ORIGINAL ARTICLE
Validation of prehospital trauma triage criteria for motor vehicle collisions Giana H. Davidson, MD, MPH, Frederick P. Rivara, MD, MPH, Christopher D. Mack, MS, Robert Kaufman, Gregory J. Jurkovich, MD, and Eileen M. Bulger, MD, Seattle, Washington
Triage of an injured patient to an appropriate trauma center can have an impact on morbidity and mortality. We sought to validate the 2012 national field triage guidelines for motor vehicle crashes. METHODS: This is a retrospective cross-sectional study using the National Automotive Sampling System Crashworthiness Data System. Vehicle damage criteria proposed as prehospital triage guidelines were correlated with injury severity. RESULTS: There were 85,761 individuals representing 29,397,234 occupants in car crashes from 2003 to 2008. Of the patients, 3.8% met physiologic Step 1 criteria with a mean Injury Severity Score (ISS) of 9.1 (95% confidence interval [CI], j3.1 to 20.9); Step 1 had a positive predictive value (PPV) of 20.8% (95% CI, 20.1Y21.4%) for severe injury (ISS 9 15). Of the patients, 0.43% met anatomic Step 2 criteria alone, with a mean ISS of 18.1 (95% CI, 16.4Y19.8) and a PPV of 48.5% (95% CI, 46.8Y50.1%). Step 3 criteria include injury mechanism; 3.7% of the patients met one of these criteria alone with a mean ISS of 5.1 (95% CI, 4.4Y5.8) and a PPV of 9.7% (95% CI, 9.3Y10.2%). Patients who met only Step 3 criteria were examined to determine crash characteristics that predict severe injury. Intrusion of greater than 12 inches had a PPV of 10.4% (95% CI, 9.5Y11.3); steering wheel collapse had a PPV of 25.7% (95% CI, 23.0Y28.4%). Older patients (age 9 55 years) who met Step 3 mechanism criteria had higher predictive values for injury for nearly all crash characteristics. CONCLUSION: Injury mechanism criteria alone predict significant injury in a substantial proportion of patients who did not meet the physiologic or anatomic criteria. Vehicular crash data could improve the ability of emergency medical service providers to triage injured occupants. Consideration of transport to a trauma center should be given for elderly patients and drivers with steering wheel collapse. (J Trauma Acute Care Surg. 2014;76: 755Y761. Copyright * 2014 by Lippincott Williams & Wilkins) LEVEL OF EVIDENCE: Epidemiologic study, level III. KEY WORDS: Triage; motor vehicle crash; trauma center; mechanism of injury. BACKGROUND:
N
ational field triage guidelines were introduced by the American College of Surgeons’ Committee on Trauma Optimal Hospital Resources document in 1986. They have been revised, most recently in January 2012 by an expert panel convened by the Centers for Disease Control and Prevention (CDC).1Y5 The goal of these recommendations is to aid prehospital emergency medical service (EMS) providers in identifying patients at risk for severe injury and transporting them to appropriate facilities. The decision to transport a patient to a trauma center impacts outcomes. In a recent study, severely injured trauma patients treated at Level I trauma centers were shown to have a 25% reduction in mortality compared with patients treated at nontrauma centers,6 leading the CDC expert panel to suggest that patients meeting Step 1 or 2 criteria should Submitted: July 8, 2013, Revised: October 15, 2013, Accepted: October 15, 2013. From the Harborview Injury Prevention and Research Center (G.H.D., F.P.R., C.D.M., R.K., E.M.B.), and the Departments of Surgery (G.H.D., E.M.B.), Pediatrics (F.P.R.), and Epidemiology (F.P.R.), University of Washington, Harborview Medical Center, Seattle, Washington; and Department of Surgery (G.J.J.), University of Colorado School of Medicine, Denver Health and Hospitals Authority, Aurora, Colorado. The views expressed are those of the authors and do not represent the views of NHTSA. Address for reprints: Giana H. Davidson, MD, MPH, University of Washington, Box 356410, Seattle, WA 98195; email: [email protected]. DOI: 10.1097/TA.0000000000000091
be preferentially triaged to a Level I trauma center. Step 3 and 4 patients are recommended to be triaged to any trauma center with the assistance of a medical control. Various systems have been used to define the severity and extent of injuries. The Abbreviated Injury Scale (AIS) is widely accepted and ranks the severity of injury by body region. The Injury Severity Score (ISS) uses an ordinal scale (1Y75) derived from AIS scores and has been accepted as a method to determine overall severity of injury.7 The National Study on Costs and Outcomes of Trauma defined major trauma as an ISS of 16 or greater. For patients with blunt trauma, survival has been shown to decrease with increasing ISS (915) and age (954 years).8 Thus, it is vital that injured patients are matched to appropriate level trauma centers and emphasis should be placed on identifying older patients who may have severe injuries early in triage.9,10 The triage decision scheme for patients in motor vehicle collisions has undergone revisions to optimize field triage protocols used by EMS systems.4,5 The goal is to match severity of injury with level of care to reduce both undertriage and overtriage. The National Expert Panel of Field Triage used a threshold of 20% positive predictive value (PPV) to use an indicator to predict severe injury.4,5 Severe injury was defined as an ISS greater than 15, major surgery, or intensive care unit admission to place a criterion in discussion for inclusion in the decision scheme in mechanism of injury.11 Step 1 criteria
J Trauma Acute Care Surg Volume 76, Number 3
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
755
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
focus on the patient’s physiologic response, including vital signs and level of consciousness; Step 2 identifies specific anatomic injuries likely requiring specialty treatment. Mechanism of injury and vehicular damage are assessed in Step 3. High-speed impacts and severe injury are linked; however, the usefulness of estimating vehicular speed and $V (or force of the crash) has been limited owing to the challenges in estimation.12 Step 4 includes age criteria, burn patients, and those with specific medical comorbidities. Patients meeting any of the Step 3 or 4 criteria should be transported to the closest trauma center, with medical control helping to decide if bypassing a closer hospital for a higher level trauma center is needed.11 The purposes of this study were to validate the triage guidelines for motor vehicle crashes using the National Automotive Sampling System Crashworthiness Data System (NASS/ CDS) and to determine the likelihood of significant injury based on each step in the triage guidelines. We hypothesized that the Step 3 mechanisms alone predict significant occult injury.
PATIENTS AND METHODS Data Source We completed a cross-sectional study using the NASS/ CDS from 2003 to 2008. CDS is a nationwide crash data collection program sponsored by US Department of Transportation National Highway Traffic Safety Administration (NHTSA). A CDS crash must fulfill the following requirements: must be police reported, must involve a harmful event including property damage or personal injury resulting from a crash, and must involve one towed passenger car or light truck or van in transport on a traffic way. Data are collected in 27 primary sampling units throughout the United States on vehicles involved in approximately 5,000 crashes annually, with an oversampling of crashes resulting in fatalities and serious injuries.13 These data can be weighted to represent all police-reported, tow-away crashes in the United States. Injuries to vehicle occupants are documented and scored using AIS.14 Relevant crash and occupant information, based on crash reconstruction data, was extracted from the NASS database, including age, sex, seat position, intrusion at occupant’s seat position, maximum intrusion in the crash vehicle, death of a passenger in same compartment, restraint, crash type, $V, air bag deployment, ejection, entrapment (defined as 920-minute extrication), ISS, initial recorded vital signs, transport to a Level I trauma center, and specific injuries as described later. The institutional review board of the University of Washington approved this study.
