COMPLEX THORACIC INJURIES

COMPLEX AND CHALLENGING PROBLEMS IN TRAUMA SURGERY

0039-6109/96 $0.00

+ .20

COMPLEX THORACIC INJURIES J. David Richardson, MD, Frank B. Miller, MD, Eddy H. Carrillo, MD, and David A. Spain, MD

Thoracic injuries remain a leading cause of death in trauma victims, and many patients with chest injuries die at the scene of injury. Problems related to cardiac or great vessel injuries, which are outside the scope of this article, are certainly among the "complex" thoracic injuries. We have focused our attention on four specific problems that we have found to be difficult to manage, either from a judgment or a treatment standpoint. These complex problems are (1) the diagnosis and management of air leakage not associated with pneumothorax, (2) the management of major thoracic esophageal injuries, (3) problems related to penetrating trauma, particularly those pulmonary resections for trauma, and (4) the diagnosis and management of retained hemothorax and empyema, especially in the patient with blunt trauma. EVALUATION AND MANAGEMENT OF EXTRA-ANATOMIC AIR

Patients who have air in the subcutaneous tissues, mediastinum, or pericardium are frequently encountered with either blunt or penetrating trauma. The presence of air in these extraneous locations may or may not be associated with pneumothorax and may occur in the constellation of multiple injuries. Air in these extra-anatomic locations may be entirely innocuous; however, it may also herald the presence of a serious injury to the esophagus or tracheobronchial tree in addition to the lung itself. This problem may not be complex in the treatment required but frequently demands wise judgment to properly evaluate the cause of abnormal air collections and to prioritize diagnosis and treatment, particularly in the multiply injured patient. Whether air in the mediastinum should be evaluated at all is a valid question. We have used the term extra-anatomic air to describe air occurring outside its normal location. This may include pneumomediastinum, pneumopericar-

From the Department of Surgery, University of Louisville, Louisville, Kentucky

SURGICAL CLINICS OF NORTH AMERICA VOLUME 76 * NUMBER 4 * AUGUST 1996

725

726

RICHARDSON et a1

dium, or subcutaneous air. The incidence of pneumomediastinum, pneumopericardium, or subcutaneous air in the chest or neck in trauma patients has not been reported, but it is a frequent occurrence in patients with both blunt and penetrating injuries. Several different mechanisms allow air to enter the mediastinum, pericardium, or subcutaneous tissues of the chest or neck (1)through a perforation of the trachea, bronchus, or esophagus, (2) from an interstitial lung injury, (3) from a facial injury through the fascia1 planes of the neck, (4)from the retroperitoneal space through the diaphragmatic hiatus, and (5) from the outside via a penetrating injury. The majority of these patients present to the hospital with subcutaneous air along the chest wall. This air is frequently adjacent to rib fractures or penetrating injuries. In this situation, the air usually comes from the lung or from outside the chest and a pneumothorax is frequently present. Treatment consists of placement of a thoracostomy tube and consideration of a diagnostic evaluation to exclude a major airway or esophageal injury. Routine use of bronchoscopy, esophagoscopy, and esophogography is not needed in all these patients, but one should always maintain a degree of suspicion for esophageal or major airway injury. If a pneumothorax is present, we generally do not evaluate the patient for visceral injuries based on the mere presence of extra-anatomic air if the injury is not in proximity to the mediastinum. Major tracheobronchial injuries usually present with a large air leak or failure of the lung or a lobe to properly expand. The above-mentioned diagnostic studies should be performed in patients with penetrating injuries in proximity to the esophagus, trachea, or major bronchi. Patients with blunt trauma who have other signs or symptoms of major airway or esophageal injury obviously need further diagnostic studies. If the patient does not have a demonstrable cause for extra-anatomic air such as an open penetrating wound or a pneumothorax, the suspicion for an esophageal or airway injury should be even greater (Fig. 1). Esophageal Injuries

The majority of traumatic injuries to the esophagus are from penetrating injuries and involve the cervical esophagus.53Pain is the most common symptom

Figure 1. Chest radiograph of patient with blunt chest trauma with multiple rib fractures and pulmonaty contusion. Note air in neck mediastinum and pericardium (arrows).

COMPLEX THORACIC INJURIES

727

but it may not be present at all. Crepitus is common in cervical injuries but not in thoracic or abdominal perforations. Other clinical findings are dyspnea, dysphagia, hemoptysis, and fever. Plain roentgenograms frequently show air in the subcutaneous tissue, prevertebral area, or mediastinum. Patients with blunt esophageal rupture usually have associated injuries of the thorax involving the vertebrae, lung, heart, or great vessels. Any patient with the above findings should have the esophagus investigated.2O We have recently seen two patients in our trauma unit who had cervical esophageal perforations from difficult nasotracheal intubations. Both of the perforations went unrecognized for 2 days despite the physical examination finding of subcutaneous air in the neck after the difficult intubations. The perforations were discovered only after the appearance of fever, leukocytosis, and increasing swelling of the neck. Both of these patients should have had diagnostic studies, including esophagoscopy and esophagograms, after the subcutaneous air in the neck was discovered. Because these patients may remain intubated, contrast is less useful and rigid esophagoscopy may be needed. Depending on the condition of the patient and the availability of resources, flexible fiberoptic esophagoscopy is the initial procedure of choice. Rigid esophagoscopy may be preferred in potential cervical esophageal wounds because the blind passage of the flexible scope through the cricopharyngeal muscle and proximal esophagus may result in undiagnosed injuries in the upper gullet. Data suggest that they are both excellent methods and one should use the procedure one is most expert in performing. Flexible endoscopy is probably safer and has better optics and visualization; we prefer it for injuries that might involve the intrathoracic or abdominal esophagus. If an injury is visualized, operative repair is indicated. If no injury is seen, contrast esophagography is necessary. We prefer to use a water-soluble contrast first, and, if this is negative, a barium esophagogram is performed. Esophagoscopy and esophogography are complementary, with reported individual sensitivity of 85% to 90% and nearly 100% occurring when both methods are employed. The treatment of esophageal injuries is often straightforward if the diagnosis is made early and the injury is fairly small. The operative approach depends on the level of injury and either a cervical incision, thoracotomy, or celiotomy is required to expose the esophagus. In all locations we prefer a two-layer closure with an absorbable suture closure of the mucosa and a muscle repair as well. In all areas, some type of buttress should be added to reinforce the muscular repair. Local muscle flaps or pleura may be used to buttress the repair. We routinely use gastrostomy and jejunostomy for all major esophageal injuries that are at high risk for disruption. The subsequent section on management of difficult esophageal wounds focuses on operative techniques that we have found useful for managing major esophageal wounds. Tracheobronchial Injuries

Injuries to the trachea or major bronchi can be life threatening and have a reported incidence of 0.03% in a autopsy study of 1178 trauma deaths? In this study, 27 of the 33 patients who succumbed to tracheobronchial injuries died almost immediately. However, increasing numbers of patients are arriving at hospitals in severe airway distress. Occasionally, few clinical signs or symptoms are present even with severe tracheobronchial injury. Early diagnosis and treatment are essential to preserve life and prevent complications of airway stenosis and infection in the form of mediastinitis.

