Clinical and radiographic indications for aortography in blunt chest trauma

Transcription

1 Clinical and radiographic indications for aortography in blunt chest trauma Harry B. Kram, M.D., David A. Wohlmuth, M.D., Paul L. Appel, M.P.A., and William C. Shoemaker, M.D., Los Angeles, Calif. To determine which clinical and radiographic findings are valuable in selecting patients with blunt chest trauma for aortography, we analyzed the medical records and admission chest radiographs of 76 consecutive victims of blunt chest trauma with suspected thoracic aortic rupture during the past 7 years. All patients were evaluated by history, physical examination, chest radiography, and aortography; a total of 70 clinical and radiographic findings were independently assessed in each patient. The following occurred with significantly greater frequency in patients with thoracic aortic rupture than in those without: history of significant hypotension (mean arterial pressure less than 80 mm Hg) (p < 0.04); the presence of upper extremity hypertension, bilateral lower extremity pulse deficits, or an initial chest tube output greater than 750 ml of blood (p < 0.05); and greater incidence of myocardial contusions, intra-abdominal injuries, and pelvic fractures compared with patients without thoracic aortic rupture (p < 0.05). Mediastinal widening (equal to or greater than 8 cm) shown on anteroposterior chest radiography occurred in all patients with thoracic aortic rupture; however, its specificity was only 10.6%. Radiographic signs that were helpful in indicating the presence of thoracic aortic rupture included paratracheal stripe greater than 5 mm, rightward deviation of the nasogastric tube or central venous pressure line, blurring of the aortic knob, and an abnormal or absent paraspinous stripe. Upper rib fractures and mediastinal to thoracic cage width ratios at any level did not increase diagnostic accuracy for thoracic aortic rupture in the present series. Six patients in the series died, two of whom had thoracic aortic rupture. Compfications from aortography occurred in eight patients (10.5%), one of whom required blood transfusion for severe groin hemorrhage. A branch-chain decision tree (clinical algorithm) was described as an approach to rapid diagnosis and treatment of patients with severe or high velocity deceleration blunt chest trauma. (J VAse SuR~ 1987;6: ) Injury to the thoracic aorta from a high-spced motor vehicle accident usually rcsuks in instantaneous death. Nevertheless, approximately 15% of persons suffering traumatic thoracic aortic rupture live long enough to receive medical attention? If prompt diagnosis is accomplished, most patients' lives can be saved2; without definitive operative therapy, less than 5% of such patients can be expected to survive 6 weeks after injury. 1 Aortography provides the definitive diagnosis of thoracic aortic rupture and is useful for optimal surgical management. Unfortunately, the time, cost, and morbidity associated with aortography make its application inappropriate for all victims of blunt chest From the Department of Surgery, King-Drew Medical Center, Los Angeles (Drs. Kram, Shoemaker, and Mr. Appel), the Department of Radiology, Harbor-UCLA Medical Center, Torrance (Dr. Wohlmuth), and the University of California, Los Angeles School of Medicine. Reprint requests: William C. Shoemaker, M.D., Department of Surgery, Drew Postgraduate Medical School, 1621 East 120th St., Los Angeles, CA trauma. Although widening of the mediastinum as seen on chest x-ray films has been considered the hallmark of thoracic aortic rupture, 3-8 a plethora of additional radiographic and clinical findings hac been suggested as indications for aortography? -23 Thus, patient selection criteria for aortography remain controversial and ill defined. Los Angeles County provides a unique patient population for the study of high-speed/deceleration chest trauma, as many of its inhabitants and visitors routinely travel on a complex freeway system. More than 60,000 injury-related motor vehicle accidents result in more than 1000 deaths each year in Los Angeles County. 24 The purpose of the present study was to evaluate clinical mad radiographic findings that may distinguish chest trauma victims with thoracic aortic rupture from those without to define indications for selecting patients for aortography. A total of 70 clinical and radiographic findings were independently assessed in each patient, including six measurements and eight ratios calculated from admission chest radiographs. In addition, the findings c-j!

