Correlation of transesophageal ultrasound of the pericardium with computed tomography

In this study, we assess the sensitivity and specificity of ultrasound and computed tomography (CT) for pericardial effusion and constrictive pericarditis.

This was a retrospective, institutional review board-approved, and health insurance privacy accountability act compliant study performed at a single tertiary center over a 10-year period (2001-2011) for patients who had clinical symptoms of pericarditis and had undergone both cardiac CT imaging and transesophageal echocardiogram (TEE) in a span of 2 weeks.

Inclusion criteria included patients with clinical symptoms of pericarditis, pericardial thickness measuring more than 2 mm on CT, and patients who had both cardiac CT imaging and TEE performed within 2 weeks.

Exclusion criteria included patients with pericardial thickness measuring 2 mm or less on CT, no TEE, TEE not done within 2 weeks of the thoracic CT, and calcified pericardium on CT.Computed tomographic images were retrospectively reviewed by 2 radiologists who were unaware of the TEE findings. Pericardial effusion on CT was deemed present if there was obliteration of the fat plane in the left pulmonic recess.

Statistical analysis was performed using the R statistical environment (Rstat). Intraobserver and interobserver variability was estimated using Cohen κ- statistic (Cohen).

Forty-three cases constituted the study population (28 men and 15 women; mean age, 55 years; age range, 22-82 years). Twenty-one patients had pathologic confirmation of pericarditis.The findings for CT and TEE were discrepant in 10 cases. Intraobserver variability Cohen κ statistic was 0.855. Interobserver variability Cohen κ statistics were 0.54 and 0.49.

Computed tomography is sensitive to pericardial effusion and pericardial thickening, whereas TEE seems insensitive to isolated pericardial thickening.