728

RICHARDSON et al

The majority of these patients with blunt or penetrating injury present in severe distress or in extremis. It is generally obvious that a major airway and/ or ventilatory problem exists. Depending on the level of injury, endotracheal intubation should be attempted and if successful may temporarily improve the problem. Bilateral tube thoracostomies should be placed if the patient remains in distress and there is any suspicion of a pneumothorax. Occasionally emergency or urgent sternotomy or thoracotomy must be performed to obtain and secure an airway. All patients with penetrating injuries near the trachea or major bronchi should have bronchoscopy to diagnose an injury. In experienced hands fiberoptic and rigid types of bronchoscopy are both very ~ensitive.~ We perform fiberoptic bronchoscopy as our procedure of choice. We use the "armed" or "sheathed" bronchoscope, in which an endotracheal tube is threaded over the flexible scope. It can then be advanced to secure or stabilize an airway if an injury is found. This eliminates the need to remove the bronchoscope and attempt reintubation with the attendant risk of losing the airway in critical cases. Stable patients with blunt tracheal or major bronchial injuries are less common. In a series of 13 patients with blunt tracheobronchial disruptions at our institution, 85% had subcutaneous emphysema and 77% had d y ~ p n e a . ~ ~ Pain, hemoptysis, and aphonia are also frequent findings. Radiographic findings commonly include air in the neck, mediastinum, or pleural space, depending on the level of the injury. In one series of 12 patients, all patients with tracheal or laryngeal injuries had deep cervical emphysema seen on plain radiograph^.^^ All the patients with bronchial injuries had a pneumothorax; either mediastinal air or a complicated pneumothorax (total lung collapse or tension) was uniformly present as well. Because 80% of bronchial ruptures occur within 2.5 cm of the carina, not all cases present with a pneumothorax. This is especially true on the left side, where the main bronchus has a longer mediastinal course. Additional radiologic abnormalities, which may suggest bronchial disruption, include extraluminal position of the endotracheal tube tip, overdistention or round appearance of the endotracheal balloon, and the fallen lung sign.49The latter, described by Oh in 1969, refers to the peripheral, rather than central, collapse of the lung which can occur when the normal central anchoring attachments of the lung are disrupted. If a pneumothorax is suspected, a tube thoracostomy should be placed. A large air leak or inability to expand the lung strongly suggests a major airway injury. In stable patients with subcutaneous air without a pneumothorax, we recommend placement of a tube thoracostomy if the patient requires ventilatory support. If a patient has unilateral rib fractures or a penetrating injury, then placement of a single tube on the side of injury may suffice. The treatment of major airway disruptions has been well described. Cervical tracheal injuries can be repaired through a collar incision. Injuries to the proximal intrathoracic portion of the trachea may be approached through a collar incision with a partial upper sternotomy. Simple penetrating wounds may be repaired by two-layer suture technique after dkbriding the wound edges. Blunt injuries often result in complete tracheal ring disruption that leads to loss of normal architectural structure, requiring resection of one or more tracheal rings. End-to-end anastomosis is required in this situation. More distal tracheal injuries require a right thoracotomy for repair. The use of bronchoplastic technique such as used with a sleeve resection may occasionally be needed to salvage a lung with a mainstem bronchus injury. Pulmonary resection is avoided whenever possible but occasionally must be performed for severe crush injuries to the mainstem bronchus (see section on pulmonary resection).

COMPLEX THORACIC INJURIES

729

Pneumomediastinum and Pneumopericardium

Many patients with pneumomediastinum or pneumopericardium have other signs or symptoms that have been previously mentioned and require diagnostic evaluation to detect potential esophageal or tracheobronchial injuries (Fig. 2). When mediastinal air is present, physical examination may reveal a systolic crunch as described by Hammon in 1939. Pneumopericardium may have a ”bruit de moulin,” as described in 1844 by Bricheteau. Patients with blunt injuries who have radiologic findings of air in the mediastinum or pericardial space and no other evidence of esophageal or major airway injury can be safely observed.” In this situation, facial fractures should be considered as a source of the mediastinal air.z3 Although pain may occur, these patients are generally asymptomatic and need no treatment. The air resolves with time and rarely requires drainage. On rare occasions, tension pneumopericardium may occur.” This entity has been described in both blunt and penetrating injuries. The air is easily seen in the pericardial space by plain chest radiographs. Air is more compressible than fluid, so more air is required to cause tamponade, and the radiographic picture of air surrounding the heart is usually quite striking. Treatment consists of drainage of the pericardial air. Pericardiocentesis can be life saving and may be all that is required. A pericardial window with tube drainage or stemotomy/thoracotomy to establish drainage and repair the air leak into the pericardial space may be necessary, depending on the origin of the air. Another interesting problem that presents a dilemma for trauma surgeons is the presence of a pneumopericardium in stable patients with stab wounds near the heart. When no signs of tamponade are seen, the need for aggressive treatment is questioned by some. Demetriades et all3described 20 such patients, all with knife injuries of the chest and a pneumopericardium seen on the initial roentgenogram. All patients were stable and were assessed clinically and with serial chest films, electrocardiograms, and echocardiograms. One patient developed a tension pneumopericardium 36 hours after injury and required operative intervention. The other 19 made an uneventful recovery, although five had electrocardiogram evidence of pericarditis and two had small pericardial effu-

Figure 2. Pneumopericardium (arrows) seen on chest radiograph in patient with blunt chest trauma.

730

RICHARDSON et a1

sions seen on echocardiogram. They advocated nonoperative treatment with hospitalization and serial studies until the pneumopericardium resolved. We recommend a diagnostic pericardial window on initial presentation of these patients3" A sternotomy would follow only if blood were found in the pericardial space. A pericardial window provides a rapid, definitive diagnosis and is less expensive than the management protocol of multiple serial studies as described above. In summary, all patients with subcutaneous, mediastinal, or pericardial air should be suspected of having an esophageal or major tracheobronchial injury. Unstable patients generally need rapid airway control via endotracheal intubation, tracheostomy, or sternotomy/thoracotomy. Once patients are stable, diagnostic studies consisting of bronchoscopy, esophagoscopy, and esophagography should be considered. Stable patients with no evidence of esophageal or major tracheobronchial injury other than subcutaneous, mediastinal, or pericardial air can be safely observed. The yield from these diagnostic studies is admittedly low, but the failure to diagnose an esophageal injury early may result in the patient's death. In a recent unpublished review of patients evaluated at our institution with extra-anatomic air, we detected two patients with major injuries out of 45 patients studied. The two esophageal injuries detected had no other signs present. Two patients with injuries of the cervical esophagus secondary to traumatic intubation had only minimal crepitus in the neck and had delayed diagnosis. These cases underscore the need for a vigorous diagnostic approach, primarily with fiberoptic endoscopy, to exclude major perforations in the airway or esophagus. Intubated patients need rigid esophagoscopy to evaluate the cervical esophagus for perforation. Management of Esophageal Injuries

The diagnostic evaluation of potential esophageal injuries has been mentioned previously and is not discussed further. Three issues are discussed: (1) the general principles of managing proven esophageal injuries, (2) the treatment of injuries that are detected early, and (3) the management of injury diagnosis made in a delayed fashion. Treatment Principles

Several principles are important to the management of all esophageal injuries. The first principle of treatment is to close the esophagus when possible to prevent further contamination of the mediastinum. In early-recognized injuries, this can usually be easily accomplished by primary repair of the esophagus. If the leak cannot be closed, a suction tube should be placed within the esophagus to remove secretions. The injury must be properly drained initially, and adequate drains must be left in place to control either an injury that cannot be closed or a leak following repair. A soft chest tube or right angle tube may be useful. If contamination of the mediastinum has occurred, the mediastinitis must be treated by appropriate aggressive surgical means. The mediastinal spaces must be opened and drained and any purulent or necrotic tissue must be dkbrided. A delay of even a few hours in the diagnosis of an esophageal injury can lead to significant mediastinitis. Control of reflux through the use of a gastrostomy tube is very useful in distal injuries. The addition of a feeding jejunostomy to provide enteral nutrition

