Defining the Full Spectrum of Pediatric Firearm Injury and Death in the United States

Objective:

To characterize the full spectrum of pediatric firearm injury in the United States by describing fatal and nonfatal injury data epidemiology, vulnerable populations, and temporal trends.

Background:

Firearm injury is the leading cause of death in children and adolescents in the United States. Nonfatal injury is critical to fully define the problem, yet accurate data at the national level are lacking.

Methods:

A cross-sectional study combining national firearm injury data from the Centers for Disease Control (fatal) and the National Trauma Data Bank (nonfatal) between 2008 and 2019 for ages 0 to 17 years. Data were analyzed using descriptive and χ² comparisons and linear regression.

Results:

Approximately 5000 children and adolescents are injured or killed by firearms each year. Nonfatal injuries are twice as common as fatal injuries. Assault accounts for the majority of injuries and deaths (67%), unintentional 15%, and self-harm 14%. Black youth suffer disproportionally higher injuries overall (crude rate: 49.43/million vs White, non-Hispanic: 15.76/million), but self-harm is highest in White youth. Children <12 years are most affected by nonfatal unintentional injuries, 12 to 14 years by suicide, and 15 to 17 years by assault. Nonfatal unintentional and assault injuries, homicides, and suicides have all increased significantly (P < 0.05).

Conclusions:

This study adds critical and contemporary data regarding the full spectrum and recent trends of pediatric firearm injury in the United States and identifies vulnerable populations to inform injury prevention intervention and policy. Reliable national surveillance for nonfatal pediatric firearm injury is vital to accurately define and tackle this growing public health crisis.

In the United States, death due to firearm-related injury is now the leading cause of mortality in children and youth 0 to 19 years, surpassing death from motor vehicle crashes in 2019.1 Furthermore, American children are 21 to 23 times more likely to be killed with firearms than in other high-income countries.2 This is a clearly growing national public health problem and prevention efforts are urgently needed. In the Centers for Disease Control and Prevention’s (CDC) public health approach to injury prevention, defining the problem based on the evaluation of injury data is the first step that should be undertaken before prevention strategies are implemented.3 Including nonfatal firearm injury, data are essential to define the full spectrum of pediatric firearm injury to inform prevention and policy more accurately. Nonfatal injuries are significantly more common than fatal injuries in children and can still cause severe injury, physical disability, psychological sequelae, and burden health care systems.4–8

Fatal firearm injury data are well-captured in national surveillance reporting systems such as the CDC’s Web-based Injury Statistics and Query Reporting System (WISQARS); however, nonfatal firearm injury estimates are no longer reported due to unstable estimates.9 Available sources of national nonfatal injury data each have their limitations (Supplemental Digital Content Table 1, http://links.lww.com/SLA/E462).10 Children injured by firearm injuries are commonly evaluated at trauma centers, either through direct arrival or transfer. The National Trauma Data Bank (NTDB) captures hospital-level data from the majority of trauma centers in the United States and has been shown to capture 2.4 times more pediatric cases than the Nationwide Emergency Department Sample (NEDS), thus making it the most comprehensive currently available national data source for nonfatal pediatric firearm injury data.11 As firearm victims that die at the scene (particularly suicides) are not taken to trauma centers and, therefore, not captured in NTDB, accurate fatal firearm data are lacking in this database.

The goal of this study is to fully define the spectrum of pediatric fatal and nonfatal firearm injury by combining 2 comprehensive databases to provide national-level data to guide prevention efforts and inform policies to address this public health crisis.

METHODS

A cross-sectional population-based study of national pediatric firearm injuries in the United States was conducted utilizing the NTDB for nonfatal data and WISQARS for fatal data during the time period of 2008 to 2019. As both the NTDB and WISQARS databases contain only deidentified patient information, this study was exempt from IRB approval and the requirement for informed consent was waived. The NTDB contains over 7 million patient records of trauma hospital admissions at over 900 trauma centers, making it the largest trauma registry available within the United States.12 Compiled registry data are available as research data sets (Trauma Quality Programs Participant Use Files). From these research data sets, firearm injuries for patients 0 to 17 years of age were queried based on the International Classification of Disease (ICD-9 or 10) external cause of injury codes (Ecodes). Ecodes include firearm-injury intents, which are broadly categorized as “Unintentional”, “Suicide/Intentional self-harm”, “Assault”, and “Other, including legal intervention” and ”Undetermined” (Supplemental Digital Content Table 2, http://links.lww.com/SLA/E463). Additional extracted data included demographics, geographic region, fatal versus nonfatal injuries, and shooting intent. To prevent overlap with fatal injuries obtained from the WISQARS database, patients with fatal firearm injury data extracted from NTDB were excluded.

