Diagnostic accuracy of a simple qualitative ultrasound assessment for the diagnosis of empty stomach in the adult: A supplementary analysis of a prospective observer‐blind randomized crossover study

It has been reported that qualitative ultrasound assessment performed in the semi‐upright position had high sensitivity to detect gastric fluid volume >1.5 mL.kg−1. Nevertheless, the diagnostic accuracy of qualitative assessment for the diagnosis of empty stomach (fluid volume <0.8 mL.kg−1) has not been assessed. We aimed to assess the diagnostic accuracy of simple qualitative ultrasound assessment with and without head‐of‐bed elevation to 45° for the diagnosis of an empty stomach. We also aimed to determine the diagnostic accuracy of a composite ultrasound scale and clinical algorithm.


| INTRODUCTION
Point-of-care gastric ultrasound should provide easy, rapid, and reliable assessment of gastric contents status to assist the clinician in selecting the most appropriate anesthetic strategy based on the risk of aspiration.Solid and thick liquid contents are readily identifiable via ultrasound and consistently indicate high-risk gastric contents. 1,2wever, ultrasound interpretation becomes more challenging when dealing with fluid contents.first, differentiating between an empty antrum and one containing clear fluids is not always straightforward via ultrasound, 3,4 and second, the relationship between gastric fluid volume and the risk of regurgitation and aspiration remains a topic of debate. 5 is nevertheless considered that a gastric fluid volume >1.5 mL.6][7] We therefore recently assessed the ability of gastric ultrasound to detect gastric fluid volume >1.5 mL.kg À1 , 8 and found that the ultrasound visualization of clear fluids in volunteers lying down in the supine position with the head of the bed at 45 allowed the diagnosis of gastric fluid volume >1.5 mL.kg À1 with a sensitivity of 91% and a positive predictive value of 71%. 8The diagnostic accuracy of this simple qualitative assessment was not significantly different from that of a composite ultrasound scale nor from that of a clinical algorithm including calculation of gastric fluid volume. 8 addition to this result, it would be relevant to determine the diagnostic accuracy of the ultrasound visualization of an empty antrum in the semirecumbent position for the diagnosis of an "empty stomach".Gastric fluid volume <0.8 mL.kg À1 can be considered as a relevant threshold value to define an "empty stomach", because gastric fluid volume ≥0.8 mL.kg À1 is infrequent in elective fasted patients; 9 furthermore, it has been established that volume of aspirate ≥0.8 mL.kg À1 may have detrimental pulmonary consequences, 10,11 and that 0.8 mL.kg À1 of water injected into the esophagus could lead to aspirated volume up to 0.8 mL.kg À1 in fresh human cadavers in the sniffing position. 12Thus, gastric fluid volume ≥0.8 mL.kg À1 may favor the occurrence of aspiration when associated with other risk factors, 13,14 while volume <0.8 mL.kg À1 is associated with very low risk of aspiration and pulmonary injury.
The aim of this supplementary analysis of a previously published trial was to determine the diagnostic accuracy of a simple qualitative ultrasound examination in the semirecumbent position for the diagnosis of empty stomach defined as gastric fluid volume <0.8 mL.kg À1 . 8,15,16[18][19] 2 | METHODS

| Study design
We performed a supplementary analysis of a prospective randomized observer-blind crossover trial that had received institutional ethics committee approval (Comité de Protection des Personnes Sud-Ouest et Outre-Mer II, N 2019-A01333-54, Chairperson Professor Bettina Couderc, Toulouse, France) and had been registered in the public registry ClinicalTrials.govNo. NCT04097236. 8The study was conducted between November 27, 2019 and September 6, 2021.Written informed consent had been obtained from each volunteer.This secondary analysis was planned in the protocol submitted to the ethics committee and was registered in ClinicalTrials.gov.The present report follows the Standards for Reporting Diagnostic Accuracy (STARD) statement. 20

| Participants
Healthy adult volunteers were included in the study; inclusion criteria were: American Society of Anesthesiologists physical status 1, age ≥ 18 years, fasting duration for clear fluids ≥2 h, and that of solids ≥6 h.Volunteers with history of gastrointestinal surgery, digestive disease, gastroparesis, insulin-dependent diabetes mellitus, volunteers with obesity (body mass index ≥30 kg.m À2 ), those taking medications affecting gastric motility, and pregnant or breastfeeding women, were not included. 8

