A, Adjusted probability of admission to the non–trauma center according to a patient’s insurance type. B, Adjusted probability of admission to the non–trauma center according to the type of hospital where a patient was initially seen. These adjusted probabilities were calculated using the estimates of the multivariate logistic regression model summarized in Table 3. Error bars represent 95% CIs. Metro indicates Metropolitan Statistical Area, a high population density according to the US Census Bureau.
eAppendix. Supplemental Methods.
Delgado MK, Yokell MA, Staudenmayer KL, Spain DA, Hernandez-Boussard T, Wang NE. Factors Associated With the Disposition of Severely Injured Patients Initially Seen at Non–Trauma Center Emergency DepartmentsDisparities by Insurance Status. JAMA Surg. 2014;149(5):422-430. doi:10.1001/jamasurg.2013.4398
Copyright 2014 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Trauma is the leading cause of potential years of life lost before age 65 years in the United States. Timely care in a designated trauma center has been shown to reduce mortality by 25%. However, many severely injured patients are not transferred to trauma centers after initially being seen at non–trauma center emergency departments (EDs).
To determine patient-level and hospital-level factors associated with the decision to admit rather than transfer severely injured patients who are initially seen at non–trauma center EDs and to ascertain whether insured patients are more likely to be admitted than transferred compared with uninsured patients.
Design, Setting, and Participants
Retrospective analysis of the 2009 Nationwide Emergency Department Sample. We included all ED encounters for major trauma (Injury Severity Score, >15) seen at non–trauma centers in patients aged 18 to 64 years. We excluded ED discharges and ED deaths. We quantified the absolute risk difference between admission vs transfer by insurance status, while adjusting for age, sex, mechanism of injury, Injury Severity Score, weekend admission and month of visit, and urban vs rural status and median household income of the home zip code, as well as annual ED visit volume and teaching status and US region.
Main Outcomes and Measures
Inpatient admission vs transfer to another acute care facility.
In 2009, a total of 4513 observations from 636 non–trauma center EDs were available for analysis, representing a nationally weighted population of 19 312 non–trauma center ED encounters for major trauma. Overall, 54.5% in 2009 were admitted to the non–trauma center. Compared with patients without insurance, the adjusted absolute risk of admission vs transfer was 14.3% (95% CI, 9.2%-19.4%) higher for patients with Medicaid and 11.2% (95% CI, 6.9%-15.4%) higher for patients with private insurance. Other factors associated with admission vs transfer included severe abdominal injuries (risk difference, 15.9%; 95% CI, 9.4%-22.3%), urban teaching hospital vs non–teaching hospital (risk difference, 26.2%; 95% CI, 15.2%-37.2%), and annual ED visit volume (risk difference, 3.4%; 95% CI, 1.6%-5.3% higher for every additional 10 000 annual ED visits).
Conclusions and Relevance
Patients with severe injuries initially evaluated at non–trauma center EDs were less likely to be transferred if insured and were at risk of receiving suboptimal trauma care. Efforts in monitoring and optimizing trauma interhospital transfers and outcomes at the population level are warranted.
