Murder-Suicides Perpetrated by Adolescents: Findings from the National Violent Death Reporting System

Abstract

Murder-suicide is a rare and especially tragic form of violence. Nascent literature has characterized murder-suicide perpetrated by adults, but no study has focused on adolescents who perpetrated murder-suicide. This study evaluated all cases of murder-suicide perpetrated by persons under age 21 that were included in the National Violent Death Reporting System, and categorized each incident using Joiner’s perversion of virtue framework of murder-suicide. Forty-seven incidents of murder-suicide perpetrated by adolescents were identified, representing 56 victims. Most cases did not have sufficient information to be categorized within a particular perversion of virtue. Among the 20 cases that did have sufficient detail to be categorized, incidents were identified as perversions of self-control (n=11), justice (n=8), and mercy (n=1). Most perpetrators were male, and the majority of victims were female. Additional victim, perpetrator, and incident characteristics are described. Our findings extend previous work by employing a coding system based on a theoretically-derived framework for categorizing “genuine” murder-suicide and related behavior among adolescents. This study also characterized the nature of this complex and tragic sequence of behavior among adolescents, who are at elevated risk of suicide, and who may benefit from prevention efforts that effectively address means safety, suicidality, and intimate partner violence.

Introduction

Murder-suicide is a rare event, and the incidence in the United States has remained relatively constant at approximately 0.2–0.3/100,000 per year (Eliason, 2009). Despite the low base rate of murder-suicide, there are myriad catastrophic and durable consequences for families and communities in the wake of murder-suicide incidents in addition to the individual lives lost (Joiner, 2014; Podlogar, Gai, Schneider, Hagan, & Joiner, 2018). These consequences include far-reaching social and emotional impacts for those who survive the death of loved ones, as well as substantial logistical and financial burdens associated with complex investigations and postvention that follow after a murder-suicide incident (Joiner, 2014). Nascent scholarship has evaluated characteristics of murder-suicide perpetrators and victims across several contexts, including comparisons of military and civilian perpetrators (Patton, McNally, & Fremouw, 2017), in adults of different ages (Eliason, 2009), and within the context of intimate partnerships (Malphurs & Cohen, 2002; Salari & Sillito, 2016). However, murder-suicide perpetrated by adolescents has not been widely addressed in the literature and may offer insight into the nature of murder-suicide across different developmental contexts.

Prior research has compared murder-suicide to homicide only and to suicide only, but murder-suicide is a distinct event that should be examined separately since it has characteristics that distinguish it from both homicide and suicide (Panczak et al., 2013; Podlogar et al., 2018; Joiner, 2014). Joiner (2014) asserts that murder-suicide is characterized by suicide as a primary motivation that leads to a “perversion of virtue,” which occasions the perception that it is necessary, just, or merciful to murder another person before enacting a pre-determined suicidal plan (p. 5). In this framework, suicidal intent is the primary antecedent to murder-suicides (Joiner, 2014). Joiner (2014) posits that “genuine” murder-suicide is not primarily an impulsive or homicidal act but rather a sequence of primary suicidal intent that results in the perversion of four virtues: mercy, justice, duty, and glory. He also defined two related but distinct categories that include the perversion of self-control and the perversion of fate (Joiner, 2014; Podlogar et al., 2018). In contrast to homicide, studies of murder-suicide have noted relatively lower rates of drug and alcohol intoxication among murder-suicide perpetrators and victims (e.g., Eliason, 2009; Marzuk, Tardiff, & Hirsch, 1992).

It remains unclear as to what extent murder-suicide is best conceptualized, and under what conditions, as a subtype of homicide or a subtype of suicide (Podlogar et al., 2018). Large and colleagues (2009) noted that “homicide-suicide appears to be closer in epidemiological terms to homicide than suicide in regions with high rates of homicide and measures to reduce homicide in these regions may also reduce homicide-suicide.” This observation is also reflected in Joiner’s (2014) definition of “neighboring but distinct” (p. 177) categories of murder-suicide, one of which is perversion of self-control. The perversion of self-control, or “taking things into one’s own hands” (p. 178), suggests that a person engages in homicide without pre-existing suicidal intent, tries and fails to escape the consequences of having committed murder and thus enacts a suicidal behavior to control one’s destiny (Joiner, 2014). Therefore, in cases where the perversion of self-control underlies murder-suicide, efforts to prevent homicide broadly would be potentially most effective. In contrast, cases of “genuine” murder-suicide (i.e., where suicide is the primary motivation; Joiner, 2014) may be best prevented via efforts focused on mitigating the risk of suicide.

