a b s t r a c t

Introduction: As cities nation-wide combat gun violence, with less than 20% of shots fired reported to police, use of acoustic gunshot sensor (AGS) technology is increasingly common. However, there are no studies to date in- vestigating whether these technologies affect outcomes for victims of gunshot wounds (GSW). We hypothesized that the AGS technology would be associated with decreased prehospital transport time.

Methods: All GSW patients from 2014 to 2016 were collected from our institutional registry and cross-referenced with local police department data regarding times and locations of AGS alerts. Each GSW incident was catego- rized as related or unrelated to an AGS alert. Admission data, trauma outcomes, and prehospital time were then compared.

Results: We analyzed 731 patients. Of these, 192 were AGS-related (26%) and 539 were not (74%). AGS-related patients were more likely to be female (p b 0.01), have a higher injury severity score (ISS) (p b 0.01), and require an operation (p = 0.03). ventilator days (p b 0.05) and hospital length of stay (p b 0.01) was greater in the AGS cohort. Mortality, however, did not differ between groups (p = 0.5). On multivariable analysis, both Total prehospital time and on-scene time were lower in the AGS group (p b 0.01).

Conclusion: Our study suggests reduced transport times, decreased prehospital and emergency medical service on-scene times with AGS technology. Additionally, despite higher ISS and use of more Hospital resources, mortal- ity was similar to non-AGS counterparts. The potential of AGS technology to further decrease prehospital times in the urban setting may provide an opportunity to improve outcomes in trauma patients with Penetrating injuries.

(C) 2019

1. Introduction

The United States has the highest homicide rate of all developed countries according to the United Nations Office on Drugs and Crime, with a rate of 4.9 per 100,000 inhabitants [1]. Most homicides in the US are committed with firearms [2], with firearm-related deaths ap- proaching 40,000 in 2016 [3]. In response to this, new enforcement strategies have been developed in order to help police departments both reduce the number of gun-related crimes while increasing arrests for these offenses. However, a barrier to the enforcement of gun-related crime is that less than 20% of gunshots are reported to police [4]. To ad- dress this problem, police departments nationwide have deployed a va- riety of strategies to increase gunshot identification capabilities and improve incident response time.

Abbreviations: AGS, acoustic gunshot sensor; AIS, Abbreviated Injury Scale; GSW, gunshot wound; ISS, injury severity score; OPD, Oakland Police Department; EMS, emergency medical services; LOS, length of stay; VIF, variance inflation factor.

? This paper was presented at the 2017 American College of Surgeons (ACS) Clinical

Congress Scientific Forum in San Diego, Ca.

* Corresponding author.

E-mail address: [email protected] (G. Beattie).

Acoustic gunshot sensor (AGS) technology has emerged as one of the strategies used by law enforcement agencies to increase the identi- fication of gunshots within a community and improve response time to the scene [5-7]. AGS technology also has been employed as a strategy to locate and target harmful environmental practices such as animal poaching [8] and blast fishing [9]. Application of AGS technology in the urban setting has many potential benefits including leading to an in- crease in the number of arrests for firearm discharges, act as a deterrent for future incidents, and “reduce the detrimental effects (e.g. injuries, fear, disinvestments) of shots being fired in urban settings [5]“. The Oak- land California Police Department adopted the use of AGS technology (ShotSpotter Inc., Newark, CA) in 2006 [10]. The ShotSpotter AGS sys- tem consists of acoustic sensors that are placed in strategic locations throughout the city (15-20 sensors per square mile) to allow the detec- tion and triangulation of gunshots. The sensors capture the time, loca- tion, and audio of any noise that might represent a gunshot. This data is then evaluated by a proprietary algorithm as well as by human ana- lysts at an “incident review center” to confirm whether each event is truly a gunshot. This process takes approximately 45 s. Once an incident has been confirmed as a gunshot, the location can be uploaded to a 911- dispatch center or pushed directly to officers in the field using a Smartphone application [6,7,11,12].

https://doi.org/10.1016/j.ajem.2019.10.042

0735-6757/(C) 2019

While the cost-effectiveness and ability of AGS technology to reduce crime rates remain controversial, it has yet to be determined what, if any, effect the system has on outcomes for victims of gun violence. We hypothesized that with earlier detection of gunshot wounds using AGS technology, Prehospital providers could be dispatched sooner to these victims, decreasing prehospital time and improving clinical outcomes.

