Objectives. We estimated the frequency and examined the characteristics of intimate partner homicide and related deaths in 16 US states participating in the National Violent Death Reporting System (NVDRS), a state-based surveillance system.
Methods. We used a combination of quantitative and qualitative methods to analyze NVDRS data from 2003 to 2009. We selected deaths linked to intimate partner violence for analysis.
Results. Our sample comprised 4470 persons who died in the course of 3350 intimate partner violence–related homicide incidents. Intimate partners and corollary victims represented 80% and 20% of homicide victims, respectively. Corollary homicide victims included family members, new intimate partners, friends, acquaintances, police officers, and strangers.
Conclusions. Our findings, from the first multiple-state study of intimate partner homicide and corollary homicides, demonstrate that the burden of intimate partner violence extends beyond the couple involved. Systems (e.g., criminal justice, medical care, and shelters) whose representatives routinely interact with victims of intimate partner violence can help assess the potential for lethal danger, which may prevent intimate partner and corollary victims from harm.
Intimate partner violence (IPV) is a serious public health problem that affects millions in the United States. IPV is defined as physical violence, sexual violence, stalking, or psychological aggression (including coercive tactics) by a current or former intimate partner.1 IPV that is severe enough to lead to injury or significant harm is primarily but not always perpetrated by men. Estimates from the 2010 National Intimate Partner and Sexual Violence Survey indicate that more than 74 million people in the United States have experienced IPV (physical violence, sexual violence, stalking) at some point in their lives, and more than 12 million in the previous 12 months.2 In the United States, IPV disproportionately affects women, especially racial/ethnic minorities.2,3
The most extreme form of IPV is intimate partner homicide (IPH). In 2007, intimate partners committed 14% of all US homicides, and 70% of those victims were female.4 Although IPH has decreased during the past 15 to 20 years,4 it remains a disturbing possibility for people experiencing abusive relationships. Across studies, major risk factors for IPH consistently include previous domestic violence, unemployment, access to firearms, estrangement, threats to kill, threats with a weapon, previous nonfatal strangulation, a stepchild in the home (if the victim is female), and previous mental health problems of the perpetrator (for homicide–suicide).5,6 Of these, previous IPV is the strongest predictor.6 Furthermore, homicides followed by suicide of the perpetrator are more than twice as likely to be committed by former or current spouses as by other perpetrators and are significantly more likely to involve firearms than other weapons.6,7 It is estimated that one third of IPHs in the United States involve suicide of the perpetrator, who is most often male.6,8
The Centers for Disease Control and Prevention estimates that societal costs resulting from IPV victimization approach $6 billion annually.9 Such cost estimates and scientific studies of IPH have largely focused on intimate partners of the perpetrator (e.g., spousal homicides). However, a substantial portion of IPV-related homicide victims are not the intimate partners themselves. These corollary victims may be family members, friends, neighbors, persons who intervene in IPV, law enforcement responders, or bystanders. Previous studies10,11 have used the term “collateral victims” to refer to non–intimate partner victims in situations stemming from IPV. Because of the colloquial usage of “collateral” and out of concern for the negative connotations associated with the word, we selected the word “corollary” to refer to non–intimate partner victims whose death is connected to IPV.
Few studies have examined corollary victims or included them in analyses of IPH.10,11 In one exception, a British study examined murder connected to intimate partner conflict and found that 37% of the 166 victims were not intimate partners of the murderer.10 Instead, the victims were children of the intimate partner, allies (e.g., relatives, neighbors, friends, lawyers connected to the abuse victim), or new partners. In the United States, it is difficult to estimate the magnitude of corollary victimization. Existing data systems, such as the Uniform Crime Reports, often use categories that do not provide the details necessary to understand the relationships among the victims and offenders.12 For example, if an ex-husband kills his ex-wife’s new partner, the new partner may be categorized as an acquaintance. Furthermore, same-sex intimate partners are categorized as acquaintances in lieu of a more specific designation (e.g., romantic partner, domestic partner). At state and local levels, investigations conducted by fatality review boards may reveal the proportion of collateral victims resulting from IPV, but those figures are often not widely reported.
