Domestic violence is considered the most life-threatening, traumatic, and harmful family problem existing in today’s society (Roberts 2007). The World Health Organization (2005) collected data from over 24,000 women in 10 countries and identified that ever-partnered women (women who were, or had ever been, married, in a common-law relationship, or in a dating relationship) have a lifetime prevalence rate of physical violence ranging from 13 to 61 %, with the most common rates ranging from 23 to 49 %. An estimated 1.5 million women are victims of intimate partner violence each year in the United States (Tjaden and Thoennes 2000). In 2007, Canadian police received reports of 40,200 incidents of spousal violence, totaling 12 % of all police-reported violent crime (Statistics Canada 2009). Statistics Canada determined that the Canadian average rate of police-reported spousal violence stood at 188 per 100,000 in 2007. True rates of domestic violence are much higher, as the majority of incidents go unreported (Loewenberg 2005).

Alpert et al. (1997) define adult intimate partner violence or domestic violence as “intentional violent or controlling behavior by a person who is currently, or was previously, in an intimate relationship with the victim” (p. S4). Violence or abuse depicts a pattern of behaviors intentionally used by one person to gain control over another, inclusive but not limited to physical aggression, threats, intimidation, neglect, sexual assault, social isolation, verbal attacks, and restriction to resources (Alpert et al. 1997). Other researchers define violence by the act itself such as physical violence, emotional or psychological violence, and/or sexual violence (WHO 2005).

Once violence is present within a relationship, it often escalates with time (Frye et al. 2006). Risks indicative of severe domestic violence include prior sexual assault, stalking, controlling behaviors, separation from the victim, lack of child access, violence during pregnancy, violence escalation, child abuse, death threats to the victim or the child, homicide attempts, threats of suicide, isolation, and barriers to help-seeking (Adams 2007; Humphreys 2007). Previous strangulation, prior use of a weapon to scare or injure the victim, past incidences resulting in severe victim injury, extreme jealousy, and a perpetrator’s possession of a weapon are other factors signifying severe risk. When violence escalates to an extreme degree, domestic homicide may occur (Adams 2007; Campbell 1995; Websdale 1999).

Domestic Homicide

Domestic homicide is the killing of a family or household member, or intimate partner by another (Turvey 2008). In terms of the frequency of domestic homicide, it was noted that in the United States approximately 1,800 adults are killed annually as a result of domestic homicide (Adams 2007). The 2007 domestic homicide rate in Canada was four per million spouses (Statistics Canada 2009).

Domestic homicide often results from an accumulation of rage, long-standing turmoil, and conflict (Turvey 2008). Common risk factors for domestic homicide include a history of domestic violence, estrangement (victim attempts to leave the relationship), obsessive-possessiveness, prior police involvement, and the perpetrator having a criminal history. Other factors include threats to kill, substance abuse issues, protection orders, child custody disputes, a perpetrator coping with a mental illness, hostage-taking, perpetrator’s step children in the home, changes in circumstances (e.g., loss of employment), and victim fear (Campbell et al. 2003; Daly and Wilson 1996; Ontario Domestic Violence Death Review Committee [Ontario DVDRC] 2009).

Children and Domestic Homicide

Intimate partners may not be the only victims in a domestic homicide. In many cases, children are directly impacted by a domestic homicide on the account that they may lose a parent or both parents due to the crime, they may have witnessed or been exposed to the violence or death, or they may become a homicide victim themselves (Jaffe and Juodis 2006). Approximately 3,300 children lose parents to domestic homicide every year in the United States (Lewandowski et al. 2004).

The number of children affected by the extreme violence in domestic homicide increases when one considers that roughly three attempted-homicides occur for each completed homicide (Lewandowski et al. 2004). One study examined 121 cases involving the actual homicide or attempted homicide of a woman by her intimate partner and the number of children impacted (Lewandowski et al. 2004). The study found that children witnessed 35 % of the actual homicides and 62 % of the attempted homicides. Furthermore, children discovered the bodies of their mothers in 37 % of the homicide cases and 28 % of the attempted homicide cases. The Montgomery County Domestic Violence Death Review Committee (2008) in Ohio reviewed 42 domestic homicide cases that occurred between 1995 and 2007 that resulted in 52 deaths. In 24 of the cases, children were living in the home at the time of the homicide. Fifteen of those 24 cases were documented to have children present at the time of the homicide with 53 % of these children having witnessed the event. These children were often directly involved in the homicide, with some children escaping through windows, and others getting injured as they tried to intervene. Further review revealed that 58 children lost at least one parent during the domestic incident.

