- Email: [email protected]
Measuring the Impact of Whole-Body Computed Tomography on Hospital Length of Stay in Blunt Trauma
Original Investigation
Measuring the Impact of Whole-Body Computed Tomography on Hospital Length of Stay in Blunt Trauma Jessica Chan, MD, Christopher Johnson, MPH, Gillian Beran-Maryott, BA, Janet Cortez, RN, MS, Thomas H. Greene, PhD, Raminder Nirula, MD, MPH, Marta Heilbrun, MD Rationale and Objectives: Whole-body computed tomography (WBCT) imaging has become commonplace in some emergency departments (EDs) for trauma where management is dependent on rapid diagnosis achieved through comprehensive imaging. The purpose of this study was to assess the value that computed tomography (CT) imaging contributes to trauma patients by retrospectively comparing hospital length of stay (LOS) between WBCT and selective CT imaging, while controlling for hemodynamic stability and socioeconomic considerations. Materials and Methods: This study was institutional review board approved. The institutional trauma registry database was crossreferenced with our radiology information system database to identify adult patients who sustained blunt trauma between July 2011 and June 2013 and received CT imaging. Propensity score weighting was utilized to achieve balance in baseline covariates, including demographics, hemodynamic stability, Glasgow Coma Scale, and socioeconomic factors. A generalized linear model was used to compare LOS between imaging types, and a multinomial logistic regression was utilized to analyze differences in discharge disposition. Results: A total of 2291 patients were identified of which 14.5% underwent WBCT imaging. WBCT patients had an insignificantly longer inpatient hospital LOS of 0.31 days (P = 0.54), and insignificantly higher odds of being discharged to a nursing home facility (versus home, odds ratio = 1.29 [P = 0.34]) when compared to those who received selective CT. Conclusion: WBCT imaging did not have a statistically significant effect on inpatient hospital LOS or discharge disposition. Key Words: Blunt trauma; whole-body CT imaging; hospital length of stay. © 2016 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.
INTRODUCTION
I
n the emergency department (ED), computed tomography (CT) utilization has increased threefold between 1996 and 2007, with a quarter of all CTs performed in the United States occurring in the ED (1,2). Determining the optimal use and justification for emergent testing is essential to containing the costs of medical care. A critical area of focus for imaging utilization in the ED is in the trauma setting. The early diagnostic evaluation of patients with severe trauma has become increasingly dependent on rapid and comprehensive imaging, with an emphasis on CT over the past two decades (3). CT distinguishes patients with injuries requiring intervention from patients without critical injuries, so that Acad Radiol 2016; 23:582–587 From the Department of Radiology, University of Utah, 30 North 1900 East #1A071, Salt Lake City, Utah 84132 (J.C., G.B.-M., M.H.); Division of Epidemiology (C.J., T.H.G.); Department of General Surgery, University of Utah, Salt Lake City, Utah (J.C., R.N.). Received October 8, 2015; revised January 6, 2016; accepted January 10, 2016. Address correspondence to: J.C. e-mail: [email protected] © 2016 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.acra.2016.01.013
582
the trauma team can safely focus on the acute care issues (4). With rapid data acquisition and improved image quality, wholebody computed tomography imaging (WBCT) (defined as head, C-spine, chest, abdomen, and pelvis) is being used with increasing frequency in the trauma setting (3). However, the use of WBCT imaging in patients with blunt trauma in the ED remains controversial due to (1) the proliferation of potentially unnecessary imaging, (2) the associated risk of radiation exposure, (3) the added cost of the additional imaging, (4) the added time in the scanner, and (5) the expense of further workup of incidental findings (5–7). The literature describing WBCT imaging is mostly from Europe and has found reduced mortality in severe blunt trauma patients who underwent WBCT imaging (3,4,7–9,14,15). However, the generalizability of these results to the United States remains to be determined due to the differences in our health-care systems. The purpose of our study was to examine the impact of trauma imaging strategies on patient outcomes in a US level 1 trauma center. Primary end points included hospital length of stay (LOS) and discharge disposition, while secondary endpoints included intensive care unit (ICU) LOS and mortality rate.
Academic Radiology, Vol 23, No 5, May 2016
IMPACT OF WHOLE-BODY COMPUTED TOMOGRAPHY
METHODS AND MATERIALS
sity score weights based on the average treatment effect for the treated (ATT) weighting scheme (those receiving WBCT were assigned a weight of 1, and those receiving selective CT were assigned a weight of (propensity score/(1 − propensity score)). ATT estimates the average treatment effect for those who actually received the treatment, in this study the WBCT. We observed propensity score overlap between imaging groups and chose to exclude subjects with propensity scores less than 0.7, as shown in Figure 1. Using all imputed data, we generated standardized differences between imaging groups both before propensity score weighting and after propensity score weighting to assess the effectiveness of our propensity score model. For our outcome of hospital discharge disposition, we performed a multinomial logistic regression on each imputation using the propensity score sample weights to assess the effect of imaging on discharge disposition (nursing home, home, or other). A P < 0.05 was used to determine statistical significance. Analyses were performed using SAS (version 9.4; SAS Institute, Cary, NC) or Stata (Stata Statistical Software, Release 13; StataCorp LP, College Station, TX). The present study was reviewed and approved by the institutional review board.
