Survival Benefit of Transfer to Tertiary Trauma Centers for Major Trauma Patients Initially Presenting to Nontertiary Trauma Centers

Survival Benefit of Transfer to Tertiary Trauma Centers for Major Trauma Patients Initially Presenting to Nontertiary Trauma Centers Tabitha Garwe, PhD, Linda D. Cowan, PhD, Barbara Neas, PhD, Timothy Cathey, MD, Brandon C. Danford, and Patrice Greenawalt RN, MS Abstract Objectives: Recent evidence suggests a measurable reduction in mortality for patients transferred from a nontertiary trauma center (Level III or IV) to a Level I trauma center, but not for those transferred to a Level II trauma center. Whether this can be generalized to a predominantly rural region with fewer tertiary trauma care resources is uncertain. This study sought to evaluate mortality differences for patients initially presenting to nontertiary trauma centers in a predominantly rural region depending on transfer status. Methods: This was a retrospective cohort study of patients initially presenting to 104 nontertiary trauma centers in Oklahoma and meeting the state s criteria for major trauma. Patients dying within 1 hour of emergency department (ED) arrival at the nontertiary trauma center were excluded. The exposure variable of interest was admission status, which was categorized as either transfer to a tertiary (Level I or II) trauma center within 24 hours or admission to a nontertiary trauma center. Propensity scores were used to minimize the selection bias inherent in the decision to admit or transfer a patient for higher-level care. Multiple logistic regression was used to generate three propensity score models: probability of transfer to either a Level I or II, Level I only, and Level II only. Propensity scores were then included as a covariate in multivariable Cox regression models assessing outcome differences between admitted and transferred patients. The outcome of interest was 30-day mortality, defined as death at either the nontertiary trauma center or the tertiary trauma center within 30 days of arrival at the initial Level III IV center s ED. Results: A total of 6,229 patients met study criteria, of whom 2,669 (43%) were transferred to tertiary trauma centers. Of those transferred, 1,422 patients (53%) were transferred to a Level I trauma center. Crude mortality was lower for patients transferred to tertiary trauma centers compared to those remaining at nontertiary trauma facilities (hazard ratio [HR] = 0.59; 95% confidence interval [CI] = 0.48 to 0.72). After adjusting for the propensity to be transferred, Injury Severity Score (ISS), presence of severe head injury, and age, transfer to a tertiary trauma center was associated with a significantly lower 30-day mortality (HR = 0.38; 95% CI = 0.30 to 0.50) compared to admission and treatment at a nontertiary trauma center. The observed survival benefit was similar for patients transferred to a Level I trauma center (HR = 0.36; 95% CI = 0.20 to 0.4) and those transferred to a Level II center (HR = 0.45; 95% CI = 0.33 to 0.61). Conclusions: This study suggests a survival benefit among patients initially presenting to nontertiary trauma centers who are subsequently transferred to tertiary trauma centers compared to those remaining in nontertiary trauma centers, even after adjusting for variables affecting the likelihood of transfer. Although this survival benefit was larger for patients treated at a Level I trauma center, Level II trauma centers in a region with few tertiary trauma resources demonstrated a measurable benefit as well. ACADEMIC EMERGENCY MEDICINE 2010; 17:1223 1232 ª 2010 by the Society for Academic Emergency Medicine From the Trauma Division (TG), Protective Health Services (TC, PG), Oklahoma State Department of Health, Oklahoma City, OK; the Department of Biostatistics and Epidemiology (LDC, BN), University of Oklahoma Health Sciences Center, Oklahoma City, OK; and the University of Oklahoma (BCD), Oklahoma City, OK. Received March 9, 2010; revision received April 22, 2010; accepted April 27, 2010. Supervising Editor: David M. Cline, MD. Address for correspondence and reprints: Tabitha Garwe, PhD; e-mail: Tabitha-Garwe@ouhsc.edu. ª 2010 by the Society for Academic Emergency Medicine ISSN 1069-6563 doi: 10.1111/j.1553-2712.2010.00918.x PII ISSN 1069-6563583 1223

1224 Garwe et al. OUTCOMES IN TRAUMA PATIENT TRANSFERS The goal of an organized trauma system is to reduce mortality and morbidity by ensuring injured patients have access to the most appropriate level of trauma care in the least amount of time. 1 A component of an organized trauma system is classification of acute care facilities within a state or region based on resource depth. The American College of Surgeons Committee on Trauma (ACS-COT) recommends a four-level verification system of acute care facilities. 2 ACS-COT verified Level I and II trauma centers are considered tertiary or definitive care centers and offer the most comprehensive care for severely injured patients. Level III and Level IV trauma centers are considered nontertiary trauma centers and are located primarily in rural areas. Oklahoma is a predominantly rural state with a statewide, mandatory, inclusive trauma system. Legislation passed in 1999 assigned the Oklahoma State Department of Health (OSDH) considerable authority to lead the development and implementation of a sustainable statewide trauma system. Development included a mandatory four-level classification of all acute care hospitals based on the depth of trauma and emergency operative care resources and was completed in 2001. While differences between ACS-COT and Oklahoma s four-level classifications are slight, one variation is the ACS-COT requirement for immediate, in-house (24-hour) availability of certain trauma physician specialists, while Oklahoma s classification system requires on-call availability. Trauma system growth in Oklahoma also included development and implementation of prehospital triage, transport, and interfacility transfer protocols to ensure the most severely injured patients are treated at the most appropriate level of trauma care. 2 The timely interhospital transfer of severely injured trauma patients to Level I II trauma centers is recommended due to scarce resources in the rural areas. Despite these guidelines, in many regions there is some evidence of substantial variability in transfer practices among Level III IV trauma centers (nontertiary). 