Trends in firearm-related injuries by residential zip code were mapped using ArcGIS software, version 10.4 (Esri). Spatial clustering was tested using the Hot Spot Analysis (Getis-Ord Gi*) tool and the Cluster and Outlier Analysis (Anselin Local Moran’s I) tool. Significant spatial clustering of patients who were initially treated in nondesignated facilities was identified both on the west side of Chicago and in the southern parts of Chicago and Cook County.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Lale A, Krajewski A, Friedman LS. Undertriage of Firearm-Related Injuries in a Major Metropolitan Area. JAMA Surg. 2017;152(5):467–474. doi:10.1001/jamasurg.2016.5049
What are the prevalence and outcomes of firearm-related injuries undertriaged to facilities without specialized trauma units?
In this analysis, 1 in 6 persons with firearm-related injuries who met national anatomic triage criteria were undertriaged to facilities without specialized trauma units, but an increase in mortality was not observed in this group. Spatial disparities were observed regarding the prevalence of undertriage of firearm-related injuries.
Better regional coordination among hospitals and frequent trauma care system assessments are needed to improve prehospital triage of firearm-related injuries.
National anatomic triage criteria prescribe specific transport rules for injured patients. However, there is limited information about patients with firearm-related injuries undertriaged to nondesignated facilities (ie, hospitals without specialized trauma teams or units), including what clinical outcomes are achieved and how many are transferred to a higher level of care. Without these data, it is difficult to make informed regional or national policy decisions about triage practices. Undertriage of firearm-related injuries is a good model for evaluating the undertriage of patients with trauma because the anatomic triage criteria for patients with firearm-related injuries are simple.
To evaluate the prevalence, spatial distribution, and clinical outcomes of undertriage of firearm-related injuries.
Design, Setting, and Participants
This study is a retrospective analysis of firearm-related injuries in residents of Cook County, Illinois, from January 1, 2009, to December 31, 2013. Outpatient and inpatient hospital databases were used. Participants included patients with International Classification of Diseases, Ninth Revision, Clinical Modification firearm-related cause-of-injury codes. Data were collected all at once in August 2014. Data analysis took place from March 12, 2015, to February 1, 2016.
Main Outcomes and Measures
Undertriaged cases were defined as patients who met the national anatomic triage criteria for transfer to higher-level trauma center care. Spatial distribution, injury severity, and clinical outcomes, including death, were analyzed.
Of the 9886 patients included in this analysis, 8955 (90.6%) were male, 7474 (75.6%) were African American, and 5376 (54.4%) were aged 15 to 24 years.In Cook County, Illinois, where there are 19 trauma centers, 2842 of 9886 (28.7%) firearm-related injuries were initially treated in nondesignated facilities. Among the 4934 cases with firearm-related injury who met the anatomic triage criteria, 884 (17.9%) received initial treatment at a nondesignated facility and only 92 (10.4%) were transferred to a designated trauma center. Significant spatial clustering was identified on the west side of Chicago and in the southern parts of Chicago and Cook County. In the multivariable models, patients treated in nondesignated facilities were less likely to die than were patients treated in designated trauma centers.
Conclusions and Relevance
Undertriage of firearm-related injuries was much more prevalent than expected. Although the likelihood of dying during hospitalization was greater among patients treated in designated trauma centers, these patients were substantially in worse condition across all measures of injury severity. A smaller proportion of patients treated in designated trauma centers died during the first 24 hours of hospitalization. This study highlights the need for better regional coordination, especially with interhospital transfers, as well as the importance of assessing the distribution of emergency medical services resources to make the trauma care system more effective and equitable.
Each year in the United States, more than 110 000 persons are injured by firearms and more than 30 000 of the wounded die from their injuries. Quiz Ref IDThe “Guidelines for Field Triage of Injured Patients”1 recommends that gunshot wounds proximal to the elbows and knees be treated in specialized trauma units even if it means bypassing a nearby hospital. Studies show that injured patients who are treated in designated trauma centers (ie, facilities with specialized trauma units) have lower in-hospital, 30-day, and 90-day mortality rates,2-4 a finding that supports this triage recommendation. However, undertriage to hospitals without specialized trauma teams or units occurs for a variety of reasons, including prehospital triage errors,5 a lack of trauma units near the site of the injury,6,7 and diversion by trauma units of emergency medical services (EMS) personnel to nondesignated facilities. The undertriage of firearm-related injuries is a good model for evaluating the undertriage of patients with trauma because the anatomic triage criteria for patients with firearm-related injuries are simple.
