Abstract
Central venous pressure (CVP) monitoring is valuable for guiding fluid management in critically ill patients admitted to the ICU. However, direct measurement via an intravenous catheter is invasive, time-consuming, and labor-intensive. Noninvasive ultrasound vessel measurements, such as internal jugular vein (IJV) and inferior vena cava (IVC) collapsibility index, offer alternatives but are affected by respiratory and anatomical factors. Static vessel diameters may provide a simpler, more reliable method, yet few studies have assessed their combined predictive value for estimating CVP. Critically ill, spontaneously breathing ICU patients received central venous pressure monitoring and ultrasound assessment of the transverse and anteroposterior diameter(TD and APD)of the IJV and CCA, along with IVC diameter (IVCD). The dataset was randomly divided into a training set and a validation set. Correlations between each vessel diameter and CVP were analyzed using linear regression. A multivariable linear regression model was then developed to predict CVP. Receiver operating characteristic (ROC) analysis was performed to evaluate the ability of the model to identify patients with low CVP (< 8 mmHg). A total of 181 patients were enrolled. IJV-APD (r = 0.57), CCA-APD (r = 0.49), IVCD (r = 0.58) showed significant linear positive correlations with CVP. Using these correlative variables from univariate linear analysis, both IJV-APD (β = 1.789, p < 0.05) and ICVD (β = 1.334, p < 0.05) effectively predict the actual CVP values by multivariable linear regression. A predictive model combining IJV-APD and IVCD measurements accurately identified patients with CVP < 8 mmHg, with high AUC both in the training set (AUC = 0.807) and validating set (AUC = 0.749). We found that IJV-APD, CCA-APD and ICVD were associated with the level of CVP in critically ill patients. The predictive model incorporating IJV-APD and IVCD can effectively identify patients with low CVP (< 8 mmHg) in clinical practice.