3 A few studies have evaluated short-term outcome differences between patients transferred from Level III IV trauma centers to Level I II trauma centers and those transported directly from the scene of injury to a Level I or II trauma center. 4 11 There is, however, little information regarding the effect on outcomes for transferred patients compared to those who remained at nontertiary trauma centers. Newgard et al. 12 found lower mortality in patients transferred from the emergency department (ED) of a Level III or IV to a Level I or II trauma center compared to those admitted to Level III IV trauma centers. This lower mortality was greatest among patients transferred to Level I trauma centers (odds ratio [OR] = 0.62; 95% confidence interval [CI] = 0.40 to 0.95), but there was no measurable benefit for patients transferred to Level II trauma centers (OR = 0.82; 95% CI = 0.47 to 1.43). Whether these findings can be generalized to a maturing trauma system in a predominantly rural state, with fewer tertiary trauma care centers, and where a significant number of patients receive definitive care in Level III trauma centers, is uncertain. In Oklahoma, some Level III trauma centers in the rural areas act as regional trauma referral centers. The objective of this study was to determine whether there is a difference in short-term mortality between patients with major trauma transferred to a Level I or II trauma center (tertiary) from a Level III or IV trauma center (nontertiary) compared to major trauma patients initially presenting to and receiving definitive care at Level III or IV trauma centers. METHODS Study Design This was a retrospective cohort study of major trauma patients presenting to 104 non-tertiary trauma centers from January 1, 2006 through December 2007. This study was approved by the Oklahoma State Department of Health and the University of Oklahoma Health Sciences Center Institutional Review Boards. Study Setting and Population As of December 2007 there were 75 Level IV trauma centers (primarily located in the rural areas), 29 Level III trauma centers, two Level II trauma centers, and one Level I trauma center in Oklahoma. Only state-licensed acute care facilities are designated trauma centers; nondesignated hospitals include federal facilities (Veterans Administration Hospitals), Indian Health Service, and specialty hospitals. The State of Oklahoma has a mandatory, inclusive trauma system composed of eight trauma regions. The three tertiary trauma centers are located in the metropolitan areas of two of the eight regions. The only Level I trauma center is located centrally and is ACS verified, while the two Level II trauma centers (of which one is ACS verified) are located in the northeastern part of the state. Patients meeting Oklahoma s definition of major trauma (Appendix I) were identified from the Oklahoma State Trauma Registry. All state-licensed, acute care facilities are required to report data to the state trauma registry on a monthly basis. Additionally, there is a financial incentive to submit complete and accurate data to the registry. Trained trauma registrars at each of the participating hospitals abstract data submitted to the state registry. Patient records from different facilities are routinely linked within the database. A unique identification number is then assigned to records related to the same event. The study included all injured patients presenting to Level III IV trauma centers and meeting major trauma criteria during the 2-year period. Patients were only eligible if they were admitted to a Level III or IV trauma center or required transfer for higher level of trauma care (Level I or II trauma centers) within 24 hours of injury. Instead of restricting eligibility to the most severely injured patients, all patients meeting the criteria for major trauma were included so that results could be more broadly generalized. Patients were excluded if they were transferred from an out-of-state or non state-licensed facility (n = 391) or died in the ED within 1 hour of arrival (n = 68). The 1-hour criterion was chosen because the probability of transfer was lowest (0.1) during the first hour of ED

ACAD EMERG MED November 2010, Vol. 17, No. 11 www.aemj.org 1225 arrival and was the same ( 0.2) in each of the subsequent hours thereafter, up to 5 hours. Study Protocol The outcome of interest was 30-day mortality, defined as death at either the nontertiary trauma center or the tertiary trauma center within 30 days of arrival at the initial Level III ED. The exposure variable of interest was admission status and the type of trauma center in which the patient received definitive care and was categorized as either transfer to a tertiary trauma center or admission to a nontertiary trauma center. Transfer was defined as interfacility transfer from a Level III or IV trauma center to one of the three tertiary trauma centers (i.e., Level I or II) within 24 hours of ED arrival. A risk-adjusted analysis of the association of mortality and transfer from one nontertiary trauma center to another suggested no survival differences (hazard ratio [HR] = 0.71; 95% CI = 0.5 to 1.3), and therefore, patients transferred between nontertiary trauma centers were included in the comparison group (patients admitted to Level III IV trauma centers). Unadjusted comparisons between patients transferred to tertiary trauma centers and those admitted to nontertiary trauma centers were performed using the independent Student s t-test or Mann-Whitney Wilcoxon tests for continuous variables and Pearson s chi-square test for categorical variables. Patients are not randomly selected for interfacility transfer. Rather, those patients with more severe injuries, and hence a higher mortality risk, are more likely to be transferred. This potential selection bias was adjusted using propensity scores. When differences exist in baseline characteristics and these differences are not accounted for in analyses of outcomes, biased estimates of associations between potential risk factors and outcomes may result. 