Left out in the cold: Homicide amongst persons experiencing homelessness

BACKGROUND

There is no American population-level study comparing the characteristics of homicides involving victims who were vs were not experiencing homelessness at time of death. We aim to identify variables surrounding homeless homicide that are unique, and intervenable.

METHODS

In this retrospective cohort study, we reviewed the National Violent Death Reporting System (NVDRS) from 2003 to 2018 and compared the characteristics surrounding homicides of victims who were not-homeless (cohort 1) vs experiencing homelessness (cohort 2) at death. We utilized the available perpetrator data to characterize the average perpetrator for each cohort. We considered housing status to be our primary predictor and recorded NVDRS variables, such as age of victims and likelihood to know perpetrators, to be our primary outcomes.

RESULTS

81,212 Homicide Victims and 60,982 Homicide Perpetrators were included in analysis. Homeless cohort victims were more likely younger, White, male, and to have a known mental health or substance abuse disorder. PEH were also more likely to have co-morbid mental health and substance abuse disorders but were roughly half as likely to be getting treatment for said disorders. Circumstances surrounding incidents, including geographic location, mechanism of injury, and premeditation, varied; homeless cohort victims were more likely to die in random acts of violence, but were less likely to die via firearm.

CONCLUSIONS

There is room for targeted interventions against homeless homicides.

LEVEL OF EVIDENCE

Epidemiological, Level IV.

Research utility and limitations of textual data in the National Violent Death Reporting System: a scoping review and recommendations

BACKGROUND

Many studies of injury deaths rely on mortality data that contain limited contextual information about decedents. The National Violent Death Reporting System (NVDRS) is unique among such data systems in that each observation includes both quantitative variables and qualitative texts (called "narratives") abstracted from original source documents. These narratives provide rich data regarding salient circumstances that can be used to inform prevention efforts. This review provides a comprehensive summary of peer-reviewed research using NVDRS narratives over the past 20 years, including the limitations of these texts and provides recommendations on utilizing and improving narrative quality for researchers and practitioners.

MAIN BODY

Studies that used narratives to examine deaths related to suicide, homicide, undetermined intent, accidental firearm, or legal intervention were identified by a title/abstract screening, followed by a full-text review. The search was conducted on English-language, peer-reviewed literature and government reports published from 2002 to 2022 in PubMed, PsycInfo, Scopus, and Google Scholar. Abstracted elements focused on the methodologies used to analyze the narratives, including approaches to explore potential biases in these texts. Articles were abstracted independently by two reviewers, with disagreements resolved through consensus discussion. During the 20-year period, 111 articles used narratives. Two-thirds studied suicide (n = 48, 43%) and homicides (n = 25, 23%). Most studies analyzed the narratives using manual review (n = 81, 73%) and keyword searches (n = 9, 8%), with only 6 (5%) using machine learning tools. Narratives were mainly used for case finding (n = 49, 44%) and characterization of circumstances around deaths (n = 38, 34%). Common challenges included variability in the narratives and lack of relevant circumstantial details for case characterization.

CONCLUSION

Although the use of narratives has increased over time, these efforts would be enhanced by detailed abstraction of circumstances with greater salience to injury research and prevention. Moreover, researchers and practitioners would benefit from guidance on integrating narratives with quantitative variables and standardized approaches to address variability in the completeness and length of narratives. Such efforts will increase the reliability of findings and set the stage for more widespread applications of data science methods to these texts.

Who owned the gun in firearm suicides of men, women, and youth in five US states?

One way to reduce firearm suicide is to keep household guns away from a person at risk for suicide. To learn who owned (and presumably controlled access to) the guns used in suicide and which broad gun type they were, we examined National Violent Death Reporting System (NVDRS) data from 2015 to 2017 for five US states that supplied information on gun owner in over 80% of firearm suicides (AK, IA, NH, UT, WI). For adult males, 88% used their own gun; for women, 52% used their own gun and 32% used their partner's gun; for youth ages 18-20, 42% used their own gun, 43% used a family member's, and 8% used a friend's; for children, 19% used their own gun (usually a long gun) and 79% used a family member's gun. Almost 3/4 of firearm suicides involved a handgun, ranging from 62% for youth to 92% for women. In times of suicide risk, interventions for a youth should address not only the parents' guns, but those of other family members and the youth's own rifle or shotgun. For a woman, interventions need to address her own and her partner's guns. For a man, locking guns alone will confer little protection if he controls the keys or combination. Storing firearms-or a critical component-away from home or having someone else control the locks may be safer. Five NVDRS states provided useful data on who owned the gun used in firearm suicides. More NVDRS states should follow suit.