In the United States, nearly all research on firearm-related injuries has used data from trauma centers. Some of these studies have evaluated the association between adverse outcomes and prehospital modes of transport and found that police and personal vehicles have transported 10% or more patients arriving at a trauma center.8-10 However, these data sets do not capture patients who are treated in nondesignated facilities (ie, hospitals without specialized trauma teams or units). Available data on the undertriage and treatment of injured patients in nondesignated facilities offered little information about who is undertriaged, how many are transferred to designated trauma centers, and what clinical outcomes are achieved. Without such data, it is difficult to make informed regional or national policy decisions about triage practices.
The present study describes firearm-related injuries that were treated in both designated trauma centers and nondesignated facilities. Inpatient and outpatient hospital databases were used to identify patients who qualified, according to anatomic triage criteria, for higher-level trauma care but instead received treatment in nondesignated facilities. The study aimed to evaluate the prevalence, spatial distribution, and clinical outcomes of the undertriage of firearm-related injuries.
This retrospective analysis focused on residents of Cook County, Illinois, who sustained a firearm-related injury and initially received care at a hospital in Cook County or its collar counties from January 1, 2009, to December 31, 2013. Our data sets captured both inpatient and outpatient cases, and both databases were derived from billing records and represented a census of patients who were treated in nondesignated facilities. The inpatient database included patients who were treated for 24 hours or more, while the outpatient database included patients who were treated in emergency departments for less than 24 hours but were not admitted into the hospitals. Both data sets contained information on patient demographics, exposure, health outcomes, and economic outcomes. Review of the annual state audit of hospitals revealed that the hospitals in the data sets provided 96.5% of all patient admissions statewide.11
Data were collected in August 2014. Data analysis took place from March 12, 2015, to February 1, 2016. The institutional review board of the University of Illinois at Chicago provided approval, which included a waiver of patient consent, for this study.
For this analysis, residents of Cook County were included on the basis of their residential zip codes, which were available for 99.9% of the 9886 patients in the data sets. These patients were initially treated in a hospital located in the Illinois EMS regions covering Cook County and its collar counties. We established these inclusion criteria to adjust for the flow of patients to hospitals across county lines, which frequently occurs because many catchment areas are not restricted to county jurisdictions. Firearm-related injuries were identified using the following International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) cause-of-injury codes (E codes): E922.0-3 and E922.9 (unintentional injury), E955.0-955.4 (suicide and self-inflicted injury), E965.0-965.4 (intentional injury), E970 (injury due to legal intervention), and E985.0-4 (unknown intent).
We used probabilistic linkage methods to identify transfers between hospitals. No information was deleted. All data were linked to the earliest index record and analyzed in aggregate. Transfers involved the same patient treated in multiple acute care hospitals on sequential dates. For the transfer cases, we used a multistage approach (3 stages) to allow for fuzzy matching criteria. The idea behind multistage matching is to start the process using only “high probability” variables, which will result in low false-match rates. Each subsequent step adjusts the linkage criteria for any remaining cases, allowing for greater variability in the linkage criteria (including fuzzy terms) and thus less certainty of a true match among the remaining unmatched cases. We used the following variables for matching: date of birth, sex, race, ethnicity, residential zip code, date of admission, date of discharge, and cause of injury. The initial primary injury ICD-9-CM diagnosis was also required to be listed as a diagnosis in the transfer hospital records.
We analyzed the ICD-9-CM diagnosis codes (N codes; up to 28 fields with diagnosis codes per patient) for each patient to assess body region and type of injury on the basis of the Barell classification matrix.12Quiz Ref ID Undertriaged cases were defined as patients who met the anatomic triage criteria for transfer to a trauma unit1 but were treated initially in a nondesignated facility. These cases included N codes for the following: penetrating injuries to the head, neck, torso, and extremities proximal to elbow and knee; flail chest; 2 or more proximal long-bone fractures; crushed, degloved, or mangled extremity; amputation proximal to wrist and ankle; pelvic fractures; open or depressed skull fracture; or paralysis.1 Physiological triage criteria (low systolic blood pressure, tachypnea, or low Glasgow Coma Scale score) were not analyzed because prehospital physiological measures were not available in these data sets. However, these measures frequently change between the prehospital setting (where the triage decision is made) and the point of hospital admission (when data are collected and reported in the billing databases).