Reviews of domestic homicides in the United States and Canada point to the fact that children may become homicide victims (Jaffe et al. 2012; Jaffe and Juodis 2006). For example, there were 572 domestic violence fatalities in Washington State between January 1997 and June 2010. Of those 572 incidences, 44 children were killed (Fawcett 2010). In the state of Florida, domestic homicides accounted for 15 % of all the state’s murders in 2006 with a total of 31 child deaths (Florida Domestic Violence Fatality Review Team 2007). The Ontario DVDRC (2009) in Canada identified 184 domestic homicide incidents that occurred in the province between 2002 and 2008. There were a total of 253 deaths, including 23 child deaths. These statistics illustrate that children can also be victims in a domestic homicide and that it is not just intimate partners that can be in danger of death.

Motivations for Killing Children

There may be multiple motivations behind killing a child in the context of domestic violence situations (Ewing 1997). Children can be used as a direct target for retaliation against an intimate partner and are killed for the purpose of inflicting harm on the child’s other parent. In other cases, the children, the (ex)partner, and the perpetrator are killed in the context of domestic violence. These cases of familicide (killing of multiple family members), may represent situations where the perpetrator is very controlling, but also very dependent on family members (Ewing 1997). Some authors have suggested that the perpetrator may be overwhelmed by shame and a sense that they have not lived up to their gender role expectations as a husband and father (Websdale 2010).

[If the] perpetrator feels that his domination of the family is threatened, often by family members’ threats to leave and/or report his abuse to others, he may resort to homicidal violence in a misguided effort to maintain his control and prevent a complete rupture of the family unit. (Websdale 2010, p. 135)

In a significant number of these types of cases, the perpetrator commits suicide after the homicide (Websdale 1999). However, some children may not be the target of the homicide or a victim of a familicide, but may become a victim through their presence or their attempt to protect the other parent from violence. These children, who attempt to intervene in acts of domestic violence against a parent, may become victims of homicide just by being in the wrong place at the wrong time (Jaffe et al. 2012; Jaffe and Juodis 2006).

Characteristics of Child Homicide in Domestic Violence Situations

There has been some research that has identified particular characteristics or factors associated with child deaths in a domestic violence context (Division of Criminal Justice Services Office of Justice Research and Performance 2008; Marleau et al. 1999; Websdale 1999). One study found that perpetrator unemployment, an actual separation or the threat of separation, psychological instability, and substance abuse were factors that may have played a role in the killing of a child in a domestic violence incident (Marleau et al. 1999). Websdale (1999) identified three antecedents to child domestic homicide, which include a history of child abuse, prior family involvement with agencies, and domestic violence within the family. Other important factors include past child abduction or threats of abduction, and threats of killing the child (Websdale 1999).

A report from the Division of Criminal Justice Services Office of Justice Research and Performance (2008) identified a number of characteristics of child homicide in domestic violence situations. Children were documented as victims in 36 homicides, totaling nearly 27 % of domestic homicides and more than 4 % of all homicides. Over 90 % of child victims in these incidences were 4 years of age or younger. Documentation noted a history of physical abuse in 13 child victims. Neglect and/or inappropriate care giving was noted in 11 child homicide cases. In reviews of child homicides in other jurisdictions there appears to be a pattern of multiple risk factors including a history of domestic violence, parental mental health problems, and substance abuse (Brandon 2009). Reviews of the literature on paternal filicide (killing of children by a father) point to more men killing children as an act of retaliation for the mother leaving the marriage (Bourget et al. 2007).

Current Study

As noted by the literature, children may be at risk of homicide in the context of domestic violence. Through the identification of child specific risk factors, professionals, agencies, and the community can further efforts to protect these children. In an effort to determine child specific risk factors, case summaries were reviewed from the Ontario DVDRC. The Ontario DVDRC, established in 2002, is a multi-disciplinary advisory committee that assists the Office of the Chief Coroner for the Province of Ontario with the investigation and review of deaths involving domestic violence. Using historical information, interviews with family members, police reports, and multiple agency files pertaining to the conduct of perpetrators, victims, and their respective families, the Ontario DVDRC conducts a review and makes recommendations on how to prevent further deaths in similar circumstances. The Ontario DVDRC (2008) defines domestic homicide as, “homicides that involve the death of a person, and/or his/her child(ren) committed by the person’s partner or ex-partner from an intimate relationship” (p. 31).

