Optimizing the D-dimer Threshold Used to Exclude Venous Thromboembolism

Abstract

Background: A D-dimer threshold <500ug/L has high negative predictive value (NPV) for venous thromboembolism (VTE), but is non-specific. Two strategies increase the specificity and utility (defined as the proportion of patients with a negative test) of D-dimer testing: 1) using a higher D-dimer threshold with increasing age (IAIT Strategy); and 2) using a high threshold in low clinical pretest probability (CPTP) patients and the standard threshold in moderate CPTP patients (CPTP Strategy). It is unknown whether the gain in specificity of the IAIT Strategy is simply due to using a higher threshold in some patients and whether the CPTP Strategy has better diagnostic accuracy than the IAIT Strategy. Methods: In a retrospective analysis of 1649 outpatients with suspected VTE, I compared the diagnostic accuracy of the IAIT Strategy to 1) its opposite: using a higher D-dimer threshold with decreasing age (DAIT strategy); 2) using a higher D-dimer threshold in all patients (Median Age Strategy); and 3) the CPTP Strategy. Results: The NPV of both the IAIT and DAIT Strategies was 99.6% and the NPV of the Median Age Strategy was 99.7%. The utility was almost identical in the IAIT and DAIT Strategies (50.9% vs. 50.6%) and greater in the Median Age Strategy (53.9%, p<0.001). The NPV of the CPTP and IAIT Strategies were 99.6% and 99.7%, respectively. The utility was higher in the CPTP Strategy than the IAIT Strategy (56.1% vs. 50.9%, p<0.001). Conclusions: The NPV and utility of using a higher D-dimer threshold in older patients (IAIT Strategy) is the same as using a higher D-dimer threshold in younger patients. The CPTP Strategy had the greatest utility while maintaining a high NPV and therefore appeared to be the optimal strategy of D-dimer interpretation.