Using natural language processing to improve suicide classification requires consideration of race

OBJECTIVES

To improve the accuracy of classification of deaths of undetermined intent and to examine racial differences in misclassification.

METHODS

We used natural language processing and statistical text analysis on restricted-access case narratives of suicides, homicides, and undetermined deaths in 37 states collected from the National Violent Death Reporting System (NVDRS) (2017). We fit separate race-specific classification models to predict suicide among undetermined cases using data from known homicide cases (true negatives) and known suicide cases (true positives).

RESULTS

A classifier trained on an all-race dataset predicts less than half of these cases as suicide. Importantly, our analysis yields an estimated suicide rate for the Black population comparable with the typical detection rate for the White population, indicating that misclassification excess is endemic for Black suicide. This problem may be mitigated by using race-specific data. Our findings, based on the statistical text analysis, also reveal systematic differences in the phrases identified as most predictive of suicide.

CONCLUSIONS

This study highlights the need to understand the reasons underlying suicide rate differences and for further testing of strategies to reduce misclassification, particularly among people of color.

Affordable housing through the low-income housing tax credit program and intimate partner violence-related homicide

The most severe outcome of intimate partner violence (IPV) is IPV-related homicide. Access to affordable housing may both facilitate exit from abusive relationships and reduce financial stress in intimate relationships, potentially preventing IPV-related homicide. We examined the association of the availability of rental housing through the Low-Income Housing Tax Credit (LIHTC) program, a federal program providing tax incentives to support the development of affordable housing, with IPV-related homicide and assessed whether this association differed by eviction rates at the state-level. We used 2005-2016 National Violent Death Reporting System, LIHTC Property, and Eviction Lab data for 13 states and compared the rate of IPV-related homicide in state-years with ≥30 to state-years with <30 LIHTC units per 100,000 population, overall and stratified by eviction rates. We conducted analyses in fall 2020. Adjusting for potential state-level confounders, the rate of IPV-related homicide in state-years with ≥30 LIHTC units per 100,000 population was lower than in state-years with <30 LIHTC units per 100,000 population (RR = 0.89, 95% CI 0.81, 0.98). The reduction in the rate of IPV-related homicide was slightly larger in state-years with higher eviction rates (≥3500 evictions per 100,000 renter population; RR = 0.83, 95% CI 0.74, 0.93) compared to state-years with lower eviction rates (<3500 evictions per 100,000 renter population; RR = 0.91, 95% CI 0.81, 1.03). Overall, at the state-level, increased availability of affordable housing through the LIHTC program was associated with lower rates of IPV-related homicide. Increasing the availability of affordable housing may be one tool for preventing IPV-related homicide.

A Typology of Civilians Shot and Killed by US Police: a Latent Class Analysis of Firearm Legal Intervention Homicide in the 2014-2015 National Violent Death Reporting System

Approximately 1000 people are killed by police acting in the line of duty each year. Historically, research on these deaths, known as legal intervention homicides (LIH), has been limited by data that is either contextually rich but narrow in scope and not readily available to the public (e.g., police department reports from a single city), or detail-poor but geographically broad, large, and readily available (and maintained by federal agencies) (e.g., vital statistics and supplemental homicide reports). Over the past 5 years, however, researchers have turned to the National Violent Death Reporting System (NVDRS), which captures nearly all lethal police shootings in participating states while providing detailed incident and victim information. The current study extends prior work on police-involved lethal shootings in three important ways. First, we use latent class analysis to construct a data-driven, exhaustive, mutually exclusive typology of these events, using NVDRS data 2014-2015. Second, rather than fitting some, but not all cases into predefined sub-types, every case is assigned membership to a particular emergent class. Third, we use a validated case identification process in NVDRS to identify incidents of lethal police-involved shootings. Seven classes emerge. Classes differ across important incident and victim characteristics such as the event that brought the victim and law enforcement together, the highest level of force used by the victim against law enforcement, and the kind of weapon, if any, used by the victim during the incident. Demographic variables do not distribute uniformly across classes (e.g., the latent class in which the victim appeared to pose minimal threat to law enforcement was the only class in which the plurality of victims was a non-white race). Our approach to generating these typologies illustrates how data-driven techniques can complement subjective classification schemes and lay the groundwork for analogous analyses using police encounter data that include fatal and non-fatal outcomes.

Suicide, Alcohol Intoxication, and Age Among Whites and American Indians/Alaskan Natives

BACKGROUND

Among American Indians/Alaskan Natives (AI/ANs), suicides are disproportionately high among those younger than 40 years of age. This paper examines suicide and alcohol intoxication (postmortem BAC ≥ 0.08 g/dl) by age among Whites and AI/ANs to better understand the reasons for the high rate of suicide among AI/ANs for those younger than 40.

