Patients with blunt trauma having both a chest radiograph and thoracic computed tomography.
Kaiser ML, Whealon MD, Barrios C, Dobson SC, Malinoski DJ, Dolich MO, Lekawa ME, Hoyt DB, Cinat ME. Risk Factors for Traumatic Injury Findings on Thoracic Computed Tomography Among Patients With Blunt Trauma Having a Normal Chest Radiograph. Arch Surg. 2011;146(4):459-463. doi:10.1001/archsurg.2011.56
We sought to identify risk factors that might predict acute traumatic injury findings on thoracic computed tomography (TCT) among patients having a normal initial chest radiograph (CR).
In this retrospective analysis, Abbreviated Injury Score cutoffs were chosen to correspond with obvious physical examination findings. Multivariate logistic regression analysis was performed to identify risk factors predicting acute traumatic injury findings.
Urban level I trauma center.
All patients with blunt trauma having both CR and TCT between July 1, 2005, and June 30, 2007. Patients with abnormalities on their CR were excluded.
Main Outcome Measure
Finding of any acute traumatic abnormality on TCT, despite a normal CR.
A total of 2435 patients with blunt trauma were identified; 1744 (71.6%) had a normal initial CR, and 394 (22.6%) of these had acute traumatic findings on TCT. Multivariate logistic regression demonstrated that an abdominal Abbreviated Injury Score of 3 or higher (P = .001; odds ratio, 2.6), a pelvic or extremity Abbreviated Injury Score of 2 or higher (P < .001; odds ratio, 2.0), age older than 30 years (P = .004; odds ratio, 1.4), and male sex (P = .04; odds ratio, 1.3) were significantly associated with traumatic findings on TCT. No aortic injuries were diagnosed in patients with a normal CR. Limiting TCT to patients with 1 or more risk factors predicting acute traumatic injury findings would have resulted in reduced radiation exposure and in a cost savings of almost $250 000 over the 2-year period. Limiting TCT to this degree would not have missed any clinically significant vertebral fractures or vascular injuries.
Among patients with a normal screening CR, reserving TCT for older male patients with abdominal or extremity blunt trauma seems safe and cost-effective.
In recent years, the number of thoracic computed tomographic (TCT) images obtained among patients with blunt trauma has risen dramatically.1,2 More than one-third of all CT images are intended to elucidate chest anatomy.3 This has led to growing concern about the potentially detrimental effects of obtaining additional images. Patients who are exposed to radiation incur a small associated risk of cancer.4 Patients, institutions, and society as a whole assume the added economic burden of obtaining and interpreting these studies.5
A 2009 study5 from our institution demonstrated that TCT after a normal initial screening chest radiograph (CR) in patients with blunt trauma rarely identifies clinically significant injury, termed occult injury. However, the possibility of aortic injury (and the reported insensitivity of CRs for this potentially lethal injury) has led many physicians to err on the side of caution and obtain TCT if questions remain.6 Establishing indications to pursue TCT after a normal CR could spare patients unnecessary radiation exposure while safely conserving health care resources. We sought to identify risk factors that might predict acute traumatic injury findings on TCT among patients with blunt trauma having a normal initial screening CR.
We performed a retrospective analysis of the trauma registry at our urban level I trauma center. All patients with blunt trauma having a normal initial screening CR who subsequently underwent TCT between July 1, 2005, and June 30, 2007, were included. Patients with any traumatic abnormality on their CR, including but not limited to pneumothorax, hemothorax, pulmonary contusion, abnormal mediastinal contour, or any evidence of fracture, were excluded. At our institution, adult patients typically undergo a single anteroposterior digital CR in the supine position during maximal inspiration (if cooperative) as part of the primary evaluation. Inspiration and expiration radiographs are not routinely used to diagnose pneumothoraces in the trauma bay. The main outcome measure was the finding of any acute traumatic abnormality on TCT as noted on the final report of the attending radiologist. Risk factors analyzed included age, sex, hypotension (defined as systolic blood pressure <90 mm Hg), Glasgow Coma Scale score of 12 or less, Revised Trauma Score less than 4, and injury in other anatomic regions on the basis of Abbreviated Injury Score (AIS) scores. Cutoff AISs were selected to maximize sensitivity while coinciding with obvious evidence of injury on history, physical examination, radiographs, or focused assessment with sonography for trauma (FAST) image. Examples are given in Table 1.
Pearson product moment correlation χ2 test was used for univariate analysis. All variables with P < .20 on univariate analysis were included in a binary logistic regression analysis to identify independent risk factors for occult thoracic injury. Significance on multivariate analysis was defined as P ≤ .05. Commercially available software (SPSS version 11.0; SPSS Inc, Chicago, Illinois) was used for the statistical analysis. This study was approved by our institutional review board.