WISQARS fatal injury database tracks all fatalities within the United States based on data reported to the National Center for Health Statistics that are linked to death certificates, law enforcement reports, and medical examiner and coroner reports.9 This is a comprehensive, reliable, and publicly available data source for fatal firearm injury data. WISQARS fatal injury data were queried for all pediatric firearm fatalities over the designated study period (2008–2019). Data filters were as follows—injury type: all; intent of death: all; mechanism of death: firearm; geography: all; custom age range: <1 to 17; sex: all, race and ethnicity: all; and metro versus nonmetro: all.

Population estimates from the United States census bureau were used to calculate the prevalence of pediatric firearm injuries and fatalities for both the NTDB and WISQARS cohorts based on annual population estimates for the corresponding years. When stratified by age, sex, or race/ethnicity, specific subpopulations were used to calculate crude rates (CRs). Given potential differences in exposures, the data were stratified by age group corresponding to school level (pre: 0–5 years, elementary: 6–11 years, middle: 12–14 years, and high: 15–17 years).

A comparison of categorical variables was conducted using descriptive analyses and the Pearson χ² test was used to compare fatal and nonfatal injury CRs. Linear regression was conducted to assess for significant linear trends over time. A P value <0.05 was considered significant based on 2-sided hypothesis testing. All analyses were performed using Stata version 15.1 (Stata Corp LP).

RESULTS

Between 2008 and 2019, the NTDB captured 43,353 pediatric firearm injuries, of which 5156 (12%) resulted in death and were excluded from the analysis in order not to duplicate data, resulting in 38,197 nonfatal injuries. The WISQARS database captured 17,774 pediatric firearm-related deaths over the same time period. Over this 12-year period, on average, there were 4615 injuries by firearm to children per year, of which 1481 were fatal (32%) and twice as many (3137; 68%) were nonfatal. The proportion of firearm injuries resulting in mortality each year has remained relatively stable, ranging from 30% to 35% of all injuries each year. Total firearm injuries (fatal and nonfatal) disproportionally affected males (84%), the 15 to 17-year-old age group (76%), Black, non-Hispanic children (CR: 49.43 per million individuals vs White, non-Hispanic 15.76 per million individuals, Hispanic 14.13 per million individuals, and Other 9.8 per million individuals), and the Midwest and Southern regions of the country (67% occurred in these regions). Firearm deaths had a similar incidence in the South and Midwest (South 22.88 per million individuals, Midwest 22.65, West 17.34, and Northeast 11.05; 2008-2019), whereas nonfatal injuries were highest in the Midwest (38.92 per million individuals, South 33.75, West 30.14, and Northeast 23.49; 2008-2015).

Shooting Intent

For nonfatal injuries, an assault was the most common shooting intent (72%), followed by unintentional (20%), self-harm (3%), undetermined (5%), and legal intervention (0.6%). For fatal injuries, homicide was also the most common shooting intent (56%), followed by suicide (36%), undetermined (2%), unintentional (6%), and legal (0.5%). Self-harm had the highest case-fatality rate of all shooting intents (84%). The case-fatality rate for assault was 27% and only 12% for unintentional intent. Notably, there were significant differences between fatal and nonfatal prevalence for self-harm and unintentional intents (P < 0.001) (Fig. 1).

FIGURE 1.

Comparison of shooting intents between fatal and nonfatal pediatric firearm injuries.

When shooting intent was stratified by age, there was an age-dependent increase in the injury and death incidence for every shooting intent category (Fig. 2). Interestingly, even between the 12 to 14-year age group (middle school age) and the 15 to 17-year age group (high school age), injuries and deaths in the latter were 6 times higher from assault, 2 times higher from unintentional, and 3 times higher from self-harm. An assault was the most common type of injury in all age groups; however, nonfatal unintentional injuries were slightly more prevalent than assault in children <12 years (assault 44% vs unintentional 48%), whereas nonfatal assault injuries were markedly more common above age 12 years (assault 77% vs unintentional 16%). Homicides accounted for the majority of deaths in all age groups except the 12 to 14-year age group, in which suicide deaths were more common (homicide 39% vs suicide 52%) (Table 1).