13 Conventional response modeling using such techniques as multiple logistic regression does not usually recognize such imbalances, which may be influential due to effect measure modification or model misspecification. A propensity score is the conditional probability (between 0 and 1) of assignment to a particular group given a set of observed covariates. 13,14 In this case, a collection of variables such as patient s clinical status and trauma resources, which are potentially associated with the decision to transfer the patient, are combined into one composite variable represented by a single score. A propensity-adjusted analysis becomes a means of obtaining quasi-randomization of study groups in that, if two patients, one transferred to a tertiary trauma center and one definitively treated at a nontertiary trauma center, had the same propensity score, then a comparison of outcomes would more likely reflect the facility in which they received care rather than their underlying risk. 15 In this study, the propensity score for each patient was calculated as the probability of transfer to a tertiary trauma center. A multivariable logistic regression model was used to calculate this probability. Three separate propensity score models were generated: probability of transfer to a tertiary trauma center regardless of level, probability of transfer to the Level I trauma center, and probability of transfer to a Level II trauma center (Appendix II). Because both weak and strong confounders can be included in a propensity score model, 16 variables were retained in the model if they met a p-value less than or equal to 0.1. Because of the large number of variables considered for propensity score modeling, testing for all possible two-way (firstorder) interactions would likely yield a multitude of such associations just by chance. We therefore restricted our assessment of first-order interactions to those we had a theoretical interest in and those whose statistical significance was so strong that it would be misleading to suppress them. A more restrictive p-value of 0.05 was thus used to evaluate the statistical significance of first-order interaction terms. The Hosmer-Lemeshow test was used to assess goodness of fit for all propensity score models. Seventeen demographic, clinical, and other hospitalbased variables potentially associated with the decision to transfer a patient to a tertiary level trauma center were considered for inclusion in development of the propensity score. Patient characteristics included age, race, sex, insurance status (private, self-pay, Medicare, Medicaid, other), day of presentation (weekend vs. weekday), and time of day. Clinical factors of interest included initial ED Glasgow Coma Scale score (GCS), shock (initial ED systolic blood pressure < 100 mm Hg), 17 prehospital or ED intubation, mechanism of injury, injury location stratified in one of the following groups based on all available International Classification of Diseases, Ninth Revision (ICD-9), codes: skull fracture and or intracranial injury (ICD-9 codes 800 804, 850 854), any pelvic and or long bone (humerus, femur) fracture (ICD-9 code series 808, 812, and 820 821), any liver or splenic injury (ICD-9 code series 864 and 865), any heart and or lung injury (ICD-9 code series 861), and any pneumothorax and or hemothorax (ICD-9 code series 860). The presence of comorbid condition was coded as yes or no based on reported comorbidity ICD-9 codes. Specific injuries rather than Injury Severity Score (ISS) body regions were considered for propensity score modeling because it was unlikely that the transferring facility would have accurate ISS scores for transferred patients before all definitive diagnoses were known. All ISS body regions were considered in the outcome models only if the Abbreviated Injury Scale (AIS) value for that region was 3. An AIS value of 3 or greater reflected ISS body regions with injuries considered as serious. Other variables considered included the prehospital mode of transport (emergency medical services [EMS] vs. non-ems). Because of some heterogeneity in resources for hospitals designated at the same trauma level, particularly for trauma levels III and IV, both the designated trauma level and the actual trauma resources of the initial hospital were considered. Information on available trauma resources was obtained from the 2006 2007 OSDH Trauma Resource Assessment survey. The trauma region of the initial hospital was used to determine whether transfer was intraregional versus interregional as a surrogate measure for proximity of the initial hospital to tertiary trauma centers. A propensity-adjusted regression analysis of each covariate (independent variable) on transfer status (dependent variable) using propensity scores generated