Over the 2-year study period, 2435 patients with blunt trauma having both CR and TCT were identified (Figure). Of these, 691 patients had a traumatic abnormality identified on their CR, including 244 with widened mediastinum or abnormal mediastinal contour, 217 with rib fractures, 183 with pulmonary contusions, 113 with clavicular fractures, 68 with pneumothoraces, 45 with hemothoraces, 31 with scapular fractures, 11 with pneumomediastinum, and 1 with sternoclavicular dislocation (some patients had ≥1 injury). Exclusion of all patients with abnormality on their initial CR left 1744 patients (71.6%) for analysis, 394 (22.6%) of whom had acute traumatic findings on TCT. Some patients had more than 1 occult injury. No patient with a normal CR was ultimately diagnosed as having aortic injury. Two patients had findings suggestive of aortic transection vs artifact on TCT, but both later had negative findings on formal angiograms, ruling out injury.
Risk factors for occult thoracic injury are listed in Table 2. Among 1744 patients with a normal CR, 1532 (87.8%) had 1 or more of 4 statistically significant independent risk factors (abdominal AIS ≥3, pelvic or extremity AIS ≥2, age >30 years, and male sex), and 794 (45.5%) had 2 or more. Table 3 compares the number of injuries detected by performing TCT on all 1744 patients vs only those with risk factors for occult thoracic injury. Thirty-two pneumothoraces would have been undetected by imaging only patients with 2 or more risk factors; 2 of these 32 patients underwent tube thoracostomy, although it is unclear whether these were placed as a precaution or for immediate indications. The first was placed within an hour of arrival to the trauma bay and was removed within 1 day; the second was placed in a patient with mechanical ventilation several days after admission, when the pneumothorax became visible on daily CRs. None of 7 patients with potentially undetected hemothoraces required chest tube placement. Two patients with pneumomediastinum who would not have undergone TCT using 2 or more risk factors as a threshold had no injury noted on esophagoscopy and bronchoscopy. Of 10 patients with vertebral fractures who would not have undergone TCT using the same threshold, 3 had minor spinous or transverse process fractures, and all but 2 had substantial tenderness or pain while in the trauma bay. In the latter 2 patients without symptoms, one injury was suspected to represent artifact. The other injury prompted a consultation with an orthopedic spine specialist, who diagnosed a stable, minimally displaced compression fracture requiring no intervention or weight-bearing precautions.
This study demonstrates that among patients with blunt trauma having a normal CR at admission, occult findings on TCT are more likely in older male patients with abdominal, pelvic, or extremity injuries. Given these findings, even setting our threshold low and performing TCT in all patients with 1 or more risk factors for occult thoracic injury, TCT would have been performed in 212 fewer patients over a 2-year period without missing any clinically significant injuries. Performing TCT in all patients with 2 or more risk factors would have safely cut our total number of images by half. Although a few occult pneumothoraces and hemothoraces might have gone undiagnosed on the basis of TCT images, recent studies5,7 demonstrate that both types of occult injuries can often be managed safely without tube thoracostomy and likely do not require the degree of monitoring that we once believed was necessary. Other acute injuries, such as vertebral fractures, are almost always symptomatic, and findings such as midline tenderness alone should prompt further selective investigation. Finally, cardiac effusion, suggestive of blunt cardiac injury and possible hemopericardium, should be apparent on the FAST image routinely performed at many trauma centers. Given these data, limiting TCT to patients with 1 or more risk factors seems a reasonable and conservative recommendation.
A 2006 report from the National Research Council of the National Academies4 indicates that previous studies underestimated the deleterious effects of diagnostic radiation. It was estimated in 2007 that up to 2% of all future cancer cases in the United States might be due to radiation from CT.8 Moreover, many agitated patients with traumatic injury must undergo successive imaging attempts, and the use of dosimeters on patients during trauma assessment has demonstrated doses far exceeding our initial estimates.9 Of particular concern with TCT are cancers of the thyroid, breast, and lung, all of which are linked to radiation exposure. A typical patient with traumatic injury who undergoes irradiation at a young age incurs an increased cumulative lifetime risk of developing cancer.8 A radiation dose of 0.01 gray raises the risk of breast cancer in a 35-year-old woman by about 14%; the average TCT system delivers at least twice this dose.10 Concern about excess radiation exposure has given rise to various techniques and technologies designed to minimize radiation exposure,11 none of which are as effective as thoughtfully avoiding nonindicated imaging altogether. Pediatric surgeons have successfully pursued a culture of minimal radiation exposure in their vulnerable patient population while upholding a standard of safety and efficiency; it is reasonable to expect trauma surgeons to implement minor protocol changes in an effort to do likewise.