1. Methods

After receiving Institutional Review Board approval, a retrospective chart review using our institutional trauma registry was performed on gunshot wound (GSW) patients presenting from 2014 to 2016. Using the Oakland Police Department’s (OPD) AGS database, the times, and lo- cations of all AGS alerts over the same two year time period were ob- tained. Each GSW incident in our institutional registry was compared to the OPD database in order to identify whether an AGS alert was logged within an hour prior to the patient’s arrival at the hospital by emergency transport or within 1.5 h if the patient self-presented. If a pa- tient’s GSW incident met criteria, the location of all AGS alerts within that time frame was entered into an online distance calculator applica- tion capable of calculating the straight-line distance between two points. The straight-line distance between the AGS alert locations and the trauma registry incident location were calculated. The GSW incident was considered to be AGS related if it took place within two city blocks, or approximately 200ft of the AGS alert location. This distance determi- nation was considered liberal and appropriate based on a field evalua- tion of the AGS system published in 2002 which found the system to be accurate at triangulating to a margin of error of between 40 and 50ft depending on weapon type. GSW incidents were considered AGS unrelated if they did not meet these timing and location proximity criteria.

Exclusion criteria were GSW events from our database with missing location and time data that prevented comparison to the AGS database. Given our study focus was acute injury prehospital times, inter-facility transfers and non-acute GSW patients were also excluded. Prehospital intervention and admission data, trauma outcomes, and Prehospital time intervals that are collected and maintained within our institutions’ trauma registry were compared between AGS related and AGS unre- lated groups. prehospital interventions, reported by EMS provider doc- umentation, included backboard and/or cervical collar placement, venous access, Oxygen administration, airway measures, pleural decom- pression, extremity splinting, cardiopulmonary resuscitation, defibrilla- tion and medication administration. Each prehospital intervention was compared individually between groups. Additionally, interventions were classified into basic measures (backboard, cervical collar, periph- eral venous access, crystalloid infusion, nasal cannula, bag bask, re- straints, medication administration) and advanced measures (oral airway, endotracheal tube, cardiopulmonary resuscitation, interosseous access, defibrillation) for further inter-group analysis. As a tool to com- pare injury type and severity, the Abbreviated Injury Scale score was analyzed between groups.

Outcome measures of prehospital time intervals included: (1) Total

prehospital time - time from emergency medical service (EMS) arrival on scene to hospital arrival, (2) Transport time - EMS trauma scene de- parture to hospital arrival (3) EMS on-scene time - time from arrival on scene to departure, (4) EMS response time - time until EMS arrived on scene (Fig. 1).

Pearson chi-squared test was used for comparison of categorical var- iables and t-test for continuous variables. Linear regression analysis was used to evaluate the impact of patient factors on the various prehospital time categories. Univariable analysis was performed evaluating the as- sociation of age, sex, ISS, initial GCS, major procedures, hospital length of stay , intensive care unit LOS, ventilator days, and mortality on prehospital time categories. Significant variables on univariable analy- sis, defined as p-value b0.2, included age, sex, ISS, and major procedures.

Fig. 1. Study population. Gunshot wound victims meeting study criteria were grouped into acoustic gunshot sensor (AGS) related events and non-AGS related events.

The significant variables were subsequently included in the multivari- able Linear regression models investigating the associations between prehospital times and AGS activations. Statistical significance was de- fined as a p-value b0.05 in the multivariate model. Variance inflation factors (VIF) were calculated to assess collinearity between the inde- pendent variables in our models. VIF measurements for the indepen- dent variables were b1.2, suggesting minimal collinearity within the models. Statistical analysis was performed using Stata 13 (StataCorp LP) and RStudio (Version 1.1.463, http://www.rstudio.com/) software.

1. Results

During the study period, a total of 893 patients with gunshot injuries were identified. Of these, 162 were excluded based on inability to ex- tract sufficient time or location data, transfer from another facility, or non-acute GSW. This left 731 who met our inclusion criteria of which 192 were found to be associated with an AGS activation (26%) and 539 were not (74%) (Fig. 1). There were no differences between the two groups in regards to age, prehospital transport method (EMS vs. self transport), Glasgow coma scale (GCS), initial systolic blood pressure (SBP), heart rate, or respiratory rate. Overall, the vast majority of pa- tients were male (85%) (Table 1). injury locations for both AGS and non-AGS events were mapped in relationship to the trauma center (Figs. 2 and 3). The geographic locations of AGS and Non-AGS events to the trauma center were similar.