1226 Garwe et al. OUTCOMES IN TRAUMA PATIENT TRANSFERS from the aforementioned analyses (Appendix II) was done to assess the ability of the propensity scores to create comparable groups (i.e., admitted vs. transferred). Adjusting for the propensity to be transferred removed all differences in covariates included in the score in all three groups used in outcome analyses, with the exception of age differences between patients admitted to Level III IV trauma centers and those transferred to a Level I trauma center. Age was considered in the analyses of outcome differences as a continuous variable. Stratification based on propensity score quintiles has been suggested as an effective means of providing an unbiased estimate of the true exposure effect given there is adequate overlap in all strata. 13 16 A graphical evaluation of propensity score distribution indicated adequate overlap across the range of propensity scores in the analysis including all three analyses groups ( 10% representation of each group in each quintile). However, there was minimal overlap in the first quintile of the propensity score in the analysis assessing transfer to a Level II trauma center and in the first two quintiles of the propensity score in the analysis assessing transfer to a Level I trauma center. This suggested that patients in one study group did not have enough comparable subjects in the other group and therefore any comparisons between these groups may be subject to bias and should be interpreted with caution. We therefore explored holding out patients in these nonoverlapping strata to determine their effect on measures of association in the mortality analyses. Survival analyses were performed using propensity score adjusted Cox proportional hazards models to compare 30-day mortality outcome differences between patients who were and were not transferred. The use of a survival analysis technique takes into account time to death and therefore adjusts for any clustering of deaths particularly in the early period after injury. Time to event was defined as the time elapsed between time of ED arrival at the initial Level III IV trauma center and time of death or discharge. Surviving patients were censored at 30 days (follow-up period). The proportional hazards assumption was evaluated graphically and by including an interaction variable between time and directness of transport. In addition to propensity scores, all outcome models were adjusted for ISS, presence of a severe head injury (AIS score for head region 3), and age. Data Analysis All analyses were performed using SAS Version 9.1 (SAS 9.1, SAS Institute, Cary, NC). RESULTS A total of 6,229 patients initially presenting to nontertiary trauma centers met the study criteria, of whom 2,669 (43%) were transferred to tertiary trauma centers. Of these, 2,624 (98%) were transferred to a tertiary trauma center directly from the ED of the initial facility, with 1,422 (53%) taken to the Level I trauma center. Excluded from the multivariable models were 433 patients (6.9%) because of extensive missing data, which made imputation impossible. Of the 433 excluded from multivariable analyses, 27 were decedents. Crude mortality was the same in those excluded from and those included in the multivariable analyses of outcomes. Of the excluded cases, 142 (33%) received definitive treatment at a Level III or IV trauma center while the remaining 291 were transferred to a Level I or II trauma center. The mean age of patients excluded from analyses of outcomes was 26 years (standard deviation [SD] ± 30 years). Compared to those who remained at nontertiary trauma centers, patients transferred to tertiary trauma centers were significantly younger, less likely to be transported from the scene of injury by EMS, more likely to have traffic-related injuries, disproportionately male, and Medicaid-insured (Table 1). Compared to those patients who were transferred, patients admitted to nontertiary trauma centers were more likely to have fall-related injuries and to be Medicare-insured. A higher proportion of patients presenting to Level IV trauma centers (57%) were transferred to tertiary trauma centers than those patients initially presenting to Level III trauma centers (35%; Table 2). The overall mean length of stay at the initial facility for transferred patients was 186 minutes (SD ± 128 minutes). Clinically, patients transferred to tertiary trauma centers had more severe injuries (as measured by ISS) and lower GCS scores, were more likely to be intubated, and had fewer comorbid conditions compared to those admitted to nontertiary trauma centers (Table 2). There were a few differences between patients transferred to a Level I trauma center and those transferred to a Level II trauma center (Table 2). Compared to patients transferred to Level II trauma centers, patients transferred to a Level I trauma center were significantly younger, had a higher incidence of shock, were more likely to be intubated, had a disproportionately higher number of patients with ISS greater than or equal to 25, higher incidence of internal organ injuries (heart, lung, liver and spleen) and pelvic and long bone fractures, and were disproportionately self-pay and Medicaid-insured. An evaluation of crude mortality suggested that patients transferred to tertiary trauma centers were at a lower risk of mortality within the first 30 days following injury compared to those admitted to nontertiary trauma centers (unadjusted HR = 0.59; 95% CI = 0.48 to 0.72; Figure 1). Because the proportional hazards assumption was met (Figure 2), adjusted results are presented for the total follow-up period of 30 days. After adjusting for the propensity to be transferred, ISS, presence of severe head injury, age, and time to death or discharge, transfer to a tertiary trauma center (Level I or II) was associated with a significantly lower 30-day mortality (HR = 0.38; 95% CI = 0.30 to 0.50) compared to admission and treatment at a Level III or IV trauma center (Table 3). The observed survival benefit was similar for patients transferred to a Level I trauma center (HR = 0.36; 95% CI = 0.20 to 0.40) and for those transferred to a Level II center (HR = 0.45; 95% CI = 0.33 to 0.61). Analyses were repeated excluding patients in quintiles with minimally overlapping propensity scores, since such patients were not sufficiently similar in the