Eliason (2009) reported that there were differences between murder-suicides committed by younger vs. older adults. Among intimate partner murder-suicides, younger perpetrators (aged 18 to 44) were found to exhibit primarily homicidal intent and older perpetrators (aged 60+) primarily suicidal intent (Salari & Sillito, 2016). Prior studies have reported that the mean age of perpetrators of murder-suicide was 40 to 50 years old, which is older than those who commit homicide alone, and most samples have been skewed toward this older age group (Eliason, 2009). To our knowledge, no other studies have specifically focused on murder-suicides perpetrated by adolescents. It is possible that youth who engage in murder-suicide demonstrate motives that reflect particular aspects of the developmental context of adolescence, which may have specific implications for understanding and preventing murder-suicide in this age group.

To expand on previous efforts to categorize murder-suicide using descriptive typologies (e.g., Marzuk et al., 1992), it is necessary to consider theoretically-derived explanations of murder-suicide that can account for patterns of behavior across diverse cases and that involve different types of perpetrators. Previous research on murder-suicide has called for application of Joiner’s typology (Patton et al., 2017), which may promote a clearer understanding of murder-suicide as either a subtype of murder or suicide in particular cases.

Thus, the purpose of this study was to describe murder-suicide incidents perpetrated by adolescents aged less than 21 years (Hardin, Hackell, & Committee on Practice and Ambulatory Medicine, 2017) and apply Joiner’s (2014) perversion of virtue framework. Given the dearth of literature in this area, our study was exploratory and descriptive in nature, and sought to develop and apply a systematic coding scheme for categorizing murder-suicides using data from the National Violent Death Reporting System.

Methods

Data Source

The National Violent Death Reporting System (NVDRS) is a state-based surveillance system that links data on violent deaths (e.g., homicides, suicides) from death certificates, coroner/medical examiners, law enforcement agencies, and toxicology reports (Paulozzi, Mercy, Frazier, & Annest, 2004). Trained NVDRS abstractors code detailed information from these sources on characteristics of the victim, perpetrator, incident and precipitating circumstances. Each incident also has two qualitative narratives summarizing the findings from the coroner/medical examiner reports and law enforcement reports.

Study Population

This study used data from the NVDRS Restricted Access Database, which contains deaths from the 32 states that contributed data for one or more years from 2003 to 2016 (Alaska, Arizona, Colorado, Connecticut, Georgia, Hawaii, Illinois, Indiana, Iowa, Kansas, Kentucky, Maine, Maryland, Massachusetts, Michigan, Minnesota, New Hampshire, New Jersey, New Mexico, New York, North Carolina, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, Utah, Vermont, Virginia, Washington, and Wisconsin). Incidents were included if: (1) the incident type was classified as a single or multiple homicide followed by suicide and (2) perpetrators were less than 21 years old. The NVDRS coding manual specifies that homicides and suicides are linked to a single incident if the injuries occur within 24 hours of each other, and the source documents indicate a clear link between the deaths (Centers for Disease Control and Prevention (CDC), 2016).

Variables

Demographic characteristics available for study for the victim and perpetrator included sex, race/ethnicity, and age (years). Age of perpetrator was continuous and also dichotomized as <18 or ≥18 years. Relationship status at the time of death was categorized as: intimate partner (further split into current/unspecified and former); family, non-intimate partner (e.g., sibling); acquaintance (e.g., friend, schoolmate), and stranger or law enforcement. Homicide method was categorized as firearm or other. Other includes sharp or blunt instrument (e.g., knives, bats, clubs), hanging/strangulation, or other (e.g., hands, poisoning, motor vehicle). If a firearm was used, firearm type was classified as handgun, or shotgun/rifle. Location of the homicide was coded as house/apartment, street/sidewalk/alley, motor vehicle, or other (e.g., natural area, school).