We examined IPH data gathered between 2003 and 2009 by the National Violent Death Reporting System (NVDRS), a state-based surveillance system. Our objective was to extend the existing literature on the frequency and characteristics of IPH and on corollary homicides that occur in the context of IPV and IPH.
NVDRS is a state-based active surveillance system that collects data on all suicides, homicides, legal intervention deaths, unintentional firearm deaths, and deaths whose manner is undetermined. NVDRS data are used in the investigation and prevention of many different types of violence, including IPV,13–15 child maltreatment,16 gang violence,17 suicide,18 and homicide–suicide.19
NVDRS links data from multiple sources (e.g., death certificates, coroner or medical examiner reports, law enforcement reports) into a single incident record. States manage data collection through state health departments or a subcontracted entity, such as a medical examiner’s office. Data are coded by trained abstractors, who manually or electronically extract the data from the different types of records (e.g., death certificate files, medical examiner records). NVDRS records include an incident narrative, which is generated by the abstractor and based on the coroner or medical examiner and law enforcement reports. The narratives contain a chronology and description of circumstances leading to the death. The circumstances are defined as events that preceded or occurred during the incident and may have contributed to the infliction of a fatal injury. Circumstances are derived from the coroner or medical examiner record and police reports.20
Cases in our analysis originated from data submitted by the 16 states participating in NVDRS. Data were available for Alaska, Maryland, Massachusetts, New Jersey, Oregon, South Carolina, and Virginia (2003–2009); Colorado, Georgia, North Carolina, Oklahoma, Rhode Island, and Wisconsin (2004–2009); and Kentucky, New Mexico, and Utah (2005–2009). Deaths occurred by homicide, homicide–suicide, legal intervention, and undetermined intent. We first selected a pool of cases that included IPHs and other deaths connected to IPV. We incorporated cases in the initial pool if the records indicated any of the following circumstances: IPV related; intimate partner problem; jealousy or love triangle; other argument, conflict, or abuse; or the victim–suspect relationship was spouse or intimate partner. Next, for cases that did not specify that deaths were IPV related or where the suspect was not an intimate partner, we manually reviewed the narratives and used the Centers for Disease Control and Prevention definition of IPV to determine eligibility for inclusion in the analysis: physical violence, sexual violence, stalking, or psychological aggression (including coercive tactics) by a current or former intimate partner.1 We included cases coded as jealousy or love triangle only when they concerned an actual relationship (vs unrequited interest). We excluded incidents that involved sex or drug trades (i.e., transactional relationships rather than ongoing or former intimate partnerships) and incidents motivated by a desire to end the suffering of a terminally ill loved one, where the victim wanted to die (i.e., mercy killing).
We categorized corollary homicide victims according to a definition published by Project Safeguard: “the murder of other people that occurs in the context of a domestic violence incident (such as new intimate partners, intervening friends, family or strangers, or responding law enforcement officers).”11(p11) As shown in Figure 1, we categorized the 718 corollary victims by how they were linked to the suspect: family, defined as a blood relative of the suspect or persons connected to the suspect through a familial relationship, such as the boyfriend of a child’s mother; other intimate partner involvement, defined as being connected to the suspect through a mutual intimate partner, currently or in the past (e.g., love triangle, woman’s new partner was killed by her former partner; woman’s new boyfriend and her ex-spouse killed each other during shootout); friend or acquaintance; stranger; or police officer, slain during a response to an IPV incident.
To protect privacy while illustrating various aspects of corollary victimization, we created fictional scenarios as examples.
We identified 4470 persons who died in the course of 3350 IPV-related incidents. Of the 4470 deaths, 81% were homicides, 18% were suicides, and fewer than 1% were categorized as legal intervention or undetermined intent. Most incidents resulted in a single death (71.5%); 28.5% involved multiple deaths. The 955 multiple-death incidents (median = 2 decedents; range = 2–7 decedents) resulted in a total of 2075 decedents; 1120 of these victims were killed in addition to the primary targeted victim. These included corollary homicide victims and deaths of the perpetrator by suicide or legal intervention.
