| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Radiology, University of Texas Southwestern Medical Center at
Dallas, 5323 Harry Hines Blvd., Dallas, TX 75235-8896.
2
Department of Radiology, St. Paul Medical Center, 5909 Harry Hines Blvd.,
Dallas, TX 75235.
3
Associate Radiologists, P.C., 4544 Harding Rd., Ste. 215, P.O. Box 50254,
Nashville, TN 37205.
4
Mesquite Center Radiologists, 1011 N. Galloway Ave., Mesquite, TX 75149.
Received May 8, 2000;
accepted after revision October 26, 2000.
Presented at the annual meeting of the American Roentgen Ray Society,
Washington, DC, May 2000.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. The institutional review board approved the parallel imaging—CT immediately followed by aortography—of patients presenting with blunt chest trauma between August 1997 and August 1998. To screen patients for potential aortic injuries, we performed parallel imaging on 142 patients, and these patients comprised our patient population. CT examinations of the patients were reviewed for signs of injury by radiologists who were unaware of each other's interpretations and the aortographic results. Findings of CT examinations were classified as negative, positive, or inconclusive for injury. Aortography was performed immediately after CT. The technical and professional fees for both transcatheter aortography and helical CT were also compared.
RESULTS. Our combined kappa value for all CT interpretations was 0.714. The aortographic sensitivity and negative predictive value were both 100%. Likewise, the sensitivity and negative predictive value of CT were 100%. The total costs of performing aortography were estimated at approximately $402,900, whereas those for performing helical CT were estimated at $202,800.
CONCLUSION. Helical CT has a sensitivity and negative predictive value equivalent to that of aortography. Using CT to eliminate the possibility of mediastinal hematoma and to evaluate the cause of an abnormal aortic contour in a trauma patient allows us to use aortography more selectively. Avoiding the performance of unnecessary aortography will expedite patient care and reduce costs. We report the results of our experience with CT and how our center successfully made this transition in the initial examination of patients with serious thoracic trauma.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We were also interested in the potential financial savings that could be gained using CT instead of routine aortography in the clinical examination of these patients.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Patients
During the study period, 3870 patients with blunt trauma presented at our
trauma center. Conventional chest radiographs cleared 3605 of these patients
of possible serious mediastinal trauma. The remaining 265 patients required
further imaging to exclude potentially acute thoracic aortic injury. One
patient died of an autopsy-proven aortic rupture before further imaging could
be performed. A total of 264 patients were imaged with either helical CT or
aortography or both. Of these patients, 95 bypassed the CT procedure and were
examined with aortography alone because of their age, hemodynamic instability,
or inability or unwillingness to give consent. Consent for aortography was
refused by 27 patients or their responsible family members or was declined by
the referring trauma physician. These patients were examined with CT alone.
Our final study population was the remaining 142 patients who underwent
parallel imaging.
CT
Our technique was developed from a hybrid of specific parameters that each
of the individual CT observers desired. All the CT examinations were performed
on one of three scanners of the same make and model (PQ 6000; Picker
International, Cleveland, OH). One hundred forty milliliters of iohexol
(Omnipaque 300; Nycomed, Princeton, NJ) at a concentration of either 300 mg
I/mL or 350 mg I/mL, depending on the estimated body weight of the patient,
was injected at 2 mL/sec through at least an 18-gauge antecubital fossa IV
line. The latter concentration was reserved for patients who were estimated to
weigh more than 250 lb (112.5 kg). After a scan delay of 45 sec, 3-mm helical
images were acquired from approximately 1 cm above the aortic arch through the
inferior pulmonary vein; 8-mm images were then acquired through the remainder
of the lung bases (and the abdomen and pelvis, if indicated); and, finally,
immediate delayed 5-mm images were acquired from the thoracic inlet back down
through the aorticopulmonary window. A pitch of 1.5 was used. Images through
the arch and proximal branch vessels were reconstructed at 1.5-mm intervals.
