Adult Firearm-Related Traumatic Brain Injury in United States Trauma Centers

Abstract

Civilian firearm injury is an important public health concern in the United States. Gunshot wounds to the head (GSWH) remain in need of update and systematic characterization. We identify predictors of prolonged hospital length of stay (HLOS), intensive care unit length of stay (ICU LOS), medical complications, mortality, and discharge disposition from a population-based sample using the National Sample Program (NSP) of the National Trauma Data Bank (NTDB), years 2003–2012. Statistical significance was assessed at α < 0.001 to correct for multiple comparisons. In total, 8148 adult GSWH patients were included extrapolating to 32,439 national incidents. Age was 36.6 ± 16.4 years and 64.4% were severe traumatic brain injury (TBI; Glasgow Coma Scale [GCS] score 3–8). Assault (49.2%), handgun (50.3%), and residential injury (43.2%) were of highest incidence. HLOS and ICU LOS were 7.7 ± 14.2 and 5.7 ± 13.4 days, respectively. Overall mortality was 54.6%; suicide/self-injury was associated with the highest mortality rate (71.6%). GCS, Injury Severity Score, and hypotension were significant predictors for outcomes overall. Medicare/Medicaid patients had longer HLOS compared to private/commercial insured (mean increase, 4.4 days; 95% confidence interval [2.6–6.3]). Compared to the Midwest, the South had longer HLOS (mean increase, 3.7 days; [2.0–5.4]) and higher odds of complications (odds ratio [OR], 1.7 [1.4–2.0]); the West had lower odds of complications (OR, 0.6; [0.5–0.7]). Versus handgun, shotgun (OR, 0.3; [0.2–0.4]) and hunting rifle (OR, 0.5; [0.4–0.8]) resulted in lower mortality. Patients with government/other insurance had higher odds of discharging home compared to private/commercially insured (OR, 1.7; [1.3–2.3]). In comparison to level I trauma centers, level II trauma centers had lower odds of discharge to home (OR, 0.7; [0.5–0.8]). Our results support hypotension, injury severity, injury intent, firearm type, and U.S. geographical location as important prognostic variables in firearm-related TBI. Improved understanding of civilian GSWH is critical to promoting increased awareness of firearm injuries as a public health concern and reducing its debilitating injury burden to patients, families, and healthcare systems.

Introduction

Traumatic brain injury (TBI) is a major cause of public health and socioeconomic burden, with estimates by the U.S. Centers for Disease Control and Prevention at 2.5 million annual emergency department (ED) visits and 52,000 deaths.^1,2 Penetrating brain injury (PBI) accounts for 1.5% of TBI incidents and is the most lethal (42% of TBI deaths), in part because the majority of civilian cases are from high-velocity projectiles, for example, gunshot wounds to the head (GSWH).^3,4 Because of the highly destructive nature of ballistics to brain tissue, past studies on GSWH report the exceedingly high early-mortality rate of 71% on scene as well as an overall mortality rate of 88%. Roughly 97% of GSWH patients experience unfavorable outcome at hospital discharge (Glasgow Outcome Scale [GOS] = dead [GOS = 1], vegetative [GOS = 2], or severe disability [GOS = 3]).^5

–9

For patients who survive long enough to receive medical management, GSWH is a complex challenge and there is little literature describing the natural history of these injuries in a modern medical setting. Past investigations have suffered from small sample sizes, nonstandardized treatment algorithms, and incomplete data. As a result, there is a lack of robust substantiation for improved mortality subsequent to publication of the 2001 Guidelines for the Management of Penetrating Brain Injury.¹⁰ To date, more civilians have died from gun violence since World War II than were casualties of the Korean and Vietnam conflicts combined,⁸ yet the current body of research on PBI is mostly noncivilian. Put succinctly, the standardized guidelines include immediate restoration of circulating volume, blood pressure, oxygenation, and ventilation, aggressive antibiotic and seizure prophylaxis, and less-aggressive surgical debridement for deep-seated fragments.^11,12 Although these guidelines are widely utilized, overall patient outcomes have not been corroborated by large-scale studies. Given these factors and the increased scrutiny by the media on gun violence, an accurate characterization of GSWH on a national scale is critical.

Critical differences in setting, risk factors, and prognosis exist between the civilian and noncivilian GSWH. Civilian GSWH is typically caused by subsonic bullets, whereas noncivilian GSWH is largely a result of supersonic, military-grade bullets which transmit significantly more energy to brain tissue. As a result, noncivilian GSWH is most often fatal. The noncivilian PBI literature also has a high proportion of shell and shrapnel injuries, which are generally of lower velocity and irregularly shaped, thus imparting less kinetic energy and damage to brain tissue. This difference explains why, despite an average injury-to-hospital time of less than 45 min in the civilian setting, shell and shrapnel PBI common in the military arena has mortality rates of 10–30%, even with longer mean evacuation time of 2.3 h from battlefield to medical care.^13
–15 The impact of the weapon caliber in the civilian GSWH population remains to be analyzed to determine the significance of ballistics on functional outcome.

Predictors of outcome in civilian GSWH have been an important, and timely, topic of recent research. Consensus on outcome determinants includes age, hypotension, suicide attempt, bullet trajectory, and lesions on brain computed tomography (CT) scan, Glasgow Coma Scale (GCS) score at admission, and abnormal pupillary response.^12,16,17 Other regional reports have found localized firearm-related injury patterns driven by market trends for pistols, semiautomatic weapons, and other forms of handguns.^18
–20 These fluctuations may be specific to demographics, socioeconomic status, and geographical region. To our knowledge, no standardized reports have characterized the details of civilian GSWH on a national scale to account for shifting demographics, inter-regional discrepancies, and evolving firearm technology. Epidemiological surveillance of firearm-related injuries at the turn of the century demonstrated the healthcare burden from firearms,^21

–24 but the 10–11% of patients presenting with GSWH insofar have yet to be examined on the granular level in the firearm literature. Given the lack of standardized national reporting of patients presenting with GSWH, we provide an epidemiological update of U.S. firearm TBI over the past decade. We hypothesize that clinical assessment during triage, including hypotension and injury severity measures, along with the injury intention, type of firearm, and geographical region, may be associated with worse outcomes post-GSWH. The aim of this study is to provide an update of this subset of patients in the United States.

We use the National Sample Program (NSP) of the National Trauma Data Bank (NTDB), a prospective, curated trauma registry with the purpose of tracking and providing information about trauma care and outcomes. We performed a retrospective analysis of demographic data, risk factors, management strategies, complications, disposition, and mortality in the PBI population specifically wounded by firearms who were treated nonoperatively and operatively. It is our objective to update the epidemiology and clinical history of U.S. firearm TBI over the past decade, as well as prognosis and outcome to inform acute management of this critically injured population. Notably, we characterize our patient sample by firearm cohort and U.S. geographical locale, with implications on management and outcome.

Methods

In this study, we used the NSP of the NTDB from arrival years 2003–2012. The NSP for each year consists of a stratified sample of 100 NTDB participating hospitals based on U.S. Census region, trauma care designation, and NTDB reporting status.²⁵ Hospitals were drawn by the NTDB from the sampling universe of 453 level I or II trauma centers, and the sample size of 100 hospitals was determined by past review indicating it can be extrapolated to represent the national patient distribution.²⁶ Detailed data qualification, selection, cleaning, and standardization algorithms have been previously reported.²⁶ Because the NSP of the NTDB is a fully de-identified data set without the 18 federal Health Insurance Portability and Accountability Act (HIPAA) identifiers, the current study was classified as exempt from institutional review board review.

A total of 410,101 incidents sustaining TBI were extracted from NSP years 2003–2012 using the International Classification of Diseases, Ninth Revision (ICD-9) codes 800-801.99, 803-804.99, and 850-854.19 as previously described.^27
–29 Adult patients (variable name “age” ≥18; n = 355,388) with known penetrating TBI (ICD-9 codes E922.0-922.9, E955.0-955.4, E965.0-065.4, E970, and E985.0-985.3 [Table 1]) and sex (variable name “gender”; n = 8148) were extracted for analysis (Fig. 1). Patients with nonmissing data in our five targeted outcomes variables (mortality, inpatient medical and surgical complications, hospital length of stay [HLOS] of patients alive at discharge, intensive care unit length of stay [ICU LOS] of patients alive at discharge, and discharge destination were included in the final sample).

FIG. 1.

Flow chart identifying firearm-related TBI patients for analysis from the NSP of the NTDB, 2003–2012. NSP, National Sample Program; NTDB, National Trauma Data Bank; TBI, traumatic brain injury.

Table 1.

Summary of Extracted Firearm-Related ICD-9 E-Codes

Firearm injury and code	ICD9 E-Code description	No. of incidents
Accident		286
E922.0	Caused by handgun	114
E922.1	Caused by shotgun	29
E922.2	Caused by hunting rifle	28
E922.3	Caused by military firearm	1
E922.8	Caused by other specified firearm missile	27
E922.9	Caused by unspecified firearm missile	87
Suicide and self-inflicted		3410
E955.0	Caused by handgun	2192
E955.1	Caused by shotgun	167
E955.2	Caused by hunting rifle	138
E955.3	Caused by military firearm	8
E955.4	Caused by other and unspecified firearm	905
Assault		4009
E965.0	Caused by handgun	1682
E965.1	Caused by shotgun	125
E965.2	Caused by hunting rifle	26
E965.3	Caused by military firearm	6
E965.4	Caused by other and unspecified firearm	2170
Legal intervention		139
E970	Caused by legal intervention by firearm unspecified	139
Undetermined whether accidentally or purposely inflicted		304
E985.0	Caused by handgun	93
E985.1	Caused by shotgun	14
E985.2	Caused by hunting rifle	7
E985.3	Caused by military firearm	2
E985.4	Caused by other and unspecified firearm	188

ICD-9, International Classification of Diseases, Ninth Revision.