METHODS

Data come from the restricted 2003 to 2016 National Violent Death Reporting System (NVDRS), with postmortem information on 79,150 White and AI/AN suicide decedents of both genders who had a BAC test in 32 states of the United States.

RESULTS

Among Whites, 39.3% of decedents legally intoxicated are younger than 40 years of age, while among AI/ANs the proportion is 72.9% (p < 0.001). Multivariable logistic regression with data divided by age shows that in the 18 to 39 age group, AI/ANs are about 2 times more likely than Whites to have a postmortem BAC ≥ 0.08. Veteran status compared to nonveteran, and history of alcohol problems prior to suicide were also associated with BAC ≥ 0.08. Suicide methods other than by firearm and a report of the presence of 2 or more suicide precipitating circumstances were protective against BAC ≥ 0.08. Results for the age group 40 years of age and older mirror those for the younger group with 1 exception: Race/ethnicity was not associated with BAC level.

CONCLUSIONS

The proportion of suicide decedents with a BAC ≥ 0.08 is higher among AI/ANs than Whites, especially among those 18 to 39 years of age. However, acute alcohol intoxication does not fully explain differences in suicide age structure between AI/ANs and Whites.

Veteran Suicide Risk Factors: A National Sample of Nonveteran and Veteran Men Who Died by Suicide

Veteran suicide is a serious public health problem. Some data suggest that veteran suicide risk profiles differ from those of nonveterans. Records for veteran (n = 21,692) and nonveteran (n = 83,430) men who died by suicide were examined from 17 U.S. states using the National Violent Death Reporting System data. Seventeen precipitating factors were examined and combined through meta-analysis of proportions. Many precipitating factors were found to be less frequent for veterans. A smaller number of factors were found to be higher in the veteran population, including physical health problems. A sizable cumulative effect size (1.02) was observed, suggesting that veteran and nonveteran men show meaningful and substantive differences in their risk profiles-differences that should be considered when planning and implementing suicide prevention and intervention efforts. The conspicuous role of physical health problems among veterans who die by suicide is discussed. The article concludes with specific practice recommendations for social workers.

Research utility of the National Violent Death Reporting System: a scoping review

Background

To better understand and prevent suicide and homicide, the National Center for Injury Prevention and Control of the US Centers for Disease Control and Prevention launched the National Violent Death Reporting System (NVDRS) in six states in 2002. As of 2018, the NVDRS has been expanded to include all 50 states, the District of Columbia and Puerto Rico. The purpose of this review was to assess the research utility of the NVDRS based on studies indexed in major bibliographical databases.

Methods

We performed a scoping review of published studies that were based on data from the NVDRS, identified by searching six electronic databases: PubMed, EMBASE, Google Scholar, OVID, Scopus, and Web of Science. We examined the time trend of annual NVDRS-based research output, generated a word cloud using the keywords listed in the publications, and mapped the knowledge domains covered by NVDRS-based studies.

Results

Our review included a total of 150 studies published between 2005 and 2018. There was a marked increase in the annual number of NVDRS-based publications, with 120 (80.0%) of the 150 studies published between 2011 and 2018. Overall, 104 (69.3%) studies focused on suicide and 39 (26.0%) on homicide. Of the included studies, 100 (66.7%) were descriptive epidemiology, 31 (20.7%) were risk factor analyses, 9 (6.0%) were evaluations, 7 (4.7%) were trend analyses, and 4 (2.7%) were data quality assessments. Knowledge domain mapping identified two major clusters of studies, one on suicide and the other on homicide. The cluster on suicide was commonly linked to “circumstance,” “alcohol” and “substance abuse” and the cluster on homicide was commonly linked to “firearm,” “injury,” and “gang.” The two clusters were interlinked to overlapping networks of keywords, such as “firearm” and “mental health problem.”

Conclusions

Research utility of the NVDRS has increased considerably in recent years. Studies based on data from the NVDRS are clustered in two knowledge domains – suicide and homicide. The vast potential of the NVDRS for violence research and prevention remains to be fully exploited.

Methods and Findings from the National Violent Death Reporting System for Identifying Gang-Like Homicides, 2005-2008

BACKGROUND

The National Violent Death Reporting System (NVDRS) captures homicides that law enforcement or coroner/medical examiners deem as gang-related but the criteria used may vary across locations. Also, the existing gang-related variable likely underestimates the number of homicides that are associated with gang activity. This study utilizes NVDRS data to identify "gang-like" homicides which are not currently captured as "gang-related."

