Accuracy of physical examination in detecting central venous catheter-associated thrombosis in critically ill children.

INTRODUCTION

Lack of standardized definition impedes our ability to understand the clinical significance of asymptomatic central venous catheter (CVC) associated deep venous thrombosis (CADVT). Using standardized definitions, we aimed to determine the accuracy of physical examination in detecting CADVT in critically ill children and to identify characteristics associated with this accuracy.

MATERIALS AND METHODS

In a post hoc study, we analyzed 236 children <18 years old admitted to the pediatric intensive care unit, had an untunneled CVC and surveilled for CADVT using ultrasound with paired physical examination.

RESULTS

Of 236 paired examinations, 79 (33.5 %) had CADVT on ultrasound, while 56 (23.7 %) had signs of inflammation or venous obstruction on physical examination or CVC dysfunction. Sensitivity was 29.2 % (95 % confidence interval, CI: 19.9 %, 38.5 %), specificity was 80.2 % (95 % CI: 73.9 %, 86.4 %) and area under the receiver operating characteristic curve (AUROC) was 0.55 (95 % CI: 0.49, 0.60). When CVC dysfunction was excluded, sensitivity was lower (11.1 %; 95 % CI: 4.6 %, 17.6 %; p = 0.002), but specificity was higher (88.7 %; 95 % CI: 83.6 %, 93.7 %, p = 0.04). AUROC was 0.50 (95 % CI: 0.46, 0.54; p = 0.17). Use of point-of-care ultrasound and CVC inserted in the internal jugular vein (vs femoral vein) had lower sensitivity. Sepsis or infection and vasoactive support had lower specificity. Center of enrollment was associated with variable sensitivity.

CONCLUSIONS

Physical examination has poor accuracy in detecting CADVT in critically ill children. Despite poor accuracy, physical examination that includes assessment of CVC dysfunction, in combination with imaging, is key to understanding the clinical significance of asymptomatic CADVT.