This retrospective chart review cohort study was performed to evaluate the impact of WBCT imaging utilized in the first 24 hours after a blunt trauma presentation on inpatient hospital LOS and discharge disposition primarily, and intensive care unit (ICU) LOS and mortality rate secondarily. We are an academic Level I trauma center with 33,000 patients annually with patients from a mixture of urban and rural settings (including multiple surrounding states) with highly variable access to health-care resources. The Institutional Trauma Registry Database was accessed to identify patients ≥18 years old who sustained blunt trauma between July 2011 and June 2013. Thils institutional registry is designed to provide data to the American College of Surgeons National Trauma Data Bank (https://www.facs.org/quality%20programs/trauma/ntdb). Patients who experienced penetrating trauma, were pregnant, underwent a surgical intervention before imaging, and died before the implementation of an advanced life-sustaining treatment were excluded. Demographics, injury severity score (ISS), abbreviated injury scale (AIS), inpatient hospital LOS, ICU LOS, Emergency department (ED) LOS, mortality data, disposition location, and whether the patient was transferred from an outside institution were collected from the Institutional Trauma Registry Database (12). The list of patient encounters was cross-matched with the radiology information system (RIS) database, thereby identifying imaging records including CT imaging acquired both in-house and at any outside hospital before transfer. WBCT imaging was defined as CT imaging (with or without contrast) of at least the head, C-spine, chest, abdomen, and pelvis. Selective CT imaging was defined as any CT imaging that did not include all of these body areas. A propensity score model (propensity to receive WBCT versus selective CT, given all the variables shown in Table 1) was utilized to achieve balance in baseline covariates (Tables 1 and 2) including demographics, hemodynamic stability, and ED Glasgow Coma Scale (GCS) (13). There were missing values for the following variables, due to variations in clinical practice: white blood cell (WBC) count, red blood cell (RBC) count, platelet count, international normalized ratio (INR), partial thromboplastin time (PTT), home distance from the university, glucose, potassium, anion gap, calcium, sodium, chloride, carbon dioxide, ED pulse, ED systolic blood pressure, creatinine, and blood urea nitrogen (BUN). Accordingly, a multiple imputation procedure was utilized, and transformations to normality on missing variables were necessary. Multiple imputation was carried out using PROC MI in SAS (version 9.4, SAS Institute, Cary, NC) using five imputations. All transformed variables were then back-transformed, and all variables were rounded to match original variable formats. To create propensity scores, we performed a logistic regression analysis by each imputation, modeling the probability to receive WBCT versus selective CT and using our chosen predictors of interest (Table 1). We then generated propen-
RESULTS Overall, 2,291 trauma registry patients met the inclusion criteria. The fully imputed dataset consisted of 11,328 observations, after removing 127 subjects with propensity scores less than 0.7. 1,958 (85.5%) subjects received selective CT imaging and 333 (14.5%) received WBCT imaging. Those patients who received WBCT were younger (44.2 versus 49.0 years old), less alert (Glasgow Coma Scale [GCS] of 12.2 versus 13.9), more likely have been involved in a motor vehicle accident (72.8% versus 49.9%), tachycardiac (pulse 94.9 bpm versus 88.7 bpm), and were more likely to need resuscitative products (5.3% versus 1.6% of patients) than selective CT imaging patients (Table 1). Variables with the greatest absolute standardized difference include “received lactate test,” motor vehicle accident, calcium level, GCS, and minutes in the ED, in descending order (Table 1). All baseline variables had standardized differences after propensity score weighting less than 0.1 (Table 2), indicating that propensity score weighting worked properly at achieving balance between imaging groups. Figure 1 shows sufficient overlap between the distributions of propensity scores for the patients receiving selective CT imaging and for the patients receiving WBCT imaging, to suggest that we can be reasonably certain in the precision of our effect estimates. For our primary analysis, we assessed the effect of imaging on inpatient hospital LOS while applying sample weights based on our propensity scores. Overall, the average LOS for subjects receiving whole-body imaging was 0.31 days longer when compared to subjects receiving selective imaging (95% 583
CHAN ET AL
Academic Radiology, Vol 23, No 5, May 2016
TABLE 1. Prepropensity Weighting Balance (N = 11,318)
Variables Age, mean (SD) Home Distance to the University, mean (SD) ED Glasgow Coma Scale, mean (SD) ED Pulse, mean (SD) ED Systolic Blood Pressure, mean (SD) Glucose, mean (SD) Potassium, mean (SD) Anion Gap, mean (SD) Calcium, mean (SD) Sodium, mean (SD) BUN, mean (SD) Chloride, mean (SD) Creatinine, mean (SD) Carbon Dioxide, mean (SD) WBC Count, mean (SD) Red Blood Cell Count, mean (SD) Platelet Count, mean (SD) PTT, mean (SD) INR, mean (SD) Received FFP on Admission, % Received PRBCs on Admission, % Received Lactate Test, % Female, % Hospital Transfer, % Type of Insurance Private, % Public, % Not Funded, % Other, % Mechanism of Injury Fall, % Vehicle, % Assault, % Other, %
Selective Imaging (n = 9,774)
Whole-Body Imaging (n = 1,544)
Absolute Standardized Difference
49.0 (21.3) 262.2 (697.7) 13.9 (3.0) 88.7 (18.5) 132.6 (22.0) 122.2 (47.7) 3.9 (0.5) 10.5 (3.6) 8.9 (0.9) 139.1 (3.8) 16.3 (9.3) 106.1 (4.8) 1.0 (0.6) 22.5 (3.9) 11.7 (5.0) 4.6 (0.7) 257.2 (80.2) 28.7 (6.4) 1.2 (0.5) 2.5 1.6 16.6 33.4 41.4
44.2 (18.7) 241.0 (502.4) 12.2 (4.6) 94.9 (23.2) 126.9 (23.2) 130.1 (46.9) 3.9 (0.6) 10.3 (3.9) 8.4 (1.0) 140.0 (3.7) 15.1 (7.7) 108.2 (5.1) 0.9 (0.3) 21.5 (4.2) 13.5 (6.6) 4.5 (0.8) 253.5 (77.6) 29.2 (10.0) 1.2 (0.5) 4.0 5.3 44.6 26.4 34.8
0.237 0.035 0.440 0.299 0.255 0.168 0.054 0.060 0.481 0.228 0.146 0.415 0.048 0.242 0.298 0.195 0.047 0.058 0.108 0.088 0.205 0.638 0.154 0.136
38.8 32.8 24.5 3.9
35.4 26.6 32.7 5.3
0.070 0.137 0.183 0.068
38.1 49.9 7.1 4.9
21.7 72.8 1.9 3.6
0.364 0.483 0.249 0.067
BUN, blood urea nitrogen; ED, emergency department; FFP, fresh frozen plasma; INR, international normalized ratio; PRBC, packed red blood cell; PTT, partial thromboplastin time; RBC, red blood cell; SD, standard deviation; WBC, white blood cell.
confidence interval [CI] −0.67 to 1.28) with P = 0.5377. Prior to imputation or propensity score weighting, the absolute inpatient hospital LOS was 3.8+/− 9.4 days for selective CT and WBCT imaging cohorts, respectively. We also examined the effect of imaging on discharge disposition (nursing home, home, other). Using home discharge as our reference category, we compared discharge to nursing home versus discharge to home, and discharge to another location versus discharge to home. Although WBCT subjects were more likely to be discharged to a nursing home versus home (odds ratio [OR] = 1.29, 95% CI 0.76–2.20), this effect was not statistically significant (P = 0.34). We additionally found that patients receiving WBCT imaging were less likely to be discharged to any another location versus home (OR = 0.88, 95% CI 0.62–1.25) when compared to patients receiving selective CT imaging, although this was a significant effect (P = 0.47). 584
Regarding our secondary outcomes, of the entire cohort, only 79 patients (3.4%) died within 30 days precluding a meaningful 30-day mortality analysis. Likewise, the distribution of the ICU LOS was significantly skewed, causing the analysis to be insufficiently powered due to the dominance of a small number of outliers. Observationally, sixty-seven percent of patients who received WBCT imaging were admitted to the ICU, while 41.3% of selective CT imaging patients were admitted to the ICU. Of those patients admitted to the ICU, those who received WBCT imaging spent an average ICU LOS of 5.8 days, whereas selective CT imaging patients had an average ICU LOS of 3.3 days. DISCUSSION We found no statistically significant difference in inpatient hospital LOS and discharge location in blunt trauma patients who
Academic Radiology, Vol 23, No 5, May 2016
Figure 1.
