Journal of Trauma and Acute Care Surgery:
AAST 2013 Plenary Papers
Unrelenting violence: An analysis of 6,322 gunshot wound patients at a Level I trauma center
Lavery, Robert F. MA; Lopreiato, Maeve C. MPH; Lavery, David F. JD; Passannante, Marian R. PhD
From the New Jersey Trauma Center and Departments of Surgery Division of Trauma and Critical Care (D.H.L., R.F.L., M.C.L., D.F.L.) and Preventive Medicine and Community Health (M.R.P.), Rutgers-New Jersey Medical School, Newark, New Jersey.
Submitted: September 15, 2013, Revised: October 7, 2013, Accepted: October 9, 2013.
This study was presented at the 72nd annual meeting of the American Association for the Surgery of Trauma, September 18–21, 2013, in San Francisco, California.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text, and links to the digital files are provided in the HTML text of this article on the journal’s Web site ( www.jtrauma.com).
Address for reprints: David H. Livingston, MD, University Hospital, M234 150 Bergen Street Newark, New Jersey 07103; email: firstname.lastname@example.org.
Perceptions of violence are too often driven by individual sensational events, yet “routine” gunshot wound (GSW) injuries are largely underreported. Previous studies have mostly focused on fatal GSW. To illuminate this public health problem, we studied the health care burden of interpersonal GSW at a Level I trauma center.
Retrospective analysis of GSW injuries (excluding self and law enforcement) treated from January 2000 to December 2011. Data collected included body regions injured, number of wounds per patient, and mortality. Costs were calculated using Medicare cost-charge modifiers. Geographic information system mapping of the incident location and home addresses were determined to identify hot spot locations and the characterization of those neighborhoods.
A total of 6,322 patients were treated. There were significant increases in patients with three or more wounds (13–22%, p < 0.0001) and three or more body regions injured (6–16%, p < 0.0001). Mortality increased from 9% to 14% (p < 0.0001). Nineteen percent of the patients were never seen by the trauma service. Geographic information system mapping revealed significant clustering of GSWs. Five cities accounted for 85% of the GSWs, with rates per 100,000 ranging from 19 to 108 compared with a national rate of 20. Only 19% of the census tracts had no GSWs during the period, and 39% of the census tracts had at least one GSW per year for 12 years. Fifteen percent of the census tracts accounted for 50% of the GSWs. Seventy percent of the patients were shot in their home city, 25% within 168 m, and 55% within 1,600 m of their home. Total inpatient cost was $115 million, with cost per patient increasing more than three times over the course of the study; 75% were unreimbursed.
GSW violence remains a significant public health problem, with escalating mortality and health costs. Relying on trauma registry data seriously underestimates GSW numbers. In contrast to episodic mass casualties, routine GSW violence is geographically restricted and not random. To combat this problem, policy makers must understand that the determinants of firearm violence reside at the community level.
LEVEL OF EVIDENCE
Epidemiologic study, level II.
Gun violence continues to be a major public health problem in the United States. Unfortunately, the issue is clouded and intertwined with constituents from all sides of the controversy on gun control and access to firearms, turning a public health problem into a political debate. Further confusing the subject is that many people’s view on gun violence is shaped from the media’s coverage of individual and isolated events involving mass casualties. In reality, of the 31,000 annual fatal gunshot wound (GSW) injuries, a little more than half are self-inflicted, with the remainder being “routine” events that occur across the cities and municipalities across the United States.1
The difficulty in formulating policy to combat interpersonal firearm violence has been the lack of reliable data and obstacles to serious research. Since 1996 through congressional appropriation mechanisms, the Centers for Disease Control (CDC) stopped funding research into firearm injuries.2 Few if any other agencies were interested in supporting firearms research, and in 2011, additional congressional legislation formally extended the CDC ban to all other federal agencies such as the National Institutes of Health. This lack of research leaves policymakers with few hard data with which to formulate strategies. Existing databases often contained data primarily on homicides, which represent a small fraction of all GSW patients.
The New Jersey Trauma Center at University Hospital in Newark has had a long interest in gun violence. Firearm injuries have accounted for more than 15% of our trauma patient population for more than the past three decades. To begin to understand and illuminate this public health problem, we studied the health care burden of interpersonal GSW injuries at a Level I trauma center. We specifically attempted to define and describe the extent of the population involved, its impact on trauma center resources and finances, and the characteristics of the neighborhoods involved with the hope that this information could be used to develop effective strategies that could begin to decrease interpersonal firearm injuries.