Female patients were more likely to present as an AGS-related acti- vation than a non-activated gunshot patient (20.8% vs. 12.8%; p = 0.007). A greater percentage of AGS-related activations had an injury se- verity score (ISS) >=16 on presentation (61.3% vs. 36.5%, p = 0.004). Based on AIS score, severity of injury by body region was similar be- tween groups. However, patients associated with an AGS activation had a higher proportion of abdominal injuries (31% vs. 20%, p = 0.005) than non-AGS associated patients. There was no difference in proportion of head, face, chest, extremity or external injuries (p N 0.05) (Table 2).

Hospital length of stay was longer in the AGS group (8.4 days vs. 5.7 days; p = 0.006), as was the number of ventilator days (1.2 days

Table 1

AGS versus Non-AGS Related GSWs.

Variable AGS Related Non-AGS Related p-value

Demographics
Age	29.5 +- 0.8	30.3 +- 0.6	0.5
Gender (% female)	20.8	12.8	*0.007
EMS transport (%)	84.4	81.4	0.4
Clinical factors
GCS	13.6 +- 0.3	13.6 +- 0.2	0.9
ISS >= 16 (%)	61.3	36.5	*0.004
SBP	125.6 +- 3.0	124.8 +- 1.9	0.8
Heart rate	90.7 +- 2.1	89.9 +- 1.4	0.8
Respiratory rate	17.6 +- 0.5	17.4 +- 0.3	0.8
Outcomes
Length of stay (days)	8.4 +- 1.0	5.7 +- 0.5	*0.006
ICU days	2.2 +- 0.5	1.3 +- 0.2	0.05
Ventilator days	1.2 +- 0.3	0.7 +- 0.1	*b0.05
Operation required (%)	44.8	35.6	*0.03
Discharged home (%)	81.4	87.1	0.08
Mortality (%)	12.0	10.2	0.5
Prehospital time intervals (min)
Total Prehospital time	16.4 +- 0.4	17.9 +- 0.4	*0.03
EMS response time	7.0 +- 0.2	6.6 +- 0.2	0.2
EMS time on scene	6.5 +- 0.3	7.5 +- 0.3	*0.03
Transport time	9.9 +- 0.3	10.3 +- 0.2	0.2

AGS = Acoustic gunshot sensor; EMS = Emergency medical services; GCS = Glasgow comma scale; ISS = Injury severity scale; SBP = Systolic blood pressure in mmHg; ICU = Intensive care unit.

*Indicates statistical significance. Continuous results reported as mean +- standard error of the mean.

vs. 0.7 days; p = 0.046). There was a trend towards longer ICU stays in the AGS group, but this did not reach statistical significance. Patients who were associated with an AGS activation had a higher rate of oper- ative intervention compared to those who were not activated (44.8% vs. 35.6%; p = 0.03). There was no difference in mortality between the two study groups (12.0% vs. 10.2%; p = 0.5) (Table 1).

Prehospital interventions were recorded in 517 patients who made up the cohort for this analysis (AGS = 147, Non-AGS = 370). Frequen- cies of prehospital intervention by type and total intervention number per patient were similar between the AGS and non-AGS groups (p N 0.05, Table 3). There was no difference in numbers of basic and ad- vanced prehospital interventions in AGS and non-AGS patients (p N 0.05, Table 3).

Total prehospital time (16.4 min vs. 17.9 min; p = 0.03) and EMS on scene time (6.5 min vs. 7.5 min, p = 0.03.) were shorter for AGS activa- tions. In our multivariable regression model adjusting for significant pa- tient factors (age, sex, ISS, and major procedures), total prehospital time and on-scene time remained shorter in the AGS activation group (p = 0.007 and p = 0.004, respectively) (Tables 4a and 4b). Prehospital time ranged from 4 to 44 min in the AGS cohort and 5 min to 88 min in the non-AGS cohort. In the non-AGS cohort two patients had a total prehospital time beyond 60 min. Even after exclusion of these non- AGS outliers, EMS on-scene time remained shorter for AGS activations (p = 0.04). Additionally, on-scene time ranged from less than 1 min to 28 min in the AGS cohort and less than one minute to 86 min in the non-AGS cohort. In the non-AGS cohort two patients had an on-scene time beyond 30 min, with only one such on-scene time beyond 60 min. With exclusion of these non-AGS outliers, EMS on-scene time

Fig. 2. Acoustic Gunshot Sensor Injury Locations: Pins represent acoustic gunshot sensor injury locations in relation to the trauma center. Pins with numbers identify the number of gunshot wound events at a given injury location. Made with Maptive.com, geographic map generator.