Murder-suicide incidents were categorized using a coding scheme that was developed for this study based on Joiner’s (2014) perversion of virtue framework. Four virtue categories (mercy, justice, duty, glory) of genuine murder-suicide, and two “neighboring but distinct” categories (i.e., perversions of self-control and fate) were defined according to Joiner’s (2014) descriptions (complete definitions provided in Table 2). The structured coding scheme included explicit definitions of each category, prototypical examples to guide coders, and notes as to clear points of distinction between adjacent categories.

Based on a review of all narrative information from the coroner/medical examiner and law enforcement reports, each incident was assigned to a category indicating one of the perversions of virtue from Joiner’s (2014) framework. Four authors read all narratives to determine which code predominated. For all cases, the authors met and discussed the case until a consensus about the most appropriate category was achieved. Given the small number of cases in this dataset, we did not calculate interrater reliability formally, since most of our cases would have been included in the initial cases used to train coders in how to apply the coding scheme. We present example narratives for each observed category with alteration to protect privacy. In the narrative review, we also created several indicator variables. Four were created specifically for cases perpetrated by intimate partners: 1) whether there was a documented history of intimate partner violence (IPV); 2) whether there was an argument at the time of the incident; 3) whether there was a break-up preceding the incident; and 4) whether the incident was related to IPV-related legal problems (e.g., issuance/violation of a protection order). In addition, we created indicators for whether suicidality began prior to, or was concurrent with, the act of homicide and whether the perpetrator had a history of mental illness or psychiatric care, including substance abuse or dependence.

Analysis

Given the modest number of incidents that were available for analysis, no associations were hypothesized, and no inferential statistics were calculated. Descriptive statistics were calculated to characterize the characteristics of victims, perpetrators, and incidents. Unknown data is noted in the Table. Categories were combined to ensure a cell count of ≥ 5 per NVDRS guidelines (CDC, 2016). Data were analyzed using Stata 15.1 (StataCorp).

Results

There were 47 murder-suicide incidents and perpetrators which resulted in the death of 56 homicide victims. As is reported in Table 1, the majority of victims were female (n = 33, 58.9%) and 12 to 20 years of age (n = 39, 69.6%), and half were White (n = 28, 50.0%). Over 40% of victims were identified as a current or former intimate partner of the perpetrator (n = 23, 41.1%), and approximately one-quarter (n = 15, 26.8%) were family members of the perpetrator. Nearly all perpetrators were male (n = 44, 93.6%), the majority were 18 years of age or older (n = 36, 76.6%), and over 40% were Black (n = 20, 42.6%). For nearly a fifth of perpetrators (n = 8, 17.0%), there was evidence of suicidal intent prior to, or concurrent with, the act of homicide; and 27.7% (n = 13) had a noted history of mental illness or psychiatric care, including substance use. The vast majority of incidents (n = 42, 89.4%) involved the use of firearm, of which about 60% were handguns. Approximately half occurred at a house or apartment (n = 24, 51.1%). For cases perpetrated by intimate partners (n = 23), 43.5% (n = 10) had a documented history of IPV, 78.3% (n = 18) involved an argument at the time of the incident, 43.5% (n = 10) involved a break up preceding the incident, and 34.8% (n = 8) involved IPV-related legal problems.

Table 1.

Characteristics of murder-suicides perpetrated by adolescents aged < 21 years: National Violent Death Reporting System, 2003–2016

