Pediatric Firearm Injury Emergency Department Visits From 2017 to 2022: A Multicenter Study

We conducted a retrospective cross-sectional study using data from 9 children’s hospital EDs participating in the PECARN Registry from January 2017 to November 2022. The PECARN Registry comprises electronic health record data from every pediatric ED encounter at participating institutions, harmonized into a deidentified, central repository.¹⁶  Variables include demographics, laboratory and radiology results, International Classification of Diseases, 10th Revision, Clinical Modification diagnoses, and disposition. The date range included all available data in the PECARN Registry at the time the analysis was conducted. US Census regions of participating hospitals were: 5 Midwest, 2 West, 1 Northeast, and 1 South. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline.¹⁷  The study was approved by the institutional review boards of all study sites and the University of Utah Emergency Medical Services for Children Data Center.

We identified ED visits by children <18 years old with any International Classification of Diseases, 10th Revision, Clinical Modification diagnosis code for a firearm injury, excluding non-powder firearm injuries¹⁸  and excluding codes for subsequent visits (Supplemental Table 3).

Demographic characteristics included patient age, sex, insurance type (private, public, other/self-pay), composite race and ethnicity, and zip-code-level Child Opportunity Index (COI) 2.0. The COI 2.0 is a composite measure of neighborhood-based opportunities for children, including education, health and environment, and social and economic factors.^19,20  We analyzed COI in quintiles (very low, low, moderate, high, or very high neighborhood opportunity), as provided in the COI dataset.^21,22  Race and ethnicity were categorized by ethnicity (Hispanic or non-Hispanic) and then by race within the non-Hispanic group. Because of limited sample sizes, non-Hispanic racial categories were analyzed as Black, White, and other non-Hispanic (including American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, Multiple Races, and other). Race and ethnicity were examined as social constructs and included in analyses because of differential COVID-19 pandemic experiences and previously described differences in pediatric firearm injury rates.^3,23,24 

Clinical characteristics included triage acuity defined by emergency severity index,²⁵  the injury severity score (0–8, 9–15, ≥16),^26,27  body region (head or neck, chest, abdominal or pelvic content, extremities or pelvic girdle, face, external),²⁸  the intent of injury based on diagnosis codes, ED disposition (admitted/observation/transferred, discharged, died, other), hospital length of stay (in days) among the subgroup admitted to the hospital, and death (in the ED or after admission to the hospital). If a visit had codes for multiple intents, we assigned each ED visit to one intent as follows: we classified ED visits with undetermined and assault codes as assaults, undetermined and accidental codes as accidental, undetermined and self-inflicted codes as self-inflicted, and all other combinations as “multiple” intents.

The exposure of interest was time period, defined as before the pandemic (January 1, 2017 to February 29, 2020) and during the pandemic (March 1, 2020 to November 30, 2022). The outcome was firearm injury ED visits per 30 days for each of the time periods.

We described sociodemographic and clinical characteristics of the firearm injury visits during each time period using counts and percentages, analyzing differences using χ-square tests.

We calculated observed 30-day firearm injury visit rates for each time period. To estimate expected 30-day visit rates during the pandemic time period, we fit multivariable Poisson regression models to visit data from the prepandemic time period. A Pearson scale parameter was used to account for overdispersion. Poisson models were chosen in favor of negative binomial models because of better model fit estimated by log likelihood statistics. Models included the number of months since January 2017 to account for temporal trends. A more complex temporal effect, such as an autoregressive trend, was not considered because of an insufficient number of prepandemic observations. Covariates additionally included month and site to account for seasonal and geographic trends. The dependent variable was the rate of firearm injury visits per 30 days. We estimated 95% confidence intervals (CIs) for the 30-day firearm injury visit rates during the pandemic time period. Rate ratios (RRs) and 95% CIs of observed to expected visits were calculated by treating the number of observed visits as a fixed parameter and calculating the ratio between observed visits and the estimated visit rate and 95% confidence limits.

