Adaptive D-dimer cut-offs: very low missed-embolism rates in 12,194 patients

Design

Systematic review and random-effects meta-analysis of prospective diagnostic management studies

Population

12,194 patients with suspected pulmonary embolism in six studies (prevalence 7.0% to 19.2%)

Primary outcome

Diagnostic failure (VTE or PE-related or unexplained death) within 3 months after rule-out without CT

Effect

0.12% overall (95% CI 0.06 to 0.26); 0.50% in the adaptive window (95% CI 0.20 to 1.11)

Fixed D-dimer cut-offs of 500 ng/mL are safe but exclude few patients. Adaptive strategies, which raise the cut-off with age or according to clinical probability, would spare more CT scans, but earlier post hoc validations raised safety doubts. This meta-analysis pooled six prospective management studies with 12,194 patients and suspected pulmonary embolism, where an adaptive threshold determined who got imaging. Pulmonary embolism prevalence ranged from 7.0% to 19.2%.

Diagnostic failure, defined as venous thromboembolism or death from pulmonary embolism or an unexplained cause within 3 months, occurred in 0.12% (95% CI 0.06 to 0.26; I-squared 0%) of patients in whom embolism was ruled out without CT. For patients whose D-dimer was above 500 but below the adaptive threshold, the pooled failure rate was 0.50% (95% CI 0.20 to 1.11). The algorithm saved imaging in 60.8% on average (95% CI 49.2 to 71.2, with very high heterogeneity).

These strategies were applied in studies that also used clinical probability scores and, in some, other criteria. The method needs a validated clinical probability assessment first, and the findings do not apply to pregnancy or to children, which were excluded.

• Assess clinical probability with a validated score before using any D-dimer rule-out.
• Consider age-adjusted or probability-adjusted cut-offs where your local protocol allows them.
• Do not apply D-dimer rule-out to high-probability patients; go to imaging.
• Do not use these thresholds in pregnancy or children on this evidence.
• Make sure your laboratory's D-dimer assay is the one validated for these cut-offs.

Why it matters

It answers the safety question that was holding back wider use of adaptive thresholds.

Don't overread it

The studies excluded pregnant women and children, relied on a prior clinical probability assessment and used particular D-dimer assays; it does not support unconditional rule-out.

The statistics, in plain English

A failure rate of 0.12% means roughly 1 in 800 patients ruled out later turned out to have thromboembolism or died unexplained within 3 months. The adaptive window carried 0.50%, about 1 in 200, with an interval up to 1.11%. A very high I-squared for efficiency shows the proportion of scans spared varied widely between studies.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.