 134 victims. They killed children of all ages, but they killed young children aged from 0 to 1 and 1 to 4, more frequently. Male children were killed slightly more often than female children. Single children were killed more frequently than multiple children and, sometimes, mothers killed themselves. Mothers killed children by many means, but by neglecting more often than fathers did.
Fathers killed 109 victims. They killed children who were young, but less often and younger than mothers did. Older children, from 6 onwards, were killed more frequently by fathers than by mothers. Male children were killed more frequently by fathers and more frequently than by mothers. Single children and multiple children were killed by fathers and sometimes a partner or former partner and themselves. Fathers killed the children by many means but more often by beating than did mothers. Fathers were the more frequent perpetrator of the deaths of indigenous children.
Stepfathers killed 41 victims. They killed very young children and single children, except for one stepfather who killed two children. Male children were killed more than female children, and more often than by either mothers or fathers. Stepfathers' violence, which is appeared to be a sudden explosive outburst, is likely to have been a culmination of previous violent behaviours in the domestic setting. As perpetrators, they occurred in numbers that were disproportionate to their numbers in the community (AIFS, 2018).
Fathers and mothers acting together killed 10 victims. Stepfathers and mothers acting together killed nine victims. To date, there is little if any research investigating collaborative parental killings.

Perpetrator factors
Perpetrator factors were shared among most of the perpetrators. The characteristics were the presence of mental illness, being a perpetrator or a victim of domestic violence, being separated from a partner, experiencing substance abuse, having a criminal history and having a history of childhood abuse or trauma. These characteristics varied according to the perpetrator's parental role group, forming distinctly different profiles for each group. Using the national study profiles was as follows: Mothers were characterised as being mentally ill (52%), separated from or with no current partner (45%), having a history of crime (30%), being a victim of domestic violence (23%), and having drug (22%) and alcohol (11%) abuse issues.
A further characteristic, that of suffering childhood abuse, was identified by Eriksson et al. (2016), Eriksson et al. (2019) and Sachmann and Johnson (2014), but none of these were large studies and the occurrence of this characteristic awaits confirmation from studies with larger populations.
The frequency of these characteristics outlined in the profiles may be an underestimate. The frequency of many of these characteristics, in particular being a perpetrator of domestic violence, suffering a mental illness and being separated from a partner, was higher in the state studies, partly due to design problems with the NHMP and partly due to the differences in definition between state and federal bodies.
Australian research has investigated the relationship between these characteristics. Johnson and Sachmann (2018) have linked each one to the others. They propose that abuse in childhood, or childhood trauma, lays the foundation for all of the other factors, and they suggest that the constellation of factors, termed a constellation of stressors by Stroud (2008), the English researcher, merges into an escalating spiral of stressors that ends with the killing of the child.

International comparisons
The features of perpetrators' profiles, domestic violence, mental illness, parental separation and substance abuse have been identified internationally (Bourget et al., 2007); abuse as a child and criminal history are more recent features of the profile with both being confirmed in the UK (Dobash & Dobash, 2018;Pritchard et al., 2013). Unemployment has been noted in overseas studies (Klier et al., 2019) but not in Australia.

Explanations
In the early international filicide research, perpetrators' motives were proposed as explanations for their actions. A model of various categories of motives was erected (Resnick, 1969(Resnick, , 1970 and expanded over time (Bourget et al., 2007). However, motivational analysis of perpetrators' behaviour had theoretical and operational flaws (Sidebotham, 2013) and fell into disuse.
Currently, explanations have focused on constellations of social and psychological stressors and the relationships between them. Stroud (2008) depicted an interacting and escalating process, and Sachmann and Johnson (2018) described the escalation process in detail. While all see childhood experiences as the start of the perpetrator's malfunctioning, Johnson and Sachmann (2018) believe it is the factor that leads to all the others, both social and psychological. Mathews and Abrahams (2018) support this argument, but suggest the stressors should be extended to better reflect the complexity of the causation. They have included the stressors of national history and culture which, for them as South Africans, includes a history of colonial oppression of citizenswhich also has relevance to Australia.
Researchers investigating the sub-categories of filicide have also fleshed out further stressors for each group of perpetrators, showing that perpetrators in each sub-category share common stressors but have additional ones that are different. Klier et al. (2019) exemplify this in their study on perpetrators of single neonaticide as compared with perpetrators of serial neonaticide. The differences, slight as they were, confirm Sidebotham's view (2017) that the pattern of stressors must be identified for each group of perpetrators and matched to relevant services if prevention is to be achieved.

