Sex-related differences in D-dimer levels for venous thromboembolism screening

Justin J. Reagh, Hui Zheng, Uwe Stolz, Blair A. Parry, Anna M. Chang, Stacey L. House, Nicholas J. Giordano, Jason Cohen, Adam J. Singer, Samuel Francis, Jürgen H. Prochaska, Eli Zeserson, Philipp S. Wild, Alexander T. Limkakeng, Elizabeth L. Walters, Frank LoVecchio, Daniel Theodoro, Judd E. Hollander, Christopher Kabrhel, Gregory J. Fermann

Research output: Contribution to journalArticlepeer-review

7 Scopus citations


Background: D-dimer is generally considered positive above 0.5 mg/L irrespective of sex. However, women have been shown to be more likely to have a positive D-dimer after controlling for other factors. Thus, differences may exist between males and females for using D-dimer as a marker of venous thromboembolic (VTE) disease. We hypothesized that the accuracy of D-dimer tests may be enhanced by using appropriate cutoff values that reflect sex-related differences in D-dimer levels. Methods: This research is a secondary analysis of a multicenter, international, prospective, observational study of adult (18+ years) patients suspected of VTE, with low-to-intermediate pretest probability based on Wells criteria ≤ 6 for pulmonary embolism (PE) and ≤ 2 for deep vein thrombosis (DVT). VTE diagnoses were based on computed tomography, ventilation perfusion scanning, or venous ultrasound. D-dimer levels were tested for statistical difference across groups stratified by sex and diagnosis. Multivariable regression was used to investigate sex as a predictor of diagnosis. Sex-specific optimal D-dimer thresholds for PE and DVT were calculated from receiver operating characteristic analyses. A Youden threshold (D-dimer level coinciding with the maximum of sensitivity plus specificity) and a cutoff corresponding to 95% sensitivity were calculated. Statistical difference for cutoffs was tested via 95% confidence intervals from 2,000 bootstrapped samples. Results: We included 3,586 subjects for analysis, of whom 61% were female. Race demographics were 63% White, 27% Black/African American, and 6% Hispanic. In the suspected PE cohort, 6% were diagnosed with PE, while in the suspected DVT cohort, 11% were diagnosed with DVT. D-dimer levels were significantly higher in males than females for the PE-positive group and the DVT-negative group, but males had significantly lower D-dimer levels than females in the PE-negative group. Regression models showed male sex as a significant positive predictor of DVT diagnosis, controlling for D-dimer levels. The Youden thresholds for PE patients were 0.97 (95% CI = 0.64 to 1.79) mg/L and 1.45 (95% CI = 1.36 to 1.95) mg/L for females and males, respectively; 95% sensitivity cutoffs for this group were 0.64 (95% CI = 0.20 to 0.89) and 0.55 (95% CI = 0.29 to 1.61). For DVT, the Youden thresholds were 0.98 (95% CI = 0.84 to 1.56) mg/L for females and 1.25 (95% CI = 0.65 to 3.33) mg/L for males with 95% sensitivity cutoffs of 0.33 (95% CI = 0.2 to 0.61) and 0.32 (95% CI = 0.18 to 0.7), respectively. Conclusion: Differences in D-dimer levels between males and females are diagnosis specific; however, there was no significant difference in optimal cutoff values for excluding PE and DVT between the sexes.

Original languageEnglish
Pages (from-to)873-881
Number of pages9
JournalAcademic Emergency Medicine
Issue number8
StatePublished - Aug 2021


  • D-dimer
  • age-adjusted
  • deep vein thrombosis
  • diagnostic testing
  • logistic regression
  • pulmonary embolism
  • receiver operating characteristic
  • sex-adjusted
  • venous thromboembolism


Dive into the research topics of 'Sex-related differences in D-dimer levels for venous thromboembolism screening'. Together they form a unique fingerprint.

Cite this