Of all homicides in the analysis (n = 3619), intimate partners and corollary victims represented 80% and 20% of victims, respectively; fewer than 1% of corollary victim deaths were categorized as undetermined intent. The remaining victims (n = 849) were IPV perpetrators who committed suicide during a homicide–suicide incident or were killed by police in a legal intervention; fewer than 1% of perpetrator deaths were categorized as undetermined intent. Descriptions of decedents are presented in Table 1 and Figure 1.
Characteristics of Intimate Partner and Corollary Homicide Victims: National Violent Death Reporting System, United States, 2003–2009
Characteristics | Overall (n = 4470), % or Mean ±SD (Range) | Intimate Partner Victims (n = 2903), % or Mean ±SD (Range) | Corollary Victims (n = 718), % or Mean ±SD (Range) | Perpetrator Victims (n = 849), % or Mean ±SD (Range) |
Gender | ||||
Female | 55.1 | 77.0 | 27.3 | 3.9 |
Male | 44.9 | 23.0 | 72.7 | 96.1 |
Race/ethnicity | ||||
White, non-Hispanic | 53.0 | 52.0 | 49.3 | 59.7 |
Black, non-Hispanic | 33.0 | 35.3 | 33.3 | 24.7 |
Hispanic | 9.0 | 7.8 | 12.1 | 10.4 |
Other/unknown | 5.0 | 4.9 | 5.3 | 5.2 |
Age, y | 37.9 ±14.7 (0–92) | 38.5 ±13.1 (11–90) | 29.4 ±17.8 (0–92) | 43.1 ±13.8 (16–92) |
Note. Data were available for AK, MD, MA, NJ, OR, SC, and VA (2003–2009); CO, GA, NC, OK, RI, and WI (2004–2009); and KY, NM, and UT (2005–2009).
Intimate partners accounted for 80% (n = 2903) of the homicide deaths in our sample. These victims were current or former intimate partners of the suspect and predominantly female (77%). The mean age of the intimate partner victims was 38.5 years (SD = 13.1; range = 11–90 years; Table 1, Figure 1). More than half of the victims (54.1%) were killed with a firearm, followed by sharp instrument (25%); hanging, strangulation, or suffocation (8.4%); blunt instrument (5.3%); personal weapons (e.g., fists, 3.2%); and other weapons (4%). About 81% of the incidents occurred in a residence.
Corollary victims represented 20% (n = 718) of the homicide deaths in the sample. Corollary victims were predominantly male (72.7%), with a mean age of 29.4 years (SD = 17.8; range = 0–92). Figure 2 shows the age distribution of corollary homicide and IPH victims. One quarter (25.4%) of corollary victims were aged 17 years or younger (Table 1, Figure 1). Nearly half of corollary victims who were family members of the suspect were minors, and more than one third were elementary school aged or younger. Approximately 70% of corollary victims were killed with a firearm, followed by a sharp instrument (12.4%), and other weapons (17.3%). More than three quarters (78.6%) of corollary homicides occurred in a residence.
This category comprised homicide victims who were children, parents, siblings, other family members of the suspect (including stepfamily), and persons connected through a family relationship (e.g., boyfriend of child’s mother). More than one third (38%) of family member homicide victims were aged 11 years or younger, and 48.3% were aged 17 years or younger. In many cases, the victims simply had the misfortune of being in the home at the time of the incident. In some cases, the homicide victim attempted to intervene in an IPV situation and was killed. An example: Victim 1 (male, aged 14 years) was killed during a domestic incident between his mother and father (suspect). The suspect was drinking throughout the day and got into an argument with his wife. The argument became physical, and the suspect pulled out a gun and began threatening his wife; he then shot and wounded her. Victim 1 tried to grab the gun and was shot in the process.
