A Geospatial Analysis of Severe Firearm Injuries Compared to Other Injury Mechanisms: Event Characteristics, Location, Timing, and Outcomes

Factors Associated with Discharge to Skilled Nursing Facility Following Gunshot Wounds

INTRODUCTION

Firearm injury is a public health crisis. Most victims are minorities in underserved neighborhoods. Measuring firearm injury by mortality underestimates its impact, as most victims survive to discharge. This study was done to determine if race and insurance status are associated with discharge disposition for gunshot wound (GSW)-related trauma.

METHODS

Using the 2019 Trauma Quality Improvement Program database, we identified GSW patients with Abbreviated Injury Scale (AIS) = 1-3. Exclusion criteria included patients who died in hospital and routine home discharge. We compared discharge patterns of patients based on demographics (age, gender, race, ethnicity, payor, AIS, hospital designation, and length of stay [LOS]) and injury severity. Multivariable logistic regression models identified factors associated with discharge disposition.

RESULTS

Our sample included 2437 patients with GSWs. On univariable analysis, Black patients were more likely to discharge to home with home health (64.1% Black versus 34.7% White; P < 0.001). White patients were more likely to discharge to skilled nursing facility (SNF) (51.4% White versus 44.6% Black; P < 0.001). Controlling for age, race, Latin ethnicity, primary payor, LOS, AIS severity, and injury severity score factors independently associated with discharge to SNF included age (0.0462, P < 0.001), Medicaid (1.136, P < 0.0003), Medicare (1.452, P < 0.001), and LOS (0.03745, P < 0.001).

CONCLUSIONS

Postacute care following traumatic injuries is essential to recovery. Black GSW victims are more likely to be discharged to home health than White patients, who are more likely to be discharged to SNF. Targeted programs to reduce barriers to appropriate aftercare are necessary to eliminate this bias and improve the care of underserved populations.

Social Vulnerability and Firearm Violence: Geospatial Analysis of 5 US Cities

BACKGROUND

Firearm violence is now endemic to certain US neighborhoods. Understanding factors that impact a neighborhood's susceptibility to firearm violence is crucial for prevention. Using a nationally standardized measure to characterize community-level firearm violence risk has not been broadly studied but could enhance prevention efforts. Thus, we sought to examine the association between firearm violence and the social, structural, and geospatial determinants of health, as defined by the Social Vulnerability Index (SVI).

STUDY DESIGN

In this cross-sectional study, we merged 2018 SVI data on census tract with shooting incidents between 2015 and 2021 from Baltimore, Chicago, Los Angeles, New York City, and Philadelphia. We used negative binomial regression to associate the SVI with shooting incidents per 1,000 people in a census tract. Moran's I statistics and spatial lag models were used for geospatial analysis.

RESULTS

We evaluated 71,296 shooting incidents across 4,415 census tracts. Fifty-five percent of shootings occurred in 9.4% of census tracts. In all cities combined, a decile rise in SVI resulted in a 37% increase in shooting incidents (p < 0.001). A similar relationship existed in each city: 30% increase in Baltimore (p < 0.001), 50% in Chicago (p < 0.001), 28% in Los Angeles (p < 0.001), 34% in New York City (p < 0.001), and 41% in Philadelphia (p < 0.001). Shootings were highly clustered within the most vulnerable neighborhoods.

CONCLUSIONS

In 5 major US cities, firearm violence was concentrated in neighborhoods with high social vulnerability. A tool such as the SVI could be used to inform prevention efforts by directing resources to communities most in need and identifying factors on which to focus these programs and policies.

The Geography of Injuries in Trauma Systems: Using Home as a Proxy for Incident Location

INTRODUCTION

Effective trauma system organization is crucial to timely access to care and requires accurate understanding of injury and resource locations. Many systems rely on home zip codes to evaluate geographic distribution of injury; however, few studies have evaluated the reliability of home as a proxy for incident location after injury.

