Injury Characteristics, Outcomes, and Health Care Services Use Associated With Nonfatal Injuries Sustained in Mass Shootings in the US, 2012-2019

Key Points Question What are the injury characteristics, outcomes, and health care services use among patients who have sustained nonfatal injuries in civilian public mass shootings? Findings In this case series of 403 patients who sustained nonfatal injuries in 13 consecutive mass shootings (defined as ≥10 individuals injured) from 31 hospitals in the US (2012-2019), 252 patients (62.5%) had firearm injuries, 147 (36.5%) were admitted to a hospital, 148 (40.7%) underwent emergency department procedures, 95 (23.6%) underwent 1 surgical procedure, and 42 (10.4%) underwent multiple surgical procedures. Meaning These findings suggest that the overall burden of mass shootings should not be limited to the number of deaths but should also incorporate nonfatal injuries, including those due to firearms and other trauma.


Introduction
Civilian public mass shootings (CPMSs) punctuate the firearm violence epidemic in the US and cause substantial numbers of deaths and injuries. Such shootings are the most common mass casualty events in the US, 1 and they are rising in frequency, 2 more than tripling from 2010 to 2019 compared with the previous decade. 3 In 2021, there were nearly 700 CPMSs, 4 and they continue to increase despite the COVID-19 pandemic. 5 Mass shootings are a global problem, but the US claims 38% of the world's 50 most deadly mass shootings 1 and 31% of global perpetrators. 6 Nonfatal gunshot wounds (GSWs) account for most firearm injuries in the US, 7 yet most firearm violence studies focus on deaths, 8 including those about mass shootings. [9][10][11][12] Every day, more than 230 people sustain a nonfatal GSW in the US, or 1 every 7 minutes. 8 The clinical importance of nonfatal GSW injuries by assault is amplified because most fatalities (61.2%) are suicides and most deaths (76.6%) occur outside the hospital. 7 For every firearm-related fatality of all types in the US (not just those from CPMSs), 2.5 injured individuals are treated for nonlethal GSWs. 7 Historically, for mass shootings alone, 1.5 to 1.6 patients sustain nonfatal GSWs for every death. 3,13 To our knowledge, no study has comprehensively described nonfatal injuries (GSWs and non-GSWs) in CPMSs across multiple sites, and no study has described non-GSW injuries for individuals injured in CPMSs. Previous studies 14,15 have reported only injuries from trauma registries, without an accounting of nontrauma activations, patients treated and released from the emergency department (ED), and noninjured patients from CPMSs. We herein report the injury characteristics, outcomes, and resource use of individuals who survive CPMS.

Methods
In this case series, we identified 21 CPMSs with injuries from July 20, 2012, to August 31, 2019-15 from public databases and 6 from media and site investigators-based on the Congressional Research Service criteria (public setting, civilians injured indiscriminately, motive not for criminal or other gain). 16 Although there have been more CPMSs in this period with fewer injuries, we focused on these because they had 10 or more injuries per shooting (treated at 53 receiving hospitals). We contacted local physicians in trauma and emergency medicine to participate. Data were available for 13 of 21 shootings (61.9%) and 31 of 53 primary recipient hospitals (58.5%). Sites, injured individuals, and exclusions are listed in the Figure. This study was classified as nonhuman participant research and did not require institutional review board (IRB) approval at the central hub (University of California, Irvine). Each data site obtained IRB approval. Individual patient consent was not required or obtained owing to use of deidentified data. This study followed the reporting guideline for case series.
From the public database-identified CPMS, we identified nearby hospitals that were likely to have received injured individuals. The senior authors (M.P.C., C.K.K., and M.I.L.) then contacted ED or trauma directors from their professional networks, and they identified injured individuals they received and additional local hospitals that likely received them. We then contacted champions at those hospitals to find additional injured individuals. Investigators retrieved data from their electronic medical record system and trauma registry (if available) into REDCap, version 11.2.4 We used best-practice medical record abstraction methods. 19 We trained abstractors, made specific definitions of cases, defined all variables, used standardized abstract fields with specific definitions, used continuous sampling of all eligible patients, had strategies to deal with missing or conflicting data, and obtained IRB approval at each site. We did not test interrater reliability because core investigators could not remotely access the data at the multiple sites to protect privacy. The abstractors were not blinded to the purpose of the study, but because there was no comparison of groups, there would be no potential for bias in data recording.
Because triage category was thought important for disaster planning, we extrapolated the missing ESI for the 171 patients based on diagnoses, admission data, services used, surgical procedures, ED procedures, and ISS. 20 Without initial vital signs, we assigned patients an ESI of 3 rather than 2 (lower acuity) when uncertain. Individuals with GSWs were assumed to have an ESI of at least 3, given the high-risk mechanism, potential for consultations, and high likelihood of laboratory testing and intravenous fluid resuscitation.
We collected hospital charges for the index hospitalization where available, as a surrogate for cost. We estimated cost from published ratios of costs to charges for US hospitals 21 and additional professional fees using published ratio estimates. 22 We converted costs to 2021 US dollars using inflation rates. 23 To report disabilities at discharge, investigators reviewed discharge summaries, operative notes, and diagnoses to record functional limitation(s) in use of the hand or arm, walking, cognition, or breathing as specified by a priori explicit definitions given to site investigators on how to determine if patients suffered from each of these disabilities. In addition, we assigned functional disability if patients had arm or leg fractures, because these would be splinted, even if not reported.
We assumed cognitive impairment for patients diagnosed with traumatic brain injury. Breathing impairment on discharge was assumed for diagnoses of lung contusion, diaphragm injury, multiple rib fractures, hemothorax, pneumothorax, acute respiratory failure, or acute pulmonary embolism.