Analysis This was a retrospective cross-sectional study including all patients in the NASS CDS database from 2003 to 2008 to determine the likelihood of serious trauma based on vehicle damage sustained in a crash as described in Step 3 of the CDC field triage decision scheme.5 The field triage guidelines are a decision scheme for EMS providers. Step 1 rapidly identifies critical injuries based on physiologic data. Patients that meet Step 1 criteria should be transported to a trauma center.15 Physiologic criteria alone are specific but not sensitive in identifying severe injuries. Those 756
that do not meet Step 1 criteria are evaluated by Step 2 of the triage guidelines. Step 2 identifies injuries that warrant evaluation and treatment at a trauma center15 (Fig. 1). We used the following AIS codes to identify Step 2 criteria injuries: flail chest, two or more long-bone fractures, amputation, pelvic fracture, significant skull fracture, or paralysis. Patients who do not meet Step 1 or 2 criteria may still have severe occult injuries. Step 3 of the decision scheme identifies patients by injury mechanisms and vehicle telemetry. For auto crashes, these include intrusion greater than 12 inches at the occupant site, greater than 18 inches at any site in the vehicle (including the roof ), partial or complete ejection, death of another occupant in the same compartment, and vehicle telemetry data consistent with a high risk of injury or the actual $V reading transmitted from the crashed vehicle. We included all of these variables with the exception of telemetry data, which are not available to first responders. In addition to the CDC Step 3 criteria, we examined complete steering wheel collapse (defined as two half sections of the steering wheel deformed axially downward, beyond the hub), roof crush of 12 or 18 inches anywhere in the vehicle, and extrication time longer than 20 minutes to determine if these additional criteria predict significant injury. We defined moderate injury as an ISS greater than 8 and ISS greater than 15 representing severe injury. We used multiple imputation to handle the bias due to missing data using a method of multiple imputation by chained equations.16,17 Variables in the imputation model included age, sex, ISS, regional AIS, Glasgow Coma Scale (GCS), ejection, belt use, airbag availability, hospital stay, death in vehicle, intrusion, plane of contact, telemetry data including $V and vital signs. We used a stepwise decision scheme to calculate the proportion of patients involved in a motor vehicle collision that met each step criterion in sequential order. Of those who did not meet Step 1 criteria, we calculated the proportion of patients that met Step 2 criteria. Those who did not meet Step 1 or 2 criteria were used to determine the proportion that met Step 3 criteria (Fig. 1). We determined the mean ISS and proportion of severely injured in each tier. We did not examine Step 4 criteria because information on comorbidities was not present in the NASS database; however, we did look separately at the predictive value of Step 3 criteria on older adult (age Q 55 years) crash patients. To validate the use of mechanism as a predictor of significant injury, we identified patients who did not meet Step 1 (vital sign) or Step 2 criteria (specific injury patterns) and isolated those who met Step 3 criteria alone. We calculated sensitivity and specificity of the individual mechanism criteria to identify predictive values of moderate and severe injury for patients who do not meet Step 1 or 2 criteria the during initial triage. We repeated this process for the population of patients older than 54 years (Tables 2 and 3). Statistical analysis was generated using STATA SE 11.2 (Stata Corp, College Station, TX) and SAS (SAS software, version 9.2 of the SAS System for Windows, Cary, NC) appropriately accounting for weighted survey and imputed data.
RESULTS This study examined weighted data of 85,761 individuals representing 29,397,234 patients in car crashes in the United * 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
Figure 1. CDC guidelines for trauma triage, 2012.5
States from 2003 to 2008. The mean age of the study population was 33 years, with 46% female; the proportion with older than 55 years and younger than 15 years were 14.5% and 21.5%, respectively. The mean ISS for the entire population was 1.5 (95% confidence interval [CI], 1.3Y1.7) (Table 1). For the entire population, 3.8% met criteria for Step 1 (GCS score G 14, systolic blood pressure G 90 mmHg, or respiratory rate G 10 or 929 breaths/min). The mean ISS of those who met Step 1 criteria was 9.1 (95% CI, j3.1 to 20.9). Of those who did not meet Step 1 criteria, 0.43% met Step 2 criteria (penetrating injury, flail chest, Q2 long-bone fractures, amputation, unstable pelvic fracture, skull fracture, or paralysis), and the mean ISS for this group was 18.1 (95% CI, 16.4Y19.8). An additional 3.7% of the patients who did not meet Step 1 or 2 criteria were in a vehicle that met Step 3 criteria (intrusion 9 12 inches
at the occupant’s site or 918 inches anywhere, ejection, death in the same passenger compartment). The mean ISS of this group was 5.1 (95% CI, 4.4Y5.8), with 21.9% having an ISS of 9 or greater and 9.7% of patients having an ISS greater than 15 (95% CI, 7.1Y12.1). Of those who did not meet any of the field triage criteria Steps 1 through 3, there were only a small proportion of patients with moderate injury (ISS 9 8, 2.2%; 95% CI, 1.8Y2.5%) or severe injury (0.6%; 95% CI, 0.6Y0.8%) (Fig. 2). Predictive values were calculated to determine criteria most likely to be predictive of significant injury as determined by ISS. In evaluating the entire population, we found that Step 1 had a PPVof 20.8% (95% CI, 20.1Y21.4%) for ISS greater than 15 and a PPV of 26.1% (95% CI, 25.4Y26.8%) for an ISS greater than 8. Not surprisingly, meeting Step 2 criteria had a high predictive value for an ISS of greater than 15 and 8 (48.5%;
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
757
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
TABLE 1. Characteristics of Occupants Involved in Crashes, National Automotive Sampling System Crash Worthiness System, 2003 to 2008 Proportion (95% CI) Sex, female Age Age 9 55 y Age G 18 y ISS ISS 9 8 ISS 9 15 Intrusion 12 inches, occupant site Intrusion 18 inches, anywhere Death in vehicle Steering wheel collapse Roof crush 12 inches, anywhere Roof crush 18 inches, anywhere Ejection Entrapment
46.1% (44.5Y47.8%) Mean, 33.36 (32.9Y33.8) 14.5% (13Y16.1%) 21.5% (20.4Y22.5%) Mean, 1.5 (1.34Y1.66) 4.1% (3.5Y4.8%) 1.9% (1.6Y2.2%) 1.7% (1.3Y2.2) 1.5% (0.1Y0.2%) 1.0% (0.8Y1.1%) 0.2% (0.01Y0.3%) 1.1% (0.7Y1.4%) 0.3% (0.2Y0.4%) 1.8% (1.4Y2.2%) 8.6% (6.2Y11%)
95% CI, 46.8Y50.1%; and 92.9%; 95% CI, 91.2Y94.5%, respectively). Meeting Step 3 criteria without meeting Step 1 or 2 criteria had a PPV of 9.7% (95% CI, 9.3Y10.2%) for ISS greater than 15 and a PPV of 22% (95% CI, 21.2Y2.9%) for ISS greater than 8 (Table 2). We found that Step 3 does predict significant injury in a substantial proportion of patients who did not meet any previous criteria. Therefore, we examined the population who met Step 3 criteria alone (patients meeting Step 1 or 2 criteria were eliminated from the analysis) to determine vehicle crash characteristics that best predict moderate and severe occult injury. We found that for an ISS of greater than 15, intrusion of more than 12 inches at the occupant site had a PPV of 10.4% (95% CI, 9.5Y11.3%) and a sensitivity of 41.1% (95% CI, 39.2Y43.1%). Death in the vehicle had a PPV of 21.4% (95% CI, 20.5Y22.4%), and ejection had a PPV of 9.8% (95% CI, 9.1Y10.5%). In addition to published triage criteria, we included steering wheel collapse, which had the highest PPV of 25.7% (95% CI, 23Y28.4%). PPVs of moderate injury (ISS 9 8) showed a similar trend (Table 2). For older patients (age 9 54 years), we found that for severe injury (ISS 9 15), PPVs were higher for nearly all vehicle markers compared with the general population with a higher sensitivity for intrusion at the occupant site (47.8%; 95% CI, 42.4Y53.3%) and death in the vehicle (28.5%; 95% CI, 24.9Y32.1%). Similar results were found with vehicle markers for moderate injury with intrusion of 12 inches at the occupants site in an elderly crash patient being twice that of the general public (PPV, 53.02% vs. 23.8%) (Table 3).
