Firearm Injuries in Lower Opportunity Neighborhoods During the COVID Pandemic

A multicenter, retrospective, cross-sectional study was conducted on firearm injuries involving children aged <18 years in LA County. Once institutional review board approval was granted, data from the LA County Trauma and Emergency Medicine Information System (TEMIS) Registry were abstracted from January 1, 2018, to December 31, 2021. TEMIS contains demographic and clinical data from trauma activations at 15 LA County trauma centers. Data in TEMIS were abstracted from the electronic health record at each trauma center by trained trauma registrars utilizing a standardized manual of operations.

March 19, 2020, was the date LA County’s stay-at-home order was issued, and this date was chosen as the time point defining our two cohorts: Prepandemic (January 1, 2018–March 19, 2020) and during pandemic (March 20, 2020–December 31, 2021). Demographic data, mechanism of firearm injury, Injury Severity Score (ISS), discharge capacity, and quarterly rate of firearm injuries were included. Encounters that were missing data for sex, race, or home zip code were excluded from analysis (Fig 1).

Demographic data included were age, sex, and race/ethnicity. Age was divided into three categories: 0 to 4 years old, 5 to 14 years old, and 15 to 17 years old. We grouped ages 15 to 17 years into one category because these patients are triaged as adults in the LA County emergency medical services system. Mechanism of firearm injury was further classified on the basis of intent, including assault, intentional self-inflicted, accidental discharge, and intent undetermined. ISS is calculated in TEMIS and reported as a numeric value that classifies the degree of injury by evaluating dysfunction of multiple organ systems.²²  We categorized ISS into one of the following standard classifications: mild (ISS <9), moderate (ISS 9–15), severe (ISS 16–24), and profound (ISS ≥25).

Patient discharge capacity was classified as deceased, permanent handicap (limitations from the injury expected to last >1 year), temporary handicap (required admission to the hospital for injuries sustained), and preinjury capacity (discharged from the ED with minimal or no injury) as determined by trained trauma registrars and reported in TEMIS. Mortality rates within each age group were determined.

The rate of firearm injuries was categorized into yearly quarters on the basis of the date of injury. Each quarter included three months grouped as follows: January to March, April to June, July to September, and October to December. The first quarter of 2020 was adjusted to end on March 19 and the second quarter beginning on March 20 to reflect the two cohorts.

COI was designated by home zip code using the COI 2.0 database.²³  COI is a continuous composite index of neighborhood opportunity, encompassing 29 variables relevant to healthy development in children. These variables are categorized into three domains: education, health and environment, and social and economic. A weighted scale is used to average each domain to establish an overall COI score that ranges from 0 to 100. Categorical child opportunity levels are then created by ranking all US census tracts by overall average scores and dividing them into quintiles (very low, low, moderate, high, and very high).²⁴  COI 2.0, typically measured at the census tract level, was mapped to the zip code level by the creators of COI and used in our study.²³ 

The primary outcome of interest was firearm injury volume as defined as the rate of injuries per quarter. Secondary outcomes included mechanism of firearm injury, ISS, discharge capacity, and COI.

Continuous variables were described as medians with interquartile ranges. Categorical variables were described by frequencies and proportions and analyzed by χ² tests. Nonparametric tests were used because our data did not follow a normal distribution.

The rate of firearm injuries during prepandemic and during-pandemic periods were examined as a rate per 100 000 children on the basis of the 2021 US estimated population for LA County.²⁵  The rate of firearm injuries was then compared with interrupted time series (ITS) analysis using the autoregressive integrated moving average model to account for the observed seasonality. ITS modeling was selected because it is a robust method for evaluating interventions or critical events at a population level, accounting for underlying trends, autocorrelation, and seasonality.²⁶  The impact of the stay-at-home order was evaluated by the change in the level (immediate step change) and the trend of the quarterly rates of firearm injuries.

We adhered to the Strengthening the Reporting of Observational Studies in Epidemiology guideline.²⁷  All analyses were conducted with 2-sided significance, α = .05. Analysis was done using SAS software 9.4 (SAS Institute, Inc, Cary, NC).

A total of 7693 trauma activations occurred during the study period, of which 544 (7.1%) were for firearm injuries. After exclusion (Fig 1), 530 (6.9% of all trauma activations; 97.4% of firearm injuries) remained, with 260 in the prepandemic period (26.6 months) and 270 (21.4 months) in the during-pandemic period. There were not any children aged <1 year with firearm injuries during the study time frame. The majority of injuries occurred in males of late teenage years (Table 1). Age, sex, and race/ethnicity were not different between the two periods (P = .79, P = .32, P = .76). There was no difference in mechanism of firearm injuries between the two periods. Although not different between the two periods, both periods had 50% or more of the cohort that was classified as mild on the basis of ISS.

Although the rate of firearm injuries fluctuated during the study periods, there was not a statistically significant increase in the overall rate of injuries between the two study periods after the stay-at-home order (P = .13) (Fig 2). We found the rate of firearm injuries within COI quintiles significantly changed, with a disproportionate increase in those from very low COI neighborhoods (P = .01) during the pandemic (Fig 3).