Trauma is the most common cause of years of life lost for those younger than 65 years, and it exacts $406 billion per year in costs, more than heart disease or cancer.1,2 Acute care in specialized trauma centers has been shown to reduce mortality by 25% in patients with major trauma.3 While direct transport from the scene of injury to a designated trauma center is optimal, patients who are taken to non–trauma center emergency departments (EDs) and who are then transferred in a timely fashion to trauma centers have reduced mortality compared with those who are hospitalized at the non–trauma center.4,5 Regionalized trauma systems have been developed during the past 3 decades to optimize population-level outcomes by facilitating the direct transport or transfer of patients with severe injuries to designated trauma centers.6,7
Despite the development of trauma systems, between 30% and 50% of patients with major injuries are still hospitalized in non–trauma centers8- 10 and may not receive optimal care. Even with correct application of emergency medical services field triage guidelines, at least 15% of severely injured patients will be undertriaged and transported to a non–trauma center.11 In addition, 16% of Americans lack geographic access to a trauma center within 60 minutes by emergency medical services transport.12 For these patients, transfer from a non–trauma center ED after stabilization represents another opportunity to ensure that severely injured patients get optimal care for their injuries.4,5
The factors associated with the disposition of severely injured patients initially seen at non–trauma center EDs in terms of hospitalization in the center vs transfer to a higher level of care are not well characterized. Previous studies13- 15 examining the transfer of trauma patients to multiple trauma centers only examine transfers from lower-level trauma centers to higher-level trauma centers. These studies provide a limited picture because non–trauma centers, which were not included in these studies, account for 75% of US hospitals.16 Until recently, a major barrier to examining trauma transfer patterns has been a dearth of national data sources that include trauma encounters in non–trauma center EDs because these EDs are not typically included in trauma registries.17,18
We analyzed the 2009 Nationwide Emergency Department Sample (NEDS) from the Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality, to determine patient-level and hospital-level factors associated with the decision to admit severely injured trauma patients rather than transfer them to a higher level of care. We hypothesized that insured patients would be more likely to be admitted to non–trauma centers than transferred to a higher level of care because of greater perceived reimbursement potential. Better understanding the patient and system factors associated with the decision to transfer will help to develop interventions to further optimize trauma systems and may provide insights for improving regionalized emergency care for other time-sensitive conditions.
This study was deemed exempt from review by the Stanford School of Medicine Institutional Review Board because it involved the analysis of publicly available, existing data without identifiable private information. We performed a retrospective analysis of the 2009 NEDS to determine which patient-level and hospital-level factors are associated with the decision to admit rather than transfer patients having major injuries in non–trauma center EDs in the United States. The NEDS is the largest all-payer ED database in the United States, capturing ED encounters that result in discharge or transfer and ED encounters that result in admission to the same hospital.17,19 The 2009 NEDS contains 29 million ED visits from 964 hospital-based EDs in 29 states weighted to provide national estimates of ED care. Along with standard patient and hospital administrative data, the data set contains a variable that classifies hospitals as non–trauma centers or as trauma centers (level I, II, or III).19
We extracted all non–trauma center ED encounters for major trauma (Injury Severity Score [ISS], >15) in patients aged 18 to 64 years identified from the cohort of ED encounters with diagnosis codes for trauma (International Classification of Disease, Ninth Revision, Clinical Modification codes 800-959). We excluded patients 65 years and older given that they would all be eligible for Medicare, confounding the influence of age and insurance status on the decision to transfer. We followed a validated algorithm for identifying patients with valid injury diagnosis codes used in trauma studies by excluding patients who had sole trauma diagnoses of superficial injuries, late effects of injury, foreign bodies, and burns.20,21 The 2009 NEDS provides an ISS calculated from each patient’s injury diagnosis codes. The ISS is a validated anatomical scoring system that provides a measure of the overall severity of injury for patients with multiple injuries.22 We excluded patients who did not have a NEDS ISS. We also excluded patients who had a primary payer listed as “no charge.” Finally, we excluded encounters that resulted in leaving against medical advice, discharge to home, transfer to skilled nursing care or home health care, or death in the ED.
The primary outcome of interest was inpatient admission to the non–trauma center vs transfer to another acute care facility according to primary payer status, including self-pay, Medicaid, Medicare, private insurance, and other commercial insurance (typically automobile or workers’ compensation insurance). Based on previous studies4,13- 15 and because of the regulation of the Emergency Medical Treatment and Labor Act (EMTALA) requiring that transfers out of the ED must go to a higher level of care, we assumed that all transfers of severely injured patients out of non–trauma centers would go to trauma centers.