CONCLUSION It is important to quickly identify patients in motor vehicle crashes with a high risk of severe injury. In our study, patients who met Step 1 of the CDC field triage criteria had a large range of ISSs. Previous studies from the New York state trauma registry show a reduced odds of death on transport to a 758
Level 1 trauma center if meeting physiologic criteria.18 Our nationwide study showed that nearly 4% of the patients have physiologic parameters that support transport to a trauma center with a large range of injury scores. Patients who met Step 2 anatomic criteria were a small proportion of the entire population (0.43%) but had a high likelihood of having significant injuries requiring specialty care at a trauma center. Step 2 had the highest PPV for having a significant injury, which is not surprising as our definition of significant injury relies on ISS alone. This research supports a recent multicenter retrospective study by Newgard et al.,19 which found that Step 3 criteria plays a major role in identifying patients with significant injury and risk of death who otherwise do not meet Step 1 or 2 criteria. In the current study, 3.7% of patients were identified by Step 3 criteria alone, with 22% presenting with an ISS greater than 8 and 9.7% with an ISS greater than 15. This is particularly important because it indicates that a large proportion of patients may have occult injuries not easily recognized by first responders. This supports training for first responders in the assessment of component intrusion at the patient’s location, including steering wheel collapse for drivers and maintaining a high index of suspicion for occult injury in patients who meet Step 3 criteria alone but do not meet physiologic or anatomic criteria. This may be particularly important in elderly patients who may not show early signs of shock such as tachycardia due to medications. Newgard et al.19 showed that undertriaged patients were likely to be older, have less physiologic derangement, and have a higher rate of surgical intervention. In those 55 years or older, a higher PPV was found for nearly all markers of moderate and severe injury when compared with the general population. Intrusion and death in the vehicle are more sensitive markers for injury for the older population than the general public. Identification of occult severe injuries in older patients is critical for the appropriate triage of this population. Assessment of injury mechanisms improves EMS provider ability to appropriately triage to a trauma centers. Previous studies have varied in their support for death in the vehicle as a marker for significant injury.20Y22 Death within the same occupant compartment was found to be a good predictor of severe injury within the current guidelines (PPV, 21.4%; 95% CI, 20.5Y22.4%) and had the highest predictive value (PPV, 39%; 95% CI, 36.1Y42.5%) in the older population. The utility of measuring intrusion has been controversial because there is significant variation in vehicular design and construction leading to improved protection in the transfer of energy to the occupant over time. In addition, vehicular mismatch is associated with severe injury confounding the singular measure of intrusion alone.23 Extrication time greater than 20 minutes was dropped from the 2009 decision scheme because the panel recognized that various factors including weather, vehicle type, darkness, and need for additional equipment all confound accuracy of recorded times.24 We found that patients meeting criteria for entrapment had a PPV for severe injury similar to ejection and intrusion criteria. Entrapment is typically caused by significant vehicular deformation and integrity loss surrounding the occupant, both markers already included in the triage decision scheme. To gain better measurements of cabin intrusion and presumably improved ability to assess risk of significant * 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
Figure 2. NASS CDS crash data including patients who met criteria following triage guidelines for each step. These data are survey data and imputed to account for missing values. A patient included in Step 3 will not have met criteria for Step 1 or 2. However, a patient in Step 1 may have met criteria for Step 2 and/or 3.
injury, future research should include these data in the development of Advanced Automatic Collision Notification technologies. In addition to the factors included in the triage guidelines, we examined additional crash characteristics that may be predictive of severe occult injury. We found that complete steering wheel collapse had the highest predictive value for severe injury (25.7%; 95% CI, 23Y28.4%) in patients who met Step 3 criteria alone and should be given consideration for future revisions of the triage guidelines. Roof crush alone was
comparable in its PPVs for injury to intrusion elsewhere in the vehicle. This supports its inclusion into the most current revision guidelines.5 The National Expert Panel of Field Triage is responsible for revising the decision scheme. This working subgroup used an ISS greater than 15, prompt operative care, intensive care unit admission, and case fatality rates in determining the appropriate injury mechanisms for inclusion. Criterion for the discussion of inclusion of mechanism of injury was a 20%
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
759
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
TABLE 2. Predictive Values, Sensitivity, and Specificity of Triage Criteria in Predicting Moderate or Severe Injury, Entire Population PPV, % (95% CI) Predictive values for ISS 9 8, entire population Step 1 26.1 (25.4Y26.8) Step 2 criteria (not Step 1) 92.9 (91.2Y94.5) Step 3 criteria (not Step 1 or Step 2) 22.0 (21.2Y22.9) Predictive values for ISS 9 15, entire population Step 1 20.8 (20.1Y21.4) Step 2 criteria (not Step 1) 48.5 (46.8Y50.1) Step 3 criteria (not Step 1 or Step 2) 9.7 (9.3Y10.2) Predictive values for ISS 9 8, for those who met only Step 3 criteria Intrusion 12 inches, occupant site 23.8 (22Y25.5) Intrusion 18 inches, anywhere 18.7 (17.2Y20.2) Death in vehicle 34.7 (33.6Y35.8) Steering wheel collapse 45.2 (41.9Y48.5) Roof crush 12 inches, anywhere 22.0 (20.3Y23.7) Roof crush 18 inches, anywhere 19 (16.3Y21.7) Ejection 23.1 (21.7Y24.4) Entrapment 22.3 (20.8Y23.8) Predictive values for ISS 915, for those who met only Step 3 criteria Intrusion 12 inches, occupant site 10.4 (9.5Y11.3) Intrusion 18 inches, anywhere 7.6 (7.0Y8.3) Death in vehicle 21.4 (20.5Y22.4) Steering wheel collapse 25.7 (23.0Y28.4) Roof crush 12 inches, occupant site 8.2 (7.4Y8.9) Roof crush 18 inches, anywhere 6.5 (5.7Y7.3) Ejection 9.8 (9.1Y10.5) Entrapment 10.3 (9.4Y11.2)
NPV, % (95% CI)
Sensitivity, % (95% CI)
Specificity, % (95% CI)
96.7 (96.8Y97.0) 97.1 (97.0Y97.1) 97.8 (97.8Y97.9)
24.0 (23.6Y24.5) 12.1 (11.7Y12.5) 27.9 (27.1Y28.8)
97.0 (96.9Y97.1) 99.9 (99.9Y99.9) 97.03 (96.9Y97.1)
98.8 (98.8Y98.9) 99 (99.0Y99.1) 99.4 (99.4Y99.4)
41.7 (40.9Y42.5) 18 (17.3Y18.6) 37.6 (36.4Y38.8)
96.9 (96.8Y97.0) 99.8 (99.8Y99.8) 96.6 (96.5Y96.7)
79.1 (78.2Y80.0) 76.3 (75.3Y77.3) 79.6 (78.6Y80.5) 78.2 (77.4Y79.1) 78.0 (77.0Y78.9) 77.8 (76.9Y78.7) 78.7 (77.6Y79.8) 78.1 (77.1Y79.1)
41.6 (40.1Y43.0) 28.3 (27.1Y29.6) 17.3 (16.5Y18.0) 1.99 (1.8Y2.2) 17.2 (15.9Y18.6) 6 (5.1Y6.9) 41.0 (39.5Y42.6) 26.4 (25.3Y27.6)
62.4 (60.2Y64.5) 65.2 (63.1Y67.3) 90.8 (90.4Y91.2) 99.3 (99.2Y99.4) 82.7 (81.8Y83.7) 92.8 (92.2Y93.3) 61.4 (59.6Y63.2) 74.1 (72.4Y75.7)
90.7 (90.2Y91.1) 89.2 (88.6Y89.8) 91.7 (91.2Y92.2) 90.4 (90.0Y90.9) 89.9 (89.4Y90.5) 90 (89.5Y90.5) 90.3 (89.7Y90.9) 90.5 (89.9Y91.0)
41.1 (39.2Y43.1) 26.2 (24.9Y27.5) 21.1 (22.8Y25.4) 2.6 (2.3Y2.9) 14.5 (13.3Y15.6) 4.6 (4.1Y5.1) 39.2 (37.3Y41.1) 27.6 (25.9Y29.3)
61.8 (59.9Y63.7) 65.8 (64.0Y67.7) 90.5 (90.1Y90.8) 99.2 (99.1Y99.3) 82.4 (81.5Y83.3) 92.8 (92.3Y93.3) 60.9 (59.3Y62.5) 74.1 (72.7Y74.1)
PPV to predict severe injury.24 From our study, death within the vehicle and steering wheel collapse had a PPV of greater than 20% for an ISS greater than 15 meeting Step 3 criteria alone. NASS/CDS had no available inpatient data. It would be
beneficial in future studies to have other markers for severe injury including the need for urgent operative intervention, transfusion, specialty care, or intensive care admission in future prospective studies. A limitation of this study, despite the