Statistical Analysis
On completion of data entry, the deidentified data were aggregated and converted to Excel, version 15.45 (Microsoft Corporation), for analysis. This aggregation step precluded identification of individuals and ensured patient privacy.

Results
We included 403 patients from 13 CPMS (           Step

Discussion
Three other studies 14,15,24 have reported injuries from CPMSs in the US. In this largest case series to date, we report the injuries, ED and hospital resources used, and outcomes of 403 patients injured in 13 mass shootings in the US from 2012 to 2019. This study highlights the high ratio of injuries to deaths common to CPMSs, along with the burden of injury and associated use of health care services.
These estimates are higher than the previously reported ratios of 1 The proportion of publicly insured or uninsured patients in our study (64.3%) was the same as that for all firearm injuries. 27 Our distribution of CPMS ESI triage categories reflects substantially higher acuity than a US population treated in the ED for nondisaster incidents. 28 Previous work on all US

Limitations
This study has some limitations. We were unable to collect data from 8 CPMSs owing to lack of hospital research infrastructure, lack of staff owing to the COVID-19 pandemic, refusal of some health systems to allow physicians to participate in the project for public relations reasons, age of some medical records, and legacy electronic medical record systems. Without a list of the destination hospitals for individuals involved in CPMSs, our reliance on personal networks to identify recipient hospitals may mean that we missed data on patients at some community hospitals. Some patients, especially in Las Vegas, were not registered because electronic medical record systems during disasters often cannot keep up with volume and pace. 31,32 We were unable to validate data entry or to describe κ values owing to local IRB and privacy restrictions.
Hospital charges were unavailable for 55.3% of patients owing to changes in financial systems for older shootings. The ISS values were unavailable for 54.6% of patients because only hospitals with trauma registries could report this. We did not have access to long-term disability status for injured patients. We were unable to track patients' readmissions (if any) to other hospitals beyond the initial sites of index hospitalizations. We also did not gather data on the reasons for readmissions or subsequent surgical procedures or services involved. However, comments indicate that some of these were repeated orthopedic procedures, as would be expected. Our extrapolations of physical and cognitive disabilities from reported diagnoses could be challenged, although site investigators specifically reported patients with disabilities at discharge.
We did not gather data from CPMSs that had fewer than 10 injured individuals during this period, because this would have made IRB approval and data collection infeasible. Our report may not reflect the consequences among all individuals injured in CPMSs. Given the methodological limitations of a case series, future observational studies permitting statistical comparisons would be useful for further investigation of the injury characteristics and outcomes among individuals who survive CPMSs.

Conclusions
Mass shootings in the US cause enormous burden to patients, EDs, hospitals, and society at large.
Nearly 6-fold more individuals are injured than those who die. Although two-thirds sustain GSWs, one-third have other injuries. One-third require admission to a hospital and almost half are readmitted. More than one-third of patients with GSWs undergo surgery, and almost one-half have a disability at discharge. Hospital charges are substantial. These results can inform preparation and responses for prehospital, ED, and hospital care. Given the limitations of our data collection, we recommend establishing a national data registry that addresses the consequences of mass shootings.