	%(n)
Victim (n=56)
Sex
Female	58.9 (33)
Male	41.1 (23)
Race/ethnicity
White, non-Hispanic	50.0 (28)
Black, non-Hispanic	30.4 (17)
All other racesa	19.6 (11)
Age (years)
Mean (SD)	19.1 (3.8)
Median (IQR)	18.5 (16.0–21.5)
12–20	69.6 (39)
21–28	20.4 (17)
Relationship to perpetratorb
Intimate partner	41.1 (23)
Current/unspecified	65.2 (15)
Former	34.8 (8)
Family, non-intimate partner	26.8 (15)
Acquaintance	10.7 (6)
Stranger, law enforcement	8.9(5)
Unknown	12.5 (7)
Perpetrator (n=47)
Sex
Female	6.4 (3)
Male	93.6 (44)
Race/ethnicity
White, non-Hispanic	36.2 (17)
Black, non-Hispanic	42.6 (20)
All other racesa	21.3 (10)
Age (years)
Mean (SD)	18.4 (1.5)
Median (IQR)	19.0 (18.0–20.0)
15–17	23.4 (11)
18–20	76.6 (36)
Suicidality prior to or concurrent with homicide	17.0 (8)
History of mental illness, psychiatric care	27.7 (13)
Incident (n=47)
Method
Firearm	89.4 (42)
Other (e.g., sharp or blunt instrument)	10.6 (5)
Firearm type
Handgun	59.5 (25)
Shotgun, rifle	23.8 (10)
Unknown	16.7 (7)
Location
House, apartment	51.1 (24)
Street/road, sidewalk, alley	14.9 (7)
Motor vehicle	10.6 (5)
Other (e.g., high school, natural area)	23.4 (11)
‘Perversion of Virtue’ Classification (n=47)
Category
Self-control	23.4 (11)
Justice	17.0 (8)
Mercy	2.1 (1)
Unable to Categorize	57.4 (27)
Relationship Characteristics (n=23)
Documented history of intimate partner violence	43.5 (10)
Argument at the time of incident	78.3 (18)
Break-up preceding incident	43.5 (10)
Intimate partner violence-related legal problems	34.8 (8)

Note: n=47 murder-suicide events/perpetrators, n=56 victims (victims do not include the suicide decedents)

^aAll other races includes Hispanic; American Indian/Alaska Native, non-Hispanic; 2+ races, non-Hispanic; Asian/PI, non-Hispanic; Other, non-Hispanic

^bCategories for relationships: ‘Family, non-intimate partner’ includes sibling and other family member; ‘Acquaintance’ includes acquaintance, friend, other person known to victim, schoolmate

More than half of cases (n = 27, 57.4%) could not be classified using the categories developed based on Joiner’s (2014) typology due to a lack of detail available in the NVDRS narratives. Among the cases with sufficient detail to categorize (n = 20), the majority represented a perversion of self-control (n = 11), whereby suicidal intent was not primary, and emerged subsequent to an act of homicide. Nine cases of genuine murder-suicide were able to be classified, including eight cases that were coded as perversions of justice, and one case that was coded as a perversion of mercy. Example narratives are provided in Table 2.

Table 2.

Example narratives for each category of murder-suicide

Category	Definition	Example Narratives
Genuine Murder-Suicide
Perversion of Justice	Murder is perpetrated to achieve justice from the perspective of the perpetrator. This means that killing certain person(s) represents rational punishment for past wrongs, and/or interrupts/prevents future injustice via continued violation of the perpetrators sense of justice.	Example 1: 16-year-old female victim walking on street when former intimate partner came out yelling at her and then shot her and then himself. Multiple threats had been made in previous days via text message that the perpetrator would kill the victim and then himself because she broke up with him. There was a protective order in place to ensure no contact between perpetrator and victim.
		Example 2: Male perpetrator shot 15-year-old female friend because she rejected his romantic advances, and then shot himself in the head. Perpetrator reported to have been told he was going to die prior to this incident
Perversion of Mercy	Murder is perpetrated to prevent the perceived suffering/misery of another, prior to the perpetrator enacting their own suicide.	Not provided (n = 1 in sample)
	Example (Joiner, 2014): A perpetrator killing a dependent child or other dependent person who would be potentially without a caregiver/provider after the perpetrator died by suicide.
Perversion of Heroic Glory	Murder is perpetrated prior to suicide as a means of being remembered or recognized for accomplishing something “great” or “important” prior to death by suicide.	None identified in sample
	Example (Joiner, 2014): School shootings, such as Columbine, are the example that Joiner uses often for this. For example, the perpetrators of the Columbine school shooting engaged in murder after deciding upon suicide with the intent of surpassing the violence perpetrated by Timothy McVeigh.
Perversion of Duty	Murder is perpetrated due to a perception that doing so honors a perceived responsibility or obligation to others. Can involve duties rooted in loyalty, dignity, or decorum.	None identified in sample
	Example (Joiner, 2014): A caregiver of someone with pervasive disability murders that person prior to suicide to ultimately discharge their duty to protect or care for this person.
Neighboring but Distinct Categories
Perversion of Self-Control	Perpetrator dies by suicide after committing murder and then ending up in a situation that forces them to choose between death and the loss of control. In such cases, suicide is not intended or considered prior to murder, and emerges only after the consequences of a murderous act are realized/manifest.	Example 3: 24-year-old female shot by 18-year-old boyfriend after an argument driving down the road with their kids. Perpetrator turned violent, produced a firearm, and killed the victim. Then he stopped the car and shot himself in the head.
		Example 4: 28-year-old male police officer shot in the street by 17 year old male perpetrator. Victim responded to an auto theft by the perpetrator; as more officers arrived on scene, the perpetrator then shot himself and had a noted history of mental illness.
Perversion of Fate	Victim is killed during the enactment of another’s suicidal behavior, but in the absence of any homicidal intent on the part of the person who died by suicide.	None identified in sample
	Example (Joiner, 2014): A person who jumped to their death and fell on another person who died as a result of their injuries.