| Study protocol
The study protocol has been described in detail elsewhere. 8Briefly, volunteers attended two distinct study sessions in a randomized order, defined by the position of the upper section of the bed that was either angled at 45 or 0 (no elevation).Within each session, three tests were performed.Each test corresponded to a different volume of water, among the following: 0, 50, 100, 150, or 200 mL.
These three volumes of water were the same in both sessions; they were randomized using computer-generated lists as well as their order within each session so that the order differed from one session to another.The allocations were concealed using opaque, sealed, and coded envelopes until consent was obtained from the volunteers.investigator performed qualitative ultrasound examination of the gastric antrum and measurement of the antral cross-sectional area as described further, blinded to the ingested volume of water.For this ultrasound examination, the head of the bed was set at either 45 or 0 , according to the session.The volunteer was then encouraged to ambulate, but was not allowed to consume any food or beverages.
The second and third tests were then conducted at 90-minute intervals following the same procedures, but with different volumes of water for each test.For the second and third tests, similar to the first test, an ultrasound examination was first conducted to confirm the empty status of the gastric antrum in both the supine and right lateral decubitus positions with head-of-bed elevation.If the gastric antrum was not empty, the test was postponed for 30 min, and was performed only after ensuring that the antrum was empty in both positions.

| Gastric ultrasound scan
Ultrasound examination of the gastric antrum was conducted using a portable ultrasound scanner (FUJIFILM Sonosite, Inc., Bothell, WA) equipped with a low-frequency probe (2-5 MHz).The examination was performed in a standardized sagittal plane that passed through the left lobe of the liver and the abdominal aorta.The volunteer first lay in the supine position and then in the right lateral decubitus position, with the head of the bed elevated or not depending on the session.An empty antrum was identified by visualizing a flat antrum without any contents and juxtaposed walls, while clear fluid contents were identified by visualizing a dilated antrum with anechoic contents. 2The average value of three consecutive measurements of the maximal longitudinal (D1) and anteroposterior (D2) diameters of the antrum performed between antral contractions, from serosa to serosa was used to calculate the antral cross-sectional area, according to the following formula: antral area = π Â D1 Â D2/4. 21trasound gastric content status was determined from the collected data based on a simple qualitative assessment for interpreting gastric ultrasound, whereby "empty stomach" was defined as empty antrum seen in the supine position (positive test), while clear fluid contents visualized in the supine position corresponded to increasedto-high gastric fluid volume (negative test). 2,15,16 addition, ultrasound gastric content status was determined according to a composite ultrasound scale and a clinical algorithm (Figure 1).
According to the composite ultrasound scale, "empty stomach" was defined as empty antrum in both the supine and the right lateral decubitus positions, and empty antrum in the supine position but with clear fluids seen in the right lateral decubitus position and antral area ≤ 340 mm 2 in the supine position. 17,18Increased-to-high gastric fluid volume was defined as either clear fluid contents seen in the supine position or clear fluid contents seen only in the right lateral decubitus position with antral area > 340 mm 2 in the supine position. 17,18cording to the clinical algorithm, "empty stomach" was defined as either empty antrum in both the supine and the right lateral decubitus positions or as calculated gastric fluid volume <0.8 mL.kg À1 when clear fluid contents was visualized in the supine or in the right lateral decubitus position.Increased-to-high gastric fluid volume was defined as calculated gastric fluid volume ≥0.8 mL.kg À1 . 16,19The following mathematical model was used for gastric fluid volume calculation: volume (ml) = 27 + 0.146 Â antral area in the right lateral decubitus position (mm 2 ) -1.28 Â age (year). 16

| Statistical analysis
The sample size had been calculated to show that qualitative ultrasound assessment in the semirecumbent position allowed the detection of gastric fluid volume >1.5 mL.kg À1 with a sensitivity ≥90%. 22,23e inclusion of 20 volunteers, in whom three ultrasound assessments were made in each position, had been planned. 8It was predetermined that if a volunteer withdrew prematurely from the study or if ultrasound evaluation of the gastric contents was inconclusive, the data collected would be excluded from the analysis.In such cases, a new volunteer would be recruited to replace the participant who left prematurely or whose ultrasound evaluation was inconclusive.Statistical analyses were performed using MedCalc ® version 12.1.4.0 for Windows (MedCalc Software, Ostend, Belgium).Continuous variables were expressed as median [interquartile range, IQR] and categorical variables were expressed as number (%).The diagnostic accuracy of each assessment (three methods each in both head-of-bed position) for the diagnosis of gastric fluid volume <0.8 mL.kg À1 was expressed as sensitivity, specificity, and negative and positive predictive value, with 95% confidence intervals (95% CI) using the Wilson Score Interval with Yate's continuity correction.Sensitivity and specificity of the six assessments were compared using Cochran's Q test followed by post-hoc pairwise comparisons applying the Dunn-Bonferroni procedure, 24 as appropriate.For each test, p < .05 was statistically significant.

