Background: As utilization of computed tomography (CT) scans in the evaluation of trauma patients increases, pneumothoraces (PTXs) seen on CT but not on chest X-ray (CXR), known as occult PTXs (OPTXs), are becoming more prevalent. The incidence of PTXs simply missed on CXR among OPTXs is unclear. A previous retrospective review of CXRs at our institution generally confirmed the occult versus missed designation, but lower fidelity images may have biased this determination. Thus, we repeated this evaluation using the high-quality images and improved the methodology.
Methods: The 70 Digital Imaging and Communications in Medicine (DICOM)-quality CXR images were randomly selected from two prospectively collected trauma databases including 22 normal, 5 overt PTX, and 43 study OPTX images. All CXR images were corroborated with multidetector CT imaging. Two blinded fellowship-trained radiologists reviewed and evaluated all the images on an IMPAX viewer.
Results: All images were deemed "adequate" except for one CXR by a single reviewer. For PTX diagnosis, agreement was 60% for overt PTXs, 86% for normal CXRs, and 81% for study OPTXs, yielding a kappa statistic of 0.51 (95% confidence interval, 0.22-0.81) indicating moderate agreement. Considering only the cases where the reviewers agreed, 80% of the study OPTXs were truly occult versus missed (95% confidence interval, 63-92%). In the 7 missed PTXs, subcutaneous emphysema (5), pleural line (3), and deep sulcus sign (2) were detected.
Conclusion: We estimate that 80% of PTXs considered occult in the trauma room were truly occult. The most common missed sign was subcutaneous emphysema. PTXs are poorly assessed by CXR, and accurate diagnosis should focus on other imaging modalities.