Pediatric Radiology

Originals

9 Springer-Verlag 1993 Pediatr Radiol (1993) 23:1-5

CT of blunt chest trauma in children D . M a n s o n I, P. S. B a b y n I, S. Palder 2, K. B e r g m a n 2

1Departments of Diagnostic Imaging, Hospital for Sick Children, Toronto, Ontario, Canada 2 General Surgery, Hospital for Sick Children, Toronto, Ontario, Canada Received: 14 April 1992/Accepted: 17 November 1992

Abstract. While trauma is still the leading cause of death

in the pediatric age range, it is surprising how little the CT appearances of pediatric chest injury have b e e n investigated in the literature. We have reviewed the CT findings of blunt chest trauma in 44 children for w h o m chest CT examinations were requested to investigate the extent of intrathoracic injury. We noted a propensity for pulmonary contusions to be located posteriorly or posteromedially, and for t h e m to be anatomically nonsegmental and crescentic in shape. This is possibly attributable to the relatively compliant anterior chest wall in children. The CT appearances of other m a j o r thoracic injuries are described, including pulmonary lacerations, p n e u m o t h o races, malpositioned chest tubes, mediastinal hematomas, aortic injury, tracheobronchial injury, hemopericardium, and spinal injuries with paraspinal fluid collections. Children demonstrating findings incidental to the actual injury yet important to the subsequent therapy are also presented. We conclude that, in the event of clinically significant blunt chest trauma, the single supine chest examination in the trauma r o o m is insufficient to adequately identify the extent of intrathoracic injury. With the exception of concern for aortic injury for which aortography is indicated, a dynamically enhanced CT scan of the thorax should be p e r f o r m e d as clinically significant findings m a y result in altered therapy.

Trauma remains the leading cause of death in childhood and a significant cause of serious morbidity, especially in industrialized countries where m o t o r vehicle accidents are commonplace. Mortality rates in children ranging from 7-25 % [1, 2] have b e e n attributed to intrathoracic injury alone in some series. As well, thoracic injury in conjunction with other organ systems injury m a y increase mortality rates to high as 35 % [3]. Evaluation of degree of

Correspondence to: Dr. D. Manson, Department of Diagnostic Imaging, Hospital for Sick Children, 555 University Ave., Toronto, Ontario, Canada, MSG-IX8

intrathoracic trauma is frequently limited to the chest radiograph, usually p e r f o r m e d using a single supine A P portable technique in the trauma room, not u n c o m m o n l y with extensive supportive and resuscitative artifacts which cannot be m o v e d although they m a y obscure underlying pathology. This suboptimal film is frequently the radiographic evidence upon which significant therapeutic decisions are made. While use of CT scanning to evaluate extent of intra-cerebral and intra-abdominal injury is well established, very little has b e e n published concerning the CT appearances of chest injury from blunt chest trauma in children. We reviewed all of the chest CT scans p e r f o r m e d at our institution on trauma patients over a 10 year period in order to observe what patterns of thoracic injury may exist and how the CT scan may contribute to the overall m a n a g e m e n t of the traumatized child. Patients and m e t h o d s

Forty-four children (M 28, F 16) aged 6 months to 16 years (mean 8.2 y) who had undergone CT scans of the chest for blunt trauma over a 10 year period were included in this study. The CT examinations and corresponding presenting chest X-rays were retrospectively reviewed by two radiologists blinded to the child's history and clinical course. The chest X-ray was evaluated first and findings were recorded. The chest CT scan was then reviewed. These findings were then matched to the patient's history and clinical course from the hospital record. CT scans were performed on either a GE 8800 or GE 9800 (General Electric Medical Systems, Milwaukee) CT scanner. Twenty-five scans were performed as standard dynamically enhanced complete thoracic examinations using 1 cm contiguous slices, while the remaining 19 scans used a number of different techniques designed to optimize imaging of mediastinal structures. This most commonly consisted of 5-10 mm contiguous slices through the major vessels of the mediastinum with dynamic intravenous contrast enhancement. The CT was performed within 6 h of the accident in 27 children, while the chest radiograph was performed within 6 h of the CT scan in 32 children. Abdominal CT examinations as well as chest CT examinations were performed in 25/44 children. Pulmonary contusions are defined as areas of alveolar opacity in association with recent trauma [3] in an otherwise healthy child. Contusions were considered major if they involved _>50 % of a lung segment, or if there were multiple smaller contusions within a single lobe, as defined by Sivit, et al. [2].