COMPLEX THORACIC INJURIES

731

is an essential adjunct, especially in patients with a high risk for leak or in those whose injury cannot be closed primarily. Finally, we believe that the use of an adjunct maneuver such as flap coverage or primary closure should be considered on virtually every esophageal injury. We routinely buttress primary repairs with either muscle or pleural flaps. Although one can argue that this may result in overtreatment of some wounds, the prevention of even one delayed leak in 20 cases would warrant such an approach in our opinion. This is particularly true in thoracic esophageal injuries, where a leak can be a disaster. Injuries with delayed recognition may be treated by primary closure with a muscle flap. Early-Recognized Injuries

Injuries to the thoracic esophagus are relatively uncommon and are often seen with associated injuries. Penetrating wounds constitute over 90% of the traumatic wounds; blunt thoracic esophageal injuries are relatively uncommon. Once the esophageal wound is detected, it should be approached through a thoracotomy. The side of the thoracotomy may be determined by the presence of other injuries that require treatment. If a contrast study or esophagoscopy confirms an esophageal injury and no other reasons to operate are noted, we let the level of injury dictate the side of operation. In distal esophageal wounds, we use a left thoracotomy. For more proximal injuries, we prefer a right thoracotomy. Once the injury is identified and the esophagus properly mobilized, we attempt two-layer closure. An absorbable suture closure of the mucosa is used, followed by reapproximation of the muscular layer. We use a nonabsorbable suture closure of the muscular layer. This closure is then buttressed or bolstered with a flap. If the mediastinum is not grossly inflamed, a pleural flap may be elevated to cover the repair. If the wound is large and its repair is questionable, we may use a muscle buttress. The latissimus dorsi may be used if a high suspicion of delayed leak is held. Intercostal muscle bundles are not reliable in our experience. We frequently use diaphragm flaps as well. Management of Esophageal Wounds with Delayed Recognition

Because there are no specific signs of esophageal wounds, it is not unusual that delayed diagnosis of such injuries occurs. The primary repair of laterecognized wounds may be difficult if not impossible. Suture repair may not be technically feasible, and the sutures may cut through if primary closure is attempted. A variety of techniques have been used in this situation: No attempt at direct closure: intraesophageal suction tube and tube drainage T-tube placement in the perforation Exclusion and diversion in continuity Cervical esophagostomy without distal ligation of esophagus Esophageal resection Flap closure (pleural, muscle, omentum) The number of methods employed may be in some part due to the different injuries encountered but also speaks to the unsatisfactory nature of most reported techniques.'O We have not used resection as primary treatment unless

732

RICHARDSON et a1

the esophagus has underlying disease, and this seldom is the case in trauma patients. Diversion and exclusion are frequently mentioned as options when the injury cannot be repaired.50However, the creation of a cervical esophagostomy represents a major problem in and of itself. If the patient survives, a subsequent major reconstruction is needed to close the esophagostomy. The creation of a distal obstruction is unphysiologic when the surgical goal is to close a perforation. We use cervical esophagostomy only as a last resort in patients who may die of mediastinitis. Since we have begun more aggressive closure with muscle flaps, a diverting esophagostomy has been infrequently required. We have previously reported on the use of muscle flaps to primarily close difficult esophageal It should be emphasized that this technique does not buttress a suture closure but uses the muscle as an onlay flap to close the defect directly. We have used rhomboid muscle in patients with a previous thoracotomy who did not have available latissimus dorsi musclez6(Fig. 3). The latter muscle

Figure 3. A, The rhomboid muscle is approached through a parascapular incision. B, The rhomboid major muscle is divided proximally (along the dotted lines) to provide a generous portion of muscle to close a proximal esophageal defect. C, The rhomboid muscle flap is brought through the chest wall by excising a portion of the fourth rib. It then is sutured in place to close the esophageal defect directly. (From Richardson JD, Tobin GR: Closure of esophageal defects. Arch Surg 129541-548, 1994, copyright 1994, American Medical Association; with permission.)

COMPLEX THORACIC INJURIES

733

geal Tear

Diaph&m% Heart

DiaphAgm .

Heart

Aorta

Figure 4. A, A distal esophageal defect amenable to diaphragm flap closure. B,A posteriorly based flap is elevated and double-armed sutures are placed inside the defect and sutured directly onto the diaphragmatic flap itself. C,The flap is sutured in place as the knots are tied, and the diaphragm is primarily closed with two layers of nonabsorbable sutures. (From Richardson JD, Tobin GR: Closure of esophageal defects. Arch Surg 129:541-548, 1994, copyright 1994, American Medical Association; with permission.)

may be used in early cases, particularly in a robust muscular patient. Distal esophageal wounds have been repaired by the diaphragm in a number of cases (Fig. 4). Despite theoretical concerns about the use of the diaphragm-the possibility of phrenic nerve injury and the potential contamination of the abdomen-we have encountered no problems with its use. The pleural hap has been less reliable for primary closure, particularly in patients with pleural sepsis and significant inflammation. After the closure of the defect with muscle flaps, the area is widely drained. We frequently employ a right-angle chest tube to ensure that the area in proximity to the wound is drained. A water-soluble contrast study is obtained at 5 to 7 days. Small residual leaks may persist (in about one quarter of our patients so treated), but they have uniformly healed when a muscle flap was used. We continue to study the patient with contrast until healing has occurred. We do not perform further procedures such as cervical diversion or resection unless the patient clearly fails to respond or is worsening. Our results with this approach have been excellent. It has led us to believe that surgeons rely too much on drainage alone when treating delayed esophageal injuries. We believe a more aggressive approach at attempting to eliminate further contamination is appropriate, with the use of aggressive measures to close the esophagus using muscle flaps.

734

RICHARDSON et a1

The concept of primary muscle flap closure is also useful as primary treatment for complex wounds that are recognized early. Two examples are outlined below: A 28-year-old policeman was shot in the line of duty with a .45-caliber handgun. The entrance wound was in the right chest and the patient arrived in the emergency department with stable vital signs. Following tube thoracostomy, a chest radiograph showed a large bullet in the middle mediastinum. A Gastrografin swallow showed the bullet lodged in the esophagus (Fig. 5). A right thoracotomy was performed, and the bullet was removed from the esophagus. There was a large entrance hole and a smaller exit wound on the left side of the esophagus that had devitalized the esophageal wall in the area of the exit wound. After debridement, only 50% of the esophageal wall remained, and closure of both lesions was not feasible without severely constricting the lumen. We created a latissimus dorsi flap, which was used to buttress the entrance wound, which was primarily repaired. It was then rolled 180 degrees around the esophagus to primarily close the exit wound as an onlay flap. At days 5 and 7 no leak was seen on esophagography, and the patient was allowed to eat. He was discharged on day 10, and subsequent follow-up for several years showed no subsequent narrowing or problems of any kind. A schizophrenic woman shot herself with a .357 magnum pistol in the lateral lower chest. The injury destroyed the lateral segment of the left lobe of the liver, penetrated the right diaphragm, and created a tangential wound in the distal esophagus that was 6 cm long, with absence of most of the right wall. Remarkably, the heart and aorta were not injured.

Figure 5. A large bullet (arrow) is present in the wall of the esophagus. Metal fragments are seen on the right side of the esophagus, which was the entry point into the lumen. When the large bullet was removed, more than half of the circumferential wall of the esophagus was destroyed, making primary closure impossible without complete obstruction of the lumen.