Fig. 3. Non-Acoustic Gunshot Sensor Injury Locations: Pins represent non-acoustic gunshot sensor injury locations in relation to the trauma center. Pins with numbers identify the number of gunshot wound events at a given injury location. Made with Maptive.com, geographic map generator.

remained shorter for AGS activations (p = 0.03). EMS response time and transport times did not differ between activated and non- activated gunshot patients (p = 0.2 and p = 0.2, respectively).

1. Discussion

In the United States firearm related injuries continue to significantly impact public health, not only through premature death and disability, but economically. Alone, firearm related deaths and injury cost an esti- mated $50 billion per year in medical and lost productivity costs [13].A study of national firearm deaths from 2010 to 2012 found on average more than 32,000 people die each year from gunshot-related injuries (annual age-adjusted rate of 10.2 per 100,000) with 67,197 people per year receiving medical treatment (annual age-adjusted rate of 21.6 per 100,000) [13]. The city of Oakland, despite improving trends in gun violence, continues to remain one of the nation’s toped ranked vio- lent cities. During the study period of 2014, 2015, and 2016, Oakland re- ported non-fatal shooting and homicide totals as 495, 425 and 415, respectively [14]. To combat gun violence the city has employed a multi-faceted violence reduction strategy with the AGS system as an adjunct.

The AGS system offers many potential advantages to law enforce- ment as well as medical care providers. While earlier detection of

gunshots may improve Response times to violent crime incidents by law enforcement [5-7], the effectiveness of an AGS system on clinical outcomes for victims of gunshot-related trauma remains undetermined. Our hypothesis was that the AGS system would be associated with de- creased prehospital time and subsequently improved outcomes in vic- tims of gun violence. The results show improved prehospital and on- scene times for EMS providers responding to victims of gunshots that were detected by the AGS system. Furthermore, our data demonstrates the improvement in EMS on-scene and total prehospital times in the AGS cohort is not explained by differences in volume or type of prehospital interventions or anatomic region of injury, and that the sta- tistical significance of these findings remains after accounting for injury severity. These findings suggest a potential benefit of AGS technology on hospital transport times.

Prior research has demonstrated improved police response times to gunfire locations detected by AGS systems [6]. More rapid law enforce- ment response and scene containment has the potential to allow EMS providers to focus on efficient triage and minimization of on-scene time. This factor may have contributed to our findings of reduced on- scene time in the AGS associated patients. Additionally, despite the AGS cohort having a greater ISS, as well as longer hospital stays, more ventilator days, and more operative interventions, there was no differ- ence in mortality when compared to the non-activated gunshot victims.

Table 2

Body Region of Injury and Severity: AGS versus Non-AGS Related GSWs.

Table 4a

Multivariable Linear Regression for Total Prehospital Time.

AIS Injury Variable	AGS Related	Non-AGS Related	p-value		Characteristic	B	95% CI	p-value	VIF
	n = 191	n = 538			AGS activation	-1.66	[-2.86, -0.46]	0.007	1.01
Head					Age	0.03	[-0.02, 0.08]	0.2	1.01
Head Injury, n (%)	25 (13)	45 (8)	0.06		Sex	2.52	[0.83, 4.21]	0.003	1.01
AIS, %: >=2	100	100	1.0		ISS	0.004	[-0.05, 0.05]	0.9	1.15
AIS, %: >=3	88	92	0.7		Operation	-0.42	[-1.72, 0.89]	0.5	1.15
Face Face Injury, n (%)	13 (7)	38 (7)	1.0	R2 = 0.03, p = 0.002. CI = confidence interval for B. VIF = variance inflation factor. AGS = Acoustic gunshot sensor. ISS = injury severity scale.
AIS, %: >=2	92	90	1.0

cardiac [20] and thoracic injuries [21,22], hemodynamically unstable patients [17,18], those with higher injury severity [23] and poor proba- bility of survival [24]. In hypotensive patients with torso gunshot wounds a delay to operative intervention beyond 10 min increases the risk of mortality 3-fold [25]. Additionally, prior study of major Trauma victims demonstrated a 5% increased odds of dying for every minute of increased prehospital time [26], supporting that even the modest im- provement in prehospital times associated with AGS technology can positively impact trauma outcomes. Our findings demonstrate the pos- sibility of AGS utilization to lessen prehospital time in penetrating trauma potentially mitigating preventable traumatic death.