METHODS

A set of criteria recommended by a panel of experts in gang violence, was applied to homicides collected in the NVDRS. These criteria, termed "gang-like" characteristics, were developed in order to better identify homicides consistent with gang activity. The narratives of the identified cases were then reviewed to refine the operational standard. After the reviews were complete, the typology was modified to finalize the operationalization of "gang-like" homicides.

RESULTS

A total of 481 gang-like homicides were identified using the "gang-like" criteria. This represents an increase of almost 69% over the 696 gang-related homicides captured in NVDRS dataset. Gang-like and gang-related homicides combined represented 6.6% of homicides that occurred from 2005 to 2008. Among the 16 states included in this analysis, Colorado (15.5%) and Oklahoma (14%) had the highest percentage of homicides that were either gang-related or gang-like. Maryland had the greatest relative increase (227.3%) between gang related and gang-like homicides.

CONCLUSION

The new "gang-like" variable complements the existing "gang-related" variable by providing an automated, standardized way to identify homicides that have circumstances consistent with gang activity. This new variable might be useful to states and localities seeking an efficient way to monitor homicides potentially resulting from gang activity. Additional efforts are needed to standardize the reporting of homicides associated with gang activity.

Precipitating Circumstances of Suicide and Alcohol Intoxication Among U.S. Ethnic Groups

BACKGROUND

Our goal was to assess the prevalence of 9 different types of precipitating circumstances among suicide decedents, and examine the association between circumstances and postmortem blood alcohol concentration (BAC ≥ 0.08 g/dl) across U.S. ethnic groups.

METHODS

Data come from the restricted 2003 to 2011 National Violent Death Reporting System, with postmortem information on 59,384 male and female suicide decedents for 17 U.S. states.

RESULTS

Among men, precipitating circumstances statistically associated with a BAC ≥ 0.08 g/dl were physical health and job problems for Blacks, and experiencing a crisis, physical health problems, and intimate partner problem for Hispanics. Among women, the only precipitating circumstance associated with a BAC ≥ 0.08 g/dl was substance abuse problems other than alcohol for Blacks. The number of precipitating circumstances present before the suicide was negatively associated with a BAC ≥ 0.08 g/dl for Whites, Blacks, and Hispanics.

CONCLUSIONS

Selected precipitating circumstances were associated with a BAC ≥ 0.08 g/dl, and the strongest determinant of this level of alcohol intoxication prior to suicide among all ethnic groups was the presence of an alcohol problem.

A Comparison of Suicides and Undetermined Deaths by Poisoning among Women: An Analysis of the National Violent Death Reporting System

The study compared the prevalence of common suicide risk factors between poisoning deaths classified as injuries of undetermined intent or suicides among women. Data were derived from the 2003-10 National Violent Death Reporting System. Multiple logistic regression assessed the factors associated with 799 undetermined deaths (relative to 3,233 suicides). Female decedents with lower education, a substance use problem, and a health problem were more likely to be classified as undetermined death. Older women, those with an intimate partner problem, financial problem, depressed mood, mental health problem, attempted suicide, and disclosed intent to die were less likely to be classified as undetermined death. The present study raises the possibility that many (perhaps most) undetermined female poisoning deaths are suicides.

Sex differences in suicide incident characteristics and circumstances among older adults: surveillance data from the National Violent Death Reporting System--17 U.S. states, 2007-2009

Each year in the U.S. more than 7,000 adults aged 60 years and older die of suicide and as the population ages, these numbers are expected to increase. While sex is an important predictor of older adult suicide, differences between males and females are often overlooked due to low occurrence, particularly among women. The National Violent Death Reporting System (NVDRS) bridges this gap by providing detailed information on older adult suicide by sex in 17 US states (covering approximately 26% of the U.S. population). NVDRS data for 2007-2009 were used to characterize male (n = 5,004) and female (n = 1,123) suicide decedents aged 60 years and older, including incident characteristics and circumstances precipitating suicide. Stratification of NVDRS data by sex shows significant differences with regard to the presence of antidepressants (19% and 45% respectively), opiates (18%, 37%), and 14 precipitating circumstances concerning mental health, interpersonal problems, life stressors and a history of suicide attempts. No differences were found for alcohol problems, suicide/other death of family or friends, non-criminal legal problems, financial problems, or disclosure of intent to take their own life. The findings of this study demonstrate the value of using comprehensive surveillance data to understand sex-specific suicide circumstances so that opportunities for targeted prevention strategies may be considered.