Fig.1. CT scans of the thorax displayed at lung windows in two different children demonstrating the typical crescentic shape and posterior/posteromedial location of parenchymal contusions (arrows) Fig.2. Typical appearances of parenchymallacerationsdemonstrating a "cystic" parenchymal air collection with an air41uid level Fig.3. Chest tubes positioned along posterior chest wall with interposed lung between the tubes (arrows) and significantresidual anterior pneumothoraces Fig.4. CT demonstrating discontinuity (arrow) in right posterolateral wall of trachea adjacent to endotracheal tube with extensive

mediastinal and subcutaneous emphysema. This tear was confirmed at surgery Fig. 5. CI" demonstrating clinically unsuspected pericardial hematoma Fig.6. CT myelogram showing a dural tear with leakage of contrast into the mediastinum and extrapleural space, explaining the wide mediastinum on the inital chest radiograph. Large arrow points to a residual pocket of intrathecal contrast displaced by the intraspinal hematoma

3 Pulmonary lacerations are defined a intraparenchymal air collections [3] frequently containing air fluid levels [6].

Results

Parenchymal abnormalities of the lung were the most common abnormalities detected on CT examination, taking the form of either pulmonary contusions or lacerations. Pulmonary contusions were demonstrated in 28/44 children examined. Pulmonary contusions were anatomically located either posteriorly or posteromedially in the lung in 26/28 children. As well, the contusions did not fit within the segmental anatomy of the lung in 22/28 children, and proved to have a relatively crescentic shape in 15/28 children, paralleling the thoracic cage in a subpleural location (Fig. 1). Concommittant abdominal visceral injury was often seen in conjunction with major pulmonary contusions. In 11/16 children with major right lower and/or right middle lobe contusions, underlying liver injury (3 liver contusions, 4 liver lacerations, 3 subcapsular hematomas) or right kidney injury (1 contusion) was demonstrated. Conversely, 10/11 children with liver or right kidney injury also had major right lower and/or right middle lobe contusions. As well, of the 5 children who had major RLL and/or RML contusions with no abdominal injury, 4/5 had other significant injuries (3 midthoracic spine injury, 1 retroperitoneal and i mediastinal hematoma). However, only 4/11 children with major LLL and/or lingular contusion demonstrated splenic or left kidney inj ury. Pulmonary lacerations represented the next most common parenchymal injury, seen in six children whose CT scans demonstrated nine separate lacerations (Fig.2). None of these were appreciated on the initial supine chest X-ray, an observation noted by others [3]. Four of the lacerations were situated paravertebrally in the lower hemithorax. Sixteen children demonstrated pneumothoraces on CT. On reviewing the initial chest X-ray, pneumohtoraces were either missed or underestimated in 10/16 children when compared with the CT scan. In 12/16 patients with pneumothoraces, chest tubes were incorrectly situated along the posterior pleural margin (Fig. 3) with interposed lung between the chest tube and the significant residual anterior pneumothorax. One child had a persistent tension pneumothorax on CT despite placement of a chest tube, probably because of poor positioning of the chest tube. Mediastinal injuries included one child with aortic injury, one child with tracheobronchial injury, and ten children with mediastinal fluid collections. The one child who experienced an aortic injury has been reported previously [5]. This 12 year-old child was referred to us for six days of persistent mediastinal widening on the chest X-ray. The CT performed on admission demonstrated focal widening of the aorta with a false lumen, and the subsequent angiogram demonstrated a pseudoaneurysm at the aortic isthmus. One child demonstrated a tracheal tear (Fig. 4). The intial chest X-ray demonstrated extensive subcutaneous

