ROCKSTARS: Ultrasound vs Chest X-ray in the Detection of Traumatic Pneumothorax

brownsound 2

Picture this…A 45 year old male is activated as a trauma after falling off 30 feet off scaffolding. He is flown from the scene and the report from the paramedics notes concern for spinal cord injury, as he has no sensation below the nipple line and flaccid extremities. On primary exam the resident notes diminished breath sounds on the right, however he asks the question to the rest of the trauma team: “are the decreased breath sounds from a pneumo/hemothorax or from a partially paralyzed/weakened diaphragm, in a patient with a high c-spine injury”. In a minute of clinical reasoning in an otherwise algorithmic setting, the resident challenges his clinical exam (inspection, palpation, auscultation) and places an ultrasound on the patient acquiring the following images:

Video: Lung point–the most specific sign for pneumothorax on thoracic ultrasound.

Lung Sliding

Image 1: M-mode of lung point showing both seashore and stratosphere

A chest tube is placed and the remainder of the trauma survey proceeds.

At our institution, we wait for a chest x-ray that shows no pneumothorax before sending a patient to the CT scanner. This is despite the studies that show the sensitivity of supine chest x-rays is 28 % to 75%.[i] This begs the question, would a bedside ultrasound be quicker and more sensitive to evaluate for a pneumothorax. A quick review of the literature says a definitive yes.

Three articles are at the core of the US for pneumothorax evidence. First in 2005, Blaivas et al published a paper that used EM attendings in a prospective single blinded trial to evaluate 176 patients for pneumothorax. They used 4 protocol views in each hemithorax (2nd intercostal mid clavicular, 4th intercostal ant. Axillary, 6th intercostal midaxillary and 6th intercostal post axillary) and evaluated only for lung sliding. No M-mode, no doppler. Their results were compared to the Trauma attending read of the supine CXR and the Radiologist’s view of the chest CT.

In this study: the sensitivity and specificity for thoracic US was :

Ultrasound (95% CI) CXR (95% CI)
Sensitivity 98.1% (89.9%-99.9%) 75.5% (61.7-86.2%)
Specificity 99.2% (95.6%-99.9%) 100% (97.1%-100%)

Wilkerson and Stone in 2009 published a meta-analysis of 4 trials including the Blaivas trial noted above. This meta-analysis looked at EM physicians as US operators in the analyses of pneumothorax or no pneumothorax in the setting of trauma. This study again found superior outcomes for thoracic ultrasound:

Ultrasound CXR
Sensitivity 86%-98% 28%-75%
Specificity 97%-100% 100%

Lastly in 2012, Hyacinthe et al. published a paper that aimed to assess the ability of thoracic ultrasound to detect, on arrival, the occurrence of common thoracic lesions in a cohort of chest trauma patients. This is likely the most relevant study as the methods consisted of a prospective observational cohort study where two separate EM physicians were used. First the physician taking care of the patient primarily used the Clinical Exam (Inspection, palpation, percussion and auscultation) and chest x-ray to determine the presence of pneumothorax. The physician performed their exam, looked at the supine chest x-ray then was asked to give the patient a score of how likely they are to have a thoracic lesion (0=no chance, 3=sure presence of lesion). A separate EM attending blinded to the initial exam and CXR then performed a thoracic ultrasound in both lung fields, including the upper, middle and lower parts of the anterior and lateral regions of both chest walls. Pneumothorax was defined as the absence of lung sliding or by the presence of lung point. The performing physician then, in a similar way, recorded their findings on a scale from 0-3. Results for this study included the sensitivity and specificity for each modality however, given the scales of probability entered by each physician, the more encompassing statistic is the area under the curve (AUC) for each modality.

Ultrasound CE+CXR
Sensitivity 53% 19%
Specificity 95% 100%
Area Under the Curve, mean (95% CI) 0.75 (0.67-0.83) 0.62 (0.54-0.70)

These articles make a strong case for the increased use of ultrasound in trauma. In the hands of an experienced user, a bilateral thoracic ultrasound takes 2-4 minutes, is arguably shorter than the time to call an x-ray tech, shoot the x-ray, develop the images and walk down the hall to view them, not to mention the test is overwhelmingly more sensitive.


[i] Gentry Wilkerson, R. and Stone, M. B. (2010), Sensitivity of Bedside Ultrasound and Supine Anteroposterior Chest Radiographs for the Identification of Pneumothorax After Blunt Trauma. Academic Emergency Medicine, 17: 11–17. doi: 10.1111/j.1553-2712.2009.00628.x

[ii]Blaivas, M., Lyon, M. and Duggal, S. (2005), A Prospective Comparison of Supine Chest Radiography and Bedside Ultrasound for the Diagnosis of Traumatic Pneumothorax. Academic Emergency Medicine, 12: 844–849. doi: 10.1197/j.aem.2005.05.005

[iii]Hyacinthe AC, Broux C, Francony G, et al. Diagnostic accuracy of ultrasonography in the acute assessment of common thoracic lesions after trauma.Chest. 2012;141(5):1177–83


Leave a Reply