PATIENTS AND METHODS
We conducted a retrospective review of all patients treated for GSW injuries at the New Jersey Trauma Center at University Hospital in Newark, New Jersey, from January 1, 2000, until December 31, 2011. The population of interest was those patients who sustained GSW secondary to interpersonal violence. Patients were excluded if they sustained self-inflicted GSW (suicide or unintentional), were injured by law enforcement (LE), or sustained injury from BB or pellet guns. LE officers who sustained non–self-inflected GSW injuries were included. GSW patients were identified using the trauma registry and through the hospital emergency department (ED) financial database. The medical records of all patients identified though ED billing records were reviewed to make certain that they represented a new and unique patient and not a subsequent visit of a previously identified patient or one that met exclusion criteria. Data collected included demographics, body regions injured and number of wounds, Injury Severity Score (ISS), need for and type of operative intervention, inpatient disposition, and mortality. The data on the number of wounds have been prospectively captured in the trauma registry since 1997. For the analysis of the number of wounds, we excluded those patients injured by shotgun. The number of GSWs was defined as follows. “Through-and-through” injuries or those where the missile tract connecting entrance and exit wounds was documented or could be identified were counted as one GSW. Cases where wounds that could not be definitively connected whether they were in the same or different body compartments were counted as separate GSWs. Scene location was obtained from the emergency medical service runsheets. To investigate incidents with multiple patients where patients often arrive both by ambulance and private vehicle, we defined a cluster as two or more patients who arrive within 15 minutes of each other.
Hospital cost per patient was obtained using net charges for the population. The net charges were then converted to net costs using the cost-to-charge ratio provided by the hospital accounting department. The consumer price index for medical services was annualized for the period 2000 to 2011.3 The net percentage change in the index was then compared with the net percentage change in patient charges for that same period. Physician, postdischarge outpatient, and any subsequent readmission costs were not captured, and all cost data are solely related to the index hospital encounter.
GSW scene and the patient home locations were extracted from the trauma registry and ED record and mapped by converting the locations to latitude and longitude using an online geocoding service.4 Points were aggregated to 2010 US Census tracts and municipalities in New Jersey to determine population incidence rates by both home and scene. Points were aggregated to 2010 US Census tracts in Essex and Union county and to municipalities in New Jersey to determine population incidence rates by both home and scene. The US Census Bureau defines census tracts as relatively permanent statistical subdivisions of a larger geographic location designed to be relatively homogeneous units with respect to population characteristics, economic status, and living conditions and usually contain between 1,200 and 8,000 with an average of approximately 4,000 inhabitants.5 Census tract polygon files for 2010 were obtained from the US Census Bureau, as were 2010 population estimates. Geographic information system (GIS) data warehouse polygon files for municipality and county boundaries as well as land use polygon files were obtained from the New Jersey Department of Environmental Protection Agency. All mapping, calculation of spatial statistics, and spatial analyses were performed using ArcMap version 10.0 (ESRI, Redlands, CA).6
Kernel density maps were generated, and cluster analysis was performed. The average nearest neighbor statistic was calculated for scene and home maps, which compares actual mean distances between points to a hypothetical random distance to determine if the points are clustered. To identify incidence-based census tract–based clusters, the Anselin Local Moran’s I statistic was calculated for incidence maps (significance at α = 0.05). The Anselin Local Moran’s I tool works by identifying spatial clusters of features with attribute values similar in magnitude based on a weighted feature (incidence). The band collection tool, which computes correlation and covariance between raster maps (here generated as kernel density maps) to determine if they are of equal distribution, was used to compare the point distributions of homes by selected features.6
Statistical analysis for this study was performed using SAS 9.2 (SAS Institute, Cary, NC), SPSS 20 (IBM, Armonk, NY), and MATLAB 8.1 (MathWorks, Natick, MA). Continuous data are presented as mean and SD, and categorical data are presented as numbers and percentages. χ2 and Kruskal-Wallis tests were used when comparing categorical data, and t tests and analysis of variance were used when comparing continuous data. A stepwise logistic regression was used to investigate whether there were associations in mortality associated with the GSW (yes/no) with various independent variables (e.g., age, ISS, number of wounds). Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) are reported.
During the 12-year study period, the New Jersey Trauma Center treated 6,613 patients sustaining GSW. Two hundred ninety patients were excluded (41 injured by LE, 128 self-inflicted suicide, and 122 self-inflicted unintentionally). The remaining 6,322 patients were patients of interpersonal firearm violence and are the study population of interest. Ninety-eight percent were injured by handguns. The mean number of patients treated was 527 per year (range, 389–640). The temporal changes in GSW are can be found in the Figure, Supplemental Digital Content 1 ( http://links.lww.com/TA/A319). While there are peaks and valleys, there is a constant flow of patients during the time frame of the study, ranging from 1.06 to 1.75 patients per day.
The population was overwhelmingly male (92%) and young (27  years). The distribution of race/ethnicity was as follows: black (86%), Hispanic (9%), white (4%), and Asian (1%). Compared with the demography of the region, there was a significant and disproportionate representation by young African-American males. The age distribution of the patients is shown in Figure 1. One half of the population was between the ages of 20 years and 29 years. Most of those in the teenage cohort (ages 15–19 years) tended to be older (mean age, 18 years). There were no significant changes over time for age (p = 0.82), sex (p = 0.66), and race/ethnicity (p = 0.17) during the 12-year study period.