IMPACT OF WHOLE-BODY COMPUTED TOMOGRAPHY
Overlap in propensity scores between selective scan and whole-body scan imaging types.
underwent WBCT imaging versus selective CT imaging within the first 24 hours of presentation to the hospital. Although mortality and ICU LOS data were collected, they were insufficiently powered for meaningful analysis. There are important differences between our study and the European retrospective studies that populate the literature. At our institution, only 14.5% of blunt trauma patients received WBCT imaging, a much lower utilization rate than that found in the largest comparable European study, with a 32% utilization rate (3,4). However, the decision to obtain WBCT imaging is at the discretion of the treating team in our emergency rooms, while it is often part of an algorithm, if not standardized in many European emergency rooms. Our mortality rates were too low to yield a meaningful analysis. Multiple European studies have found a 30-day mortality benefit in WBCT imaging patients (3,4,8). These studies have focused on severe blunt trauma (requiring ICU admission), whereas our dataset includes all severity levels of blunt trauma. Furthermore, our average inpatient hospital LOS was approximately 4 days, whereas a representative European study had an average inpatient hospital LOS of 30 days, which could be due to the higher severity patients in their study, but more likely due to differences in how the medical systems are structured and what is considered a “inpatient hospital” stay (9). Despite the differences between our study and the European retrospective studies, the LOS outcomes are largely the same. A recent meta-analysis of the European retrospective studies also found no effect of WBCT on hospital LOS, (14). Likewise preliminary data presented at RSNA 2015 of the prospective randomized controlled trial (REACT-2 trial) evaluating the effect of WBCT imaging on blunt trauma patients,
found that patient outcomes were no different for those who underwent WBCT (15). Many prior retrospective studies used a marker for injury severity called the ISS or trauma and injury severity score to control for injury severity between patient populations (10–12). Imaging identifies injuries that contribute to calculating the ISS; thus, the ISS is confounded by the amount of imaging and is not an independent variable of injury severity. In other words, more injuries can be found in patients with more imaging. For example, a patient may have a chest CT performed, finding a nondisplaced rib fracture, which would not have been found otherwise, consequently increasing the overall ISS. The Yeguiayan, et al (8) study is the most similar to ours, a prospective multi-institutional observational study performed in France. The authors describe the results of the FIRST (French Intensive care Recorded in Severe Trauma) study. The authors were interested in examining the impact of emergency care on hospital mortality of patients with severe blunt trauma. Acknowledging that the ISS and TRISS scores are created using information generated by the imaging, the authors argue that an adjustment for ISS, which is used clinically to estimate the risk of death, could overestimate the benefit of the WBCT. These authors used initial physiologic variables, GCS, treatments like transfusions and mechanical ventilation in their propensity model. We found similar differences in the population that underwent WBCT compared to the selective CT, in that the population was younger, had had more severe mechanisms of injury, and worse physiologic measures. Unlike that study, our patients also had a lower GCS and required more resuscitation. This would suggest that the 585
CHAN ET AL
Academic Radiology, Vol 23, No 5, May 2016
TABLE 2. Postpropensity Weighting Balance (N = 11,318)
Age, mean (SD) Home Distance to the University, mean (SD) ED Glasgow Coma Scale, mean (SD) ED Pulse, mean (SD) ED Systolic Blood Pressure, mean (SD) Glucose, mean (SD) Potassium, mean (SD) Anion Gap, mean (SD) Calcium, mean (SD) Sodium, mean (SD) BUN, mean (SD) Chloride, mean (SD) Creatinine, mean (SD) Carbon Dioxide, mean (SD) WBC Count, mean (SD) Red Blood Cell Count, mean (SD) Platelet Count, mean (SD) PTT, mean (SD) INR, mean (SD) Received FFP on Admission, % Received PRBCs on Admission, % Received Lactate Test, % Female, % Hospital Transfer, % Type of Insurance Private, % Public, % Not Funded, % Other, % Mechanism of Injury Fall, % Vehicle, % Assault, % Other, %
Selective Imaging (n = 9,774)
Whole-Body Imaging (n = 1,544)
Absolute Standardized Difference
43.8 (7.7) 248.0 (223.8) 12.1 (1.9) 95.6 (9.2) 126.3 (9.3) 130.1 (19.8) 3.9 (0.2) 10.2 (1.6) 8.4 (0.5) 140.0 (1.5) 14.9 (2.9) 108.4 (2.3) 0.9 (0.2) 21.3 (1.7) 13.5 (2.4) 4.5 (0.3) 252.3 (31.5) 29.1 (3.6) 1.2 (0.2) 4.9 4.7 45.1 26.4 36.6
44.2 (18.7) 241.0 (502.4) 12.2 (4.6) 94.9 (23.2) 126.9 (23.2) 130.1 (46.9) 3.9 (0.6) 10.3 (3.9) 8.4 (1.0) 140.0 (3.7) 15.1 (7.7) 108.2 (5.1) 0.9 (0.3) 21.5 (4.2) 13.5 (6.6) 4.5 (0.8) 253.5 (77.6) 29.2 (10.0) 1.2 (0.5) 4.0 5.3 44.6 26.4 34.8
0.026 0.018 0.017 0.040 0.032 <0.001 0.009 0.045 0.094 0.007 0.024 0.072 0.010 0.040 0.004 0.032 0.019 0.003 0.028 0.045 0.028 0.010 <0.001 0.038
34.4 26.8 34.2 4.6
35.4 26.6 32.7 5.3
0.022 0.005 0.032 0.031
21.5 72.9 1.8 3.8
21.7 72.8 1.9 3.6
0.005 0.002 0.008 0.011
BUN, blood urea nitrogen; ED, emergency department; FFP, fresh frozen plasma; INR, international normalized ratio; PRBC, packed red blood cell; PTT, partial thromboplastin time; RBC, red blood cell; SD, standard deviation; WBC, white blood cell.
