What is the best management strategy for patients with suspected deep vein thrombosis?

The most cost-effective algorithm for managing patients with suspected deep vein thrombosis (DVT) was identified, although several are nearly as good. The main message is that the best approach uses a combination of a validated clinical decision rule, D-dimer test, and venous ultrasound. (LOE = 1b)

Goodacre S, Stevenson M, Wailoo A, Sampson F, Sutton AJ, Thomas S. How should we diagnose suspected deep-vein thrombosis? QJM 2006;99:377-388.  [PMID:16728440]

Cost-effectiveness analysis

Government

Various (meta-analysis)

These authors systematically reviewed the literature and identified 18 different strategies for managing patients with suspected DVT. They then evaluated a hypothetical group of 1000 patients using each algorithm, applying reasonable estimates of test accuracy, treatment effectiveness, and the cost of testing and treatment. This included the cost of applying a clinical decision rule, such as the Wells rule. They estimated a mean survival of 11.6 quality adjusted life years (QALYs) after diagnosis of DVT at age 60. The estimates were reasonable and most of the algorithms used some combination of the Wells rule, D-dimer, and venous ultrasound. The percentage of patients with proximal DVT who would be treated appropriately by the algorithms ranged from 90.1% to 99.5%, and the patients without DVT treated inappropriately ranged from 0.6% to 6.0%. The optimal algorithm used a latex D-dimer test as the initial screen. If patients were D-dimer negative and had a low or intermediate Wells score, DVT was considered ruled out. If they were D-dimer negative, but at high risk based on the Wells score, an above knee venous ultrasound was ordered. If the ultrasound was positive, they were treated; if negative, a repeat ultrasound was ordered. Patients who were D-dimer positive, underwent ultrasound: if positive, they were treated; if negative, they had repeat ultrasound. Several other algorithms were similarly cost-effective.