Domestic violence has consistently been acknowledged as a significant public health concern, and is linked to adverse physical, emotional, economic effects on victims and society in general (World Health Organization WHO 2016). This form of violence involves intentional, violent, and/or controlling behaviour perpetrated by an individual towards a current or former intimate partner. The definition encompasses acts such as physical and verbal attacks, intimidation, threats, isolation, and sexual assaults and can affect individuals of all ethnic, racial, and socio-economic backgrounds (Alpert et al. 1997). In the United States, an average of 20 individuals per minute experience physical domestic violence, which equates to an annual rate of more than 10 million victims (Black 2011). In 2015, 54% of all violent victimizations were perpetrated by an intimate partner (United States Department of Justice 2016).

Although victimization by domestic violence is not exclusively experienced by women, a woman’s risk of victimization is four times greater than that of men (Statistics Canada 2013). In 2016, 79% of victims of reported domestic violence were women (Statistics Canada 2016). Apart from the experience of domestic violence, women are more likely to experience a higher degree of victimization, increased severity of violence, and greater susceptibility to injury and lethality stemming from the violence (Black 2011). Violence of this nature rarely occurs in isolation; rather, it often develops in conjunction with recurrent abuse patterns that have been present within the home (Kuijpers et al. 2012). An escalation of this violence may result in domestic homicide (Adams 2007).

Domestic Homicide

At its extreme, domestic violence may result in domestic homicide, which involves the killing of intimate partners and/or family members in the context of domestic violence (Turvey 2008). Common risk factors associated with homicides in these cases include a history of domestic violence and separation (Ontario Domestic violence Death Review Committee Ontario DVDRC 2017). In Canada, there were 960 domestic homicides reported between 2003 and 2013 (Statistics Canada 2015). As with domestic violence, women are disproportionately affected by domestic homicide. According to Statistics Canada, the rate of domestic homicide in 2017 was five times greater for women than for men (Statistics Canada 2017). On a global scale, approximately 30, 000 women were killed by an intimate partner in 2017 (United Nations Office on Drugs and Crime 2018). As homicides of this nature have mostly been perpetrated by men, and primary victims have been predominantly female, (Websdale 1999; Ontario Domestic Violence Death Review Committee Ontario DVDRC. 2009), the current study will focus on male perpetrators and female victims.

Experiences of Children in the Context of Domestic violence and Homicide

Although domestic violence is rooted in intimate partnerships, it can also directly and indirectly affect children. Children can be exposed to domestic violence and its aftermath in a wide variety of ways (Jaffe et al. 2011). Some children may intervene or distract their parents and are at risk of injury or death in the crossfire (Jaffe et al. 2012). A study based on data collected from 112 studies in 96 countries estimated that over one billion children, aged 2-17 years, are exposed to some form of violence, including domestic violence (Hillis et al. 2016).

Children who are exposed to domestic violence are at an increased risk of experiencing other forms of maltreatment (defined as including physical, emotional, and sexual abuse, neglect, negligence, and any other form of abusive behaviour (World Health Organization 2010). In Canada, exposure to domestic violence is among the most frequently experienced forms of maltreatment experienced by children, with 34% of cases substantiated on an annual basis (Trocmé et al. 2010). The risk of physical abuse often increases with the severity and frequency of violence experienced by the mother (Bancroft et al. 2012). Further, children can be impacted by domestic homicide through experiencing the loss of a parent, as well as suffering from the negative repercussions associated with exposure to this extreme violence (Jaffe et al. 2012; 2017). Children can be both directly (e.g., physical aggression) and indirectly (“witnessing” violence at home) exposed to violence, which can influence externalizing behaviours (Wolfe et al. 2003; Fleckman et al. 2016). Exposure to this violence can also profoundly affect a child’s development and later life course as they navigate through their own relationships (Alisic et al. 2017; Graham-Bermann and Perkins 2010; Richards et al. 2008).

