DIAGNOSTIC ACCURACY AND FINANCIAL IMPLICATIONS OF AGE-ADJUSTED D-DIMER STRATEGIES FOR THE DIAGNOSIS OF DEEP VENOUS THROMBOSIS IN THE EMERGENCY DEPARTMENT

Background: Multiple D-dimer cutoffs have been suggested for older patients to improve diagnostic specificity for venous thromboembolism. These approaches are better established for pulmonary embolism. Objectives: We evaluated the diagnostic performance and compared the health system cost for previously suggested cutoffs and a new D-dimer cutoff for low-risk emergency department (ED) deep venous thrombosis (DVT) patients. Methods: We conducted a retrospective cohort study in two large EDs involving patients aged >50 years who had low pretest probability for DVT and had a D-dimer performed. The outcome was a diagnosis of DVT at 30 days. We evaluated the diagnostic accuracy and estimated the difference in cost for cutoffs of 500 ng/mL and the age-adjusted (age x 10) rule. A derived cutoff of 1000 ng/mL was also assessed. Results: Nine hundred and seventy-two patients were included (median age 66 years; 59.5% female); 63 (6.5%) patients were diagnosed with DVT. The conventional cutoff of < 500 ng/mL demonstrated a sensitivity of 100% (95% confidence interval [CI] 94.3-100%) and a specificity of 35.6% (95% CI 32.5-38.8%). The age-adjusted approach increased specificity while maintaining high sensitivity. A new cutoff of 1000 ng/mL demonstrated improved performance: sensitivity 100% (95% CI 94.3-00%) and specificity 66.3% (95% CI 63.2-69.4%). Compared to the conventional approach, both the 1000 ng/mL cutoff and the age-adjusted cutoffs could save healthcare dollars. A cutoff of 1000 ng/mL could have saved 310 ED length of stay hours and $166,909 (Canadian dollars) in our cohort, or an average savings of 0.32 h and $172 per patient. Conclusions: Among patients aged > 50 years with suspected DVT, the age-adjusted D-dimer and a cutoff of 1000 ng/mL improved specificity without compromising sensitivity, and lowered the health care system cost compared to that for the conventional approach. (C) 2019 Elsevier Inc. All rights reserved.