Of note, 19% of the patients identified were not captured by the trauma registry but were found only through the analysis of ED billing records. The majority of patients had very peripheral or tangential GSWs, although 9% were admitted to nontrauma services. Thus, relying on trauma registry data alone would have seriously underestimated the overall volume of interpersonal GSW violence by one fifth and decreased the number of GSW clusters recorded. Of the 5,105 patients seen by the trauma service, 43% were admitted, 9% were dead on arrival or died in the trauma resuscitation area, and 37% were treated and released from the ED.
There was a significant escalation in the extent and severity of injury observed during the period. The percentage of patients with three or more wounds, indicating two or more GSWs, increased from 10% in 2000 to 2001 to 23% in 2011 (p < 0.0001). In addition, the percentage of patients with injuries to three or more Abbreviated Injury Scale (AIS)–defined body regions also doubled (p < 0.001) (Fig. 2). Overall mortality for the entire patient population was 11% with a significant increase over time (9–14%; p < 0.0001) (Fig. 3). Further analysis of serious injury (AIS score ≥ 3) patterns demonstrate that the incidence of brain and spinal cord injuries almost doubled. This increase is important as the overall mortality of those patients with a spinal cord injury was 11% and of those with a brain injury was 65%. While the AIS severity for the extremity body region did not change, the incidence of simultaneous upper- and lower-extremity injuries more than doubled from 4% to 9% (p < 0.001). The percentage of patients sustaining abdominal injuries had a smaller but still significant increase compared with chest, peripheral vascular, and major truncal vascular injuries.
A hierarchical logistic regression was performed on the data to identify the impact of various independent variables (age, ISS, number of wounds, maximum AIS by body region, ventilator days, and blood units administered) on mortality. Patients with shotgun wounds were excluded from this analysis. Overall, the model was able to correctly predict 93% of the population’s status on discharge, although it was clearly better in predicting survivors (97% of the survivors) compared with nonsurvivors (57% of the deaths). Mortality was significantly associated with older patients (OR, 1.03; 95% CI, 1.01–1.04), blood units administered (OR, 1.04; 95% CI, 1.03–1.05), head AIS score (OR, 2.64; 95% CI, 2.46–2.83), chest AIS score (OR, 2.17; 95% CI, 2.04–2.31), abdominal AIS score (OR, 1.39; 95% CI, 1.03–1.55), and multiple wounds (OR, 1.11; 95% CI, 1.06–1.17).
The effect of this escalating injury severity on the trauma center was substantial. Cluster analysis demonstrates that there was a significant increase in two or more patients arriving within 15 minutes of another (Fig. 4). A further analysis demonstrates that 9% of GSW clusters involve three or more patients and the incidence of these larger clusters increased from 1% in 2000 to 16% in 2011. There was also a noticeable shift in the timing of arrival for the GSW patients, with a higher proportion arriving during “primetime” hours between 4:00 pm and 12:00 midnight (46% in 2000 vs. 61% in 2011) compared with after midnight (45% in 2000 vs. 30% in 2011).
Seventy-five percent of all patients admitted required at least one operative procedure during their hospitalization. There were no changes in the need for or the type of procedure performed during the study period. The total number of procedures per patient was not captured, and it is unknown whether the aggregate number of operative intervention changed over time.
The mean and median intensive care unit (ICU) and hospital length of stays, ventilator days, and blood products administered are shown in the Table, Supplemental Digital Content 2 ( http://links.lww.com/TA/A320). During the course of the study, an increased number of patients required ICU resources (83 in 2003 vs.187 in 2011) and blood products (56 in 2000 vs. 110 in 2011). In addition to more patients requiring ICU care, their individual lengths of stay increased over time (Fig. 5). The number of blood and blood products followed a more variable pattern. The upswing in the total number of blood products per patient administered after 2005 was related to the institution of a massive transfusion protocol that improved delivery of blood and blood products to the patient. Thus, those patients who required blood products received more units. Eighty-two percent of all patients were discharged to home, with 4% going to rehabilitation and 5% going to correctional facilities.
The total cost for this patient population was $115 million. The cost for admitted patients rose 323% over time (Table, Supplemental Digital Content 3, http://links.lww.com/TA/A321), which is significantly greater than the annualized change in medical care inflation (153%) during the same period.3 Seventy-five percent was unreimbursed (Table, Supplemental Digital Content 4, http://links.lww.com/TA/A322). The data on reimbursement were obtained through hospital finance and the physician practice plan data of the New Jersey Medical School and likely is the “best-case scenario” with respect to potential recoupable dollars. Patients requiring operative intervention are more resource intensive than those who do not. Thus, we analyzed the costs in those patients requiring operation (Table, Supplemental Digital Content 5, http://links.lww.com/TA/A323). Not surprisingly, patients with neurosurgical or damage-control laparotomy were most costly, although there was considerable variability observed in all categories.
Of the GSW injuries in the data set, 5,964 (90%) had home address and 3,918 (59%) had scene data sufficient to be mapped. The distance from scene to home was calculated for all patients (3,667, 55%) who had both locations known and the data mapped. The average distance was 1.92 miles (95% CI, 1.75–2.08; range 0–107). Among Essex and Union County residents, the distance from home to scene did not change significantly during the 12-year period (annual mean range, 1.12–1.42 miles; p = 0.68). Seventy percent of the patients were shot in the city where they lived. More interestingly, 15% were injured at their home address, one quarter were shot within one city block (168 m), and 55% within a mile of their home.
