Anesthesiology. 2026 Feb 23. doi: 10.1097/ALN.0000000000006005. Online ahead of print.
ABSTRACT
BACKGROUND: Preoperative gastric ultrasonography serves as an objective tool for evaluating the risk of pulmonary aspiration. However, measurements obtained from gastric ultrasonography at the abdominal aorta (AA) plane and the inferior vena cava (IVC) plane may vary substantially. This study aims to identify the measurement plane that most closely reflects the actual gastric volume (GV), thereby informing clinical practice.
METHODS: All participants initially underwent gastric ultrasonography at both the AA and IVC planes following a fasting period, representing the low GV state. Subsequently, they consumed apple juice at a volume of 2.3 ml/kg, after which gastric ultrasonography was repeated to represent the high GV state. The predicted ingested volume (PIV) was calculated as the predicted GV after ingestion minus the baseline predicted GV. Comparisons were made between AA plane and IVC plane measurements in both low GV and high GV. The agreement between the PIV and the actual ingested volume (AIV) was assessed for the AA plane, IVC plane, higher-measured gastric volume plane, and lower-measured gastric volume plane, along with their detection rates for high aspiration risk.
RESULTS: Ultimately, 196 volunteers were included in the final statistical analysis. The results showed that, in both low and high GV state, gastric ultrasonography measurements at the AA plane and the IVC plane differed significantly (P< 0.001). Among all measurement planes, only the higher-measured gastric volume plane showed no statistically significant difference between the predicted and actual GV, with the smallest bias (bias = -4.27 ml, P= 0.076).
CONCLUSIONS: Notable differences exist between gastric ultrasonography measurement planes. When applying existing predictive models, measurements obtained from the higher-measured gastric volume plane provide greater accuracy. Differentiating between measurement planes during model development may enhance the predictive accuracy of such models.
PMID:41730171 | DOI:10.1097/ALN.0000000000006005