Demographic and clinical variables of interest were extracted for multi-variable analysis to include age, sex, race, health insurance status, medical history, U.S. geographical region, site of incident, intention of injury, type of firearm, isolated TBI, American College of Surgeons (ACS) trauma center designation, hypotension in the ED (systolic blood pressure [SBP] <90 mm Hg), GCS score, stratified Injury Severity Score (ISS), and ED disposition. The Charlson Comorbidity Index (CCI) was utilized as a summative score of overall comorbidity burden, calculated using the standard comorbidity weights as previously described in large, retrospective TBI studies.^30

–34 Demographic and clinical variables that were missing and/or marked as not known/not recorded from the NSP were coded as “unknown.”

Statistical analysis

Descriptive variables are presented using proportions for categorical variables and means and standard errors (SEs) for continuous variables. Group differences were assessed using Pearson's chi-squared test (X ²) for categorical variables and analysis of variance for continuous variables. For analyses with individual cell counts <5, Fisher's exact test was used in place of X ². Categorical outcome variables (e.g., mortality, complications, and discharge destination) were assessed using binary logistic regression. Continuous outcome variables (e.g., HLOS, ICU LOS) were assessed using linear regression. Multi-variable analyses were adjusted for demographic and clinical variables (age, sex, race, CCI score, health insurance status, U.S. geographical region, site of incident, intention of injury, type of firearm, isolated TBI, ACS trauma center designation, hypotension in the ED, GCS score, and ISS). Odds ratios (ORs) and associated 95% confidence intervals (CIs) were reported for logistic regressions, and mean differences (B), SEs, and 95% CI were reported for linear regressions. Statistical significance was assessed at α = 0.001. All analyses were performed using the Statistical Package for Social Sciences (SPSS; version 23; IBM Corporation, Chicago, IL).

Results

Demographics and clinical characteristics

From admission years 2003–2012, the NSP recorded 355,388 adult TBI incidents. Firearm-related injuries accounted for 8148 events (Table 2), which corresponded to 32,439 events in the U.S. population nationally, or roughly 2.2% of adult TBIs presenting to U.S. trauma centers. Incidence by year showed increased data following standardization of the trauma registry in 2008, with a stable trend in incidence over time since 2008 (Fig. 2). Patients were aged 36.6 ± 16.4 years of age, 86.5% male, and 52.9% Caucasian. Age 18–29 years (45.5%) was of greatest incidence, with few medical comorbidities as evidenced by a mean CCI score of 0.15 ± 0.59 (Table 2). Identifiable insurance statuses were self-pay/unbilled (31.3%), Medicare/Medicaid (15.2%), private/commercial (14.8%), and government/other (10.4%).

FIG. 2.

Left panel: the number of patient events included. Middle panel: proportions of patients with complications, discharge to home, and deaths. Right panel: the hospital LOS and ICU LOS of hospitalized patients alive at discharge (mean ± SE). *Year in which the National Trauma Data Standard was implemented. ICU, intensive care unit; LOS, length of stay; SE, standard error; TBI, traumatic brain injury. Color image is available online at www.liebertpub.com/neu

Table 2.

Demographic Data and Injury Characteristics of 8148 Adults With Firearm-Related TBI^a

Variable	Value^b
Mean age in years (SD)	36.6 (16.4)
M/F ratio (%)	7050 (86.5)/1098 (13.5)
Mean CCI score (SD)^c	0.15 (0.59)
Race
White	4308 (52.9)
Black	2559 (31.4)
Other	1281 (15.7)
Health insurance
Private/commercial	1208 (14.8)
Medicare/Medicaid	1242 (15.2)
Government/other	848 (10.4)
Self-pay/unbilled	2548 (31.3)
Unknown	2302 (28.3)
Region
Midwest	1303 (16.0)
Northeast	391 (4.8)
South	3221 (39.5)
West	1609 (19.7)
Unknown	1624 (19.9)
ACS trauma center
Level I	4237 (52.0)
Level II	1476 (18.1)
Unknown and not applicable	2435 (29.9)
Site of incident
Residential	3517 (43.2)
Street or highway	1822 (22.4)
Industry, mine, or farm	42 (0.5)
Recreational or sport	100 (1.2)
Public building	320 (3.9)
Other and unknown	2347 (28.8)
Mechanism of intent
Accident	286 (3.5)
Suicide and self-inflicted	3410 (41.9)
Assault	4009 (49.2)
Legal intervention	139 (1.7)
Undetermined	304 (3.7)
Type of firearm
Handgun	4102 (50.3)
Shotgun	336 (4.1)
Hunting rifle	201 (2.5)
Other and unspecified	3565 (43.4)
Isolated TBI
No	5621 (69.0)
Yes	908 (11.1)
Unknown	1619 (19.9)
ED SBP
<90	1961 (24.1)
≥90	5623 (69.0)
Unknown	564 (6.9)
ED GCS score
3–8	5247 (64.4)
9–12	332 (4.1)
13–15	2288 (28.1)
Unknown	281 (3.4)
ISS group
0–8	243 (3.0)
9–15	840 (10.3)
16–24	2407 (29.5)
25–75	3844 (47.2)
Unknown	814 (10.0)
ED disposition
Died	1652 (20.3)
Home	103 (1.3)
Observation	38 (0.5)
General ward	509 (6.2)
Step-down	120 (1.5)
ICU	4039 (49.6)
OR	1254 (15.4)
Transfer	22 (0.3)
Other/unknown	411 (5.0)

The age range of the patients was 18–89.

Data represent number of patients (%), unless indicated otherwise.

The range of the CCI score is 0–8.

TBI, traumatic brain injury; SD, standard deviation; M/F, male/female; CCI, controlled cortical impact; ACS, American College of Surgeons; ED, emergency department; SBP, systolic blood pressure; GCS, Glasgow Coma Scale; ISS, Injury Severity Score; ICU, intensive care unit; OR, operating room.

Injury intentions were assault (49.2%), suicide/self-inflicted (41.9%), accident (3.5%), and legal intervention (1.7%). African Americans accounted for the greatest proportion of GSWH by assault mechanism (52.1%). Assault was the most common mechanism in age 18–49 years (57.3%); suicide/self-inflicted was most common in age ≥50 years (72.5%; Fig. 3). Handgun (50.3%) was more often the weapon of injury, and residential area was the most frequent scene of injury (43.2%). The majority of GSWH victims presented with additional trauma not limited to the head and face (69.0%). Median initial GCS score was 3 (interquartile range, [IQR], 3–14); mean GCS was 6.8 ± 5.8. Median ISS was 21 (IQR, 16–25); mean ISS was 20.7 ± 13.9. Most patients had severe TBI (GCS, <8; 64.4%), followed by mild TBI (GCS, 13–15; 28.1%) and moderate TBI (GCS, 9–12; 4.1%). Thirty-one percent of patients were hypotensive (SBP, <90 mm Hg). ED dispositions of highest incidence were ICU (49.6%), death (20.3%), and operating room (15.4%).

FIG. 3.

Left panel: the number of firearm-related TBI events in each adult age group. Right panel: proportions of firearm-related TBI attributed to accident, suicide/self-inflicted, assault, legal intervention, and undetermined intent in each age group. TBI, traumatic brain injury. Color image is available online at www.liebertpub.com/neu

Regional characteristics and trauma center level

Across U.S. regions, most injuries occurred in the South (39.5%), followed by West (19.7%), Midwest (16.0%), Northeast (4.8%), and unknown (19.9%). Univariate analyses indicated high incidence of suicide/self-inflicted in the Midwest (47.8%) and the South (45.1%), compared with the high rates of assault in the Northeast (61.6%), the West (53.7%), and unknown (60.6%; Fig. 4). A smaller proportion of Caucasian patients was observed in the West (47.5%) and Northeast (37.9%) compared to other races. A higher proportion of African Americans was observed in the Northeast (45.5%) and South (37.3%); the category of “other races” was highest comparatively in the West (33.3%; Fig. 5). Handgun injuries were particularly frequent in the Northeast (74.2%; Fig. 6). Fifty-two percent of GSWH patients were admitted to level I trauma centers, 18.1% to level II, and the remainder were undetermined. Admission to level I trauma centers was most common in the Midwest (68.1%) followed by Northeast (64.7%); admission to level II was most common in the West (32.4%) followed by Northeast (27.6%); and undetermined trauma center level was greatest in the South (42.1%).

FIG. 4.

Injury intention and geographical region by U.S. Census. Color image is available online at www.liebertpub.com/neu

FIG. 5.

Patient race and geographical region by U.S. Census. Color image is available online at www.liebertpub.com/neu

FIG. 6.

Type of firearm and geographical region by U.S. Census. Color image is available online at www.liebertpub.com/neu

Mortality

In total, 4446 deaths occurred in the cohort, which corresponded to an overall mortality rate of 54.6%. By location, 37.2% of deaths occurred in the ED, whereas the remainder occurred following triage and during inpatient hospitalization. Mortality rates by motive were highest for suicide/self-inflicted (71.6%) followed by legal intervention (45.3%), assault (41.1%), and accidents (36.7%). Regarding weapon of choice, handguns were associated with the highest relative mortality (57.8%), followed by hunting rifle (48.7%) and shotgun (31.9%), with mortality from other/unspecified at 53.3%. By age, the elderly (60+ years) fared worst with a 73.3% mortality rate, followed by age 50–59 years (63.3%), 40–49 years (54.6%), 18–29 years (50.2%), and 30–39 years (49.3%). Mortality rates by geographical region, and by trauma center, were largely comparable; by geographical region: the Northeast (58.8%), Midwest (57.4%), South (55.0%), West (53.0%), and unknown (52%); by trauma center: level II (58.7%), level I (55.0%), and undetermined (51.3%). Mortality rate did not vary by year of admission (Fig. 2). Low GCS associated with increased mortality rate (Fig. 7).