Individual models were constructed for each of the following sociodemographic characteristics: age, sex, race and ethnicity, insurance type, and COI quintile. Models for each characteristic included the same covariates as the overall model, with the addition of the characteristic of interest and an interaction with the number of months since 2017. The interaction term allowed for the modeling of temporal trends for each level of the characteristic, resulting in better-fitting models. RRs and 95% CIs of observed to expected visits during the postpandemic period were calculated for each level of the characteristic. Visits missing information about a particular characteristic (missing race and ethnicity, for example) were excluded from related models but included in other subgroup models when possible. One hospital had data quality issues for race and ethnicity, insurance type, and COI, so visits from that hospital were excluded from models using those characteristics.

Estimated visit rates and 95% CIs were plotted alongside observed visit rates for each calendar month for all firearm injury visits and by age group, sex, race and ethnicity, and COI quintile.

We used a significance level of .05, with no adjustments made for multiple comparisons. We performed all analyses using SAS/STAT software version 9.4 (SAS Institute Inc, Cary, NC, USA).

There were 1904 firearm injury ED visits by children during the study period (52.3% by adolescents 15–17 years old, 80.0% by males, 63.5% by non-Hispanic Black children; Table 1). Of firearm injury ED visits, 81.1% were by children with public insurance and 63.8% were by children from very low COI areas. There were 694 (36.4%) firearm injury ED visits before the COVID-19 pandemic (January 2017 to February 2020) and 1210 (63.6%) firearm injury ED visits during the pandemic (March 2020 to November 2022).

There were no statistically significant differences before versus during the pandemic for insurance payer distribution, COI distribution, intent of injury, injury severity score, body region, ED disposition, or hospital length of stay. Emergency severity index level 1 visits (representing the highest triage acuity) increased from 38.5% prepandemic to 46.4% during the pandemic (P = .027). Deaths increased from 3.1% of prepandemic visits to 6.1% of visits during the pandemic (P = .007).

Before the pandemic onset, 18.0 pediatric firearm injury ED visits occurred per 30 days. During the pandemic, firearm injury ED visits increased to 36.1 per 30 days, which was twice the expected rate based on extrapolated prepandemic trends, with an observed to expected RR of 2.09 (95% CI 1.63–2.91; Table 2, Fig 1).

During the pandemic, firearm injury ED visits per 30 days were higher than expected among children 10 to 14 years old (observed to expected RR 2.61, 95% CI 1.69–5.71) and 15 to 17 years old (RR 2.09, 95% CI 1.51–3.38; Fig 2). During the pandemic, firearm injury ED visits per 30 days were higher than expected for both female (RR 2.46, 95% CI 1.55–6.00) and male patients (RR 2.00, 95% CI 1.53–2.86). Firearm injury ED visits per 30 days were 2.30 times higher (95% CI 1.30–9.91) than expected for Hispanic children, 1.88 times higher (95% CI 1.34–3.10) than expected for Black, non-Hispanic children, and 2.21 times higher (95% CI 1.65–3.35) than expected for publicly insured children. Firearm injury ED visits were higher than expected by children from COI areas classified as very low (RR 2.30, 95% CI 1.64–3.86), low (RR 2.52, 95% CI 1.38–14.21), and high (RR 4.59, 95% CI 2.02–infinity; Fig 3).

Within a multicenter pediatric ED data registry, rates of firearm injury ED visits by children doubled during the COVID-19 pandemic, compared with expected levels based on prepandemic trends. Firearm injury ED visits increased relative to expected levels among specific demographic groups: children >10 years old, both males and females, and Hispanic and Black non-Hispanic children. Firearm injury visits by children from very low, low, and high COI areas increased compared with expected levels based on prepandemic trends.