TABLE 3. Predictive Values, Sensitivity, and Specificity of Triage Criteria in Predicting Moderate or Severe Injury, Occupants 55 Years and Older PPV, % (95% CI) Predictive values for ISS 9 8, for those that met only step 3 criteria Intrusion 12 inches, occupant site 53.02 (48.7Y57.4) Intrusion 18 inches, anywhere 32.7 (27.6Y37.9) Death in vehicle 56.9 (53.7Y60.1) Steering wheel collapse 54.2 (45.3Y63.0) Roof crush 12 inches, anywhere 35.4 (28.2Y42.5) Roof crush 18 inches, anywhere 60.4 (50.0Y70.8) Ejection 23.5 (18.9Y28.0) Entrapment 22.7 (18.3Y27.1) Predictive values for ISS 915, for those who met only Step 3 criteria Intrusion 12 inches, occupant site 28.8 (24.3Y33.3) Intrusion 18 inches, anywhere 9.3 (8.1Y10.5) Death in vehicle 39.3 (36.1Y42.5) Steering wheel collapse 36.7 (28.5Y44.9) Roof crush 12 inches, occupant site 6.2 (4.9Y7.4) Roof crush 18 inches, anywhere 5.5 (3.3Y7.7) Ejection 16.7 (12.6Y20.5) Entrapment 9.3 (7.4Y11.2)
760
NPV, % (95% CI)
Sensitivity, % (95% CI)
Specificity, % (95% CI)
70.93 (67.8Y74.1) 59.11 (55.2Y63.0) 65.3 (61.7Y68.8) 61.5 (58.4Y64.5) 60.6 (57.2Y64.0) 63.3 (60.3Y66.3) 53.9 (50.8Y56.9) 55.9 (52.6Y59.3)
54.7 (50.6Y58.8) 23.81 (19.5Y28.1) 25.6 (22.9Y28.3) 0.76 (0.58Y0.93) 18.1 (13.7Y22.5) 12.5 (8.0Y17.0) 20.2 (17.3Y23.0) 15.1 (13.1Y17.0)
69.54 (66.1Y73.0) 69.3 (65.9Y72.6) 87.8 (86.3Y89.4) 99.6 (0.93) 79.2 (76.3Y82.1) 94.9 (93.9Y95.8) 58.7 (53.4Y64.0) 67.7 (83.0Y86.9)
79.2 (76.7Y81.8) 70.3 (66.9Y73.7) 79.3 (76.4Y82.1) 76.1 (73.6Y78.6) 71.7 (68.6Y74.7) 74.4 (71.8Y77.1) 72.4 (69.6Y75.3) 71.0 (68.0Y74.0)
47.8 (42.4Y53.3) 10.9 (9.3Y12.4) 28.5 (24.9Y32.1) 0.8 (0.59Y1.1) 5.1 (4.1Y6.0) 1.8 (1.2Y2.4) 23.13 (19.1Y27.2) 9.9 (8.5Y11.4)
62.7 (59.2Y66.3) 66.5 (63.2Y69.9) 86.1 (84.6Y87.6) 99.6 (99.4Y99.7) 75.6 (72.5Y78.8) 90.1 (97.6Y98.8) 63.7 (59.0Y68.4) 69.4 (65.0Y73.8)
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
fact that the use of trauma scores is a well-established method of validating triage,21,25Y27 is the use of ISS alone as a marker for significant injury. Intrusion alone predicts severe injury in patients not meeting physiologic or anatomic criteria moderately well (10.4% and 7.6% for 12 inches and 18 inches of intrusion, respectively). While there was poor sensitivity for steering wheel collapse, steering wheel collapse has a PPV of 25.7% for all ages and 36.7% for those 54 years or older. Further distribution and implementation of Advanced Automatic Collision Notification technology may need to assess measurements of interior vehicular deformation that would provide better feedback to dispatch appropriate resources and should be currently used in on-scene assessments. Consideration of transport to a trauma center should be given to those found to have steering wheel collapse because this was associated with a high risk of significant injury. AUTHORSHIP G.H.D. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, drafting of the manuscript, and critical revisions. F.P.R. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, drafting of the manuscript, and critical revisions. C.D.M. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, and drafting of the manuscript. R.K. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, and drafting of the manuscript. G.J.J. contributed in the study conception and design, analysis and interpretation of data, and critical revisions. E.M.B. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, and drafting of the manuscript, critical revisions.
ACKNOWLEDGMENT We thank Jin Wang, MS, from the Harborview Injury Prevention and Research Center for her statistics and programming assistance. She did not receive financial compensation for her work.
DISCLOSURE This work was supported by National Institute of Child Health and Human Development grant 1T32HD057822-01AZ that was awarded to the Pediatric Injury Research Training Program and performed for the Crash Injury Research and Engineering network (CIREN) project at the University of Washington/Harborview Injury Prevention and Research Center in cooperation with the US Department of Transportation/National Highway Traffic Safety Administration (USDOT/NHTSA). Funding has been provided by NHTSA under Cooperative Agreement Number DTNH22-05H-11001 and DTNH22-10-H-00295.
REFERENCES 1. Mackersie RC. History of trauma field triage development and the American College of Surgeons criteria. Prehosp Emerg Care. 2006;10(3):287Y294. 2. Field categorization of trauma patients. Bull Am Coll Surg. 1986;71:17Y21. 3. Hospital and prehospital resources for optimal care of the injured patient. Committee on Trauma of the American College of Surgeons. Bull Am Coll Surg. 1986;71(10):4Y23. 4. Sasser SM, Hunt RC, Sullivent EE, et al. Guidelines for field triage of injured patients. Recommendations of the National Expert Panel on Field Triage. MMWR Recomm Rep. 2009;58(RR-1):1Y35.
5. Centers for Disease Control and Prevention (CDC). Guidelines for Field Triage of Injured Patients Recommendations of the National Expert Panel on Field Triage. MMWR Recomm Rep. 2012;61:1Y23. 6. Mackenzie EJ, Rivara FP, Jurkovich GJ, et al. A national evaluation of the effect of trauma center care on mortality. N Engl J Med. 2006;354:366Y378. 7. Baker SP, O’Neill B, Haddon W Jr, Long WB. The Injury Severity Score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974;14(3):187Y196. 8. Champion HR, Copes WS, Sacco WJ, et al. The Major Trauma Outcome Study: establishing national norms for trauma care. J Trauma. 1990; 30(11):1356Y1365. 9. Sampalis JS, Denis R, Lavoie A, et al. Trauma care regionalization: a process-outcome evaluation. J Trauma. 1999;46(4):565Y579; discussion 579Y581. 10. 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(2):288Y295; discussion 295Y286. 11. Center for Disease Control and Prevention. MMWR Recomm Rep. 2008; 57:1Y34. 12. Center for Disease Control and Prevention. Recommendations From the Expert Panel: Advanced Automatic Collision Notification and Triage of the Injured Patient. Atlanta, GA: US Department of Health and Human Services; 2008. 13. NASSN. 2011. Available at: www-nr.nhtsa.dot.gove/department/nrd-30/ ncsa/NASS.html. Accessed March 2012. 14. AAAM Abbreviated Injury Scale, 2005 Revision. Barrington, IL: Association for the Advancement of Automotive Medicine; 2005. 15. Wuerz RTJ, Smith JS. Accuracy of trauma triage in patients transported by helicopter. Air Med J. 1996;15:168Y170. 16. van Buuren S, Boshuizen HC, Knook DL. Multiple imputation of missing blood pressure covariates in survival analysis. Stat Med. 1999;18(6):681Y694. 17. Royston P. Multiple imputation of missing value: update. Stata J. 2005; 5(2):181Y201. 18. Henry MC. Trauma triage: New York experience. Prehosp Emerg Care. 2006;10(3):295Y302. 19. Newgard CD, Zive D, Holmes JF, et al. A multisite assessment of the American College of Surgeons Committee on Trauma field triage decision scheme for identifying seriously injured children and adults. J Am Coll Surg. 2011;213(6):709Y721. 20. Henry MC, Hollander JE, Alicandro JM, Cassara G, O’Malley S, Thode HC Jr. Incremental benefit of individual American College of Surgeons trauma triage criteria. Acad Emerg Med. 1996;3(11):992Y1000. 21. Knopp R, Yanagi A, Kallsen G, Geide A, Doehring L. Mechanism of injury and anatomic injury as criteria for prehospital trauma triage. Ann Emerg Med. 1988;17(9):895Y902. 22. Palanca S, Taylor DM, Bailey M, Cameron PA. Mechanisms of motor vehicle accidents that predict major injury. Emerg Med (Fremantle). 2003; 15(5Y6):423Y428. 23. Mandell SP, Mack CD, Bulger EM. Motor vehicle mismatch: a national perspective. Inj Prev. 2010;16(5):309Y314. 24. Wang SW SS, Jurkovich GJ. Universal Medical Rescue Protocol Changed: ‘‘High Speed Auto Crash’’ Changed to ‘‘High Risk Auto Crash’’ in the Field Triage Decision Scheme. Paper Number 09-0207. Publisher National Highway Traffic Safety Administration: Available at http://www-nrd.nhtsa.dot.gov/ pdf/esv/esv21/09-0207.pdf. 25. Norcross ED, Ford DW, Cooper ME, Zone-Smith L, Byrne TK, Yarbrough DR 3rd. Application of American College of Surgeons’ field triage guidelines by pre-hospital personnel. J Am Coll Surg. 1995;181(6):539Y544. 26. Cooper ME, Yarbrough DR, Zone-Smith L, Byrne TK, Norcross ED. Application of field triage guidelines by pre-hospital personnel: is mechanism of injury a valid guideline for patient triage? Am Surg. 1995;61(4):363Y367. 27. Kreis DJ Jr, Fine EG, Gomez GA, Eckes J, Whitwell E, Byers PM. A prospective evaluation of field categorization of trauma patients. J Trauma. 1988;28(7):995Y1000.
