Diagnostic accuracy of point-of-care ultrasound for catheter-related thrombosis in children

Background Compared with consultative US performed by the radiology department, point-of-care US performed by non-radiology physicians can accurately diagnose deep venous thrombosis in adults. Objective In preparation for a multicenter randomized controlled trial, we determined the accuracy of point-of-care US in diagnosing central venous catheter-related thrombosis in critically ill children. Materials and methods Children <18 years old with a central venous catheter who were admitted to the intensive care unit were enrolled. Consultative and point-of-care compression ultrasounds with Doppler were done on the vein where the catheter was inserted within 24 h after insertion. Repeat US was obtained within 24 h of removal of the catheter. All images were centrally, blindly and independently adjudicated for thrombosis by a team of pediatric radiologists. Chance-corrected agreement between readings was calculated. Results From 84 children, 152 pairs of consultative and point-of-care ultrasounds were analyzed. A total of 38 (25.0%) consultative and 17 (11.2%) point-of-care ultrasounds were positive for thrombosis. The chance-corrected agreement between consultative and point-of-care ultrasounds was 0.17 (standard error: 0.07; P = 0.008). With consultative US as a reference, the sensitivity of point-of-care US was 28.1% (95% confidence interval: 13.7%-46.7%) with a specificity of 91.8% (95% confidence interval: 84.4%-96.4%). A catheter in the subclavian vein was associated with discordant readings (adjusted odds ratio: 4.00; 95% confidence interval: 1.45-13.94). Conclusion Point-of-care US, when performed by non-radiology physicians and centrally adjudicated by pediatric radiologists in the setting of a multicenter randomized controlled trial, may not accurately diagnose catheter-related thrombosis in critically ill children.