2 Volume 6 Number 2 August 1987 Aortography with blunt chest trauma 169 ~lable I. Symptoms and signs in 76 patients with suspected traumatic thoracic aortic rupture With Without aortic aortic injury injury Total Symptoms No. of patients Chest pain Midscapular back pain Dyspnea Hoarseness Cough Hemoptysis Dysphagia Coma Signs No. of patients Hypotension 5* 7 I2 External chest injury Lower extremity pulse deficit 2 * 0 2 Systolic ejection murmur Upper extremity hyper- 3* 0 3 tension Paraplegia Subcutaneous emphysema Pericardia[ tamponade Hematocrit <30% Initial chest tube output 2 ~ 0 2 >750 mt NOTE: Symptoms were unobtainable in 17 patients secondary to coma. Note that the presence of unexplained hypotension, lower extremity pulse deficits, and upper extremity hypertension occurred more frequently in patients with aortic injury, than in those without aortic injury. Initial chest tube outputs greater than 750 ml were also seen with greater frequency h~ patients with thoracic aortic injury than in those without thoracic aortic injury. ~p < of this and other reviews were used to construct an,~orithm for the initial management of patients with possible thoracic aortic rupture, including indications for aortography. PATIENTS AND METHODS Patients. During the 7-year period from January 1978 to January 1985, 76 consecutive patients with suspected thoracic aortic rupture were evaluated by history, physical examination, chest radiography, and aortography. Patients suspected of having great vessel injuries other than the thoracic aorta were excluded from the study. Patients' ages ranged from 14 to 76 years (37 _+_ 17 years [mean + standard deviation]); 62 patients were male and 14 were female. The medical records of all 76 patients with suspected thoracic aortic rupture were retrospectively analyzed for the presence or absence of relevant clinical findings, which are listed in Tables I and II. Radiographic analysis. All patients underwent Table II. Clinical data in 76 patients with suspected traumatic thoracic aortic rupture With Without aortic ao~ ic injury injury Total No. of patients Age (meanyr_~ SD) Cause Motor vehicle accident Motorcycle accident Motor vehicle vs pedestrian accident Fall Thoracic injuries Myocardial contusion 5 ~ Rib fracture Flail chest Sternal fracture Tracheobronchial injury Diaphragm injury Esophageal injury Thoracic spine fracture Other injuries Pelvic fracture 3 ~ 6 9 Intra-abdominal injury 4* 9 13 Extremity fracture Skull fracture Cervical spine fracture Lumbar spine fracture No. of deaths (%) 2 (20.0) 4 (6.1) 6 (7.9) Complications from aortogra- 0 (0) 8 (12.I) 8 (10.5) phy (%) *Note that myocardial contusions, pelvic fractures, and intraabdominal injuries occurred more frequently in patients with aortic injury than in those without aortic injury (p < 0.05). supine or slight head-up anteroposterior chest radiography in the emergency room, usually obtained with a source-image distance of 100 cm. The initial chest radiographs of 64 patients were retrospectively analyzed for relevant findings; 12 films could not be retrieved for review, having been transferred to another institution or lost. All of the latter radiographs were from patients subsequently found not to have thoracic aortic rupture. Radiographic analysis was performed by one of us (D. A. W.), who was unaware of the patient's name, age, history, physical examination, previous radiographic analysis, results ofaortography, eventual outcome, and of the total number of patients with and without thoracic aortic rupture. Specific radiographic findings were sought by means of a checklist consisting of 20 variables, which are listed in Table III. Six measurements were performed on each chest radiograph evaluated. These included the mediastinal width at the level of the aortic knob, the greatest mediastinal width, the width of the cardiac silhouette, as well as the width of the thoracic cage at the