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
761
Validation of prehospital trauma triage criteria for motor vehicle collisions Giana H. Davidson, MD, MPH, Frederick P. Rivara, MD, MPH, Christopher D. Mack, MS, Robert Kaufman, Gregory J. Jurkovich, MD, and Eileen M. Bulger, MD, Seattle, Washington
Triage of an injured patient to an appropriate trauma center can have an impact on morbidity and mortality. We sought to validate the 2012 national field triage guidelines for motor vehicle crashes. METHODS: This is a retrospective cross-sectional study using the National Automotive Sampling System Crashworthiness Data System. Vehicle damage criteria proposed as prehospital triage guidelines were correlated with injury severity. RESULTS: There were 85,761 individuals representing 29,397,234 occupants in car crashes from 2003 to 2008. Of the patients, 3.8% met physiologic Step 1 criteria with a mean Injury Severity Score (ISS) of 9.1 (95% confidence interval [CI], j3.1 to 20.9); Step 1 had a positive predictive value (PPV) of 20.8% (95% CI, 20.1Y21.4%) for severe injury (ISS 9 15). Of the patients, 0.43% met anatomic Step 2 criteria alone, with a mean ISS of 18.1 (95% CI, 16.4Y19.8) and a PPV of 48.5% (95% CI, 46.8Y50.1%). Step 3 criteria include injury mechanism; 3.7% of the patients met one of these criteria alone with a mean ISS of 5.1 (95% CI, 4.4Y5.8) and a PPV of 9.7% (95% CI, 9.3Y10.2%). Patients who met only Step 3 criteria were examined to determine crash characteristics that predict severe injury. Intrusion of greater than 12 inches had a PPV of 10.4% (95% CI, 9.5Y11.3); steering wheel collapse had a PPV of 25.7% (95% CI, 23.0Y28.4%). Older patients (age 9 55 years) who met Step 3 mechanism criteria had higher predictive values for injury for nearly all crash characteristics. CONCLUSION: Injury mechanism criteria alone predict significant injury in a substantial proportion of patients who did not meet the physiologic or anatomic criteria. Vehicular crash data could improve the ability of emergency medical service providers to triage injured occupants. Consideration of transport to a trauma center should be given for elderly patients and drivers with steering wheel collapse. (J Trauma Acute Care Surg. 2014;76: 755Y761. Copyright * 2014 by Lippincott Williams & Wilkins) LEVEL OF EVIDENCE: Epidemiologic study, level III. KEY WORDS: Triage; motor vehicle crash; trauma center; mechanism of injury. BACKGROUND:
N
ational field triage guidelines were introduced by the American College of Surgeons’ Committee on Trauma Optimal Hospital Resources document in 1986. They have been revised, most recently in January 2012 by an expert panel convened by the Centers for Disease Control and Prevention (CDC).1Y5 The goal of these recommendations is to aid prehospital emergency medical service (EMS) providers in identifying patients at risk for severe injury and transporting them to appropriate facilities. The decision to transport a patient to a trauma center impacts outcomes. In a recent study, severely injured trauma patients treated at Level I trauma centers were shown to have a 25% reduction in mortality compared with patients treated at nontrauma centers,6 leading the CDC expert panel to suggest that patients meeting Step 1 or 2 criteria should Submitted: July 8, 2013, Revised: October 15, 2013, Accepted: October 15, 2013. From the Harborview Injury Prevention and Research Center (G.H.D., F.P.R., C.D.M., R.K., E.M.B.), and the Departments of Surgery (G.H.D., E.M.B.), Pediatrics (F.P.R.), and Epidemiology (F.P.R.), University of Washington, Harborview Medical Center, Seattle, Washington; and Department of Surgery (G.J.J.), University of Colorado School of Medicine, Denver Health and Hospitals Authority, Aurora, Colorado. The views expressed are those of the authors and do not represent the views of NHTSA. Address for reprints: Giana H. Davidson, MD, MPH, University of Washington, Box 356410, Seattle, WA 98195; email: [email protected]. DOI: 10.1097/TA.0000000000000091
be preferentially triaged to a Level I trauma center. Step 3 and 4 patients are recommended to be triaged to any trauma center with the assistance of a medical control. Various systems have been used to define the severity and extent of injuries. The Abbreviated Injury Scale (AIS) is widely accepted and ranks the severity of injury by body region. The Injury Severity Score (ISS) uses an ordinal scale (1Y75) derived from AIS scores and has been accepted as a method to determine overall severity of injury.7 The National Study on Costs and Outcomes of Trauma defined major trauma as an ISS of 16 or greater. For patients with blunt trauma, survival has been shown to decrease with increasing ISS (915) and age (954 years).8 Thus, it is vital that injured patients are matched to appropriate level trauma centers and emphasis should be placed on identifying older patients who may have severe injuries early in triage.9,10 The triage decision scheme for patients in motor vehicle collisions has undergone revisions to optimize field triage protocols used by EMS systems.4,5 The goal is to match severity of injury with level of care to reduce both undertriage and overtriage. The National Expert Panel of Field Triage used a threshold of 20% positive predictive value (PPV) to use an indicator to predict severe injury.4,5 Severe injury was defined as an ISS greater than 15, major surgery, or intensive care unit admission to place a criterion in discussion for inclusion in the decision scheme in mechanism of injury.11 Step 1 criteria
J Trauma Acute Care Surg Volume 76, Number 3
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
755
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
focus on the patient’s physiologic response, including vital signs and level of consciousness; Step 2 identifies specific anatomic injuries likely requiring specialty treatment. Mechanism of injury and vehicular damage are assessed in Step 3. High-speed impacts and severe injury are linked; however, the usefulness of estimating vehicular speed and $V (or force of the crash) has been limited owing to the challenges in estimation.12 Step 4 includes age criteria, burn patients, and those with specific medical comorbidities. Patients meeting any of the Step 3 or 4 criteria should be transported to the closest trauma center, with medical control helping to decide if bypassing a closer hospital for a higher level trauma center is needed.11 The purposes of this study were to validate the triage guidelines for motor vehicle crashes using the National Automotive Sampling System Crashworthiness Data System (NASS/ CDS) and to determine the likelihood of significant injury based on each step in the triage guidelines. We hypothesized that the Step 3 mechanisms alone predict significant occult injury.
PATIENTS AND METHODS Data Source We completed a cross-sectional study using the NASS/ CDS from 2003 to 2008. CDS is a nationwide crash data collection program sponsored by US Department of Transportation National Highway Traffic Safety Administration (NHTSA). A CDS crash must fulfill the following requirements: must be police reported, must involve a harmful event including property damage or personal injury resulting from a crash, and must involve one towed passenger car or light truck or van in transport on a traffic way. Data are collected in 27 primary sampling units throughout the United States on vehicles involved in approximately 5,000 crashes annually, with an oversampling of crashes resulting in fatalities and serious injuries.13 These data can be weighted to represent all police-reported, tow-away crashes in the United States. Injuries to vehicle occupants are documented and scored using AIS.14 Relevant crash and occupant information, based on crash reconstruction data, was extracted from the NASS database, including age, sex, seat position, intrusion at occupant’s seat position, maximum intrusion in the crash vehicle, death of a passenger in same compartment, restraint, crash type, $V, air bag deployment, ejection, entrapment (defined as 920-minute extrication), ISS, initial recorded vital signs, transport to a Level I trauma center, and specific injuries as described later. The institutional review board of the University of Washington approved this study.