decision to use WBCT, rather than selective CT, is based on a clinical picture that predicts more severe injuries, both in our institution, and in other settings where the decision is not proscribed. An important difference in the studies, however, is that ours included all levels of trauma, and was not limited to severe trauma, possibly accounting for differences in the significance of the predictor variables. The benefit of WBCT may be different, compared to the risk of overutilization, radiation exposure, cost and detection of incidental findings, when it is unlikely that mortal injuries are present. Our study has several limitations. In addition to being a retrospective study, a significant limitation is due to an indication bias. As described above, the trauma team uses clinical and physiologic information to make the decision for imaging intensity. The trauma team is more likely to perform more imaging on more severely injured patients. These patients are 586
more likely to have a longer inpatient hospital stay and to be discharged to a care facility rather than home. Controlling for surrogate markers of injury severity, particularly vital signs, GCS, and resuscitation volume, the propensity score is intended to mitigate the effect of indication bias, without being confounded by the imaging itself. It is possible that our model may fail to account for the critical confounding variables and suffer from a bias due to these omitted variables. We chose a propensity score to identify and normalize possible confounding variables between the patient populations including severity of injury (assessed using BP, HR, GCS, CBC, chemistry panel, volume of resuscitation products), and basic demographics (age, gender, insurance type) (Table 1). We draw in patients from a large geographic area, so we also controlled for home distance from the inpatient hospital because we felt that transportation home could affect dispositioin type
Academic Radiology, Vol 23, No 5, May 2016
IMPACT OF WHOLE-BODY COMPUTED TOMOGRAPHY
and length of stay. The significant overlap in the propensity curves of Figure 1 suggest that our analysis successfully normalized the patient populations with regards to the predictor variables. Yeguiavan, et al (8) also addresses this concern and argues, after adding ISS as a covariate in their regression model, that the mortality reduction benefit assigned to the WBCT on the basis of ISS or TRISS stratification is likely overestimated. As with any method of covariate adjustment method, propensity score adjustment only reduces bias, and makes an observational study more like a randomized study, to the extent that all sources of confounding are accounted for in the model. Laboratory and other clinical values have been presumed to serve as surrogates for measures of severity, as has been shown in previous retrospective studies (8) and will be tested in the ongoing prospective REACT-2 trial (15). Other limitations include the variation of practice amongst trauma providers in deciding which patients receive selective CT imaging versus WBCT imaging. However, this model of practice is reflective of the majority of American EDs. There were no statistically or clinically significant differences in inpatient hospital LOS and discharge disposition of the average blunt trauma patient who underwent WBCT imaging versus selective CT imaging, but subset analysis may yield differences in outcomes that guide CT imaging utilization in the ED. A subset analysis of patients with head trauma and those patients who required ICU admission may yield different outcomes.
tional Sciences, National Institutes of Health, through Grant 8UL1TR000105 (formerly UL1RR025764).
CONCLUSIONS We retrospectively analyzed a large population of blunt trauma patients at a single level 1 trauma center and found that there was no statistically or clinically significant difference in inpatient hospital LOS or discharge location in patients who underwent WBCT. ACKNOWLEDGMENTS This investigation was primarily supported by funding from the Harvey L. Neiman Health Policy Institute, and also by the University of Utah Study Design and Biostatistics Center, with funding in part from the National Center for Research Resources and the National Center for Advancing Transla-
REFERENCES 1. Kocher KE, Meurer WJ, Fazel R, et al. National trends in use of computed tomography in the emergency department. Ann Emerg Med 2011; 58:452–62 e3. doi:10.1016/j.annemergmed.2011.05.020. 2. Rao VM, Levin DC, Parker L, et al. Trends in utilization rates of the various imaging modalities in emergency departments: nationwide Medicare data from 2000 to 2008. J Am Coll Radiol 2011; 8:706–709. doi:10.1016/ j.jacr.2011.04.004. 3. van Vugt R, Kool DR, Deunk J, et al. Effects on mortality, treatment, and time management as a result of routine use of total body computed tomography in blunt high-energy trauma patients. J Trauma Acute Care Surg 2012; 72:553–559. doi:10.1097/TA.0b013e31822dd93b. 4. Huber-Wagner S, Lefering R, Qvick LM, et al. Effect of whole-body CT during trauma resuscitation on survival: a retrospective, multicentre study. Lancet 2009; 373:1455–1461. doi:10.1016/S0140-6736(09)60232-4. 5. Asha S, Curtis KA, Grant N, et al. Comparison of radiation exposure of trauma patients from diagnostic radiology procedures before and after the introduction of a panscan protocol. Emerg Med Australas 2012; 24:43– 51. doi:10.1111/j.1742-6723.2011.01504.x. 6. Inaba K, Branco BC, Lim G, et al. The increasing burden of radiation exposure in the management of trauma patients. J Trauma 2011; 70:1366– 1370. doi:10.1097/TA.0b013e3181ebb4d4. 7. Salim A, Sangthong B, Martin M, et al. Whole body imaging in blunt multisystem trauma patients without obvious signs of injury: results of a prospective study. Arch Surg 2006; 141:468–473, discussion 73-5. doi:10.1001/archsurg.141.5.468. 8. Yeguiayan JM, Yap A, Freysz M, et al. Impact of whole-body computed tomography on mortality and surgical management of severe blunt trauma. Crit Care 2012; 16:R101. doi:10.1186/cc11375. 9. Weninger P, Mauritz W, Fridrich P, et al. Emergency room management of patients with blunt major trauma: evaluation of the multislice computed tomography protocol exemplified by an urban trauma center. J Trauma 2007; 62:584–591. doi:10.1097/01.ta.0000221797.46249.ee. 10. Baker SP, O’Neill B, Haddon W, Jr, et al. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1974; 14:187–196. 11. Kahloul M, Bouida W, Boubaker H, et al. Value of anatomic and physiologic scoring systems in outcome prediction of trauma patients. Eur J Emerg Med 2014; 21:125–129. doi:10.1097/MEJ.0b013e32836188ce. 12. Baker SP, O’Neill B, Haddon W, Jr, et al. The injury Severity Score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1974; 14:187–196. 13. Austin P. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res 2011; 46:399–424. 14. Jiang L, Ma Y, Jiang S, et al. Comparison of whole body computed tomography vs selective radiological imaging on outcomes in major trauma patients: a meta-analysis. Scand J Trauma Resusc Emerg Med 2014; 22:54. 15. Caputo ND, Stahmer C, Lim G, et al. Whole-body computed tomographic scanning leads to better survival as opposed to selective scanning in trauma patients: a systematic review and meta-analysis. J Trauma Acute Care Surg 2014; 77:534–539.
