Abstract
Assessing intravascular volume status in the critically ill patient remains a challenge for intensivists, and the accuracy of such estimation based on bedside examination alone is reported to be nearly a coin toss. In this retrospective study we sought to validate a previously recommended chest radiographic vascular pedicle width (VPW) ≥70 mm for identifying cardiogenic pulmonary oedema (CPO). We additionally assessed whether novice physicians-in-training can reliably measure the VPW. The study included intensive care patients with an existing pulmonary artery catheter. Three independent raters performed measurements of VPW from chest radiographs obtained within three hours of pulmonary artery occlusion pressure measurements. In 80 patients enrolled, a VPW cut-off of ≥70 mm had a 55% sensitivity, 88% specificity, 81% positive predictive value, 69% negative predictive value and 73% accuracy for identifying patients with CPO. Receiver operating characteristic curve analysis showed an area under the curve of 0.72 (95% confidence interval 0.61 to 0.84) for VPW in discriminating CPO from non-cardiogenic pulmonary oedema. Kappa statistics for inter-rater reliability showed Kappa=0.41, 0.42 and 0.85 for each pair of the three raters. In conclusion, the previously accepted VPW cut-off of ≥70 mm is reasonably accurate in discriminating CPO from non-cardiogenic pulmonary oedema. VPW can be measured by physicians-in-training with a comparable performance to previous studies utilising expert radiologists.