Analysis This was a retrospective cross-sectional study including all patients in the NASS CDS database from 2003 to 2008 to determine the likelihood of serious trauma based on vehicle damage sustained in a crash as described in Step 3 of the CDC field triage decision scheme.5 The field triage guidelines are a decision scheme for EMS providers. Step 1 rapidly identifies critical injuries based on physiologic data. Patients that meet Step 1 criteria should be transported to a trauma center.15 Physiologic criteria alone are specific but not sensitive in identifying severe injuries. Those 756
that do not meet Step 1 criteria are evaluated by Step 2 of the triage guidelines. Step 2 identifies injuries that warrant evaluation and treatment at a trauma center15 (Fig. 1). We used the following AIS codes to identify Step 2 criteria injuries: flail chest, two or more long-bone fractures, amputation, pelvic fracture, significant skull fracture, or paralysis. Patients who do not meet Step 1 or 2 criteria may still have severe occult injuries. Step 3 of the decision scheme identifies patients by injury mechanisms and vehicle telemetry. For auto crashes, these include intrusion greater than 12 inches at the occupant site, greater than 18 inches at any site in the vehicle (including the roof ), partial or complete ejection, death of another occupant in the same compartment, and vehicle telemetry data consistent with a high risk of injury or the actual $V reading transmitted from the crashed vehicle. We included all of these variables with the exception of telemetry data, which are not available to first responders. In addition to the CDC Step 3 criteria, we examined complete steering wheel collapse (defined as two half sections of the steering wheel deformed axially downward, beyond the hub), roof crush of 12 or 18 inches anywhere in the vehicle, and extrication time longer than 20 minutes to determine if these additional criteria predict significant injury. We defined moderate injury as an ISS greater than 8 and ISS greater than 15 representing severe injury. We used multiple imputation to handle the bias due to missing data using a method of multiple imputation by chained equations.16,17 Variables in the imputation model included age, sex, ISS, regional AIS, Glasgow Coma Scale (GCS), ejection, belt use, airbag availability, hospital stay, death in vehicle, intrusion, plane of contact, telemetry data including $V and vital signs. We used a stepwise decision scheme to calculate the proportion of patients involved in a motor vehicle collision that met each step criterion in sequential order. Of those who did not meet Step 1 criteria, we calculated the proportion of patients that met Step 2 criteria. Those who did not meet Step 1 or 2 criteria were used to determine the proportion that met Step 3 criteria (Fig. 1). We determined the mean ISS and proportion of severely injured in each tier. We did not examine Step 4 criteria because information on comorbidities was not present in the NASS database; however, we did look separately at the predictive value of Step 3 criteria on older adult (age Q 55 years) crash patients. To validate the use of mechanism as a predictor of significant injury, we identified patients who did not meet Step 1 (vital sign) or Step 2 criteria (specific injury patterns) and isolated those who met Step 3 criteria alone. We calculated sensitivity and specificity of the individual mechanism criteria to identify predictive values of moderate and severe injury for patients who do not meet Step 1 or 2 criteria the during initial triage. We repeated this process for the population of patients older than 54 years (Tables 2 and 3). Statistical analysis was generated using STATA SE 11.2 (Stata Corp, College Station, TX) and SAS (SAS software, version 9.2 of the SAS System for Windows, Cary, NC) appropriately accounting for weighted survey and imputed data.
RESULTS This study examined weighted data of 85,761 individuals representing 29,397,234 patients in car crashes in the United * 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
Figure 1. CDC guidelines for trauma triage, 2012.5
States from 2003 to 2008. The mean age of the study population was 33 years, with 46% female; the proportion with older than 55 years and younger than 15 years were 14.5% and 21.5%, respectively. The mean ISS for the entire population was 1.5 (95% confidence interval [CI], 1.3Y1.7) (Table 1). For the entire population, 3.8% met criteria for Step 1 (GCS score G 14, systolic blood pressure G 90 mmHg, or respiratory rate G 10 or 929 breaths/min). The mean ISS of those who met Step 1 criteria was 9.1 (95% CI, j3.1 to 20.9). Of those who did not meet Step 1 criteria, 0.43% met Step 2 criteria (penetrating injury, flail chest, Q2 long-bone fractures, amputation, unstable pelvic fracture, skull fracture, or paralysis), and the mean ISS for this group was 18.1 (95% CI, 16.4Y19.8). An additional 3.7% of the patients who did not meet Step 1 or 2 criteria were in a vehicle that met Step 3 criteria (intrusion 9 12 inches
at the occupant’s site or 918 inches anywhere, ejection, death in the same passenger compartment). The mean ISS of this group was 5.1 (95% CI, 4.4Y5.8), with 21.9% having an ISS of 9 or greater and 9.7% of patients having an ISS greater than 15 (95% CI, 7.1Y12.1). Of those who did not meet any of the field triage criteria Steps 1 through 3, there were only a small proportion of patients with moderate injury (ISS 9 8, 2.2%; 95% CI, 1.8Y2.5%) or severe injury (0.6%; 95% CI, 0.6Y0.8%) (Fig. 2). Predictive values were calculated to determine criteria most likely to be predictive of significant injury as determined by ISS. In evaluating the entire population, we found that Step 1 had a PPVof 20.8% (95% CI, 20.1Y21.4%) for ISS greater than 15 and a PPV of 26.1% (95% CI, 25.4Y26.8%) for an ISS greater than 8. Not surprisingly, meeting Step 2 criteria had a high predictive value for an ISS of greater than 15 and 8 (48.5%;
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
757
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
TABLE 1. Characteristics of Occupants Involved in Crashes, National Automotive Sampling System Crash Worthiness System, 2003 to 2008 Proportion (95% CI) Sex, female Age Age 9 55 y Age G 18 y ISS ISS 9 8 ISS 9 15 Intrusion 12 inches, occupant site Intrusion 18 inches, anywhere Death in vehicle Steering wheel collapse Roof crush 12 inches, anywhere Roof crush 18 inches, anywhere Ejection Entrapment
46.1% (44.5Y47.8%) Mean, 33.36 (32.9Y33.8) 14.5% (13Y16.1%) 21.5% (20.4Y22.5%) Mean, 1.5 (1.34Y1.66) 4.1% (3.5Y4.8%) 1.9% (1.6Y2.2%) 1.7% (1.3Y2.2) 1.5% (0.1Y0.2%) 1.0% (0.8Y1.1%) 0.2% (0.01Y0.3%) 1.1% (0.7Y1.4%) 0.3% (0.2Y0.4%) 1.8% (1.4Y2.2%) 8.6% (6.2Y11%)
95% CI, 46.8Y50.1%; and 92.9%; 95% CI, 91.2Y94.5%, respectively). Meeting Step 3 criteria without meeting Step 1 or 2 criteria had a PPV of 9.7% (95% CI, 9.3Y10.2%) for ISS greater than 15 and a PPV of 22% (95% CI, 21.2Y2.9%) for ISS greater than 8 (Table 2). We found that Step 3 does predict significant injury in a substantial proportion of patients who did not meet any previous criteria. Therefore, we examined the population who met Step 3 criteria alone (patients meeting Step 1 or 2 criteria were eliminated from the analysis) to determine vehicle crash characteristics that best predict moderate and severe occult injury. We found that for an ISS of greater than 15, intrusion of more than 12 inches at the occupant site had a PPV of 10.4% (95% CI, 9.5Y11.3%) and a sensitivity of 41.1% (95% CI, 39.2Y43.1%). Death in the vehicle had a PPV of 21.4% (95% CI, 20.5Y22.4%), and ejection had a PPV of 9.8% (95% CI, 9.1Y10.5%). In addition to published triage criteria, we included steering wheel collapse, which had the highest PPV of 25.7% (95% CI, 23Y28.4%). PPVs of moderate injury (ISS 9 8) showed a similar trend (Table 2). For older patients (age 9 54 years), we found that for severe injury (ISS 9 15), PPVs were higher for nearly all vehicle markers compared with the general population with a higher sensitivity for intrusion at the occupant site (47.8%; 95% CI, 42.4Y53.3%) and death in the vehicle (28.5%; 95% CI, 24.9Y32.1%). Similar results were found with vehicle markers for moderate injury with intrusion of 12 inches at the occupants site in an elderly crash patient being twice that of the general public (PPV, 53.02% vs. 23.8%) (Table 3).
CONCLUSION It is important to quickly identify patients in motor vehicle crashes with a high risk of severe injury. In our study, patients who met Step 1 of the CDC field triage criteria had a large range of ISSs. Previous studies from the New York state trauma registry show a reduced odds of death on transport to a 758
Level 1 trauma center if meeting physiologic criteria.18 Our nationwide study showed that nearly 4% of the patients have physiologic parameters that support transport to a trauma center with a large range of injury scores. Patients who met Step 2 anatomic criteria were a small proportion of the entire population (0.43%) but had a high likelihood of having significant injuries requiring specialty care at a trauma center. Step 2 had the highest PPV for having a significant injury, which is not surprising as our definition of significant injury relies on ISS alone. This research supports a recent multicenter retrospective study by Newgard et al.,19 which found that Step 3 criteria plays a major role in identifying patients with significant injury and risk of death who otherwise do not meet Step 1 or 2 criteria. In the current study, 3.7% of patients were identified by Step 3 criteria alone, with 22% presenting with an ISS greater than 8 and 9.7% with an ISS greater than 15. This is particularly important because it indicates that a large proportion of patients may have occult injuries not easily recognized by first responders. This supports training for first responders in the assessment of component intrusion at the patient’s location, including steering wheel collapse for drivers and maintaining a high index of suspicion for occult injury in patients who meet Step 3 criteria alone but do not meet physiologic or anatomic criteria. This may be particularly important in elderly patients who may not show early signs of shock such as tachycardia due to medications. Newgard et al.19 showed that undertriaged patients were likely to be older, have less physiologic derangement, and have a higher rate of surgical intervention. In those 55 years or older, a higher PPV was found for nearly all markers of moderate and severe injury when compared with the general population. Intrusion and death in the vehicle are more sensitive markers for injury for the older population than the general public. Identification of occult severe injuries in older patients is critical for the appropriate triage of this population. Assessment of injury mechanisms improves EMS provider ability to appropriately triage to a trauma centers. Previous studies have varied in their support for death in the vehicle as a marker for significant injury.20Y22 Death within the same occupant compartment was found to be a good predictor of severe injury within the current guidelines (PPV, 21.4%; 95% CI, 20.5Y22.4%) and had the highest predictive value (PPV, 39%; 95% CI, 36.1Y42.5%) in the older population. The utility of measuring intrusion has been controversial because there is significant variation in vehicular design and construction leading to improved protection in the transfer of energy to the occupant over time. In addition, vehicular mismatch is associated with severe injury confounding the singular measure of intrusion alone.23 Extrication time greater than 20 minutes was dropped from the 2009 decision scheme because the panel recognized that various factors including weather, vehicle type, darkness, and need for additional equipment all confound accuracy of recorded times.24 We found that patients meeting criteria for entrapment had a PPV for severe injury similar to ejection and intrusion criteria. Entrapment is typically caused by significant vehicular deformation and integrity loss surrounding the occupant, both markers already included in the triage decision scheme. To gain better measurements of cabin intrusion and presumably improved ability to assess risk of significant * 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
Figure 2. NASS CDS crash data including patients who met criteria following triage guidelines for each step. These data are survey data and imputed to account for missing values. A patient included in Step 3 will not have met criteria for Step 1 or 2. However, a patient in Step 1 may have met criteria for Step 2 and/or 3.
