**Background:** Bedside ultrasound (US) assessment of gastric content and volume has become a useful tool for rapid, non-invasive evaluation of aspiration risk, particularly in emergency and perioperative settings. However, its reproducibility in critically ill patients receiving enteral nutrition had not been established. This study aimed to estimate intra-observer and inter-observer agreement in US assessment of gastric content and volume in this population.
**Methods:** A prospective correlation study was conducted at Hospital Universitario Nacional de Colombia (Bogotá) between December 2020 and February 2021. Eligible patients were >18 years, admitted to the ICU on mechanical ventilation, and receiving enteral nutrition for at least 4 hours. Exclusion criteria included refusal to participate, pregnancy/postpartum, recent laparotomy, or discontinuation of enteral nutrition before US assessment. A total of 41 patients were included from 198 eligible. Each examiner performed 64 gastric US (total n=128). Examiners were EC1 (third-year anesthesiology resident) and EC2 (anesthesiologist/intensivist), each having performed at least 50 prior gastric US. US was performed with patients in semi-recumbent (30–45°) and right lateral decubitus positions using a 5 MHz convex-array transducer. Qualitative assessment classified the antrum as grade 0 (empty in both positions), grade 1 (fluid only in right lateral decubitus), or grade 2 (fluid in both positions). Quantitative assessment measured craniocaudal and anteroposterior diameters to calculate cross-sectional area (CSA) and gastric volume using the formula: volume (mL) = 27.0 + 14.6 × right lateral CSA (cm²) – 1.28 × age (years). For inter-observer assessment, measurements were blinded and performed within one hour. For intra-observer assessment, the procedure was repeated one hour after the first evaluation. Agreement was assessed using Cohen's Kappa, ICC, Lin's CCC, and Bland-Altman analysis.
**Key Results:** Mean age was 56.5 years (SD ±12.6); 63.4% were men. Hypertension (36.5%), type 2 diabetes (26.8%), and cancer (4.8%) were the most common comorbidities. Respiratory failure due to severe COVID-19 was the cause of ICU admission in 92.6% of patients. All patients were on invasive mechanical ventilation; 39.0% were receiving vasopressors. In supine position, both examiners classified gastric content identically: grade 0 in 1 US (1.5%), grade 1 in 4 (6.2%), and grade 2 in 59 (92.1%), with almost perfect agreement (Kappa = 1.0, SE = 0.11). In right lateral decubitus, classification changed in 3 US (4.6%) for both examiners. For intra-observer variability in antrum area measurement, EC1 showed CCC = 0.95 (95% CI 0.940–0.977), mean difference −0.47 cm² (SD ±1.64), 95% limits of agreement −3.70 to 2.75 cm²; EC2 showed CCC = 0.94 (95% CI 0.922–0.973), mean difference −0.18 cm² (SD ±2.18), limits −4.47 to 4.09 cm². For gastric volume intra-observer: EC1 CCC = 0.95 (95% CI 0.941–0.978), mean difference −7.9 mL (SD ±23.5), limits −54.1 to 38.1 mL; EC2 CCC = 0.94 (95% CI 0.922–0.974), mean difference −2.76 mL (SD ±31.8), limits −65.2 to 59.7 mL. Intra-observer Kappa for qualitative assessment was 0.74. For inter-observer variability: antrum area CCC = 0.84 (95% CI 0.778–0.911), mean difference −0.86 cm² (SD ±3.38), limits −7.50 to 5.78 cm²; gastric volume CCC = 0.84 (95% CI 0.782–0.913), mean difference −12.3 mL (SD ±49.2), limits −108.8 to 84.0 mL. ICC for cross-sectional area was 0.969 (EC1), 0.948 (EC2), and 0.872 (inter-observer). Mean gastric volume was 101.06 ± 81.94 mL (EC1) and 113.44 ± 100.22 mL (EC2). In 75% of grade 2 cases, gastric volume exceeded 1.5 mL/kg.
**Clinical Implications:** This study demonstrates that bedside US assessment of gastric content and volume has good intra- and inter-observer reliability in critically ill patients receiving enteral nutrition, supporting its potential use as a non-invasive tool for monitoring enteral nutrition tolerance and aspiration risk. The finding that most patients had gastric volumes >1.5 mL/kg highlights the high baseline aspiration risk in this population. The study also suggests that formal training may improve quantitative measurement precision, as the examiner without formal gastric US training showed a statistically significant mean difference in repeated volume measurements (−7.99 mL, p=0.009) compared to the trained examiner (−2.77 mL, p=0.497), though this difference was considered clinically acceptable. Limitations include the single-center design, predominance of COVID-19 patients (92.6%), and potential variability from regurgitation (5 patients) and peristalsis. Further studies are needed to determine whether US-measured gastric volume can predict enteral nutrition intolerance and to establish optimal training requirements.