Can routine chest radiography be used to diagnose mild COPD? A nested case-control study.

OBJECTIVES: To determine whether mild stage chronic obstructive pulmonary disease (COPD) can be detected on chest radiography without substantial overdiagnosis. METHODS: A retrospective nested case-control study (case:control, 1:1) was performed in 783 patients scheduled for cardiothoracic surgery who underwent both spirometry and a chest radiograph preoperative. Diagnostic accuracy of chest radiography for diagnosing mild COPD was investigated using objective measurements and overall appearance specific for COPD on chest radiography. Inter-observer variability was investigated and variables with a kappa >0.40 as well as baseline characteristics were used to make a diagnostic model which was aimed at achieving a high positive predictive value (PPV). RESULTS: Twenty percent (155/783) had COPD. The PPV of overall appearance specific for COPD alone was low (37-55%). Factors in the diagnostic model were age, type of surgery, gender, distance of the right diaphragm apex to the first rib, retrosternal space, sternodiaphragmatic angle, maximum height right diaphragm (lateral view) and subjective impression of COPD (using both views). The model resulted in a PPV of 100%, negative predictive value (NPV) of 82%, sensitivity of 10% and specificity of 100% with an area under the curve of 0.811. CONCLUSIONS: Detection of mild COPD without substantial overdiagnosis was not feasible on chest radiographs in our cohort.