METHODS

We analyzed data from a multicenter prospective cohort collected from 2017 to 2021. Injured patients with both home and incident zip codes were included. Outcomes included discordance and differential distance between home and incident zip code. Associations of discordance with patient characteristics were determined by logistic regression. We also assessed trauma center catchment areas based on home versus incident zip codes and variation regionally at each center.

RESULTS

Fifty thousand one hundred seventy-five patients were included in the analysis. Home and incident zip codes were discordant in 21,635 patients (43.1%). Injuries related to motor vehicles (aOR: 4.76 [95% CI 4.50-5.04]) and younger adults 16-64 (aOR: 2.46 [95% CI 2.28-2.65]) were most likely to be discordant. Additionally, as injury severity score increased, discordance increased. Trauma center catchment area differed up to two-thirds of zip codes when using home versus incident location. Discordance rate, discordant distance, and catchment area overlap between home and incident zip codes all varied significantly by geographic region.

CONCLUSIONS

Home location as proxy for injury location should be used with caution and may impact trauma system planning and policy, especially in certain populations. More accurate geolocation data are warranted to further optimize trauma system design.

Characteristics and neighborhood-level opportunity of assault-injured children in Milwaukee

BACKGROUND

Multiple studies have explored demographic characteristics and social determinants of health in relation to the risk of pediatric assault-related injuries and reinjury. However, few have explored protective factors. The Child Opportunity Index (COI) uses neighborhood-level indicators to measure 'opportunity' based on factors such as education, social environment, and economic resources. We hypothesized that higher 'opportunity' would be associated with less risk of reinjury in assault-injured youth.

METHODS

This was a single-institution, retrospective study at a Level 1 Pediatric Trauma Center. Trauma registry and electronic medical record data were queried for children ≤ 18 years old with assault-related injuries from 1/1/2016 to 5/31/2021. Reinjured children, defined as any child who sustained more than one assault injury, were compared to non-reinjured children. Area Deprivation Index (ADI), a marker of socioeconomic status, and COI were determined through census block and tract data, respectively. A post-hoc analysis examined COI between all assault-injured children, unintentionally injured children, and a state-based normative cohort representative of non-injured children.

RESULTS

There were 55,862 traumatic injury encounters during the study period. Of those, 1224 (2.3%) assault injured children were identified, with 52 (4.2%) reinjured children and 1172 (95.8%) non-reinjured children. Reinjured children were significantly more likely to be older (median age 15.0 [IQR 13.8-17.0] vs. median age 14.0 [IQR 8.8-16.0], p < 0.001) and female (55.8% vs. 37.5%, p = 0.01) than non-reinjured children. COI was not associated with reinjury. There were also no significant differences in race, ethnicity, insurance status, ADI, or mechanism and severity of injury between cohorts. Post-hoc analysis revealed that assault-injured children were more likely to live in areas of lower COI than the other cohorts.

CONCLUSIONS

Compared to children who sustained only one assault during the study period, children who experienced more than one assault were more likely to be older and female. Furthermore, living in an area with more or less opportunity did not influence the risk of reinjury. However, all assault-injured children were more likely to live in areas of lower COI compared to unintentionally injured and a state-based normative cohort. Identification of factors on a social or environmental level that leads to assaultive injury warrants further exploration.

Small area analysis methods in an area of limited mapping: exploratory geospatial analysis of firearm injuries in Port-au-Prince, Haiti

BACKGROUND

The city of Port-au-Prince, Haiti, is experiencing an epidemic of firearm injuries which has resulted in high burdens of morbidity and mortality. Despite this, little scientific literature exists on the topic. Geospatial research could inform stakeholders and aid in the response to the current firearm injury epidemic. However, traditional small-area geospatial methods are difficult to implement in Port-au-Prince, as the area has limited mapping penetration. Objectives of this study were to evaluate the feasibility of geospatial analysis in Port-au-Prince, to seek to understand specific limitations to geospatial research in this context, and to explore the geospatial epidemiology of firearm injuries in patients presenting to the largest public hospital in Port-au-Prince.

