Diagnostic yield of ultrasound and rule-out accuracy of D-dimer in fast-track diagnostic strategies for deep venous thrombosis
J Madsen, M Schmidt, M Wurtz, K Bonnesen, S A J SchmidtAbstract
Background and aim
Deep venous thrombosis (DVT) requires prompt diagnostic workup to prevent pulmonary embolism and post-thrombotic syndrome. In 2023, a Danish region implemented a fast-track diagnostic pathway for suspected DVT. This pathway allowed general practitioners to refer patients directly to the nearest regional hospital for same-day D-dimer testing and compression ultrasonography (CUS). Referrals were accepted on weekdays between 08:00 and 15:00, without prior evaluation in the emergency department. This approach accelerated DVT diagnostics but increased the use of radiologic resources. In 2025, one regional hospital revised this pathway by systematically implementing the Wells DVT score to improve the diagnostic yield of CUS. Patients with a Wells score of 1–2 underwent D-dimer testing, with CUS reserved for those with elevated D-dimer levels, whereas a score ≥3 prompted immediate CUS without prior D-dimer testing. We compared the diagnostic yield of CUS and the rule-out performance of D-dimer across different fast-track strategies.
Methods
We conducted a multicenter cross-sectional diagnostic accuracy study of primary care patients aged ≥18 years with suspected DVT who received CUS in a Danish region during 2021–2025 (n=26,500). Data on imaging, laboratory, and diagnoses were obtained from a regional data warehouse. Diagnostic yield was defined as the proportion of CUS that were positive. Using age-adjusted cut-offs, we assessed the negative predictive value (NPV) of D-dimer for ruling out DVT. Outcomes were stratified by time periods (before vs. after fast-track implementation) and fast-track strategies (initial vs. modified).
Results
Diagnostic yield of CUS was 7.4% (974 DVTs/13,198 CUSs) before and 10.3% (888/8,588) after implementation of fast-track pathway. It was 10% (845/8,347) with the initial fast-track strategy and 18% (43/241) with the modified strategy. The NPV of D-dimer was 95% (95% confidence interval 94–96%) before fast-track implementation, 96% (96–97%) overall after fast-track implementation, and 97% (96–97%) with the initial fast-track strategy.
Conclusion
The modified fast-track strategy, incorporating systematic use of the Wells DVT score and D-dimer–based risk stratification, nearly doubled the diagnostic yield of CUS from 10% to 18%. D-dimer demonstrated consistently high rule-out performance across diagnostic strategies. The modified strategy appears to be the most cost-effective approach, and broader implementation may reduce resource utilisation.