and mediastinal emphysema with large pneumothoraces. The CT demonstrated discontinuity of the right posterolateral wall of the trachea several centimetres above the carina. At surgery, a 1 cm tear was seen in the trachea at this site and was oversewn without sequelae. Ten children demonstrated mediastinal fluid collections. Four of these were pericardial hemorrhages, four were paraspinal and related to spinal injury, and two were of unknown origin. In three of the four children with pericardial hematomas, the diagnosis was initially made by the CT findings (Fig.5), while one was clinically suspected and drained before the CT study. Although the child in Fig. 5 did have periportal low density in the liver on her abdominal CT scan, none of the children demonstrated the triad of pericardial hemorrhage, distended central veins and periportal low density of the liver suggesting tamponade as described by Goldstein, et al. [7]. Four children demonstrated mediastinal/paravertebral fluid collections related to spinal injury. Two of these four demonstrated significant neurologic deficits related to thoracic spine injury, and CT myelography was performed as well as chest CT. In these children, intrathecally administered contrast agent leaked out of the subarachnoid space into the posterior mediastinal and extrapleural spaces (Fig. 6), explaining the widened mediastinum and poor definition of the aortic contour seen on the initial chest X-rays. The other two children did not have CT myelography performed as they had no neurologic symptoms and their posterior mediastinal fluid collections were most likely hematomas related to adjacent vertebral body fractures. Four CT examinations demonstrated findings which were incidental to the history of trauma. One child demonstrated in ill-defined right apical opacity on the initial chest radiograph which subsequently proved to be an azygous lobe fissure. A second child demonstrated apparent widening of the mediastinum on the initial chest radiograph which proved to be a normally located right aortic arch on the CT examination. A third child with known Marfan's syndrome and mediastinal widening underwent CT examination to exclude a mediastinal hematoma and to help confirm the integrity of the tortuous aorta. A fourth child presented with blunt trauma and a chest Xray pattern suggestive of a large pulmonary contusion and effusion. CT examination demonstrated the large effusion but also demonstrated a large partially calcified paravertebral tumour which at biopsy proved to be a benign ganglioneuroma.

Discussion

CT examination of the abdomen in the clinical setting of blunt abdominal trauma is a well accepted procedure, even though comtemporary surgical management of the vast majority of abdominal injuries is expectant. Very little attention has been paid to use of CT of the chest in children with blunt chest trauma. Sivit, et al. [2] detailed the clinical importance of chest injuries in children that can be diagnosed on abdominal CT scans when looking at the lower lung sections. In that study, children with chest injuries had