FIG. 7.

Overall rates of mortality and discharge to home by the TBI severity as assessed by initial GCS in the ED (mean % ± SE). ED, emergency department; GCS, Glasgow Coma Scale; SE, standard error; UNK, unknown. Color image is available online at www.liebertpub.com/neu

Multi-variable analysis indicated that intent of injury, type of firearm, site of incident, age, health insurance, isolated TBI, ED hypotension, TBI severity, and ISS of 25–75 or unknown were statistically significant predictors of death (Table 6). GCS score 3–8 was associated with greatest risk of death (OR, 39.4; [31.4–49.5]; p < 0.001), followed by unknown GCS (OR, 10.6; [7.4–15.2]; p < 0.001), ISS 25–75 (OR, 8.1; [4.5–14.7]; p < 0.001), and ED hypotension (OR, 5.1; [4.3–6.2]; p < 0.001). Of note, after multiple comparisons correction (threshold, p = 0.001), suicide/self-inflicted was marginally associated with higher odds of death (OR, 1.7; [1.2–2.6]; p = 0.004) compared to accidental injury. Relative to handgun injuries, shotgun (OR, 0.3; [CI, 0.2–0.4]; p < 0.001) and hunting rifle (OR, 0.5; [0.4–0.8]; p < 0.001) showed lower odds of mortality. Patients who were self-pay/unbilled (OR, 2.0; [1.6–2.5]; p < 0.001) or with unknown insurance (OR, 1.7; [1.4–2.1]; p < 0.001) were less likely to survive their injuries compared to private/commercially insured.

Hospital and intensive care unit lengths of stay

Mean HLOS was 13.9 ± 17.9 days for those who survived their injuries; when ICU admits were analyzed separately, mean ICU LOS was 8.6 ± 17.4 days. Patients who died during hospitalization had mean HLOS of 2.2 ± 5.7 days and mean ICU LOS of 2.5 ± 4.8 days. Multi-variable linear regression analysis indicated the following predictors for prolonged HLOS in survivors: suicide/self-inflicted, Medicare/Medicaid, the South geographical region, hypotensive status unknown, TBI severity, and ISS of 16–24, 25–75, or unknown; GCS of 3–8 was the only significant predictor for prolonged ICU LOS (Tables 3 and 4). ISS 22–75 (mean increase, 11.8 days; [9.4–14.3]; p < 0.001), GCS score 3–8 (mean increase, 11.0 days; [9.7–12.3]; p < 0.001), GCS score 9–12 (mean increase, 9.3 days; [7.3–11.4]; p < 0.001), ISS unknown (mean increase, 8.8 days; [6.0–11.5]; p < 0.001), GCS unknown (mean increase, 6.3 days; [3.2–9.3]; p < 0.001), ISS 16–24 (mean increase, 5.9 days; [3.6–8.2]; p < 0.001), and unknown hypotensive status (mean increase, 5.0 days; [2.4–7.7]; p < 0.001) were the greatest drivers of prolonged HLOS. Beginning in 2008, hospital LOS and ICU LOS did not vary with admission year (Fig. 2).

Table 3.

Multi-variate Analysis of LOS in the Hospital After Adult Firearm-Related TBI

Parameter	B	SEM	95% CI	p value^a
Mechanism of intent
Accident	Reference	NA	NA	NA
Suicide and self-inflicted	4.4	1.4	1.8 to 7.1	0.001
Assault	3.4	1.3	0.9 to 6.0	0.009
Legal intervention	3.5	2.3	−0.9 to 7.9	0.117
Undetermined	4.0	1.9	0.1 to 7.8	0.042
Type of firearm
Handgun	Reference	NA	NA	NA
Shotgun	1.7	1.2	−0.6 to 4.0	0.144
Hunting rifle	1.5	1.7	−1.9 to 4.8	0.384
Other and unspecified	0.7	0.6	−0.5 to 1.9	0.231
Site of incident
Residential	Reference	NA	NA	NA
Street or highway	0.4	0.8	−1.1 to 1.9	0.592
Industry, mine, or farm	1.0	3.6	−6.1 to 8.2	0.773
Recreational or sport	−2.4	2.4	−7.0 to 2.3	0.316
Public building	1.6	1.4	−1.1 to 4.3	0.255
Unknown	0.5	0.7	−0.8 to 1.9	0.447
Age (per-unit increase)	0.02	0.02	−0.02 to 0.06	0.373
CCI (per-unit increase)	−0.4	0.5	−1.5 to 0.6	0.424
Sex
Male	Reference	NA	NA	NA
Female	−1.8	0.8	−3.4 to −0.3	0.020
Race
White	Reference	NA	NA	NA
Black	−0.5	0.7	−1.9 to 0.9	0.462
Other	−0.2	0.8	−1.8 to 1.4	0.802
Health insurance
Private/commercial	Reference	NA	NA	NA
Medicare/Medicaid	4.4	0.9	2.6 to 6.3	<0.001
Government/other	0.3	1.1	−1.7 to 2.4	0.750
Self-pay/unbilled	−1.3	0.8	−3.0 to 0.3	0.113
Unknown	0.8	0.9	−0.9 to 2.5	0.359
Region
Midwest	Reference	NA	NA	NA
Northeast	1.6	1.5	−1.3 to 4.6	0.283
South	3.7	0.8	2.0 to 5.4	<0.001
West	0.3	1.0	−1.6 to 2.2	0.730
Unknown	4.1	2.5	−0.8 to 9.1	0.101
ACS trauma center
Level I	Reference	NA	NA	NA
Level II	−1.1	0.8	−2.6 to 0.5	0.181
Unknown and not applicable	0.4	0.6	−0.8 to 1.7	0.494
Isolated TBI
No	Reference	NA	NA	NA
Yes	−1.4	1.1	−3.6 to 0.8	0.212
Unknown	−2.1	2.5	−7.0 to 2.7	0.393
ED SBP
≥90	Reference	NA	NA	NA
<90	0.4	1.1	−1.8 to 2.5	0.744
Unknown	5.0	1.3	2.4 to 7.7	<0.001
ED GCS score
13–15	Reference	NA	NA	NA
9–12	9.3	1.1	7.3 to 11.4	<0.001
3–8	11.0	0.7	9.7 to 12.3	<0.001
Unknown	6.3	1.6	3.2 to 9.3	<0.001
ISS
0–8	Reference	NA	NA	NA
9–15	1.7	1.3	−0.8 to 4.2	0.188
16–24	5.9	1.2	3.6 to 8.2	<0.001
25–75	11.8	1.2	9.4 to 14.3	<0.001
Unknown	8.8	1.4	6.0 to 11.5	<0.001

The statistical significance threshold was set at p = 0.001 to account for multiple comparisons.

LOS, length of stay; TBI, traumatic brain injury; SEM, standard error of the mean; CI, confidence interval; NA, not applicable; CCI, controlled cortical impact; ACS, American College of Surgeons; ED, emergency department; SBP, systolic blood pressure; GCS, Glasgow Coma Scale; ISS, Injury Severity Score.

Table 4.

Multi-variate Analysis of LOS in the ICU After Adult Firearm-Related TBI

Parameter	B	SEM	95% CI	p value^a
Mechanism of intent
Accident	Reference	NA	NA	NA
Suicide and self-inflicted	0.6	1.7	−2.7 to 4.0	0.714
Assault	1.8	1.7	−1.5 to 5.2	0.276
Legal intervention	0.4	2.9	−5.4 to 6.2	0.899
Undetermined	2.2	2.5	−2.7 to 7.1	0.372
Type of firearm
Handgun	Reference	NA	NA	NA
Shotgun	0.1	1.4	−2.7 to 2.9	0.931
Hunting rifle	−0.4	2.0	−4.2 to 3.5	0.848
Other and unspecified	−0.6	0.7	−2.1 to 0.8	0.404
Site of incident
Residential	Reference	NA	NA	NA
Street or highway	−0.06	1.0	−2.0 to 1.9	0.952
Industry, mine, or farm	0.4	4.7	−8.7 to 9.6	0.929
Recreational or sport	−2.5	3.2	−8.8 to 3.8	0.442
Public building	0.3	1.7	−3.0 to 3.5	0.870
Unknown	0.9	0.9	−0.8 to 2.5	0.315
Age (per-unit increase)	0.001	0.026	−0.050 to 0.053	0.956
CCI (per-unit increase)	−0.5	0.7	−1.9 to 0.8	0.454
Sex
Male	Reference	NA	NA	NA
Female	−2.0	1.0	−3.9 to −0.1	0.037
Race
White	Reference	NA	NA	NA
Black	1.2	0.9	−0.6 to 2.9	0.192
Other	−0.7	1.0	−2.7 to 1.4	0.507
Health insurance
Private/commercial	Reference	NA	NA	NA
Medicare/Medicaid	1.2	1.1	−1.0 to 3.3	0.304
Government/other	1.5	1.3	−1.1 to 4.1	0.254
Self-pay/unbilled	−1.5	1.0	−3.6 to 0.6	0.155
Unknown	0.1	1.1	−1.9 to 2.2	0.896
Region
Midwest	Reference	NA	NA	NA
Northeast	−0.3	2.0	−4.2 to 3.5	0.865
South	0.5	1.0	−1.5 to 2.5	0.634
West	−0.5	1.2	−2.8 to 1.9	0.688
Unknown	4.2	2.8	−1.2 to 9.7	0.128
ACS trauma center
Level I	Reference	NA	NA	NA
Level II	−1.2	1.0	−3.1 to 0.7	0.226
Unknown and not applicable	−0.8	0.8	−2.3 to 0.7	0.317
Isolated TBI
No	Reference	NA	NA	NA
Yes	−0.2	1.4	−2.9 to 2.5	0.865
Unknown	−3.2	2.7	−8.5 to 2.2	0.250
ED SBP
≥90	Reference	NA	NA	NA
<90	0.4	1.4	−2.3 to 3.2	0.747
Unknown	2.2	1.6	−1.0 to 5.3	0.173
ED GCS score
13–15	Reference	NA	NA	NA
9–12	3.6	1.2	1.2 to 6.0	0.003
3–8	5.0	0.8	3.5 to 6.5	<0.001
Unknown	2.4	2.0	−1.5 to 6.3	0.232
ISS
0–8	Reference	NA	NA	NA
9–15	2.0	2.5	−2.8 to 6.8	0.421
16–24	5.0	2.3	0.4 to 9.6	0.032
25–75	7.7	2.4	3.1 to 12.4	0.001
Unknown	6.0	2.5	1.0 to 10.9	0.018

The statistical significance threshold was set at p = 0.001 to account for multiple comparisons.