587
Measuring the Impact of Whole-Body Computed Tomography on Hospital Length of Stay in Blunt Trauma Jessica Chan, MD, Christopher Johnson, MPH, Gillian Beran-Maryott, BA, Janet Cortez, RN, MS, Thomas H. Greene, PhD, Raminder Nirula, MD, MPH, Marta Heilbrun, MD Rationale and Objectives: Whole-body computed tomography (WBCT) imaging has become commonplace in some emergency departments (EDs) for trauma where management is dependent on rapid diagnosis achieved through comprehensive imaging. The purpose of this study was to assess the value that computed tomography (CT) imaging contributes to trauma patients by retrospectively comparing hospital length of stay (LOS) between WBCT and selective CT imaging, while controlling for hemodynamic stability and socioeconomic considerations. Materials and Methods: This study was institutional review board approved. The institutional trauma registry database was crossreferenced with our radiology information system database to identify adult patients who sustained blunt trauma between July 2011 and June 2013 and received CT imaging. Propensity score weighting was utilized to achieve balance in baseline covariates, including demographics, hemodynamic stability, Glasgow Coma Scale, and socioeconomic factors. A generalized linear model was used to compare LOS between imaging types, and a multinomial logistic regression was utilized to analyze differences in discharge disposition. Results: A total of 2291 patients were identified of which 14.5% underwent WBCT imaging. WBCT patients had an insignificantly longer inpatient hospital LOS of 0.31 days (P = 0.54), and insignificantly higher odds of being discharged to a nursing home facility (versus home, odds ratio = 1.29 [P = 0.34]) when compared to those who received selective CT. Conclusion: WBCT imaging did not have a statistically significant effect on inpatient hospital LOS or discharge disposition. Key Words: Blunt trauma; whole-body CT imaging; hospital length of stay. © 2016 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.
INTRODUCTION
I
n the emergency department (ED), computed tomography (CT) utilization has increased threefold between 1996 and 2007, with a quarter of all CTs performed in the United States occurring in the ED (1,2). Determining the optimal use and justification for emergent testing is essential to containing the costs of medical care. A critical area of focus for imaging utilization in the ED is in the trauma setting. The early diagnostic evaluation of patients with severe trauma has become increasingly dependent on rapid and comprehensive imaging, with an emphasis on CT over the past two decades (3). CT distinguishes patients with injuries requiring intervention from patients without critical injuries, so that Acad Radiol 2016; 23:582–587 From the Department of Radiology, University of Utah, 30 North 1900 East #1A071, Salt Lake City, Utah 84132 (J.C., G.B.-M., M.H.); Division of Epidemiology (C.J., T.H.G.); Department of General Surgery, University of Utah, Salt Lake City, Utah (J.C., R.N.). Received October 8, 2015; revised January 6, 2016; accepted January 10, 2016. Address correspondence to: J.C. e-mail: [email protected] © 2016 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.acra.2016.01.013
582
the trauma team can safely focus on the acute care issues (4). With rapid data acquisition and improved image quality, wholebody computed tomography imaging (WBCT) (defined as head, C-spine, chest, abdomen, and pelvis) is being used with increasing frequency in the trauma setting (3). However, the use of WBCT imaging in patients with blunt trauma in the ED remains controversial due to (1) the proliferation of potentially unnecessary imaging, (2) the associated risk of radiation exposure, (3) the added cost of the additional imaging, (4) the added time in the scanner, and (5) the expense of further workup of incidental findings (5–7). The literature describing WBCT imaging is mostly from Europe and has found reduced mortality in severe blunt trauma patients who underwent WBCT imaging (3,4,7–9,14,15). However, the generalizability of these results to the United States remains to be determined due to the differences in our health-care systems. The purpose of our study was to examine the impact of trauma imaging strategies on patient outcomes in a US level 1 trauma center. Primary end points included hospital length of stay (LOS) and discharge disposition, while secondary endpoints included intensive care unit (ICU) LOS and mortality rate.
Academic Radiology, Vol 23, No 5, May 2016
IMPACT OF WHOLE-BODY COMPUTED TOMOGRAPHY
METHODS AND MATERIALS
sity score weights based on the average treatment effect for the treated (ATT) weighting scheme (those receiving WBCT were assigned a weight of 1, and those receiving selective CT were assigned a weight of (propensity score/(1 − propensity score)). ATT estimates the average treatment effect for those who actually received the treatment, in this study the WBCT. We observed propensity score overlap between imaging groups and chose to exclude subjects with propensity scores less than 0.7, as shown in Figure 1. Using all imputed data, we generated standardized differences between imaging groups both before propensity score weighting and after propensity score weighting to assess the effectiveness of our propensity score model. For our outcome of hospital discharge disposition, we performed a multinomial logistic regression on each imputation using the propensity score sample weights to assess the effect of imaging on discharge disposition (nursing home, home, or other). A P < 0.05 was used to determine statistical significance. Analyses were performed using SAS (version 9.4; SAS Institute, Cary, NC) or Stata (Stata Statistical Software, Release 13; StataCorp LP, College Station, TX). The present study was reviewed and approved by the institutional review board.
