Can pulmonary venous hypertension be graded by the chest radiograph?

The chest radiograph is widely used for the diagnosis and follow-up of patients with pulmonary venous hypertension (PVH) caused by rheumatic heart disease. Observer accuracy and interobserver agreement on the radiographic diagnosis of PVH is rarely assessed. We have studied this accuracy and agreement in 120 consecutive patients.

Patients were selected on the basis of mean pulmonary capillary wedge pressures (PCWP) obtained in the basal state. A chest radiograph was obtained within the 24 h before catheterization and was assessed by two observers blinded to the diagnosis and PCWP. PVH was graded as absent (PCWP < 12 mmHg); mild (PCWP 12-19 mmHg; redistribution of blood flow and hilar haze); moderate (PCWP 20-25 mmHg; septal oedema, pleural effusion) and severe (PCWP > 25 mmHg; alveolar oedema). Observer and interobserver agreement was quantified by using the Kappa (kappa) statistic. The radiological assessment coincided with the measured PCWP in 52% of the patients according to observer 1 and 43% of the patients according to observer 2 (kappa 0.36 and 0.24), indicating poor agreement with haemodynamic values. Interobserver agreement was poor (kappa 0.3-0.35). Observer accuracy and interobserver agreement increased when the radiographic criteria were used to distinguish patients with normal PCWP from those with abnormal pressures when 85% and 78% of radiographs were correctly classified (kappa 0.6 and 0.5) by observers 1 and 2, respectively. Overall agreement was significant (kappa 0.6). In differentiating patients with a PCWP > 20 mmHg from those with a PCWP < 20 mmHg, 78% and 75% of radiographs were correctly classified, giving kappa values of 0.6 and 0.5. Overall observer agreement was significant (kappa 0.6).

The chest radiograph is useful in detecting the presence of PVH and in distinguishing significant elevations in PCWP (> 20 mmHg) from mildly elevated PCWP. There is poor correlation between haemodynamic and radiographic findings when more precise grading of the severity of PVH is attempted.