Ultrasound diagnosis of occult pneumothorax.

OBJECTIVES Pneumothorax can be missed by bedside radiography, and computed tomography is the current alternative. We asked whether lung ultrasound could be of any help in this situation. DESIGN Retrospective study. SETTING The medical intensive care unit of a university-affiliated teaching hospital. PATIENTS All patients admitted to the intensive care unit are routinely scanned with whole-body ultrasound (including screening for pneumothorax) and chest radiography. The study population included 200 consecutive undifferentiated intensive care unit patients who received a chest computed tomography scan in addition to ultrasound and chest radiograph. Forty-seven consecutive cases of radioccult pneumothorax were compared with 310 consecutive hemithoraces free from pneumothorax in the intensive care unit. INTERVENTIONS None. MEASUREMENTS AND RESULTS Three signs were investigated at the anterolateral chest wall in supine patients: lung sliding, the A line sign, and the lung point. A total of 357 hemithoraces were analyzed in this study, 47 with occult pneumothorax and 310 controls. Four of the 47 cases of pneumothorax were excluded from the final analysis (parietal emphysema) as well as eight of the 310 controls (large dressings), leaving a final study population of 345 hemithoraces in 197 patients. Feasibility was 98%. Ultrasound scans in all 43 examinable patients with pneumothorax showed absent lung sliding, 41 of 43 patients had the A line sign, and 34 exhibited a lung point. Among 302 analyzable controls, 65 had absent lung sliding, 16 of them showed an A line sign, and none showed a lung point. For the diagnosis of occult pneumothorax, the abolition of lung sliding alone had a sensitivity of 100% and a specificity of 78%. Absent lung sliding plus the A line sign had a sensitivity of 95% and a specificity of 94%. The lung point had a sensitivity of 79% and a specificity of 100%. CONCLUSIONS For the diagnosis of occult pneumothorax, ultrasound can decrease the need for computed tomography.