Firearm injuries in Missouri

Firearm deaths continue to be a major public health problem, but the number of non-fatal firearm injuries and the characteristics of patients and injuries is not well known. The American College of Surgeons Committee on Trauma, with support from the National Collaborative on Gun Violence Research, leveraged an existing data system to capture lethal and non-lethal injuries, including patients treated and discharged from the emergency department and collect additional data on firearm injuries that present to trauma centers. In 2020, Missouri had the 4th highest firearm mortality rate in the country at 23.75/100,000 population compared to 13.58/100,000 for the US overall. We examined the characteristics of patients from Missouri with firearm injuries in this cross-sectional study. Of the overall 17,395 patients, 1,336 (7.7%) were treated at one of the 11 participating trauma centers in Missouri during the 12-month study period. Patients were mostly male and much more likely to be Black and uninsured than residents in the state as a whole. Nearly three-fourths of the injuries were due to assaults, and overall 7.7% died. Few patients received post-discharge services.


Introduction
Firearm injuries and deaths continue to be a major public health problem, resulting in 48,830 deaths in the US in 2021, of which 53.8% were suicides and 42.9% were homicides [1].However, the number of non-fatal firearm injuries and the characteristics of patients and injuries is not well known, as documented by a recent report from NORC at the University of Chicago [2].The existing databases do not collect or report on clinical information such as severity of injuries and their outcomes, nor do they provide data that better contextualize injuries including community characteristics, individual risk factors, co-morbid illnesses, substance abuse or

Methods
The ACS Trauma Quality Improvement Program (TQIP) collects data for the purposes of performance benchmarking from over 700 centers, representing more than 800 distinct trauma programs, across the U.S., capturing an estimated 90%-95% of all level 1 and 2 (major trauma centers) and a less comprehensive sample of level 3 trauma centers.Prior studies indicate that verified and/or designated trauma centers care for approximately 70% of medically treated firearm injuries in the U.S [5].Data collected are specified in the National Trauma Data Standard [6] and include patient and injury characteristics, processes of care and outcomes among all patients admitted, transferred to that center, or who died in hospital.Notably, prior to this initiative, the ACS TQIP only collected data on people with firearm injuries who met these criteria; patients assessed in the emergency department and discharged home were not routinely captured.
We recruited volunteer participation from the entire TQIP sample and provisionally enrolled 165 trauma centers to participate in this program; 128 ultimately contributed data, of which 11 were in Missouri.Centers had to participate in ACS TQIP over the duration of the study and agreed to collect data elements listed below on all individuals treated with firearm injuries (including those discharged alive from the emergency department).Centers were not provided any funding to support additional data collection.Centers were recruited through electronic mailings to the trauma directors in all ACS TQIP centers, holding a webinar for potential centers, ACS COT newsletters, and direct contact by study investigators with trauma directors.

Study population
Patients eligible for the study were individuals of any age arriving alive at a participating trauma center in Missouri and with residential zip codes in Missouri between March 1, 2021 external datasets (e.g., the Economic Innovation Group's Distressed Community Index).Authors received special privileges to access these data, as they were the Principal Investigators, Co-Principal Investigators, or staff working on the original grant issued by the National Collaborative on Gun Violence Research (NCGVR) to the ACS COT.While the exact dataset used for this manuscript cannot be shared, a de-identified version can be made available for researchers interested in conducting their own analyses.Please email traumaquality@facs.orgfor more information on how to request this dataset.and February 28, 2022 who had sustained an injury due to a firearm.The study was approved by Advarr CIRBI TM .Consent for the registry to abstract data from the medical record was waived by the IRB.Centers started participation at varying times during the study period and may not have contributed 12-months of data.

