Interpersonal violence is a leading cause of both fatal (Centers for Disease Control and Prevention (CDC), 2024a) and nonfatal (Centers for Disease Control and Prevention (CDC), 2024b) injuries among children and young adults in the United States. Firearm injuries, specifically, disproportionately affects younger Americans, both with regard to fatal and nonfatal injuries (Goldstick et al., 2021; Goldstick et al., 2019a; Kaufman et al., 2021). Moreover, violence among children and young adults is a critical driver of health disparities, with racial and sexual minorities showing substantially higher rates of violence injuries (Centers for Disease Control and Prevention (CDC), 2024c). Those trends extend to firearm violence as well with, for example, Black and Hispanic White Americans showing substantially higher rates of firearm deaths than their Non-Hispanic White counterparts (Rees et al., 2022; Vargas et al., 2023). These disparities reflect broader structural inequalities and non-uniform distribution of social determinants of health (Degli Esposti et al., 2022; Kim, 2019). These facts collectively point to a critical need to leverage all opportunities to reduce youth violence, and firearm violence specifically, and the correspondent health disparities.

Urban emergency departments (EDs) are an important touch point for young people at high risk for violence. Adolescents and young adults entering the ED consistently show higher rates of violence than the general population; for example, the proportion of high school youth reporting past-year fighting has decreased (e.g., 33.2 % in 2001; 18.3 % in 2021) (Centers for Disease Control and Prevention (CDC), 1991–2021), but up to three quarters of adolescents entering the ED report peer violence (Walton et al., 2009), including over half of those not presenting for assault-related injuries (Bohnert et al., 2015). Furthermore, young people presenting to the ED for violence injuries are at substantial risk for violent re-injury, with over one-third returning within two years, and nearly one-fourth of those presenting for other reasons returned within a year (Cunningham et al., 2015). Not only are adolescents and young adults presenting to the ED at elevated risk for violence, but the visit itself may represent an opportunity to catalyze behavior change. Prior research supports this supposition, with previous successes of ED-based interventions for violence (Walton et al., 2010; Cunningham et al., 2012; Cunningham et al., 2013), and health behaviors such as alcohol use (Barata et al., 2017; Gentilello et al., 2005), and other drug use (Bogenschutz et al., 2014; Madras et al., 2009).

While ED-based violence epidemiology has been studied previously (Cunningham et al., 2015; Carter et al., 2015; Cunningham et al., 2009; Wiebe et al., 2011; Ranney et al., 2016), violence dynamics in the U.S. have shifted in recent years—including firearm violence specifically. Violence mortality has sharply increased in recent years, with rates increasing 23.6 % from 2014 to 2021, driven largely by increases in homicides, which have increased 58.5 % during that time (Centers for Disease Control and Prevention (CDC), 2024d). Although violence mortality rates are uniformly higher across most age groups in 2021 (e. g., relative to 2014), these changes are largest among young people (Centers for Disease Control and Prevention (CDC), 2024d). Trends in firearm violence have been analogous, but more pronounced, increasing 41.7 % from 2014 to 2021, with a correspondent increase of 83.9 % among firearm homicides (Centers for Disease Control and Prevention (CDC), 2024d). Similarly, increases in firearm deaths are most pronounced among young people, as more gravely reflected by the fact that firearm deaths recently overtook motor vehicle crash deaths as the leading cause of death among children and young adults (Goldstick et al., 2022). Taken together, this suggests it is critical to update violence rates—and associated factors—among young people presenting to the ED.

In this report, we describe baseline results from the Screening to Predict Young Adults at Risk for Firearm Violence (SPARK) project, a multi-site prospective longitudinal study of young adults entering the ED that aims to externally validate and improve the SaFETy (Serious fighting, Friend weapon carrying, community Environment, and firearm Threats) score (Goldstick et al., 2017), a four-item screen tool derived to predict future firearm violence risk in the ED. We began by describing rates of different types of violence, substance use, and mental health symptoms. Next, we estimated how sociodemographic, community, and substance use/mental health factors were associated with recent firearm assault experience (shot or threatened, as victim or aggressor). We hypothesized that violence rates among a general ED sample of young adults would be high, and that firearm assault is associated with a number of psychosocial factors measured in the baseline survey.

The SPARK recruitment flow is shown in Fig. 1. Among those approached, 25.4 % refused study participation prior receiving the baseline survey. Those refusing prior to baseline were 44.4 % male, and had an average age of 21.4 (SD = 1.9). In total, 1506 young adults enrolled in the study and completed the baseline survey—649 in Seattle (one ED), 318 in Flint (one ED), and 539 in Philadelphia (two EDs). In total, 61.4 % of those completing the baseline survey self-reported as female, 35.9 % male, and 2.7 % non-conforming. Regarding race, the sample is predominantly Black (41.7 %) or White (33.6 %), followed by Multiracial (9.8 %) and Asian (7.2 %). About one-third of the sample was in school. There are differences across sites in these demographics, e.g., Seattle showed a greater proportion male, lower proportion Black, and higher proportion Asian and Multiracial; full demographics by site are in Table 1. Nearly 1-in-5 reported keeping in firearm in their car and/or household in their lifetime, while 1-in-9 reported doing so within the past six months; over 1-in-10 reported past-six-month firearm carriage (Table 1). Of note, among those reporting past-six-month firearm carriage, 35.0 % reporting doing so 20+ times; 68.7 % reported carrying at least “3–5 times”).

