PURPOSE: We sought to assess the yield of chest roentgenography for the detection of pneumothorax among hospitalized patients with pleural effusion who have undergone diagnostic or therapeutic thoracentesis. SUBJECTS AND METHODS: We performed a prospective study of 506 thoracentesis procedures in 370 patients. After the procedure, each operator filled out a note recording patient data and the characteristics of the thoracentesis. A chest radiograph was performed within 12 hours after the procedure in all patients. RESULTS: Eighteen (4%) pneumothoraces occurred in 17 patients, 9 (2%) of which required chest tube drainage. Of the 488 patients without symptoms, only 5 (1%) developed a pneumothorax, only 1 of which required chest tube drainage. By contrast, of the 18 patients with symptoms, 13 developed a pneumothorax, 8 of which required chest tubes. There were two independent predictors of pneumothorax: presence of symptoms (odds ratio [OR] = 250; 95% confidence interval [CI]: 65 to 980) and male gender (OR = 5.4; 95% CI: 1.9 to 69). CONCLUSIONS: Among the symptom-free patients in our sample, the risk of developing pneumothorax with clinical consequences was so low that the practice of routine chest roentgenography may not be justified.
PURPOSE: We sought to assess the yield of chest roentgenography for the detection of pneumothorax among hospitalized patients with pleural effusion who have undergone diagnostic or therapeutic thoracentesis. SUBJECTS AND METHODS: We performed a prospective study of 506 thoracentesis procedures in 370 patients. After the procedure, each operator filled out a note recording patient data and the characteristics of the thoracentesis. A chest radiograph was performed within 12 hours after the procedure in all patients. RESULTS: Eighteen (4%) pneumothoraces occurred in 17 patients, 9 (2%) of which required chest tube drainage. Of the 488 patients without symptoms, only 5 (1%) developed a pneumothorax, only 1 of which required chest tube drainage. By contrast, of the 18 patients with symptoms, 13 developed a pneumothorax, 8 of which required chest tubes. There were two independent predictors of pneumothorax: presence of symptoms (odds ratio [OR] = 250; 95% confidence interval [CI]: 65 to 980) and male gender (OR = 5.4; 95% CI: 1.9 to 69). CONCLUSIONS: Among the symptom-free patients in our sample, the risk of developing pneumothorax with clinical consequences was so low that the practice of routine chest roentgenography may not be justified.