3 170 Kram et al. 7oumal of VASCULAR SURGERY AI2 E o v -r' I- Q --8.J Z T~ laj O [] NORMAL MEDIASTINAI WIDTH ~WIDE MEOIASTINUM WITHOUT AORTIC INJURY [] WIDE MEDIASTINUM WITH AORTIC INJURY T AT AORTIC KNOB /.//./././ 2/ 4, / / / AT GREATEST TRANSVERSE DIMENSION Fig. 1. Mediastinal widths, as measured on initial chest radiography, in patients with suspected thoracic aortic rupture. Bars represent mean _+ SD. Group I patients had normal mediastinal widths (less than 8 cm) and no aortic injuries (N = 7), group II patients had mediastinal widening and no aortic injuries (N = 47), and group III patients had mediastinal widening and thoracic aortic rupture (N = 10). Note that although both group II and III patients had significantly greater mediastinai widths than patients in group I (p < 0.05 by analysis of variance [ANOVA]), there was no significant difference in mcdiastinal widths between patients in group II vs. group III. Table III. Radiographic signs present on initial chest x-ray films in patients with suspected traumatic thoracic aortic rupture With Without aortic aortic injury injury Total No. of patients Mediastinal width >8 cm 10 ~ Paratracheal stripe >5 mm 6? 4 10 Nasogastric tube deviated to right 2/2t 0/5 2/7 CVP line deviated to right 2/4 + 0/11 2/i5 Blurring of aortic knob Abnormal or absent paraspinous stripe 67 i4 20 Trachea deviated to right i Opacification of AP space Apical cap Depression of left main stem bronchus >40 degrees Hemothorax Pneumothorax :~ Pneumomediastinum 0 i 1 Lung contusion Clavicular fracture Sternal fracture First rib fracture 2 I0 12 Second rib fracture 1 10 i i Multiple rib fractures Bilateral rib fractures *Mediastinal widening greater than 8 cm seen on chest x-ray films of all patients with aortic injuf. tp < 0.05 for the differences in frequency, of radiographic signs between patients with and without aortic injury. level of the aortic knob, at its maximum, and at the diaphragm. These measurements were entered into a Hcwlctt-Packard 1000 computer and eight ratios were calculated for each radiograph, which are listed in Table IV. Statistical analysis. The relative frequency of clinical and radiographic findings in patients with and without thoracic aortic rupture were compared by analyzing the difference between two proportions. 2s Radiographic measurements and ratios were stratified into three groups: patients with a mcdiastinal width less than 8 cm (group I), patients with a mediastinal width greater than or equal to 8 cm but without thoracic aortic rupture (group II), and patients with a mediastinal width greater than 8 cm and thoracic aortic rupture (group III). The measurements and ratios of each group were compared by means of analysis of variance (ANOVA) and Neuman-Kculs test for significance. Statistical significance was assumed at the 95% confidence level. The sensitivity, specificity, false positive rate, false negative rate, and predictive value of diagnosing thoracic aortic rupture with the use of a mediastinal width greater than or equal to 8 cm were calculated by standard formulas. 26'27 RESULTS Clinical findings. Symptoms were unobtainable in 17 patients (22.4%) as a result of coma caused by closed head injury. Of the 59 patients in whom symptoms could be elicited, 37 (62.7%) complained of chest pain; only seven of these patients (11.9%) had thoracic aortic rupture (Table I). However, mid ~ scapular back pain was found in 12 conscious patients (20.3%), three of whom had thoracic aortic rupture. Of the nine patients with complaints of midscapular back pain without thoracic aortic rupture, eight had thoracic spine fractures; no patients with thoracic aortic rupture had a thoracic spine fracture. Thus, four conscious patients (6.8%) in the present series had midscapular back pain from causes other than thoracic spine fracture; thoracic aortic rupture was discovered subsequently in three of these patients. A history of significant hypotension (mean arterial pressure less than 80 mm Hg) occurred with significantly greater frequency in patients with thoracic aortic rupture than in those without thoracic aortic rupture (p < 0.04) (Table I). Hypotension occurred in 12 patients (15.8%), five of whom (41.7%) subsequently were found to have thoracic aortic rupture.

