Nutritional support is fundamental in the management of critically ill patients, being enteral nutrition the nutritional support technique of choice in these patients [16, 17]. However, its use and enteral nutrition intolerance may be associated with several adverse events, such as nausea, vomiting, abdominal pain and pulmonary aspiration [18]. So far, there is no standard method to assess tolerance to this therapy in ICU patients and, in general, decisions on its use in this clinical setting are based on expert recommendations and the assessment of the clinical condition of the patient.

Bedside US offers several advantages in the management of critically ill patients. In this regard, the US assessment of gastric content strategy developed by Perlas et al. [1] and Van de Putte et al. [5], and that has also been described in multiple studies [19, 20], has become an adequate method to assess the risk of pulmonary aspiration prior to intubation in critically ill patients. In addition, although the risk of pulmonary aspiration and the US measurement of gastric content have been poorly studied in this type of patients, initial data suggest that this method has an adequate performance in the evaluation of gastric content in critically ill patients.

As mentioned above, there are only a few studies reporting data on inter-observer agreement in US assessment of gastric content. For example, Mackenzie et al. [19], in a randomized controlled clinical trial conducted in 45 healthy adult patients and in which three expert sonographers were asked to interpret 45 US and classify gastric content into three categories (presence, absence, not visible), reported that inter-observer agreement was good for the interpretation when the patient was in a right lateral decubitus position (kappa coefficient = 0.91) and moderate for the general interpretation and subxiphoid windows (kappa coefficients = 0.64 and 0.72, respectively)[19]. Similarly, Johnson et al. [21], in a randomized controlled trial in which three ultrasound experts (the one who performed the ultrasound and examiners A and B) evaluated 60 gastric US performed in healthy adult patients, found an almost perfect inter-rater agreement in the fluid content group (ICC = 0.950) and a good inter-rater reliability for the solid content and empty stomach (fasted) groups (ICC = 0.781 and 0.761, respectively) [21].

Although these studies were not specifically conducted in critically ill patients receiving enteral nutrition, their results in terms of inter-observer agreement are similar to the results of our study, where inter-rater agreement was almost perfect (kappa coefficient = 1). Furthermore, unlike the studies by Mackenzie et al. [19] and Johnson et al. [21], intra-observer agreement in the assessment of gastric content by means of US was also evaluated in the present study, obtaining a good level of agreement in both examiners (kappa coefficient = 0.74).

On the other hand, Kruisselbrink et al.[22], in a randomized controlled clinical trial carried out in Ontario in 22 healthy patients and in which three sonographers with previous experience in gastric US performed a standard US assessment of gastric volume, found an “almost perfect” intra-observer and inter-observer reliability in the ultrasound assessment of the cross-sectional area of the antrum (ICC = 0. 96 to 0.99 and 0.96, respectively); a finding similar to that reported in the present study, where intra-observer agreement (ICC = 0.969 and 0.948 for EC1 and EC2, respectively) and inter-observer agreement (ICC = 0.872) were also “almost perfect” in the US measurement of the cross-sectional area of the antrum in critically ill patients receiving enteral nutrition.

In our study, gastric content in the antrum was classified as grade 2 in 92.19% of the gastric US (n = 59), and gastric fluid volume was > 1.5 mL/kg in 75% of these cases. In the same vein, mean gastric volume was 101.06 ± 81.94 mL (95% CI 80.25–121.81) and 113.44 ± 100.221 mL (95% CI 87.98–138.89) in EC1 and EC2, respectively. These findings suggest that critically ill patients receiving enteral nutrition through a nasogastric tube are at high risk of pulmonary aspiration [1, 2, 5].

Additionally, the present study compared the intra-observer agreement in US assessment of gastric content of an examiner trained and certified in gastric US with the intra-observer agreement of an examiner who, despite having performed at least 50 qualitative and quantitative US assessments of gastric volume and content, had not received any formal training in gastric US. In the case of the qualitative assessment of the gastric content by means of US, intra-observer agreement was similar in both examiners; however, in the case of quantitative assessment, the mean difference in gastric volume was statistically significant (− 7.99 mL ± 23.56 mL; p-value = 0.009) in the examiner without training in gastric US, while in the examiner with training in gastric US this difference was not significant (− 2.77 mL ± 31.89 mL; p-value = 0.497). This difference in the means of gastric volume can be considered acceptable, taking into account that basal gastric secretions generate fasting gastric residual volumes of up to 1.5 mL/kg without this representing a significant risk of pulmonary aspiration (approximately 100–110 mL in the average adult population) [1, 2, 20].

Although the training specifications necessary for the performance of a reproducible gastric US assessment have not yet been defined, Arzola et al. cited by Kruisselbrink et al. [22], suggest that the sonographer should perform at least 33 gastric US under the supervision of an expert to achieve an accuracy rate of 95% in the qualitative US evaluation of gastric content. On the other hand, according to these authors, since quantitative US evaluation of gastric volume requires additional steps, it is logical to assume that the examiner must have performed a much higher number of gastric US to achieve a similar accuracy rate [22]. Considering the above, further studies are required to establish the optimal learning curve necessary for trainees to perform an adequate US assessment of gastric content.

This is one of the first studies that evaluates intra-observer and the inter-observer agreement in US assessment of gastric volume and content (being the latter classified in grades) in critically ill patients receiving enteral nutrition, which is undoubtedly its main strength, since the results reported here may contribute to the validation of this gastric content classification method in this type of patients and, therefore, its implementation in clinical practice.

On the other hand, this study has some limitations. First, it is a nonrandomized observational single-center study; however, its sample size is appropriate for the assessment of intra-observer and inter-observer agreement. Second, the COVID-19 pandemic had a great impact on the study population (critically ill patients on mechanical ventilation and receiving enteral nutrition), since 95.31% of the patients who were considered eligible for inclusion were admitted to the ICU due to respiratory failure resulting from severe COVID-19. Third, during the position changes required for achieving a proper US assessment of the gastric content and volume, five patients regurgitated the enteral nutrition formula, which, besides peristaltic contractions and gastric emptying, adds an element of variability in successive gastric ultrasound measurements.