LOS, length of stay; ICU, intensive care unit; TBI, traumatic brain injury; SEM, standard error of the mean; CI, confidence interval; NA, not applicable; CCI, controlled cortical impact; ACS, American College of Surgeons; ED, emergency department; SBP, systolic blood pressure; GCS, Glasgow Coma Scale; ISS, Injury Severity Score.

Injuries from suicide/self-inflicted mechanism demonstrated longer HLOS (mean increase, 4.4 days; [1.8–7.1]; p < 0.001) relative to unintentional injuries. In comparison to patients with private/commercial insurance, Medicare/Medicaid insurance was associated with prolonged HLOS (mean increase, 4.4 days; [2.6–6.3]; p < 0.001). By regional classification, patients treated in the South had prolonged HLOS (mean increase, 3.7 days; [2.0–5.4]; p < 0.001) compared to the Midwest. Multi-variate analysis of ICU LOS of patients who survived their injuries found GCS score 3–8 as the only significant predictor of longer ICU LOS (mean increase, 5.0 days; [3.5–6.5]; p < 0.001) relative to GCS score 13–15.

Medical complications

In total, 1412 patients (17.3%) suffered medical complications during hospitalization, which included pneumonia, acute respiratory distress syndrome (ARDS), cardiac arrest, deep venous thrombosis (DVT), coagulopathy, acute kidney injury or renal failure, pulmonary embolism, myocardial infarction, cerebrovascular complication, decubitus ulceration, and urinary tract infection. Pneumonia (5.2%), ARDS (5.0%), cardiac arrest (5.0%), and DVT (1.7%) were most common. Multi-variable analysis indicated that geographical region, TBI severity, and ISS were significant predictors for medical complications (Table 5). Compared to ISS 0–8, unknown ISS was associated with the greatest odds of complications (OR, 14.2; [4.5–45.2]; p < 0.001), followed by ISS 25–75 (OR, 10.9; [3.4–34.4]; p < 0.001) and ISS 16–24 (OR, 10.1; [3.2–31.9]; p < 0.001). Multiple comparisons correction (threshold, p = 0.001) indicated that, relative to handgun, other/unspecified weapon of injury had higher risk of complications (OR, 1.3; [1.1–1.5]; p < 0.001). Compared to the Midwest, the South had higher odds of complications (OR, 1.7; [1.4–2.0]; p < 0.001), and the West had lower odds of complications (OR, 0.6; [0.5–0.7]; p < 0.001). Complication rate did not vary by year of admission (Fig. 2).

Table 5.

Multi-variate Analysis of Predictors of Complications After Adult Firearm-Related TBI

Parameter	Odds ratio (95% CI)	p value^a
Mechanism of intent		0.367
Accident	Reference	NA
Suicide and self-inflicted	0.8 (0.6–1.2)	0.274
Assault	0.9 (0.7–1.3)	0.748
Legal intervention	0.7 (0.4–1.3)	0.220
Undetermined	0.9 (0.6–1.4)	0.753
Type of firearm		0.002
Handgun	Reference	NA
Shotgun	1.1 (0.8–1.5)	0.707
Hunting rifle	0.9 (0.6–1.3)	0.582
Other and unspecified	1.3 (1.1–1.5)	<0.001
Site of incident		0.393
Residential	Reference	NA
Street or highway	1.0 (0.8–1.2)	0.889
Industry, mine, or farm	1.8 (0.8–4.0)	0.125
Recreational or sport	0.7 (0.4–1.3)	0.308
Public building	1.1 (0.8–1.4)	0.717
Unknown	0.9 (0.8–1.1)	0.268
Age (per-unit increase)	1.0 (0.99–1.00)	0.094
CCI (per-unit increase)	1.1 (1.0–1.2)	0.264
Sex
Male	Reference	NA
Female	0.9 (0.8–1.1)	0.245
Race		0.144
White	Reference	NA
Black	0.9 (0.7–1.0)	0.067
Other	1.0 (0.8–1.2)	0.981
Health insurance		0.005
Private/commercial	Reference	NA
Medicare/Medicaid	1.2 (0.9–1.4)	0.153
Government/other	1.0 (0.7–1.2)	0.769
Self-pay/unbilled	0.9 (0.7–1.1)	0.297
Unknown	0.8 (0.7–1.0)	0.044
Region		<0.001
Midwest	Reference	NA
Northeast	0.8 (0.6–1.1)	0.161
South	1.7 (1.4–2.0)	<0.001
West	0.6 (0.5–0.7)	<0.001
Unknown	0.7 (0.3–1.5)	0.409
ACS trauma center		0.136
Level I	Reference	NA
Level II	0.9 (0.8–1.1)	0.252
Unknown and not applicable	0.9 (0.8–1.0)	0.061
Isolated TBI		0.094
No	Reference	NA
Yes	0.9 (0.7–1.0)	0.099
Unknown	0.6 (0.3–1.2)	0.153
ED SBP		0.165
≥90	Reference	NA
<90	0.9 (0.7–1.0)	0.058
Unknown	0.9 (0.7–1.2)	0.624
ED GCS score		<0.001
13–15	Reference	NA
9–12	3.3 (2.5–4.4)	<0.001
3–8	2.2 (1.8–2.6)	<0.001
Unknown	2.1 (1.5–3.1)	<0.001
ISS		<0.001
0–8	Reference	NA
9–15	5.5 (1.7–17.8)	0.004
16–24	10.1 (3.2–31.9)	<0.001
25–75	10.9 (3.4–34.3)	<0.001
Unknown	14.2 (4.5–45.2)	<0.001

The statistical significance threshold was set at p = 0.001 to account for multiple comparisons.

TBI, traumatic brain injury; CI, confidence interval; NA, not applicable; CCI, controlled cortical impact; ACS, American College of Surgeons; ED, emergency department; SBP, systolic blood pressure; GCS, Glasgow Coma Scale; ISS, Injury Severity Score.

Table 6.

Multi-variate Analysis of Predictors of Death After Adult Firearm-Related TBI

Parameter	Odds ratio (95% CI)	p value^a
Mechanism of intent		<0.001
Accident	Reference	NA
Suicide and self-inflicted	1.7 (1.2–2.6)	0.004
Assault	1.0 (0.7–1.4)	0.945
Legal intervention	0.9 (0.5–1.6)	0.643
Undetermined	1.5 (0.9–2.4)	0.120
Type of firearm		<0.001
Handgun	Reference	NA
Shotgun	0.3 (0.2–0.4)	<0.001
Hunting rifle	0.5 (0.4–0.8)	<0.001
Other and unspecified	1.1 (0.9–1.2)	0.509
Site of incident		<0.001
Residential	Reference	NA
Street or highway	1.2 (1.0–1.5)	0.051
Industry, mine, or farm	1.8 (0.6–5.8)	0.289
Recreational or sport	0.9 (0.5–1.7)	0.723
Public building	1.0 (0.7–1.3)	0.777
Unknown	0.8 (0.6–0.9)	0.002
Age (per-unit increase)	1.0 (1.01–1.02)	<0.001
CCI (per-unit increase)	0.9 (0.8–1.0)	0.057
Sex
Male	Reference	NA
Female	0.9 (0.7–1.1)	0.240
Race		0.899
White	Reference	NA
Black	1.0 (0.9–1.3)	0.660
Other	1.0 (0.8–1.3)	0.767
Health insurance		<0.001
Private/commercial	Reference	NA
Medicare/Medicaid	1.0 (0.8–1.3)	0.992
Government/other	1.3 (1.0–1.7)	0.081
Self-pay/unbilled	2.0 (1.6–2.5)	<0.001
Unknown	1.7 (1.4–2.1)	<0.001
Region		0.054
Midwest	Reference	NA
Northeast	1.5 (1.1–2.2)	0.020
South	1.0 (0.8–1.2)	0.974
West	0.9 (0.7–1.1)	0.362
Unknown	0.8 (0.4–1.7)	0.621
ACS trauma center		0.033
Level I	Reference	NA
Level II	1.3 (1.1–1.6)	0.010
Unknown and not applicable	1.1 (0.9–1.3)	0.273
Isolated TBI		<0.001
No	Reference	NA
Yes	2.5 (2.0–3.2)	<0.001
Unknown	0.9 (0.5–1.9)	0.865
ED SBP		<0.001
≥90	Reference	NA
<90	5.1 (4.3–6.2)	<0.001
Unknown	2.3 (1.8–2.9)	<0.001
ED GCS score		<0.001
13–15	Reference	NA
9–12	3.5 (2.4–5.1)	<0.001
3–8	39.4 (31.4–49.5)	<0.001
Unknown	10.6 (7.4–15.2)	<0.001
ISS		<0.001
0–8	Reference	NA
9–15	1.6 (0.8–2.9)	0.162
16–24	1.6 (0.9–3.0)	0.109
25–75	8.1 (4.5–14.7)	<0.001
Unknown	3.0 (1.6–5.6)	<0.001

The statistical significance threshold was set at p = 0.001 to account for multiple comparisons.