The study included various measures of injury severity, including the New Injury Severity Score (NISS), which indicates serious injuries (eg, NISS≥16) and is based on the Abbreviated Injury Scale.13 The NISS is calculated at the time of discharge and based on all N codes for injuries identified during hospitalization. For the statistical modeling, we used the Trauma Mortality Prediction Model as a measure of injury severity.14 To characterize comorbidities, we used the Elixhauser Comorbidity Index.15
All statistical analyses were conducted using SAS software, version 9.4 (SAS Institute Inc). As part of the descriptive analysis, we compared the demographic characteristics, geospatial trends, temporal trends, injury severity, and hospital course of treatment measures for firearm-related injuries that were treated at nondesignated facilities with those treated at designated trauma centers. In addition, we mapped trends in firearm-related injuries by residential zip code using ArcGIS software, version 10.4 (Esri) and tested spatial clustering using the Hot Spot Analysis (Getis-Ord Gi*) tool and the Cluster and Outlier Analysis (Anselin Local Moran’s I) tool. Appropriate nonparametric (Wilcoxon rank sum test) and parametric (t test) tests were used to compare continuous variables, and the χ2 test was used to compare categorical variables.
Four multivariable logistic regression models were developed to evaluate the relationship between in-hospital mortality and treatment in the nondesignated facilities and designated trauma centers: (1) a full model with all the patients, (2) a subset including only patients who met the anatomic triage criteria, (3) a subset including only inpatient cases, and (4) a subset including only outpatient cases. Statistical evaluation of covariates, as well as a priori knowledge, was used to determine the inclusion of covariates in the final models. No evidence of multicolinearity among the independent variables was indicated. P < .05 (2-tailed) was considered statistically significant.
Quiz Ref IDOf the 9886 firearm-related injuries that were initially treated in hospitals in Cook County and its collar counties from January 1, 2009, to December 31, 2013, there were 2842 patients (28.7%) who received care at nondesignated facilities and 7044 patients (71.3%) who received care in designated trauma centers. Of the 9886 patients included in the analysis, 8955 (90.6%) were male, 7474 (75.6%) were African American, and 5376 (54.4%) were aged 15 to 24 years (Table 1). The most common cause of firearm-related injury was intentional assault (6267 [63.4%]) followed by accidental injury (2458 [24.9%]), but the latter cause was more common among patients treated in nondesignated facilities than in designated trauma centers (31.2%  vs 22.3% ; P < .001). Suicides were rare among patients treated in both the nondesignated facilities (22 [0.8%]) and designated trauma centers (76 [1.1%]).
Quiz Ref IDMapping the number and proportion of all firearm-related injuries that were initially treated at nondesignated facilities (by residential zip code) shows that residents living on the west side of Chicago and in the southern parts of Chicago and Cook County were disproportionately affected (Figure). Significant spatial clustering of patients who were treated in nondesignated facilities, adjusted for total population, was identified in both the west side of Chicago and the southern parts of Chicago and Cook County. This pattern persisted when we limited our analysis to only firearm-related injuries that met anatomic triage criteria.
Compared with patients who were treated in designated trauma centers, patients who were treated at nondesignated facilities had a substantially lower mean NISS—the proportion who were placed on a ventilator, the proportion who required surgical intervention, and the percentage who died during hospitalization (Table 2). Quiz Ref IDAlthough patients who were treated in nondesignated facilities showed clinical signs of less severe injuries, 884 (31.1%) of these patients met the anatomic triage criteria for a higher level of care (Table 3).
Among the 4934 Cook County residents with firearm-related injuries who met the anatomic triage criteria, approximately 1 (884 [17.9%]) in 6 received treatment at a nondesignated facility. Within this subgroup of patients who met anatomic triage criteria, those treated at designated trauma centers (4050 [82.1%]) consistently had indicators of more serious injuries compared with those treated at nondesignated facilities (884 [17.9%]). Consider these numbers for patients at designated trauma centers vs nondesignated facilities, respectively: mean length of stay, 6.2 vs 3.5 days; required mechanical ventilation, 18.6% vs 2.6%; injuries to multiple body regions, 53.1% vs 30.7%; and serious injuries (NISS ≥16), 35.6% vs 6.1%. Of the 884 patients who met triage criteria but were initially treated in nondesignated facilities, only 92 (10.4%) were transferred to a higher level of care (Table 3). Of the 59 patients with serious injuries (NISS ≥16) who were treated at nondesignated facilities, only 1 patient was transferred to a trauma unit.