Indeterminate and positive studies were transferred to a workstation (Voxel Q,
Picker International) for aortographic-equivalent reconstructions.
The CT examinations were monitored by one of four staff radiologists trained in either thoracic imaging or trauma radiology. All CT examinations were then independently reviewed on both the workstation and on film by at least two of these radiologists.
All the CT observers were free to manipulate window and level settings at
their discretion and were unaware of each other's interpretations and of the
aortographic results. Interobserver variability among the CT observers was
assessed by the kappa (
) statistic.
Each CT examination was checked for direct signs (e.g., pseudoaneurysm, intimal flap, pseudocoarctation, dissection, contrast material extravasation) and indirect signs (e.g., contour anomaly, hemomediastinum, peribranch vessel blood) of acute traumatic aortic injury that have been previously described in the literature [8, 14, 15]. For the purpose of this study, we did not consider blood isolated to the anterior mediastinal compartment a sign of potential vascular injury. Findings for all CT examinations were classified as negative, positive, or inconclusive for acute aortic or branch vessel injury. Studies judged to be inconclusive were those showing hemomediastinum either alone or in conjunction with subtle aortic contour anomalies; those showing peribranch vessel hematoma; and those compromised by motion, pulsation artifact, streak artifact, or suboptimal contrast medium bolus.
Aortography
Immediately after completion of CT, transcatheter aortography was performed
with a 6-French high-flow pigtail catheter introduced via a femoral arterial
puncture using a standard Seldinger technique and was advanced over a
guidewire into the root of the aorta.
Early in the study, we used single-plane aortography, acquiring images digitally (Multistar; Siemens Medical Systems, Iselin, NJ) in two projections, usually, 30° right anterior oblique and approximately 40° left anterior oblique, depending on the patient's anatomy. Each projection was obtained with the injection of 60-70 mL of undiluted contrast material (Omnipaque 300; Nycomed) administered at a rate of 30-35 mL/sec for a 2-sec injection or a total of approximately 140 mL. Images were acquired at a rate of six frames per second for 4-6 sec.
Later in the study, images were acquired by rotational aortography (Dynavision; Siemens Medical Systems), a procedure during which the X-ray tube and image receptor are rotated axially around the patient at approximately 15° per sec while images are obtained every 2.5° of rotation. Images were reviewed in a cine format as well as in subtracted and native (unsubtracted) modes. The length of the injection was determined by adding the time estimated for contrast material to flow from the root of the aorta to the level of the diaphragm (usually 0.5-1.0 sec) and a rotational acquisition time of approximately 3 sec. The rate of injection was determined by estimating the flow in the aorta, typically 25-35 mL/sec. The total dose of contrast material ranged between 100 and 160 mL (including 10-20 mL used to position the catheter and to assess blood flow). All aortograms were interpreted by trained and experienced interventional radiologists unaware of the results of the CT scans of the patient's chest. The aortograms were interpreted by the same radiologist who performed the study.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The mechanism of injury involved a motor vehicle for almost 88% of the patients. Seventy-eight percent of the injuries resulted from motor vehicle collisions. Motorcycle and motor—pedestrian collisions were responsible for 3.5% and 6.5% of the injuries to patients, respectively. Falls ranging between 6 and 30 ft (1.829-9.144 m; mean, 17.4 ft [5.304 m]) accounted for about 9% of the injuries. The remaining 3% of injuries stemmed from a variety of mishaps (e.g., waterskiing accident, helicopter crash, airplane crash, and industrial cave-in).
CT Data
Of the 142 CT examinations performed, 121 were interpreted as negative, and
120 had accompanying negative aortograms. The aortogram on a 53-year-old man
involved in a motor vehicle collision was prospectively interpreted as
revealing an apparent traumatic pseudoaneurysm at the root of an aberrant
right subclavian artery. The patient was treated conservatively. The
interventional radiologist subsequently reviewed the case and considered the
findings to be inconclusive for an acute injury (Fig.