AIS, %: >=3 Chest	38	28	0.5
Chest Injury, n (%)	43 (23)	87 (16)	0.06
AIS, %: >=2	98	98	1.0
AIS, %: >=3	93	86	0.5
Abdomen Abdomen Injury, n (%)	59 (31)	110 (20)	*0.004
AIS, %: >=2	100	97	0.9
AIS, %: >=3	83	90	0.2
Extremity Extremity Injury, n (%)	83 (43)	194 (36)	0.08
AIS, %: >=2	93	94	0.8
AIS, %: >=3	61	66	0.5
External

In practice the AGS system appears more sensitive in detecting mul- tiple gunshot fires [5,7]. The greater ISS in the AGS cohort may be a re-

External Injury, n (%)	131 (69)	396 (74)	0.2	sult of the AGS system’s skewed detection of multiple gunshot events,
AIS, %: >=2	17	20	0.5	targeting those patients with more severe GSW related trauma. Addi-
AIS, %: >=3	4	2	0.2	tionally, there is likely higher potential for bystander injuries with mul-

AGS = Acoustic gunshot sensor; AIS = Abbreviated injury scale.

*Indicates statistical significance. Severity listed in percentages.

While a causal relationship cannot be assumed, the association seen with AGS technology between reduced prehospital times and possible advantageous patient outcomes may positively impact clinical out- comes in urban environments with high rates of penetrating trauma.

In the urban setting, where patients can be rapidly transported to dedicated trauma centers, the debate over minimization of prehospital times to improve patient outcomes in the undifferentiated trauma pa- tient persists [15-19]. However, in penetrating trauma reduced prehospital time has demonstrated improved survival in penetrating

Table 3

Pre-hospital Interventions AGS versus Non-AGS Related GSWs.

Pre-hospital Intervention	AGS Related	Non-AGS Related	p-value
	n = 147 (%)	n = 370 (%)
Backboard	26 (18)	83 (22)	0.3
Cervical Collar	28 (19)	88 (24)	0.3
PIV	115 (78)	304 (82)	0.4
Crystalloid	117 (80)	304 (82)	0.6
Nasal Cannula O2	21 (14)	36 (10)	0.2
NRBM/Bag Mask	63 (43)	152 (41)	0.8
Oral Airway	7 (5)	19 (5)	1.0
ETT	4 (3)	15 (4)	0.6
CPR	10 (7)	30 (8)	0.8
Restraints	1 (0.7)	3 (0.8)	1.0
Extremity Splint	3 (2)	4 (1)	0.7
IO	8 (5)	19 (5)	1.0
Pleural Decompression	3 (2)	10 (3)	0.9
Defibrillation	0 (0)	1 (0.3)	1.0
Medication Administration	2 (1)	17 (5)	0.1
Intervention #, mean (range)
Total	2.8 (0-7)	2.9 (0-10)	0.3
Basic*	2.5 (0-5)	2.7 (0-5)	0.3
Advanced**	0.2 (0-4)	0.3 (0-5)	0.7

AGS = Acoustic gunshot sensor; PIV = Peripheral venous access; NRBM = Non- rebreather mask; ETT = Endotracheal tube; CPR = cardiopulmonary resuscitation; IO = Interosseous access.

*Basic pre-hospital interventions include: backboard, cervical collar, PIV, crystalloid infu- sion, nasal cannula, bag bask, restraints, medication administration.

**Advanced pre-hospital interventions include: oral airway, ETT, CPR, IO, defibrillation.

tiple gunshot events, which may account for the higher proportion of female victims in the AGS group (though males made up the majority of GSW victims in both groups). Males are more likely overall to be vic- tims of firearm violence and face a six times higher Death rate due to gunshot injury as a result [3]. However evaluation of penetrating trau- matic injuries in women demonstrate crossfire incidents (bystander in- jury) at a rate if 22% [27]. We suspect that females were often bystanders during multiple gunshot events and therefore, were more likely to be included in the AGS group.