Children may also be victims of homicides, which can occur in the context of domestic violence that is occurring between their parents. In these cases, children may not be the primary targets of the perpetrator (Lawrence 2004). Cases involving child victims are more likely to occur during the period of separation and as part of an ultimate act of revenge by the perpetrator (Jaffe and Juodis 2006; Dawson 2015). In Canada between the years 2010-2015, 8% of children were victims of domestic violence-related fatalities (Dawson et al. 2018).

Risk Factors

Research is extensive on the impact of domestic violence exposure on children’s development. In fact, research on the effects of domestic violence on children has increased almost twenty-fold since the early 1990s (Jaffe et al. 2011). In contrast, a child’s risk of lethality in the context of domestic violence is not well documented, although some risk factors have been identified. Common risk factors that increase the risk of domestic homicide among intimate partners include a history of domestic violence and an actual or pending separation and some research has found that these factors likewise place children at risk of homicide (Dawson et al. 2018; Ontario Domestic violence Death Review Committee Ontario DVDRC. 2015). In addition, the presence of mental health-related challenges has often been identified in domestic homicide perpetrators (Jackson 2012; Sillito and Salari 2011). The presence of current child custody/access disputes among parents has also been identified as a risk factor for domestic homicides of children (Dawson et al. 2018).

Other research has also pinpointed risk factors related to child homicide. An examination of domestic homicide cases in the U.S. from 1999-2004 found that a higher percentage of cases where children were killed had parents in an intact relationship, perpetrators had exhibited suicidal intent, and they were more likely to involve biological children (Sillito and Salari 2011). Moreover, three antecedents to child domestic homicides have been identified as including: a prior history of child abuse, previous agency involvement with the family, and the existence of prior domestic violence (Websdale 1999). Apart from these studies, research has been limited on domestic homicide risk of children (Hamilton et al. 2013). Research has shown, however, that children can be considered at risk if their mothers are at risk (e.g., Olszowy et al. 2013) because of the overlapping risk factors for children and adult domestic homicide.

Exposure Reduction Framework

The exposure reduction framework can be applied in exploring the ways in which domestic homicides, including those of children, can be prevented. According to this framework, the prevention of domestic homicides depends on the identification of structures in place for victims of domestic violence in order to facilitate the reduction of risk. Mechanisms through which this may occur can include: assisting victims with ending an abusive relationship, providing support with overcoming obstacles to safety, and acknowledging the attitudes and behaviours of the perpetrator (Dawson et al. 2009; Dugan et al. 2003). According to the framework, decreased exposure of victims to perpetrators of domestic violence reduces the likelihood of domestic homicides from occurring through the provision of opportunities to leave the relationship (Reckdenwald and Parker 2010). Some of these opportunities include the availability of domestic violence -oriented resources, policies, and societal shifts that help victims leave an abusive relationship (Dawson et al. 2009; Dugan et al. 2003). However, research has also documented a potential retaliation effect in cases of severely violent relationships when too little, or ineffective prevention resources are implemented. That is, ineffectively implemented interventions may increase perpetrators’ aggression without appropriately reducing exposure (Dugan et al. 2003). These effects may be further compounded when children are involved due to the added pressure for facilitation of contact or engagement with perpetrators, as perpetrators are presumed innocent until the risk they pose is ascertained in criminal and family court.

Ultimately, it is thought that reducing exposure to domestic violence with appropriate service provision can curtail the occurrence of domestic homicides. Drawing on this framework and on research that has shown that a child’s risk is parallel to their mother’s risk, it can be inferred that an increased awareness of child-related risk factors can aid in the prevention of child domestic homicides. Therefore, appropriately implemented interventions that are tailored to the unique needs and risks of children (e.g., supervised visitation, parallel parenting, age-appropriate safety planning) are paramount (Jaffe et al. 2008; Poole et al. 2008).