Injury characteristics were evaluated, stratified by each COI quintile (Table 2). Across all COI quintiles during the pandemic, there were fewer or no change in the number of fatalities from firearm injuries: very low, 17 to 12; low, 8 to 4; moderate, 4 to 2; and high, 2 to 2. Also, fewer injuries were classified as profound on the basis of ISS: very low, 27 to 19; low, 11 to 9; moderate, 8 to 2; high, 5 to 4; and very high, 1 to 0. During the pandemic, 14% of firearm injuries in high COI neighborhoods resulted from intentional self-inflicted injuries, whereas self-inflicted injuries only made up 4% of firearm injuries during the prepandemic period.

The overall trend of firearm injuries within a large metropolitan area did not significantly increase after the stay-at-home order issued in response to the pandemic. Comparison between the prepandemic and during-pandemic periods did not reveal differences in patient demographics, mechanism of injury, injury severity, or patient discharge capacity. We identified a disproportionate increase in firearm injuries in patients residing in neighborhoods of very low COI, but no differences were observed in other COI quintiles.

After the pandemic, the rate of firearm injuries increased on a national scale.^7,28,29  In contrast, reports from regional- and state-level studies found both unchanged rates and increased rates of firearm injuries during the pandemic.^10,30  Our findings, in a large region, revealed no increase during the pandemic. Despite the contrast in the rates of firearm injuries, there is consistency in the distribution of demographics, with a higher incidence of injuries in teenagers, males, and because of assaults.^18,28,31,32  Our study demographics aligned with the findings of those previous studies, which did not change in response to the pandemic.

In agreement with current literature, our findings further support the growing knowledge of the substantial influence community environments have on children.^12–14  Neighborhoods have traditionally been measured on the level of disadvantage, with focus on vulnerability and factors that are not necessarily specific to children.³³  Overall, COI was developed to assess child-focused factors, providing several distinct features separating it from others’ indices. One unique feature of COI is the educational domain, which encompasses the level of access to education across childhood ages and the quality of these educational opportunities. Further distinguishing COI, the health and environment domain extends beyond a child’s home, including access to healthy food, access to green space, and neighborhood walkability.²³ 

Children in neighborhoods with lower COI are at a disadvantage because of limited availability of resource and opportunities in domains that are vital for children.²¹  Before the pandemic, communities of lower socioeconomic status experienced firearm injuries at higher rates,^19,34,35  although these community assessments were not based upon COI. Only recently has COI been incorporated in studies of traumatically injured pediatric populations, because we only identified one previous study that used COI in association with firearm injuries in children.³⁶  This study from Kwon et al reports that children of lower COI neighborhoods were more likely to be evaluated for firearm injuries than children of very high COI neighborhoods.³⁶  Although consistent with our findings, this study does not assess the consequence of the pandemic as reflected in our study. Our findings suggest that the pandemic resulted in further hardship for this population as businesses, schools, and care facilities closed. Resulting from these closures, many people were likely left with suboptimal child care options.

Ultimately, a combination of potential factors likely contributed to the increased rate of firearm injuries isolated to children of very low COI neighborhoods. The increase in firearm purchasing^3,37  and unsafe storage^3,5  during the pandemic led to an increase in firearms in households with children. In the context of an unprecedented hardship on neighborhoods with limited resources, the increase in childhood exposure to firearms in lower COI quintiles may be attributable to the higher frequency of single-headed households in lower COI neighborhoods.²³  A single source of household income combined with broad closures of schools and early education centers may have resulted in increased unsupervised time of children during the pandemic. Thus, our findings highlight an important disparity in childhood risk for firearm injuries on the basis of neighborhood opportunities that would otherwise be missed if rates were not stratified by measures such as COI.

The solution to this national crisis is a substantial undertaking. Our results identify a population of children who may encounter increased adversity during times of crisis. Understanding that a patient’s risk of experiencing a firearm injury during times of crisis is associated with available neighborhood resources can enrich provider–family conversations about firearms. Additionally, this knowledge allows health care providers to emphasize the importance of firearm safety to parents, particularly in regard to children.

Our study used a large clinical trauma database that allowed for examination of trends but did not permit for more granular patient-level characteristics. Because our investigation only pertained to firearm injuries, it is not a good assessment of suicide rates during this time frame. We recognize that our analysis of the trend in rates of pediatric firearm injuries without a control group does not account for other events that may have influenced the outcome. However, the use of the prepandemic cohort as the control and the use of an ITS are a methodologically acceptable design for studying effects of an event. Additionally, this database includes only trauma activations captured by hospital trauma registrars and may not include some patients who were not documented as such or those who did not present to the ED for evaluation, such as patients who died at the scene. The COI performs better at the level of census tract, so it is possible there is misclassification of COI at the zip code level. Furthermore, the performance of this study on a regional level is also a potential limitation. Nonetheless, the differential association found between neighborhood opportunity and rates of firearm injuries remains relevant nationally, particularly given the size of our sample population. Despite the limitations of this study, the availability of data related to intent of injury and home zip codes allowed us to understand the neighborhoods that were most affected.

Although there was no overall increase in pediatric firearm injuries during the pandemic, there was a disproportionate increase among children from very low COI neighborhoods. Our findings highlight that the pandemic was associated with further hardship to this population that may be related to availability of community-level resources. Further examination of this community disparity should be a focus for future education, intervention, and development.

COI

Child Opportunity Index

ED

emergency department

ISS

Injury Severity Score

ITS

Interrupted Time Series

LA

Los Angeles

TEMIS

Trauma and Emergency Medicine Information System