All statistical analyses used the NEDS sampling strata and discharge weights to produce nationally weighted patient-level estimates and standard errors that account for clustering of patients among hospitals. We tabulated baseline characteristics by primary payer status. We then tabulated the primary outcome of inpatient admission vs transfer by baseline characteristics and compared unadjusted differences using χ2 test for categorical variables and linear regression for continuous variables. Finally, we used sample-weighted multivariate logistic regression to estimate the absolute risk difference of inpatient admission vs transfer to another acute care facility according to primary payer status by calculating average marginal effects. We adjusted for patient-level variables, including age, sex, ISS, mechanism of injury, weekend admission and month of visit, median household income of the patient’s home zip code, urban vs rural status and median household income of the home zip code, and body regions of injuries with an Abbreviated Injury Score of 3 or higher (an indicator of severe injury and a component of the ISS). We adjusted for hospital-level variables, including ownership, annual ED visit volume, urban vs rural status, and teaching status. We used available software (STATA 12.0; StataCorp LP) for all statistical analyses. The eAppendix in the Supplement provides additional information on our statistical methods and assumptions.
A total of 15 048 observations were available for analysis, representing 64 789 adult (≥18 years) major trauma (ISS, >15) patient encounters in 756 non–trauma center EDs in 2009. After excluding elderly patients, 6148 observations remained, representing 26 429 encounters for patients aged 18 to 64 years. After excluding patients who were discharged from the ED or who died in the ED, 4513 observations comprised the study population, representing 19 312 patient encounters for major trauma in 636 non–trauma center EDs in 2009 who were admitted as inpatients to the same non–trauma center hospital (54.5%; 95% CI, 50.8%-58.2%) or who were transferred to another acute care facility (45.5%; 95% CI, 41.8%-49.2%).
The mean patient age was 44 years, and the mean ISS was 19. Most patients were privately insured (43.3%), followed by those who were uninsured (20.6%), those covered by Medicaid (15.1%), those covered by Medicare (11.6%), and those with other commercial insurance (9.3%) (Table 1). Most patients lived in metropolitan areas (72.7%) and had severe head and neck injuries (70.1%). Falls were the most common cause of injury (35.8%), followed by motor vehicle crashes (25.4%). Uninsured patients tended to be younger but did not have major clinical differences compared with the overall population of insured patients (Table 2). Among patients with insurance, those covered by Medicare (most of whom qualify because of disability) had a higher incidence of falls and severe head and neck injuries.
In unadjusted analysis, patients who were admitted rather than transferred were more likely to be older, have insurance, be injured as a result of a fall, and have a severe injury to the chest or abdomen (Table 3). These patients were more likely to be treated on a weekday, in higher-volume EDs, and in metropolitan teaching hospitals located in the Northeast, South, and West.
After adjustment with multivariate regression, many of these factors remained associated with increased risk of admission compared with transfer (Table 4). Compared with patients who were uninsured, those who were covered by Medicaid had a rate of admission vs transfer that was 14.3% (95% CI, 9.2%-19.4%) higher. The absolute risk of admission vs transfer was also higher among patients with other types of insurance compared with patients without insurance: these values were 13.2% (95% CI, 7.5%-18.9%) for Medicare, 11.2% (95% CI, 6.9%-15.4%) for private insurance, and 13.1% (95% CI, 6.6%-19.6%) for other commercial insurance (Table 4 and the Figure).
After multivariate analysis, clinical characteristics associated with increased risk of admission vs transfer were older age (≥35 years) and severe injury to the abdomen (Table 4). Whereas patients with severe injury to the abdomen were admitted 15.9% (95% CI, 9.4%- 22.3%) more often, patients with severe injuries to the head and neck were transferred 13.2% (95% CI, 6.6%-19.7%) more often.
Higher annual ED visit volume and teaching status also remained independently associated with a greater risk of admission vs transfer. The risk of admission vs transfer was most strongly predicted by whether the hospital ED was a teaching hospital. Compared with metropolitan non–teaching hospitals, the risk of admission rather than transfer was 26.2% (95% CI, 15.2%-37.2%) higher if a severely injured patient was initially seen at a metropolitan teaching hospital. Conversely, the risk of admission was 20.4% (95% CI, 9.8%-31.1%) less if a patient was initially seen at a rural hospital rather than a metropolitan non–teaching hospital (Table 4 and the Figure).