Hospital discharge disposition

Among patients who survived their injuries, 2240 (27.5%) were discharged to home and 1462 (17.9%) to rehabilitation or skilled nursing facilities. By mechanism of intent, rates of return to home were: assault (70.5%), accident (64.6%), undetermined (59.5%), legal intervention (57.9%), and a much lower return-home rate for suicide/self-inflicted (35.8%). Patients with suicide/self-inflicted mechanism of intent had the lowest odds of return to home (OR, 0.5; [0.3–0.7]; p < 0.001) compared to accidental injury. Discharge rates were comparable across firearm type and site of incident. Return to home rates were, on average, higher for younger patients: 18–29 years (66.1%), 30–39 years (62.9%), 40–49 years (59.6%), and lower for 50–59 years (46.7%) and 60+ years (29.9%). Return to home rates were comparable across the geographical regions: the South (64.4%), Northeast (62.7%), West (61.6%), and Midwest (58.0%); region of unknown (53.5%) was lower than the South and West. Patients at level II trauma centers were less likely to return home than those at level I trauma centers (53.8% vs. 62.1%). By insurance category, self-pay/unbilled patients were most likely to go home (66.9%) followed by government/other (64.7%) and unknown (64.0%), whereas Medicare/Medicaid (51.1%) and private/commercial (50.2%) patients had the lowest rates of return to home. Rate of discharge to home did not vary by year of admission (Fig. 2). Low GCS associated with decreased rate of discharge to home (Fig. 7).

Multi-variable analysis indicated that mechanism of intent of injury, age, health insurance, trauma center level, ED hypotension, TBI severity, and ISS were significant predictors of return to home (Table 7). In comparison to GCS score 13–15, GCS scores 9–12 (OR, 0.4; [0.3–0.6]; p < 0.001), 3–8 (OR, 0.4; [0.3–0.4]; p < 0.001), and unknown (OR, 0.4; [0.2–0.6]; p < 0.001) were the strongest drivers of decreased odds of returning home, followed by government/other insurance (OR, 1.7; [1.3–2.3]; p < 0.001). After multiple comparisons correction (threshold, p = 0.001), suicide/self-inflicted injuries were associated with lower odds of returning home (OR, 0.5; [0.3–0.7]; p < 0.001) compared to accidental injury. Patients with government/other insurance had higher odds of returning home (OR, 1.7; [1.3–2.3]; p < 0.001) compared to those who were privately or commercially insured. Relative to patients from level I trauma centers, level II patients were less likely to return home (OR, 0.7; [0.5–0.8]; p < 0.001).

Table 7.

Multi-variate Analysis of Predictors of Hospital Discharge Disposition After Adult Firearm-Related TBI

Parameter	Odds ratio (95% CI)	p value^a
Mechanism of intent		<0.001
Accident	Reference	NA
Suicide and self-inflicted	0.5 (0.3–0.7)	<0.001
Assault	1.3 (0.9–1.9)	0.138
Legal intervention	0.9 (0.5–1.6)	0.630
Undetermined	1.1 (0.6–1.8)	0.789
Type of firearm		0.193
Handgun	Reference	NA
Shotgun	1.4 (1.0–2.0)	0.031
Hunting rifle	1.0 (0.6–1.6)	0.994
Other and unspecified	1.1 (0.9–1.2)	0.521
Site of incident		0.003
Residential	Reference	NA
Street or highway	0.9 (0.7–1.1)	0.240
Industry, mine, or farm	0.6 (0.2–1.6)	0.280
Recreational or sport	2.0 (1.0–4.1)	0.058
Public building	1.1 (0.8–1.6)	0.594
Unknown	1.3 (1.1–1.6)	0.014
Age (per-unit increase)	0.8 (0.8–0.9)	<0.001
CCI (per-unit increase)	0.9 (0.8–1.1)	0.196
Sex
Male	Reference	NA
Female	0.9 (0.7–1.1)	0.139
Race		0.785
White	Reference	NA
Black	1.1 (0.9–1.3)	0.522
Other	1.1 (0.8–1.3)	0.616
Health insurance		<0.001
Private/commercial	Reference	NA
Medicare/Medicaid	0.9 (0.7–1.2)	0.550
Government/other	1.7 (1.3–2.3)	<0.001
Self-pay/unbilled	1.4 (1.1–1.8)	0.005
Unknown	1.4 (1.1–1.7)	0.011
Region		0.094
Midwest	Reference	NA
Northeast	1.0 (0.7–1.5)	0.980
South	1.3 (1.0–1.7)	0.016
West	1.2 (0.9–1.6)	0.216
Unknown	0.8 (0.4–1.7)	0.590
ACS trauma center		<0.001
Level I	Reference	NA
Level II	0.7 (0.5–0.8)	<0.001
Unknown and not applicable	0.9 (0.8–1.1)	0.242
Isolated TBI		0.145
No	Reference	NA
Yes	0.7 (0.5–1.0)	0.053
Unknown	0.9 (0.4–1.7)	0.706
ED SBP		<0.001
≥90	Reference	NA
<90	0.6 (0.4–0.8)	<0.001
Unknown	1.1 (0.7–1.5)	0.687
ED GCS score		<0.001
13–15	Reference	NA
9–12	0.4 (0.3–0.6)	<0.001
3–8	0.4 (0.3–0.4)	<0.001
Unknown	0.4 (0.2–0.6)	<0.001
ISS		<0.001
0–8	Reference	NA
9–15	1.6 (1.1–2.3)	0.018
16–24	1.1 (0.8–1.5)	0.711
25–75	0.6 (0.4–0.9)	0.004
Unknown	0.8 (0.6–1.2)	0.356

The statistical significance threshold was set at p = 0.001 to account for multiple comparisons.

Discussion

Civilian firearm-related TBI is an important public health concern with growing recognition from both the medical community and the public. Previous studies have documented the incidence of GSWH in various urban and rural communities.^8,35

–38 However, the absence of a single data set that incorporates weapon characteristics, patient demographics, inpatient course, and acute outcome has precluded large-scale analyses of specific risk factors for prolonged LOS, increased complications, and acuity of discharge destination after GSWH. The present study utilized the NSP of the NTDB to characterize the epidemiology and clinical history of U.S. firearm TBI over the past decade with a targeted interest on firearm cohort, U.S. geographical locale, and acute management strategies. To our knowledge, this study is the first to define predictors of poor outcome after GSWH across regional centers to reflect national trends of care.

Demographics and mortality

The difference in incidence between years 2003–2007 and 2008–2012 likely reflects a limitation of data acquisition during the early years and more standardized data collection with implementation of the National Trauma Data Standard (NTDS) beginning 2008.^39,40 In our analysis of 8148 adult GSWH patients triaged to a level I or II trauma center, over 60% were classified as severe TBI by GCS score, over 50% were assaults, and over 30% of patients were hypotensive (SBP, <90 mm Hg). Handgun was the most prevalent firearm type, causing half of all injuries. These findings are largely consistent with previous reports that small firearms cause approximately half of 50,000 TBI deaths each year in the United States with high rates of early mortality.^38,41 Overall mortality without accounting for out-of-hospital deaths in the current study was slightly more than half, and over 60% of these deaths occurred during inpatient hospitalization, conforming to national reports that less than half of patients arriving alive to the ED with GSWH ultimately survive their injuries at hospital discharge.^5,7,9,42 In the literature, age, CT findings of basal cistern obliteration and intraventricular hemorrhage, and GCS score at admission are consistent predictors of mortality after GSWH.^{8,9,43

–47}

Not surprisingly, the majority of GSWH patients were ages 18–29 years. In contrast to the higher incidence of interpersonal violence in patients <40 years of age (62%), suicide/self-inflicted injuries were more frequently observed in patients >40 years of age (65%). We found that mortality rate increased with age, and older patients were less likely to be discharged home. Severe TBI by GCS score conferred the greatest odds of death, followed by unknown GCS, ISS 25–75, ED hypotension, unknown ISS, insurance type, injury intention, and firearm type. Patients categorized as “unknown” likely include those who expired before completing their injury severity assessment in the ED, and thus an “unknown” GCS or ISS can be a marker of mortality. Similarly, patients who were self-pay/unbilled or with unknown insurance had higher rates of mortality than those with identifiable insurance. Suicide/self-inflicted GSWH conferred greater odds of death compared to unintentional injury; by firearm type, shotgun and hunting rifle were associated with lower odds of mortality compared to handgun. Shotgun wounds carry a wide severity spectrum owing to the fact that pellets scatter and behave as individual low-velocity projectiles, even at the range of a few meters.^48,49 Whereas shotgun GSWH carries lower mortality, our analysis showed that shotgun-injured patients experienced longer stays in both the ICU (mean increase, 3.5 days) and hospital (mean increase, 4.2 days). High-energy rifle bullets carry high lethality and these patients rarely survive to be admitted, unless GSWH from rifle shots are accidental hits from long range or ricocheted striking at low velocity.^50,51

Length of stay and discharge disposition

Not surprisingly, lower GCS score, increasing ISS, ED hypotension, and increasing age were also negative predictors for discharge to home; relative to patients of unintentional injury, suicide/self-inflicted injury patients were half as likely to return home. Patients admitted at level II trauma centers had an OR of 0.7 for being discharged home compared to level I trauma centers, a finding consistent with previous studies.⁵² Several explanations regarding this gap in functional outcome have been proposed, including higher volume and higher acuity of patients observed in level I centers, more experienced medical personnel, in-house neurosurgeons, and the immediate availability of the operating room.^53

–56

Patients with Medicare/Medicaid had longer stays in both the hospital and ICU compared to private/commercial insurance. Medicare/Medicaid patients are often more health burdened at baseline, constituting a higher health-risk bracket and may require longer monitoring during their hospitalization. Patients who were self-pay/unbilled had shorter HLOS stay (mean decrease, 2.4 days) compared to those who were privately/commercially insured, which can be explained by both the high mortality rate of unbilled individuals and also by the gap in medical access for uninsured patients.^57

–60 Interestingly, relative to private/commercial insurance, patients with government-sponsored insurance were more likely to be discharged home. Although the reason for this finding is unclear, it is possible that patients insured by government agencies have set care standards before a patient is deemed ready for discharge, as well as an organized network of referrals and/or placement to low-acuity residential facilities, in-home health services and social/community services.