ACAD EMERG MED November 2010, Vol. 17, No. 11 www.aemj.org 1227 Table 1 Demographic and Injury Characteristics in Major Trauma Patients Initially Presenting to Nontertiary Trauma Centers by Transfer Status Variables Admitted to Nontertiary Facility (n = 3,560) Transferred to Level I (n = 1,422) Transferred to Level II (n = 1,247) Mean (±SD) age, yr 54.6 (±24.8)* 27 (±23) 37 (±26) Sex: Male 2,069 (58)* 960 (67) 834 (67) Race White 3,039 (85.4) 1,077 (76) 1,049 (84) Black African American 227 (6.4) 92 (6.5) 70 (5.6) Native American 112 (3.2) 95 (6.7) 101 (8.1) Other 182 (5.1) 158 (11.1) 27 (2.2) Primary payer Private insurance 1,103 (31) 491 (34.5) 443 (35.5) Self-pay 693 (19.5) 358 (25.2) 272 (21.8) Medicaid 256 (7)* 338 (23.8) 230 (18.4) Medicare 1,307 (36.7)* 134 (9.4) 241 (19.3) Other 201 (5.7) 101 (7.1) 61 (4.5) Injury mechanism MVC 790 (22.2)* 453 (32) 350 (28) MC Ped BC 237 (6.7)* 142 (10) 112 (9) GSW Stab 195 (5.5) 77 (5.4) 62 (5) Fall 1,767 (49.6)* 421 (29.6) 444 (35.6) Assault struck 266 (7.5) 134 (9.4) 94 (7.5) Other 305 (8.6) 195 (13.7) 185 (14.8) Time and day of initial ED presentation 6 AM 12 Noon 789 (22.5) 251 (17.9) 255 (20.9) 12 Noon 6 PM 1,267 (36.1) 495 (35.3) 448 (36.7) 6 PM midnight 1,001 (28.5) 480 (34.2) 380 (31.1) Midnight 6 AM 453 (12.9) 178 (12.7) 139 (11.4) Weekend 1,670 (46.9) 716 (50.4) 642 (51.5) Prehospital mode of transport EMS 2,323 (65.3)* 672 (47.3) 577 (46.3) Non-EMS 1,237 (34.7) 750 (52.7) 670 (53.7) Values in parentheses indicate percentage or SD for mean values. BC = bicyclist; EMS = emergency medical services; GSW = gunshot wound; MC = motorcyclist; MVC = motor vehicle crash; Ped = pedestrian. *Significant difference between those who remained at nontertiary trauma centers and those who were transferred. Significant difference between those transferred to a Level I center and those transferred to a Level II center. baseline variables included in the propensity score. Exclusion of these patients accentuated the survival benefit in patients transferred to a Level I trauma center (HR = 0.29; 95% CI = 0.20 to 0.40), but the survival benefit for patients transferred to a Level II trauma center remained essentially unchanged (HR= 0.44; 95% CI = 0.32 to 0.62). Figure 2 summarizes these results. To compare our results to those of a similar prior study, 12 we repeated these analyses after excluding all ED deaths occurring in nontertiary trauma centers and used the same method of analysis (logistic regression). This subanalysis yielded qualitatively similar results (i.e., survival benefit) for patients transferred to a Level I trauma center but not for patients transferred to a Level II trauma center. When all deaths in EDs of the nontertiary trauma centers were excluded in this analysis, there was no longer a significant difference in 30- day mortality between those transferred to Level II trauma centers and those who received definitive care in a Level III or IV trauma center (OR = 0.77; 95% CI = 0.54 to 1.11). DISCUSSION The interfacility transfer of patients is recommended for patients whose injuries are beyond the initial trauma center s resources. These recommendations are designed to ensure that seriously injured patients have access to tertiary (comprehensive) trauma care regardless of their proximity to these resources. Oklahoma is a large and predominantly rural state, and thus it is not practical to transport all patients with major injury directly from the scene to a tertiary trauma center. Patients injured in rural parts of the state are often evaluated first in local nontertiary trauma centers where medical providers make a decision between attempting to provide definitive care or transferring the patient to a higher level trauma center. There is little information on the outcome benefit of transferring patients for higher level trauma care and differences in tertiary trauma care outcomes. Although transferred patients in our study had more severe injuries, results suggest that there was a survival benefit associated with transfer of such patients to tertiary trauma centers. Earlier studies evaluating differences between Level I and Level II trauma center outcomes may have been limited in their methodologies, which were based primarily on comparing outcomes using the survival probability (Trauma and Injury Severity Score [TRISS]) derived from the Major Trauma Outcome Study. 18 21 This methodology was subsequently shown to be unreliable, with a high misclassification

1228 Garwe et al. OUTCOMES IN TRAUMA PATIENT TRANSFERS Table 2 Clinical and Initial Hospital Characteristics in Major Trauma Patients Initially Presenting to Nontertiary Trauma Centers by Transfer Status Variable Admitted to Nontertiary Facility (n = 3,560) Transferred to Level I (n = 1,422) Transferred to Level II (n = 1,247) Trauma level of initial hospital Level III 2,599 (73) 836 (58.8) 560 (44.9) Level IV 961 (27) 586 (41.2) 687 (55.1) Trauma resources of initial hospital No trauma physician specialist 489 (13.7) 338 (23.8) 300 (24.1) Limited GS Ortho 270 (7.6) 197 (13.9) 207 (16.7) GS Ortho 941 (26.5) 464 (32.6) 508 (40.9) GS Ortho Neuro 484 (13.6) 92 (6.5) 182 (14.7) GS Ortho Neuro Other 1,374 (38.6) 331 (23.3) 45 (3.6) Mean ISS (±SD) 12.6 (±6.4)* 16.2 (±9.7) 15.1 (±8.4) ISS Group 1 15 2,555 (71.8) 802 (56.4) 720 (57.7) 16 24 767 (21.5) 361 (25.4) 363 (29.1) 25 238 (6.7) 259 (18.2) 164 (13.2) Severe head injury (AIS 3) 1,039 (29.2) 517 (36.4) 564 (45.2) Initial ED mean GCS (±SD) 14.3 (±2.8)* 13.7 (±3.1) 13.9 (±2.8) Initial ED sbp < 100 mm Hg 266 (7.7) 166 (13.1) 102 (9.2) EMS or ED intubation 281 (8)* 192 (13.6) 153 (12.3) Comorbid condition present 1,163 (32.3)* 299 (21) 240 (19.3) Injury distribution Skull fracture intracranial injury 1,203 (33.9)* 666 (47) 561 (45) Pneumo hemothorax 452 (12.7) 211 (15) 138 (11) Heart lung 267 (7.5)* 184 (13) 145 (11.7) Liver spleen 160 (4.5)* 152 (10.7) 73 (6) Pelvic femur humerus 1,354 (38)* 498 (35) 271 (21) Overall 30-day mortality 285 (8) 68 (4.8) 64 (5.1) Values in parentheses indicate percentage or SD for mean values. AIS = Abbreviated Injury Score; GS = general surgery; GCS = Glasgow Coma Scale; ISS = Injury Severity Score; Neuro = neurosurgery; Ortho = orthopedics; sbp = systolic blood pressure. *Significant difference between those who remained at nontertiary trauma centers and those who were transferred. Significant difference between those transferred to a Level I center and those transferred to a Level II center. Figure 1. Thirty-day cumulative risk* of death by survival time and admit status. *Adjusted for propensity score (for interhospital transfer), presence of severe head injury, ISS, and age.