injury, future research should include these data in the development of Advanced Automatic Collision Notification technologies. In addition to the factors included in the triage guidelines, we examined additional crash characteristics that may be predictive of severe occult injury. We found that complete steering wheel collapse had the highest predictive value for severe injury (25.7%; 95% CI, 23Y28.4%) in patients who met Step 3 criteria alone and should be given consideration for future revisions of the triage guidelines. Roof crush alone was
comparable in its PPVs for injury to intrusion elsewhere in the vehicle. This supports its inclusion into the most current revision guidelines.5 The National Expert Panel of Field Triage is responsible for revising the decision scheme. This working subgroup used an ISS greater than 15, prompt operative care, intensive care unit admission, and case fatality rates in determining the appropriate injury mechanisms for inclusion. Criterion for the discussion of inclusion of mechanism of injury was a 20%
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
759
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
TABLE 2. Predictive Values, Sensitivity, and Specificity of Triage Criteria in Predicting Moderate or Severe Injury, Entire Population PPV, % (95% CI) Predictive values for ISS 9 8, entire population Step 1 26.1 (25.4Y26.8) Step 2 criteria (not Step 1) 92.9 (91.2Y94.5) Step 3 criteria (not Step 1 or Step 2) 22.0 (21.2Y22.9) Predictive values for ISS 9 15, entire population Step 1 20.8 (20.1Y21.4) Step 2 criteria (not Step 1) 48.5 (46.8Y50.1) Step 3 criteria (not Step 1 or Step 2) 9.7 (9.3Y10.2) Predictive values for ISS 9 8, for those who met only Step 3 criteria Intrusion 12 inches, occupant site 23.8 (22Y25.5) Intrusion 18 inches, anywhere 18.7 (17.2Y20.2) Death in vehicle 34.7 (33.6Y35.8) Steering wheel collapse 45.2 (41.9Y48.5) Roof crush 12 inches, anywhere 22.0 (20.3Y23.7) Roof crush 18 inches, anywhere 19 (16.3Y21.7) Ejection 23.1 (21.7Y24.4) Entrapment 22.3 (20.8Y23.8) Predictive values for ISS 915, for those who met only Step 3 criteria Intrusion 12 inches, occupant site 10.4 (9.5Y11.3) Intrusion 18 inches, anywhere 7.6 (7.0Y8.3) Death in vehicle 21.4 (20.5Y22.4) Steering wheel collapse 25.7 (23.0Y28.4) Roof crush 12 inches, occupant site 8.2 (7.4Y8.9) Roof crush 18 inches, anywhere 6.5 (5.7Y7.3) Ejection 9.8 (9.1Y10.5) Entrapment 10.3 (9.4Y11.2)
NPV, % (95% CI)
Sensitivity, % (95% CI)
Specificity, % (95% CI)
96.7 (96.8Y97.0) 97.1 (97.0Y97.1) 97.8 (97.8Y97.9)
24.0 (23.6Y24.5) 12.1 (11.7Y12.5) 27.9 (27.1Y28.8)
97.0 (96.9Y97.1) 99.9 (99.9Y99.9) 97.03 (96.9Y97.1)
98.8 (98.8Y98.9) 99 (99.0Y99.1) 99.4 (99.4Y99.4)
41.7 (40.9Y42.5) 18 (17.3Y18.6) 37.6 (36.4Y38.8)
96.9 (96.8Y97.0) 99.8 (99.8Y99.8) 96.6 (96.5Y96.7)
79.1 (78.2Y80.0) 76.3 (75.3Y77.3) 79.6 (78.6Y80.5) 78.2 (77.4Y79.1) 78.0 (77.0Y78.9) 77.8 (76.9Y78.7) 78.7 (77.6Y79.8) 78.1 (77.1Y79.1)
41.6 (40.1Y43.0) 28.3 (27.1Y29.6) 17.3 (16.5Y18.0) 1.99 (1.8Y2.2) 17.2 (15.9Y18.6) 6 (5.1Y6.9) 41.0 (39.5Y42.6) 26.4 (25.3Y27.6)
62.4 (60.2Y64.5) 65.2 (63.1Y67.3) 90.8 (90.4Y91.2) 99.3 (99.2Y99.4) 82.7 (81.8Y83.7) 92.8 (92.2Y93.3) 61.4 (59.6Y63.2) 74.1 (72.4Y75.7)
90.7 (90.2Y91.1) 89.2 (88.6Y89.8) 91.7 (91.2Y92.2) 90.4 (90.0Y90.9) 89.9 (89.4Y90.5) 90 (89.5Y90.5) 90.3 (89.7Y90.9) 90.5 (89.9Y91.0)
41.1 (39.2Y43.1) 26.2 (24.9Y27.5) 21.1 (22.8Y25.4) 2.6 (2.3Y2.9) 14.5 (13.3Y15.6) 4.6 (4.1Y5.1) 39.2 (37.3Y41.1) 27.6 (25.9Y29.3)
61.8 (59.9Y63.7) 65.8 (64.0Y67.7) 90.5 (90.1Y90.8) 99.2 (99.1Y99.3) 82.4 (81.5Y83.3) 92.8 (92.3Y93.3) 60.9 (59.3Y62.5) 74.1 (72.7Y74.1)
PPV to predict severe injury.24 From our study, death within the vehicle and steering wheel collapse had a PPV of greater than 20% for an ISS greater than 15 meeting Step 3 criteria alone. NASS/CDS had no available inpatient data. It would be
beneficial in future studies to have other markers for severe injury including the need for urgent operative intervention, transfusion, specialty care, or intensive care admission in future prospective studies. A limitation of this study, despite the
TABLE 3. Predictive Values, Sensitivity, and Specificity of Triage Criteria in Predicting Moderate or Severe Injury, Occupants 55 Years and Older PPV, % (95% CI) Predictive values for ISS 9 8, for those that met only step 3 criteria Intrusion 12 inches, occupant site 53.02 (48.7Y57.4) Intrusion 18 inches, anywhere 32.7 (27.6Y37.9) Death in vehicle 56.9 (53.7Y60.1) Steering wheel collapse 54.2 (45.3Y63.0) Roof crush 12 inches, anywhere 35.4 (28.2Y42.5) Roof crush 18 inches, anywhere 60.4 (50.0Y70.8) Ejection 23.5 (18.9Y28.0) Entrapment 22.7 (18.3Y27.1) Predictive values for ISS 915, for those who met only Step 3 criteria Intrusion 12 inches, occupant site 28.8 (24.3Y33.3) Intrusion 18 inches, anywhere 9.3 (8.1Y10.5) Death in vehicle 39.3 (36.1Y42.5) Steering wheel collapse 36.7 (28.5Y44.9) Roof crush 12 inches, occupant site 6.2 (4.9Y7.4) Roof crush 18 inches, anywhere 5.5 (3.3Y7.7) Ejection 16.7 (12.6Y20.5) Entrapment 9.3 (7.4Y11.2)
760
NPV, % (95% CI)
Sensitivity, % (95% CI)
Specificity, % (95% CI)
70.93 (67.8Y74.1) 59.11 (55.2Y63.0) 65.3 (61.7Y68.8) 61.5 (58.4Y64.5) 60.6 (57.2Y64.0) 63.3 (60.3Y66.3) 53.9 (50.8Y56.9) 55.9 (52.6Y59.3)
54.7 (50.6Y58.8) 23.81 (19.5Y28.1) 25.6 (22.9Y28.3) 0.76 (0.58Y0.93) 18.1 (13.7Y22.5) 12.5 (8.0Y17.0) 20.2 (17.3Y23.0) 15.1 (13.1Y17.0)
69.54 (66.1Y73.0) 69.3 (65.9Y72.6) 87.8 (86.3Y89.4) 99.6 (0.93) 79.2 (76.3Y82.1) 94.9 (93.9Y95.8) 58.7 (53.4Y64.0) 67.7 (83.0Y86.9)
79.2 (76.7Y81.8) 70.3 (66.9Y73.7) 79.3 (76.4Y82.1) 76.1 (73.6Y78.6) 71.7 (68.6Y74.7) 74.4 (71.8Y77.1) 72.4 (69.6Y75.3) 71.0 (68.0Y74.0)
47.8 (42.4Y53.3) 10.9 (9.3Y12.4) 28.5 (24.9Y32.1) 0.8 (0.59Y1.1) 5.1 (4.1Y6.0) 1.8 (1.2Y2.4) 23.13 (19.1Y27.2) 9.9 (8.5Y11.4)
62.7 (59.2Y66.3) 66.5 (63.2Y69.9) 86.1 (84.6Y87.6) 99.6 (99.4Y99.7) 75.6 (72.5Y78.8) 90.1 (97.6Y98.8) 63.7 (59.0Y68.4) 69.4 (65.0Y73.8)
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
J Trauma Acute Care Surg Volume 76, Number 3
Davidson et al.