Isolated gunshot wounds to the head versus polytrauma

The majority of patients suffered additional trauma not limited to the head and face. A higher proportion of polytraumatic injuries were from assault, whereas a majority of isolated TBI and face injuries were suicide/self-inflicted. In the past several decades, demand for pistols has increased in part because of their increased capacity (vs. the six-round capacity of revolvers); accordingly, the number of patients with multiple GSWs in assaults increased as well.⁶¹ A recent study found that in urban communities of low socioeconomic status, multiple GSWs was associated with higher mortality and longer ICU stay.⁶² National data indicate that approximately one in four Americans shot during interpersonal violence do not survive.⁶³ For assaultive GSWH, we report a mortality rate of 41%. In the context of all GSWH events, isolated TBIs have higher mortality than nonisolated (74% vs. 52%), partly attributed to the close-range lethality by suicide/self-inflicted injury compared with being shot multiple times at farther range. We observed no difference in HLOS and ICU LOS in isolated versus polytraumatic injuries for patients surviving to hospital discharge.

Medical complications

Similar to the other outcomes assessed in this study, GCS and ISS scores were the strongest predictors of medical complications. Pneumonia (5%), ARDS (5%), cardiac arrest (5%), and DVT (2%) were most frequently observed, which aligns with the fact that severe brain injury, mechanical ventilation, depressed consciousness, aspiration, and secondary cerebral insults are prominent risk factors for pulmonary complications when the central nervous system/respiratory axis is altered after severe TBI.^64

–67 Previous evidence recommended that the prevention or reversal of such complications should be prioritized during critical care to improve outcome.⁶⁸ In comparison to handgun, other/unspecified firearms had a higher complication rate; differences in ballistics may play a role and deserve additional investigation in future studies. Inherent missile characteristics (mass, shape, and construction), velocity (longitudinal and rotational), and fragmentation tendency during impact (hollow point or soft point) confer differing injury patterns that alter clinical course and risk for complications.^69,70

Distinctions by U.S. region and trauma center designation

This study demonstrated a few regional differences of interest. Compared with patients treated in the Midwest, those in the South had a higher complication rate whereas the West had a lower complication rate. The South also had a prolonged HLOS (mean increase, 3.7 days) relative to the Midwest; a higher complication rate would naturally result in longer hospitalization. Although mortality rates did not vary across the United States as was expected, the discrepancies in regional complication rate and hospitalization length are worth noting. Although great efforts at standardization have been made since the Guidelines for the Management of Penetrating Brain Injury was published in 2001, GSWH represent a unique subset of TBIs in need of evidence-based-guidelines on optimal management.^71
–73 Recently, there is a trend toward less-aggressive debridement of bone and missile fragments and more-aggressive antibiotic prophylaxis. Whether the best surgical approach is craniectomy or craniotomy and the survival benefit from intracranial pressure (ICP) monitoring are still under debate, with varied reports on morbidity and mortality.^12,74
–76 Other explanations previously put forth include differences in trauma center distribution and services expenditure, varied stages of regionalization for trauma-care delivery, and urban-rural pre-hospital care.^77

–80

Cost of gunshot injuries

Firearm ownership policies worldwide correlate, and often are commensurate with, national firearm-related injury rates. The correlation between gun ownership and homicide/suicide rates has been previously affirmed by the International Crime Survey in 14 countries including the United States.⁸¹ Other studies showed similar findings, with the highest comparative rate of firearm-related injuries in the United States.^82,83 On the other hand, lower gun availability has not been shown to spur an increase in other methods of homicide and suicide. Whereas rates of both homicide and suicide by firearm rise with gun ownership, it has a stronger impact on suicides than homicides, and, given the particularly high mortality associated with a suicidal intent, reducing gun availability may decrease risk of firearm-related deaths.⁸⁴

Cook and colleagues estimated the national cost of acute-care and follow-up treatment to be $17,000 per gunshot injury, which in 1994 alone produced $2.3 billion in lifetime costs; assaults accounted for 74% of the aggregate cost.⁸⁵ Considering that U.S. taxpayers paid nearly $1.1 billion (49%) of these costs, firearm injuries create a substantial burden to the healthcare system.^86
–88 Further, the financial burden per injury is substantially greater for gunshot trauma to the head, with the mean cost of treatment in these patients totaling $117,338 per individual.⁸⁹ Firearms rank second only to motor vehicles among all consumer products in their consumption of public healthcare funds.

Limitations of the study

This study is not without limitations. First, it only included data submitted by level I and II trauma centers participating in the NTDB, and therefore does not capture GSWH at non-ACS-designated trauma centers, and injuries without sufficient severity to triage to higher-acuity care. GSWH patients who expire at the scene, or en route to the hospital, are likewise not captured by the NTDB. Therefore, the true incidence and prevalence of GSWH in U.S. trauma centers are likely several times higher than recorded in the current analysis, on both extremes of the injury severity spectrum. The increased incidence from 2007 to 2008 was attributed to improved standardization and uniformity of information collection by the trauma registry, which limited the potential to accurately compare the periods 2003–2007 with 2008–2012. Specifics of the hospital management course, such as critical care management, are not recorded in the NTDB and represent a significant limitation to the understanding of the optimal management course for distinct GSWH injury profiles, as well as protocol differences relating to U.S. regions. It is known that significant practice variation exists among even trauma centers of the same region and acuity, which is another shortcoming of retrospective analyses, given that comparative effectiveness cannot be readily determined.

For the current analysis, we focused on the demographic and clinical injury effects on acute outcome after GWSH, and did not delve into the myriad of surgical interventions appropriate for various GWSH etiologies and severities, which should be the subject of future studies of similar scale. Associations between the type of neurosurgical intervention and outcomes, such as length of stay, complications, disposition, and mortality, constitute topics of ongoing study to improve management, prognosis, and recovery post-GSWH. In addition, the sequelae of medical complications, and subsequent interventions, are not recorded by the NTDB and represent a gap in our understanding of GSWH outcome in the United States. We are also limited to 30-day outcomes in the NTDB and are unable to determine the long-term outcome of patients contained in this study nor improvement or decompensation in their trajectory of recovery.

Another limitation of the national database, particularly for a patient population with high injury severity and mortality at initial triage, may be the lack of complete data reporting. Although we perform multi-variable regressions to identify predictors for poor outcome, we are limited by the data submitted to the NSP of the NTDB. Improved quality assurance to minimize unknown data would aid future studies in the GSWH population. Gunshot injuries characterized as separate entities (i.e., homicide vs. suicide) can often comprise interconnecting crimes (i.e., homicides-suicides).⁸³ Such a distinction does not exist in the NTDB nor the majority of retrospective studies to date; hence, improved characterization of interconnecting crimes would provide a more accurate understanding of firearm epidemiology in the United States and possibly provide evidence for formulating effective policies at reducing GSWH.

Conclusions

Older age, hypotension, and injury severity (GCS and ISS) are associated with poor prognosis after firearm-related TBI. Isolated GSWH and handgun injuries are associated with higher mortality. Suicide/self-inflicted GSWHs are more likely to suffer prolonged hospitalization and decreased odds of returning home. Compared with level I trauma centers, GSWH patients treated at level II trauma centers have lower odds of being discharged home. Patients with Medicare/Medicaid insurance show longer hospitalizations than those with private/commercial insurance, patients of self-pay/unbilled insurance have higher mortality, and patients of government/other insurance are more likely to be discharged home. Practice differences for GSWH management likely exist among U.S. geographical locales and warrant further investigation. Improved understanding of civilian GSWH is critical to promoting increased awareness of firearm injuries as a public health concern and reducing its injury burden to patients, families, and healthcare systems.

Footnotes

Acknowledgments

Committee on Trauma, American College of Surgeons, NTDB NSP 2003-12, Chicago, IL. The content reproduced from the NTDB NSP remains the full and exclusive copyrighted property of the American College of Surgeons. The American College of Surgeons is not responsible for any claims arising from works based on the original data, text, tables, or figures.

Author Disclosure Statement

No competing financial interests exist.

References

Faul

, Xu

, Wald

M.M.

, and Coronado

V.G.

(2010). Traumatic Brain Injury in the United States: Emergency Department Visits, Hospitalizations, and Deaths, 2002–2006. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Injury Prevention and Control: Atlanta, GA.

Roozenbeek

, Maas

A.I.R.

, and Menon

D.K.

(2013). Changing patterns in the epidemiology of traumatic brain injury. Nat. Rev. Neurol., 9, 231–236.

Laskowitz

, and Grant

(2015). Translational Research in Traumatic Brain Injury. CRC: Boca Raton, FL.

Bizhan

, Mossop

, and Aarabi

J.A.

(2015). Surgical management of civilian gunshot wounds to the head. Handb. Clin. Neurol., 127, 181–193.

Siccardi

, Cavaliere

, Pau

, Lubinu

, Turtas

, and Viale

G.L.

(1991). Penetrating craniocerebral missile injuries in civilians: a retrospective analysis of 314 cases. Surg. Neurol., 35, 455–460.

[No authors listed]. (2001). Part 2: Prognosis in penetrating brain injury. J. Trauma, 51, 2 Suppl., S44–S86.

Zafonte

R.D.

, Wood

D.L.

, Harrison-Felix

C.L.

, Valena

N.V.

, and Black

(2001). Penetrating head injury: a prospective study of outcomes. Neurol. Res., 23, 219–226.