COMPLEX THORACIC INJURIES

735

The liver wound was treated with resectional dhbridement and a left thoracotomy performed. The wound in the esophagus could not be satisfactorily repaired without closing the esophageal lumen. A diaphragmatic flap was created to directly close the defect. A study on the sixth postoperative day showed a small leak that was well drained. By day 10 the leak had closed. The patient did well and had not developed a stricture after 18 months follow-up. The use of these flaps is an important adjunct in the management of complex esophageal injuries. PULMONARY RESECTIONS AFTER PENETRATING INJURIES TO THE CHEST

The principles and practice of nonoperative therapy for penetrating thoracic trauma have been previously detailed in a number of series. Observation or placement of chest tube thoracostomy, adequate volume resuscitation, occasional need for respiratory support, and serial chest roentgenograms are the only treatments required in 80% to 85% of 37, 52 For the remaining patients, urgent thoracotomy for control of bleeding, repair of tracheobronchial disruptions, relief of cardiac tamponade, and control of associated injuries can be life saving. Historically, only a small number of patients requiring thoracotomy undergo formal pulmonary resections, usually secondary to severe injuries to the pulmonary parenchyma or the pulmonary hilum. Pulmonary resections in 19, 39 Death most comthese patients carry a mortality rate from 30% to 5Oy0.~. monly is secondary to uncontrollable hemorrhage; air embolism has also been recognized as a serious and fatal complication of penetrating injuries to the lung and pulmonary hilum.I6 Surgical treatment of battle casualties with penetrating thoracic trauma over the last 50 years shows improved survival with early operation combined with aggressive thoracic pr0cedures.5~This experience, gained in the battlefields, has stimulated significant progress in surgical techniques and overall management of trauma victims with complex thoracic injuries in the civilian population. Initially, pulmonary resection was almost never required for civilian injuries but was occasionally needed for military wounds. The incidence of high-velocity wounds encountered in civilian practice has skyrocketed, necessitating a marked increase in the use of pulmonary resection as definitive therapy in trauma patients. Incidence Complete pneumonectomies or extensive pulmonary resections are indicated in only a few patients undergoing emergency thoracotomy after penetrating thoracic trauma.39, 54 In our own review of 259 patients who underwent urgent thoracotomy after penetrating chest trauma? 34 patients (13%)underwent lobectomies and 9 patients (4%)underwent pneumonectomies. This incidence of pulmonary resection is high for a civilian series based on historical trends but reflects the change in weaponry and subsequent wounds encountered. CLINICAL PRESENTATION AND DIAGNOSIS OF PATIENTS REQUIRING URGENT THORACOTOMY

A systolic blood pressure of 80 mm Hg or less on arrival at the hospital is present in 67% to 75% of patients with injuries involving hilar vascular struc-

736

RICHARDSON et a1

tures, and associated injuries are seen in 75% of patients?, 54 After initial clinical evaluation, the insertion of a thoracostomy tube, the monitoring of blood loss, and the evaluation of the chest roentgenograms, one should be able to select that 15%to 20% group of patients who require urgent thoracotomy. The chest roentgenogram can be extremely useful to determine location of missile fragments or the persistence of a large retained hemothorax ("white chest"). The presence of missile fragments near the pulmonary hilum or the presence of a large hemothorax should arouse suspicion of a major vascular injury in the mediastinum or pulmonary hilum. initial Evaluation and Management

Because of the critical nature of these injuries, in most patients the diagnostic evaluation should be very limited. Persistent attempts to document specific injuries before operation are not only not indicated but are extremely dangerous, add little to the decision-making process, and place the patient in imminent risk of death. It is important to emphasize that in order to diminish the morbidity and mortality in these patients, an organized approach is very important. Patients who present in an unstable condition after sustaining penetrating chest trauma and who remain hemodynamically unstable despite resuscitation should be taken to the operating room immediately (Fig. 6). This deviation from "traditional" management should be made by a senior surgeon with experience in the management of thoracic trauma. In our institution these patients, upon arrival to the trauma room, are met by a senior surgical resident and/or the in-house trauma surgeon, so the decisions regarding further care after initial resuscitation are made in a rather expeditious way. Initial management and resuscitation are carried out following the guidelines and principles of Advanced Trauma Life Support. Deviation from "traditional" management is done mainly because a management strategy designed to manage patients on a more selective basis is not applicable to this group of patients, and the nature of these injuries usually requires urgent operation for definitive treatment. The decision to perform a thoracotomy in these patients should be deter-

Hemorrhage

Cardiac Tamponade

I

Aspiration Pericardial Window

Thoracotomy +Hemithorax with major blood loss

Median Sternotomy or Left Thoracotomy

J STABILIZATION

1

Assess Other Structures

Figure 6. Algorithm for the management of the unstable patient with penetrating wounds to the chest.

COMPLEX THORACIC INJURIES

737

Figure 7. A, Single-gunshot wounds to the right chest with retained hemothorax (“white chest”) in a 22-year-old man. 6,The postoperative radiograph shows that the mediastinum is slightly deviated to the right. No thoracostomy tube was left in the thoracic cavity. The incision was extended through the sternum to facilitate exposure, and the sternum was closed with a single No. 5. stainless steel wire.

mined by history, clinical assessment, and the initial AP chest roentgenogram. Amount of drainage through a chest tube thoracostomy should not be used as the only criterion to decide on an operation. Persistence of an undrained hemothorax with bullet fragments near the pulmonary hilum and cavitary injuries to the lung parenchyma should arouse suspicion as to the extent of the injury (Fig. 7A).We use a 42 Fr chest tube in all patients with penetrating chest trauma. If after placement of the chest tube a large hemothorax is noted and the patient sustains a drop in blood pressure greater than 10 mm Hg, the chest tube is clamped and the patient is taken immediately to the operating room. In more stable patients, if the trajectory of the missile is in an AP plane, a lateral chest film may help to define the location of the bullet. However, no time should be wasted on unnecessary roentgenograms. Because injuries to the pulmonary

738

RICHARDSON et a1

hilum should be suspected preoperatively, these patients should be taken to the operating room as expeditiously as possible. In the operating room, resuscitation is continued with placement of lines necessary for fluid administration and initial monitoring. The use of pulmonary artery catheters is not justified, and arterial line placement should not delay needed urgent thoracotomy. All these patients should be prepped and draped while still awake, especially those for whom there is a concern about air embolism secondary to positive-pressure ventilation. Once the patient is intubated, one should be ready to perform a thoracotomy immediately and obtain control of the pulmonary hilum as soon as possible. Occasionally, a double-lumen (Carlens) tube can be used, but this technique requires more expertise and time and is very seldom indicated in these patients. In hypotensive patients, an anterolateral thoracotomy is performed, and this incision usually gives good exposure to the lung and pulmonary hilum. If there is need for better exposure, the incision can be extended through the sternum without entering the contralateral pleura. Median sternotomy, the standard incision for most operative procedures on the heart, ascending aorta, and anterior mediastinum, is increasingly employed for selected cases of chest trauma.'* The major limitation of median sternotomy is access to the left lower lobe; however, it provides easy access to the pulmonary hilum. We recommend the use of median sternotomy in patients with bilateral anterior thoracic wounds and when there is a question about cardiac involvement. The posterolateral thoracotomy incision is generally used in more stable patients or when there is a need to explore the posterior mediastinum. Positioning of the patient is more difficult, the physiologic stress placed on hypotensive patients is not well tolerated, and a double-lumen tube is ideal to prevent blood and secretions from flooding the contralateral lung. Unfortunately, the ideal situation is often not possible with urgent thoracotomy. Once the thoracic cavity has been entered, the extent of the injuries should be assessed and the need to control the pulmonary hilum determined. If the pulmonary hilum needs to be controlled, hilar cross-clamping usually is done with a Satinsky clamp placed around the hilum and proximal to the injury, incorporating the pulmonary artery, both pulmonary veins, and the main bronchus. Alternatively, a hilar snare (Roummel tourniquet) can be used for securing rapid vascular control of the pulmonary h i l ~ mGood . ~ ~ control of the hilum is equivalent to a physiologic pneumonectomy. It is at this point that the most critical decision should be made regarding the best way to handle these injuries. Usually these patients have sustained significant blood losses, prolonged hypotension, hypothermia, uncorrected acidosis, and occasionally air emboli, and they do not tolerate prolonged operations and further blood loss. Attempts to control hemorrhage without proximal vascular control usually provide deceptive control of the hemorrhage. In proximal injuries or in injuries too close to the hilar clamp, the pericardium can be opened and intrapericardial control of individual pulmonary vessels can be obtained. If the hilar cross-clamp provides good control of the hemorrhage, one or two attempts should be made to repair the vascular injuries. Bronchial injuries are usually not difficult to repair if good vascular control has been obtained and if no associated vascular injuries are present. If these attempts to control hemorrhage or repair vascular injuries are futile, then a stapled pneumonectomy should be performed to prevent the cycle of hemorrhage-hypotension-acidosis-hypothermiaand death. In our review of 259 patients9who underwent urgent thoracotomy after penetrating chest trauma, 16% required major pulmonary resections. Wienceck and Wilson54reported a 15% incidence of hilar injuries in their series. Mortality was 66% and 50%,