1A,1B).
We regarded the findings for these 121 patients as our true-negatives. There
were no proven false-negative findings.
|
|
The CT examinations of seven patients were interpreted as positive. Six of these patients had accompanying positive aortograms. The CT scan of an 80-year-old man injured in a motor vehicle collision showed a triangular contour abnormality along the lateral aspect of the aortic arch that persisted on delayed images and was suggestive of a possible subtle injury. Although no evidence of a paraaortic hematoma was adjacent to this region, no evidence of atherosclerotic plaque was found in this vicinity or elsewhere in the aorta. The aortogram did appear somewhat unusual but was eventually interpreted as negative. The consulting cardiovascular—thoracic team also was concerned about the appearance of the aorta on both studies. However, the surgeons elected not to operate and continued to treat the patient medically with ß-adrenergic blocking agents without complications (Fig. 2A,2B). Despite the clinical indications, we believe the CT finding for this patient to be false-positive. With the exception of this patient, all the CT scans prospectively interpreted as positive showed variable amounts of middle mediastinal blood. The CT scans of six patients revealed periisthmus contour abnormalities, and five revealed intimal flaps.
|
|
For the purpose of this study, we regarded an inconclusive CT as being a potentially positive finding for injury. Fourteen CT examinations were prospectively interpreted as inconclusive for potentially acute aortic or branch vessel injuries because of the presence of hemomediastinum or various contour abnormalities. Eleven of these patients had accompanying negative aortograms.
The three remaining patients with inconclusive CT examinations included a 25-year-old man who had fallen from a second-story window. His aortogram was initially interpreted as inconclusive for a possible acute injury. The patient was treated medically during the night. The interventional radiologist reviewed the findings the next morning with one of his colleagues. Their consensus was that the findings of study were negative for an acute injury. The aortogram of a 61-year-old woman injured in a motor vehicle collision was initially interpreted as positive for an acute aortic injury. Because of complicating medical and surgical problems, the patient was treated conservatively and stabilized with ß-adrenergic blocking agents. Nine days later, the cardiovascular—thoracic team requested a follow-up aortogram to help them to reassess the injury in preparation for the upcoming surgery. As in the patient's initial aortogram, the follow-up examination showed extensive atherosclerotic disease but was interpreted as showing no acute injury. The aortographic findings of the remaining patient, a 52-year-old man injured in a motor vehicle collision, were indeed positive for acute injury (Fig. 3A,3B,3C).
|
|
|
Our conservatism regarding the inconclusive interpretations led to the creation of several false-positive examinations, decreasing our rate of specificity. Scrutiny of the rationale for the various inconclusive, prospective interpretations revealed problems typical of an early learning curve. More than 50% of the inconclusive interpretations occurred during the first 20 weeks of the study. Most of these interpretations were related to variable appearances in normal mediastinal anatomy, such as the ductus bump, the left supreme intercostal vein, and the bronchial artery infundibulum (six examinations). The interpreters of the CT examinations prospectively commented that these normal anatomic structures were likely responsible for the apparent contour abnormalities, but they erred on the side of conservatism. The CT scan of a 19-year-old man injured in a high-speed motor vehicle collision was judged to be inconclusive on the basis of technical factors—a combination of pulsation and streak artifact around the superior pericardial recess. Atherosclerotic plaque mimicked the appearance of an intimal flap on the CT scan of a 64-year-old man injured in a motor vehicle collision. As the study progressed, fewer examinations were interpreted as inconclusive on the basis of such technical factors. The remaining examinations revealed indirect signs of mediastinal trauma—hemomediastinum, peribranch vessel hematoma, or contour abnormalities—that would warrant further aortographic investigation or confirmation at most centers.