ACAD EMERG MED November 2010, Vol. 17, No. 11 www.aemj.org 1229 Figure 2. HRs for 30-day in-hospital mortality in patients transferred to Level I or II compared to patients admitted to nontertiary trauma centers using unadjusted and propensity score-adjusted Cox regression analyses. L1 = Level I; LII = Level II. *Excludes quintiles 1and 2; **excludes the first quintile. (h) 95% CIs; ( ) HR. HR = hazard ratio. Table 3 Multivariable Analyses (Cox Regression) of the Association of Directness of Transport and 30-Day Mortality in Major Trauma Patients Transferred to Tertiary Trauma Centers 30-Day Mortality Variables Level I or II Level I Level II Transfer 0.38 (0.3 0.5) 0.36 (0.21 0.4) 0.45 (0.33 0.61) Propensity score quintile 1.60 (1.4 1.8) 1.50 (1.3 1.8) 1.07 (0.97 1.2) Severe head injury 1.50 (1.2 1.9) 1.50 (1.1 1.9) 1.52 (1.2 1.9) ISS 1.05 (1.04 1.06) 1.05 (1.04 1.06) 1.05 (1.04 1.06) Age 1.03 (1.02 1.04) 1.03 (1.02 1.04) 1.01 (1.01 1.02) Data are reported as HR (95% CI). ISS = Injury Severity Score; HR = hazard ratio. percentage, particularly in patients with severe injuries (ISS 20). 22 In this study, we adjusted for both injury severity and the selection bias inherent in the decision to admit patients to nontertiary facilities or transfer them for higher-level trauma care. The benefit of transferring patients to Level I trauma centers has been previously demonstrated. 12 Newgard et al., however, did not find any measurable benefit for patients transferred to Level II trauma centers. Our results demonstrate an overall measurable benefit for patients transferred to Level II trauma centers, although this was 38% less than that observed for patients transferred to a Level I trauma center. This divergence in findings between this study and prior studies is likely explained by differences in the availability of tertiary trauma care resources as well as some of the exclusion criteria, particularly for early deaths. Only three acute care facilities are considered tertiary trauma centers in our study (50% fewer than studied by Newgard et al. 12 ), and we excluded patients dying within 1 hour of initial ED arrival, while Newgard et al. excluded all initial facility ED deaths. When the same exclusion criteria were applied, we also observed no measurable mortality benefit for patients transferred to Level II trauma centers. While the demonstrated outcome benefit of transferring patients for higher level care validates the interfacility transfer recommendations, a few points warrant further discussion. We previously reported in a similar population that patients transferred to a Level I trauma center are at higher risk for in-hospital mortality compared to patients transported directly from scene. 11 Thus, these results should not be construed to imply that the Oklahoma trauma system is functioning optimally. Efforts should continue to be focused on reducing the time between initial ED arrival at nontertiary

1230 Garwe et al. OUTCOMES IN TRAUMA PATIENT TRANSFERS facility and arrival at the tertiary trauma center. Patients in our study stayed, on average, 3 hours at the initial nontertiary trauma center before transfer. Ideally the decision to transfer a patient should be based on the patient s clinical status and the resources available at the initial facility. The results of this study, however, suggest a lack of uniformity regarding interhospital transfer practices among Level III and IV trauma centers (Appendix II). Other nonclinical factors were associated with the decision to transfer, particularly for patients transferred to a Level I trauma center. Patients without private (commercial) insurance were more likely to be transferred to a Level I trauma center. Patient age was independently associated with the transfer decision regardless of whether the patient was transferred to a Level I or II trauma center. Older patients were less likely to be transferred. While not all elder injured patients require tertiary trauma care, age alone should not be used as a determinant for interhospital transfer. Other studies have reported similar findings, particularly patient age and insurance status. 