Central adjudication of venous ultrasound in VTE screening trials: reasons for failure

Background: The accuracy of screening ultrasound for venous thrombosis in asymptomatic patients is still a matter of debate. The VENUS study evaluated the accuracy of centrally adjudicated venous ultrasound against venography in patients after major orthopedic surgery and found the sensitivity of ultrasound to be poor for both proximal and distal deep vein thrombus (DVT). Objectives: To evaluate whether thrombus characteristics such as location or size influence the diagnostic performance of centrally adjudicated venous ultrasound. Methods: All false negative sonograms of the VENUS study were re-evaluated against the corresponding venograms. Discrepancies were categorized into types of diagnostic failures. Within these categories, thrombus characteristics such as location, length or size of thrombus were evaluated. Results: One hundred and twelve pairs of discrepant ultrasound and venography documents were compared with 28 pairs with concordant results. Discrepancies were caused by local documentation failure (37.5%), failure of the ultrasound method (43.7%) and failure of the central adjudication process (18.7%). The overall size of thrombi was small, which caused about 40% of all sonographic failures with a detection threshold of five Marder points, a thrombus length of 9.5 cm and a number of 3.5 pathological compression manoeuvres. Proximal or distal location of DVT did not affect thrombus detection. Conclusion: If centrally adjudicated ultrasound is to be used in future VTE screening trials, training of local sonographers and central adjudicators needs to be intensified, because asymptomatic DVTs seem to be small and ultrasound sensitivity depends on the number of pathological compression manoeuvres documented in the ultrasound document. In contrast, distal or proximal thrombus location itself does not influence sensitivity.