Data abstraction
Data were abstracted from the electronic health record by trained personnel at each trauma center including registrars, clinicians, and research staff.In addition to the standard data collected for TQIP, the abstractors also extracted available additional data specific to this study.This new data included (See S1 File): • Demographic characteristics: education, veteran status, and caregivers (pediatric patients) • Risk factors: illicit substance use and intoxication, history of or newly identified/diagnosed mental illness, prior arrests/involvement in the criminal justice system, Adverse Childhood Experiences.
• Circumstances: Context/preceding events (assaults: altercation, commission of a crime, drug or gang-related, bystander, mass violence, intimate partner violence, child abuse; suicide attempts: life stressors, declining mental illness, terminal medical illness, suicide-homicides; unintentional: playing, cleaning, handling, hunting, accidental discharge when unaware of firearm presence), specific location of injury occurrence, perpetrator-victim relationships.
• Firearms (for self-inflicted and unintentional injuries): type implicated in the injury, ownership, access and storage at time of the incident (for self-inflicted and unintentional injuries).
• Functional outcomes, medical needs and services at time of discharge.
The direction provided to participating centers was to abstract these data from the EMS, emergency department, or inpatient medical record.There was no expectation for additional interview of patients to capture data that was not otherwise routinely collected during the course of care.Together, this study required two modifications to existing data abstraction for trauma centers: 1) expanded inclusion criteria to capture patients discharged from the ED to include those that may not have qualified for TQIP inclusion and 2) expanded data abstraction to include additional information on the context of firearm injuries.

Data analysis
Sex was imputed since the original variable had a high proportion of missingness among patients due to abstraction error at two centers.The alcohol screen result was also imputed for all patients (originally 17.7% missing overall).
Race, mental illness, injury intent, drug screen, and discharge services were all variables in which more than one choice could be selected by the coders and were recoded to ease interpretation.Patients who were reported to have more than one racial category were grouped together.Injury intent was originally a variable in which a patient could have multiple intents, and the context of injury variable was used to help inform the mutually exclusive categories used in these tables.
Drug screen results were also reported here in mutually exclusive categories.Patients positive for only one drug were represented in the relevant category.A separate category for patients who were positive for cannabinoid in addition to any other drug listed was created, in addition to a separate category for patients who were positive for more than one drug excluding cannabinoid.
Post-discharge service variables (rehabilitation/post-discharge needs, home health needs, and psychosocial ancillary services) were also "select all that apply" variables; within each postdischarge service variable, those with more than one service were counted in each category that was applicable (non-mutually exclusive categories).A variable was also created to indicate whether a patient had any post-discharge service in any of the three service categories (rehabilitation/post-discharge needs, home health needs, and psychosocial ancillary services) vs. none.
An urbanicity variable was created to further explore the comparisons between Missouri patients and patients in other states.By linking the Rural-Urban Commuting Codes (RUCA) [7] via ZIP code to our patient data, we categorized zip codes of their residence with a RUCA code of 1 as urban, and all other codes (2-10) as rural.
To provide additional community context around the patient and injury, we linked data from the Distressed Community Index (DCI) to patient records meeting our inclusion criteria via ZIP code.The DCI is a validated index of prosperity that includes variables related to education, housing, unemployment, poverty, and changes in business establishments [8].The scale ranges from 0-100 and is sorted into quintiles with the highest scores representing the most distressed communities.DCI is not calculated for ZIP codes with less than 500 residents.DCI data is publicly available and utilizes the U.S. Census Bureau's American Community Survey 5-Year Estimates as well as the Census Bureau's Business Pattern's dataset for 2016 and 2020.
P-values were calculated using Pearson's Chi Squared tests for mutually exclusive categorical variables and two-sided t-tests for numerical variables.In cases where there were fewer than 10 observations per cell, Fisher's exact tests were used.P-values were considered significant at the 0.05 alpha level.All data cleaning and tables were done in RStudio 4.2.2.

Demographic factors
Data were collected on 17,395 patients of whom 1,336 (7.7%) were treated at one of the 11 participating trauma centers in Missouri.As shown in Table 1, these patients were mostly young (median age 29.0) adult males (83.0%).Firearm injured patients in Missouri were much more likely to be Black (78.8% compared with 11.8% of state residents) and much less likely to be White (15.0%compared with 82.5% of state residents).More than half of the patients in Missouri were recorded as being uninsured, compared to 9.3% in the state as a whole.The majority (72.9%) of firearm injuries were due to assaults with few due to self-inflicted injuries, since most such firearm injuries result in death at the scene.
In the 708 patients for whom context of injury data was available, assault injuries most commonly occurred in the context of community violence.Shootings related to interpersonal and random violence were also common.Self-inflicted injuries occurred in the context of intoxication in 25.8% patients, and were associated with a personal crisis in almost half, while mental illness was reported in 48.4%.Unintentional injuries were usually associated with handling the firearm.