More than 90% of mapped home locations fell within a 6-mile radius of the trauma center, and five cities (Newark, Irvington, East Orange, Elizabeth, and Orange) accounted for more than 87% of the injured patients (Table, Supplemental Digital Content 6, http://links.lww.com/TA/A324). The GSW rate (per 100,000 residents) in those cities ranged from 19 to 108, with two cities, Newark and Irvington, having a mean annual incidences greater than 100 injuries per 100,000 population compared with the national average of 20.1,7
When examining the census tract data, the 12-year cumulative incidence of GSW injury rates ranged from 0 to 39.5 per 1,000 (Fig. 6). With the use of the home address, there were 60 census tracts (19%) in Essex and Union counties free of firearm injury during the entire study period. Conversely, 39% of the census tracts had at least one shooting per year during the study period, and 15% of the census tracts accounted for just less than half of all GSW injuries. With the use of the GIS analysis, neither home nor scene address had any bearing on the GSW injury severity pattern or mortality for any period.
The 12-year scene point map and 12-year home point map were clustered based on average nearest neighbor results (p < 0.05). The 12-year mean incidence rate per census tract was used for cluster identification using the Anselin Local Moran’s I tool. High incidence clustering occurred in neighborhoods covering central and southwest Newark as well as neighboring areas in Irvington and East Orange (Fig. 6). Seventy-two (23%) of the census tracts in the mapped area were in the highly clustered area (12-year incidence range, 9–40 injuries per 1,000 residents), and one outlier was found (a census tract with a much lower than expected incidence rate given its high-incidence neighbors). The clustered high-incidence census tracts were smaller in size and population than non–high-incidence census tracts. Population density was lower in high-incidence tracts, but both housing density and house vacancy were greater in high-incidence census tracts. Census tracts with high-incidence clustering had greater proportions of black/African-American residents, female-headed households, and residents living below poverty as well as lower education attainment and lower median household incomes. Non–high-incidence clustering census tracts had an older median age among residents and a greater proportion of residents 65 years or older.
The care of patients of firearm violence is an integral and unfortunately somewhat unique aspect of the trauma surgeon’s practice. The United States far outstrips most other countries in both the number of firearms and the number of firearm patients.7 The data presented in our study provide a 12-year analysis of the effect of interpersonal firearm violence on the individual, the trauma center, and the neighborhood from which it originated. It also demonstrates that the “day-to-day” firearm violence is far different from how it is portrayed in the media as it is neither episodic nor random but rather constant and circumscribed by geography and demography.
The data presented also begin to fill in an important gap in our knowledge of GSW violence. In 1996, the US Congress interdicted the CDC from funding studies in firearm violence.2 This lack of funding led to a stagnation in scholarship on this topic.8 The ban on federal funding of firearm related research was later extended to all federal agencies. It is only in January 2013 that President Obama through executive orders directed federal agencies to improve basic knowledge related to (1) causes of firearm violence; (2) interventions, which might be effective; and (3) strategies to minimize its public health burden. This resulted in an Institute of Medicine consensus report outlining future strategies to combat gun violence.9 In addition, the CDC was ordered to resume support of firearm-related violence research. Previously, health care providers and policy makers, faced with more than 30,000 deaths annually, were forced to make medical and policy decisions in a vacuum, the antithesis of evidence-based medicine.
Without robust funding, researchers in this arena were relegated to use existing administrative databases. In contrast, the majority of the data here was collected prospectively and included all patients injured by firearms. One commonly used database is the National Violent Death Reporting System (NVDRS). One limitation of the NVDRS database is that it is populated with only those patients sustaining fatal injuries. As these patients are more easily captured by various administrative databases, they may be overweighted as to their importance. Other databases, such as the National Electronic Injury Surveillance System, are a national probability sample of hospital EDs but provide no information on any inpatient hospitalizations following their ED visit. Estimates of nonfatal firearm injuries caused by interpersonal violence were difficult to ascertain and varied widely. In one estimate of firearm violence from 1993 to 1998 using these administrative databases, the nonfatal-to-fatal case ratio was recorded as 4:1.10 Another using the National Electronic Injury Surveillance System database estimated the rate at 2.6:1.11 A more recent study readily admitted that the data on nonfatal injuries were difficult to discern, but their estimate of the nonfatal-to-fatal case ratio was only 2:1.12 In contrast, a New Mexico ED surveillance system found that only 4% of all firearm injuries between 2008 and 2010 died.13 In the data reported here, death occurred in only 10% of unintentional firearm injuries and is much closer to more recent data by the CDC, which estimated that nonfatal GSWs account for just fewer than 80% of all GSW injuries.7 More importantly, using trauma registry data alone or an administrative database such as the National Trauma Data Bank would have missed 19% of all GSW patients and 26% of nonfatal injuries treated at our institution.