This retrospective chart review cohort study was performed to evaluate the impact of WBCT imaging utilized in the first 24 hours after a blunt trauma presentation on inpatient hospital LOS and discharge disposition primarily, and intensive care unit (ICU) LOS and mortality rate secondarily. We are an academic Level I trauma center with 33,000 patients annually with patients from a mixture of urban and rural settings (including multiple surrounding states) with highly variable access to health-care resources. The Institutional Trauma Registry Database was accessed to identify patients ≥18 years old who sustained blunt trauma between July 2011 and June 2013. Thils institutional registry is designed to provide data to the American College of Surgeons National Trauma Data Bank (https://www.facs.org/quality%20programs/trauma/ntdb). Patients who experienced penetrating trauma, were pregnant, underwent a surgical intervention before imaging, and died before the implementation of an advanced life-sustaining treatment were excluded. Demographics, injury severity score (ISS), abbreviated injury scale (AIS), inpatient hospital LOS, ICU LOS, Emergency department (ED) LOS, mortality data, disposition location, and whether the patient was transferred from an outside institution were collected from the Institutional Trauma Registry Database (12). The list of patient encounters was cross-matched with the radiology information system (RIS) database, thereby identifying imaging records including CT imaging acquired both in-house and at any outside hospital before transfer. WBCT imaging was defined as CT imaging (with or without contrast) of at least the head, C-spine, chest, abdomen, and pelvis. Selective CT imaging was defined as any CT imaging that did not include all of these body areas. A propensity score model (propensity to receive WBCT versus selective CT, given all the variables shown in Table 1) was utilized to achieve balance in baseline covariates (Tables 1 and 2) including demographics, hemodynamic stability, and ED Glasgow Coma Scale (GCS) (13). There were missing values for the following variables, due to variations in clinical practice: white blood cell (WBC) count, red blood cell (RBC) count, platelet count, international normalized ratio (INR), partial thromboplastin time (PTT), home distance from the university, glucose, potassium, anion gap, calcium, sodium, chloride, carbon dioxide, ED pulse, ED systolic blood pressure, creatinine, and blood urea nitrogen (BUN). Accordingly, a multiple imputation procedure was utilized, and transformations to normality on missing variables were necessary. Multiple imputation was carried out using PROC MI in SAS (version 9.4, SAS Institute, Cary, NC) using five imputations. All transformed variables were then back-transformed, and all variables were rounded to match original variable formats. To create propensity scores, we performed a logistic regression analysis by each imputation, modeling the probability to receive WBCT versus selective CT and using our chosen predictors of interest (Table 1). We then generated propen-
RESULTS Overall, 2,291 trauma registry patients met the inclusion criteria. The fully imputed dataset consisted of 11,328 observations, after removing 127 subjects with propensity scores less than 0.7. 1,958 (85.5%) subjects received selective CT imaging and 333 (14.5%) received WBCT imaging. Those patients who received WBCT were younger (44.2 versus 49.0 years old), less alert (Glasgow Coma Scale [GCS] of 12.2 versus 13.9), more likely have been involved in a motor vehicle accident (72.8% versus 49.9%), tachycardiac (pulse 94.9 bpm versus 88.7 bpm), and were more likely to need resuscitative products (5.3% versus 1.6% of patients) than selective CT imaging patients (Table 1). Variables with the greatest absolute standardized difference include “received lactate test,” motor vehicle accident, calcium level, GCS, and minutes in the ED, in descending order (Table 1). All baseline variables had standardized differences after propensity score weighting less than 0.1 (Table 2), indicating that propensity score weighting worked properly at achieving balance between imaging groups. Figure 1 shows sufficient overlap between the distributions of propensity scores for the patients receiving selective CT imaging and for the patients receiving WBCT imaging, to suggest that we can be reasonably certain in the precision of our effect estimates. For our primary analysis, we assessed the effect of imaging on inpatient hospital LOS while applying sample weights based on our propensity scores. Overall, the average LOS for subjects receiving whole-body imaging was 0.31 days longer when compared to subjects receiving selective imaging (95% 583
CHAN ET AL
Academic Radiology, Vol 23, No 5, May 2016
TABLE 1. Prepropensity Weighting Balance (N = 11,318)
Variables Age, mean (SD) Home Distance to the University, mean (SD) ED Glasgow Coma Scale, mean (SD) ED Pulse, mean (SD) ED Systolic Blood Pressure, mean (SD) Glucose, mean (SD) Potassium, mean (SD) Anion Gap, mean (SD) Calcium, mean (SD) Sodium, mean (SD) BUN, mean (SD) Chloride, mean (SD) Creatinine, mean (SD) Carbon Dioxide, mean (SD) WBC Count, mean (SD) Red Blood Cell Count, mean (SD) Platelet Count, mean (SD) PTT, mean (SD) INR, mean (SD) Received FFP on Admission, % Received PRBCs on Admission, % Received Lactate Test, % Female, % Hospital Transfer, % Type of Insurance Private, % Public, % Not Funded, % Other, % Mechanism of Injury Fall, % Vehicle, % Assault, % Other, %
Selective Imaging (n = 9,774)
Whole-Body Imaging (n = 1,544)
Absolute Standardized Difference
49.0 (21.3) 262.2 (697.7) 13.9 (3.0) 88.7 (18.5) 132.6 (22.0) 122.2 (47.7) 3.9 (0.5) 10.5 (3.6) 8.9 (0.9) 139.1 (3.8) 16.3 (9.3) 106.1 (4.8) 1.0 (0.6) 22.5 (3.9) 11.7 (5.0) 4.6 (0.7) 257.2 (80.2) 28.7 (6.4) 1.2 (0.5) 2.5 1.6 16.6 33.4 41.4
44.2 (18.7) 241.0 (502.4) 12.2 (4.6) 94.9 (23.2) 126.9 (23.2) 130.1 (46.9) 3.9 (0.6) 10.3 (3.9) 8.4 (1.0) 140.0 (3.7) 15.1 (7.7) 108.2 (5.1) 0.9 (0.3) 21.5 (4.2) 13.5 (6.6) 4.5 (0.8) 253.5 (77.6) 29.2 (10.0) 1.2 (0.5) 4.0 5.3 44.6 26.4 34.8
0.237 0.035 0.440 0.299 0.255 0.168 0.054 0.060 0.481 0.228 0.146 0.415 0.048 0.242 0.298 0.195 0.047 0.058 0.108 0.088 0.205 0.638 0.154 0.136
38.8 32.8 24.5 3.9
35.4 26.6 32.7 5.3
0.070 0.137 0.183 0.068
38.1 49.9 7.1 4.9
21.7 72.8 1.9 3.6
0.364 0.483 0.249 0.067
BUN, blood urea nitrogen; ED, emergency department; FFP, fresh frozen plasma; INR, international normalized ratio; PRBC, packed red blood cell; PTT, partial thromboplastin time; RBC, red blood cell; SD, standard deviation; WBC, white blood cell.
confidence interval [CI] −0.67 to 1.28) with P = 0.5377. Prior to imputation or propensity score weighting, the absolute inpatient hospital LOS was 3.8+/− 9.4 days for selective CT and WBCT imaging cohorts, respectively. We also examined the effect of imaging on discharge disposition (nursing home, home, other). Using home discharge as our reference category, we compared discharge to nursing home versus discharge to home, and discharge to another location versus discharge to home. Although WBCT subjects were more likely to be discharged to a nursing home versus home (odds ratio [OR] = 1.29, 95% CI 0.76–2.20), this effect was not statistically significant (P = 0.34). We additionally found that patients receiving WBCT imaging were less likely to be discharged to any another location versus home (OR = 0.88, 95% CI 0.62–1.25) when compared to patients receiving selective CT imaging, although this was a significant effect (P = 0.47). 584
Regarding our secondary outcomes, of the entire cohort, only 79 patients (3.4%) died within 30 days precluding a meaningful 30-day mortality analysis. Likewise, the distribution of the ICU LOS was significantly skewed, causing the analysis to be insufficiently powered due to the dominance of a small number of outliers. Observationally, sixty-seven percent of patients who received WBCT imaging were admitted to the ICU, while 41.3% of selective CT imaging patients were admitted to the ICU. Of those patients admitted to the ICU, those who received WBCT imaging spent an average ICU LOS of 5.8 days, whereas selective CT imaging patients had an average ICU LOS of 3.3 days. DISCUSSION We found no statistically significant difference in inpatient hospital LOS and discharge location in blunt trauma patients who
Academic Radiology, Vol 23, No 5, May 2016
Figure 1.