For the purpose of this study, 84 cases reviewed by the committee were selected and divided into three separate groups: (a) no child in the home (cases where children do not reside within the family system), (b) no child target (cases where a child resides within the family system, but no attempt was made on his/her life), and (c) child target (cases where a child who resides within the family system was killed or an attempt was made on his/her life). The children in each case were either the biological or stepchildren of the perpetrator and victim. Each individual case was reviewed for potential risk factors and compared across groups. This study was completed with the goal to further knowledge of the unique factors that may specifically place a child at risk for homicide in the context of domestic violence. It is anticipated that such information can support agencies in being more effective in their attempts to prevent domestic homicides.

Method

Participants

The current retrospective study analyzed 84 domestic homicide case summaries obtained from the Ontario DVDRC from 2003 to 2009. These 84 cases represented 90 % of the overall sample from the Ontario DVDRC and were selected based on the study’s inclusion criteria (cases consisting of a heterosexual primary relationship, the perpetrator and victim being between the ages of 18 and 65, and the perpetrator gender being male). Cases are discussed using the terms perpetrator (the person committing the offense), primary victim (adult female partner in the heterosexual relationship who is the primary target of the homicidal violence), and child (individual 18 years of age and under).

All 84 cases were reviewed for child involvement and were subsequently divided into three groups. Group 1 represented “No Child in the Home” defined by the complete absence of children within the family system, neither the perpetrator nor the primary victim had biological and/or adopted children within their direct care. Group 2, “No Child Target” included cases where the perpetrator and/or primary victim had biological and/or adopted child(ren) within the family system, but these children were not attacked or targeted. Group 3, “Child Target” included cases where a child, who resided within the family system, was murdered during an incident of domestic homicide or an attempt was made on their life. Group 1 acted as a control or comparison group, providing an opportunity to exclude predisposing factors common to all domestic homicide cases. Group 2 and Group 3 aided in the identification of predisposing factors specific to child homicide as it allowed for the differentiation between cases when children were present in the family system and targeted or not targeted by the perpetrator.

The present study utilized the Ontario DVDRC database and individual case reports to examine the unique factors that place children at risk of homicide in the context of domestic violence. The information in the database and case reports were previously gathered by the Ontario DVDRC through reviews of police files, files obtained from professionals and agencies involved with the perpetrator and victim(s), and interviews of friends, family members, and co-workers. The amount of information in each case varied as a result of the thoroughness of police investigations and information available on file.

The Ontario DVDRC database was developed from coding information from the multi-disciplinary review of each of these cases. The information coded was the background of the perpetrator and victim, nature of the homicide, and risk factors present. The 12 most common risk factors were selected and are summarized with their definition in the Appendix. Due to the small sample size of child homicide cases (n = 13), only the most common factors were examined as others had such low frequencies that any statistical comparison would not have been meaningful. The definitions in the Appendix have been developed through committee discussion and consensus over the past 8 years. The only factor not self-explanatory was the factor related to perpetrator depression. This factor was often obvious through file review and descriptions of the perpetrator by friends and family; however, the depression was not always diagnosed by health professionals. Therefore, the committee established two categories that differentiated perpetrator depression as viewed by family, friends, and others and perpetrator depression that was diagnosed by a health professional. Information and definitions of all risk factors are available in the appendix of the Ontario DVDRC (2010) eighth annual report.

Procedure

To ensure reliability when extracting the data of interest from the Ontario DVDRC database and case histories, the research assistant for the DVDRC, who had been part of the team who developed the database for the committee (4 years experience), and a graduate student in counseling psychology each coded the data independently for 30 cases. The overall reliability on all factors coded was .95. The graduate student then coded the remaining cases. In some cases, there was missing data from the official DVDRC database which necessitated a review of the case history for factors which required judgment rather than simply transcribing the information, for example the number of contacts with different community agencies. Cases were then divided into three groups: No Child in the Home, No Child Target, and Child Target as described above. Factors that were not available in 50 % or more of the cases were removed from any analysis. Demographic information was analyzed to identify general case characteristics. The various factors identified were then compared amongst groups using, where appropriate, chi-square and t-tests. Initially, the three groups were compared using chi-square. This analysis was followed by a two-group comparison (No Child in the Home and Child Involvement [combining Child Target and No Child Target]) using chi-square and t-tests for continuous variables, such as the number of agencies contacted.