RESULTS

To overcome limited mapping penetration, multiple data sources were combined. Boundaries of informally developed neighborhoods were estimated from the crowd-sourced platform OpenStreetMap using Thiessen polygons. Population counts were obtained from previously published satellite-derived estimates and aggregated to the neighborhood level. Cases of firearm injuries presenting to the largest public hospital in Port-au-Prince from November 22nd, 2019, through December 31st, 2020, were geocoded and aggregated to the neighborhood level. Cluster analysis was performed using Global Moran's I testing, local Moran's I testing, and the SaTScan software. Results demonstrated significant geospatial autocorrelation in the risk of firearm injury within the city. Cluster analysis identified areas of the city with the highest burden of firearm injuries.

CONCLUSIONS

By utilizing novel methodology in neighborhood estimation and combining multiple data sources, geospatial research was able to be conducted in Port-au-Prince. Geospatial clusters of firearm injuries were identified, and neighborhood level relative-risk estimates were obtained. While access to neighborhoods experiencing the largest burden of firearm injuries remains restricted, these geospatial methods could continue to inform stakeholder response to the growing burden of firearm injuries in Port-au-Prince.

Burden of firearm-related injuries as associated secondary diagnosis in the United States from National Inpatient Sample Dataset

There is little known about the differences, from the point of view of healthcare resource utilization, between non-fatal versus fatal firearm-related injuries. We undertook this research project utilizing the National Inpatient Sample (NIS) database to address this critical knowledge gap. Our aims for this study were to describe the patterns of FRI in the United States during the period of 2016-2019 and to evaluate the patient-centered outcomes in the survivor (non-fatal injuries) versus the non-survivor (fatal injuries) groups. We used the National Inpatient Sample (NIS) Database, 2016-2019 (5) (~ 20% of United States hospitalizations) to identify patients with an associated diagnosis of firearm-related injuries (FRI) [Gibson T et al (2016) in Agency for Healthcare Research and Quality 2016-02]. We found that the individuals from the lowest quartile of annual household income, males, young Americans, and racial minorities were disproportionally affected The non-survivor (fatal injuries) group had a shorter length of stay in the hospital by 5.1 days (95% CI - 5.64 to - 4.58, p value =  < 0.01), the higher median cost of hospitalization by $8903 (95% confidence interval $311.9 to $17,494.2, p value = 0.04), and a higher median cost of hospitalization per day by $41,576.74 (95% confidence interval $ 40,333.1 to $42,820.3, p value =  < 0.01). In conclusion, the individuals from the lowest quartile of annual household income, males, young Americans and racial minorities were disproportionally affected. Firearm-related injuries pose a persistent healthcare cost burden with the cumulative and per day cost of hospitalization for fatal injuries being significantly higher than the non-fatal injuries despite a shorter hospital LOS.

Comparing forms of neighborhood instability as predictors of violence in Richmond, VA

Violence events tend to cluster together geospatially. Various features of communities and their residents have been highlighted as explanations for such clustering in the literature. One reliable correlate of violence is neighborhood instability. Research on neighborhood instability indicates that such instability can be measured as property tax delinquency, yet no known work has contrasted external and internal sources of instability in predicting neighborhood violence. To this end we collected data on violence events, company and personal property tax delinquency, population density, race, income, food stamps, and alcohol outlets for each of Richmond, Virginia's 148 neighborhoods. We constructed and compared ordinary least-squares (OLS) to geographically weighted regression (GWR) models before constructing a final algorithm-selected GWR model. Our results indicated that the tax delinquency of company-owned properties (e.g., rental homes, apartments) was the only variable in our model (R2 = 0.62) that was associated with violence in all but four Richmond neighborhoods. We replicated this analysis using violence data from a later point in time which yielded largely identical results. These findings indicate that external sources of neighborhood instability may be more important to predicting violence than internal sources. Our results further provide support for social disorganization theory and point to opportunities to expand this framework.