Among those who died, patients who were treated in designated trauma centers were more likely to have injuries to the torso and multiple body parts than were those treated in nondesignated facilities. In addition, fatal injuries treated in designated trauma centers were more likely to meet anatomic triage criteria and had much higher NISS than fatalities treated in nondesignated facilities. However, NISS is not always a good predictor of injury severity—particularly, fatal injury. Overall, a few of the fatal firearm-related injuries had a NISS of 16 or greater, and a high NISS was rare among those who died at nondesignated facilities. Only 6 (6.6%) of the 91 fatal firearm-related injuries treated in nondesignated facilities and only 258 (36.8%) of the 701 fatal firearm-related injuries treated in designated trauma centers had a NISS of 16 or greater.
Of the 884 patients at nondesignated facilities who met anatomic triage criteria, 73 (8.3%) subsequently died. Seven (9.6%) of those deaths occurred after being transferred to a designated trauma center, 66 (90.4%) occurred at the initial nondesignated hospital, and 15 (20.6%) died more than 1 day after admission. In contrast, the 4050 patients at designated trauma centers who met anatomic triage criteria had a higher mortality rate (642 deaths [15.9%]). There were additional deaths among patients who did not meet anatomic triage criteria: 18 (19.8%) at nondesignated facilities and 59 (8.4%) at designated trauma centers.
We developed multivariable models to test our prediction that meeting anatomic triage criteria would carry a higher risk of mortality at a nondesignated facility than at a designated trauma center. Table 4 presents the unadjusted and adjusted odds ratios for all patients (0.30 [0.24-0.37] and 0.54 [0.43-0.69], respectively) as well as the stratified models exclusive to inpatient and outpatient cases who met anatomic triage criteria. Patients treated in nondesignated facilities were less likely to die than patients treated in designated facilities. The direction of the association remained constant in all the models.
Most firearm-related injury research is conducted using trauma center data, which inadvertently exclude a large proportion of firearm-related injuries treated in nondesignated facilities. In Cook County, which has 19 trauma centers, almost 3 in 10 gunshot wounds—including 1 in 6 that met anatomic triage criteria—were treated outside of designated trauma centers. However, the occurrence of undertriage was not uniformly distributed across the geographic region but was substantially more pronounced in certain neighborhoods on the west side of Chicago and in southern parts of Chicago and Cook County. Regional trauma care systems and anatomic triage criteria exist to dictate a standard of care.1 The treatment of patients who meet the criteria and/or dying of a firearm-related injury in a community hospital, especially if the patient can be stabilized for transport, should be a rare event in a county with 19 trauma centers. Our findings raise 2 major questions.
First, why are patients with firearm-related injury being treated in nondesignated facilities? Patients who arrived at a lower level of care and met the anatomic triage criteria were either undertriaged by EMS personnel or dropped off by a personal vehicle. Policy measures can more effectively address the former issue. However, we suspect, based on our communications with physicians from community hospitals in Cook County and the peer-reviewed literature,10,16 that civilian transport is the major underlying cause for these findings. However, given the thousands of shootings in Chicago that happen 5 or more miles away from a trauma center,17 EMS personnel may also contribute by undertriaging injured patients when trauma centers are too far away.6 This scenario is especially relevant in some of the most economically disadvantaged municipalities in southern Cook County that have limited EMS resources. In these communities, long travel times to a trauma center can take an ambulance out of service for hours, affecting potential response times for other emergency cases (eg, stroke) that could be treated in local hospitals.
Second, how do nondesignated facilities properly manage injured patients who meet the triage criteria for a higher level of care once they arrive? Community hospitals should be better integrated into regional trauma care systems, especially in terms of communication and transfer protocols. Without established or improved regional coordination, a patient who arrives at a nondesignated facility can be blocked from transfer by a higher-level designated trauma center or may go unrecognized by the community hospital staff as someone who needs a higher level of care. In this study, almost 90% of patients who met the anatomic triage criteria were not transferred to a designated trauma center. Furthermore, among the 91 patients who died during hospitalization in nondesignated facilities, 73 (80.4%) met the anatomic triage criteria for a higher level of care. Given the volume of gun violence in Cook County, our study shows there is room for improvement of the regional trauma care system.