COMPLEX THORACIC INJURIES

739

respectively, after pneumonectomies in those series; however, increasing experience from our institution as well as others indicates the advantage of an early pneumonectomy in these patients if an extensive pulmonary resection is required. Pulmonary resection of a lesser magnitude than a pneumonectomy may be indicated in several situations. High-velocity bullet wounds may cause a cavitation of the periphery of the lung that destroys most of the lung parenchyma. These injuries are usually peripheral because patients who have central highvelocity wounds usually die at the scene. In this situation, a wedge resection with a stapler or a lobectomy may be needed to stop ongoing bleeding and to ”dbbride” devitalized lung tissue. In our experience, these types of pulmonary resections are well tolerated unless they are associated with other injuries that produce an additional blood loss and coagulopathy. Mortality and Morbidity

The mortality rate for injuries to the pulmonary hilum or devastating injuries to the lung parenchyma remains high owing to exsanguination. However, systemic air emboli have become an important factor, despite the fact that their diagnosis is difficult to make in the acute care setting. ”Pleural shock,” almost certainly air embolism, was an entity described at the turn of the century.4I It is characterized by the sudden onset of cardiovascular collapse, seizures, and, at times, sudden death occurring as a complication of thoracentesis, pulmonary resection, or therapeutic pneumothorax. If one suspects systemic air emboli, one should (1) maintain or obtain control of the pulmonary hilum; (2) place the patient in the Trendelenburg position so that the apex of the heart is the highest point in the chest; (3) aspirate air from the tip of the left ventricle and aorta; (4) start open cardiac massage; and (5) maintain perfusion pressure with fluids and vasopressors. It is important to maintain an adequate diastolic pressure because coronary blood flow to the left ventricle occurs in diastole. Air embolism should be corrected before attention is returned to the original injury. Postoperative complications are common in patients with devastating thoracic injuries that require pneumonectomy or lobectomy as part of their surgical management. Tominaga et a148report that in patients who required pulmonary resections after isolated thoracic trauma, the average number of ventilator days was 3.8, average intensive care unit (ICU) stay was 5 days, and average length of hospitalization was 10.1 days. In our experience? the most common complication, pneumonia, was present in 87% of patients, and respiratory failure in 62% who required pulmonary resection. Two patients (8%)developed bronchopleural fistulas, which were successfully managed with chest tube thoracostomy and high frequency jet ventilation (HFJV). Although the incidence of bronchial stump blowout did not increase with the addition of positive pressure in experimental animals, we believe that positive pressure does increase bronchopleural fistulas. If positive pressure and positive end-expiratory pressure (PEEP) are required, they must be used. We emphasize the use of an increased rate and lower tidal volume in postpneumonectomy patients. Thompson et a147and Bowling et a17have shown that pulmonary resections in this select group of patients carry a mortality of 75% to 100% (44% in our experience). And more specifically, the combination of shock, resuscitation, and emergency pneumonectomy is associated with a high mortality rate (69% in our series). Of the patients who presented in shock, one half exsanguinated in the operating room owing to multiple associated and thoracic injuries. The rest of

740

RICHARDSON et a1

the patients died in a combination of pulmonary edema and right ventricular failure with pulmonary hypertension, despite "adequate" resuscitation. These hemodynamic abnormalities were also noted in all patients who died 48 hours after completion of the operation with successful control of hemorrhage and hypotension. Experimental studies'* have shown that during hypovolemic shock there is a greater increase in pulmonary vascular resistance than systemic vascular resistance, and this relationship remains unchanged despite successful resuscitation. Possible mechanisms include the shock-induced t h r ~ m b o x a n eand ~~ leukocyte on the pulmonary circulation. This persistent pulmonary hypertension results in acute right heart failure and eventual left ventricular dysfunction by displacement of the cardiac septum. It is our clinical impression that some of this downward spiraling cascade can be prevented by earlier pneumonectomy. Thus, early senior input into these cases is needed. If the lung cannot be salvaged, a pneumonectomy is preferable to prolonged attempts at vascular repair in the hilum. Cardiac arrhythmias constitute the most common cardiac complication after major thoracic procedures. Most are supraventricular, and of these atrial fibrillation is the most common.51The incidence is 20% to 25% following a pneumonectomy. We use diltiazem for supraventricular tachycardia and esmolol and digoxin for rapid atrial fibrillation. Adenosine, the drug of choice in many patients with primary cardiac disease, has been effective in only 45% of our patients. Postoperative hemorrhage is another common complication after thoracotomy for complex thoracic injuries. Initial management of postoperative hemorrhage in these patients entails correction of both washout and hypothermiainduced coagulopathy and platelet dysfunction. Once these measures have been instituted and abnormal clotting parameters corrected, persistent excessive bleeding mandates return to the operating room. This diagnosis should be entertained early to prevent recurrence of hypothermia and coagulopathy as a result of continued bleeding. In many patients the actual source of the hemorrhage is not found. However, draining the retained clot and irrigating the thoracic cavity help to remove factors that tend to perpetuate coagulopathy. The importance of lung expansion in the prevention of thoracic bleeding cannot be overemphasized. Thus, if a pneumonectomy is required and the patient is coagulopathic, the space may fill with blood rather than serum or exudate. On several occasions, we have packed the chest following pneumonectomy to achieve hemostasis. Pack removal is accomplished after correction of clotting factors. Hypothermia is another serious and common complication. Even mild hypothermia (core temperature 34 to 36.5" C) initiates undesirable changes, including the following: (1) peripheral vasoconstriction and metabolic acidosis; (2) shift of the oxyhemoglobin curve to the left; (3) increase in heart rate, cardiac output, and mean arterial pressure secondary to catecholamine release; (4) progressive and generalized slowing of body enzyme systems; (5) "cold diuresis" secondary to increased central intravascular volume from severe peripheral vasoconstriction; (6) a coagulopathy due to a reversible depression in platelet production of thromboxane A, and release of adenosine diphosphate, which aggregates other platelets in the area and increases the bulk of the platelet plug. Hypothermia therefore inhibits production of elements necessary for primary hemostasis. With the use of new stapling devices such as the TA-35 with a staple size of 4.8, bronchopleural fistulas are observed in 3% to 5% of patients undergoing pneurnonect~my.~~ Early fistulas, 1 to 2 days postoperatively, occur because of a technically poor closure of the bronchial stump. A fistula is manifested by a