Firearm injury epidemiology in children and youth in Ontario, Canada: a population-based study

BACKGROUND AND OBJECTIVE

Despite firearms contributing to significant morbidity and mortality globally, firearm injury epidemiology is seldom described outside of the USA. We examined firearm injuries among youth in Canada, including weapon type, and intent.

DESIGN

Population-based, pooled cross-sectional study using linked health administrative and demographic databases.

SETTING

Ontario, Canada.

PARTICIPANTS

All children and youth from birth to 24 years, residing in Ontario from 1 April 2003 to 31 March 2018.

EXPOSURE

Firearm injury intent and weapon type using the International Classification of Disease-10 CM codes with Canadian enhancements. Secondary exposures were sociodemographics including age, sex, rurality and income.

MAIN OUTCOMES

Any hospital or death record of a firearm injury with counts and rates of firearm injuries described overall and stratified by weapon type and injury intent. Multivariable Poisson regression stratified by injury intent was used to calculate rate ratios of firearm injuries by weapon type.

RESULTS

Of 5486 children and youth with a firearm injury (annual rate: 8.8/100 000 population), 90.7% survived. Most injuries occurred in males (90.1%, 15.5/100 000 population). 62.3% (3416) of injuries were unintentional (5.5/100 000 population) of which 1.9% were deaths, whereas 26.5% (1452) were assault related (2.3/100 00 population) of which 18.7% were deaths. Self-injury accounted for 3.7% (204) of cases of which 72.0% were deaths. Across all intents, adjusted regression models showed males were at an increased risk of injury. Non-powdered firearms accounted for half (48.6%, 3.9/100 000 population) of all injuries. Compared with handguns, non-powdered firearms had a higher risk of causing unintentional injuries (adjusted rate ratio (aRR) 14.75, 95% CI 12.01 to 18.12) but not assault (aRR 0.84, 95% CI 0.70 to 1.00).

CONCLUSIONS

Firearm injuries are a preventable public health problem among youth in Ontario, Canada. Unintentional injuries and those caused by non-powdered firearms were most common and assault and self-injury contributed to substantial firearm-related deaths and should be a focus of prevention efforts.

Demographics and Incident Location of Gunshot Wounds at a Single Level I Trauma Center

Introduction

Kansas has seen a steady increase in the rate of firearm deaths and injuries. Little is known surrounding the demographic and geospatial factors of these firearm-related traumas. The purpose of this study was to describe the overall incidence of firearm-related traumas, identify high injury locations, and examine any racial/ethnic disparities that may exist.

Methods

A retrospective review was conducted of all patients 14 years or older who were admitted with a gunshot wound (GSW) to a Level I trauma center between 2016 and 2017.

Results

Forty-nine percent of patients were Caucasian, 26.5% African American, and 19.6% Hispanic/Latino. Hispanic/Latino patients were the youngest (25.8 ± 8.8 years) and Caucasians were the oldest (34.3 ± 14.1 years, p = 0.002). Compared to Caucasian patients, African American (42.0%) and Hispanic/Latino (54.1%) patients were more likely to be admitted to the intensive care unit (ICU; p = 0.034) and experienced longer ICU lengths of stay (2.5 ± 6.3 and 2.4 ± 4.7 days, p = 0.031, respectively). African American patients (96.0%) experienced more assaults, while Caucasians were more likely to receive gunshot wounds accidentally (26.9%, p = 0.001). More African American (86.0%) and Hispanic/Latino (89.2%) patients were injured with a handgun and Caucasians sustained the highest percentage of shotgun/rifle related injuries (16.1%, p = 0.012). Most GSWs occurred in zip codes 67202, 67203, 67213, 67211, and 67214. Geographical maps indicated that GSWs occur in neighborhoods with low-income and high minority residents and in the downtown and nightclub areas of the city.