4 Volume 6 Number 2 August 1987 Aortography with bluet chest trauma 171 1atfle lv. Ratios of mediastinal to thoracic width and mediastinal to cardiac width at level of aortic knob, diaphragm, and at the greatest thoracic width Level Group I Group H Group III Mediastinal width (at aortic knob)/thoracic width Aortic knob 0.30 ± ± ± 0.07 Diaphragm 0.26 ± ± ± 0.05 Greatest thoracic width 0.26 _ ± ± 0.05 Mediastinal width (greatest)/thoracic width Aortic knob 0.30 ± ± ± 0.06 Diaphragm 0.26 ± _ _ Greatest thoracic width 0.26 ± ± ± 0.05 Mediastinal/cardiac width Aortic knob 0.51 ± ± 0.08 Greatest mediastinal width ± ± 0.08 Group I = patients without mediastinal widening; Group II = patients with mediastinal widening without aortic injury; Group III = patients with mediastinal widening and aortic injury. NOTE: Values listed are mean _+ 1 standard deviation. Note that there was no significant difference of any ratio between groups II and III. The pseudocoarctation syndrome, defined as the triad of upper extremity hypertension, lower extremity pulse deficit, and mediastinal widening found on chest radiography, s was not seen in any patients, with or without thoracic aortic rupturc. Howevcr, cithcr uppcr extremity hypertension or the presence of bilateral lower extremity pulse deficits occurrcd with greater frequency in patients with thoracic aortic rupture than in those without (p < 0.01). External signs of chest trauma, including abrasions and ecchymoses, werc scen in 21 patients (27.6%) (Table I); four of these patients had thoracic aortic rupture. Of morc importance, 6 of 10 patients with thoracic aortic rupture had no external evidence of chest trauma. Only two patients in the series had initial chest tubc outputs greater than 750 ml; both of these pat' 'nts subsequently were found to have thoracic aortic rupture. There was no significant difference in patients' age or origin of injuu in thosc with and without thoracic aortic rupture (Table II). However, myocardial contusions, pelvic fractures, and intraabdominal injuries occurred more frequently in patients with thoracic aortic rupture than in those without (p < 0.05). Other associated injuries did not occur with a significantly grcatcr frequency in paticnts with thoracic aortic rupture. Radiographic findings. Mcdiastinal widening greater than 8 cm was present on the radiographs of all 10 patients with thoracic aortic rupture (Tablc Ili); no other radiographic sign was seen on the radiographs of all patients who had thoracic aortic rupture. Thus mcdiastinal widening had a sensitivity of 100% and a false negative rate of zero in the present series. Mediastinal widening was also seen on the radiographs of 54 patients without thoracic aortic rupture. Mediastinal hemorrhage seconda~ to thoracic vertebral fracture was determined to be the cause of the mediastinal widening in eight of these patients. Thus mediastinal widening was only 10.6% specific in diagnosing thoracic aortic rupture with a false positive rate of 89.4% and a positive predictive value of 14.5%. Of the 18 other radiographic signs evaluated, five were seen with significantly greater frequency on radiographs of patients with thoracic aortic rupture compared with those without (p < 0.05); these included right paratracheal stripe greater than 5 mm, deviation of the nasogastric tube to the right, deviation of the central venous pressure line to the right, blurring of the aortic knob, and abnormal or absent paraspinous stripe (Table III). Rightward tracheal deviation was seen on radiographs of i i patients, eight of whom did not have thoracic aortic rupture. Seven of eight of the latter patients were more than 55 years of age and had varying degrees of aortic atherosclerosis and tortuosity. Many patients had first, second, multiple, or bilateral rib fractures seen on their initial chest radiograph. However, the presence of rib fractures in no way helped to differentiate patients with thoracic aortic rupture from those without (Table III). Analysis of radiographic measurements and ratios. No radiographic measurement or ratio was as sensitive as the mediastinal width in diagnosing patients with thoracic aortic rupture (Fig. 1 and Table IV). Even patients without thoracic aortic rupture usually had mediastinal to thoracic cage width ratios greater than 0.25, thereby decreasing specificity for this variable. None of the seven patients with

