Canadian Association of General Surgeons and American College of Surgeons Evidence Based Reviews in Surgery. 18. Treatment of occult pneumothoraces from blunt trauma

Question: Can occult pneumothoraces be safely observed without the need for a chest tube? Design: A randomized controlled trial. Setting: Two trauma centres in the United States. Patients: Thirty-nine patients with 44 pneumothoraces (defined as a pneumothorax seen on abdominal CT scan but not on an anteroposterior chest x-ray as read by the trauma chief resident or attending staff member) were enrolled. Intervention: Within 6 hours of admission, patients were randomized to receive a chest tube (n = 18, group 1) or observation (n = 21, group 2). Chest tubes remained in place for a median 3 days (range 1-12 d). The main outcome measures were: Respiratory distress, pneumothorax progression, pneumonia, retained hemothorax and chest tube insertion. Results: One nonintubated patient with a chest tube developed respiratory distress, and 3 who were being observed had respiratory distress without pneumothorax after these were removed from suction; 3 patients without chest tubes had pneumothorax progression, 2 while being ventilated. No differences were statistically significant. Conclusions: Pneumothoraces can be safely observed in patients with blunt trauma regardless of the need for positive pressure ventilation. The primary outcome measures were respiratory distress and pneumothorax progression; pneumonia, retained hemothoraces and a requirement for a chest tube placement were also tracked (Table 1). No significant differences in any outcomes were observed. One nonintubated patient with a chest tube developed respiratory distress from stridor, and 3 patients who were being observed developed respiratory distress without PTX progression. Thus, the rates of respiratory distress were 5.5% (1/18) and 14.3% (3/21) for the respective groups. This is actually a relative risk increase of 233% for the observation group, even though the difference was not statistically significant. Confidence intervals (CIs) are a way of measuring the precision of an estimate.15 They provide the range of values within which the true difference is likely to reside. If the range is wide, even though "O" lies within the range, the clinician is less likely to consider 2 treatments equivalent. The CI may be calculated with a formula or estimated with the "rule of 9." In the observed group of patients, the estimated upper limit of the 95% CI for respiratory distress could be as high as 9/21 = 0.429, or 43% of patients. If, indeed, this were the true rate of respiratory distress in the observed