Conclusions

Most GSW victims were older Caucasian males. Racial differences were noted and injury locations concentrated in certain locations.

Estimating Nonfatal Gunshot Injury Locations With Natural Language Processing and Machine Learning Models

IMPORTANCE

Nonfatal gunshot injuries are the most common firearm injury, but where they frequently occur remains unclear owing to data limitations. Natural language processing can be applied to medical text narratives of gunshot injury records to classify injury location and inform prevention efforts.

OBJECTIVE

To examine the performance of natural language processing (NLP) and machine learning models to predict nonfatal gunshot injury locations and generate new national estimates of the locations in which these injuries occur.

DESIGN, SETTING, AND PARTICIPANTS

Cross-sectional study of data from the National Electronic Injury Surveillance System Firearm Injury Surveillance Study on nonfatal gunshot injuries that occurred in the US between January 1, 1993, and December 31, 2015. The unweighted sample included 59 025 gunshot injuries that were initially treated at emergency departments. Data were analyzed from June 1, 2019 to July 24, 2020.

MAIN OUTCOMES AND MEASURES

The primary outcomes were classification of injury location and subsequent estimation of nonfatal gunshot injury location. The NLP was used to generate 6 sets of predictors, and 4 machine learning models were trained to classify the missing locations: multinomial support vector machines, lasso regression, XgBoost gradient descent, and feed-forward neural networks. For each of the 6 sets of NLP predictors, 70% of records with locations were randomly sampled to form the training set and the remaining 30% of records composed the test set. The best-performing model was validated by comparing the predicted locations were with those from existing national estimates of nonfatal gunshot injuries stratified by location and intent.

RESULTS

The unweighted sample included 59 025 nonfatal gunshot injuries; patients with these injuries were predominantly male (n = 52 630, [89.2%]), of Black race/ethnicity (n = 29 304 [49.6%]), and young (15-24 years; n = 27 037 [45.8%]). In total, 54 089 nonfatal gunshot injuries that were weighted to approximate national estimates were included in the analysis. Existing national estimates suggest that the most prevalent nonfatal gunshot injury location is the home (n = 14 764 [23.4%]), followed by the street or highway (n = 14 402 [22.9%]), and other public places (n = 7276 [11.6%]). After implementation of NLP classification, the most frequent gunshot injury location was the street or highway (n = 27 200 [46.1%]), followed by the home (n = 23 738 [37.7%]), and other public places (n = 10 439 [15.1%]).

CONCLUSIONS AND RELEVANCE

The findings of this study suggest that NLP and machine learning models may be useful for classifying gunshot injury location and that most nonfatal gunshot injuries occur in the street or highway rather than in the home; these findings can inform future firearm injury prevention efforts.

Youth and weapons: Patterns, individual and neighborhood correlates of violent crime arrests in Miami-Dade County, Florida

This study uses arrest data from the Miami-Dade County Police Department (n = 13 districts) for 2014-2017 to identify census tracts in which weapon violence arrests among young people aged 10-24 is high, and models area-level predictors of the total number of violent weapons crimes by census tract (greenness and socioeconomic status indices, population density and weapon dealers), after adjusting for arrestee-level factors (age, race and gender). Combined, handguns and firearms accounted for 23.2% (n = 1330) of all arrests (including murder and non-negligent manslaughter, rape, robbery, and aggravated assault). Arrests for weapon -related violent crime are concentrated in census tracts located in the north and south neighborhoods of Miami-Dade County. Findings indicate that arrestee factors and a neighborhood greenness index are more important than population density, weapon dealers and poverty in predicting arrests for weapon-related violent crime at the census tract level.

Death and long-term disability after gun injury: a cohort analysis

BACKGROUND

Gun injury accounts for substantial acute mortality worldwide and many others survive with lingering disabilities. We investigated whether additional health losses beyond mortality can also arise for patients who survive with long-term disability.

