LAUSR

RATIONALE AND OBJECTIVES Basal lung opacities are frequently observed on supine chest x-ray (SCXR) of intensive care patients, causing insecurity among clinicians and radiologists. We sought to determine the diagnostic accuracy of SCXR for basal pneumonia. MATERIALS AND METHODS We identified 172 patients who received both SCXR and computed tomography within 1 hour. Two readers examined the SCXR and rated findings in both basal zones according to the following scale: 0 = "no pneumonia," 1 = "possible pneumonia," 2 = "highly suspected pneumonia." Computed tomography served as standard of reference. Sensitivity, specificity, and positive and negative predictive values (PPV/NPV) were calculated once pooling 0 and 1 as negative and once pooling 1 and 2 as positive finding. RESULTS When pooling 0 and 1 as negative, sensitivity was 0.45 (right)/0.38 (left), specificity was 0.94/0.97, PPV was 0.76/0.79, and NPV was 0.81/0.84. When pooling 1 and 2 as positive, sensitivity was 0.80/0.75, specificity was 0.62/0.58, PPV was 0.45/0.35, and NPV was 0.88/0.89. The most common findings in false-positive cases were combined pleural effusions and lower lobe atelectasis. CONCLUSIONS Interpreting only highly suspicious basal opacities as pneumonia considerably increases the PPV with almost constant NPV. Clinicians and radiologists should be aware of the limitations of SCXR regarding basal pneumonia.