There are several limitations of our study, including the retrospec- tive study design. In particular, location and time data relied on docu- mentation by EMS providers during transport. It is unknown why this data was not recorded for some patients, possibly resulting in bias with exclusion of these patients. However, the lack of this data occurs for both AGS and non-AGS patients and would not lend to selective bias in one group. There are questions regarding the accuracy of the AGS system itself. A field study conducted in 2002 showed that the AGS system was able to detect gunshots in 81% of trial events and trian- gulate these gunshots in 84% of the events [5]. However, this was a rel- atively small sample size (31 events) with a homogenous mix of gun types tested. When broken down into shotguns, handguns, and assault rifles, the system was able to detect shotgun fire at a rate of 90%, pistol fire at a rate of 85%, and assault rifle fire at only 63%. Additionally, the system was most accurate only when multiple shots were fired. Only 50% of one-shot events were detected by the AGS system. Despite this data, the study concluded that the AGS system “could contribute to the reduction and prevention of community crime and disorder [5].” In our study, we found that though much of Oakland is covered by the

Table 4b

Multivariable Linear Regression for On-Scene Time.

Characteristic	B	95% CI	p-value	VIF
AGS activation	-1.22	[-2.06, -0.39]	0.004	1.01
Age	0.03	[0.00001, 0.06]	0.05	1.01
Sex	0.69	[-0.49, 1.86]	0.3	1.01
ISS	0.008	[-0.03, 0.04]	0.5	1.15
Operation	-0.73	[-1.64, 0.19]	0.1	1.15

R² = 0.03, p = 0.004. CI = confidence interval for B. VIF = variance inflation factor. AGS = Acoustic gunshot sensor. ISS = injury severity scale.

system, only 26% of our cases were connected with an AGS activation, bringing up the question of accuracy in the field. In addition, the AGS technology utilized in the city of Oakland during the study period was limited to outdoor gunfire detection. Gunshot wound victims may also move or be moved after the event, which could further limit AGS local- ization in real time. However, even if the AGS system is skewed towards detection of outdoor multiple gunshot events and victims remaining on scene, improved transport times among those with extensive gunshot trauma may significantly impact preventable trauma related mortality. Our data supports this given no increased mortality in the AGS cohort despite a higher baseline injury severity.

The results of our study reinforce the feasibility to further decrease prehospital times in the urban setting, providing an opportunity to im- prove outcomes in trauma patients with penetrating injuries. In partic- ular, the AGS system appears to provide this benefit to victims of GSW related trauma by reducing EMS prehospital and on-scene times. AGS technology may be of most use in victims of multiple GSW injuries, who otherwise would have likely faced a higher mortality rate.

1. Conclusion

Our study suggests reduced transport times, on-scene and total prehospital times, with AGS technology. In severe penetrating trauma, where reduced prehospital time provides additional Survival benefit, AGS employment may be a method to minimize Preventable death. Fu- ture studies examining the use of this technology are warranted to see if further implementation and improvement in AGS systems would be of benefit in areas with high rates of gun violence.

Disclosures

The authors have no disclosures to make.

Acknowledgements

None.

Funding

This research did not receive any specific grant from funding agen- cies in the public, commercial, or not-for-profit sectors.

References

1. United Nations Office on Drugs and Crime (UNODC). Global Study on Homicide 2013. United Nations Publication, Sales No. 14.IV.1. 2014. Accessed December 5, 2018.
2. Cooper A, Smith E. Homicide in the U.S. known to law enforcement, 2011. Bureau Justice Stat 2013:1-17.
3. Xu J, Murphy SL, Kochanek KD, Bastian B, Arias E. Deaths:final data for 2016. CDC Na- tional Vital Statistics Reports. 2018;67(5):1-76. Available at https://www.cdc.gov/ nchs/data/nvsr/nvsr67/nvsr67_05.pdf. Accessed December 4 2018.
4. Carr JB, Doleac JL. The geography, incidence, and underreporting of gun violence: new evidence using shotspotter data. Washington, DC: Brookings Institution; 2016; 1-17.
5. Watkins C, Mazerolle LG, Rogan D, Frank J. Technological approaches to controlling random gunfire: results of a gunshot detection system field test. Policing: Int J Police Strat Manag 2002;25:345-70.
6. Choi KS, Librett M, Collins TJ. An empirical evaluation: gunshot detection system and its effectiveness on police practices. Police Pract Res 2014;15(1):48-61.
7. Mazerolle LG, Watkins C, Rogan D, Frank J. Random gunfire problems and gunshot detection systems. National Institute of Justice Research in Brief 1999:1-9.
8. Gonzalez-Castano FJ, Alonso JV, Costa-Montenegro E, Lopez-Matencio P, Vicente- Carrasco F, Parrado-Garcia FJ, et al. Acoustic sensor planning for gunshot location in national parks: a pareto front approach. Sensors 2009;9:9493-512.
9. Showen R, Dunson C, Woodman GH, Christopher S, Lim T, Wilson SC. Locating fish bomb blasts in real-time using a networked acoustic system. Mar Pollut Bull 2018; 128:496-507.
10. “City of Oakland ShotSpotter Report.” City of Oakland. Available at: http://www2. oaklandnet.com/government/o/OPD/a/data/shotspotter/index.htm. Accessed De- cember 4, 2018.
11. “Contributing to positive outcomes.” ShotSpotter, http://www.shotspotter.com.
12. “Technology.” ShotSpotter. Available at: http://www.shotspotter.com. Accessed De- cember 4, 2018.
13. Fowler KA, Dahlberg LL, Haileyesus T, Annest JL. Firearm injuries in the United States.