Risk factors for firearm injury
We examined a number of factors, as recorded in the medical record, which are known to be associated with firearm injuries as shown in Table 2.About 1 in 6 patients was intoxicated at the time of the emergency department (ED) assessment.Drug testing was not done on 71.8% of patients in Missouri, but among those tested only 22% were negative for any drugs.The most common drug found on toxicology assessment was cannabis.Prior mental illness was recorded in 12.6% of patients but was missing in 29.0%.The injuries occurred in a variety of settings, most commonly in homes, streets, motor vehicles and commercial areas.Nearly three-fourths (72.2%) of patients with prior assault injuries had sustained a prior gunshot wound.

Characteristics of treatment
Overall, 43.4% of firearm patients treated at these trauma centers were discharged from the hospital alive; 7.7% of patients presenting to trauma centers in Missouri with firearm injuries died in the hospital, with 66 of the 103 fatalities dying in the ED (Table 3).Only 12.3% of patients received services on discharge from the hospital; only 4.3% received social work or mental health services and only 5.4% received any services related to violence.We also examined how these treatments varied across age groups (Table 4).Few of those 65 and older were discharged from the ED, and more this age required skilled nursing facility (SNF) or other long-term care on discharge from the hospital.Those in the middle age group were significantly less likely to receive any services on discharge from the hospital.
When we examined patient outcomes by insurance status, stratifying for injury intent, there were no significant differences in mortality (Table 5).However, those who were injured by assault and covered by Medicaid or Medicare were less likely to be discharged home from the ED compared to those in the other two insurance groups.Those with commercial insurance had the lowest likelihood of receiving any post-discharge services.

Socioeconomic community factors
Patients with firearm injuries in Missouri lived in communities with much higher DCI scores and were much more likely to live in the most distressed communities (69.4% vs 24.6%) compared to the state as a whole (Table 6).Older adults were less likely to live in the highest DCI quintile compared to those 65 and younger (42.9% vs 70.0%).Patients who sustained firearm injuries from assault had the highest mean DCI and were most likely to live in the most distressed communities.Patients in urban areas were nearly twice as likely to come from the most distressed quintile compared to those living in rural areas.

Rural and urban differences
Using the RUCA codes for residence zip code, 7.7% of the patients were from rural Missouri.Firearm injured patients living in rural areas were somewhat older than those in urban areas (Table 7).They were much less likely to be Black compared to patients in urban areas (18.4% vs 83.9%) Patients living in urban areas were more likely to be uninsured.Three-quarters of patients in urban areas were injured in assaults compared with less than half in rural areas.Fewer patients in urban areas received any post-discharge services.

