Analysis of an algorithm incorporating limited and whole-leg assessment of the deep venous system in symptomatic outpatients with suspected deep-vein thrombosis (PALLADIO): a prospective, multicentre, cohort study

Background: Compression ultrasonography is the mainstay of diagnosis of deep-vein thrombosis (DVT) of the legs. Compression ultrasonography can be extended to the entire deep venous system (whole-leg) or restricted to the proximal veins only (limited), and the two approaches are clinically equivalent. We aimed to assess the diagnostic value of an algorithm combining whole-leg and limited compression ultrasonography. Methods: We did a prospective, multicentre, cohort study at eight centres in five countries. Consecutive outpatients aged 18 years or older with suspected DVT underwent D-dimer measurement and pretest clinical probability assessment. DVT was ruled out without further testing if pretest probability was unlikely and D-dimer was negative (group 1). Patients in whom either pretest probability was likely or who were positive for D-dimer underwent limited compression ultrasonography only (group 2). Finally, patients in whom pretest probability was likely and who had a positive measurement for D-dimer underwent extended whole-leg compression ultrasonography (group 3). All patients in whom DVT was ruled out were followed up for 3 months. The primary outcome was the incidence of objectively recorded venous thromboembolism. The primary analysis included all patients managed according to the study protocol. This study is registered with ClinicalTrials.gov, number NCT01412242. The final results are reported here. Findings: Between March 1, 2011, and July 31, 2014, 1348 consecutive outpatients were referred for this study, of whom 1162 were eligible to participate. After pretest probability assessment and D-dimer testing, 351 were in group 1, 401 in group 2, and 410 in group 3. Limited compression ultrasonography was positive in 12 (3%) patients in group 2 and extended whole-leg compression ultrasonography was positive in 200 (49%) patients in group 3. 82 (39%) of all DVT diagnosed at baseline were isolated distal thromboses. 26 protocol violations were reported. Thus, 351 patients from group 1, 371 patients in group 2, and 202 patients in group 3 who had been excluded for DVT by the algorithm were included in the primary analysis at 3 months. One, four, and three DVTs were reported, respectively. Thus, the 3-month incidence of venous thromboembolism in untreated patients after a negative diagnostic strategy was 0·87% (95% CI 0·44-1·70). Interpretation: An algorithm combining limited and whole-leg compression ultrasonography could be a reliable, safe, and convenient method for diagnostic management of outpatients with clinically suspected DVT. Funding: None.