METHODS

We conducted a population-based individual patient analysis of adults injured by firearms who had received emergency medical care in Ontario, Canada, from Apr. 1, 2002, to Apr. 1, 2019. Longitudinal cohort analyses were evaluated through deterministic linkages of individual electronic patient files. The primary outcome was death or subsequent application for long-term disability in the years after hospital discharge.

RESULTS

In total, 8313 patients were injured from firearms, of which 3020 were injured from intentional incidents and 5293 were injured from unintentional incidents. A total of 2657 (88.0%) patients with intentional gun injury and 5089 (96.1%) patients with unintentional gun injury survived initial injuries. After a mean 7.75 years of follow-up, patients surviving intentional injuries had a disability rate twice as high as patients surviving unintentional injuries (19.7% v. 10.1%, p < 0.001), equivalent to a hazard ratio of 2.01 (95% confidence interval 1.80-2.25). The higher risk of long-term disability for survivors after intentional gun injury was not explained by demographic characteristics, extended to survivors treated and released from the emergency department, and was observed regardless of whether the incident was self-inflicted or from interpersonal assault. Half of the disability cases were identified after the first year. Additional predictors of long-term disability included a lower socioeconomic status, an urban home location, arrival by ambulance transport, a history of mental illness and a diagnosis of substance use disorder.

INTERPRETATION

Our study shows that gun death statistics underestimate the extent of health losses from long-term disability, particularly for those with intentional injuries. Additional and sustainable follow-up medical care might improve patient outcomes.

Firearm assault injuries by residence and injury occurrence location

Research on spatial injury patterns is limited by a lack of precise injury occurrence location data. Using linked hospital and death records, we examined residence and injury locations for firearm assaults and homicides in or among residents of King County, Washington, USA from 1 January 2010 to 31 December 2014. In total, 670 injuries were identified, 586 with geocoded residence and injury locations. Three-quarters of injuries occurred outside the census tract where the victim resided. Median distance between locations was 3.9 miles, with victims 18-34 having the greatest distances between residence and injury location. 40 of 398 tracts had a ratio of injury incidents to injured residents of >1. Routine collection of injury location data and homelessness status could decrease misclassification and bias. Researchers should consider whether residential address is an appropriate proxy for injury location, based on data quality and their specific research question.

The epidemiology of firearm injuries managed in US emergency departments

Background

Firearm-related injuries cause significant morbidity and mortality in the United States (US), consuming resources and fueling political and public health discourse. Most analyses of firearm injuries are based on fatality statistics. Here, we describe the epidemiology of firearm injuries presenting to US emergency departments (EDs).

Methods

We performed a retrospective study of the Healthcare Cost and Utilization Program Nationwide Emergency Department Survey (NEDS) from 2009 to 2012. NEDS is the largest all-payer ED survey in the US containing approximately 30 million annual records. Results include survey-adjusted counts, proportions, means, and rates, and confidence intervals of age-stratified ED discharges for firearm injuries.

Results

There were 71,111 (se = 613) ED discharges for firearm injuries in 2009; the absolute number increased 3.9% (se = 1.2) to 75,559 (se = 610) in 2012. 18-to-44-year-olds accounted for the largest proportion of total injuries with 52,187 (se = 527) in 2009 and 56,644 (se = 528) in 2012—a 7.2% (se = 1.6) relative rate increase and an absolute increase of 3.3/100,000 (se = 0.7). Firearm injuries among children < 5-years-old increase 16%, and 19% among children 5-to-9-years-old. 136,112 (se = 761)—or 48.2%—of those injured were treated and discharged home without admission; 106,927 (se = 755) were admitted. Firearm deaths represented one-third of all trauma mortality. Three-quarters of those injured resided in neighborhoods with median incomes below $49,250.

Conclusions

Firearm injuries increased from 2009 to 2012, driven by adults aged 18-to-44-years-old, and disproportionately impacting lower socioeconomic communities. Injuries also increased among young children. Firearm injuries remain a continued public health challenge, and a significant source of ED morbidity and mortality.