In all, there were 14 CT examinations with false-positive findings. Based on these conclusions and the data described, CT at our institution had a sensitivity of 100% and a negative predictive value of 100% for the detection of acute aortic injuries.
Interobserver Variability
The aortographers were aware of each other's interpretations but were
unaware of the CT results. However, the CT observers were not aware of either
each other's interpretations or the aortographic results. We measured the
degree of interobserver variability among the four CT observers with the kappa
statistic, which is a measure of agreement among observers. The value is
scaled to be 0 if the amount of agreement is what would be expected by chance
alone and if there is perfect agreement. For intermediate values, we used the
interpretation suggested by Landis and Koch
[16]. The kappa value for
positive CT examinations was very high (=0.913), as was that for
negative examinations (=0.724). The combined kappa value for all CT
interpretations was 0.714, a value indicative of substantial agreement.
Aortographic Data
A total of 132 aortograms were interpreted as negative. Among the nine
aortograms prospectively interpreted as positive and the one judged to be
inconclusive were seven surgically proven aortic tears and the three
false-positive examinations discussed earlier. Thus, the sensitivity and
negative predictive value for transcatheter aortography at our institution
were also 100%.
In the interest of completeness, we should report on those 95 patients who bypassed CT and underwent aortography because of the established institutional review board criteria.
Four of these patients had acute aortic injuries that were revealed at aortography and surgically confirmed. During the study period, 11 patients sustained aortic injuries; this total represents a prevalence of approximately 5% in our trauma patients.
Cost Savings
CT is not only less invasive but also less costly than transcatheter
aortography. In-house technologists are available 24 hr a day to perform CT,
but aortography requires more labor and resources than CT as well as the
additional expenses of overtime and nursing services. In addition to these
factors, the difference in the combined technical and professional charges
between the two examinations at our institution is approximately $500.
During the time of our study, the total cost of the aortograms was estimated at approximately $402,900. This figure includes the cost for the 142 patients who underwent parallel imaging and the 95 patients that bypassed CT. The total cost of the CT examinations was estimated at approximately $202,800. This figure includes the cost for the additional 27 patients not imaged with aortography. By excluding the presence of mediastinal hematoma and confirming that a patient had a normal aorta, the use of CT feasibly could have reduced the number of patients needing to undergo aortography to 21—seven patients whose CT examinations showed positive findings and 14 whose scans showed inconclusive findings—for a cost of approximately $36,000.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Granted, the incidence of acute traumatic aortic injury at any single trauma center, including ours, is low. More useful information may be obtained from the data generated from multicenter trials than from data generated from a study conducted at a single institution. However, knowing the results reported in the literature does not guarantee being able to reproduce those results at one's own center. The trauma physicians and cardiovascular—thoracic surgeons at our own institution were reluctant to accept CT in lieu of aortography without our first assessing its accuracy. This reluctance was overcome by our study comparing the results of both imaging modalities.
There is, of course, a selection bias in our data because the trauma surgeons could, at their discretion, both veto the option to perform CT before aortography and override the exclusion criteria. For this reason, we purposely focused on only those 142 patients who actually did undergo parallel imaging.
As reported earlier, all CT interpreters involved in the study were unaware of each other's interpretations and of the aortographic results. Throughout the course of our study, our interpreters had little interobserver variability, and our combined kappa values revealed substantial interobserver agreement. Obviously, the introduction of any new modality or means of evaluating a particular disease process means that radiologists are faced with a learning curve. Our CT interpreters were no exception. This learning curve, combined with our willingness to err on the side of conservatism, resulted in a CT specificity of only 89%. However, of greater importance are the facts that CT had 100% sensitivity and a 100% negative predictive value for the detection of possible acute aortic injuries.
Our technique may be regarded as somewhat stringent. The decision to obtain thincollimated 3-mm images reconstructed every 1.5 mm stemmed from our concern that variable appearances in the ductus bump potentially could mimic the appearance of injuries. We also hoped this technique might facilitate the detection of peribranch vessel hematoma or branch vessel injuries.