COMPLEX THORACIC INJURIES

741

massive air leak with associated subcutaneous emphysema and respiratory distress. Late bronchopleural fistulas (8 to 10 days) reflect failure of healing or infection in the chest cavity. The patient coughs up serosanguinous fluid, and danger of flooding the contralateral lung is present. The affected pleural space should be drained and placed dependently. Airway control occasionally is necessary with selective intubation of the normal lung. If the general condition of the patients allows, early fistulas should be managed surgically with reclosure of the bronchial stump. During primary closure in this unstable group of patients, routine coverage of the bronchial stump is not done in delayed bronchial closure. However, mobilization of neighboring tissues for flap coverage should be used routinely. Pleura may be used, but we prefer muscle flaps. Fistulas that present late usually are secondary to persistent infection in the thoracic cavity. They initially can be managed with drainage of the chest cavity and, if the patients need mechanical respiratory support, HFJV improves gas exchange and is associated with low mean and peak airway pressures. These changes reduce trauma and gas flow through the fistula site and facilitate healing8 Empyema should be suspected in any patient with persistent fever, pleural effusion, and/or failure of closure of a bronchopleural fistula. Thoracentesis, with Gram's stain and culture of the fluid obtained, confirms the diagnosis and guides selection of antibiotics. Pleural fluid with a pH less than 7.20 and glucose less than 40 mg/dL strongly suggests empyema requiring drainage. The goals of treatment of empyema in the postpneumonectomy patient are (1)early diagnosis, (2) appropriate and specific antimicrobial therapy, and ( 3 ) drainage with obliteration of the pleural space by tube thoracostomy and closed or open drainage. We have found that in critically ill patients with persistent infection in the thoracic, cavity, with or without bronchopleural fistula, open drainage using the Eloesser flap techniqueI5 is quite useful. Some patients may require mobilization of omentum or muscle flaps to obliterate the pleural space. Postoperative Management

Monitoring the patient's oxygenation should be routine after pneumonectomy. Arterial lines and pulse oximetry are quite useful. We believe that continuous mixed venous oxygen saturation (SVOJis indicated in almost all patients who undergo extensive pulmonary resections to optimize cardiorespiratory function and to monitor oxygen delivery and consumption. Many patients require initial mechanical ventilatory support after pneumonectomy. Upon arrival in the ICU, the patient should be initially placed in an inspired oxygen fraction of 100% and weaned to 40%, depending on the Svo2 and arterial saturation of oxygen (Sao,). Tidal volume of 600 to 800 (8 to 10 mL/ kg), an intermittent mandatory ventilation (IMV) rate of 8 or to maintain the PCO, around 35 mm Hg, and 4 to 5 cm H,O PEEP may be used. Pressure support can be added to overcome the imposed work of breathing; patients can be ventilated or weaned using this technique. These patients may be kept paralyzed for the first 24 to 36 hours postoperatively to avoid sudden increases in airway pressure and prevent further barotrauma. Routine use of HFJV has been proposed in these patients to improve gas exchange and reduce the "adverse" effects of conventional mechanical ventilation. However, HFJV has some significant limitations, including (1) CO, retention and (2) low-flow and underpowered systems with insufficient alveolar recruitment. The ventilation of the remaining lung is maintained by an increase in the rate of breathing. With hyperventilation, the efficiency of oxygen uptake in-

742

RICHARDSON et a1

creases?’ An increase in the ratio of the tidal volume to the functional residual capacity occurs and leads to an improved mixing of inspired gases. Compliance is reduced with subsequent increase in the work of breathing, this becoming the main reason why some patients are difficult to wean from ventilatory support after extensive pulmonary resections. After pneumonectomy for trauma, the pleural space is closed without drainage46unless packing of the chest has been used. Adjustment (a negative pressure of 2 to 4 cm HzO on inspiration and a positive pressure of 2 to 4 cm HzO on exhalation) may be made simply by thoracentesis and removal of air until the trachea is in the midline at the sternal notch. The rate of accumulation of fluid and complete absorption of the air from the cavity are variable. Obliteration of the pleural space is usually accomplished by re-expansion of the contralateral lung, narrowing of the intercostal spaces, and elevation of the hemidiaphragm. Generally, the process is complete within 3 to 4 weeks (Fig. 7B). Other important factors in the postoperative care of these patients include proper positioning, chest physiotherapy, tracheal suctioning, control of pain, and contingency plans for a possible prolonged intubation and ventilation. Tracheal suctioning should be integral to the care of any patient who is not able to mobilize secretions spontaneously. Tracheal suctioning should be done extremely carefully, ideally by the medical staff. Pushing a catheter through a pneumonectomy stump may result in a bronchopleural fistula, infection of the pneumonectomy space, multiple subsequent operations, and eventually death. If tracheal suctioning does not clear the airways of secretions, bronchoscopy is indicated. Precautions should be taken during bronchoscopy to avoid bronchospasm (instillation of local anesthetic), hypoxia (use high FIo2), and trauma (achieve adequate sedation) to the airway. The use of an epidural catheter for pain control after correction of coagulopathy is an integral part of the management of trauma patients, as it is with electively operated thoracic patients. PLEURAL FLUID COLLECTIONS FOLLOWING BLUNT CHEST TRAUMA The problem of retained hemothorax in the patient with penetrating trauma is well recognized. Similarly, the problem with empyema following penetrating trauma is well understood. In the past several years, we have recognized with increased regularity problems associated with undrained pleural collections in the trauma patient with blunt injury. Frequently, these patients are severely injured, have multiple rib fractures and other associated injuries, are ventilator dependent, and have a confusing diagnostic picture. Is the chest opacity pneumonia, atelectasis, pleural effusions, empyema, or some combination of these processes? Furthermore, the development of advanced imaging techniquesz8* 29, 34 and minimal access s ~ r g e r yhas ~ ~created ,~ controversy in the management of chest trauma. CT scans performed for other indications (e.g., abdominal evaluation) often diagnose pleural fluid collections and parenchymal abnormalities that are not fully appreciated on routine chest radiography.”, 28, 29, 44 Chest CT scan may provide additional information to help differentiate the previous diagnostic puzzles between pulmonary contusion, pneumonia, pleural effusion, retained hemothorax, and empyema (Fig. 8). Earlier recognition of these problems has increased the complexity of decision making in chest trauma: (1) What is the role of CT scan and is it too sensitive? (2) Does every retained hemothorax need to be evacuated? (3) What is the role of thoracoscopy? (4) What is the role of

COMPLEX THORACIC INJLJRIES

743

Figure 8. CT scan of the chest in a blunt trauma patient demonstrating parenchymal consolidation and loculated pleural fluid collections.

surgical intervention in a patient with respiratory failure and pleural collections on CT scan (i.e., is the process empyema whose drainage may aid the respiratory fa$lure)? Retained Hemothorax

Retained hemothorax occurs in 5% to 30% of patients with chest trauma and is a major risk factor for the development of empyema.I4 If untreated, it can lead to fibrothorax, lung entrapment, and impaired pulmonary function. Therefore, accurate diagnosis and early treatment are crucial if morbidity and in particular mortality are to be reduced.", 14, 43 Conventional chest radiographs in trauma patients are often technically inadequate, may underestimate the size of a hemothorax, and may not distinguish pulmonary contusion from retained hemothorax?*,29 Chest CT scan has proven very effective in defining these problems. We have used it primarily in patients with blunt or penetrating trauma who have residual abnormalities not clearly defined by chest radiography. CT scan may delineate retained hemothorax from a parenchymal process. If a hemothorax is small (<200 to 300 mL) and the patient is clinically uncompromised, it may be observed, especially if the pleural cavity has not been violated and potential contamination has not occurred. If drainage is necessary (symptomatic patient or large hemothorax), the first step is to ensure proper placement of an adequate chest tube (Fig. 9). A 42 Fr chest tube should be used to drain a hemothorax. If the tube is malpositioned or inoperable, a second tube should be placed through a different skin incision. Radiologic guidance may be helpful if the hemothorax is loculated or in an unusual location. If a second tube fails to resolve the hemothorax quickly (within 24 hours), we have opted for surgical intervention. Some have advocated use of thrombolytics in this situation3;however, their use is often contraindicated in the multiple trauma patient. Bleeding, allergic reactions, and significant pain and fever may confuse the diagnosis and management. Thrombolytics may be a reasonable alternative in poor-risk pa-