We used data from a comprehensive baseline survey to describe violence rates, and other service needs (e.g., substance use, mental health) among a general population of young adults entering EDs in three cities, and examined factors associated with firearm assault. These results update previous ED-based studies of violence epidemiology against a backdrop of increasing national rates of violent injury, particularly firearm violence. Half of surveyed young adults presenting to urban EDs reported recent violence experience, and over one-seventh report recent firearm assault. Risky substance use and mental health symptoms were also commonly reported, across sites, underscoring a general ED sample of young adults as a high-need population. These findings point to the ongoing injury prevention opportunity in urban EDs; there are effective ED-based violence prevention programs (Carter et al., 2022; Carter et al., 2016) and these efforts may be informed by these contemporary results.

Previously published work from Project SPARK showed strong cross-sectional associations between the SaFETy score and self-reported firearm violence (Goldstick et al., 2024), and the results presented here suggest there are other important factors associated with firearm assault experience. Several factors not included in the original SaFETy score (Goldstick et al., 2017)—such as peer/parental support, family conflict and ACEs, retaliatory and prosocial attitudes, and mental health symptoms—showed strong associations with firearm assault in this sample. As standalone results, these associations point to the importance of social exposures and attitudes/norms in modulating firearm assault, which can inform interventions. In addition, they point to high-need subpopulations, such as those with more ACEs, greater family conflict, and drug use/mental health comorbidities.

While prior studies measuring firearm violence and a large set of sociodemographic factors are limited, these results do affirm prior results. In particular, associations between firearm violence and retaliatory attitudes, PTSD, firearm carriage, and drug use are consistent with prior work (Carter et al., 2015). Given the previously discussed national increases in firearm violence rates, particularly among children and young adults, results confirming similarities to dynamics observed in a sample from a decade earlier (Carter et al., 2015) provide important updates to the literature. At the same time, we presented associations not shown in previous ED-based studies of firearm violence, such as those between self-reported firearm assault and family conflict, parental support, positive peer behaviors, and prosocial attitudes. Taken together, these results provide important contemporary context for future ED-based studies of firearm violence.

Replication of the previous machine-learning-based longitudinal analyses that derived the score in a general sample of young adults entering the ED, as opposed to the higher-risk sample in which the score was derived, and with a broader set of self-reported measures, is a critical next step for Project SPARK. Our baseline results confirm the importance of those next steps, and suggest optimism for their success. In particular, the high rates of violence—including firearm assault experience—show that general ED samples of young adults continue to be a population at elevated risk for violence. Thus, prospective ascertainment risk is critically important, particularly among those not presenting for reasons that suggest elevated violence risk (i.e., violent injury). This critical need is underscored by the fact that the SaFETy score is being measured in clinical sites already (Goyal et al., 2024), even without existing external validation. As described above, the strong associations identified here suggest, in combination with previously reported associations between the SaFETy score and self-reported firearm violence (Goldstick et al., 2024), there may be room for improved violence prediction. The forthcoming prospective validation will attempt to fill this critical gap.

Outside of firearm violence, our results point to other significant service needs of young adults entering the ED in all three cities. Sizable proportions reported severe violence involvement (both aggression and victimization), and about 1-in-10 report recent firearm carriage; this combination points to potential escalation of violence severity and critical opportunities for firearm safety counseling, particularly given previous research showing that firearm violence is often preceded by lower-level violence (Goldstick et al., 2019b). High rates of substance use and positive screen for PTSD, depression, and anxiety were also observed; these rates are substantially higher than corresponding rates in the general population. For example, 38.2 % of those age 18–25 reported past-year cannabis use in 2022 (Substance Abuse and Mental Health Services Administration (SAMHSA), 2024); we found nearly half of the sample reporting risky use—a higher threshold than any past-year use. Similarly, prevalence of PTSD among young adults is under 5 % (National Institute of Mental Health (NIMH), 2024), but nearly one-third of this sample screened positive (PCL ≥ 30) for possible PTSD, suggesting substantially higher rates. The fact that these other services needs are all strongly associated with violence involvement in this sample further points to important prevention priorities among a general ED sample of young adults.

We acknowledge some limitations of this work. Importantly, all of the data presented here are self-reported, uncorroborated by objective sources. Our study did include an NIH certificate of confidentiality, but there is a possibility of misreporting, particularly with regard to higher risk behaviors. Second, relatedly, participants had the option of skipping questions. We felt this was an important human subjects protection, given the sensitive nature of some questions, but does raise the possibility of underreporting on the highest risk behaviors (e.g., ~1.7% missing on the firearm assault outcome, which may be a higher-risk subset). Third, individuals refusing study participation prior to receiving the baseline survey were more likely to be men than those who did complete the baseline survey, raising minor concerns about selection bias. Fourth, the wording of the question related to firearms in ones household/car does not preclude that the firearm was owned and primarily handled by another household member; nevertheless we feel this question serves as an important proxy for firearm access in ones’ home/car. Finally, the cross-sectional nature of this analysis undermines arguments about predictive power and whether the covariates preceded the outcomes. Nevertheless, these results comport with a priori expectations (e.g., showing strong convergent validity), and are an important next step in understanding the worsening problem of firearm violence in the US. Longitudinal results from Project SPARK will be critical to validating these results.

Violence—including severe violence, firearm violence, partner violence, and non-partner violence—is common among a general ED sample of young adults, and those with elevated violence risk also have other service needs. In addition, familial, peer, and community factors play strong roles in determining firearm assault risk. Taken together, this suggests it is possible to characterize those at highest risk, and leverage that information to inform intervention development, and efficiently disseminate resources.