Accuracy of chest radiography in predicting pulmonary hypertension: a case-control study.

Objective: To assess the accuracy of chest radiography (CXR) in predicting pulmonary hypertension (PH).

Methods: We studied 108 consecutive patients with suspected PH who underwent right heart catheterization (RHC). All were PH treatment naives. Hemodynamic criteria included a mean pulmonary artery pressure >25 mmHg at rest, and a mean pulmonary wedge pressure <15 mmHg. Postero-anterior and lateral CXR were obtained shortly before RHC. To avoid a selection bias which could be introduced by examining only patients with suspected PH, we included in the analysis the CXR of 454 additional patients with different diagnosis: 57 with left heart failure (LHF) and pulmonary venous hypertension at RHC, 197 with chronic obstructive pulmonary disease, and 200 non-obstructed controls. CXR were examined independently by 4 raters, who were blinded to clinical, hemodynamic, and spirometric data. The diagnosis of PH was made if a prominent main pulmonary artery was associated with anyone of: isolated enlargement of right ventricle, right descending pulmonary artery >16 mm in diameter, pruning of peripheral pulmonary vessels.

Results: Eighty-two patients had PH confirmed at RHC. Weighted sensitivity of CXR was 96.9% (95% confidence interval, 94.9 to 98.2%), and weighted specificity 99.8% (95% confidence interval, 99.6 to 99.9%). By considering the 165 patients who underwent RHC, weighted sensitivity of CXR was unchanged, and weighted specificity decreased to 99.1%. None of the patients with PH were misclassified as having LHF, and vice versa.

Conclusions: CXR is accurate in predicting PH. It may aid clinicians in selecting patients with suspected PH for hemodynamic ascertainment.