While our data found no demographic or geographic differences between the patients who died and those who survived, the overweighting of fatal GSWs in various databases may introduce unknown degrees of bias. NVDRS data are also only available from a limited number of states and are not necessarily nationally representative. Most importantly, the NVDRS data are dependent on partnerships among state health department NVDRS teams, medical examiners, and multiple LE departments and jurisdictions. Data sharing and communication among those entities responsible for providing data to accurately populate the NVDRS registry can be problematic. One example is a recent analysis of gang homicides in five cities across the United States.14 In that study, gang activity in Newark was reported to be responsible for only 10% of all firearm homicides, and only 20% were related to drugs. These data belie our experience at the New Jersey Trauma Center. Thus, while the NVDRS and other administrative databases are important, their completeness and limitations should serve as an impetus for improved data collection as it specifically relates to nonfatal gun violence.
The increase in number of wounds per patient as well as the increase number of patients per incident is also concerning. A study of firearm deaths in Philadelphia 20 years ago found a similar trend. At that time, they found significant changes in weaponry from revolvers, which traditionally hold six rounds to automatic handguns with a 50% increase in the number of bullets available.15 Our data show a similar trend with even more wounds per patient and suggest several possible explanations for the increase. The first could be that perpetrators have improved marksmanship. Absent improved accuracy, the increase can only be explained by an increased number of bullets fired during the incident. These increased numbers of bullets can be caused by an increased number of perpetrators per incident, the ability to fire more bullets per incident, or both. An increase in the number of perpetrators cannot be discounted as the crucial linkage between the medical and LE data is missing. The other possibility is related to the trend toward routine increased magazine size (≥13 bullets). It should be noted that this increase in magazine size is not to be confused with the true high capacity clips (e.g., 30 rounds), which were used in several of the episodic mass casualty shootings.
The effect of the constant flow of firearm patients on trauma center resources is also substantial. Of the patients admitted to the hospital, 2,044 (71%) required at least one operative procedure. With the exception of those patients undergoing damage-control laparotomy, we did not collect data on repeat procedures for a single patient or subsequent procedures performed following the index hospitalization. Thus, this group of patients likely required even greater operating room resources. In addition, with the clustering of patients, we have noted an increased number of patients going to the operating room simultaneously and an increased need for additional attending surgeons in the operating room. All of which put stress on trauma center resources. The notable shift in timing of arrival to earlier in the evening may have a deleterious effect on trauma center operating room use and it begins to cut into and interrupt the elective schedule. ICU resources continued to increase over time, which may have a negative effect on the nontrauma services by limiting elective surgical cases requiring ICU care or turning down transfers from outside institutions.
The aggregate cost of caring for our patients is staggering and is consistent with other published series.16–18 Ordog et al.19 in a study from Los Angeles the 1980s estimated hospital costs for GSW patients to be approximately $250 million with more than 90% of the cost borne by the public. Cook et al.18 estimated the overall cost of GSW injuries in the United States (in 1994 dollars) to be $2.3 billion, with a mean acute care per case of $17,000 per patient, approximately half having some insurance. There are differences in both treatment costs and payer mix between self-inflicted or unintentional injuries and those caused by assaults, which likely explain the more favorable payer mix in that study . Brown et al.20 in a study of orthopedic GSW injuries in New Orleans found that 96% were uninsured. Our data demonstrated a clear increase in cost during the period of the study, with 75% being unreimbursed. We believe that this 75% figure is conservative and is based on estimated costs and not charges. The primary defect in the cost data is that the charge-to-cost modifier is for the entire hospital population and represents the “average hospitalized patient,” not just those injured by guns. It is possible and quite likely that these patients are not average and use greater and more costly resources. The strength of these data is that we calculated costs instead of relying on hospital charges for patients, and while not perfect, they are a much better approximation of the financial impact to the institution. These data do not capture physician charges or subsequent outpatient visits or hospitalizations. Thus, the actual cost of the direct medical care for this population exceeds the numbers presented here. The combination of a greater-than-expected increase in cost during the study period and the extremely low reimbursement rates inflicts a large resource strain on the hospital and the community. That burden is not decreasing with time. In addition, the neighborhoods most affected by the costs of firearm violence are those least equipped to handle it.