IMPACT OF WHOLE-BODY COMPUTED TOMOGRAPHY
Overlap in propensity scores between selective scan and whole-body scan imaging types.
underwent WBCT imaging versus selective CT imaging within the first 24 hours of presentation to the hospital. Although mortality and ICU LOS data were collected, they were insufficiently powered for meaningful analysis. There are important differences between our study and the European retrospective studies that populate the literature. At our institution, only 14.5% of blunt trauma patients received WBCT imaging, a much lower utilization rate than that found in the largest comparable European study, with a 32% utilization rate (3,4). However, the decision to obtain WBCT imaging is at the discretion of the treating team in our emergency rooms, while it is often part of an algorithm, if not standardized in many European emergency rooms. Our mortality rates were too low to yield a meaningful analysis. Multiple European studies have found a 30-day mortality benefit in WBCT imaging patients (3,4,8). These studies have focused on severe blunt trauma (requiring ICU admission), whereas our dataset includes all severity levels of blunt trauma. Furthermore, our average inpatient hospital LOS was approximately 4 days, whereas a representative European study had an average inpatient hospital LOS of 30 days, which could be due to the higher severity patients in their study, but more likely due to differences in how the medical systems are structured and what is considered a “inpatient hospital” stay (9). Despite the differences between our study and the European retrospective studies, the LOS outcomes are largely the same. A recent meta-analysis of the European retrospective studies also found no effect of WBCT on hospital LOS, (14). Likewise preliminary data presented at RSNA 2015 of the prospective randomized controlled trial (REACT-2 trial) evaluating the effect of WBCT imaging on blunt trauma patients,
found that patient outcomes were no different for those who underwent WBCT (15). Many prior retrospective studies used a marker for injury severity called the ISS or trauma and injury severity score to control for injury severity between patient populations (10–12). Imaging identifies injuries that contribute to calculating the ISS; thus, the ISS is confounded by the amount of imaging and is not an independent variable of injury severity. In other words, more injuries can be found in patients with more imaging. For example, a patient may have a chest CT performed, finding a nondisplaced rib fracture, which would not have been found otherwise, consequently increasing the overall ISS. The Yeguiayan, et al (8) study is the most similar to ours, a prospective multi-institutional observational study performed in France. The authors describe the results of the FIRST (French Intensive care Recorded in Severe Trauma) study. The authors were interested in examining the impact of emergency care on hospital mortality of patients with severe blunt trauma. Acknowledging that the ISS and TRISS scores are created using information generated by the imaging, the authors argue that an adjustment for ISS, which is used clinically to estimate the risk of death, could overestimate the benefit of the WBCT. These authors used initial physiologic variables, GCS, treatments like transfusions and mechanical ventilation in their propensity model. We found similar differences in the population that underwent WBCT compared to the selective CT, in that the population was younger, had had more severe mechanisms of injury, and worse physiologic measures. Unlike that study, our patients also had a lower GCS and required more resuscitation. This would suggest that the 585
CHAN ET AL
Academic Radiology, Vol 23, No 5, May 2016
TABLE 2. Postpropensity Weighting Balance (N = 11,318)
Age, mean (SD) Home Distance to the University, mean (SD) ED Glasgow Coma Scale, mean (SD) ED Pulse, mean (SD) ED Systolic Blood Pressure, mean (SD) Glucose, mean (SD) Potassium, mean (SD) Anion Gap, mean (SD) Calcium, mean (SD) Sodium, mean (SD) BUN, mean (SD) Chloride, mean (SD) Creatinine, mean (SD) Carbon Dioxide, mean (SD) WBC Count, mean (SD) Red Blood Cell Count, mean (SD) Platelet Count, mean (SD) PTT, mean (SD) INR, mean (SD) Received FFP on Admission, % Received PRBCs on Admission, % Received Lactate Test, % Female, % Hospital Transfer, % Type of Insurance Private, % Public, % Not Funded, % Other, % Mechanism of Injury Fall, % Vehicle, % Assault, % Other, %
Selective Imaging (n = 9,774)
Whole-Body Imaging (n = 1,544)
Absolute Standardized Difference
43.8 (7.7) 248.0 (223.8) 12.1 (1.9) 95.6 (9.2) 126.3 (9.3) 130.1 (19.8) 3.9 (0.2) 10.2 (1.6) 8.4 (0.5) 140.0 (1.5) 14.9 (2.9) 108.4 (2.3) 0.9 (0.2) 21.3 (1.7) 13.5 (2.4) 4.5 (0.3) 252.3 (31.5) 29.1 (3.6) 1.2 (0.2) 4.9 4.7 45.1 26.4 36.6
44.2 (18.7) 241.0 (502.4) 12.2 (4.6) 94.9 (23.2) 126.9 (23.2) 130.1 (46.9) 3.9 (0.6) 10.3 (3.9) 8.4 (1.0) 140.0 (3.7) 15.1 (7.7) 108.2 (5.1) 0.9 (0.3) 21.5 (4.2) 13.5 (6.6) 4.5 (0.8) 253.5 (77.6) 29.2 (10.0) 1.2 (0.5) 4.0 5.3 44.6 26.4 34.8
0.026 0.018 0.017 0.040 0.032 <0.001 0.009 0.045 0.094 0.007 0.024 0.072 0.010 0.040 0.004 0.032 0.019 0.003 0.028 0.045 0.028 0.010 <0.001 0.038
34.4 26.8 34.2 4.6
35.4 26.6 32.7 5.3
0.022 0.005 0.032 0.031
21.5 72.9 1.8 3.8
21.7 72.8 1.9 3.6
0.005 0.002 0.008 0.011
BUN, blood urea nitrogen; ED, emergency department; FFP, fresh frozen plasma; INR, international normalized ratio; PRBC, packed red blood cell; PTT, partial thromboplastin time; RBC, red blood cell; SD, standard deviation; WBC, white blood cell.
