The accuracy of thoracic ultrasound in differentiating transudative from exudative effusions: A meta-analysis.

BACKGROUND

Literature on the diagnostic accuracy of thoracic ultrasound (TUS) in differentiating transudates from exudates is sparse. We sought to summarize available evidence on the accuracy of TUS at distinguishing between transudative and exudative pleural effusions.

STUDY DESIGN AND

METHODS

We included studies of patients with pleural effusions who underwent TUS and a thoracentesis with pleural fluid analysis. We performed meta-analyses using a bivariate model to estimate and compare summary sensitivities, specificities, and positive and negative predictive values. Summary receiver operating characteristic (ROC) plots were created to visualize the overall diagnostic accuracy of TUS for complex and anechoic patterns.

RESULTS

We included five studies in the final analysis. A total of 1422 effusions were included in the final analysis (491 anechoic transudates, 300 anechoic exudates, 34 echogenic transudates, and 597 echogenic exudates). An echogenic TUS pattern for detecting an exudate demonstrated a pooled sensitivity of 0.71 (95 % CI 0.57 to 0.82) and specificity of 0.92 (95 % CI 0.59 to 0.99), with a diagnostic odds ratio of 34.82 (95 % CI 4.23 to 286.42). The overall performance showed an area under the curve (AUC) of 0.81 and a partial AUC (pAUC) of 0.77. Significant heterogeneity was observed across studies (I2 = 90.9 % for sensitivity and 91.9 % for specificity).

INTERPRETATION: An echogenic TUS pattern can reliably identify an exudative effusion, but an anechoic TUS pattern cannot reliably identify a transudative effusion. In cases of diagnostic uncertainty, thoracentesis and pleural fluid analysis are still needed to correctly classify the effusion.