FIGURE 2.

Prevalence of injuries and deaths stratified by shooting intent.

TABLE 1.

Demographics Stratified by Shooting Intent

	Nonfatal Injuries; n (%)	CR Per 1 Million	Fatal Injuries; n (%)	CR	P
Black, Non-Hispanic
Assault	17,768 (62.8)	132.17	5645 (20.0)	41.99	0.034
Unintentional	2782 (9.8)	20.69	332 (1.2)	2.47	—
Self-harm	227 (0.8)	1.69	506 (1.8)	3.76	—
Other/undetermined	871 (3.1)	6.48	162 (0.6)	1.21	—
White, Non-Hispanic
Assault	2486 (17.1)	5.18	1852 (12.7)	3.86	0.013
Unintentional	3394 (23.3)	7.08	578 (4.0)	1.21	—
Self-harm	763 (5.2)	1.59	4944 (33.9)	10.31	—
Other/undetermined	369 (2.5)	0.77	182 (1.2)	0.38	—
Hispanic
Assault	5113 (54.6)	24.05	2144 (22.9)	10.09	0.155
Unintentional	883 (9.4)	4.15	123 (1.3)	0.58	—
Self-harm	151 (1.6)	0.71	661 (7.1)	3.11	—
Other/undetermined	208 (2.2)	0.98	76 (0.8)	0.36	—
Other race/ethnicity
Assault	1810 (48.7)	31.16	224 (6.0)	3.86	0.004
Unintentional	549 (14.8)	9.45	41 (1.1)	0.71	—
Self-harm	113 (3.0)	1.95	251 (6.8)	4.32	—
Other/undetermined	710 (19.1)	1.24	17 (0.5)	0.29	—
0–5 yr
Assault	863 (25.2)	2.99	851 (24.9)	2.95	0.844
Unintentional	1082 (31.6)	3.75	363 (10.6)	1.26	—
Self-harm	36 (1.1)	0.12	0	0	—
Other/undetermined	182 (5.3)	0.47	44 (1.3)	0.15	—
6–11 yr
Assault	1296 (32.1)	4.40	878 (21.7)	2.98	0.687
Unintentional	1299 (32.2)	4.41	174 (4.3)	0.59	—
Self-harm	50 (1.2)	0.17	94 (2.3)	0.32	—
Other/undetermined	204 (5.1)	0.51	44 (1.1)	0.15	—
12–14 yr
Assault	3358 (39.2)	22.47	1118 (13.1)	7.48	0.001
Unintentional	1710 (20.0)	11.44	174 (2.0)	1.16	—
Self-harm	289 (3.4)	1.93	1471 (17.2)	9.84	—
Other/undetermined	351 (4.1)	1.69	85 (1.0)	0.57	—
15–17 yr
Assault	21,660 (54.2)	142.08	7040 (17.6)	46.18	<0.001
Unintentional	3517 (8.6)	23.07	363 (0.9)	2.38	—
Self-harm	879 (2.2)	5.77	4810 (12.0)	31.55	—
Other/undetermined	1421 (3.6)	6.75	265 (0.7)	1.74	—

Crude rate per million individuals.

Percentages are described within each subcategory.

P values compare nonfatal and fatal.

When shooting intent was analyzed by race/ethnicity, clear disparities were identified, and minority races were disproportionally affected across the board. Black, non-Hispanic children, and youth were most frequently injured or killed in every category except suicide deaths, comprising 58% of the assault and unintentional injuries. This discrepancy was most apparent regarding assault: Black children were 10.5 times more likely to be killed by assault and 25.5 times more likely to be injured by assault, but only 2 to 3 times more likely to be killed or injured by unintentional shootings compared with their White, non-Hispanic counterparts (race-adjusted CRs). Nonfatal assault and unintentional injuries were also disproportionally higher for children of “Other” races compared with White. Interestingly, a fatal assault was high in Hispanic children (second to Black), but this ethnic group was the least frequently injured of all race/ethnic groups by unintentional and self-harm intents (Table 1).