Current Study

As evidenced by research in the field, children who live in an environment of domestic violence may be at risk of homicide. Although previous research on child-specific risk factors has been limited, some research has supported a history of domestic violence, an actual or pending separation, and the presence of perpetrator mental health-related challenges as risk factors for child domestic homicides (Dawson et al. 2018; Jackson 2012; Sillito and Salari 2011). Furthermore, a study that examined data from a review of 84 domestic homicide cases in Ontario, Canada, found that a higher number of agencies were involved with the family prior to the homicide (Hamilton et al. 2013). As a result, agencies, professionals, and members of the community are in a unique position to identify risk factors and engage in efforts to prevent child domestic homicides from occurring. This study sought to examine child-specific risk factors for domestic homicide, through comparing domestic homicide cases with and without the presence of children Specifically, the study examined cases where children did and did not exist as part of the family unit, and of those, cases where children were killed and not killed. The degree of agency involvement in these cases was also examined.

To determine risk factors for domestic homicide that are specific to children and the agencies involved, a review of Ontario Domestic Violence Death Review Committee (DVDRC) cases was conducted. Established in 2002, the DVDRC is a multidisciplinary committee of experts in the field who assist the Office of the Chief Coroner by reviewing deaths that occur as a result of domestic violence. The Committee uses historical information, interviews with family and friends, police reports, and agency files related to the perpetrator, victim, and other family members, to conduct a review of the cases and make recommendations with the objective of preventing future domestic homicides from occurring. The Committee has specified 41 risk factors for domestic homicide, obtained from literature in the field specific to the risk of repeated or lethal domestic violence. Consensus among committee members is required in order to include identification of these risk factors within the reviewed cases. Detailed definitions of these risk factors can be obtained from annual reports that are publicly available on the website of the Office of the Chief Coroner (https://www.mcscs.jus.gov.on.ca/english/DeathInvestigations/office_coroner/PublicationsandReports/coroners_pubs.html).

This study selected 140 cases reviewed by the DVDRC and divided these cases into two separate groups: (a) No Children (cases where no biological or step children exist within the family system) and (b) Children (cases where there is a child who exists within the family system, irrespective of whether a direct or indirect attempt was made on their life). This study operated with the awareness that children are at risk by living in proximity to domestic violence regardless of whether they were directly harmed, and that children are at risk for negative outcomes irrespective of their degree of exposure, such as physical injury, long-term mental health, behavioural, academic problems, as well as the difficulties associated with the loss or one or both caregivers (Stanley et al. 2018). Moreover, surviving children may have been spared from the homicide simply due to being physically absent from the homicide scene. A subsequent analysis was done with respect to cases where children were killed or where children survived the attempted homicide. The children in each case were either biological or stepchildren of the primary victim and/or perpetrator. Cases were examined for risk factors as well as agency involvement with families where domestic homicides took place. The objective of the study was to examine those factors that increase the risk of children for domestic homicide, as well as the number of agencies involved in cases involving children.

Method

Sample

This study analysed 140 domestic homicide case summaries that were obtained from the Ontario DVDRC database between the years 2003 and 2016, alongside individual case summaries and reports to identify unique risk factors that heighten the risk of domestic homicide for children.

The Ontario DVDRC database was developed through coding information from the Committee’s review of each case, based on files obtained from various agencies and professionals who were involved with the families, as well as interviews with family, friends, and other key individuals. Information that was coded included: demographic characteristics of the perpetrator and victim, circumstances of the homicide, and risk factors that were present in the case. The amount of information that was available differed among cases based on the amount of information available on file and the meticulousness of police investigations. The definitions for the risk factors have been developed by the DVDRC based on discussion and mutual agreement. Information and definitions pertaining to all risk factors are available in the appendix of the Ontario DVDRC’s 2016 annual report. The cases in the database came from two coding forms that are used by the Committee to organize the data pertaining to the homicide at intake.

DVDRC risk factor coding form

The DVDRC risk factor coding form is used by the Committee to identify and code information specific to the DVDRC’s 41 risk factors, with specification of whether the risk factor was present (P), absent (A), or unknown (Ukn) based on the information contained within the case reports.