Discussion

Using Joiner’s (2014) perversion of virtue framework with a sample of murder-suicides perpetrated by adolescents, we found that over half of cases did not contain sufficient narrative detail to be categorized. For cases able to be categorized, perversions of self-control and justice were most prevalent. Interestingly, perversion of self-control was most common in this sample of murder-suicides, although Joiner (2014) classifies it as a “neighboring but distinct” rather than “genuine” murder-suicide, where suicidal intent is the primary antecedent to the murder-suicide. We categorized only one case as perversion of mercy and identified no cases of perversion of glory or duty. Taken together, these findings may challenge the validity of these theoretical subtypes of murder-suicide. However, given the small sample (that notably did not include a school shooting) and focus on adolescent perpetrators, this question remains open to debate. Perversion of glory or duty may be identified in a larger sample of murder-suicides or in a sample not restricted by perpetrator age. There is evidence that younger perpetrators of murder-suicides exhibit primarily homicidal intent while older perpetrators exhibit primarily suicidal motives (Salari & Sillito, 2016), perhaps suggesting that “genuine” murder-suicide as conceptualized by Joiner is likely to be identified among an older sample. It is also possible that narrative information in NVDRS does not capture the amount of detail required to validate Joiner’s framework. Additional research with a larger sample of cases of murder-suicide is needed to adequately assess the utility of NVDRS in validating Joiner’s framework. To this end, an expanded analysis of murder-suicides perpetrated by individuals of all ages is currently underway by our research team.

Consistent with previous studies of murder-suicide, our findings indicated a clear gendered pattern of murder-suicide perpetration in adolescents, such that perpetrators were most often male and demonstrated motives and actions that unfolded in the context of intimate partner violence and relationship-related stressors, with the majority of victims being female. A prior study of newspaper articles describing 45 murder-suicides in North America found 3 themes: domestic desperation (e.g., failing to provide economic security), workplace justice (e.g., disgruntled worker seeking justice for being mistreated or wronged), and school retaliation (e.g., pay back for perceived insults and bullying; Oliffe et al., 2015). This study highlighted men’s use of violence when they perceived a loss of control or hopelessness in their lives or perceived threats to their masculine identities (Oliffe et al., 2015). The themes identified from Oliffe and colleagues (2015) parallel the perversion of the virtue of justice, as described by Joiner (2014), which was the most common type of genuine murder-suicide that was identified in this study.

The vast majority of perpetrators in this sample were men. This is in line with research on adolescent homicide showing that perpetrators are primarily men (Coyne-Beasley, Schoenbach, & Herman-Giddens, 1999; Shields, Hunsaker, Matnani, & Davis, 2014). In the case of murder-suicide, it may be that men are at greater risk of developing primary suicidality in the wake of relationship problems, and thus are more likely to engage in the sequence of behavior that characterizes genuine murder-suicide. While this cannot be tested with these data, the findings suggest that men are more inclined to engage in murder-suicide compared to women. Previous research has noted that men may be at greater risk of suicide than women in the aftermath of relationship breakdown due to role inflexibility, potential loss of child custody, thwarted desire for control, and a lack of meaningful social support (Scourfield & Evans, 2015). However, adolescents – particularly those under 18 – may not be as likely to suffer from the same types of psychosocial stressors and consequences as adult men (e.g., loss of housing, financial liabilities), and thus it is important to note how the perversion of self-control was relatively more common in this sample than the perversion of justice. This may suggest that the perpetration of murder-suicide within the developmental context of adolescence is gendered for reasons that reflect broader social discourses, rather than specific consequences within the lives of individual adolescents. Norms around conventional masculinity may still play a critical role for these male perpetrators of murder-suicide, who may be reacting to rejection (e.g., from an intimate partner) or perceived failure by using power, dominance, and violence to assert one’s masculinity (Oliffe et al., 2015). In addition, other contextual factors, such as the availability of weapons, appear to be potent influences upon the nature of murder-suicide perpetrated by young people.