Despite adjustment for patient-level, injury-level, and hospital-level characteristics, insured patients and those with initial care in higher-volume urban teaching hospitals had a significantly increased risk of hospitalization in a non–trauma center rather than transfer to a potentially higher level of care. Severely injured trauma patients require a broad spectrum of diagnostic, critical care, and surgical services to optimize outcomes. Therefore, regionalized trauma systems were established beginning in the 1980s to facilitate the field triage and interhospital transfer of severely injured trauma patients to designated trauma centers.23 Given that patients hospitalized in non–trauma centers have worse outcomes than those transferred to trauma centers,3- 6 our findings suggest that insured patients may receive worse care. While unmeasured patient preferences may partially explain this phenomenon, our findings raise the possibility that insured patients are disproportionately being kept at non–trauma centers because of better reimbursement potential compared with uninsured patients.
The first studies examining the association between insurance status and interhospital transfer in the 1980s documented the phenomenon of “dumping” uninsured patients on publicly owned tertiary hospitals.24- 26 Many of these patients were transferred without any stabilizing treatment.25,26 In response to concerns about patient dumping, Congress in 1986 enacted the EMTALA.27 This law imposes specific obligations on hospitals that offer emergency services to provide a medical screening examination for patients initially seen at EDs regardless of an individual’s ability to pay. Hospitals are then required to provide stabilizing treatment for patients found to have an emergent condition. According to the EMTALA, if a hospital is unable to stabilize or treat a patient within its capability or if the patient requests, an appropriate transfer should be implemented.27
Our study is the first to date to use national ED data to show that severely injured patients initially seen at non–trauma center EDs who are uninsured are more likely to be appropriately transferred, whereas patients who are insured are more likely to hospitalized in the non–trauma center. Previous research on this phenomenon has been limited because of a lack of data from non–trauma centers, which account for more than 75% of the hospitals in the United States.16 Three previous studies13- 15 examining trauma registry data found that patients who had noncommercial insurance or who were uninsured were more likely to be transferred from lower-level trauma centers (level III) to tertiary care trauma centers (level I or II). Level III trauma centers have transfer agreements with level I and II trauma centers for patients with exceptionally severe injuries. Level III trauma centers do not have the full availability of surgical subspecialists as level I and II trauma centers but have the resources for emergency resuscitation, surgery, and intensive care for most trauma patients.16 Therefore, our findings demonstrating disparities in transfer patterns of severely injured patients by insurance status who are initially seen at non–trauma center EDs are even more concerning because these hospitals do not have the critical care trauma resources that level III trauma centers have.
While the focus with the implementation of the EMTALA has been to ensure that uninsured patients are not dumped on tertiary care public hospitals, our findings suggest that additional policies are needed to ensure that critically injured insured patients are not inappropriately retained in non–trauma centers rather than transferred to a higher level of care. This is especially necessary when regionalized care for critically ill patients with certain conditions, including trauma, has been demonstrated to improve outcomes. The need to ensure an appropriate level of care extends beyond trauma patients. For example, transfer of low-birth-weight infants to high-volume neonatal intensive care units has been shown to improve outcomes, but infants covered by insurance are less likely to be transferred.28- 30
A subset of critically ill or injured patients likely exists whom physicians at non–trauma centers are willing to admit to their hospital. However, if it is found that a patient is uninsured, with a low likelihood of recouping the costs of providing intensive care, this may tip the scale to transfer the patient.31,32 Likewise, transferring rather than not transferring insured patients is associated with financial loss for the transferring hospital.33 Therefore, policies that allow the sharing of reimbursement between the transferring hospital and the receiving hospital may be a solution to neutralizing the financial conflict of interest for transferring hospitals.34 Such policies have been proposed for the regionalization of out-of-hospital cardiac arrest.35 To facilitate this, reimbursement policies and quality measurement activities need to view the care of an emergency condition originating at one hospital and followed by transfer to another hospital as a single “episode of care” rather than separate health care encounters.36- 38 This would better allow regional acute care systems such as trauma systems to reduce disparities in transfer patterns by insurance status and to ensure that patients are being optimally regionalized.
