Adjusting How We Diagnose Deep Vein Thrombosis

Age-adjusted D-dimer Cutoff Levels to Rule Out Deep Vein Thrombosis: a Prospective Outcome Study (ADJUST-DVT)

NCT02384135

University Hospital, Geneva

3,300 participants

Multiple sites in Canada, France, and Switzerland

This is a prospective cohort management study examining the utility of an age-adjusted D-dimer cutoff in combination with clinical pretest probability for the diagnosis of deep vein thrombosis (DVT). The age-adjusted cutoff is defined as “negative” as follows: if patient’s age is less than 50 years, D-dimer is less than 500 µg/L; if patient’s age greater than 50 years old, D-Dimer less than age multiplied by 10. This contrasts with prior DVT diagnostic studies in which a fixed D-dimer cutoff of 500 µg/L was applied across all age groups.

Outpatients presenting to the emergency department with suspected DVT will have their clinical probability of DVT assessed using the two-level Wells score for DVT. Patients with a non-high (unlikely) probability will have a high-sensitivity D-dimer test performed, and those with a negative age-adjusted D-dimer will be considered as not having DVT and will not receive anticoagulation therapy. Those with a high (likely) probability, or those with unlikely probability but positive age-adjusted D-dimer, will have compression ultrasonography to assess for proximal DVT. The primary outcome measure is the risk of venous thromboembolism (VTE; including DVT or symptomatic pulmonary embolism [PE]) at three months in patients with D-dimer levels between the usual fixed cutoff and the age-adjusted cutoff who are not anticoagulated on the basis of this strategy (i.e., those in whom DVT is ruled out).

Exclusion criteria will include those in whom suspicion for DVT is raised more than 48 hours after hospital admission, pregnancy, other indications for anticoagulation besides VTE, concomitant suspicion for PE, and life expectancy less than three months. The D-dimer assay used is a high-sensitivity quantitative assay based on ELISA (VIDAS D-dimer Exclusion) or immunoturbidimetric test (Innovance D-dimer).

D-dimer results have previously been extensively validated for the exclusion of DVT, particularly in combination with clinical prediction rules.¹  The cutoff value for categorizing D-dimer results as negative has generally been set at a fixed, low level of 500 µg/L to achieve higher sensitivity and negative predictive value. However, D-dimer results are not specific to VTE and are known to increase with age, which may limit their specificity in older patients.

In this study, the authors have implemented a management algorithm to validate the specificity and safety of using an age-adjusted D-dimer cutoff in combination with the dichotomized Wells DVT score. A similar management algorithm was validated in the ADJUST-PE study for the diagnosis of PE using age-adjusted D-dimer and two-level clinical probability (Wells or revised Geneva score) and was found to increase the proportion of patients with suspected PE who could safely be excluded on the basis of D-dimer without chest imaging (6.4-29.7%).²  Furthermore, a recent individual patient-data meta-analysis suggested that approaches using graduated D-dimer cutoffs adjusted to age or differing clinical pretest probability for DVT, instead of a fixed cutoff for all non–high-probability patients, demonstrated good specificity and negative predictive value.³ 

Given the established relationship (and often coexistence) between DVT and PE,⁴  along with the expected age-related increase in baseline D-dimer, this study is a natural next step from the aforementioned ADJUST-PE study.

Strengths of this study include its large sample size, pragmatic design using only a simple and widely used clinical prediction rule, and multicenter international design. Limitations include the use of only two high-sensitivity D-dimer assays (whereas other recent studies using graduated cutoffs, including ADJUST-PE, have used up to six different assays), which may impact its external generalizability. Additionally, its nonrandomized design precludes comparison with a control group using the traditional 500 µg/L cutoff.

However, this study will have important implications for clinical care and resource utilization for primary care and emergency medicine physicians, and its results will be eagerly anticipated.