Roentgenography remains the standard imaging modality after thoracic surgery. Trials from intensive medicine proved a high accuracy of ultrasound examination in the diagnosis of various conditions. The assumption was that ultrasound examination could reduce the number of roentgenograms after thoracic surgery. This prospective study compared ultrasound examinations performed by thoracic surgeons with roentgenograms in the diagnosis of pneumothorax and pleural effusion after noncardiac thoracic surgery. Patients received 2 ultrasound scans, the first on the day of surgery and the second before chest tube removal. A total of 297 patients underwent 545 examinations; 336 ultrasound scans (61.6%) showed neither pneumothorax nor pleural effusion. Pneumothorax was detected on 69 roentgenograms and 51 ultrasound scans. Both modalities showed positive results in 32 cases and negative results in 395 cases (Cohen's κ, 53.4%). Ultrasound missed 37 clinically irrelevant pneumothoraces. Roentgenograms missed 19 pneumothoraces; 15 of themwere clinically irrelevant. Sensitivity and specificity were 59.4% and 95.9% in the first examination and 50.0% and 94.8% in the second examination, respectively. Pleural effusion was detected on 169 roentgenograms and117 ultrasound scans. Both modalities showed positiveresults in 88 cases and negative results in 336 cases (Cohen's κ, 49.6%). Ultrasound scans missed 81 pleural effusions; except for 5 cases, the clinical decisions would not have changed. Roentgenograms missed 29 clinically irrelevant pleural effusions. Sensitivity and specificity were 44.4% and 92.6% in the first examination and 60.9% and 91.3% in the second examination, respectively. Given high specificities, a large share of results without pneumothorax and pleural effusion, and mismatch analysis, we could reduce the number of roentgenograms by 61.6% by using ultrasound as a primary imaging modality.
Read full abstract