4 lower mean trauma scores than those without, and were more clinically unstable. That study also demonstrated mortality rates of 1.3 % in children with no chest injury, 10.8 % in children with significant unilateral chest injury, and 40 % with significant bilateral lung or mediastinal injury, as determined by the CT scan. As well, Wagner, et al. [8] reviewed their experience in adults, and concluded that patients with significant lung injury as seen on the CT scan had a significantlyhigher requirement for ventilatory support. Rhea, et al. [9] also described a population of 65 adults who demonstrated abnormalities in the lower chest on abdominal CT scans. Seven of these patients had abnormalities demonstrable by CT alone that required some form of therapy. In an effort to clarify the clinical significance of the parenchymal contusions seen in our study, we attempted to correlate the degree of severity of parenchymal injury with ventilatory requirements. However, most (80 % ) of our population who received mechanical ventilation did so because of CNS injury. The need to hyperventilate these children therefore resulted in uninterpretable statistics concerning parenchymal injury. The appearances of pulmonary parenchymal injury in children are probably best explained by the natural compfiance and plasticity of the pediatric thorax, which results in a scenario for intrathoracic injury different from the adult. While rib factures are common in adult injury, significant intrathoracic injury is frequently seen in children with surprisingly little thoracic cage injury [1, 10, 13, 16, 18]. This pliability probably accounts for the propensity for lung contusions to be located posteriorly and espedally posteromedially, as the more compliant anterior chest is freer to move, compressing the more posterior aspect of the lungs against the less complaint posterior ribs and vertebral column. This would also explain the crescentic appearances of many contusions, as the blunt pressure is distributed along the less mobile regions of lung. As well, since blunt trauma is frequently distributed over a large area, it is not surprising that the underlying lung injury would not conform to segmental lung anatomy. These observations may be helpful when trying to differentiate contusion from aspiration. The association of right basilar lung injury with right sided abdominal injuries is not surprising in the perspective of blunt trauma. This partially explains why other authors who have found similar findings [2, 9] have recommended that routine scans of the lower chest be performed as part of an abdominal CT examination in the case of blunt trauma. A prospective study is needed to verify if left sided injuries do follow a different pattern than right sided injuries as we have seen. The finding of a paravertebral "cystic" lesion with an air-fluid level in the traumatized patient has been the topic of several papers [3, 11, 17]. We have included these as parenchymal lacerations, however whether these lesions represent true lacerations [3], air trapped in the medial pleural space, air trapped in the inferior pulmonary ligament [17], or air trapped in the posterior mediastinum [11] is still a matter of academic debate, as these generally resolve without therapy as long as adjacent organ injury such as an esophageal tear has been excluded. This study suffers from several deficiencies because of its retrospective nature. For example, the relative in-

cidence of the types of injury seen in this study may appear somewhat surprising, such as the high incidence of pericardial hematomas. Relative rates of injuries of various types cannot be determined based on this study, however, as the inclusion criteria were biased towards those children with blunt trauma severe enough to warrant a chest CT. In this institution chest CT is not performed as part of a protocol but only performed at the discretion of the surgeon responsible for the patient. Nevertheless, while it has been stated cardiac trauma occurs infrequently in children [6, 16], recent studies [14] have suggested that cardiac contusion may be more frequent than previously thought, possible explaining the high incidence of pericardial hematomas. While we were able to appreciate these blood collections by CT, echocardiography remains the procedure of choice if a pericardial hematoma is suspected clinically. The adult literature has been somewhat controversial for some time now concerning the use of CT in diagnosing major vascular, or more specifically, aortic injury in blunt chest trauma. Little has been published in the pediatric literature other than the recent article by Spouge, et al. [5], suggesting that a large proportion of CT scans in children for this purpose will be non-diagnostic. We found it quite useful in our single case, however, as noted, this child was studied some time after his injury when he was stable and cooperative. It should be noted that the purpose of our review was to evaluate the use of CT in delineating intrathoracic injury other than aortic injury. When clinical or radiographic concern for aortic injury exists, the procedure of choice remains aortography. The question as to what types of chest injury are severe enough to warrant a chest CT remains relatively unanswered by this study because of its retrospective nature and because of the subjectivity of the inclusion criteria. While one is tempted to suggest that any child with an abnormal chest X-ray in the setting of blunt chest trauma should undergo a chest CT once aortic injury has been excluded, this needs to be verified with a prospective study. Nevertheless, this review does provide evidence that the use of CT in the child after an episode of blunt chest trauma can provide extremely useful information to the clinicians. The extent of lung injury can be estimated and since extent of lung injury has been shown to correlate with the subsequent clinical course [8, 9], this should prove helpful in anticipating the need for aggressive ventilatory support. The size of pneumothoraces and relative positions of chest tubes can be critical in children who may subsequently require positive pressure ventilation, either in the operating room or in the ICU, for pulmonary or CNS complications. Evaluation of mediastinal widening in the child can be difficult, as a supine chest X-ray in a child with a normal thymus can produce spurious mediastinal widening with poor definition of the aortic arch. Upright positioning for a chest X-ray may be impossible for some children, such as those with pelvic or spinal fractures, however the CT can differentiate widening caused by a normal thymus from a mediastinal hematoma. Mediastinal widening may also be seen in children who have posterior mediastinal fluid collections, most of which are unrelated to aortic injury in the pediatric population. Spinal frac-