Radiation exposure is not the only problematic aspect of TCT imaging. Nephropathy may occur secondary to intravenous contrast. Additional time may be spent imaging, delaying definitive treatments. Occult injuries identified on TCT may prompt the use of increased resources that may be unnecessary (eg, hospital or intensive care unit admission). Finally, relevant to the current economic health care crisis, overutilization of TCT increases costs. At our institution, the charges for TCT with contrast are $874, and radiologist reading fees account for an additional $251. Forgoing imaging on 212 patients herein with no risk factors for occult thoracic injury would have resulted in a cost savings of almost $250 000 over the 2-year period.
It has been argued that the risk of missing a potentially devastating aortic injury far outweighs any downsides of thoracic imaging. However, no patient in our study with a completely normal CR was ultimately diagnosed as having an aortic injury. Evidence shows an 8% incidence of aortic rupture on TCT that was not apparent on CR: in a 1998 study by Demetriades et al,6 only 9 patients demonstrated aortic disruption on TCT among 112 patients experiencing high-speed deceleration injury. Six of the 9 patients had a traumatic abnormality visible on CRs, ranging from fractures of the ribs, clavicle, and sternum to hemothoraces and left apical capping. Since publication of that study, the advent of digital CRs in many trauma centers has drastically increased the ease with which injuries can be diagnosed on radiograph alone. Similar to the findings herein, a 2001 study by Omert et al12 failed to detect any aortic abnormalities in patients having blunt trauma with normal CRs.
The strengths of this study are that it includes many patients from a comprehensive registry with numerous data points to answer a clinically relevant question. The primary limitation of this study is its retrospective nature. Inherent in the retrospective approach is uncertainty about when CR and TCT studies were viewed and the reports dictated. At our center, screening CRs obtained in the trauma bay are reviewed before CT imaging by the attending surgeon (C.B., D.J.M., M.O.D., M.E.L., D.B.H., and M.E.C.), the senior trauma resident, or both. As a result, some CRs judged as normal may later be interpreted by the attending radiologist as having subtle evidence of injury. Using our database, it was impossible to retrospectively determine how many abnormal CRs were originally misread as normal or what were the slight findings on these particular CRs. Alternatively, the radiologist may have reviewed both the CR and TCT images simultaneously once both were completed; therefore, the results of one may have influenced the interpretation of the other. A final weakness is the use of AISs (which are generally complete only after CT imaging) as surrogates for other findings available while still in the trauma bay. We acknowledge that not all patients having diagnoses with associated AISs will have obvious external evidence of injury on physical examination, particularly in the case of obtunded, obese, or pediatric patients. A prospective clinical trial will be necessary to thoroughly address this question and is under way at our institution.
In conclusion, while it is well established that TCT is indicated in the presence of high-risk deceleration injury,6 the use of TCT has expanded beyond this clinical indication in many trauma centers. In situations in which the mechanistic details are unclear, TCT is most likely to reveal occult thoracic injury not seen on CRs in older male patients with clinically significant abdominal, pelvic, or extremity injuries. Reserving TCT for patients with 1 or more risk factors predicting occult thoracic injury may safely conserve resources, while minimizing radiation exposure among a vulnerable population.
Correspondence: Marianne E. Cinat, MD, Department of Surgery, University of California, Irvine, 333 City Blvd W, The City Tower, Ste 705, Orange, CA 92868 (firstname.lastname@example.org).
Accepted for Publication: March 30, 2010.
Author Contributions: Dr Kaiser had full access to all the data in the study and takes full responsibility for the integrity of the data and the data analysis. Study concept and design: Kaiser, Whealon, Barrios, and Cinat. Acquisition of data: Kaiser, Barrios, Malinoski, Dolich, Lekawa, Hoyt, and Cinat. Analysis and interpretation of data: Kaiser, Whealon, Dobson, and Cinat. Drafting of the manuscript: Kaiser, Whealon, Dobson, and Cinat. Critical revision of the manuscript for important intellectual content: Barrios, Malinoski, Dolich, Lekawa, Hoyt, and Cinat.
Financial Disclosure: None reported.
Previous Presentation: This study was presented as an e-poster at the 81st Annual Meeting of the Pacific Coast Surgical Society; February 14, 2010; Kapalua, Hawaii.
Additional Contributions: Xuan-Mai Nguyen, PhD, performed independent statistical review.