DVDRC data summary form

The other coding form is a data summary form, which is used to provide a summary on all information pertaining to the case, including information specific to the victim(s) and perpetrator). This form also provides information on the involvement of 34 different service providers from various sectors (e.g., justice system, child protection, mental health).

Out of the 289 cases reviewed by the DVDRC between 2003– 2016, 140 cases were selected that met this study’s inclusionary criteria, which were based on the following: a heterosexual relationship between the primary victim and perpetrator, the perpetrator and victim being up to the age of 55 years (inclusive), and the perpetrator being male. Cases with female perpetrators and same-sex couples were excluded due to small sample sizes. The age restriction criterion was applied in order to better reflect cases where the perpetrator and primary victim were of age to have minor children and exclude cases involving older couples. Cases with older adults tend to be a distinct group of homicide victims and perpetrators, usually with a history of physical and mental disorders (Bourget et al. 2010). For the purposes of the study, the term “perpetrator” is used to denote the individual committing the offense, and the primary victim refers to the female intimate partner of the perpetrator. “Child” was used to classify any individual 18 years of age and under.

Cases were divided into a “No Children” group and a “Children” group based on a careful review of the homicide case summaries. The “No Children” group represented the absence of children in the family unit where neither the victim nor the perpetrator had any children. “Children” was the category used to reflect cases where children existed within the family unit as biological, adopted, or step-children of the primary victim and/or perpetrator, whether or not they were in their direct care and resided within their home. This included cases where children were killed, as well as ones where they were not killed. Cases which included adult children and minor children who were not the kin of the victim or perpetrator were not included. Likewise, cases were removed when it could be inferred that the victim and perpetrator had no contact or access to the children. Group 1 (No Children) served as a control group and thereby allowed for the exclusion of predisposing factors that are common to all domestic homicides, whereas Group 2 (Children) provided an opportunity to examine all cases involving children, with the awareness that all children living in an environment of domestic violence are at equivalent levels of risk, irrespective of whether they have or have not been directly harmed.

The researcher took an oath of confidentiality and was granted approval from the Western University Ethics Review Board prior to commencing this study. In order to maintain confidentiality, cases were identified by numbers. All cases were stored on a password-encrypted computer in a locked room at the university and were not transported outside of the room. All data analyses were performed on the same encrypted computer.

Procedure

The data in the DVDRC database was coded by a research associate and graduate research assistants who were all familiar with the database. The database has existed since 2003 and is continuously updated with incoming data related to domestic homicide cases. Where there was missing data, case histories were carefully reviewed and a judgment was made by the researcher on specific variables (e.g., whether children existed within the family system). Cases were then divided into the two groups: No Children (n = 101) and Children (n = 39). Demographic information was used to investigate general case characteristics. Variables related to risk (e.g., risk factors, risk assessment, agency involvement) were compared among the groups using chi-square and t-test analyses. The analysis was then followed by a comparison of the cases in the Children group, through segregation of cases where children were killed (n = 20), and ones where they were not killed (irrespective of whether they were physically present in the home) (n = 81), again through conducting analyses using chi-square and t-tests for continuous variables.

Results

General Case Characteristics

The study utilized separate chi-square analyses with the two groups (No Children × Children) on variables that related to the case characteristics and the primary victim and perpetrator’s relationship (Type of Case, Type of Relationship, Length of Relationship; see Table 1). The results revealed no significant differences between groups with respect to Type of Case. Significant differences between the groups were found for Type of Relationship (χ2 (2) = 25.7, p < 0.001) and Length of Relationship (χ2 (2) = 24.9, p = 0.001). The highest percentage of cases where children existed within the family unit involved legal spouses (65%), which was significantly higher than for cases where no children existed (23%). No statistically significant difference was found with respect to common-law relationships for both groups (26% for No Children and 22% for Children). Differences between the groups were found for the Boyfriend/Girlfriend category, involving 49% of the No Children group and 13% of the Children group. Most cases where children existed (50%) and where they did not exist (72%) involved relationships that were one to ten years in length. Differences were found for relationships that were less than one year in length, with 21% of cases with no children and 4% of cases with children falling in this category. Furthermore, the two groups also differed in relation to relationships that were 11 years or longer, encompassing 46% of cases with children and 5% of cases with no children.