Unfortunately, the technique did not prove helpful in this regard. No unexpected branch vessel injuries were detected on either CT or aortography during the course of our study. However, it is likely that the subtle injury detected on the CT examination illustrated in Figure 3A,3B,3C would have otherwise gone unrecognized without this particular technique. In addition, as discussed earlier, there is somewhat of a learning curve when it comes to the appreciation of the variable appearance of normal mediastinal vascular structure (ductus bump, bronchial and third intercostal artery infundibula, and superior intercostal vein). Thin-collimated images aid in gaining appreciation of such variable appearances. As the radiologist becomes more familiar with the appearances of these structures, it is possible that less stringent imaging protocols may be used but potentially at the expense of missing otherwise subtle injuries.
Perhaps another peculiarity to our technique is the second run of 5-mm images through the thoracic aorta. These immediate delayed images have proven most helpful in resolving questions involving images degraded by motion or pulsation artifacts, especially those images obtained in patients with tachycardia. This technique has substantially reduced the number of limited or suboptimal studies.
It is also our preference to inject contrast medium through right-arm IV lines whenever possible. Left-arm injections are often complicated by streak artifact as the bolus traverses the thoracic inlet vessels, often obscuring the branch vessel origins. However, the immediate delayed images obtained through the chest often minimize this artifact as well, allowing better delineation of these structures. ECG leads are routinely moved to the patients' shoulders, and nasogastric tubes are removed whenever possible to reduce the streak artifacts often created by these devices.
Our rate of contrast medium injection is somewhat slower than that used by many other institutions [8, 9, 11, 13, 14]. Our goal was to provide both a vascular phase study of the mediastinum, and, at the same time, a parenchymal phase study of the abdomen. The latter is important for radiologists to diagnose confidently traumatic hepatic and splenic injuries. However, as the technology has continued to evolve, even better aortic opacification can be achieved on today's faster CT scanners at injection rates of 3-4 mL/sec without having to sacrifice evaluation of the subdiaphragmatic viscera.
Although often helpful to our clinical colleagues in preparing for surgical reconstruction, the reformatted and reconstructed images have proven to be of little diagnostic value. We initially transferred all inconclusive and positive CT examinations to a workstation for aortographic reformation and reconstruction. However, these images did nothing to aid us in our diagnosis. Those studies that were clearly positive on the two-dimensional axial images were likewise positive on the reconstructions. It was our hope that the reformations would direct our diagnosis in the more problematic two-dimensional examinations with inconclusive findings. Such was not the case. We have since abandoned the routine reformation of CT scans of the chests of trauma patients. However, these images are made available at the request of the operating surgeon. The reformatted images that will be available with the newer multidetector scanners will likely be of greater diagnostic value.
Attempts should always be made to minimize the volume of contrast material administered to a patient without compromising the diagnostic yield of the study. Those trauma patients who underwent parallel imaging received a substantial quantity of nonionic contrast material. A total of between 280 and 300 mL of contrast material was administered for both examinations. Such a contrast medium load may be cause for concern, especially in elderly patients and those requiring subsequent operative intervention. Sequential serum creatinine levels were followed by the clinical services in less than 50% of our patients during the first 3 days after the contrast-enhanced imaging studies. None of these patients showed any laboratory evidence of compromised renal function. A review of the medical records of patients who did not have follow-up serum creatinine levels drawn likewise showed no reported cases of acute tubular necrosis, renal failure, or subsequent dialysis. The concern over the contrast medium volume may be more apparent than real. Perhaps the use of nonionic contrast material for both examinations, the average age of the trauma patients, and their relative hyperdynamic status and fluid resuscitation requirements somewhat protect the patients from the potential nephrotoxicity of the large bolus of contrast medium. Further studies with a larger patient population are necessary before such a hypothesis can be substantiated.