3,23 This suggests a broader problem than just regional variation in transfer practices. In addition to nonclinical factors associated with transfer decisions, there was a lack of uniformity between patients transferred to a Level I and those transferred to a Level II trauma center in some clinical factors such as presence of shock and intubation status. If mortality reduction is to be realized on a broader scale, objective and standardized interfacility transfer protocols will likely need to be implemented across all nontertiary facilities to reduce variability in transfer practices. Interfacility transfer guidelines exist in the State of Oklahoma. Reinforcement of these guidelines and continued education of medical providers in nontertiary trauma centers is warranted. LIMITATIONS Patients were those initially presenting to nontertiary trauma centers in a predominantly rural state with a mandatory trauma system, and hence these findings may not be generalizable to urban trauma systems, regions without an organized trauma system, or patients transported directly to tertiary trauma centers. The study did not identify specific subgroups of patients or patients with injuries or combinations of injuries who would benefit from interfacility transfer, and therefore, these findings should not be construed to mean that all transferred patients would benefit from transfer. While propensity scores were used to minimize the selection bias inherent in the decision to admit or transfer a patient to a higher-level care trauma center, completely eliminating this bias is impossible. Unmeasured factors, such as provider experience and the quality, timing, and appropriateness of care at the initial facility, were not available, yet may have affected both the decision to transfer and the outcomes. CONCLUSIONS There appears to be a survival benefit among patients initially presenting to nontertiary trauma centers who are subsequently transferred to tertiary trauma centers, compared to those remaining in nontertiary trauma centers. Further research should focus on whether there are specific subgroups of patients who would especially benefit from transfer to tertiary trauma centers to prevent excessive triage to tertiary trauma centers. Identification of such subgroups would allow for more efficient utilization of trauma care resources and hence better trauma system performance. References 1. Eastman AB, Lewis FR Jr, Champion HR, Mattox KL. Regional trauma system design: critical concepts. Am J Surg. 1987; 154:79 87. 2. American College of Surgeons Committee on Trauma. Resources for the Optimal Care of the Injured Patient. Chicago, IL: American College of Surgeons, 1999. 3. Newgard CD, McConnell KJ, Hedges JR. Variability of trauma transfer practices among non-tertiary care hospital emergency departments. Acad Emerg Med. 2006; 13:746 54. 4. Nathens AB, Maier RV, Brundage SI, Jurkovich GJ, Grossman DC. The effect of interfacility transfer on outcome in an urban trauma system. J Trauma. 2003; 55:444 9. 5. Rogers FB, Osler TM, Shackford SR, Cohen M, Camp L, Lesage M. Study of the outcome of patients transferred to a Level I hospital after stabilization at an outlying hospital in a rural setting. J Trauma. 1999; 46:328 33. 6. Obremskey W, Henley MB. A comparison of transferred versus direct admission orthopedic trauma patients. J Trauma. 1994; 36:373 6. 7. Rivara FP, Koepsell TD, Wang J, Nathens A, Jurkovich GA, MacKenzie EJ. Outcomes of trauma patients after transfer to a Level I trauma center. J Trauma. 2008; 64:1594 9. 8. Sampalis JS, Denis R, Frechette P, Brown R, Fleiszer D, Mulder D. Direct transport to tertiary trauma centers versus transfer from lower level facilities: impact on mortality and morbidity among patients with major trauma. J Trauma. 1997; 43:288 95. 9. Hartl R, Gerber LM, Iacono L, Ni Q, Lyons K, Ghajar J. Direct transport within an organized state trauma system reduces mortality in patients with severe traumatic brain injury. J Trauma. 2006; 60:1250 6. 10. Young JS, Bassam D, Cephas GA, Brady WJ, Butler K, Pomphrey M. Interhospital versus direct scene transfer of major trauma patients in a rural trauma system. Am Surg. 1998; 64:88 91. 11. Garwe T, Cowan L, Neas B, Sacra JC, Albrecht R. Directness of transport of major trauma patients to a Level I trauma center: a propensity-adjusted survival analysis of the impact on short-term mortality. J Trauma. 2010; [Epub ahead of print]. 12. Newgard CD, McConnell KJ, Hedges JR, Mullins RJ. The benefit of higher level of care transfer of injured patients from nontertiary hospital emergency departments. J Trauma. 2007; 63:965 71.