Table 1 Demographic information/general case characteristics
Full size table

Additional analyses were completed to examine children killed by dividing the Children group into two groups based on whether the children were killed: No Child Target (child was not killed) and Child Target (child was killed) to examine if any differences existed between them with respect to the degree of child involvement. No significant differences were found between these groups for all three categories (Type of Case, Type of Relationship, and Length of Relationship).

Agency Contact

An independent samples t-test was used to compare the two groups (No Children × Children) to determine if there were any significant differences in the average number of all agency contacts (domestic violence-related and otherwise) for the primary victim only, perpetrator only, and for both the victim and perpetrator (see Table 2). No significant differences were found between the two groups. Likewise, when the two Child groups were compared (No Child Target × Child Target), no significant differences were found between the groups.

Table 2 Average number of agencies involved
Full size table

To determine the types of formal and informal domestic violence-related supports that were sought by the families involved in these cases, both groups (No Children × Children) were compared for the number of agency reports made using a chi-square analysis (see Table 3). Results were significant for Child Protection Reports (χ2(2) = 8.2, p < 0.05), indicating that 21% of cases where children existed involved reports made to child protection services in contrast to these reports made in 3% of cases where there were no children. This percentage represents a single case within this group that involved an adolescent primary victim who experienced child protection involvement within her family of origin. Significant differences were found for Legal Counsel/Services Reports (χ2(2) = 7.5, p < 0.05, with 31% of cases where children existed and 13% of cases with no children involved reports made to legal counsel or legal services. No other significant relationships were found for any of the other agency reports.

Table 3 Formal and informal agency reports
Full size table

The cases were further examined for distinction between child-related cases (No Child Target × Child Target). No significant differences were found between these two groups with respect to formal and informal reports, with the exception of informal reports from the “Family Members” category, which was present in a higher number of No Child Target cases; χ2(2) = 6.7, p < 0.05.

Risk factors, risk assessment, risk management, and safety planning

An independent samples t-test compared the two groups (No Children × Children) to determine if there were any significant differences in the average number of risk factors between the two groups, utilizing the 41 risk factors from the Ontario DVDRC. Results indicated no significant differences with the average number of risk factors between both groups, although both groups had the presence of 11 or more risk factors. A subsequent analysis with the child-focused groups (No Child Target x Child Target) was also found to be insignificant.

The cases were examined for differences across the two groups (No Child in the Home × Children) with respect to risk factors for domestic homicide as identified by the Ontario DVDRC in Canada (see Table 4). All 41 risk factors were initially analyzed, with low-frequency risk factors (present in five or fewer cases for both categories) being removed. The analyses were conducted using chi-square and Fisher’s exact tests for those variables with low expected cell counts. Significant results using chi-square analyses were found for Youth of Couple (χ2 (2) = 20.5, p < 0.01), with a higher percentage (33%) of cases with no children than child-specific cases (5%) having this risk factor, as well as for History of Violence/Threats to Children (χ2 (2) = 28.5, p < 0.01) with 32% of child-specific cases having this risk factor. As expected, none of the cases without children had the presence of this risk factor. Significant, although slightly weaker relationships were found for: Perpetrator Abused/Witnessed Violence χ2 (2) = 7.7, p < 0.05), with a higher percentage of cases with no children having this risk factor (36% vs. 17%) and Presence of Step-Children (χ2 (2) = 8.1, p < 0.05), with 17% of child-specific cases having this risk factor.