The geographic mapping demonstrates that both the patient’s domicile and incident locations are colocated and did not change significantly during the course of the study. Many of the neighborhood characteristics associated with gun violence have been well documented for decades and were observed in those census tracts, which had consistently high numbers of patients. These include lower socioeconomic status, increased percentage of residents living below the poverty line, increased percentage of female-headed households, increased housing density, and increased numbers of vacant lots. One novel finding was that of an increase in brownfield areas within those high-performing census tracts. Branas et al.21 and Garvin et al.22 reported that vacant lots were associated with an increase in gun violence, which could be positively impacted on by urban development. Newark has undergone a significant amount of development during the past decade with the removal of several high-density housing projects and the building of single family homes on many previous vacant lots. Despite this, our data would suggest that it had little effect on GSW violence (location and incidence), and a more in-depth study is needed to determine what effect this development had in our community. One possibility for the marked disparity between those census tracts with increased gun violence compared with others is that the population who has the ability to move from undesirable neighborhoods do so, resulting in an increased percentage of people who are involved in activities associated with gun violence or have no other options.23,24
As with many other series on gun violence, there are several limitations to our study. The first is that this is a single-center retrospective study in a densely populated urban city with a history of gun violence. While the data are not population based as strictly epidemiologically defined, we are the only trauma center in our catchment area and are confident that our data (exclusive of patients pronounced dead at the scene) represent well more than 95% of the serious GSW injuries. In addition, before and during the 12-year period of the study, several other hospitals in the Newark metropolitan area closed, and we postulate that our data represent the overwhelming majority of “nonserious” GSW injuries as well. A specific strength of our study is the 12-year longitudinal analysis and its ability to connect patient-centric injury data for all GSW patients with actual home and scene address to understand the neighborhood contribution to our patient population. This allows us to identify and examine changes over time, which may not be apparent with more abbreviated studies. Another significant limitation on the analysis and understanding of these injuries is the data silos, which exist between the health care and LE communities. This is especially true for the vast majority of patients who sustain nonfatal GSW injuries. We believe that contemporaneous and real-time direct data linkages, which contain the health care information, incident circumstances, victim-offender relationships, as well as gang and drug affiliations, are both possible25 and remain largely unexplored. These data could be exceedingly useful to form and shape public policy and prevention strategies. To explore this concept further, during December 2010, we piloted a “real-time” link between the New Jersey Trauma Center and LE. LE data were obtained for 21 shooting incidents, which occurred during a 7-day period in Newark. These incidents involved 24 alleged individuals. Two people could not be not linked, indicating that they were either treated at another facility or more likely were uninjured in the event. One patient was pronounced dead at the scene. Twenty-one patients were linked to the trauma registry and were treated at our trauma center. Eleven were discharged from the ED. Of the remaining 10 patients, all required operative management, 1 died in the operating room, 1 died in the surgical ICU, and 7 were discharged. Two of the three identified perpetrators were treated previously for GSW injuries. Unfortunately, because of changes in LE personnel and administrative priorities, this project could not be continued, and we believe it remains a vital and unappreciated concept to combat gun violence.
In summary, the data presented here demonstrate the unrelenting violence secondary to firearms treated at an urban Level I trauma center for more than a decade. These data unfortunately are emblematic and representative of the overall lack of progress on gun violence across the United States. These data also represent the “routine” gun violence, which is observed across America. Acknowledging that this day-to-day violence is just has “horrific” as those episodic and more “newsworthy” mass casualty incidents is needed to restore the rational dialog on treating gun violence as a public health crisis as suggested more than two decades ago. The presidential executive order reinstituting the CDC’s ability to fund gun research and the resulting Institute of Medicine report9 outlining research priorities are positive steps to bypass and circumvent congressional constipation and meddling, which has resulted in a near two-decade hiatus on firearm research. We strongly urge that interdiction of other federal agencies to fund this type of research needs to be lifted as well. Lastly, no discussion of firearm violence would be complete without some mention on the unabated proliferation of firearms in the United States in the past 20 years.26 While the politics and effects of firearm legislation and regulation are complex and beyond the scope of the data presented here, it is obvious to all health care providers that without guns, there would be no GSWs and the health care consequences of these injuries would be eliminated. The decline in both interpersonal as well as self-inflicted GSWs in Australia two decades following a strong governmental and societal response to a series of mass casualty events culminating in 1996 should give one food for thought as to what is possible.27 In his presidential address to the Eastern Association for the Surgery of Trauma in 1993, Dr. C. William Schwab hypothesized that it might take 20 years to turn the tide of public opinion on gun violence, a time frame similar to the success regarding legislation controlling tobacco and auto safety.28 One can only hope that we have merely taken a 20-year detour, and with continued data and research, we might see declining firearm violence in only one more decade.
D.H.L. and R.F.L. conceived of the study and contributed to the data analysis, manuscript execution and review, and literature search. M.C.L. contributed and performed all GIS studies and the analyses of all GIS data, literature review, and critical manuscript editing. D.F.L. contributed all economic analysis and critical manuscript editing. M.R.P. participated in the study design, data and statistical analysis, and critical manuscript editing.
The authors declare no conflicts of interest.
2. Kellerman AL, Rivara FP . Silencing the Science on Gun Research. JAMA. 2013; 309: 549–550.
3. Federal Reserve Economic Data. Available at: http://research.stlouisfed.org/fred2
, Economic Research Division, Federal Reserve Bank of St. Louis. Consumer Price Index for All Urban Consumers: Medical Care. Accessed July 2013.
8. Ladapo JA, Rodwin BA, Ryan AM, Trasande L, Blustein J . Scientific publications on firearms in youth before and after congressional action prohibiting federal research funding. JAMA. 2013; 310: 533–535.
9. National Research Council. Priorities for Research to Reduce the Threat of Firearm-Related Violence. Washington, DC: The National Academies Press; 2013; .