744

RICHARDSON et a1 PLEURAL FLUID COLLECTION DIAGNOSTIC THORACENTESIS

I RETAINED HEMOTHORAX 1 I

I

mp===qI LARGE

RESPIRATORY FAILURE

I

1

TUBE DRAINAGE f RADIOLOGIC

DECORTICATION

5+ RESOLUTION

DRAINAGE, CONVERT TO THORACDTOMY

Figure 9. Algorithm for diagnosis and treatment of patients with pleural fluid collections following blunt chest trauma.

tients with primarily isolated chest injuries, but we do not use them. Thoracoscopy is our initial mode of treatment for the removal of retained blood or clot in the majority of trauma patients, especially within the first 2 to 4 days. Once the clot has been allowed to organize and adhere to the lung and pleura, thorascopic removal becomes more difficult.27,32 This has led some to advocate even earlier use of thoracoscopy, that is, if the first chest tube fails to resolve the hemothorax. Presumably, earlier use increases the likelihood of success. Although it is "less invasive," thoracoscopy is not without significant risk, related primarily to the need for general anesthesia and single lung ventilation. This often requires relinquishing of a controlled airway to place a double-lumen endotracheal tube, which may be difficult under the best of circumstances. Patients with impaired pulmonary function may not tolerate the single lung ventilation required for thoracoscopy. These factors must be taken into consideration when contemplating thoracoscopy during preoperative preparation. If thoracoscopy is unsuccessful, we then use the findings to proceed with a limited thoracotomy directly over the hemothorax. Intraoperative cultures are routinely obtained. Following evacuation, the lung is fully examined and any fibrous tissue is removed. The lung is then fully expanded, and adequate tube drainage is established. Pain control and clearance of secretions are the main objectives postoperatively. This problem is relatively straightforward in patients with penetrating injuries. However, in the blunt trauma patient with multiple injuries, we use more caution in taking patients to the operating room early. This may lead to a higher incidence of true empyema in blunt trauma patients than previously recognized. Empyema

Empyema, defined as infection in the pleural space, occurs in 5% to 10% of patients with penetrating chest trauma but is probably underdiagnosed follow-

COMPLEX THORACIC INJURIES

745

ing blunt trauma. Although many factors predispose the trauma patient to empyema, including incomplete expansion of the lung, retained hemothorax, presence of a foreign body (chest tube), pneumonia, and immunosuppression, the incidence has decreased dramatically with the increased use of antibiotics. The microbiology of empyema has changed as well. Currently, the most common isolate is Staphylococcus mucus, although many patients have culture-negative purulent collections.2,6, l7 Gram-negative organisms are common in patients with concomitant pneumonia. Treatment of empyema depends primarily on the stage at diagnosis.’, Stage 1 is characterized as the exudative stage and can often be treated with simple chest tube drainage and antibiotics. Patients in stage 2, the fibrinopurulent stage, usually require surgical drainage and often decortication. In the organizing phase or stage 3, patients are best treated by thoracotomy, decortication, drainage, and appropriate antibiotic therapy. However, we do not believe that the typical trauma patient has an exudative phase. This is more typical of a parapneumonic process than an infected hematoma, which characterizes most traumatic empyema cases. Therefore, a more aggressive treatment approach may be more appropriate for traumatic empyema than parapneumonic empyema. The increased use of CT scan in critically ill patients, however, often finds complex thoracic changes that are difficult to categorize. This is especially true after blunt chest trauma. A typical clinical scenario is the following: A patient sustains blunt chest and abdominal trauma. A tube thoracostomy is performed for hemothorax or pneumothorax associated with rib fraqtures. The patient undergoes a laparotomy for intra-abdominal bleeding. Postoperatively, the patient develops fever, leukocytosis, and ventilator dependence. He or she may have positive sputum cultures and abnormal chest radiographs. A CT of the abdomen is performed for suspected intra-abdominal abscess. The abdominal portion is negative, but there appears to be either a pneumonia, a retained hemothorax (with or without secondary infection), or an empyema present within the injured hemithorax. These patients usually have some pleural thickening due to rib fractures and the tube thoracostomy. There are often changes that may represent some combination of retained hemothorax, consolidated pneumonia, or an empyema (see Fig. 8). If a patient has no other reason for fever, leukocytosis, and ventilator dependence, we have been very aggressive about investigating these findings (see Fig. 9). A diagnostic thoracentesis is performed at the time of CT, using this as a guide. However, fluid often cannot be obtained. If the Gram’s stain or culture is positive, adequate drainage must be established. Because the purulence is loculated, a chest tube may not be able to enter the space. Finally, if the space is entered, the material is often fibrinous and does not drain. Therefore, traditional diagnostic techniques for parapneumonic empyema may not be applicable to post-traumatic empyema. Further, negative cultures occur in 25% to 55% of patients with empyema. Thus, the failure to obtain positive cultures should not delay treatment. The treatment of empyema is said to vary with the stage or chronicity. However, we generally find that post-traumatic empyema requires a thoracotomy for proper treatment. If a small, well-loculated collection is present and it can be drained with a tube, this is adequate treatment. If the patient does not improve, then operation should be undertaken. If the patient is diagnosed late, has a large empyema or thick peel on CT scan, or has major volume loss,

746

RICHARDSON et a1

operation should be immediately performed. Open drainage with small rib resection was formerly used in poor-risk patients but is rarely indicated for trauma patients. Some evidence suggests that complete removal of the peel associated with empyema may not be necessary.6,31Neff et a131used CT-guided drainage on 10 patients with intrathoracic abscess. At 3-month follow-up, four patients had normal CT scans, whereas the remaining six had mild changes. A long-term follow-up of 15 pediatric patients with empyema who were treated nonoperatively found no abnormalities on pulmonary function tests or exercise tolera n ~ eBlock . ~ ~ et a16 used radiologically guided percutaneous drainage for posttraumatic empyema in 12 patients with excellent results. However, all patients had primarily isolated penetrating chest trauma. These interesting findings probably do not apply to the multiply-injured, ventilator-dependent trauma patient. Also, we have found several of these patients to have necrotic lung or pulmonary abscess that required resection and would not have responded to nonoperative therapy.44Delaying definitive therapy by relying on antibiotics and nonoperative techniques only increases morbidity and mortality? We have found these techniques useful in an occasional postoperative patient with a small, loculated, persistent fluid collection.

References 1. Andrews NC, Parker EF, Shaw RR, et al: Management of nontuberculous empyema. Am Rev Respir Dis 85:35, 1962 2. Ashbaugh DG: Empyema thoracis. Factors influencing morbidity and mortality. Chest 99:1162, 1991 3. Aye RW, Froese DE, Hill L D Use of purified streptokinase in empyema and hemothorax. Am J Surg 161:560, 1991 4. Baumgartner F, Shepphard B, devirgilio C, et al: Tracheal and main bronchial disruptions after blunt chest trauma: Presentation and management. Ann Thorac Surg 50: 569,1990 5. Bertelsen S, Howitz P: Injuries of the trachea and bronchi. Thorax 27188, 1972 6. Block EF, Kirton OC, Windsor J, et a1 Guided percutaneous drainage for posttraumatic empyema thoracis. Surgery 117282, 1995 7. Bowling R, Mavroudis C, Richardson JD, et al: Emergency pneumonectomy for penetrating and blunt trauma. Am Surg 51:136, 1985 8. Carlon GC: Clinical experience with high frequency jet ventilation. Crit Care Med 9:42, 1981 9. Carrillo EH, Block EFJ, Zeppa R, et al: Urgent lobectomy and pneumonectomy after penetrating thoracic trauma. Eur J Emerg Med 1:126, 1994 10. Cheadle W, Richardson J D Options in management of trauma to the esophagus. Surg Gynecol Obstet 155:380, 1982 11. Coselli JS, Mattox KL, Beall AC: Reevaluation of early evacuation of clotted hemothorax. Am J Surg 148:786, 1984 12. Cryer HG, Mavroudis C, Yu J, et al: Shock, transfusion and pneumonectomy. Ann Surg 212197, 1990 13. Demetriades D, Charalambides D, Pantanowitz D, et al: Pneumopericardium following penetrating chest injuries. Arch Surg 125:1187, 1990 14. Eddy AC, Luna GK, Copass M: Empyema thoracis in patients undergoing emergent closed tube thoracostomy for thoracic trauma. Am J Surg 157494, 1989 15. Eloesser L: An operation for tuberculous empyema. Surg Gynecol Obstet 60:1096,1935 16. Estrera AS, Pass LJ, Platt MR Systemic arterial air embolism in penetrating lung injury. Ann Thorac Surg 50:257, 1990 17. Fallon WE: Post-traumatic empyema. J Am Coll Surg 179:483, 1994