Prev Med 2015;79:5-14.

1. Muhammad D. Oakland’s successful gun violence reduction strategy. National Insti- tute for Criminal Justice Reform. 2018:1-9. Available at: http://nicjr.org. Accessed October 3, 2019.
2. Smith MR, Conn AKT. Prehospital care - scoop and run or stay and play? Injury Int J Care Injured 2009;40S4:S23-6.
3. Liberman M, Mulder D, Lavoie A, Denis R, Sampalis JS. Multicenter Canadian study of prehospital trauma care. Ann Surg 2003;237(2):153-60.
4. Harmsen AMK, Giannakopoulos GF, Moerbeek PR, Jansma EP, Bonjer HJ, Bloemers FW. The influence of prehospital time on trauma patients outcome: a systematic re- view. Injury Int J Care Injured 2015;46:602-9.
5. Brown JB, Rosengart MR, Forsythe RM, Reynolds BR, Gestring ML, Hallinan WM, et al. Not all prehospital time is equal: influence of scene time on mortality. J Trauma Acute Care Surg 2016;81:93-100.
6. Ruelas OS, Tschautscher CF, Lohse CM, Sztajnkrycer MD. Analysis of prehospital scene times and interventions on mortality outcomes in a national cohort of pene- trating and blunt trauma patients. Prehosp Emerg Care 2018;22:691-7.
7. Gervin AS, Fischer RP. The importance of prompt transport in salvage of patients with penetrating heart wounds. J Trauma 1982;22:443-8.
8. Ivatury RR, Nallathambi MN, Roberge RJ, Rohman M, Stahl W. Penetrating thoracic injuries: infield stabilization vs. prompt transport. J Trauma 1987;27:1066-73.
9. Swaroop M, Straus DC, Agubuzu O, Esposito TJ, Schermer CR, Crandall ML. Pre- hospital transport times and survival for hypotensive patients with penetrating tho- racic trauma. J Emerg Trauma Shock 2013;6(1):16-20.
10. McCoy CE, Menchine M, Sampson S, Anderson C, Kahn C. Emergency medical ser- vices out-of-hospital scene and transport times and their association with mortality in trauma patients presenting to an urban Level I trauma center. Ann Emerg Med 2013;61(2):167-74.
11. Feero S, Hedges JR, Simmons E, Irwin L. Does out-of-hospital EMS time affect trauma survival? Am J Emerg Med 1995;13(2):133-5.
12. Meizoso JP, Ray JJ, Karcutskie 4th CA, Allen CJ, Zakrison TL, Pust GD, et al. Effect of time to operation on mortality for hypotensive patients with gunshot wounds to the torso: the golden 10 minutes. J Trauma Acute Care Surg 2016;81:685-91.
13. Sampalis JS, Denis R, Lavoie A, Frechette P, Boukas S, Nikolis A, et al. Trauma care re- gionalization: a process-outcome evaluation. J Trauma 1999;46(4) 565-79-81.
14. Jacovides CL, Bruns B, Holena DN, Sims CA, Wiebe DJ, Reilly PM, et al. Penetrating trauma in urban women: patterns of injury and violence. J Surg Res 2013;184(1): 592-8.