Discussion
In this analysis of 1336 patients with Missouri residential zip codes treated at 11 trauma centers in Missouri, patients were mostly male and much more likely to be Black and uninsured than residents in the state as a whole.Nearly three-fourths of the injuries were due to assaults, and overall 7.7% died.Few patients received post-discharge services.
A striking finding is the much higher levels of distress in communities in which individuals are experiencing firearm injuries compared to residents of the state as a whole.This is especially true for injuries among patients in urban areas.While firearm injuries have long been associated with poverty, urban areas in Missouri with firearm injuries represent extremely distressed and deprived communities.Combined with the fact that the vast majority of firearm injuries in these urban areas were to Black males, the data demonstrate the effects of longstanding structural racism on the epidemiology of firearm injuries [9][10][11].
In this study, only 3.8% had a prior arrest although 114 of the total (11.8%) had a prior assault.Given the high degree of socioeconomic disadvantage of the areas in which these patients lived, this is surprising compared to other studies which have found half of firearm injured patients had an arrest in the prior 2 years [12].This difference may be due to the large amount of missing data in our study for prior assaults and prior arrests.
There was a marked dearth of services for the people being discharged alive from the hospital, especially those with injuries due to assaults.Very few had social work services post-discharge, and even fewer had access to violence prevention programs.Such services are critically important in order to both prevent injury recidivism as well as to help patients and families recover from their trauma [13].More than one-half of survivors of firearm violence develop Post Traumatic Stress Disorder and/or depression [14] and need interventions starting in trauma centers [15].It should start with inpatient screening and intervention followed by transition to outpatient care.Without any interventions after discharge, individuals hospitalized with firearm injuries have a 30-fold greater chance of being re-hospitalized with a firearm injury and a 7.3-fold greater chance of dying from a firearm injury than those in the general population [16].While hospital-based violence intervention programs are often suggested as interventions to prevent recidivism and subsequent violence [17], the currently available evidence suggests otherwise [18,19].
One of the concerns for firearm injured patients in Missouri is that a large proportion of firearm injured patients are not insured, which can result in further barriers to obtaining follow-up care.While Missouri expanded Medicaid in 2021, the high percentage of firearm injured patients without insurance is of great concern given the potential benefits of Medicaid expansion that include reduction of suicide rates and improving access to inpatient rehabilitation after acute hospitalization among those who are injured by firearms [20,21].
This study has a number of important implications.The amount of missing data for variables that can be used to guide prevention efforts was large.We do not think this was a failure of chart abstraction by the coders but lack of collection and documentation of this data by health care providers.Efforts to direct resources for primary and second prevention in communities will require that health systems collect this information.Health systems also need to view firearm-injured patients different than other trauma patients, for example those injured in motor vehicle crashes.The high rate of recidivism and risk of subsequent violence and violent death requires that health systems develop, test and implement effective programs to respond to this problem.The clear association with firearm injuries and deaths with socioeconomic distress and poverty is well-known and will require investment in the most distressed neighborhoods in communities.

Limitations
This study does have important limitations.First, many patients with firearm injuries die at the scene, especially those who used a firearm to attempt suicide, 90% of which result in death [22].These individuals and characteristics unique to the patients that experience firearm suicide are thus largely missing from these data, reflected in that only 4.3% of the patients in this study had self-inflicted injuries.Second, while the authors believe that the additional data collected on patients admitted to trauma center hospitals is important to understand the risks and circumstances of their injuries, and significant training of trauma center data personnel occurred, the data missingness may represent inadequate resources, training and practices of the entire healthcare team, all of whom gather medical and social histories and document in the EHR.Additionally, the extra time required of trauma personnel to locate and extract the additional data may have limited their ability to do so, notably during the data collection period when trauma centers reported an increase in trauma volume as a result of the COVID-19 pandemic.
In addition, the total number of patients included in this study is an under representation of all the patients treated at participating centers, as centers started participation at varying times during the study period and may not have contributed 12-months of data.The amount of time required to extract data was significant and may have been difficult for centers especially since it occurred when hospitals may have been very busy with patients during the pandemic and no additional funding was available to support this data collection at the center level.In summary, this study describes patients with firearm injuries living in Missouri and treated at ACS trauma centers, enhancing our knowledge of their injuries, clinical care, characteristics, and injury circumstances.The results emphasize the need for interventions at multiple levels for people living in distressed communities to reduce the burden of firearm-related harms, in addition to highlighting potential opportunities to improve access to care and evidence-informed services among those who survive firearm injuries.

Table 1 .
(Continued) 1. Subcategories are not mutually exclusive, and percentages calculated among patients where parent variable was reported as present.2.Random "indicates an act in which the suspect is not concerned with who is being harmed, just that someone is being harmed, such as a person who shoots randomly at passing cars from a highway bridge."https://doi.org/10.1371/journal.pone.0294737.t001

Table 2 . Pre-injury factors of firearm injury patients with Missouri zip codes.
Subcategories are not mutually exclusive, and percentages calculated among patients where parent variable was reported https://doi.org/10.1371/journal.pone.0294737.t002

Table 4 . Hospital course and outcomes of firearm injury patients of firearm patients by age group.
2. Discharged/transferred to court/law enforcement, left against medical advice or discontinued care https://doi.org/10.1371/journal.pone.0294737.t004

Table 7 .
(Continued) Jail, institutional care, mental health, etc. 2. Discharged/transferred to court/law enforcement, left against medical advice or discontinued care 3. Percentages calculated among surviving patients.Patients receiving more than one post-discharge service are represented in all relevant categories.https://doi.org/10.1371/journal.pone.0294737.t007