Table 4 Comparison DVDRC risk factors across two groups
Full size table

Overall, ten risk factors did not meet the chi-square assumption of fewer than 25% of cells having an expected count of fewer than five. As a result, the Fisher’s exact test was administered. Of these risk factors, Child Custody or Access Disputes (χ2(1) =8.1, p < 0.01), Prior Assault on Victim During Pregnancy (χ2(2) =8.7, p < 0.01), and History of Domestic violence in Current Relationship (χ2 (1) =5.4, p < 0.05), were significant with a higher percentage of cases with children having these risk factors (0% vs. 17%, 3% vs. 8%, and 67% vs. 79%, respectively). Post hoc analyses utilizing the Bonferroni correction, indicated that the significance of all results was not maintained with an adjusted alpha value (p = 0.001), with the exception of Youth of Couple and History of Violence/Threats Towards Children. Furthermore, post hoc results approached significance for Child Custody or Access Disputes and Prior Assault on Victim During Pregnancy. No significance was found for all factors upon analysis with child-focused groups (Child Target × No Child Target).

The number of cases where risk assessment, risk management, and safety planning practices were undertaken, was examined. A two-group comparison (No Children × Children) was performed using a chi-square analysis (see Table 5). Overall, results indicated no significant difference between groups with respect to Completed Risk Assessment, and Safety Planning & Risk Management. An analysis of both cases in the Child group (No Child Target × Child Target) was performed to determine if any differences existed between these two groups. Again, no significant differences were found.

Table 5 Number of cases reporting risk assessment, risk management, and safety planning
Full size table

This study investigated the risk factors related to domestic homicide faced by children who are exposed to domestic violence and the extent of agency involvement with families experiencing this violence. The research examined 140 domestic homicide cases in Ontario, Canada that were reviewed by a multidisciplinary death review committee in order to investigate the circumstances and factors that were present where children existed within family units that experienced domestic homicide and compared to cases with no children.

Overall, cases in the Children category were similar to those in the No Children category. Most significant differences found were based on expected demographics; i.e. factors associated with having children such as older couples in longer term relationships. Significant results were found pertaining to the type of relationship existing between the perpetrator and primary victim. A higher proportion of cases where children existed within the family system involved legal or estranged legal spouses, whereas a higher percentage of cases with no children involved more casual dating relationships. A higher percentage of cases with no children involved relationships that were less than one year in length and a significantly higher percentage of cases with children involved relationships that were 11 years or longer in length. Findings were not significant, however, when cases with children were compared for child involvement. Overall, these findings are aligned with research that has found that married couples with children are more likely to stay together than cohabiting parents (Social Trends Institute. 2017). Therefore, it may be more commonplace for couples with children to have a longer, more firmly established relationship status than couples with no children.

With respect to agency involvement, significant findings were found for reports made to legal counsel and/or legal services, present in a higher percentage of cases with children. Research has found that children are a motivating factor for women to seek legal intervention for challenges with child custody and support (Rhodes et al. 2011). Research has also documented the use of custody and access proceedings as abuse tactics meant to exert power and control over victims, with new opportunities for retaliation (Radford et al. 1997; Harrison 2008; Jaffe et al. 2009; Watson and Ancis 2013), which may propel women to seek legal advice. As would be expected, significant differences were also found with respect to child protection reports in 21% of cases involving children. As child maltreatment has been reported to occur in up to 60% of homes with domestic violence, this increases the likelihood of child protection involvement with these families (Lawson 2019).

In the subsequent analysis, cases where children were targeted (children killed) and children who were not targeted (children not killed) were compared. This yielded an unexpected finding for informal family reports, with a higher percentage of cases with children not directly targeted having had reports made to family members; that is, family members were aware of the violence taking place in these cases. As the likelihood of child maltreatment increases with the frequency and severity of domestic violence (Hartley 2004), extended family members may be more aware of domestic violence occurring in these cases and as a result they may have provided some form of safeguarding to children in those families where children were not harmed. Conversely, victims with children experiencing domestic violence may be more likely to seek informal supports such as family, in lieu of formal agency support which they may not access due to feared repercussions (e.g., children being removed from the home) (Ansara and Hindin 2010; Fugate et al. 2005; Sylaska and Edwards 2014), which in turn may contribute to increasing the domestic homicide risk of children. As all of these cases culminated in a death (either victim or child-related), however, this indicates that homicide risks may not have been fully recognized by family members, and they may not have been aware of where to direct victims to potential areas of support. No significant findings in the current study were found, however, with respect to the average number of agencies accessed by victims and perpetrators, although both groups in this study had an average of 11 or more risk factors, indicating that significant warning signs were likely present in these cases.