Results

Demographic Information/General Case Characteristics

Separate chi-square analyses were conducted with the three groups (No Child in the Home × No Child Target × Child Target) on variables specifically related to the case in general (Type of Case) and the relationship between the primary victim and the perpetrator (Type of Relationship, Length of Relationship, Actual Separation; see Table 1). Results indicated no significant difference between groups with the Type of Case (χ 2(2) = .073, ns), the Type of Relationship (χ 2(4) = 8.40, ns), and the Length of Relationship (χ 2(6) = 11.31, ns). Results indicated a significant difference between groups with the presence of an Actual Separation between the couple at the time of the homicide (χ 2(2) = 6.91, p < .03); however, when conducting a post hoc analysis by condensing the variables into No Child in the Home and Child Involvement (combining Child Target and No Child Target), and applying a Bonferroni correction, the results indicated no significant difference between the two groups (χ 2(1) = 3.14, ns).

Table 1 Demographic information/general case characteristics
Full size table

A three group comparison (No Child in the Home × No Child Target × Child Target) chi-square analysis was conducted on variables specifically related to the primary victim and the perpetrator (Unemployment and Criminal History; see Table 1). Results revealed a significant difference amongst groups for Unemployment with the Primary Victim (χ 2(4) = 10.46, p < .03); however when extracting a 2 × 2 matrix by condensing variables (No Child in the Home × Child Involvement) and applying a Bonferroni correction, the results indicated no significant difference between the two groups (χ 2(1) = 2.63, ns). Results indicated no significant difference between groups with Perpetrator Unemployment (χ 2(4) = 2.78, ns), Criminal History of the Primary Victim (χ 2(4) = 4.77, ns), and Criminal History of the Perpetrator (χ 2(4) = 2.90, ns).

Agency Contact

A three-group comparison (No Child in the Home × No Child Target × Child Target) was conducted using a one-way ANOVA to determine if there were significant differences with the average number of agency contacts for the family, the perpetrator only, the primary victim only, and the perpetrator and victim combined (see Table 2). Because of the exploratory nature of this study and the multiple comparisons made without a priori hypotheses, a Bonferroni correction was made and the significance level was set at p < .01. Results identified a significant difference with the number of agency contacts amongst the three groups: All Agency Contact with the family (F(2) = 7.77, p < .001); Agency Contact with Perpetrator Only (F(2) = 8.25, p < .001); Agency Contact with the Primary Victim Only (F(2) = 5.43, p < .01) and Agency Contact with Primary Victim and Perpetrator (F(2) = 7.86, p < .001). The post-hoc analyses on all significant results indicated that more agencies were in contact with the family in the Child Target group compared to the No Child in the Home group (see Table 2). To highlight the role of children drawing more professionals into the family, a subsequent analysis was completed combining the two child groups (No Child Target and Child Target, n = 40) versus cases with no children (No Child in the Home, n = 44). Overall, children in the home almost doubled the number of agencies involved and significantly differentiated this population (4.07 vs. 7.28, total number of agency contacts per case (t(82) = 3.31, p < .01), total number of agencies perpetrator was involved with (t(82) = 3.47, p < .01) and total number of agencies involved specific to the perpetrator and victim (t(66) = 2.89, p < .01)).

Table 2 Average number of total agencies involved
Full size table

When examining the percentage of cases that reported risk assessment and management strategies (e.g., a risk assessment completed; assessing child risk; safety planning and/or protection order for the primary victim, including the child in the protection order; providing a protection order for the child; restricting perpetrator access to the child; and safety planning for the child), using chi-square analyses, none of the overall comparisons were significant (see Table 3). It is interesting to note that overall, risk assessments were completed in less than 15 % of the cases. The most common risk management strategy in approximately a quarter of the cases was a protection order.

Table 3 Percentage of cases reporting safety planning
Full size table

Finally, of the 12 most common risk factors identified by the Ontario DVDRC, Sense of Fear of the Primary Victim was the only one that revealed significance (χ 2(4) = 10.61, p < .03) (see Table 4) when comparing the three types of cases. However, a post hoc chi-square analysis that examined the difference between No Child in the Home versus Child Involvement (Child Target and No Child Target combined) revealed no significant result with a Bonferroni correction (χ 2(1) = 1.33, ns). Overall, there were no risk factors that differentiated adult from child homicide cases.