Some privately insured patients may not want to be transferred to a trauma center given that many trauma centers are publicly owned and located in inner-city areas. However, while a paucity of literature exists on qualitative factors behind interhospital transfer decisions for trauma, preference has not been shown to be an important factor, at least in the medical literature.31,32 Finding that patients do not want to be transferred to a trauma center would highlight the importance of the efforts of the Centers for Disease Control and Prevention39 to educate the public about the survival benefit of trauma center care.
Given the scarcity of neurosurgeons who take call for trauma,40,41 it is not surprising that patients with severe head injuries were much more likely to be transferred out of non–trauma centers than those without severe head injuries. However, it is surprising that patients with severe injuries to the chest or abdomen were more likely to be admitted to the non–trauma center rather than be transferred. While general surgeons can manage most of these injuries, those who work at non–trauma centers are likely to have less experience with managing these complex conditions than general surgeons who work at trauma centers and are trained in trauma resuscitation.42
We also found that presentation to higher-volume, teaching hospitals located in metropolitan areas was significantly associated with a greater likelihood of being admitted rather than transferred among insured patients. While these hospitals are more likely to have several surgical specialists and intensive care resources, processes of care for trauma patients are not likely to be in place compared with designated trauma centers.42 Furthermore, a landmark 2006 national study3 comparing the outcomes of patients treated in level I trauma centers vs in high-volume non–trauma centers found a 25% relative reduction in mortality among patients hospitalized in trauma centers.
This analysis has several limitations. First, because of the limitations of administrative data among patients who had a disposition of transfer to another care facility from the ED of a non–trauma center, we cannot determine what type of hospital patients were actually transferred to. Based on the EMTALA and previous studies,4,13- 15 we assumed that severely injured patients treated in non–trauma center EDs who were transferred would be transferred to a higher level of care and would almost exclusively go to designated trauma centers rather than to another non–trauma center. Second, we were unable to determine the proximity of trauma centers to individual non–trauma centers. Third, administrative data lack more granular details on clinical and physiological characteristics; therefore, some of the variation in transfer rates may be explained by unobserved differences in these variables. Fourth, we were unable using this data set to determine whether patients who were transferred had better outcomes compared with patients who were not transferred.
In summary, we found that insured, critically injured trauma patients are much less likely to be transferred out of non–trauma center EDs than uninsured trauma patients after adjusting for patient, injury, and hospital characteristics. Given that transfer to a trauma center has been shown to reduce mortality, these insured patients may be receiving suboptimal care. Our findings suggest that encounters for time-sensitive critical illness such as trauma should be monitored at the regional level using an episode of care approach to ensure the optimal regionalization of patients according to patient need regardless of ability to pay. To reduce transfer disparities, shared reimbursement schemes may be needed to offset the potential loss of reimbursement that non–trauma centers may experience in transferring rather than admitting insured trauma patients with critical injuries.
Accepted for Publication: July 3, 2013.
Corresponding Author: M. Kit Delgado, MD, MS, Department of Emergency Medicine and Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, and Leonard Davis Institute of Health Economics, University of Pennsylvania, 934 Blockley Hall, 423 Guardian Dr, Philadelphia, PA 19104 (email@example.com).
Published Online: February 19, 2014. doi:10.1001/jamasurg.2013.4398.
Author Contributions: Dr Delgado had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Delgado, Yokell, Wang.
Acquisition of data: Yokell, Wang.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: Delgado.
Critical revision of the manuscript for important intellectual content: Yokell, Staudenmayer, Spain, Hernandez-Boussard, Wang.
Statistical analysis: Delgado, Yokell, Hernandez-Boussard.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by grant UL1 RR025744 from the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health (Drs Delgado and Wang) and award K12HL109009 from the National Heart, Lung, and Blood Institute (Dr Delgado).
Role of the Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the National Heart, Lung, and Blood Institute.
Previous Presentations: This study was presented at the 2013 Society for Academic Emergency Medicine Annual Meeting; May 17, 2013; Atlanta, Georgia; and at the AcademyHealth Annual Research Meeting; June 25, 2013; Baltimore, Maryland.
Correction: This article was corrected on May 1, 2014, for an omission in the Funding/Support section.