| RESULTS
Twenty volunteers were included, of whom 11 were women (Figure 2).Their median [IQR] age was 27 [24-29]  The main data from the qualitative assessment, the composite ultrasound scale and the clinical algorithm are summarized in Table 1.
F I G R E 2 Subject flow.Index test: qualitative ultrasound assessment.
The specificity significantly differed among the six assessments ( p < .001),while there was no significant difference regarding sensitivity.Post-hoc analysis found that the specificity of the qualitative assessment performed without elevation of the head of the bed was significantly lower than the specificity of the qualitative assessment performed in the semirecumbent position ( p < .05),than the specificity of the composite ultrasound scale performed with head-of-bed elevation ( p < .05),and then the specificity of the clinical algorithm performed with head-of-bed elevation ( p < .05)as well as without head-of-bed elevation ( p < .05;Table 2).

| DISCUSSION
We found that a very simple qualitative assessment performed in volunteers lying on their backs with the head of the bed increased to 45 had high diagnostic accuracy for the diagnosis of gastric fluid volume <0.8 mL.kg À1 .In this position, sensitivity and specificity were both approximately 90%, and specificity was significantly higher than that of qualitative assessment without head-of-bed elevation.Furthermore, neither composite ultrasound scale nor clinical algorithm had better sensitivity or specificity compared to the qualitative assessment performed in the semirecumbent position.
We thought it was relevant to determine the diagnostic accuracy of gastric ultrasound to detect low gastric fluid volume (<0.8 mL.kg À1 ) that corresponds to empty stomach, i.e. a volume of gastric fluid that T A B L E 1 Ultrasound data for qualitative assessment, composite ultrasound scale, and clinical algorithm according to the head-of-bed position.No significant difference was found whether the head of the bed was elevated or not within each approach, as well as among the three approaches within each head-of-bed elevation.Note: * † ‡ § p < .05compared to the specificity of the qualitative assessment performed with the head of the bed angled at 0 according to post-hoc analysis.
is unlikely to be associated with an increased risk of regurgitation and aspiration, since this provides valuable information for the clinician.
The results of this supplementary analysis suggest that empty antrum in the semirecumbent position corresponds to very low risk of aspiration, with high accuracy and low uncertainty, allowing the anesthesiologist to consider routine rather than rapid sequence induction technique, should general anesthesia be performed.By contrast, we have previously reported that simple qualitative assessment in the semirecumbent position allowed the detection of high-risk gastric contents (fluid volume >1.5 mL.kg À1 ) with sensitivity >90%, but specificity of only <60% and positive predictive value of 71%, reflecting high number of false positives. 8Thus, the visualization of fluid contents in this position corresponds either to gastric fluid volume >1.5 mL.kg À1 and high risk of aspiration in >70% of the cases, or, at least, to increased risk of aspiration if associated with other risk factors of regurgitation, such as gastro-esophageal reflux or obesity.
The visualization of fluid contents should therefore lead the anesthesiologist to take into consideration the risk of aspiration when choosing the anesthetic strategy.
These results help make the interpretation of gastric ultrasound easy for the clinician.In fact, qualitative assessment is performed only in the supine position with upper section of the bed elevated at 45 , while composite ultrasound scale and gastric fluid volume calculation both require positioning the patient in the right lateral decubitus position, [15][16][17][18][19] which would not be feasible in around a quarter of emergency patients. 186][17][18][19] Thus, the results of the present study may facilitate the use of gastric ultrasound technique in routine clinical practice. 25vertheless, the semi-upright position may be not achievable when the patient is lying on the operating table in the full supine position.In this position, fluid contents are distributed throughout the stomach and do not flow to the antrum and therefore only high gastric fluid volumes can be detected at the antrum.This leads to increased rate of false-positives for the diagnosis of empty stomach as we found in the present study, and consequently, to reduced diagnostic accuracy of the qualitative assessment in comparison with the examination performed in the semi-upright position. 8,23Turning the patient to the right lateral decubitus position when no elevation of the head of the bed is feasible may improve the diagnostic accuracy of gastric ultrasound, as this position favors gastric contents flowing towards the antrum. 