tures produce paravertrebral h e m a t o m a s or m a y even be associated with dural tears, especially if there are neurologic symptoms, and CT myelography or M R I may be indicated to delineate the extent of injury. We have also seen how CT m a y be the first clue to the presence of blood in the mediastinum with all of its potential implications for major vascular injury, as well as blood in the pericardial sac, with its implications for potential sudden cardiac decompensation. While CT scanning m a y not be the m e t h o d of choice for specific evaluation of pericardial h e m a t o m a s or spinal/dural injury, it is reassuring to know these can be appreciated even if they are clinically unsuspected. As well, we have seen a surprising n u m b e r of CT examinations which revealed findings incidental to the history of trauma. We conclude that the t r a u m a r o o m chest X-ray is only a gross screening examination when evaluating intrathoracic pathology. The child's compliant thoracic cage m a y not demonstrate the fractures which, in adults, serve as clues to the severity of the impacting trauma. In the presence of significant blunt chest trauma, CT can demonstrate the true extent of underlying injury which m a y be useful in subsequent management.

References 1. Eichelberger MR, Randolph JG (1981) Thoracic trauma in children. Surg Clin North Am 61:1181 2. Sivit CT, Taylor GA, Eichelberger MR (1989) Chest injury in children with blunt abdominal trauma: evaluation with CT. Radiology 171:815

3. Wagner RB, Crawford WO Jr, Schimpf PP (1988) Classification of parenchymal injuries of the lung. Radiology 167:77 4. Tocino I, Miller M (1987) Computed tomography in blunt chest trauma. J Thorac Imaging 2:45 5. Spouge A, Burrows PE, Armstrong D, Daneman A (1991) Traumatic aortic rupture in the pediatric population. Pediatr Radiol 21:324 6. Bender TM, Oh KS, Medina JL, Girdany BR (1987) Pediatric chest trauma. J Thorac Imaging 2:60 7. Goldstein L, Mirvis SE, Kostrubiak IS, Turney SZ (1989) CT diagnosis of acute pericardial tamponade after blunt chest trauma. A JR 152:739 8. Wagner RB, Crawford WO, Schimpf PR Jamieson PM, Rao K (1988) Quantitation and pattern of parenchymal lung injury in blunt chest trauma: Diagnostic and therapeutic implications. J Comput Tomogr 12:270 9. Rhea JT, Novelline RA, Lawrason J, Sachnoff R, Oser A (1989) The frequency and severity of thoracic injuries detected on abdominal CT scans of multiple trauma patients. J Trauma 29:502 10. Bonadio WA, Hellmich T (1989) Post-traumatic pulmonary contusion in children. Ann Emerg Med 18:49 11. Godwin JD, Merten DF, Baker ME (1985) Paramediastinal pneumatocele: alternative explanation to gas in the inferior pulmonary ligament. A JR 145:525 12. Kearns SR, Gay SB (1990) CT of blunt chest trauma. A JR 154:55 13. Smyth BT (1979) Chest trauma in children. J Pediatr Surg 14:41 14. Illstad ST, Tollerud D J, Weiss RG, Cox JA, Martin LW (1990) Cardiac contusion in pediatric patients with blunt thoracic trauma. J Pediatr Surg 25:287 15. Wall S, Federle M, Jeffrey RB, Brett CM (1983) CT diagnosis of unsuspected pneumothorax after blunt abdominal trauma. A JR 141:919 16. Nakayama DK, Ramenofsky ML, Rowe MI (1983) Chest injuries in childhood. Ann Surg 210:771 17. Shulman HS, Samuels TH (1983) The radiology of blunt chest trauma. J Can Assoc Radio134:204 18. Kilman JW, Charnock E (1968) Thoracic trauma in infancy and childhood. J Trauma 9:863