Background: Although D-dimer assays have high negative predictive values for the diagnosis of deep venous thrombosis, their accuracy in patients with cancer is uncertain.
Objective: To compare the clinical utility of a whole-blood D-dimer assay for the diagnosis of deep venous thrombosis in patients with and those without cancer.
Design: Retrospective analysis of three prospective studies.
Setting: Two tertiary care hospitals.
Patients: 1068 consecutive outpatients with suspected deep venous thrombosis.
Measurements: All patients underwent D-dimer testing and assessment with a priori diagnostic strategies that incorporated impedance plethysmography, compression ultrasonography, or contrast venography. Patients in whom deep venous thrombosis was not diagnosed initially were followed for 3 months for the development of thrombosis. Results of D-dimer testing were assessed according to the final diagnosis based on objective testing and clinical follow-up. Cancer status was identified at presentation.
Results: The prevalence of deep venous thrombosis was 48.8% in 121 patients with cancer and 14.6% in 947 patients without cancer. Although the sensitivity of the D-dimer assay was similar in patients with and those without cancer (86.4% [95% CI, 75.0% to 94.0%] and 82.6% [CI, 75.2% to 88.5%], respectively), the specificity was significantly lower in patients with cancer (48.4% [CI, 35.5% to 61.4%] and 82.2% [CI, 79.4% to 84.8%]), as was the negative predictive value (78.9% [CI, 62.7% to 90.4%] and 96.5% [CI, 94.9% to 97.8%]). In contrast, the likelihood ratios of a negative test result (0.28 [CI, 0.14 to 0.56] and 0.21 [CI, 0.15 to 0.31]) did not differ significantly.
Conclusions: A negative D-dimer test result in patients with cancer does not reliably exclude deep venous thrombosis because the negative predictive value of the test is significantly lower in these patients than in patients without cancer.