Predicting the need for thoracoscopic evacuation of residual traumatic hemothorax: chest radiograph is insufficient.

BACKGROUND The early removal of large residual posttraumatic hemothorax by videothoracoscopy is increasingly used to avoid the late sequelae of trapped lung and empyema. Plain chest radiography (CXR) is the tool most frequently used to select such cases for operation. Our recent experience has demonstrated that what appears to be a large retained hemothorax on CXR may turn out to be intrapulmonary or extrapleural conditions not amenable to thoracoscopic removal. Our objective was to evaluate the accuracy of CXR in detecting significant residual hemothorax and compare its clinical value to thoracic computed tomography (CT) when used to select patients for thoracoscopic evacuation. METHODS All patients requiring tube thoracostomy for traumatic hemothorax were prospectively evaluated during a 22-month period (n = 703). Patients who, on the second day after admission, demonstrated opacification on CXR involving more than the costophrenic angle were evaluated by thoracic computed tomography for the presence of undrained fluid. Second-day CXR (CXR2) results were compared with the CT findings. Incorrect interpretation was defined as a difference of more than 300 mL between the two readings. All CXR2 and CT results were reviewed in the same fashion by a radiologist blinded to the surgeon's interpretations. Data on injury mechanism, hemodynamic status, laboratory values, interventions, and outcome were collected prospectively. RESULTS Fifty-eight patients had clinically significant opacifications on CXR2. The surgeon's and radiologist's CXR2 interpretations were incorrect in 48 and 47% of the cases, respectively. The CT interpretations by the two specialists were in agreement in 97% of the cases. Management that would have been instituted on the basis of CXR2 findings was changed in 18 cases (31%). Twelve patients (21%) required early thoracoscopic evacuation of undrained collections. There was good correlation between the CT estimation and the thoracoscopically retrieved amount of blood. CONCLUSION Although CXR is useful as a screening tool, it cannot be used to reliably select patients for surgical evacuation of retained traumatic hemothorax. Decision-making should be based on thoracic CT findings.