Table 4 Comparison of the 12 most common DVDRC risk factors across three groups
Full size table

Discussion

This exploratory study focused on the risks that children face in the context of domestic homicide. The authors examined domestic homicide cases reviewed by a multi-disciplinary committee to explore unique circumstances and factors that may be present in cases where children were killed. Previous research suggested that children are at serious risk in domestic violence/homicide situations and child deaths may be associated with retaliation (e.g., perpetrators who seek revenge for their partner leaving the relationship), the perpetrator’s feelings of shame for not complying to the gender role expectations of a father or husband, and/or the perpetrator’s dependence on his family that can lead to the perpetrator committing familicide (Bourget et al. 2007; Ewing 1997; Websdale 2010). This study looked at background variables and risk factors present in cases of child domestic homicide and compared them to those evident in cases of adult domestic homicide. The study represented an effort to identify distinct factors that place children at risk of homicide in the context of domestic violence. Homicides where there were no children in the family; children present, but not killed; and children targeted were compared. Although a significant number of risk factors have been reported in past research with adult victims of domestic homicide, there has been more limited research on factors specifically predictive of child domestic homicide (Jaffe et al. 2012; Websdale 1999).

The results from this study did not reveal significant differences between domestic homicides with or without children in the family. One significant background difference identified was the number of agencies involved with each group. Specifically, when comparing the three groups (No Child in the Home vs. No Child Target vs. Child Target), the Child Target group was shown to have the most community agency involvement when compared to cases with No Child in the Home. Overall, cases with children had almost twice as many agencies involved compared to cases without children. There are a number of obvious explanations that may account for this result. First, the addition of children to the family system generally increases the number of individuals within that system, which alone may account for the overall increase in agency involvement. Second, the community at large has a higher number of services specifically mandated to assist children as a vulnerable population. As such, the family may be involved with a larger number of agencies as more are available to them. It is also important to consider that most services available to children are free or state mandated (e.g., schools, child protection service) and increase the likelihood of more professionals and agencies monitoring children’s well being (e.g., school attendance, abuse reports). Third, separating parents are more likely to be involved with family court and court-related professionals (e.g., mediators, family law lawyers) in regards to potential child custody and support conflicts.

Previous research has identified that child domestic homicide is often preceded by prior family involvement with agencies, domestic violence within the family, and a history of child maltreatment (Websdale 1999). Furthermore, research has indicated that children exposed to domestic violence are more likely to experience child maltreatment, which would increase the likelihood of child protection involvement and referral to other agencies for counseling. In many jurisdictions, including the one under study, there are formal policies that require police and teachers to report to child protection agencies if a child is living with domestic violence, which in itself is considered a form of emotional abuse and potential endangerment (Jaffe et al. 2011).

Although a larger number of agencies involved with a family should provide a greater opportunity for intervention and prevention of homicides, the present study did not reflect enhanced risk assessment and management strategies. This study focused on cases reviewed by a domestic homicide committee so every case involved a tragic outcome. There is no comparison to other cases without these outcomes in this study, but the overall low rate of risk assessment is concerning. There were no significant differences amongst the cases irrespective of children being involved. Less than 15 % of all cases showed evidence of any formal risk assessment and only 1 of the 40 cases involving children had a specific assessment of child risk. In only 10 % of all the cases there were indications of a safety plan. The most common risk management strategy reported was a protection order (24 %, 20 out of 84 cases). In over 25 % of the cases involving children, there was restricted access to the perpetrator. These findings reveal several missed opportunities for intervention and prevention strategies by professionals and indicate the importance of risk assessment and safety planning as mandatory strategies for all professionals and agencies involved in responding to domestic violence (Jaffe et al. 2008).

The case outcomes in this study indicate that opportunities for risk assessment and risk management are not being utilized on a consistent basis contrary to the recommendations of many domestic violence death review committees across North America (Jaffe et al. 2012). These systemic gaps must be addressed and resolved in an effort to prevent future homicides. As an example of this problem, only 6 out of 13 Child Target cases showed contact with Child Protection Services (CPS). Aside from the question of why CPS was only involved in less than half of these cases, only 23 % of Child Target cases showed a completed risk assessment for the adult victim and only 15 % showed evidence of child protection orders. The perpetrator’s access to the children was restricted in just half of these cases.

A major goal of the study was the exploration of differentiated risk factors for child homicides in contrast to adult homicides. The death review committee in this study examined a host of factors identified in previous research as commonly associated with domestic homicides. Twelve common factors were examined and 11 of them showed no difference amongst the three groups. One factor, the victim’s intuitive sense of fear was higher for the no child involvement group. Overall, there can be no conclusions drawn as to how to differentiate amongst the cases and the best advice is to consider children to be at potential risk if their mother is at risk. These findings support the argument that children should also be protected when community agencies believe mothers are at risk of homicide. This reinforces the necessity of including children in risk assessments, safety planning, and protection orders and establishing supervised access specific to children.