ACAD EMERG MED November 2010, Vol. 17, No. 11 www.aemj.org 1231 13. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983; 70:41 55. 14. Newgard CD, Hedges JR, Arthur M, Mullins RJ. Advanced statistics: the propensity score a method for estimating treatment effect in observational research. Acad Emerg Med. 2004; 11:953 61. 15. D Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998; 17:2265 81. 16. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997; 127(8 Pt 2):757 63. 17. Eastridge BJ, Salinas J, McManus JG, et al. Hypotension begins at 110 mm Hg: redefining hypotension with data. J Trauma. 2007; 63:291 7. 18. Champion HR, Copes WS, Sacco WJ, et al. The Major Trauma Outcome Study: establishing national norms for trauma care. J Trauma. 1990; 30:1356 65. 19. Barone JE, Ryan MC, Cayten CG, Murphy JG. Is 24-hour operating room staff absolutely necessary for Level II trauma center designation? J Trauma. 1993; 34:878 82. 20. Thompson CT, Bickell WH, Siemens RA, Sacra JC. Community hospital Level II trauma center outcome. J Trauma. 1992; 32:336 41. 21. Clancy TV, Gary MJ, Covington DL, Brinker CC, Blackman D. A statewide analysis of Level I and II trauma centers for patients with major injuries. J Trauma. 2001; 51:346 51. 22. Demetriades D, Chan L, Velmanos GV, et al. TRISS methodology: an inappropriate tool for comparing outcomes between trauma centers. J Am Coll Surg. 2001; 193:250 4. 23. Nathens AB, Maier RV, Copass MK, Jurkovich GJ. Payer status: the unspoken triage criterion. J Trauma. 2001; 50:776 83. Appendix I Oklahoma s Definition of Major Trauma Inclusion criteria (must meet at least one criteria in each category) 1. ICD-9-CM code of 800.00 959.9 2. Abbreviated Injury Scale 3; or ISS 9; or TRISS survival probability < *0.50; or Dead on arrival or died while in hospital. 3. Length of hospital stay > 48 hours; or Transferred from a lower level to a higher level trauma center with major trauma; or Dead on arrival or died while in hospital Admitted to intensive care unit; or Admitted to operating room for major surgery (head, chest, abdomen, vascular); or Hospital trauma team activated. Exclusion criteria 1. Persons who died at the scene, or 2. Any of the following as the sole type of injury Isolated orthopedic injury to the extremities due to same level fall Overexertion injuries Submersions Poisonings Asphyxiation Injuries caused by a preexisting condition (osteoporosis, etc.). ICD-9-CM = International Classification of Disease Ninth Revision, Clinical Modification; ISS = Injury Severity Score; TRISS = Trauma and Injury Severity Score. Appendix II Variables Included in the Propensity Score Models Predicting Transfer to a Level I or II, Level I, and Level II Trauma Centers Transfer to Level I or II Transfer to Level I Transfer to Level II Variable Coefficient (SE) OR (95% CI) Coefficient (SE) OR (95% CI) Coefficient (SE) OR (95% CI) Age, yr 0.04 (0.001) 0.96 (0.96 0.97) 0.05 (0.002) 0.95 (0.95 0.96) 0.02 (0.001) 0.97 (0.97 0.098) Race White Reference Reference Reference Black African 0.26 (0.13) 0.77 (0.6 0.99) 0.21 (0.157) 0.81 (0.6 1.1) 0.32 (0.16) 0.72 (0.53 0.98) American Native American 0.41 (0.14) 1.58 (1.2 2.1) 0.43 (0.171) 1.54 (1.1 2.2) 0.52 (0.17) 1.69 (1.2 2.3) Other 0.12 (0.13) 1.1 (0.87 1.44) 0.55 (0.142) 1.74 (1.3 2.3) 0.95 (0.23) 0.39 (0.24 0.61) Penetrating injury 0.23 (0.06) 0.63 (0.5 0.81) 0.24 (0.08) 0.61 (0.46 0.83) Prehospital transport EMS Reference Reference Reference Non-EMS 0.17 (0.03) 1.4 (1.3 1.6) 0.14 (0.04) 1.32 (1.12 1.54) 0.14 (0.04) 1.3 (1.1 1.5) Transferring region Intraregion Reference Reference Interregion 0.11 (0.05) 1.24 (1.0 1.5) 0.57 (0.06) 0.38 (0.25 0.40)

1232 Garwe et al. OUTCOMES IN TRAUMA PATIENT TRANSFERS Appendix II (Continued) Transfer to Level I or II Transfer to Level I Transfer to Level II Variable Coefficient (SE) OR (95% CI) Coefficient (SE) OR (95% CI) Coefficient (SE) OR (95% CI) Injury Other Reference Reference Reference Skull intracranial 0.47 (0.05) 1.8 (1.6 2.2) 0.5 (0.06) 2.6 (2.1 3.2) 0.52 (0.06) 1.6 (1.3 1.9) Internal organ* 0.06 (0.08) 1.2 (0.98 1.53) 0.1 (0.09) 1.7 (1.3 2.3) 0.07 (0.1) 1 (0.77 1.3) Pelvic long bone 0.40 (0.06) 0.77 (0.64 0.92) 0.15 (0.07) 1.35 (1.1 1.7) 0.64 (0.08) 0.5 (0.4 0.6) Initial ED sbp 0.004 (0.001) 1 (1.002 1.004)à ED or EMS intubation 0.26 (0.05) 1.7 (1.4 2.1) 0.25 (0.06) 1.66 (1.3 2.1) Comorbid present 0.11 (0.04) 1.2 (1.1 1.4) 0.25 (0.06) 1.65 (1.4 2) Trauma resources Level III vs. IV 0.54 (0.03) 0.34 (0.3 0.39) 0.36 (0.04) 0.49 (0.4 0.6) 0.74 (0.04) 0.23 (0.19 0.27) Primary payer Private insurance Reference Reference Self-pay 0.02 (0.08) 0.98 (0.83 1.1) 0.06 (0.1) 1.1 (0.87 1.29) Medicaid 0.28 (0.11) 1.32 (1.07 1.63) 0.31 (0.12) 1.37 (1.07 1.75) Medicare 0.23 (0.11) 1.26 (1.02 1.6) 0.32 (0.15) 1.39 (1.03 1.9) Other 0.15 (0.13) 1.16 (0.9 1.5) 0.36 (0.15) 1.43 (1.06 1.9) Weekend 0.06 (0.03) 1.12 (0.99 1.27) SE = standard error; sbp = systolic blood pressure; *Includes heart, lung, liver, and spleen. Femur humerus. àodds for every 10-unit increase in sbp.