Interpreting Near-Threshold Age-Adjusted D-Dimer for Suspected Pulmonary Embolism in the Emergency Department: A...

Age-adjusted D-dimer (AAD) thresholds are used to reduce false positives in older patients evaluated for pulmonary embolism (PE). However, evidence is limited regarding test behavior near the individualized cutoff ("near-threshold" zone) and the safety of excluding PE in AAD-negative cases.

We conducted a retrospective diagnostic accuracy study of adults ≥ 50 years presenting to a tertiary emergency department (2015-2019) with suspected PE who underwent D-dimer testing and definitive imaging (CT pulmonary angiography or V/Q scintigraphy). Diagnostic performance of the standard threshold (0.5 µg/mL, FEU) and AAD (age ×  0.01 µg/mL) was assessed using ROC analysis. Accuracy metrics are reported with 95% confidence intervals (CIs) via bootstrap; AUC CIs and between-curve comparisons used the DeLong method. The near-threshold zone was defined as ±0.1 µg/mL around the AAD cutoff. Clinical charts of AAD-negative PE cases were reviewed.

Among 979 patients, 162 (16.5%) had imaging-confirmed PE. Standard D-dimer showed sensitivity 99.4% (95% CI 98.0-100.0) and specificity 4.0% (2.7-5.4); AAD showed sensitivity 97.5% (94.8-99.4) and specificity 11.8% (9.6-13.9). AUCs were 0.734 and 0.727, with no significant difference (DeLong p = 0.41). Within the near-threshold window, PE prevalence was 7.7%, increasing with age (28.6% in ≥ 80 years). Three patients had PE despite AAD-negative results; all had segmental or subsegmental emboli, and no in-hospital complications were documented in the available records.

In this imaging-selected retrospective cohort, AAD improved specificity with minimal sensitivity loss. However, AAD should not be used as a stand-alone rule-out strategy and should be applied only within clinical probability-guided diagnostic pathways. Near-threshold results are not risk-free, particularly in older adults, and should prompt careful clinical reassessment when overall probability remains concerning.