One potential criticism may be our attempt to exclude older trauma patients from undergoing parallel imaging. We did so for two reasons. The first was to protect this particular subset of patients from the double load of contrast medium and to expedite their particular clinical workup. The second reason was to minimize the potential complicating variable of atheromatous disease mimicking an acute injury. Atherosclerotic disease is a well-recognized cause of a false-positive CT interpretation. The use of CT in the examination of patients with acute blunt chest trauma was relatively new to our CT interpreters, and we wanted to minimize this variable until our learning curve was better established. However, we were not given that luxury. As discussed earlier, the exclusion criteria could be overridden at the discretion of the clinicians if the information gained by both examinations might affect a patient's treatment. The age range of our patients was quite broad, including many patients in the seventh, eighth, and even ninth decade of life. Surprisingly, the variable appearances of normal mediastinal anatomy was the cause of more inconclusive findings for the CT examinations than was atherosclerotic disease.
Because of its usefulness in excluding the presence of middle mediastinal hematoma and revealing a normal aorta, helical CT has helped reduce the ratio of negative (and hence unnecessary) aortographic findings compared with positive aortographic findings from 10:1 to 5:1 in several studies [12, 18]. Since the completion of this study, our institution has adopted the routine use of helical CT in the examination of all hemodynamically stable blunt chest trauma patients who have a radiographically unclearable mediastinum. Patients with CT evidence of peribranch vessel blood, paraaortic blood, or contour abnormalities in the thoracic aorta are then examined with aortography.
Another benefit of the use of CT in this setting has been a substantial cost savings. There is an approximate $500 difference between the total cost of an aortogram and a contrast-enhanced CT scan at our institution. In the year following the completion of this study, 298 unnecessary aortograms were avoided because of our use of CT. During that same year, the number of arch aortograms performed for blunt thoracic trauma patients was reduced to 32 with an examination cost savings of approximately $149,000.
In conclusion, helical CT has a sensitivity and negative predictive value equivalent to that of aortography. Because of these valuable factors, aortography can be used selectively in examining trauma patients. Reducing the number of unnecessary aortographic examinations will expedite patient care and save money.
Acknowledgments
We thank Susan E. Carozza and David Reichel for their assistance in the
kappa value analysis and Cindy Parker for her contribution to the manuscript.
We also thank Alison Russell from Illustration/Research Services for her
preparation of the figures.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  T. G. Gleason and J. E. Bavaria Trauma to the Great Vessels Card. Surg. Adult, January 1, 2008; 3(2008): 1333 - 1354. [Full Text]
|
|
|
|
 |
|
 D. M. Warshauer, R. K. Archer, C. H. Selzman, H. S. Tamaddon, and M. A. Mauro Case 115: Aortic Pseudoaneurysm from Penetrating Superior Vena Cava Stent Radiology, June 1, 2007; 243(3): 901 - 904. [Full Text] [PDF]
|
|
|
|
|
|
M. G. M. Hunink Decision Making in the Face of Uncertainty and Resource Constraints: Examples from Trauma Imaging Radiology, May 1, 2005; 235(2): 375 - 383. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Wong, M. B. Gotway, A. D. Sasson, and R. B. Jeffrey Periaortic Hematoma at Diaphragmatic Crura at Helical CT: Sign of Blunt Aortic Injury in Patients with Mediastinal Hematoma Radiology, April 1, 2004; 231(1): 185 - 189. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. J. Macura, F. M. Corl, E. K. Fishman, and D. A. Bluemke Pathogenesis in Acute Aortic Syndromes: Aortic Aneurysm Leak and Rupture and Traumatic Aortic Transection Am. J. Roentgenol., August 1, 2003; 181(2): 303 - 307. [Full Text] [PDF]
|
|
|
|
|
|
T. G. Gleason and J. E. Bavaria Trauma to Great Vessels Card. Surg. Adult, January 1, 2003; 2(2003): 1229 - 1250. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