Kaufman

H.H.

, Makela

M.E.

, Lee

K.F.

, Haid

R.W.

Jr. , and Gildenberg

P.L.

(1986). Gunshot wounds to the head: a perspective. Neurosurgery, 18, 689–695.

Grahm

T.W.

, Williams

F.C.

Jr. , Harrington

, and Spetzler

R.F.

(1990). Civilian gunshot wounds to the head: a prospective study. Neurosurgery, 27, 696–700; discussion, 700.

10.

[No authors listed]. (2001). Part 1: Guidelines for the management of penetrating brain injury. Introduction and methodology. J. Trauma, 51, 2 Suppl., S3–S6.

11.

Bullock

, Chesnut

R.M.

, Clifton

, Ghajar

, Marion

D.W.

, Narayan

R.K.

, Newell

D.W.

, Pitts

L.H.

, Rosner

M.J.

, and Wilberger

J.W.

(1996). Guidelines for the management of severe head injury. Brain Trauma Foundation. Eur. J. Emerg. Med., 3, 109–127.

12.

Kazim

S.F.

, Shamim

M.S.

, Tahir

M.Z.

, Enam

S.A.

, and Waheed

(2011). Management of penetrating brain injury. J. Emerg. Trauma Shock, 4, 395–402.

13.

Brandvold

, Levi

, Feinsod

, and George

E.D.

(1990). Penetrating craniocerebral injuries in the Israeli involvement in the Lebanese conflict, 1982–1985. Analysis of a less aggressive surgical approach. J. Neurosurg., 72, 15–21.

14.

Aarabi

(1990). Surgical outcome in 435 patients who sustained missile head wounds during the Iran-Iraq War. Neurosurgery, 27, 692–695; discussion, 695.

15.

Hammon

W.M.

(1971). Analysis of 2187 consecutive penetrating wounds of the brain from Vietnam. J. Neurosurg., 34, 127–131.

16.

Gressot

L.V.

, Chamoun

R.B.

, Patel

A.J.

, Valadka

A.B.

, Suki

, Robertson

C.S.

, and Gopinath

S.P.

(2014). Predictors of outcome in civilians with gunshot wounds to the head upon presentation. J. Neurosurg., 121, 645–652.

17.

Aarabi

, Tofighi

, Kufera

J.A.

, Hadley

, Ahn

E.S.

, Cooper

, Malik

J.M.

, Naff

N.J.

, Chang

, Radley

, Kheder

, and Uscinski

R.H.

(2014). Predictors of outcome in civilian gunshot wounds to the head. J. Neurosurg., 120, 1138–1146.

18.

Carr

B.G.

, Schwab

C.W.

, Branas

C.C.

, Killen

, and Wiebe

D.J.

(2008). Outcomes related to the number and anatomic placement of gunshot wounds. J. Trauma, 64, 197–202; discussion, 202–203.

19.

Stone

J.L.

, Lichtor

, Fitzgerald

L.F.

, Barrett

J.A.

, and Reyes

H.M.

(1995). Demographics of civilian cranial gunshot wounds: devastation related to escalating semiautomatic usage. J. Trauma, 38, 851–854.

20.

Hargarten

S.W.

, Karlson

T.A.

, O'Brien

, Hancock

, and Quebbeman

(1996). Characteristics of firearms involved in fatalities. JAMA, 275, 42–45.

21.

Peek-Asa

, Butcher

, and Cavanaugh

J.E.

(2017). Cost of hospitalization for firearm injuries by firearm type, intent, and payer in the United States. Inj. Epidemiol., 4, 20.

22.

Fowler

K.A.

, Dahlberg

L.L.

, Haileyesus

, and Annest

J.L.

(2015). Firearm injuries in the United States. Prev. Med., 79, 5–14.

23.

Centers for Disease Control and Prevention (CDC). (1999). Nonfatal and fatal firearm-related injuries—United States, 1993–1997. MMWR Morb. Mortal. Wkly. Rep., 48, 1029–1034.

24.

Taylor

C.A.

, Bell

J.M.

, Breiding

M.J.

, and Xu

(2017). Traumatic Brain Injury-Related Emergency Department Visits, Hospitalizations, and Deaths—United States, 2007 and 2013. MMWR Surveill. Summ., 66, 1–16.

25.

American College of Surgeons. ([date unknown]). National Trauma Data Bank (NTDB) Research Dataset Admission Year 2012: User Manual. American College of Surgeons: Chicago, IL.

26.

Schoenfeld

A.J.

, Belmont

P.J.

Jr. , See

A.A.

, Bader

J.O.

, and Bono

C.M.

(2013). Patient demographics, insurance status, race, and ethnicity as predictors of morbidity and mortality after spine trauma: a study using the National Trauma Data Bank. Spine J. 13, 1766–1773.

27.

Bekelis

, Missios

, and Mackenzie

T.A.

(2015). Prehospital helicopter transport and survival of patients with traumatic brain injury. Ann. Surg., 261, 579–585.

28.

Bowman

S.M.

, Martin

D.P.

, Sharar

S.R.

, and Zimmerman

F.J.

(2007). Racial disparities in outcomes of persons with moderate to severe traumatic brain injury. Med. Care, 45, 686–690.

29.

Majidi

, Shahram

, Farhan

, and Qureshi

A.I.

(2013). Prehospital neurologic deterioration is independent predictor of outcome in traumatic brain injury: analysis from National Trauma Data Bank. Am. J. Emerg. Med., 31, 1215–1219.

30.

Charlson

M.E.

, Pompei

, Ales

K.L.

, and MacKenzie

C.R.

(1987). A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J. Chronic Dis., 40, 373–383.

31.

Yue

J.K.

, Winkler

E.A.

, Burke

J.F.

, Chan

A.K.

, Dhall

S.S.

, Berger

M.S.

, Manley

G.T.

, and Tarapore

P.E.

(2016). Pediatric sports-related traumatic brain injury in United States trauma centers. Neurosurg. Focus, 40, E3.

32.

Winkler

E.A.

, Yue

J.K.

, Burke

J.F.

, Chan

A.K.

, Dhall

S.S.

, Berger

M.S.

, Manley

G.T.

, and Tarapore

P.E.

(2016). Adult sports-related traumatic brain injury in United States trauma centers. Neurosurg. Focus, 40, E4.

33.

Zhao

, Mejaddam

A.Y.

, Chang

, DeMoya

M.A.

, King

D.R.

, Yeh

D.D.

, Kaafarani

H.M.A.

, Alam

H.B.

, and Velmahos

G.C.

(2016). Admissions for isolated non-operative mild head injuries: Sharing the burden among trauma surgery, neurosurgery, and neurology. J. Trauma Acute Care Surg.

34.

Chen

A.Y.

, Zagorski

, Parsons

, Vander Laan

, Chan

, and Colantonio

(2012). Factors associated with discharge destination from acute care after acquired brain injury in Ontario, Canada. BMC Neurol. 12, 16.

35.

Cooper

K.D.

, Tabaddor

, Allen Hauser

, †

K.S.

, Feiner

, and Factor

P.R.

(1983). The epidemiology of head injury in the Bronx. Neuroepidemiology, 2, 79–88.

36.

Jagger

, Levine

J.I.

, Jane

J.A.

, and Rimel

R.W.

(1984). Epidemiologic features of head injury in a predominantly rural population. J. Trauma, 24, 40–44.

37.

Annegers

J.F.

, Grabow

J.D.

, Kurland

L.T.

, and Laws

E.R.

Jr. (1980). The incidence, causes, and secular trends of head trauma in Olmsted County, Minnesota, 1935–1974. Neurology, 30, 912–919.

38.

Kaufman

H.H.

(1993). Civilian gunshot wounds to the head. Neurosurgery, 32, 962–964; discussion, 964.

39.

, and Baker

S.P.

(2012). Injury Research: Theories, Methods, and Approaches. Springer Science & Business Media: New York, NY.

40.

Zehtabchi

, Nishijima

D.K.

, McKay

M.P.

, and Mann

N.C.

(2011). Trauma registries: history, logistics, limitations, and contributions to emergency medicine research. Acad. Emerg. Med., 18, 637–643.

41.

Coronado

V.G.

, Xu

, Basavaraju

S.V.

, McGuire

L.C.

, Wald

M.M.

, Faul

M.D.

, Guzman

B.R.

, and Hemphill

J.D.

; Centers for Disease Control and Prevention (CDC). (2011). Surveillance for traumatic brain injury-related deaths—United States, 1997–2007. MMWR Surveill. Summ., 60, 1–32.

42.

[No authors listed]. (2000). Nonfatal and Fatal Firearm-Related Injuries—United States, 1993–1997. JAMA, 283, 47.

43.

Aarabi

, Hesdorffer

D.C.

, Ahn

E.S.

, Aresco

, Scalea

T.M.

, and Eisenberg

H.M.

(2006). Outcome following decompressive craniectomy for malignant swelling due to severe head injury. J. Neurosurg., 104, 469–479.

44.

Aldrich

E.F.

, Eisenberg

H.M.

, Saydjari

, Foulkes

M.A.

, Jane

J.A.

, Marshall

L.F.

, Young

, and Marmarou

(1992). Predictors of mortality in severely head-injured patients with civilian gunshot wounds: a report from the NIH Traumatic Coma Data Bank. Surg. Neurol., 38, 418–423.

45.

Ambrosi

P.B.

, Valença

M.M.

, and Azevedo-Filho

(2012). Prognostic factors in civilian gunshot wounds to the head: a series of 110 surgical patients and brief literature review. Neurosurg. Rev., 35, 429–436.

46.

Benzel

E.C.

, Day

W.T.

, Kesterson

, Willis

B.K.

, Kessler

C.W.

, Modling

, and Hadden

T.A.