5 172 Kram et al. ;oumal of VASCULMt~ SURGERY Fig. 2. A, Admission chest radiograph of a 36-year-old man after a high-speed automobile accident. Note mcdiastinal widening (10. i cm), deviation of the nasogastric tube to the right (lave arrow), deviation of the central venous pressure line to the right (small arrow), deviation of the trachea to the right (asterisk), blurring of the aortic knob, opacification of the aortopulmonary space, and left hemothorax. B, Aortogram obtained from patient shown in A. A left chest robe has been placed. Note aortic injury located at the levcl of the left subclavian artery. mediastinal widths less than 8 cm had thoracic aortic rupture and all 10 patients with rupture had mediastinal widths greater than 8 cm. The mean mediastinal width at the level of the aortic knob in patients with thoracic aortic rupture was cm (range 8.4 to 12.0 cm). Morbidity and mortality. Complications caused by aortography occurred in eight patients (10.5%); included were seven patients with groin hematomas and one with an intimal flap of the ascending aortic arch. Only one patient with a groin hematoma required blood transfusion to maintain a hematocrit greater than 30%. All eight aortographic-rclatcd complications occurred in patients without thoracic aortic rupture. Of the 10 patients with thoracic aortic rupture, two (20.0%) died during operation (Table II); one death was due to severe cardiac injury and the other after hypotcnsivc shock. Of the remaining 66 patients without thoracic aortic rupture, only four (6.1%)

6 Volume 6 Number 2 August 1987 Aortography with blunt chest trauma 173 ~i ~ ~i ) v ~ 9 Fig. 3. A, Admission chest radiograph of a 44-year-old woman after a high-speed automobile accident. Note mediastinal widening (12.0 cm), deviation of the nasogastric robe to the right (large arrow), widened paratracheal stripe (small arrow), deviation of the trachea to the right (asterisk), blurring of the aortic knob, opacification of the aortoputmonary space, abnormal paraspinous stripe, and bilateral hemothoraces. B, Aortogram obtained from patient shown in A. Note aortic injury located near level of the ligamenmm arteriosum; the intraluminal filling defect of the descending aorta proved to represent a large blood clot. died. Two of these paticnts dicd of head injuries, onc died of hypotensive shock, and one late death occurred because of sepsis. D~SCUSSION According to Sailer, 28 the first case of thoracic aortic aneurysm thought to be due to trauma was reported in 1556 by Vesalius. In today's era of highspeed motor vehicles and complex freeway systems, an estimated 7500 motor vehicle accident victims die each year as a result of thoracic aortic rupture. 29 The key to successful management lies in the principles of prompt diagnosis and surgical repair. 2 Data of the present series indicate that a mediastinal width greater than 8 cm seen on standard anteroposterior chest radiography is the most sensitive indicator of thoracic aortic rupture (Table III). However, the presence of thoracic aortic rupture in patients without mediastinal widening has been reported. 17,3 As pointed out by MacKenzie, 31 the thoracic aorta is covered with a periadventitial sheath and a similar sheath surrounds the large mediastinal veins; a common vascular sheath encloses both of these sheaths. Thus, for mediastinal widening from thoracic aortic rupture to occur, blood must penetrate or distort both periadventitial sheaths. In addition to mediastinal widening, other radiographic findings that were helpful in distinguishing patients with thoracic aortic rupture from those without included a right paratracheal stripe greater than 5 mm, rightward deviation of the nasogastric tube or central venous pressure line, blurring of the aortic knob, and an abnormal or absent paraspinous stripe (Figs. 2 through 4). Similar to reports by other investigators, 22'23 we found no increase in the incidence of thoracic aortic rupture in patients with upper rib fractures. However, unlike the latter reports, calculation of mediastinal-to-thoracic width ratios did not increase diagnostic accuracy for thoracic aortic rupture in the present series. The presence of unexplained hypotension appeared to be an important sign of thoracic aortic rupture. Similarly, complaints of midscapular back pain in the absence of thoracic spine fracture should