10. Gotsch KE, Annest JL, Mercy JA, Ryan GW . Surveillance for fatal and nonfatal firearm-related injuries—Unites States 1993–1998. MMWR Morb Mortal Wkly Rep. 2001; 50: 1–32.
11. Annest JL, Mercy JA, Gibson DR, Ryan GW . National estimates of nonfatal firearm-related injuries. Beyond the tip of the iceberg. JAMA. 1995; 273: 1749–1754.
12. Christoffel KK . Firearm injuries: epidemic then to endemic now. Am J Pub Health. 2007; 97: 696–699.
14. Centers for Disease Control and Prevention. Gang homicides—five U.S. cities, 2003–2008. MMWR Morb Mortal Wkly Rep. 2012; 61: 46–51.
15. Richmond TS, Branas CC, Cheney RA, Schwab CW . The case for enhanced data collection of gun type. J Trauma. 2004; 57: 1356–1360.
16. Mock C, Pilcher S, Maier R . Comparison of the costs of acute treatment for gunshot and stab wounds: further evidence of the need for firearm control. J Trauma. 1994; 36: 516–522.
17. Vassar MJ, Kizer KW . Hospitalizations for fire-arm related injuries: a population-based study of 9562 patients. JAMA. 1996; 275: 1734–1739.
18. Cook PJ, Lawrence BA, Ludwig J, Miller TR . The medical costs of gunshot injuries in the United States. JAMA. 1999; 282: 447–454.
19. Ordog GJ, Wasserberger J, Ackroyd G . Hospital costs of firearm injuries. J Trauma. 1995; 38: 291–298.
20. Brown TD, Michas P, Williams RE, Dawson G, Whitecloud TS, Barrack RL . The impact of gunshot wounds on an orthopaedic surgical service in an urban trauma center. J Orthop Trauma. 1997; 11: 149–153.
21. Branas CC, Cheney RA, MacDonald JM, Tam VW, Jackson TD, Ten Have TR . A difference-in-differences analysis of health, safety, and greening vacant urban space. Am J Epidemiol. 2011; 174: 1296–1306.
22. Garvin EC, Cannuscio CC, Branas CC . Greening vacant lots to reduce violent crime: a randomised controlled trial. Inj Prev. 2013; 3: 198–120.
23. Tita GE, Petras TL, Greenbaum RT . Crime and residential choice: a neighborhood level analysis of the impact of crime on housing prices. J Quant Criminol. 2006; 22: 299–317.
24. Kyriacou DN, Hutson HR, Anglin D, Peek-Asa C, Kraus JF . The relationship between socioeconomic factors and gang violence in the City of Los Angeles. J Trauma. 1999; 46: 334–339.
25. Branas CC, Culhane D, Richmond TS, Wiebe DJ . Novel linkage of individual and geographic data to study firearm violence. Homicide Stud. 2008; 3: 298–320.
27. Ozanne-Smith J, Ashby K, Newstead S, Stathakis VZ, Clapperton A . Firearm related deaths: the impact of regulatory reform. Inj Prev. 2004; 10: 280–286.
28. Schwab CW . Violence: America’s uncivil war—presidential address, Sixth Scientific Assembly of the Eastern Association for the Surgery of Trauma. J Trauma. 1993; 35: 657–665.
Dr. Demetrios Demetriades (Los Angeles, California): David, that was an excellent paper. This study reported some important findings with practical implications. I have some comments and questions.
Over the study period there was an increase in the proportion of patients with multiple gunshot injuries in multiple body areas. You suggested that the most likely explanation was the use of handguns with increased firing capacity. Has this suspicion been confirmed by the police? If yes, what is the message we need to convey to the legislators?
The overall mortality increased over the study period from 9% to 14%. You attributed this finding to the increased number of victims with gunshot wounds to the head and the increase in the injury severity. Have you attempted to stratify survival outcomes according to ISS groups or specific severe injuries, for example, isolated severe head trauma, chest trauma, abdominal trauma with an AIS higher than three?
Lastly, probably the most important message of the study is that we need a reliable national database on firearm injuries. Most of the existing studies use trauma registries as their source of data. However, trauma registries, as you pointed out, do not include victims treated and discharged from the emergency department, or victims who die at the scene and never reach hospital care, or victims treated at non-trauma centers.
This major limitation by the registries cause a lot of confusion. For example, there was a recent front-page article in the Wall Street Journal which reported the results of a research performed by the Howard-Hopkins Surgical Outcomes Research Center. The study concluded that over the last six years the number of gunshot wounds has been increasing but the mortality rates have fallen because of, presumably, better care.
In a recent study we did at USC using the NTDB, we found that over the same time period the incidence from gunshot injuries has been going down while the mortality remained unchanged. It is obvious that we need a reliable national database. What role do you think that our major professional organizations, such as the AAST or the ACS, should play in such a project?
As you pointed out, the president issued an executive order which lifted the previous ban on federal funding on this issue. What shall we do as professional organizations?
Dr. C. William Schwab (Philadelphia, Pennsylvania): This is a serious message and perhaps one of the most important of this meeting. This is a well-done study and it builds on several key public health design and methods papers. The conclusions are well supported.
