D-dimer is nearly always increased in patients with acute venous thromboembolism (VTE) and, therefore, a low value (i.e., negative test) has high negative predictive value. A positive result indicates the need for further diagnostic testing (e.g. ultrasound imaging if deep vein thrombosis is suspected), but it is generally believed that the predictive value of a positive result is of no clinical value. While this is true if D-dimer values are dichotomized as positive or negative, D-dimer may have diagnostically useful positive predictive value when levels are markedly elevated (i.e., positive results are not considered as a single category).
To determine the prevalence of, and likelihood ratios for, deep vein thrombosis (DVT) associated with progressively elevated D-dimer levels in patients with a suspected first DVT.
We performed a retrospective analysis of D-dimer level from 1224 banked peripheral blood samples from participants enrolled at 4 of the 5 participating clinical centres in the recently competed SELECT study, which compared two diagnostic strategies in patients with a suspected first DVT. D-dimer levels were measured using the Triage® D-dimer assay. The clinical pretest probability (CPTP) was determined in all patients using the 9 point Wells score and categorized as low, moderate or high. Patients were categorized as DVT positive if DVT was diagnosed by lower limb ultrasound (US) at initial presentation or VTE was diagnosed during three months of follow-up. Patients were categorized as DVT negative if DVT was not diagnosed at initial assessment (patients with low CPTP and negative D-dimer results did not undergo US) and there was no VTE during three months of follow-up. The diagnosis or exclusion of DVT was made independently of the Triage® D-dimer measurements (a different D-dimer assay was used at presentation in the SELECT study, and D-dimer was not used to evaluate suspected VTE during the 3 month follow-up).
The prevalence of DVT and likelihood ratios (LHR) for all patients, and according to CPTP (low, moderate or high), was determined for the following intervals of D-dimer level: <500ug/L, 500–1000ug/L, 1001–1500ug/L, 1501–2000ug/L, 2001–3000 ug/L, 3001–4000ug/L, 4001ug/L-5000ug/L and >5000ug/L.
The prevalence of DVT, and associated LHR, among all patients increased steadily with rising D-dimer levels (Table 1). At values of less than 1000ug/L, the prevalence of DVT among all patients was low (0.15% for D-dimer <500ug/L and 1.2% for 500–1000ug/L). However, at levels above 1500ug/L there was a progressive increase in the prevalence from 14.3% for 1500–2000ug/L to 64.0% at levels >5000ug/L. The associated likelihood ratios for DVT also increased according to D-dimer level from 2.6 for levels of 1500–2000ug/L to 28.0 for values >5000ug/L.
The prevalence of DVT was 2.6% for patients with low CPTP, 6.0% among those with moderate CPTP and 22.0% among patients with high CPTP. The increase in prevalence and associated LHR for DVT with increasing D-dimer levels was consistent across all three CPTP groups.
Increasing D-dimer levels are associated with a progressive increase in prevalence and LHR for DVT across all clinical pretest probability strata. The positive predictive value of markedly elevated D-dimer levels was high and may help to identify the subgroup of patients with an abnormal D-dimer level who require more intensive investigation after a normal imaging test.
Bates: Trinity Biotech: Consultancy, consultation fees and in-kind support (D-dimer kits) Other. Kearon:Alere: Consultancy.