decision to use WBCT, rather than selective CT, is based on a clinical picture that predicts more severe injuries, both in our institution, and in other settings where the decision is not proscribed. An important difference in the studies, however, is that ours included all levels of trauma, and was not limited to severe trauma, possibly accounting for differences in the significance of the predictor variables. The benefit of WBCT may be different, compared to the risk of overutilization, radiation exposure, cost and detection of incidental findings, when it is unlikely that mortal injuries are present. Our study has several limitations. In addition to being a retrospective study, a significant limitation is due to an indication bias. As described above, the trauma team uses clinical and physiologic information to make the decision for imaging intensity. The trauma team is more likely to perform more imaging on more severely injured patients. These patients are 586
more likely to have a longer inpatient hospital stay and to be discharged to a care facility rather than home. Controlling for surrogate markers of injury severity, particularly vital signs, GCS, and resuscitation volume, the propensity score is intended to mitigate the effect of indication bias, without being confounded by the imaging itself. It is possible that our model may fail to account for the critical confounding variables and suffer from a bias due to these omitted variables. We chose a propensity score to identify and normalize possible confounding variables between the patient populations including severity of injury (assessed using BP, HR, GCS, CBC, chemistry panel, volume of resuscitation products), and basic demographics (age, gender, insurance type) (Table 1). We draw in patients from a large geographic area, so we also controlled for home distance from the inpatient hospital because we felt that transportation home could affect dispositioin type
Academic Radiology, Vol 23, No 5, May 2016
IMPACT OF WHOLE-BODY COMPUTED TOMOGRAPHY
and length of stay. The significant overlap in the propensity curves of Figure 1 suggest that our analysis successfully normalized the patient populations with regards to the predictor variables. Yeguiavan, et al (8) also addresses this concern and argues, after adding ISS as a covariate in their regression model, that the mortality reduction benefit assigned to the WBCT on the basis of ISS or TRISS stratification is likely overestimated. As with any method of covariate adjustment method, propensity score adjustment only reduces bias, and makes an observational study more like a randomized study, to the extent that all sources of confounding are accounted for in the model. Laboratory and other clinical values have been presumed to serve as surrogates for measures of severity, as has been shown in previous retrospective studies (8) and will be tested in the ongoing prospective REACT-2 trial (15). Other limitations include the variation of practice amongst trauma providers in deciding which patients receive selective CT imaging versus WBCT imaging. However, this model of practice is reflective of the majority of American EDs. There were no statistically or clinically significant differences in inpatient hospital LOS and discharge disposition of the average blunt trauma patient who underwent WBCT imaging versus selective CT imaging, but subset analysis may yield differences in outcomes that guide CT imaging utilization in the ED. A subset analysis of patients with head trauma and those patients who required ICU admission may yield different outcomes.
tional Sciences, National Institutes of Health, through Grant 8UL1TR000105 (formerly UL1RR025764).
CONCLUSIONS We retrospectively analyzed a large population of blunt trauma patients at a single level 1 trauma center and found that there was no statistically or clinically significant difference in inpatient hospital LOS or discharge location in patients who underwent WBCT. ACKNOWLEDGMENTS This investigation was primarily supported by funding from the Harvey L. Neiman Health Policy Institute, and also by the University of Utah Study Design and Biostatistics Center, with funding in part from the National Center for Research Resources and the National Center for Advancing Transla-
REFERENCES 1. Kocher KE, Meurer WJ, Fazel R, et al. National trends in use of computed tomography in the emergency department. Ann Emerg Med 2011; 58:452–62 e3. doi:10.1016/j.annemergmed.2011.05.020. 2. Rao VM, Levin DC, Parker L, et al. Trends in utilization rates of the various imaging modalities in emergency departments: nationwide Medicare data from 2000 to 2008. J Am Coll Radiol 2011; 8:706–709. doi:10.1016/ j.jacr.2011.04.004. 3. van Vugt R, Kool DR, Deunk J, et al. Effects on mortality, treatment, and time management as a result of routine use of total body computed tomography in blunt high-energy trauma patients. J Trauma Acute Care Surg 2012; 72:553–559. doi:10.1097/TA.0b013e31822dd93b. 4. Huber-Wagner S, Lefering R, Qvick LM, et al. Effect of whole-body CT during trauma resuscitation on survival: a retrospective, multicentre study. Lancet 2009; 373:1455–1461. doi:10.1016/S0140-6736(09)60232-4. 5. Asha S, Curtis KA, Grant N, et al. Comparison of radiation exposure of trauma patients from diagnostic radiology procedures before and after the introduction of a panscan protocol. Emerg Med Australas 2012; 24:43– 51. doi:10.1111/j.1742-6723.2011.01504.x. 6. Inaba K, Branco BC, Lim G, et al. The increasing burden of radiation exposure in the management of trauma patients. J Trauma 2011; 70:1366– 1370. doi:10.1097/TA.0b013e3181ebb4d4. 7. Salim A, Sangthong B, Martin M, et al. Whole body imaging in blunt multisystem trauma patients without obvious signs of injury: results of a prospective study. Arch Surg 2006; 141:468–473, discussion 73-5. doi:10.1001/archsurg.141.5.468. 8. Yeguiayan JM, Yap A, Freysz M, et al. Impact of whole-body computed tomography on mortality and surgical management of severe blunt trauma. Crit Care 2012; 16:R101. doi:10.1186/cc11375. 9. Weninger P, Mauritz W, Fridrich P, et al. Emergency room management of patients with blunt major trauma: evaluation of the multislice computed tomography protocol exemplified by an urban trauma center. J Trauma 2007; 62:584–591. doi:10.1097/01.ta.0000221797.46249.ee. 10. Baker SP, O’Neill B, Haddon W, Jr, et al. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1974; 14:187–196. 11. Kahloul M, Bouida W, Boubaker H, et al. Value of anatomic and physiologic scoring systems in outcome prediction of trauma patients. Eur J Emerg Med 2014; 21:125–129. doi:10.1097/MEJ.0b013e32836188ce. 12. Baker SP, O’Neill B, Haddon W, Jr, et al. The injury Severity Score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1974; 14:187–196. 13. Austin P. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res 2011; 46:399–424. 14. Jiang L, Ma Y, Jiang S, et al. Comparison of whole body computed tomography vs selective radiological imaging on outcomes in major trauma patients: a meta-analysis. Scand J Trauma Resusc Emerg Med 2014; 22:54. 15. Caputo ND, Stahmer C, Lim G, et al. Whole-body computed tomographic scanning leads to better survival as opposed to selective scanning in trauma patients: a systematic review and meta-analysis. J Trauma Acute Care Surg 2014; 77:534–539.
587