The majority of self-harm injuries (suicide attempts and suicide) were in White, non-Hispanic children (75%). Compared with Black children, suicide deaths were 2.7 times more prevalent in White, non-Hispanic children. However, when separated from the other group, Native American, non-Hispanic youth had the highest CR for the suicide of all race/ethnic groups (17.38 per million individuals, compared with 10.31 per million individuals for White, non-Hispanic). Native American children also had the second highest rate of after Black children of fatal assaults (13.8 per 1 million individuals). This is a significant disparity given that only 1.3% of the United States under 18 population is Native American.

Temporal Trends

Shooting Intent

When all injuries and deaths were combined, overall, there was a slight decrease in incidence between 2008 and 2014 followed by a steady increase until 2017 and a slight decrease until 2019. The majority (84%) of firearm injuries from 2008 to 2019 occurred within the 3-year period from 2015 to 2017. Nonfatal unintentional injuries (β = 0.42, P < 0.001) and fatal self-harm injuries (β = 0.48, P < 0.001) increased significantly throughout the study period, whereas assault injuries initially decreased, then started to increase in 2013 (Fig. 3).

FIGURE 3.

Nonfatal and fatal temporal injury trends by shooting intent Curve estimate significance: *P <0.05; **P <0.001. Significant linear increase started in 2013 for fatal assaults, in 2014 for nonfatal assaults, and since 2008 for nonfatal unintentional and fatal self-harm injuries.

Age Group

Starting in 2014, nonfatal injury trend slopes significantly increased only for the 15 to 17-year age group (β = 10.14; P = 0.047) and fatal injury trend slopes increased significantly in all age groups except for 0 to5-year age group (0–5 years: β = 0.08; P = 0.089; 6–11 years: β = 0.13; P = 0.009; 12–14 years: β = 0.95; P < 0.001; 15–17 years: β = 2.18; P = 0.019). With fatal and nonfatal injuries combined, all shooting types increased significantly except for self-harm in children <12 years (Fig. 4). Unintentional overtook assault as the #1 shooting intent in children in the 0 to 5-year age group in 2016, and self-harm overtook unintentional as the #2 shooting intent (after the assault) in children in the 12 to 14-year age group in 2013, and these trends have remained the same way.

FIGURE 4.

Temporal trends by shooting intent for children < and >12 years. Curve estimate significance: *P <0.05; **P <0.001.

Race/Ethnic Group

Significant linear growth over the study period was noted in unintentional deaths among Black, non-Hispanic children (β = 0.863, P = 0.029) and suicide deaths among White, non-Hispanic children (β = 0.96, P < 0.001). Nonfatal self-harm in Black, non-Hispanic children declined sharply between 2014 and 2017, whereas it remained stable in the White population.

Region

Nonfatal regional trends were only analyzed until 2015, as after this year the NTDB stopped reporting regional data. In 2013, the nonfatal injury rate started to rise in the South and Midwest for all shooting intents, whereas it remained stable in the West and Northeast. Fatal regional trends were analyzed for the entire study period. Homicides increased significantly in the Midwest (β = 0.239; P = 0.05) and decreased significantly in the Northeast (β = −0.21; P = 0.02) and West (β = −0.45; P = 0.01). Suicides increased significantly in all regions except the Northeast (Northeast: β = 0.10; P = 0.11; South: β = 0.60; P < 0.001; Midwest: β = 0.58; P < 0.001; West: β = 0.44; P <0.001), and unintentional deaths had no significant change in any region.

DISCUSSION

Firearm injury is now the leading cause of death for children and adolescents living in the United States.13 We have been successful at reducing deaths from motor vehicle crashes by prioritizing it as a public health problem, and we can do the same for firearm injury.14 First, we must define the full scope of the problem, which includes nonfatal injuries and risk factors. As there is no national registry for reporting nonfatal firearm injuries, it can be challenging to find nonfatal firearm injury data. A publicly available and now commonly utilized source is the Gun Violence Archive.15 Other national nonfatal data sources come from hospital-level data from emergency departments4 or trauma centers;16,17 however, every database has its limitations when used in solitude. Hospital-level firearm data propose that nonfatal pediatric firearm injuries are 8 to 15 times more common than fatal injuries; however, as out-of-hospital deaths are not captured, this comparison is misleading as the data are incomplete. In our study, we have attempted to combine comprehensive fatal and nonfatal injury data sources to gain the most accurate understanding of national firearm injury data.