Firearms were used in about 90% of the murder-suicide incidents perpetrated by adolescents in this study. This striking number suggests the potential utility of means safety interventions, which reduce access to and increase storage of lethal methods for suicide. Means safety interventions have been shown to be effective for preventing suicide in general (Jin, Khazem, & Anestis, 2016; Yip et al., 2012) and are also likely to be relevant for preventing murder-suicide. As noted by Joiner (2014), “suicide prevention is murder-suicide prevention,” and thus engaging in efforts to safely store, restrict and remove access to firearms for at-risk youth should be considered as a way to prevent genuine murder-suicide, which is theorized to be the result of primary suicidality. The related phenomenon of the perversion of self-control is also likely to be prevented, and its effects mitigated, via robust means safety efforts. However, this related phenomenon is distinct from murder-suicide in that it represents an act of suicidal behavior that emerges from the consequences of having perpetrated a murder (e.g., imminent arrest and detention), and therefore suicide prevention per se is not a directly applicable approach for reducing the fundamental drivers of this outcome.

Given the overlap of the perversion of justice with a history of IPV, it is important to consider how interventions that prevent and mitigate teen dating violence could also help prevent the relatively rapid sequence of murder followed by a decision to enact suicide as a means of escaping the consequences of IPV and other forms of interpersonal violence. A documented history of IPV was noted in over 40% of the murder-suicides committed by intimate partners, which is likely an underestimate given evidence of underreporting of IPV (Ashley & Foshee, 2005; Sylaska & Edwards, 2014). Furthermore, nationally-representative studies show that the prevalence of IPV among adolescents is substantial and can lead to deadly outcomes (Adhia, Kernic, Hemenway, Vavilala, & Rivara, 2019; Taylor & Mumford, 2016). Effective interventions for the primary prevention of IPV include education on healthy relationships and communication skills, and bystander interventions (De Koker, Mathews, Zuch, Bastien, & Mason-Jones, 2014; Lundgren & Amin, 2015).

Limitations

Although NVDRS RAD is one of the richest and most extensive databases of murder-suicide incidents to date, this study demonstrates the limitations of the available information for the purposes of classifying incidents based on motivation and intent. Indeed, more than half of the incidents identified for this study did not include sufficient information to determine which type of perversion of virtue was antecedent to murder-suicide, if any. The amount of detail contained in the narratives varied greatly between incidents. The indicators coded (e.g., history of IPV; suicidal intent prior to, or concurrent with, the act of homicide) were limited by the accuracy and depth of detail recorded in the medical examiner and law enforcement reports, so these results likely underestimate the true prevalence of these factors. As previously mentioned, research with a larger sample of NVDRS cases of murder-suicide is needed to evaluate the utility of this data source for validating Joiner’s (2014) framework. Relatedly, the small number of cases that occurred also precluded inferential analyses with regard to the associations among perpetrator, victim, and incident characteristics. Data for this study come only from 32 states contributing varying years of data from 2003 to 2016, so our findings may have limited generalizability. The included states are not randomly selected but appear to be fairly representative of the U.S. in terms of characteristics such as geographic location and levels of gun ownership. The NVDRS is funded now to collect data in all 50 states, so future studies will benefit from a larger sample size.

Conclusion

Our findings demonstrate the gendered nature of murder-suicide perpetration and victimization among adolescents, and also suggest the relative frequency of the perversion of justice and the perversion of self-control as motivations underlying the perpetration of this behavior by adolescents. Further, murder-suicide perpetrated by adolescents appeared to occur frequently within the context of IPV and other relationship stressors.