The vast majority of cases had multiple risk markers and the homicide seemed predictable and preventable with hindsight and perhaps foresight. In 80 % of the larger sample of cases from which these homicides were drawn, there were seven or more known risk factors upon review suggesting critical information was held by professionals involved with the family (Ontario DVDRC 2008). This study suggests that few of the domestic homicide cases appeared to have a completed risk assessment, safety planning, or protection order put in place for the primary victim (13 % of all cases had a completed risk assessment, 6 % had a safety plan established for the primary victim, and 24 % had a protection order put in place for the primary victim).

Risk assessment and safety planning is imperative to predicting and preventing domestic homicide. There are multiple domestic violence risk assessment tools reported in the literature that assess the risk of repeated assault or the risk for lethality and that are essential in identifying, monitoring, and managing risk to a family experiencing domestic violence (Domestic Violence Advisory Council 2009; Kropp 2008). These tools are used by several agencies that may be involved with at-risk families including police, shelters, justice, and child protection services, with the overall goal of identifying risk and establishing effective safety plans. Other researchers suggests that these tools are not used often enough, even in extreme circumstances when abused women are coming before the court and seeking protection orders (Nichols-Hadeed et al. 2012). Although there are no tools specific to identifying risk for children in domestic violence circumstances, it can be a safe working assumption by this study’s findings that identifying a risk for lethality with the primary victim may also indicate a risk to children.

Limitations

There are notable limitations in the study that restrict the strength of the previously noted findings. This study’s use of a secondary data set to ascertain information on domestic homicide had considerable missing data. Information for this study was gathered from Ontario DVDRC case reports (a post-hoc analysis and event summary). Case reports may be limited as the Ontario DVDRC may not have access to all relevant reports and interviews. Some cases may have less information available as the family does not have a lengthy history of involvement with community agencies or the court. Cases that result in a criminal trial have more thorough information than the homicide–suicide cases, which may not be deemed a priority by investigators.

Another significant study limitation is the small sample size related to child deaths. Of the 84 domestic homicide cases gathered from the Ontario DVDRC, only 13 were Child Target cases. Statistically speaking, 13 Child Target cases may not provide sufficient power to detect differences that are particularly robust, and therefore, may not adequately identify unique factors that place a child at risk of homicide when domestic violence is present within the family system. The small sample size limited the number of risk factors that could be meaningfully examined.

Implications

In spite of the above limitations, there are some interesting findings that raise considerations for future research in this area, which is significantly lacking. Related studies in the field have not used comparison groups and as a result can only provide descriptive information about child homicides rather than differentiate factors that place children at risk (Marleau et al. 1999; Websdale 1999). This study is unique as it addresses this gap and hopefully provides a foundation for future research. Although the current study did not find variability in risk factors between adult domestic homicide and child homicide in the context of domestic violence, further research studies in this area, with larger sample sizes and comparison groups, may identify risk factors specifically related to child domestic homicide.

Aside from specific risk factors, it would be important to review existing risk assessment tools to see if they differentiate children from adults at risk in domestic violence cases. Child homicide risk factors can provide policy makers with essential information to aid in the development of enhanced practices and protocols pertaining to child safety. This knowledge can promote the use of current risk assessment tools to identify risk of homicide for both adults and children. It would be critical to educate professionals and the community at large about the importance of conducting risk assessments and establishing safety plans that include the children within the family.

The present study can serve as a springboard for a more in-depth study further identifying the unique factors that place children at risk of homicide when domestic violence is present within the family system. Future researchers could increase their sample size by collaborating with other death review committees and working towards creating a common and more extensive database. Additional comparison groups, including Child Target cases not within the context of domestic violence from child death review organizations, can examine other factors predictive of child homicide that are not in the context of domestic violence. Studies also need to examine the motives of perpetrators to commit these murders which could assist front-line professionals in their screening and risk assessment interviews (Websdale 2010).

It was evident throughout this study that there are significant gaps in community agency services and missed opportunities for intervention and prevention of homicides. Although there is no guarantee that any individual homicide could have been prevented, the number of agencies who had prior contact with family members raises critical questions. These questions are especially pointed in the face of multiple risk factors in most cases and minimal efforts at assessment and safety planning. Future research should complete a more thorough and qualitative review of agency contact with families and individuals at risk to assist in the identification and elimination of these gaps. A review of agency coordination noted through recorded incidences of information sharing and meetings amongst professionals and agencies may also provide information on how to better protect children and adult victims from the tragic outcomes in domestic homicides.