Victims were new intimate partners of the suspect’s former intimate partner (e.g., suspect’s ex-wife’s new boyfriend), former intimate partners of the suspect’s current intimate partner (e.g., suspect’s girlfriend’s ex-husband), another current intimate partner (e.g., infidelity situations, nonmonogamous relationships). An example: The female victim (victim 1) and her boyfriend (victim 2), were watching TV at the home of victim 1. Witnesses stated that the male suspect (ex-boyfriend of victim 1) began banging on the door and screaming. The victims did not open the door but told the suspect to leave or they would call the police. The suspect then kicked the door open and shot both victims.
These victims knew the couple and were present during an IPV incident. An example: Victim 1 was driving his car when the ex-boyfriend of his passenger, a female friend with whom victim 1 was not intimately involved, pulled up beside the car and began shooting. One of the bullets killed victim 1.
These homicide victims were strangers who were killed during an attempt to harm the intended victim or other persons whose relationship to the suspect was unknown. For example: Two roommates died in an apartment fire. The male suspect who set fire to the building had recently threatened his ex-girlfriend, who lived in the same apartment building but was not home at the time of the incident.
These victims were killed in the line of duty when responding to an IPV incident. An example: Victim 2 was on duty when he was shot by the male suspect, who had killed his wife (victim 1) shortly before. The suspect ambushed victim 2 as he was attempting to enter the home of victim 1 in pursuit of the suspect. Other officers on the scene fatally shot the suspect.
Ours was the first study to our knowledge to examine, across multiple US states, the frequency and characteristics of IPH and corollary homicides (involving victims other than the current or former partner) that occurred in the context of IPV. Our results extend the existing literature on IPH and corroborate previous research showing that most perpetrators in lethal IPV are male.3,21 Our findings demonstrate that the burden of IPV extends beyond violence experienced by intimate partners. Intimate partners and corollary victims represented 80% and 20% of homicide victims, respectively. Corollary homicide victims were often young; they were connected to the suspect through an intimate partner (e.g., new partner of suspect’s ex-partner) or as family members, friends, acquaintances, or police officers on the scene. Some were complete strangers to the couple involved in the IPV.
Corollary victims represent a distinct group of individuals who come into contact with IPV, either through their relationship to one or both members of a couple experiencing IPV or through proximity to the violence. Many of the corollary victims we identified were young. The percentage of child victims in IPV-related homicide is alarming and emphasizes the significant ripple effects of IPV. These data may be useful for informing children's safety policies and intervention and prevention strategies directed at families exposed to IPV.
Fatality review boards are an important resource for understanding these types of homicide. Currently, domestic violence review teams operate in 41 US states.22 Their in-depth reviews can support the development of prevention strategies to protect not only the intimate partner, but also family, friends, and others who may be exposed to the violence.
Our findings revealed interesting victimization patterns with respect to gender. Female homicide victims were typically current or former intimate partners of the suspect. We found that 77% of the intimate partner victims of IPH were female, supporting existing evidence that women are more likely than men to experience IPH and that men are more likely to perpetrate it.4 By contrast, men were overwhelmingly represented in suicide deaths and suspected perpetration of multiple-death incidents. Our data also showed that nearly three quarters of corollary homicide victims were male. Intuitively, this makes sense, because a portion of those victims are new boyfriends and partners killed by the former partner. However, some of the victims were family members and others who intervened in violence. Male socialization and beliefs about masculinity may influence men and boys to physically intervene in IPV situations, putting them at risk for serious injury or death.23,24 It is essential for communities to promote strategies that bystanders can use to safely intervene in IPV and other violence. One example is Green Dot, a program that has shown promise through training adolescents to be proactive bystanders in preventing dating and sexual violence.25 Published evaluations of Green Dot’s influence on dating violence are not available, but it is effective in increasing bystander actions among students exposed to the program.
Although homicide is a somewhat rare outcome in IPV, effective prevention strategies could avert serious and fatal injuries. These range from immediate strategies, such as intervening in current IPV to prevent serious injury or homicide, to more long-term, primary prevention efforts, which aim to keep IPV from occurring at all.