Kaufman et al,5 combined CDC’s WONDER database for fatal firearm injuries and the National Emergency Department Sample (NEDS) for nonfatal firearm injuries for children and adults. Adolescents 15 and over were grouped with adults (15–34 years) and sex and race/ethnic group data were combined for adults and children; thus demographic characterization of pediatric-specific data is not clear in this study. Unintentional injuries (fatal and nonfatal) are the most common injury type in the 0 to 14-year age group, which differs from our study that finds assault to be the most common for all age groups; furthermore, they report that these injuries decreased during the study period, which also differs from our results.

Direct comparison of injury data (Supplemental Digital Content Table 3, http://links.lww.com/SLA/E464) reveals increased numbers in all injury types in our study except nonfatal unintentional injuries. This could be a consequence of NEDS capturing more injuries from rural regions than NTDB;18 in addition, the published comment to this study states that these injuries were overestimated due to coding misclassification, as hospital coding software defaults to the “accidental” Ecode for all firearm injury unless differently specified by the coders at these hospitals.10,19 Trauma centers that contribute to the NTDB database have trained trauma registrars that do the injury coding; thus firearm injuries are carefully reviewed before assigning Ecodes (accidental vs assault).20 The largest discrepancy in trauma center coding has been shown not to be for accidental injuries but for mistaken target/bystander injuries, which are coded as assaults as there is no ICD 9 or 10 Ecode for these.20

A second similar study by Fowler et al21 evaluated pediatric firearm data by combining data from WISQARS and the National Electronic Injury Surveillance System. As the study period ended in 2014, which is before the spike in national firearm injury incidence, this data may not be fully representative of the current situation. This report identifies temporal trends for pediatric unintentional and assault injuries to be decreasing, which differs from our study. Furthermore, nonfatal firearm injury estimates are not reliable as per the CDC.22

Our study reflects the most contemporary and pediatric-specific firearm injury data to date, based on two of the most comprehensive, currently available national firearm injury databases. Since the increase in 2015, on average, nearly 1700 children and adolescents are killed by firearms and at least 3700 are injured by firearms per year. Given that unintentional injuries had the lowest case-fatality rate and thus are the least severe, it is possible that minor injuries were managed at nontrauma centers; thus the nonfatal unintentional injury data are likely underestimated in our study. Despite this limitation, when directly compared with the Gun Violence Archive data that is commonly referenced by media, our study has identified 1.4 times more pediatric firearm injuries and deaths during the same time period (Supplemental Digital Content Table 3, http://links.lww.com/SLA/E464): therefore, this problem is even worse than we think. The data presented in this study confirm some known disparities, identify a few new vulnerable subpopulations and elucidate injury trends. It also highlights the need for a reliable national surveillance system for capturing nonfatal firearm injuries given the limitations of the current databases.

Relative to White, non-Hispanic children, minorities were disproportionally affected in every shooting category, most prominently due to assault. It should be noted that mass shooting deaths make up a very small portion of the assault numbers, affecting fewer than 35 children per year.23 The literature on firearm injury to children of Hispanic and Other races is sparse, thus this study contributes new findings to further explore. The fact that unintentional shootings and self-harm are low in Hispanic children may be a consequence of the social structure and multigenerational, tight-knit family homes. It is well-documented that domestic and street violence account for the majority of assaults, which Black youth are most at-risk for.24 As there is no Ecode that differentiates assault from bystander shootings, the incidence of these is unknown.20 Continued research on root causes, risk and protective factors, and public health community-based interventions are desperately required to mitigate violence involving children of minority races.

Self-harm was most prevalent in White, non-Hispanic children, which is similar to adults and may be a consequence of this race group having the highest firearm ownership and thus access to firearms.25 In addition, fatal self-harm (suicide) was most prevalent among Native American, non-Hispanic youth, which is consistent with recent data.26 Given the low case counts reported for this race group, these numbers could be unreliable;1 however, self-harm in Native American youth has been an issue for some time and should not be dismissed.27 The association of a mental health disorder or other risk factors was not assessed in our study but continued efforts to identify risk factors and research effective suicide prevention interventions for both of these vulnerable populations—starting at the middle school level— are of utmost importance.