15,23,26Thus, empty antrum in both the supine and the right lateral decubitus position unquestionably corresponds to a very low volume of gastric fluid and an "empty stomach", as previously reported. 15,16,27However, the question remains as to how to interpret cases where the antrum is empty in the supine position but fluid contents are seen in the right lateral decubitus position, since it may correspond to gastric fluid volume >1.5 mL.kg À1 in a quarter of the patients. 16In order to discriminate between high and low gastric fluid volume in this setting, and thereby improve the diagnostic performance of the qualitative assessment, a composite ultrasound scale had been proposed, though not formally assessed. 18In this supplementary analysis, we used a threshold value of antral area measured in the supine position that had shown high diagnostic accuracy for the diagnosis of gastric fluid volume >0.8 mL.kg À1 and/or the presence of solid particles. 9However, the composite ultrasound scale without head-of-bed elevation did not provide better sensitivity or specificity than the qualitative assessment performed in this position.Conversely, we found that the clinical algorithm combining qualitative assessment and calculation of gastric fluid volume had significantly better specificity without head-of-bed elevation than the qualitative assessment performed in this position, with no significant change in sensitivity.This result suggests that the clinical algorithm with gastric fluid volume calculation is of value to improve the diagnostic accuracy of gastric ultrasound whenever examinations are performed without head-of-bed elevation (when the patient is lying on the operating table, for example), provided, however, that the patient can be examined in the right lateral decubitus position.
This study does have some limitations that have previously been discussed. 8In particular, ultrasound examinations were conducted 3 min after the volunteers consumed water, as in the study by Kruisselbrink et al. 28 This time interval may have been sufficient to allow some of the ingested water to be emptied from the stomach, which could have affected the accuracy of the assessment.Nevertheless, gastric emptying of water starts up to 4 min after ingestion; 29 therefore, it can be hypothesized that only a small fraction of the ingested volume had passed the gastric antrum when ultrasound assessment was performed, so that the accuracy of the methodology was probably only marginally affected.Another specific limitation is that sample size had been calculated to demonstrate that qualitative ultrasound assessment in the semi-upright position allowed the detection of a volume >1.5 mL.kg À1 with a sensitivity ≥90%.For this, the study had been planned to reach a prevalence of gastric fluid volume >1.5 mL.
kg À1 of approximately 55%.The prevalence of gastric fluid volume <0.8 mL.kg À1 was 25%, and was similar to that has been previously reported in emergency patients. 9Considering this prevalence, 136 measurements would have been required to demonstrate that qualitative ultrasound assessment with head-of-bed elevation allowed the diagnosis of gastric fluid volume <0.8 mL.kg À1 with a sensitivity and a specificity ≥90%. 22Nevertheless, we were able to find high sensitivity and specificity, both around 90%, of the qualitative assessment for the diagnosis of empty stomach.
To conclude, the results of this supplementary analysis of a pro- After an initial ultrasound examination to confirm that the gastric antrum was empty in both the supine and right lateral decubitus positions with head-of-bed elevation to 45 , the volunteer was instructed to consume a specific volume of water.Three minutes later, another F I G U R E 1 Composite ultrasound scale and clinical algorithm assessed in the present study for the diagnosis of gastric fluid volume <0.8 mL.kg À1 .
spective observer-blind randomized crossover study corroborate that simple qualitative examination of gastric antrum performed in volunteers lying down in the supine position had high diagnostic accuracy for the diagnosis of empty stomach provided that the head of the bed was elevated to 45 .When no head-of-bed elevation is feasible, the clinical algorithm with mathematical model for gastric fluid volume calculation should be used, although this method requires turning the patient to the right lateral decubitus position and may be less easily feasible in clinical practice.analysis, Writing -Review & Editing.Sophie Barnoud: Project administration, Writing -Review & Editing.Julien Cordoval: Investigation, Writing -Review & Editing.Dominique Chassard: Supervision, Writing -Review & Editing.
Diagnostic accuracy of each assessment for the ultrasound diagnosis of gastric fluid volume <0.8 mL.kg À1 according to the position of the head of the bed for ultrasound.