Interobserver reliability of compression ultrasound for residual thrombosis after first unprovoked deep vein thrombosis

Abstract

Accurate assessment of residual thrombosis is of clinical importance for diagnostic baseline imaging, and may be of value in risk stratification for recurrent venous thromboembolism (VTE). This study evaluated the interobserver reliability of the measurement of residual thrombosis in patients 6 months after a first unprovoked deep vein thrombosis (DVT) of the leg. All enrolled patients received two ultrasound examinations by two independent blinded ultrasound technicians 5-7 months after their first unprovoked DVT. In total, 49 patients completed the two baseline ultrasound examinations. During the examinations, the presence of residual thrombosis was evaluated. If residual thrombosis was present, a detailed description of the size and location was reported. After all ultrasound results had been collected, the interobserver agreement was calculated by use of the kappa statistics, Pearson correlation, and the Bland-Altman plot. Furthermore, the clinical implications of interobserver reliability were examined. The interobserver reliability of the assessment of whether residual thrombosis is present was very good (κ = 0.92). The interobserver reliability of the measurement of residual thrombosis was good (r2 = 0.648), with a limited number of patients being misclassified. For the assessment of the percentage of residual occlusion, the interobserver reliability was fair (r2 = 0.357). Our results suggest that the interobserver reliability for measurement of residual thrombosis is high, and that the variability introduced by interobserver reliability has minimal clinical implications. Our study is important for the use of baseline imaging for the diagnostic and prognostic management of recurrent VTE.