Our results are consistent with other studies revealing a substantial increase in pediatric firearm injuries during the pandemic, while also contributing more recent data from 2022. During the first 6 months of the pandemic, pediatric firearm injuries at 9 Midwest trauma centers increased 87% above expected rates.²⁹  During the same time frame, Gun Violence Archive data revealed a 1.90-fold increase in firearm injuries among young children <12 years old.¹¹  During the first 21 months of the COVID-19 pandemic (through 2021), pediatric firearm injury presentations to US children’s hospitals increased 52% compared with previous years.⁶  The most recent Centers for Disease Control and Prevention data from 2022 revealed that firearm injury ED visits increased among children aged 0 to 14 years, with a visit ratio of 1.49 for females and 1.44 for males, compared with 2019.¹⁴  Our data suggest this increase was driven primarily by children >10 years because we did not find significantly greater than expected increases among younger age groups. In addition to changes in injury epidemiology, it is also possible that Emergency Medical Services transport patterns may have changed during the pandemic, with differential transport of older adolescents to pediatric rather than adult trauma centers.

The reasons for rising pediatric firearm injuries during the pandemic are multifactorial, often varying by intent. In 2020, US gun and ammunition sales surged,^11 ,30  in part due to worries about potential violence and lawlessness, with many first-time owners purchasing guns and existing owners switching to less secure manners of storage in response to the pandemic.^31,32  Additionally, among households that already owned guns before the pandemic, those storing firearms unsafely (ie, loaded and unlocked) were more likely to purchase additional firearms during the pandemic than households utilizing safe storage practices.³²  Access to guns stored unsafely in the home may have contributed to increases in self-inflicted and unintentional firearm injuries.^11,30  Simultaneously, psychosocial stressors and financial strains related to the pandemic may have led to increases in assault and self-inflicted firearm injuries.^7,33–35  Although we collected intent data and found no significant difference in intent by study period, we are cautious about interpreting this data because of the extent of miscoding of intent of firearm injuries in administrative datasets.^36,37  Emerging methodologies, such as natural language processing, may improve accuracy in the identification of intent from electronic health record data.³⁶  An improved understanding of injury patterns by intent may inform specific opportunities for prevention.³⁸ 

We found that more than one-half of all firearm injury visits were incurred by Black children, both before and during the pandemic. During the pandemic, Hispanic and Black non-Hispanic children experienced disproportionate increases in firearm injuries with significant increases above expected levels. The authors of some studies did not identify any change to the racial and ethnic breakdown of pediatric firearm injuries during the pandemic,^9,10  whereas others found postpandemic increases in the proportion of firearm injuries by Black children only.^6,13  When specifically focusing on firearm assaults, racial and ethnic disparities widened substantially during the pandemic in 4 major US cities. In particular, the Black-White disparity in firearm assaults grew from a relative risk of 27.5 before the pandemic to 100.7 during the pandemic, whereas the Hispanic-White disparity grew from an RR of 8.6 to 25.8.⁷  These data suggest that firearm injury prevention efforts should be centered around the most affected communities while targeting structural racism as a driver. Examples of such efforts may include neighborhood environmental interventions associated with violence reduction, such as cleaning and greening of vacant spaces,^39,40  and community investments that address social determinants of health, such as increased access to affordable housing and educational and employment opportunities.^38,41 

We found that 4 in 5 firearm injury ED visits were by publicly insured children, and nearly two-thirds were by children from very low COI areas, illustrating the high risk of firearm injury for children living in areas of concentrated disadvantage. During the pandemic, firearm injury visits increased across all payer types and COI quartiles, with significant increases beyond expected rates among publicly insured children and children living in very low, low, and high COI areas. However, the overall distribution of firearm injury ED visits by both payer and COI did not significantly differ before versus during the pandemic. This is consistent with a study of firearm-related encounters at US children’s hospitals from March to August 2020, which also revealed no change in the distribution of COI quartiles among firearm-injured children before versus during the pandemic.¹⁰  Similarly, a study conducted at 9 Midwest trauma centers during the first 6 months of the pandemic revealed no difference in the Social Vulnerability Index among firearm-injured children before versus during the pandemic.²⁹  The Social Vulnerability Index is a census tract-level composite index that incorporates socioeconomic status, household composition and disability, minority status and language, and housing and transportation.⁴²  Increases in firearm injuries across socioeconomic groups indicate that no child in the United States is immune to the growing risks of firearm violence.