7 174 Kram et al. Journal of VASCULAR SURGERY Fig. 4. A, Admission chest radiograph of a 20-year-old man after an automobile accident. Note that in this patient, mediastinal widening (10.6 cm) and blurring of the aortic knob were the only radiographic signs of thoracic aortic rupture that were present. Physical signs and symptoms diagnostic of thoracic aortic rupture were a!so absent. B, Aortogram obtained from patient shown in A. Note aortic injury located at level of the left subclavian artery. alert the physician as to a significant possibility of thoracic aortic rupture. This does not mean that patients with mediastinal widening, midscapular back pain, and thoracic spine fracture should be excluded from undergoing aortography because thoracic aortic rupture may occur in association with thoracic spine fracture. Similarly, tracheal deviation in the elderly patient who has blunt chest trauma with a widened mediastinum seen on chest radiography may be due to an atherosclerotic, tortuous aorta. However, aortography is indicated in these patients because of the greater than 90% mortality rate for untreated victims of thoracic aortic rupture. In 1914, Copeland 32 reported a patient with thoracic aortic rupture and no external signs of injury. In the present series, 6 of 10 patients with thoracic aortic rupture had no signs of external chest trauma. Similarly, chest radiography may be normal despite complete aortic disruption. 17'3 Ruptures of the ascending aorta do not usually produce mediastinal widening because the ascending aorta is completely covered by pericardium. Rupture of the ascending aorta usually occurs above the aortic valve, resulting in hemopericardium and cardiac tamponade rather than mediastinal widening? 3 Because failure to recognize the presence of thoracic aortic rupture usually leads to death of the patient, it has been recommended that a history of severe blunt chest trauma is enough indication for immediate aortograph);ee~ However, this may not be feasible in hospitals that receive large numbers of blunt chest trauma victims. On the basis of data gathered in the present series and a review of the literature, we constructed an algorithm for the diagnosis of thoracic aortic rupture in patients with a history of severe or high velocity/deceleration chest trauma (Fig. 5). The algorithm is designed to maximize sensitivity with the use of the appropriate findings to provide rational therapy as soon as possible. The low incidence of significant complications secondary to aortography and the grave nature of undiagnosed thoracic aortic rupture justifies this aggressive diagnostic approach. Until newer and better noninvasive modalities for the diagnosis of thoracic aortic rupture are developed, the physician's high index of suspicion combined with a detailed history, physical examination, and chest radiography are of paramount importance in deciding whether aortography is indicated. ~4,;

8 Volume 6 Number 2 August 1987 Aortography with blunt chest trauma 175 1!RAPID INITIAL ~)(RE 1 ASSESSMENT[ SPIPATOR Y DISTRESS\YES... ~o '2 "[RESOS:'T~TEI No[. ]CHESTX-RAY, HISTORY,PHYS.EXAM., LAGS] i 1 YES CHESTTUB TP YES \... NOl" E... ORAX~cH~NSTETRTBE ~ ~ 750[!~U U T ieicwyi, o?e2 G?;Cii;iiT },o ~WIDENEO MEDIASTINUM,WIDENEDPARATRACHEALSTRIPE, ~ 1 HYPOTENSION.UNEXPLAINEDUPPEREXTREMITY / HYPERTENSION OR LOWER EXTREMITY PULSE DEFICfTS N01 INSERTNASOGASTRICTUBE C~P LINE, 1... \,ES ASOGASTRIC TUBE, CVP LINE... OR'OHT 7 N l ] AORTOGR ~,PH y [ 1,01 CONTINUE WORK-UP FOROTHERINJURIES Fig. 5. Algorithm for the diagnosis of traumatic thoracic aortic rupture. Entrance criteria include all patients with severe or high velocity/deceleration blunt chest trauma. though other noninvasive modalitics for the diagnosis of central and peripheral arterial trauma have had promising initial results, ss,36 chest radiography remains the most practical noninvasive screening test for thoracic aortic rupture. With the described algorithm for patients with high velocity/deceleration chest trauma, we hope that more patients with thoracic aortic rupture will be diagnosed and treated ear- Lr, ultimately leading to an increased survival rate. REFERENCES 1. Parmley LF, Mattingly TW, Manion WC, Jahnke EJ. Nonpenetrating traumatic injury of the aorta. Circulation 1958; 17: Bodily K, Perry JF, Strate RG, Fischer RP. The salvageability of patients with posttraumatic rupture of the descending thoracic aorta in a primary trauma center. J Trauma 1977;17: Rich NM, Spencer FC. Vascular trauma. Philadelphia: WB Saunders Co, 1978: Kirsh MM, Crane, ID, Kahn DR, et al. Roentgenographic evaluation of traumatic rupture of the aorta. Surg Gynccol Obstct 1970;13i: Symbas PN, Tyras DH, Ware RE, Diorio DA. Traumatic rupture of the aorta. Ann Surg 1973;178:6-I2. 