Selective D-dimer testing for diagnosis of a first suspected episode of deep venous thrombosis: a randomized trial.

BACKGROUND: D-Dimer testing is sensitive but not specific for diagnosing deep venous thrombosis (DVT). Changing the use of testing and the threshold level for a positive test result on the basis of risk for DVT might improve the tradeoff between sensitivity and specificity and reduce the need for testing. OBJECTIVE: To determine whether using a selective D-dimer testing strategy based on clinical pretest probability (C-PTP) for DVT is safe and reduces diagnostic testing compared with using a single D-dimer threshold for all patients. DESIGN: Randomized, multicenter, controlled trial. Patients were allocated using a central automated system. Ultrasonographers and study adjudicators but not other study personnel were blinded to trial allocation. (ClinicalTrials.gov: NCT00157677) SETTING: 5 Canadian hospitals. PATIENTS: Consecutive symptomatic patients with a first episode of suspected DVT. INTERVENTION: Selective testing (n = 860), defined as D-dimer testing for outpatients with low or moderate C-PTP (DVT excluded at D-dimer levels <1.0 µg/mL [low C-PTP] or <0.5 µg/mL [moderate C-PTP]) and venous ultrasonography without D-dimer testing for outpatients with high C-PTP and inpatients, or uniform testing (n = 863), defined as D-dimer testing for all participants (DVT excluded at D-dimer levels <0.5 µg/mL). MEASUREMENTS: The proportion of patients not diagnosed with DVT during initial testing who had symptomatic venous thromboembolism during 3-month follow-up and the proportion of patients undergoing D-dimer testing and ultrasonography. RESULTS: The incidence of symptomatic venous thromboembolism at 3 months was 0.5% in both study groups (difference, 0.0 percentage point [95% CI, -0.8 to 0.8 percentage points]). Selective testing reduced the proportion of patients who required D-dimer testing by 21.8 percentage points (CI, 19.1 to 24.8 percentage points). It reduced the proportion who required ultrasonography by 7.6 percentage points (CI, 2.9 to 12.2 percentage points) overall and by 21.0 percentage points (CI, 14.2 to 27.6 percentage points) in outpatients with low C-PTP. LIMITATION: Results may not be generalizable to all D-dimer assays or patients with previous DVT, study personnel were not blinded, and the trial was stopped prematurely. CONCLUSION: A selective D-dimer testing strategy seems as safe as and more efficient than having everyone undergo D-dimer testing when diagnosing a first episode of suspected DVT. PRIMARY FUNDING SOURCE: Heart and Stroke Foundation of Ontario.