fact that the use of trauma scores is a well-established method of validating triage,21,25Y27 is the use of ISS alone as a marker for significant injury. Intrusion alone predicts severe injury in patients not meeting physiologic or anatomic criteria moderately well (10.4% and 7.6% for 12 inches and 18 inches of intrusion, respectively). While there was poor sensitivity for steering wheel collapse, steering wheel collapse has a PPV of 25.7% for all ages and 36.7% for those 54 years or older. Further distribution and implementation of Advanced Automatic Collision Notification technology may need to assess measurements of interior vehicular deformation that would provide better feedback to dispatch appropriate resources and should be currently used in on-scene assessments. Consideration of transport to a trauma center should be given to those found to have steering wheel collapse because this was associated with a high risk of significant injury. AUTHORSHIP G.H.D. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, drafting of the manuscript, and critical revisions. F.P.R. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, drafting of the manuscript, and critical revisions. C.D.M. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, and drafting of the manuscript. R.K. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, and drafting of the manuscript. G.J.J. contributed in the study conception and design, analysis and interpretation of data, and critical revisions. E.M.B. contributed in the study conception and design, acquisition of data, analysis and interpretation of data, and drafting of the manuscript, critical revisions.
ACKNOWLEDGMENT We thank Jin Wang, MS, from the Harborview Injury Prevention and Research Center for her statistics and programming assistance. She did not receive financial compensation for her work.
DISCLOSURE This work was supported by National Institute of Child Health and Human Development grant 1T32HD057822-01AZ that was awarded to the Pediatric Injury Research Training Program and performed for the Crash Injury Research and Engineering network (CIREN) project at the University of Washington/Harborview Injury Prevention and Research Center in cooperation with the US Department of Transportation/National Highway Traffic Safety Administration (USDOT/NHTSA). Funding has been provided by NHTSA under Cooperative Agreement Number DTNH22-05H-11001 and DTNH22-10-H-00295.
REFERENCES 1. Mackersie RC. History of trauma field triage development and the American College of Surgeons criteria. Prehosp Emerg Care. 2006;10(3):287Y294. 2. Field categorization of trauma patients. Bull Am Coll Surg. 1986;71:17Y21. 3. Hospital and prehospital resources for optimal care of the injured patient. Committee on Trauma of the American College of Surgeons. Bull Am Coll Surg. 1986;71(10):4Y23. 4. Sasser SM, Hunt RC, Sullivent EE, et al. Guidelines for field triage of injured patients. Recommendations of the National Expert Panel on Field Triage. MMWR Recomm Rep. 2009;58(RR-1):1Y35.
5. Centers for Disease Control and Prevention (CDC). Guidelines for Field Triage of Injured Patients Recommendations of the National Expert Panel on Field Triage. MMWR Recomm Rep. 2012;61:1Y23. 6. Mackenzie EJ, Rivara FP, Jurkovich GJ, et al. A national evaluation of the effect of trauma center care on mortality. N Engl J Med. 2006;354:366Y378. 7. Baker SP, O’Neill B, Haddon W Jr, Long WB. The Injury Severity Score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974;14(3):187Y196. 8. Champion HR, Copes WS, Sacco WJ, et al. The Major Trauma Outcome Study: establishing national norms for trauma care. J Trauma. 1990; 30(11):1356Y1365. 9. Sampalis JS, Denis R, Lavoie A, et al. Trauma care regionalization: a process-outcome evaluation. J Trauma. 1999;46(4):565Y579; discussion 579Y581. 10. 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(2):288Y295; discussion 295Y286. 11. Center for Disease Control and Prevention. MMWR Recomm Rep. 2008; 57:1Y34. 12. Center for Disease Control and Prevention. Recommendations From the Expert Panel: Advanced Automatic Collision Notification and Triage of the Injured Patient. Atlanta, GA: US Department of Health and Human Services; 2008. 13. NASSN. 2011. Available at: www-nr.nhtsa.dot.gove/department/nrd-30/ ncsa/NASS.html. Accessed March 2012. 14. AAAM Abbreviated Injury Scale, 2005 Revision. Barrington, IL: Association for the Advancement of Automotive Medicine; 2005. 15. Wuerz RTJ, Smith JS. Accuracy of trauma triage in patients transported by helicopter. Air Med J. 1996;15:168Y170. 16. van Buuren S, Boshuizen HC, Knook DL. Multiple imputation of missing blood pressure covariates in survival analysis. Stat Med. 1999;18(6):681Y694. 17. Royston P. Multiple imputation of missing value: update. Stata J. 2005; 5(2):181Y201. 18. Henry MC. Trauma triage: New York experience. Prehosp Emerg Care. 2006;10(3):295Y302. 19. Newgard CD, Zive D, Holmes JF, et al. A multisite assessment of the American College of Surgeons Committee on Trauma field triage decision scheme for identifying seriously injured children and adults. J Am Coll Surg. 2011;213(6):709Y721. 20. Henry MC, Hollander JE, Alicandro JM, Cassara G, O’Malley S, Thode HC Jr. Incremental benefit of individual American College of Surgeons trauma triage criteria. Acad Emerg Med. 1996;3(11):992Y1000. 21. Knopp R, Yanagi A, Kallsen G, Geide A, Doehring L. Mechanism of injury and anatomic injury as criteria for prehospital trauma triage. Ann Emerg Med. 1988;17(9):895Y902. 22. Palanca S, Taylor DM, Bailey M, Cameron PA. Mechanisms of motor vehicle accidents that predict major injury. Emerg Med (Fremantle). 2003; 15(5Y6):423Y428. 23. Mandell SP, Mack CD, Bulger EM. Motor vehicle mismatch: a national perspective. Inj Prev. 2010;16(5):309Y314. 24. Wang SW SS, Jurkovich GJ. Universal Medical Rescue Protocol Changed: ‘‘High Speed Auto Crash’’ Changed to ‘‘High Risk Auto Crash’’ in the Field Triage Decision Scheme. Paper Number 09-0207. Publisher National Highway Traffic Safety Administration: Available at http://www-nrd.nhtsa.dot.gov/ pdf/esv/esv21/09-0207.pdf. 25. Norcross ED, Ford DW, Cooper ME, Zone-Smith L, Byrne TK, Yarbrough DR 3rd. Application of American College of Surgeons’ field triage guidelines by pre-hospital personnel. J Am Coll Surg. 1995;181(6):539Y544. 26. Cooper ME, Yarbrough DR, Zone-Smith L, Byrne TK, Norcross ED. Application of field triage guidelines by pre-hospital personnel: is mechanism of injury a valid guideline for patient triage? Am Surg. 1995;61(4):363Y367. 27. Kreis DJ Jr, Fine EG, Gomez GA, Eckes J, Whitwell E, Byers PM. A prospective evaluation of field categorization of trauma patients. J Trauma. 1988;28(7):995Y1000.
* 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
761