COMPLEX THORACIC INJURIES

747

18. Falor WH, Traylor R Extended indications for the median sternotomy incision. Am Surg 48:582, 1982 19. Fisher RP, Geiger JP, Guernsey J M Pulmonary resections for severe pulmonary contusions secondary to high velocity missile wounds. J Trauma 14:293, 1974 20. Flynn AE, Verrier ED, Way LW, et al: Esophageal perforation. Arch Surg 1241211,1989 21. Gaensler EA, Strieder JW: Progressive changes in pulmonary function after pneumonectomy. J Thorac Surg 221, 1951

22. Gross BH, Spirarny D L Computed tomography of the chest in the intensive care unit. Crit Care Clin 10:267, 1994 23. Henry CHH DDS, Hills EC: Traumatic empyema of the head, neck, and mediastinum associated with maxillofacial trauma: Case report and review. J Oral Maxillofac Surg 47876, 1989 24. Hudgens S, McGraw J, Craun M: Two cases of tension pneumopericardium following blunt chest injury. J Trauma 31:1408, 1991 25. Jones WS, Mavroudis C, Richardson JD, et al: Management of tracheobronchial disruption resulting from blunt trauma. Surgery 95:319, 1984 26. Lucas AE, Snow N, Tobin GR, et al: Use of the rhomboid major muscle flap for esophageal repair. Ann Thorac Surg 33619, 1982 27. Mancini M, Smith LM, Nein A, et al: Early evacuation of clotted blood in hemothorax using thoracoscopy: Case reports. J Trauma 34:144, 1993 28. Marts B, Durham R, Shapiro M, et al: Computed tomography in the diagnosis of blunt chest injury. Am J Surg 169688, 1994 29. McGonigal MD, Schwab CW, Kauder DR, et al: Supplemental emergent chest computed tomography in the management of blunt torso trauma. J Trauma 301431,1990 30. Miller FB, Bond SJ, Shumate CR, et al: Diagnostic pericardial window-A safe alternative to exploratory thoracotomy for suspected heart injuries. Arch Surg 122:605, 1987 31. Neff CC, vansonnenberg E, Lawson DW, et al: CT follow-up of empyemas: Pleural peels resolve after percutaneous catheter drainage. Radiology 176195, 1990 32. 0:Brien J, Cohen M, Solit R, et al: Thorascopic drainage and decortication as definitive

treatment for empyema thoracis following penetrating chest trauma. J Trauma 36:536, 1994 33. Pickard LR, Mattox KL: Thoracic trauma: General considerations and indications for 34. 35. 36. 37. 38.

thoracotomy. In Moore EE, Mattox KL, Feliciano DV (eds): Trauma. Norwalk, CT, Appleton & Lange, 1991, pp 319-326 Pillgram-Larsen J, Lovstakken K, Hafsahi G, et al: Initial axial computerized tomography examination in chest injuries. Injury 24:182, 1993 Powell RJ, Redan JA, Swan KG: The hilar snare, an improved technique for securing rapid vascular control of the pulmonary hilum. J Trauma 30:208, 1990 Redding GJ, Walund L, Walund D, et al: Lung function in children following empyema. Am J Dis Child 1M1337, 1990 Richardson J D Indications for thoracotomy in thoracic trauma. Curr Surg 42:361, 1985 Richardson JD, Tobin G R Closure of esophageal defects with muscle flaps. Arch Surg

129:541, 1994 39. Robison PD, Harman PK, Trinkle JK, et al: Management of penetrating lung injuries in civilian practice. J Thorac Cardiovasc Surg 95:184, 1988 40. Sands DL, Ledgerwood AM, Lucas CE: Pneumomediastinum on a surgical service. Am Surg 54:434,1988 41. Schlaepfer K Air embolism following various diagnostic or therapeutic procedures in diseases of the pleura and the lung. Bull Johns Hopkins Hosp 33:321, 1922 42. Shields TW: Pulmonary resections. In Shields TW (ed): General Thoracic Surgery, ed 3. Philadelphia, Lea & Febiger, 1989, p 363 43. Simon RJ, Ivatury RR: Current concepts in the use of cavitary endoscopy in the

evaluation and treatment of blunt and penetrating truncal injuries. Surg Clin North Am 75:157, 1995 44. Snow N, Bergin KT, Horrigan TP: Thoracic CT scanning in critically ill patients. Information obtained frequently alters management. Chest 971467, 1990 45. Spencer JA, Rogers CE, Westaby S Clinico-radiological correlates in rupture of the major airways. Clin Radio1 43:371, 1991

748

RICHARDSON et al

46. Suarez J, Clagett OT, Brown AL The postpneumonectomy space. J Thorac Cardiovasc Surg 57539, 1969 47. Thompson DA, Rowlands BJ, Walker WE, et al: Urgent thoracotomy for pulmonary or tracheobronchial injury. J Trauma 28:276, 1988 48. Tominaga GT, Waxman K, Scanell G, et al: Emergency thoracotomy with lung resection following trauma. Am Surg 59:834, 1993 49. Unger JM, Schuchmann GO, Grossman J, et al: Tears of the trachea and main bronchi caused by blunt trauma: Radiologic findings. AJR Am J Roentgen01 153:1175, 1989 50. Urschel HC Jr, Razzuk MA, Wood RE, et al: Improved management of esophageal perforations: Exclusion and diversion in continuity. Ann Surg 179:587, 1974 51. Wahi R, McMurtrey MJ, Decaro LF: Determinants of perioperative morbidity and mortality after pneumonectomy. AM Thorac Surg 48:33, 1989 52. Washington 8, Wilson RF, Steiger Z, et al: Emergency thoracotomy: A four-year review. Ann Thorac Surg 40:188, 1985 53. Wieman DS, Walker WA, Brosnan KM, et al: Noniatrogenic esophageal trauma. AM Thorac Surg 59:845, 1995 54. Wienceck RJ, Wilson RF Central lung injuries: A need for early vascular control. J Trauma 28:1418, 1988 55. Wilson JW: Leukocyte sequestration and morphologic augmentation in the pulmonary network following hemorrhagic shock and related forms of stress. Adv Microcirc 4:197, 1972 56. Wong C, Huval W, Hechtman LT, et al: Effect of hemorrhagic shock on endotoxininduced pulmonary hypertension and increased vascular permeability in unanesthetized sheep. Circ Shock 12:61, 1984 57. Zakharia AT: Thoracic battle injuries in the Lebanon War: Review of the early operative approach in 1992 patients. Ann Thorac Surg 40:209, 1985 Address reprint requests to J. David Richardson, MD Department of Surgery University of Louisville Louisville, KY 40292