6. Vasko IS, Raess DH, Williams TE, et al. Nonpenetrating trauma to the thoracic aorta. Surgery 1977;82: Gundry SR, Williams S, Burney RE, et al Indications for aortography in blunt thoracic trauma: a reassessment. J Trauma 1982;22: Gundry SR, Burney RE, MacKenzie JR, et al. Assessment of mcdiastinal widening associated with traumatic rupture of the aorta. J Trauma 1983;23: Holmes TW, Nertervilte RE. Complications of first rib fracture including one case each of tracheoesophagcal fistula and aortic arch aneurysm. J Thorac Cardiovasc Sure 1956;32: Richardson JD, McElrein RB, Trinkle IK. First rib fracture: a hallmark of severe trauma. Ann Surg 1975;181: McIliduffJB, Foster ED, Alley RD. Tramnatic aortic rupture: an additional roentgenographic sign. Ann Thorac Surg 1972;24: Tisnado J, Tsal FY, Als A, Roach JF. A new radiographic sign of acute rupture of the thoracic aorta: displacement of the nasogastric tube to the right. Radiology 1977;125: Gerlock AJ, Muhletaler CA, Coulan CM, Hayes PT. Traumatic aortic aneurysm: validity of esophageal tube displacement sign. AJR 1980;135: Peters DR, Gamsu G. Displacement of the right paraspinous interface: a radiographic sign of acute traumatic rupture of the thoracic aorta. Radiology 1980;134: Simeone JF, Deren MM, Cagle F. The value of the left apical cap in the diagnosis of aortic rupture. Radiology 1981; 139: Marsh DG, Sturm IT. Traumatic aortic rupture: roentgenographic indications for angiography. Ann Thorac Surg 1976;21: Sturm IT, Marsh DG, Bodily KC. Ruptured thoracic aorta: evolving radiological concepts. Surgery 1971;85: Burney RE, Gundry SIR., MacKenzie JR, et al. Chest roentgenograms in diagnosis of traumatic rupture of the aorta. Observer variations in interpretation. Chest 1984;85:605-9.

9 176 Kram et al. 7oumai of VASCULAR SURGERY 19. Woodring JH, Pulmano CM, Stevens RK. The right paratracheal stripe in blunt chest trauma. Diagn Radiol 1982; 143: Woodring JH, Lob FK, Kryscio RJ. Mediastinal hemorrhage: an evaluation of radiographic manifestations. Diagn Radiol 1984;151: Milne ENC, Imray TJ, Pistoles M, et al. The vascular pedicle and the vena azygous. Diagn Radiol 1984;153: Seltzer SE, D'Orsi C, Kirshner R, DeWeese IA. Traumatic aortic rupture: plain radiographic findings. AJR 1981;137: Kirshner R, Seltzer S, D'Orsi C, DeWeese JA. Upper rib fractures and mediastinal widening: indications for aortography. Ann Thorac Surg 1983;35: California Highway Patrol. Annual report of fatal and injury motor vehicle traffic accidents Bruning JL, Kintz BL, eds. Computational handbook of statistics, 2nd ed. Glenville: Scott, Foresman & Co, 1977: Vecchio TJ. Predictive value of a single diagnostic test in unselected populations. N Engl J Med 1966;274: McNeil BJ, Keeler E, Adelstein SJ. Primer on certain elements of medical decision making. N Engl J Med 1975;293: Sailer S. Dissecting aneurysm of the aorta. Arch Pa~.ol 1942;23: Jackson DH. Of TRAs and ROCs. (Editorial). Chest 1984;85: Wilson RF, Arbulu A, Bassett JS, Walt AJ. Acute mediastinal widening following blunt chest trauma. Arch Surg 1972; 104: MacKenzie JR. Discussion of acute treatment of traumatic aortic rupture. J Trauma 1970;11: Copeland GG. Traumatic rupture of the healthy aorta without external signs of the cause of death. JAMA 1914;63: Attar S~ Ayella RJ, McLaugblin JS. The widened mediastinum in trauma. Ann Thorac Surg 1972;13: Kirsh MM, Behrendt DM, Orringer MB, et al. The treatment of acute traumatic rupture of the aorta: a 10-year experience. Ann Surg 1976;184: Egan TJ, Neiman HL, Herman RJ, et al. Computed tomog~ raphy in the diagnosis of aortic aneurysm, dissection or traumatic injury. Radiology 1980;136: Kram HB, Shoemaker WC. Diagnosis of major periphe '~ arterial trauma by transcutaneous oxygen monitoring. Am J Surg 1984;147: