Background: The incidence of thrombosis associated with an indwelling umbilical venous catheter (UVC) in unselected newborn infants is uncertain. Although real-time and doppler flow echocardiography is commonly used to detect such thrombi, the diagnostic accuracy of this technique has never been validated against a gold standard test.

Objective: To compare real-time and doppler flow echocardiography with contrast venography in a prospective cohort of infants at the time of elective removal of the UVC.

Methods: 50 newborn infants were studied. None had clinically suspected UVC-associated thrombosis. All catheter tips had been correctly placed above the diaphragm. After parental consent, each patient underwent echocardiography a few hours prior to elective removal of the UVC. All study tapes were interpreted independently by two pediatric cardiologists without knowledge of the radiologic findings. Contrast venography via the UVC was performed at the time of catheter removal.

Results: Three radiologic studies were excluded for poor quality. Mean birthweight (SD) in the remaining 47 infants was 1962 (1090) g. Mean gestational age (SD) was 32 (5.3) weeks. Catheters remained in situ for a mean(SD) of 7.0 (2.3) days. Thrombi were detected by venography in 14 of 47 cases(30%) in the following locations: Right atrium n=3; inferior vena cava n=8; ductus venosus n=3. Cardiologist `A' correctly diagnosed 3 cases, with 3 false-positive diagnoses of thrombosis (21% sensitivity, 91% specificity). Cardiologist `B' correctly diagnosed 6 cases, with 8 false-positive diagnoses of thrombosis (43% sensitivity, 76% specificity).

Conclusion: Central venous thrombosis is common in sick newborn infants who receive umbilical venous catheters. Contrast venography appears to be required to accurately diagnose such thrombi in affected patients.