It is a snapshot of only one aspect of firearm injury in America: inner-city, inner-personal, intentional, handgun, non-lethal wounding among younger males. However, though only a hospital based only focus, it is compelling.
More importantly for those in the audience—and for the future readership—when this is published, it will serve as a template for other trauma centers, as a guide in the use of commonly available hospital-based data sets to analyze and portray the burden of gun wounding on the hospital’s bottom line, on the community, and on our nation.
Explicit in your message from Newark’s only Level I trauma center is that it is threatened by a financial burden that this patient line forces on the hospital. But implicit is the threat that it brings to all urban-based centers.
If we use only the 25 largest United States metropolitan centers and apply the $115 million cost figure, it sums to $2.8 billion of cost; $2.1 of which is uncompensated to these 25 key safety net hospitals.
At a time when recent events have continuously proven the 24-7 readiness and life-saving missions of the trauma centers and assured that most communities have a front-line for mass casualty and disaster response, it would seem that we, the AAST, and our sister and brother societies, have a political advantage to sound yet another alarm about this uncovered burden.
I have two questions.
David, were you able to find if within the handgun data the mere substitution or growth of semi-automatic pistols as opposed to revolvers led to the increased wounding and death patterns? The substitution of pistols, not necessarily higher magazine capacity, drove the change in wounding patterns in the early 1990s.
And I hate to bring this up, but has there been an improvement in shooters’ abilities with a gun that might partially be explained by a number of these people being ex-military? Though that might seem absurd, that was actually suggested as possibility as military-trained people returned to the cities and perhaps to protect their homes, businesses, etc., legally armed themselves. Thank you.
Dr. Andrew W. Mikulaschek (Fort Meyers, Florida): Yes, I would like to thank Dr. Livingston. I think that’s an important paper.
I come from a non-urban center, but we still have a significant problem. In the last 15 years I’ve spent in Fort Meyers we have a small but steady population that represents about 15% annual penetrating trauma.
As this is probably more a societal issue than anything else, did you look at the incidence of gang-related mechanism as well as possibly drug-transaction related incidences as a major focus for the use of guns?
Dr. Carnell Cooper (Baltimore, Maryland): Nice work. I am sure a number of these victims were repeat victims of violence.
The group at Lincoln published a paper in 2012 noting that their folks who had more than one injury, one admission for a victim of violence, they were five times to ten times higher to die if there was a second or a third admission.
I was wondering if you had any data in your study on how many of them were repeat victims and was their mortality increased in that group that was a repeat victim. Thanks.
Dr. David H. Livingston (Newark, New Jersey): Thank you to Dr. Demetriades and all of my discussants. A special thank you to Dr. Schwab. Your work has been inspiring over the years and we used a lot of what you have done in Philadelphia in the discussion of our data regarding this very significant public health issue.
I will try to answer as many of the questions I can in the time allowed. Drs. Demetriades and Schwab asked, “Why the increased number of wounds?” We really dont know. We also do not know if there really is a change in the ballistics or whether the switch that Dr. Schwab described in the early `90s from revolvers to handguns is operating here. Yes, available and common magazine sizes have gone up, but our data shows a lot more bullets at one time. Maybe there are also more shooters. Again we just dont know and we strongly believe this is why we need a law enforcement-Department of Justice connection with some of our data. That would greatly improve our understanding of some of these questions.
Bill, I dont think they are ex-military and I am not sure a lot of our guys could have joined the military.
Dr. Demetriades brings up some contrary-appearing data from the NTDB on firearm mortality. As he suggests and we believe, these data are the inclusions of a lot more—possibly lower volume–trauma centers seeing and capturing many more “minor” or less-severe GSW injuries. This would be similar to the 20% of patients in our high-volume center that the trauma service never even saw. Thus increasing the denominator without changing the numerator. We do strongly agree that a national database linked to law enforcement would be ideal.
With respect to question on gangs and drug-related violence asked by Dr. Cooper: Again, there is a lot of conjecture and not a lot of data available to us which would be improved upon by a linkage of data between health and law enforcement. A recent study in MMWR suggested less than 15% of all firearm violence in Newark were drug- and gang-related, and we all laughed at that. But I think no one really knows.
Our readmit rate for gunshot wounds was 6%. The recidivism rate for any previous encounter secondary to violence was about 20%. From our data, it didnt appear that it changed mortality, but I imagine eventually if you get shot enough times it will.
Our readmit rate for gunshot wounds was 6%. The recidivism rate for any previous encounter secondary to violence was about 20%. From our data, it didn’t appear that it changed mortality, but I imagine eventually if you get shot enough times it will.
We would love to replicate this study in other municipalities. I think it would be a great snapshot in America about firearm violence and important data to show to politicians and policy makers.
I’ll be more than happy to discuss this further with everyone outside. Thank you very much.
Gunshot wound; firearm violence; health care costs; trauma mortality; GIS
Supplemental Digital Content
Copyright © 2014 by Lippincott Williams & Wilkins