Lisette van Dam

Detection of UEDVT by MR-NCTI 5 89 Secondary outcomes The two independent reviewers disagreed on 5 of the 60 MR-NCTI readings. Hence, the interobserver agreement between two independent readers had a kappa value of 0.83 (95%CI 0.69-0.97). Consensus reading of the MR-NCTI scans with discrepancy between the two readers resulted in a correct positive diagnosis in four patients and a falsely negative UEDVT diagnosis in one patient. Table 5 . Characteristics of the two patients with false-negative MR-NCTI result Sex Age (years) Symptom duration (days) Constans’ score (points) D-dimer (ng/mL) MR- NCTI Contrast venography Patient 1 Female 66 6 3 1975 Negative for UEDVT Hypoplastic left internal jugular vein and intraluminal hypodensity at the confluence of the left subclavian and internal jugular vein and in the proximal hypoplastic jugular vein Patient 2 Male 52 3 2 1600 Negative for UEDVT Thrombus in right subclavian vein MR-NCTI, magnetic resonance non-contrast thrombus imaging; UEDVT, upper extremity deep vein thrombosis. DISCUSSION In this study, we showed that MR-NCTI is accurate for the diagnosis of UEDVT, with a sensitivity of 93% (95%CI 78-99%) and specificity of 100% (95%CI 88-100%). Moreover, the interobserver agreement was excellent with a kappa value of 0.83 (95%CI 0.69-0.97).