Improving the diagnostic management of upper extremity deep vein thrombosis.

Background Among patients with clinically suspected upper extremity deep vein thrombosis (UEDVT), a clinical decision rule based on the Constans score combined with D-dimer testing can safely rule out the diagnosis without imaging in approximately one-fifth of patients. Objectives To evaluate the performance of the original Constans score, an extended Constans score and an age-adjusted D-dimer positivity threshold. Methods Data of 406 patients with suspected UEDVT previously enrolled in a multinational diagnostic management study were used. The discriminatory performance, calibration and diagnostic accuracy of the Constans score were evaluated. The Constans score was extended by selecting clinical variables that may have incremental value in detecting UEDVT, conditional on the original Constans score items. The performance of the Constans rule was evaluated in combination with fixed and age-adjusted D-dimer thresholds. Results The original Constans score showed good discriminatory performance (c-statistic, 0.81; 95% confidence interval [CI], 0.76-0.85). An extended Constans score with five additional clinical items improved discriminatory performance and calibration, but this did not translate into a higher efficiency in avoiding imaging tests. Compared with a fixed threshold, age-adjusted D-dimer testing increased the proportion of patients for whom imaging and anticoagulation could be withheld from 21% to 25% (gain, 3.7%; 95% CI, 2.3-6.0%). Conclusions The Constans score has good discriminatory performance in the diagnosis of UEDVT. Age-adjusted D-dimer testing is likely to safely increase the efficiency of the diagnostic algorithm, but this approach needs prospective validation.