One immediate strategy that holds potential for preventing IPV homicide is the Danger Assessment, an instrument used to assess the likelihood of being killed or seriously injured by a current or former intimate partner.26 Retrospective testing of the revised Danger Assessment revealed that 79% of victims of attempted homicide by their partners scored in the 2 highest categories of danger (severe and extreme); only 14% of a community-drawn sample of IPV victims scored in these 2 categories.26 These results demonstrated that the Danger Assessment shows good specificity and sensitivity at identifying potential IPH victims when the lower of these 2 categories (severe danger) is used as the threshold cutoff for high risk of femicide. Lethality assessments may be used by shelter staff to prioritize admission and by criminal justice professionals to assess risk among IPV victims and take appropriate action.26 Further studies are needed to determine whether lethality assessments designed for femicide are applicable to male IPV victims as well as to IPV-related corollary homicide.
We found that most homicides occurred in homes and were committed with a firearm. Previous research has highlighted the association between firearms in the home and risk of violent death.27 For example, IPV incidents that involve a firearm are 12 times as likely to result in death as incidents that do not involve a firearm.28 IPV incidents that involve a firearm also increase the chance of an IPH followed by the suicide of the perpetrator.7,19 Some states have implemented policies aimed at reducing the likelihood of IPH. For example, as of 2008, 18 states had laws giving police the authority to remove firearms when responding to domestic violence incidents; 20 states and the District of Columbia had laws authorizing courts to remove firearms when issuing protective orders.29 However, many of these policies have not been adequately evaluated to determine their effectiveness in preventing IPH; thus, it is important to identify effective strategies for preventing the use of firearms during incidents of IPV.27
Distal and primary strategies aim to prevent IPV. Many researchers suggest that true primary prevention of IPV must target individuals before they become involved in intimate relationships, and most of the evidence-based programs for the prevention of IPV are targeted at adolescents or preadolescents. A few of these programs, such as Safe Dates30,31 and Fourth R,32 are school-based curricula targeted at young adolescents (eighth and ninth grades, respectively) and have demonstrated longitudinal effects in preventing dating violence. A recently evaluated program targeting parents and their early-adolescent children with in-home activities also shows promise for preventing adolescent dating violence.33 Dating Matters: Strategies to Promote Healthy Teen Relationships, a more comprehensive approach to prevention of adolescent dating violence, is being tested in a cluster-randomized controlled trial by the Centers for Disease Control and Prevention.34 Other programs that intervene with established adult couples and individuals have been developed and evaluated for prevention of IPV, but some are still in the testing and analysis phase, and others have not been evaluated rigorously enough to determine true effectiveness.35
We analyzed data from 16 states that participated in NVDRS, so our findings are not a national representation of intimate partner and corollary homicides. We were unable to provide an estimate of the proportion of homicides that were committed in self-defense or in the defense of another. The use of self-defense is often determined during the investigative process and legal proceedings, and the NVDRS does not have a circumstance code to reliably capture such situations.
We were also unable to include related deaths that occurred in non-NVDRS states (e.g., a man who killed his ex-partner in an NVDRS state and then committed suicide in a non-NVDRS state). Finally, we were unable to conduct comparisons between victims who were former and current intimate partners of the perpetrator because for some cases we could not determine whether the relationship was current.
Ours was the first study to our knowledge to quantify and describe intimate partner and corollary homicides in a large sample spanning multiple states. Our findings provide improved understanding of the nature of these homicides and the range of lives that are lost as a result of these incidents. Previous IPH research has, understandably, centered on the deaths of intimate partners. Our findings may help expand the focus of future studies and prevention efforts to include corollary victims.
Thousands of people in the United States die as a result of IPH, many of whom are not the IPV perpetrator or victim and too many of whom are children. Through the use of primary and secondary prevention strategies, systems whose representatives interact with IPV victims (e.g., criminal justice, children's protective services) may be better equipped to identify IPV perpetrators and help victims assess their level of lethal danger and protect themselves and possibly others from harm.
Human Participant Protection
Protocol approval was not needed because the data were derived from routine injury surveillance.