Our finding that the mortality rates were higher among patients treated in trauma centers despite controlling for known confounders (ie, injury severity, comorbidities, and patient age) was unexpected. However, patients who were treated in designated trauma centers were substantially in worse condition across all measures of injury severity than patients who were treated in nondesignated facilities. This finding may reflect the limitations of the current severity scoring systems in characterizing risk of death, or it may reflect that there are unknown variables that remain unaccounted for in the multivariable models. For example, mode of transport, transport times, and physiological measures are not accounted for in our model. Civilians may be getting patients to local community hospitals faster than patients transported by EMS, and these saved minutes may be contributing to lower mortality.10 However, a key finding of this study was that fewer patients treated in designated trauma centers died during the first 24 hours compared with patients treated in nondesignated facilities (49.8% vs 80.2%). This finding may reflect the higher quality of care in trauma centers, lowering the mortality rate associated with the immediate causes of death (eg, hypovolemia, hemorrhage). However, because designated trauma centers receive more serious injuries, their patient pool is more likely than the patient pool of nondesignated facilities to experience complications from subsequent physiological responses to the injury.14
In evaluating the undertriage in Cook County, several limitations should be considered. First, we lacked patients’ physiological triage measures (ie, blood pressure, respiratory rate, and Glasgow Coma Scale score) from the time of arrival to the hospital and from the prehospital setting, which likely results in an underestimation of the true count of undertriaged patients. Second, this lack of physiological measures affected our ability to identify patients who arrived dead or who died when resuscitation failed during the initial window of care. Most deaths in nondesignated facilities occurred during the first 24 hours of care. Third, a larger proportion of firearm-related injuries treated at nondesignated facilities were of lower injury severity. However, patients who died at the nondesignated facilities also had much lower NISS than patients who died at the designated trauma centers. While this fact may reflect gross mismanagement of minor firearm-related injuries in nondesignated facilities, it may also reflect the inadequacy of anatomic severity scales that fail to capture other important factors, such as physiological measures, comorbidities, and medical complications. It may also be the result of inaccurate recordkeeping, especially among the dead-on-arrival and failed-resuscitation cases. Fourth, although the inpatient and outpatient databases provided a greater patient capture than trauma registries, these data systems are not designed for trauma surveillance and lack some key data fields that make trauma registries valuable. Billing databases and trauma registries should be used to complement each other given that both data systems provide unique data elements. However, billing databases can be greatly improved by the capture of prehospital measures such as mode of transport, total call time, and the trauma center criteria used by EMS. Fifth, all of our data are from a single metropolitan area in Illinois surrounding a major American city, which may affect the generalizability of the analysis. However, the large proportion of firearm-related injuries treated in nondesignated hospitals despite the presence of 19 trauma centers suggests that this pattern may be relevant in other urban areas with disparate trauma coverage as well as in most rural areas and economically disadvantaged countries. Regional disparities in access to specialized trauma teams are felt in other states and abroad and have been reported in the peer-reviewed literature.18-20 Sixth, firearm-related injuries represent only a fraction of all traumatic injuries. Disparities in transport and transfer status may differ in patients with more common mechanisms of traumatic injury, such as motor vehicle crashes and falls, as has been demonstrated in other studies.21,22
Although federal law requires emergency departments to stabilize all patients that come through their doors, there is a legal exception when the benefits of transfer outweigh its risks. This study illustrates that the standard of care for the triage of patients with firearm-related injuries is working for many patients; however, there remains room for improvement, and there is a need for continual evaluation. We recommend that all trauma care systems conduct an analysis on a regular basis to inform triage policy and practice. At the administrative level, stakeholders should evaluate existing coordination plans, communications, leadership infrastructure, training of medical staff, EMS transport, and between-hospital transfer protocols. Evaluation should be done by independent stakeholders who set annual benchmarks and goals, while these stakeholders should be supported by government and private entities with the authority to implement recommendations. In addition, administrators and medical personnel from community hospitals need to be part of this process. At the practice level, there should be regular audits of EMS transports and interhospital transfers to both designated trauma centers and nondesignated facilities to assess adherence to triage protocols and quantify the adverse effects of undertriage.
Corresponding Author: Lee S. Friedman, PhD, Division of Environmental and Occupational Health Sciences, School of Public Health, University of Illinois at Chicago, 2121 W Taylor St, Room 504, Chicago, IL 60612 (email@example.com).
Accepted for Publication: November 10, 2016.
Published Online: January 18, 2017. doi:10.1001/jamasurg.2016.5049
Author Contributions: Dr Friedman had full access to all the Illinois hospital data analyzed in this study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: Lale, Friedman.
Statistical analysis: All authors.
Administrative, technical, or material support: Friedman.
Study supervision: Friedman.
Conflict of Interest Disclosures: None reported.
Create a personal account or sign in to: