OC 11.4 - Diagnosis of deep vein thrombosis using a D-dimer threshold adjusted to clinical probability: supportive real-world evidence for a change in clinical practice

Background: Patients presenting with suspected deep vein thrombosis (DVT) are risk-stratified by the Wells’ score and plasma D-dimer. However, whilst the majority of patients require imaging, only a small minority are diagnosed with DVT. Kearon and colleagues have proposed an updated algorithm, using different D-dimer thresholds based on the initial Wells’ score, to determine the need for first or second scans.

Aims: We aimed to assess the performance of this novel algorithm in a retrospective cohort, to establish the rates of DVTs that would have been missed, and the reduction ultrasound scans per patient.

Methods: We reviewed clinical records of 12,365 presentations of suspected DVT to our specialist clinic, from 1st January 2012 to 31st July 2022, for whom a complete Wells’ score and D-dimer were available.

Results: Of 12,365 presentations (Table 1), there were 1,315 DVT diagnoses. A further 20 patients were diagnosed at re-presentation within 90 days [0.2% rate of missed DVTs (95% confidence interval 0.20-0.20%)]. In this cohort, the modified algorithm (Figure 1) would have detected 1,268 DVT cases, missing 47 patients (and also the 20 patients who re-presented). This is a 0.61% rate of missed DVTs (95% confidence interval 0.61-0.62%).

The conventional algorithm would lead to 19,277 ultrasound scans [mean of 1.56 (95% confidence interval 1.55-1.57) scans per patient]. In comparison, the modified algorithm would lead to 11,248 ultrasound scans [mean of 0.91 (95% confidence interval 0.90-0.92) scans per patient]. Taken together, the modified algorithm reduces number of scans by 42%, missing one DVT diagnosis for every 171 scans saved compared to the conventional approach.

Conclusion(s): The modified algorithm can detect the vast majority of DVTs and maintain a low rate of missed DVTs, whilst very significantly reducing the number of ultrasounds required. This could save resources for healthcare systems, and travel time and uncertainty for patients.