The value of including nonfatal firearm injuries in the national conversation is most evident regarding unintentional injuries. If only fatal data are considered then this may not seem to be a major issue, as only a small proportion (6%) are affected. However, when nonfatal data are included, unintentional shootings account for at least 15%. It is estimated that more than half of United States firearm owners do not store their guns safely, access to firearms in the home increases the risk of adolescent suicides fourfold, safe firearm storage and child access protection laws reduce the risk of suicide and unintentional deaths as most of these firearms are accessed at home.28–31 Our data show that when unintentional and self-harm injuries are combined, they account for almost a third of pediatric firearm injuries that safe firearm storage could have potentially prevented. Furthermore, firearms used for assault are often also accessed at home; thus safe firearm storage education and awareness and supporting strong child access protection laws storage is vital for the prevention of pediatricfirearm injury and should be strongly promoted at a national level.

The regional data described in this study reveals that the South and Midwest have been most affected by pediatric firearm injury. There is a larger population of Black Americans in the South, but this is not true in the Midwest, which is majority White. Thus, these data are likely associated with increased firearm ownership and less restrictive firearm laws in these regions.32 Firearm injury prevention for children in these regions must be prioritized.

Our study has several limitations. First, participation within the NTDB is voluntary and limited to trauma centers, which represent more urban settings than rural,18 and thus likely have inherent selection bias. NTDB is not a surveillance system and has not been validated as a stable estimate of national nonfatal firearm injury data. However, as described, at this time we do feel that this is the most reliable data that we have available, especially for serious injuries requiring hospitalization. The national surveillance system that does exist (National Electronic Injury Surveillance System) has been shown to have unstable data estimates of nonfatal injuries, especially regarding undetermined injuries.10,33 Second, minor injury cases that were managed at home, at nontrauma centers, or discharged from the emergency room without admission at trauma centers were not accounted for in this data. Given these limitations in missing data, the CR calculations in this study are considered an estimate of true population-level data. Third, association with poverty or other social factors could not be assessed due to limited individual-level data in the WISQARS database. Fourth, there may be inaccuracies or miscoding in the shooting intents in both databases, which is a similar issue with other surveillance systems.10,19,20 Fifth, the data trends may be slightly biased given the long time frame of this study as there have been changes and expansion of both databases during the studied interval. Finally, the transition from ICD-9 to ICD-10 occurred in 2015. Although there was a change in the specific number code assigned to each intent, there was no change in the categories of intentionality between the 2 versions of ICD codes.20 Thus, the coding transition is unlikely to have contributed significantly to the increased incidence of NTDB injuries after 2015.

CONCLUSIONS

To conclude, this study adds and supports several key findings to the pediatric firearm injury literature that may help to focus research, injury prevention, and policy efforts. More than 5000 children and adolescents per year are being seriously injured or killed by firearms in the United States. Nonfatal injuries are twice as common as fatal injuries and most injuries and deaths are due to assault. Black children, males, and high school-age adolescents are disproportionally affected by all injury types and thus are the highest-risk groups. Preschool and elementary school-age children are most affected by unintentional injuries, but fortunately, these are also the least likely to be fatal. Suicide by firearm is the leading cause of death in middle school-age children and most common in Native American and White, non-Hispanic youth. Temporal trends show that in addition to the increase in injuries related to violence (assault) during the past decade, unintentional injuries and suicides have risen even more significantly, especially in regions where firearm ownership rates are highest. Our findings strongly support safe household firearm storage, which has important implications for advocacy and policy.

Although national-level data such as that presented here are a good start, there are still vital missing data on specific risk factors such as individual and neighborhood-level factors and shooting circumstances that would help injury prevention planning. Given the limitations in the current national databases and surveillance systems, we impress the importance of instituting reliable surveillance to accurately capture nonfatal firearm injuries and risk factors to address this significant public health problem as a top national priority.

Supplementary Material

sla-278-0010-s001.docx ^{(14.5KB, docx)}

sla-278-0010-s002.docx ^{(16KB, docx)}

sla-278-0010-s003.docx ^{(15.4KB, docx)}