This study also examined risk factors for domestic homicide, where notable significant results were found. As expected, a higher percentage of cases involving children had a history of violence/threats towards children. Previous abuse of children has been found to significantly increase their risk of harm after divorce, indicating a link between previous violence towards children and increased risk to children (Hardesty et al. 2008). The study also found significant differences for cases with younger victims and perpetrators (between the ages of 15 and 24 years), with a higher percentage of cases without children having this risk factor. This finding is in agreement with research that has linked young age as a risk factor for all types of police-reported violence, including domestic violence (Statistics Canada 2013).

Although results were not significant when examining differences between the two groups with respect to other risk factors following post-hoc analyses, several of these results approached significance and therefore warrant consideration. A higher percentage of cases with children involved victims who experienced prior assault(s) during pregnancy, and as expected, prior child custody and access disputes. It has been suggested that child homicides can occur as a result of relationship breakdown with the involvement of custody and access disputes. Likewise, perpetrators of these homicides are frequently motivated by jealousy and revenge and perceived loss of control (Dawson 2015), feelings that may be heightened during custody and access proceedings. As indicated by the Exposure Reduction framework, the potential loss of power and control during legal proceedings also provides opportunities for retaliation by the perpetrator. In addition, research suggests that violence against victims during pregnancy may be indicative of more dangerous perpetrator characteristics (Campbell et al. 2003). No significant findings were found, however, when the groups with children targeted and not targeted were compared for all of the risk factors. Therefore, it can be inferred that children are at risk by merely living in proximity to domestic violence, regardless of whether they were directly harmed.

The outcomes in the cases that were analyzed in this study reflect an under-utilization of opportunities to perform risk assessment, risk management, and safety planning with families experiencing domestic violence. To illustrate, only 9% of the cases with children that were analyzed in this study had formal risk assessments completed and 9% of the cases were followed up with risk management and safety planning interventions. This finding is especially crucial, as the cases indicated an average of over 11 risk factors and an average of over 4 agencies involved with the families, although the findings were not significant. The Ontario DVDRC has an arbitrary definition of seven or more risk factors as being indicative of high-risk perpetrators where the homicide should have been predictable and preventable with hindsight (Ontario Domestic violence Death Review Committee Ontario DVDRC 2017). In addition, cases involving children had a higher percentage of reports made to legal counsel and legal services, indicating missed opportunities for intervention among these agencies. Although there are multiple tools that assess risk of harm or lethality, research suggests that these tools are underutilized, even in extreme cases (Nichols‐Hadeed et al. 2012). A study that analyzed family court responses to child welfare reports found that even with evidence of domestic violence and concerns by child welfare agencies, contact with fathers was still promoted (Macdonald 2016). Therefore, research reflects a need for utilization of family violence screening tools among legal professionals (Cross et al. 2018; Ontario Domestic violence Death Review Committee Ontario DVDRC 2004). Despite the recommendations made by death review committees for increased risk assessment, risk management, and safety planning, it appears that these recommendations are not being fully implemented (Jaffe et al. 2012). Drawing on the Exposure Reduction framework, services for children may therefore be inadequate in diminishing their exposure to domestic homicide risk.

The objective of the study was to identify unique risk factors present in domestic homicide cases involving children and examine the degree of agency involvement in those cases. The study utilized 41 risk factors identified by the Ontario death review committee and examined those risk factors that were present in cases involving children exposed to domestic violence. In addition to acknowledging the risk posed to children when their mothers are at risk, professionals working with families experiencing domestic violence should also be cognizant of the heightened risk posed to children when the identified risk factors are present, irrespective of whether there are obvious forms of maltreatment. The identification of these child-specific risk factors warrants timely risk assessment, risk management, and safety planning practices with families experiencing domestic violence, as these practices play a vital role in reducing their exposure to risk associated with domestic violence and ultimately preventing domestic homicides.

