VENTILATOR-ASSOCIATED PNEUMONIA IN PATIENTS WITH ADULT-RESPIRATORY-DISTRESS-SYNDROME - CT EVALUATION

PURPOSE: To determine the diagnostic accuracy of computed tomography ( CT) for pneumonia in patients with adult respiratory distress syndrome (ARDS). MATERIALS AND METHODS: CT scans were obtained within 1 week o f bronchoscopic sampling in 31 patients receiving mechanical :ventilat ion for ARDS for more than 48 hours. Of 11 patients with pneumonia, fi ve developed symptoms less than 11 days after the onset of ARDS (early ARDS). CT scans were rated for pneumonia independently by four radiol ogists who were unaware of the clinical diagnosis. Diagnostic accuracy was defined: by means of the area under the receiver operating charac teristic curve of A(z). RESULTS: Diagnostic accuracy for pneumonia was fair (A(z) = 0.69 +/- 0.4 [standard error]) owing to 70% true-negativ e ratings (vs 59% true-positive ratings).:The generalizability coeffic ient was good (0.79). No-single CT finding was Significantly different for the presence of pneumonia. Nondependent opacities predominated in 10 (91%) of 11 patients with pneumonia and 12 (60%) of 20 without pne umonia. Nondependent opacities predominated in nine (56%) of 16 patien ts with early ARDS and 13 (87%) of 15 with late ARDS. CONCLUSION: CT h as fair diagnostic accuracy for ventilator-associated pneumonia in pat ients with ARDS owing primarily to identification of patients without pneumonia. No single CT sign was significantly different for pneumonia , but dependent atelectasis was more common in patients with early ARD S without pneumonia.