We found increases in death rates for pediatric firearm injury visits during versus before the pandemic. In contrast, during the first 6 months of the pandemic, death rates did not change for pediatric firearm injuries at 9 Midwest trauma centers.²⁹  However, a single-center study contributing data through March 2022 also found that pediatric firearm deaths increased 29% during the pandemic, in tandem with increased admissions to the operating room and ICU.¹³  The pandemic placed strains on the capacity of Emergency Medical Services and prolonged response times,⁴³  resulting in delays that might have contributed to increased death rates. Although we did not identify changes in injury intent during the pandemic in our data, self-inflicted firearm injuries are known to have high case-fatality rates,⁴⁴  and other studies have identified increases in suicide attempt-related ED visits during the pandemic, particularly among adolescent girls.⁴⁵  Thus, strategies to reduce pediatric firearm fatalities must encompass suicide prevention efforts, such as increasing access to mental health services.⁴⁶ 

A comprehensive approach to preventing pediatric firearm injuries must be rooted in public health principles.⁴⁷  At the level of the individual clinician, counseling on safe storage and distribution of gun locks may modestly improve safe firearm storage practices.⁴⁸  At the health system level, hospital-based violence intervention programs are increasingly available to assist youth at risk for violence by providing linkages to community-based services and long-term case management.^49–51  Societal-level efforts must address social and structural determinants of health by bolstering community-based and economic support.⁵²  Additionally, a substantial investment in firearm injury prevention research is needed,^53,54  given that firearm injury research has historically been funded at only 3.3% of predicted levels based on mortality burden.⁵⁵  Research is also needed to understand the long-term behavioral and physical health consequences of firearm injuries on children and to design effective interventions to mitigate those impacts.¹⁵  Finally, policy solutions may also play a role in reducing pediatric firearm injuries.^56,57  Specific types of firearm legislation, such as universal background checks and child access prevention laws, have been associated with reduced pediatric firearm mortality.^58,59  During the first year of the pandemic, states with stronger gun laws saw decreased child-involved shooting incidents, whereas states with weaker gun laws experienced increased rates of pediatric firearm violence.¹²  Investments at all levels are needed to mitigate the substantial morbidity and mortality associated with increasing pediatric firearm injuries.

There are several limitations to be considered with our work, including miscoding of the intent of firearm injuries^36,37  and potential misclassification of race and ethnicity in electronic health record data. The data do not capture children who died at the scene and were not transported to the ED. Additionally, because of large sample sizes, some statistically significant differences may not be clinically significant. Because we were unable to clearly define the populations living in the catchment areas of the participating hospitals, we were unable to calculate population-based rates of firearm-related ED visits, overall or for specific demographic subgroups. Data were collected from children’s hospitals, potentially limiting the generalizability of results to non-pediatric trauma centers, in which most firearm-injured children in the United States receive emergency care.⁶⁰  We were unable to determine to what extent the increase in ED visit rates reflected an increase in firearm injuries occurring in communities, versus increased transport of firearm-injured children to children’s hospitals relative to non-children’s hospitals during the pandemic.

Firearm injury ED visit rates by children during the COVID-19 pandemic exceeded twice the rates predicted by prepandemic trends. Visit rates were higher than expected for Black and Hispanic children, widening injury disparities that preceded the pandemic. Although children from low-opportunity neighborhoods remain at the highest risk of firearm injuries, increases in injuries spanned multiple levels of neighborhood opportunity. Our results have relevance to the development of targeted injury prevention interventions and policy considerations. A comprehensive public health approach is needed to stem the rising tide of firearm injuries in children.

Members of the PECARN Registry Study Group include Drs Elizabeth Alpern, Lynn Babcock, Lalit Bajaj, David Brousseau, Jim Chamberlain, Bob Grundmeier, Prashant Mahajan, Bashar Shihabuddin, Leah Tzimenatos, and Joe Zorc.

We would like to acknowledge Sanjana Shankar for her contributions to literature review and manuscript preparation.

CI

confidence interval

COI

Child Opportunity Index

COVID-19

coronavirus disease 2019

ED

emergency department

PECARN

Pediatric Emergency Care Applied Research Network

RR

rate ratio