Error bars indicate 95% confidence intervals. Gunshot wound was the only mechanism with a significant increase in case-fatality rates and a significant decrease in TRISS-derived survival probabilities. A, Unadjusted rates are the solid lines and adjusted rates are the dashed lines. B, Dashed lines represent simple linear trends.
For each mechanism of injury, the relative risk comparator is all other mechanisms of injury shown in the graphs. A, After adjustment for age and sex, the risk of having severe injuries over time significantly increased for gunshot wounds (6.5% per biennium) and stabbings (6.0% per biennium). B, After adjustment for age and sex, gunshot wounds was the only mechanism for which the risk of having more than 2 severe injuries (AIS score >2) significantly increased over time (4.8% per biennium). Linear trends shown only for selected mechanisms.
Sauaia A, Gonzalez E, Moore HB, Bol K, Moore EE. Fatality and Severity of Firearm Injuries in a Denver Trauma Center, 2000-2013. JAMA. 2016;315(22):2465-2467. doi:10.1001/jama.2016.5978
Death rates provide an incomplete picture of the effect of firearm injuries. To devise appropriate prevention efforts, investigations of the severity and prognosis of both fatal and nonfatal gunshot wounds (GSW) are pivotal, yet they remain scarce.1- 3 We studied temporal patterns of GSW-associated severity and mortality in a Colorado urban trauma center and of all trauma deaths occurring in its catchment area from 2000 to 2013.
We queried the state-mandated trauma registry of a level 1 trauma center (Denver Health Medical Center, DHMC) for data on injuries, cause, and severity for all patients who died in the hospital, were hospitalized, or required more than 12-hour observation from 2000 to 2013. Throughout this period, the DHMC catchment area was Denver County. To assess injury deaths at the scene (vs in-hospital), we obtained all Denver County records of trauma deaths during the same period. The Colorado Multiple Institutional Review Board approved the study with a waiver of consent.
Injury severity was quantified by the Injury Severity Score (ISS; range, 1-75), a score for multiple injuries.4 Each single injury was assigned an Abbreviated Injury Scale (AIS; 1 = minor injury to 6 = lethal injury) score, with severe defined as a score higher than 2. The maximum AIS score from the 3 most severely injured body regions was squared and added to produce the ISS. Patients with ISS higher than 25 were considered severely injured.
Poisson regression models with robust standard errors were used to estimate relative risks (RR) and risk differences (RD) of in-hospital mortality, an ISS higher than 25, and having more than 2 severe injuries (AIS score >2), by mechanism and biennium (to produce more stable estimates), adjusting for age and sex. Temporal trends by injury mechanism were assessed by including an interaction between mechanism and biennium. Linear regression was used to model temporal trends of proportion of deaths at the scene and survival probabilities derived through the Trauma and Injury Severity Score (TRISS), a benchmarking tool widely used to estimate trauma survival probabilities based on age, mechanism, admission vital signs, and injury severity.5
There were less than 6% of missing values. SAS (SAS Institute), version 9.4, was used for all analyses. All tests were 2-tailed with significance set at less than .05. The institutional review board approved the study with a waiver of consent.
From 2000 to 2013, 28 948 patients presented to the DHMC with injuries due to GSWs (5.8%), stabbings (6.3%), pedestrian accidents (6.9%), assaults (8.7%), falls (23.9%), motor vehicle crashes (26.2%), and other mechanisms (22.1%). Of these, 5.4% died. The proportions of DHMC injury admissions due to GSWs, stabbings, and assaults remained stable from 2000 to 2013, whereas falls increased (from 16.8% to 27.8%) and motor vehicle crashes decreased (from 37.0% to 19.7%) over time. Adjusted in-hospital case-fatality rates for GSWs at the DHMC significantly increased (RR per biennium, 1.06 [95% CI, 1.03-1.08]; RD, 0.51% [95% CI, 0.01%-1.02%]) (Figure 1A) and the TRISS-derived survival probabilities decreased (P = .002) (Figure 1B). All other mechanisms presented stable or opposite temporal trends for deaths and survival probability. Over time, more GSW patients had an ISS higher than 25 (RR per biennium, 1.06 [95% CI, 1.04-1.08]; RD, 1.16% [95% CI, 0.68%-1.65%]). In addition, the number of severe GSWs per patient increased significantly (RR per biennium of >2 severe injuries, 1.04 [95% CI, 1.02-1.06]; RD, 1.22% [95% CI, 0.66%-1.79%]) (Figure 2).
In Denver County during this same period, there were 4762 trauma deaths (falls, 28.4%; GSWs, 22.9%; motor vehicle crashes, 19.9%; pedestrian accidents, 4.0%; stabbings, 2.4%). Of these, 64.1% occurred in-hospital (50.9% at the DHMC) and 35.9% were pronounced outside a health care facility. Of the 1092 GSW deaths, 47.3% were in-hospital (81% at the DHMC), and this proportion did not vary significantly over time (P = .62). Only for falls was a decrease in the proportion of in-hospital deaths detected (from 81.0% to 41.6%, P = .006).
Firearm in-hospital case-fatality rates increased, contrary to every other trauma mechanism, attributable to the rising severity and number of injuries. The differential in severity and mortality is unlikely due to improved emergency medical services (ie, more severely injured patients arriving alive to the hospital vs dying in the field), as there were no changes in deaths at the scene over time. This single trauma center study has limited generalizability. A renewed attention to research and policy are needed to decrease the morbidity and mortality of GSWs.6
Corresponding Author: Angela Sauaia, MD, PhD, University of Colorado Denver, 13011 E 17th Pl, Room E-3360-C, Aurora, CO 80045 (email@example.com).
Author Contributions: Dr Sauaia had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Sauaia, Gonzalez, E. Moore.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Sauaia, E. Moore.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Sauaia, H. Moore, Bol.
Administrative, technical, or material support: Sauaia, Gonzalez, E. Moore.
Study supervision: Sauaia, E. Moore.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.