(1991). Civilian craniocerebral gunshot wounds. Neurosurgery, 29, 67–71; discussion, 71–72.

47.

Kim

K.A.

, Wang

M.Y.

, McNatt

S.A.

, Pinsky

, Liu

C.Y.

, Giannotta

S.L.

, and Apuzzo

M.L.J.

(2005). Vector analysis correlating bullet trajectory to outcome after civilian through-and-through gunshot wound to the head: using imaging cues to predict fatal outcome. Neurosurgery, 57, 737–747; discussion, 737–747.

48.

Ordog

G.J.

, Wasserberger

, and Balasubramaniam

(1988). Shotgun wound ballistics. J. Trauma, 28, 624–631.

49.

Vincent

J.M.

, and DiMaio

M.D.

(2015). Gunshot Wounds: Practical Aspects of Firearms, Ballistics, and Forensic Techniques, 3rd ed. CRC: Boca Raton, FL.

50.

Jandial

, Reichwage

, Levy

, Duenas

, and Sturdivan

(2008). Ballistics for the neurosurgeon. Neurosurgery, 62, 472–480.

51.

Swan

K.G.

, and Swan

R.C.

(1985). Wound ballistics for the civilian surgeon. Surg. Annu., 17, 163–187.

52.

Cudnik

M.T.

, Newgard

C.D.

, Sayre

M.R.

, and Steinberg

S.M.

(2009). Level I versus Level II trauma centers: an outcomes-based assessment. J. Trauma, 66, 1321–1326.

53.

Nathens

A.B.

(2001). Relationship between trauma center volume and outcomes. JAMA, 285, 1164.

54.

MacKenzie

E.J.

, Rivara

F.P.

, Jurkovich

G.J.

, Nathens

A.B.

, Frey

K.P.

, Egleston

B.L.

, Salkever

D.S.

, and Scharfstein

D.O.

(2006). A national evaluation of the effect of trauma-center care on mortality. N. Engl. J. Med., 354, 366–378.

55.

Harbrecht

B.G.

, Zenati

M.S.

, Ochoa

J.B.

, Townsend

R.N.

, Puyana

J.C.

, Wilson

M.A.

, and Peitzman

A.B.

(2004). Management of adult blunt splenic injuries: comparison between level I and level II trauma centers. J. Am. Coll. Surg., 198, 232–239.

56.

Birkmeyer

J.D.

, Stukel

T.A.

, Siewers

A.E.

, Goodney

P.P.

, Wennberg

D.E.

, and Lucas

F.L.

(2003). Surgeon volume and operative mortality in the United States. N. Engl. J. Med., 349, 2117–2127.

57.

Hadley

(2007). Insurance coverage, medical care use, and short-term health changes following an unintentional injury or the onset of a chronic condition. JAMA, 297, 1073.

58.

Haas

J.S.

, and Goldman

(1994). Acutely injured patients with trauma in Massachusetts: differences in care and mortality, by insurance status. Am. J. Public Health, 84, 1605–1608.

59.

McWilliams

J.M.

, Zaslavsky

A.M.

, Meara

, and Ayanian

J.Z.

(2003). Impact of Medicare coverage on basic clinical services for previously uninsured adults. JAMA, 290, 757–764.

60.

Sorlie

P.D.

, Johnson

N.J.

, Backlund

, and Bradham

D.D.

(1994). Mortality in the uninsured compared with that in persons with public and private health insurance. Arch. Intern. Med., 154, 2409–2416.

61.

Reedy

D.C.

, and Koper

C.S.

(2003). Impact of handgun types on gun assault outcomes: a comparison of gun assaults involving semiautomatic pistols and revolvers. Inj. Prev., 9, 151–155.

62.

Carr

B.G.

, Schwab

C.W.

, Branas

C.C.

, Killen

, and Wiebe

D.J.

(2008). Outcomes related to the number and anatomic placement of gunshot wounds. J. Trauma, 64, 197–202; discussion, 202–203.

63.

Vyrostek

S.B.

, Annest

J.L.

, and Ryan

G.W.

([date unknown]). Surveillance for Fatal and Nonfatal Injuries—United States, 2001. PsycEXTRA Dataset.

64.

Pelosi

, Ferguson

N.D.

, Frutos-Vivar

, Anzueto

, Putensen

, Raymondos

, Apezteguia

, Desmery

, Hurtado

, Abroug

, Elizalde

, Tomicic

, Cakar

, Gonzalez

, Arabi

, Moreno

, and Esteban

, Ventila Study Group. (2011). Management and outcome of mechanically ventilated neurologic patients. Crit. Care Med., 39, 1482–1492.

65.

Rincon

, Ghosh

, Dey

, Maltenfort

, Vibbert

, Urtecho

, McBride

, Moussouttas

, Bell

, Ratliff

J.K.

, and Jallo

(2012). Impact of acute lung injury and acute respiratory distress syndrome after traumatic brain injury in the United States. Neurosurgery, 71, 795–803.

66.

Bronchard

, Albaladejo

, Brezac

, Geffroy

, Seince

P.-F.

, Morris

, Branger

, and Marty

(2004). Early onset pneumonia: risk factors and consequences in head trauma patients. Anesthesiology, 100, 234–239.

67.

Baigelman

, and O'Brien

J.C.

(1981). Pulmonary effects of head trauma. Neurosurgery, 9, 729–740.

68.

Piek

, Chesnut

R.M.

, Marshall

L.F.

, van Berkum-Clark

, Klauber

M.R.

, Blunt

B.A.

, Eisenberg

H.M.

, Jane

J.A.

, Marmarou

, and Foulkes

M.A.

(1992). Extracranial complications of severe head injury. J. Neurosurg., 77, 901–907.

69.

Hollerman

J.J.

, Fackler

M.L.

, Coldwell

D.M.

, and Ben-Menachem

(1990). Gunshot wounds: 1. Bullets, ballistics, and mechanisms of injury. AJR Am. J. Roentgenol. 155, 685–690.

70.

Fackler

M.L.

, Surinchak

J.S.

, Malinowski

J.A.

, and Bowen

R.E.

(1984). Bullet fragmentation. J. Trauma, 24, 35–39.

71.

Enam

, Kazim

, Tahir

, Waheed

, and Shamim

(2011). Management of penetrating brain injury. J. Emerg. Trauma Shock, 4, 395.

72.

Esposito

D.P.

, and Walker

J.B.

(2009). Contemporary management of penetrating brain injury. Neurosurg. Q., 19, 249–254.

73.

Blissitt

P.A.

(2006). Care of the critically ill patient with penetrating head injury. Crit. Care Nurs. Clin. North Am., 18, 321–332.

74.

Shafi

, Diaz-Arrastia

, Madden

, and Gentilello

(2008). Intracranial pressure monitoring in brain-injured patients is associated with worsening of survival. J. Trauma, 64, 335–340.

75.

Rish

B.L.

, Dillon

J.D.

, Caveness

W.F.

, Mohr

J.P.

, Kistler

J.P.

, and Weiss

G.H.

(1980). Evolution of craniotomy as a debridement technique for penetrating craniocerebral injuries. J. Neurosurg., 53, 772–775.

76.

Bell

R.S.

, Mossop

C.M.

, Dirks

M.S.

, Stephens

F.L.

, Mulligan

, Ecker

, Neal

C.J.

, Kumar

, Tigno

, and Armonda

R.A.

(2010). Early decompressive craniectomy for severe penetrating and closed head injury during wartime. Neurosurg. Focus, 28, E1.

77.

Nathens

A.B.

, Brunet

F.P.

, and Maier

R.V.

(2004). Development of trauma systems and effect on outcomes after injury. Lancet, 363, 1794–1801.

78.

Welch

W.P.

, Miller

M.E.

, Welch

H.G.

, Fisher

E.S.

, and Wennberg

J.E.

(1993). Geographic variation in expenditures for physicians' services in the United States. N. Engl. J. Med., 328, 621–627.

79.

MacKenzie

E.J.

(2003). National inventory of hospital trauma centers. JAMA, 289, 1515.

80.

Bazzoli

G.J.

, Madura

K.J.

, Cooper

G.F.

, MacKenzie

E.J.

, and Maier

R.V.

(1995). Progress in the development of trauma systems in the United States. Results of a national survey. JAMA, 273, 395–401.

81.

Killias

(1993). International correlations between gun ownership and rates of homicide and suicide. CMAJ, 148, 1721–1725.

82.

Hemenway

, and Miller

(2000). Firearm availability and homicide rates across 26 high-income countries. J. Trauma, 49, 985–988.

83.

Panczak

, Geissbühler

, Zwahlen

, Killias

, Tal

, and Egger

(2013). Homicide-suicides compared to homicides and suicides: systematic review and meta-analysis. Forensic Sci. Int., 233, 28–36.

84.

Kaplan

M.S.

, and Geling

(1998). Firearm suicides and homicides in the United States: regional variations and patterns of gun ownership. Soc. Sci. Med., 46, 1227–1233.

85.

Cook

P.J.

(1999). The medical costs of gunshot injuries in the United States. JAMA, 282, 447.

86.

Miller

T.R.

, and Cohen

M.A.

(1997). Costs of gunshot and cut/stab wounds in the United States, with some Canadian comparisons. Accid. Anal. Prev., 29, 329–341.

87.

Max

, and Rice

D.P.

(1993). Shooting in the dark: estimating the cost of firearm injuries. Health Aff. 12, 171–185.

88.

Annest

J.L.

(1995). National estimates of nonfatal firearm-related injuries Beyond the tip of the iceberg. JAMA, 273, 1749–1754.

89.

Reddy

A.K.

, Baker

M.S.

, Sobel

R.K.

, Whelan

D.A.

, Carter

K.D.

, and Allen

R.C.

(2014). Survivors of self-inflicted gunshot wounds to the head: characterization of ocular injuries and health care costs. JAMA Ophthalmol. 132, 730–736.