Limitations

The study’s findings should be considered with acknowledgment of its limitations. This research utilized a secondary data set in order to gather information on domestic homicides. Information about these homicides was based on Ontario DVDRC case reports, with an analysis of risk factors and agency involvement, as well as a synopsis of events leading up to the homicide. The case reports may have differed from one another based on the amount of information that was available. The Ontario DVRC may have been restricted in its access to all pertinent information (e.g., reports, interviews). The availability of information can differ based on factors such as the degree of a family’s agency involvement, regardless if a case resulted in a criminal trial. As a result of the missing information, certain variables may have been inputted as “unknown,” and therefore may not have been reflected in the analyses. This study is also limited by its lack of detailed data pertaining to the whereabouts and circumstances of children at the time of the homicide. The information that was available was limited in its scope with respect to the location of children who were and were not killed. Furthermore, the intentions of domestic homicide perpetrators were not always clear or documented in the DVDRC database.

Another limitation of the study is the uneven distribution of cases within each respective category. The “No Children” category had 39 cases and the “Children” category had 101 cases. Within the child group, there was an uneven distribution of cases where children were directly harmed and ones where they were not present or targeted (20 and 81 cases, respectively). As a result, these discrepancies may not have allowed for sufficient detection of differences, and therefore may have had some effect on the findings of this study.

Implications

Notwithstanding the limitations, the study generated findings that offer considerations for future research and practice. This study sought to expand on previous research examining child domestic homicide risk factors (Hamilton et al. 2013; Olszowy et al. 2013), through the use of a larger sample size than previous studies had access to. Specifically, Hamilton et al. (2013) found a higher number of agency involvement among families with children in their comparison of cases. No other significant differences among cases involving children were found in their study, however, which was attributed to the small sample size of child-specific cases at the time.

Future research can explore differences among children who were directly harmed or killed and children who were not present, using larger and more evenly distributed samples to determine if there are unique risk factors for lethality that exist in these populations. Future studies should also explore the unique risks and challenges within diverse and vulnerable populations of children who have various intersectional identities. Examples of such groups include: immigrant/refugee, indigenous, as well as rural/remote/northern populations. Literature in the field has demonstrated that children from vulnerable populations (e.g., immigrant) may face additional vulnerabilities and risks (David and Jaffe 2017). A greater awareness of unique risk factors among these populations can aid in the development of more targeted interventions.

As the findings in this study revealed an extensive lack of risk assessment (e.g., professional judgment, structured tools), risk management (e.g., counselling, parenting programs), and safety planning interventions (e.g., age-appropriate safety strategies, identification of safe places and emergency contact), even in the presence of a number of risk factors for victims and children, this is an area that warrants further exploration. Evidently there is a need for more commonplace use of standardized risk assessment and management measures, and safety planning interventions among service providers working with families experiencing domestic violence. Retrospective exploration of risk factors in homicide cases allows for an enhanced examination of identified areas of need by prevention efforts (Jaffe et al. 2017). As a result, future research can more closely examine the barriers that exist to assessing risk and providing appropriate interventions, through conducting interviews with a variety of service professionals and exploring the implementation of death review committee recommendations. A retrospective examination of risk factors most frequently present across homicide cases allows for prevention efforts to focus on these areas of need. Furthermore, there is a need to increase public awareness of these risks and available resources, so that friends and family members of victims can be aware of where referrals can be made to support services.

As mentioned throughout the study, children’s risk of homicide in the context of domestic violence has not been extensively studied. It is the hope of this research that through awareness and identification of unique risk factors specific to children in the context of domestic violence, and challenges associated with providing services catered to their needs, efforts can be made by service professionals and the community to protect these children and ultimately prevent the occurrence of these types of homicides.