Perpetrators' use of services
One of the few studies investigating perpetrator's use of services (Brown et al., 2014) found most perpetrators were not isolated from services. Either they had contacted services or services had contacted them. Thus, some 100% of mothers, 71% of fathers and no stepfathers had contacted services and 90% of mothers, 71% of fathers and 56% of stepfathers had been contacted by services.
Mothers sought help mostly from General Practitioners, mental health services and counselling services. Fathers sought help from GPs and not at all from mental health services. Mothers, fathers and stepfathers were contacted by criminal justice, alcohol and drug services, and child protection.
Services did prevent children's deaths. The Victorian study showed one outer suburban local government area (LGA) area adjoining Melbourne had the highest rate of children's deaths among all Victorian LGAs. On investigation, that LGA was found to have few health and welfare services of any kind and certainly none for the influx of young families living there.
However, services struggled with perpetrators as clients. Johnson (2008), who interviewed perpetrators of familicide from WA, described many examples of police and court staff being inappropriate in their responses to perpetrators and to victims' families. Rosy Batty, whose son Luke was killed by his father, echoed Johnson (2008) findings and extended the criticism to the poor coordination of services, particularly to police and court services (Barry, 2015). The files in the Victorian Coroner's Office showed similar examples of perpetrators asking for help, but meeting inappropriate responses from professionals, many of whom were inexperienced and untrained in tackling filicide, despite being well intentioned in attempting to address this problem.

International comparisons
Overseas research (Stroud, 2008;Wilczynski, 1997) had earlier found that perpetrators use services. However, as in the Australian research, problems were found in the exchanges between perpetrators and service professionals. Stroud (2008), who like Johnson had questioned perpetrators, suggested some perpetrators approached the wrong service for their problems. Alternatively, when they did approach the relevant service, the service did not understand the problem as it was presented by perpetrators. Stroud argued that, regardless of the perpetrators' views of their contact with services, perpetrators were unclear in voicing their concerns and the danger to the child did not emerge. Bourget et al. (2007) supported this view, seeing services as being ignorant of the perpetrator's potential for filicide, even when the services were the relevant service for perpetrators to be using.

Explanations
Evidence for services preventing filicide deaths rests on research (Brown et al., 2014;Rodriguez Manriquez & Fernandez Arias, 2018) that shows a connection between the absence of services and higher rates of deaths. Little is known about why services fail or why they succeed. Bourget et al. (2007) argue they fail for two reasons. The first is that professionals in the services are largely unaware of the problem and so are ill-equipped to manage it. The second is that the services are not placed in a suitable location in the service system. They recommend new and specialised services be developed and located in the health and family law services. Hatters Friedman and Resnick (2007) propose something very similar. They place the specialist service units in the health services, a location possibly suitable for mothers but possibly less suitable for fathers and not at all for stepfathers. Appropriate and responsive service provision has not been designed or implemented.
Question 3: what have the answers to Questions 1 and 2 contributed to policies for prevention Koenraadt (2013) points out that filicide is a universal phenomenon but not a uniform one. It presents a different face in every country, necessitating research in every country to identify its incidence, nature, causes and prevention policies and programmes for it to be overcome.
The answer to Question 1 shows that filicide is a problem of concern in Australia; incidence is high for a developed country and deaths are not declining. All sub-categories of filicide occur except serial neonaticide.
The answers to Questions 1 and 2 form a beginning knowledge bank on filicide in Australia. They describe who are the victims and who are the perpetrators, and they provide vital information about them. They identify which children are most at risk and which parents are most likely to be perpetrators. They show a constellation of many problems and stressors that lead perpetrators to act, and they show how the problems interact and escalate over time until the 'and then something happened' (Stroud, 2008, p. 498) crisis point. The research shows widespread ignorance of the problem in the community, in the helping services and among the professionals who staff them.
Clearly, a comprehensive approach is required to address filicide, what Dawson (2017) terms a public health approach. Such an approach would use the knowledge gained from the research to design, produce, disseminate and implement prevention programmes, including educational programmes for professionals, organisations and communities, and programmes designed especially for each of the targeted perpetrator groups.
However, a central national organisation is required to lead what would be a major campaign. The national campaign to address men's violence to women and children is a powerful model. Thus, a national Filicide Committee or Board should be set up to fund and direct the development of a national database, a national research programme, a national education programme and a national inventory and evaluation of prevention programmes. The national Filicide Committee should link to the various relevant state committees. Prevention will be complex as filicide is scattered thinly across all Australian communities rather than concentrated in any one group or location. Prevention will have to be targeted to each of the perpetrator groups whose needs are multiple and varied.