Trust your gut? Evaluating non-expert gastric ultrasound performance – A prospective observational cohort study

BACKGROUND: Aspiration of gastric content is a serious complication of anesthesia, associated with high mortality and morbidity. Recent studies demonstrated that fasting status can be assessed accurately by gastric ultrasound. However, there is still a lack of evidence regarding the application of this technique by inexperienced examiners. We aimed to determine the accuracy of gastric ultrasound performed by medical students after a standardized training sequence. OUTCOME: Accuracy of identifying the fasting status by gastric ultrasound performed by inexperienced examiners. METHODS: In this prospective observational cohort study, five medical students performed 80 gastric ultrasound examinations on healthy, non-obese volunteers. Standardized training consisted of blended online training, one plenary lecture, and 2 h of hands-on-training. Volunteers were randomized in a 2:1:1 ratio to "fasted", "non-fasted, fluid", and "non-fasted, solid" with the examiners being blinded to the fasting status. The examiners determined the fasting status using gastric ultrasound and recorded the gastric cross-sectional area. Beforehand, an expert examiner also conducted all examinations for comparison. Sensitivity, specificity, positive and negative predictive values were calculated from the acquired data. The cross-sectional area was pooled for the expert and non-experts and the results plotted and interrater reliability assessed by the intraclass correlation coefficient (ICC). RESULTS: Data from 80 individuals were analyzed. The inexperienced examiners correctly identified all non-fasted volunteers (sensitivity 1.00,95 % CI: 0.91-1.00). They wrongly classified 18 out of 40 fasted volunteers as "non-fasted" (specificity 0.55, 95 % CI: 0.40-0.69). Positive predictive value was 0.69 (95 % CI: 0.56-0.79) and negative predictive value 1.00 (95 % CI: 0.85-1.00). The overall ICC was 0.72 (95 % CI 0.57-0.82), and 0.30 (95 % CI -0.32-0.63), 0.66 (95 % CI -0.16-0.87), and 0.27 (95 % CI -0.82-0.71), for the "fasted", "non-fasted, fluid", and "non-fasted, solid" states, respectively. CONCLUSIONS: Examiners with limited experience in ultrasound diagnostics may accurately identify a full stomach in normal-weight volunteers after a standardized training sequence. However, the detected specificity of 0.63 was low, and more focused training on the ultrasound anatomy of an empty stomach may be needed to rule out an empty stomach in a clinical scenario.