Background We sought to measure the diagnostic accuracy of D-dimer in children with suspected pulmonary embolism (PE).

Methods We queried our electronic medical record for quantitative D-dimer values obtained in all children ages 5–17 over 10 years in our 10-hospital system. Patients who had a D-dimer obtained in the evaluation of PE underwent supervised chart review to extract baseline demographics (age, sex, ethnicity), medical history, laboratory data and imaging results. PE was confirmed by imaging positive for deep vein thrombosis (DVT) or PE and excluded by imaging or no DVT or PE diagnosis within 90 days.

Results Over a 10-year period, we identified 13 792 orders for D-dimer testing in 2554 unique patients. Chart review indicated that 526 (20.6%) unique patients had D-dimer testing performed in the evaluation of PE (Cohen’s kappa=0.95, 95% CI 0.85 to 1.0). Most D-dimers (465/526, 88%) were ordered in children aged >12 years. Of these 526 children, 34 (6.4%, 95% CI) had a criterion standard positive for new or recurrent PE. The mean D-dimer value was 2104±1394 ng/mL in the 34/34 PE+ children and 586±962 ng/mL in PE− children with a sensitivity of 34/34 (100%, 89% to 100%) and a specificity of 290/492 (58%, 54% to 63%). The area under the receiver operating characteristic curve was 0.90 ((0.9)87–0.94).

Conclusions D-dimer is currently ordered in children for suspected PE in the emergency care setting, mostly in teenagers. The observed lower limit 95% CIs of 89% and 54% for diagnostic sensitivity and the specificity, respectively, suggest if used in patients with low-clinical probability, a normal D-dimer can safely exclude PE in children.