Secondary Logo

Journal Logo

Validation of a Mathematical Model for Ultrasound Assessment of Gastric Volume by Gastroscopic Examination

Perlas, Anahi MD, FRCPC*; Mitsakakis, Nicholas M.Sc, PhD; Liu, Louis MD, FRCPC; Cino, Maria MD, FRCPC; Haldipur, Nidhi BSc; Davis, Liisa BSc, RDMS; Cubillos, Javier MD; Chan, Vincent MD, FRCPC, FRCA*

doi: 10.1213/ANE.0b013e318274fc19
Technology, Computing, and Simulation
Free

INTRODUCTION: Pulmonary aspiration of gastric contents is a serious perioperative complication. Previous models of ultrasound gastric volume assessment are preliminary and have not been validated by an external “gold standard.” In the present study we propose a more accurate model based on prospective data obtained from 108 patients undergoing bedside gastric sonography and upper gastrointestinal endoscopy (UGE).

METHODS: Patients undergoing elective UGE were randomized to ingest one of 6 predetermined volumes of apple juice after an 8-hour fasting period. A cross-sectional area of the antrum in the right lateral decubitus position (Right lat CSA) was measured by a blinded sonographer following a standardized scanning protocol. Gastric fluid was subsequently suctioned under gastroscopic vision during UGE performed by a blinded gastroenterologist and measured to the nearest milliliter.

RESULTS: Data from 108 patients suggest that a previously reported model tends to overestimate gastric volume particularly at low volume states. A new best fit mathematical model to predict gastric fluid volume based on measurements of Right lat CSA is presented. This new model built on a more accurate gold standard can be used to estimate gastric volumes from 0 to 500 mL, in nonpregnant adults with body mass index < 40 kg/m2.

CONCLUSIONS: We report a new prediction model to assess gastric fluid volume using standard 2-dimentional bedside ultrasound that has several advantages over previously reported models.

From the *Department of Anesthesia, University Health Network, University of Toronto, Toronto, Ontario, Canada; Department of Anesthesia and Pain Management, University Health Network, Toronto, Ontario, Canada; and Department of Medicine, Division of Gastroeneterology, University of Toronto, Toronto, Ontario, Canada. NM is currently at Toronto Health Economics and Technology Assessment Collaborative, University of Toronto, Toronto, Ontario, Canada.

Accepted for publication September 05, 2012

This report was previously presented, in part, at the 2011 Canadian Anesthesthesiologists’ Society Meeting

Funding: This study was funded by the 2010 Canadian Anesthesiologists’ Society/ GE Healthcare Award in Peri-operative Imaging and the 2011–2013 University of Toronto, Department of Anesthesia Merit Award

See Disclosures at end of article for Author Conflicts of Interest.

Reprints will not be available from the authors.

Address correspondence to Anahi Perlas, MD, FRCPC, University Health Network, University of Toronto, Toronto Western Hospital, Department of Anesthesia, 399 Bathurst St. Toronto, Ontario, M5T 2S8, Canada. Address e-mail to anahi.perlas@uhn.on.ca

Pulmonary aspiration of gastric contents is a serious perioperative complication leading to significant morbidity and mortality.1–3 Sedation and general anesthesia depress both the tone of the lower esophageal sphincter and upper airway protective reflexes, increasing the risk of pulmonary aspiration in subjects with significant gastric content.4

Bedside ultrasound can be used clinically to differentiate an empty stomach from one with fluid or solid gastric content.5–9 An empty stomach carries a negligible aspiration risk, and the sonographic diagnosis of an empty stomach is a qualitative one; no volume assessment is required.10 Solid stomach content is associated with a high aspiration risk and poor patient outcome. The diagnosis of solid content is based on sonographic appearance alone (a qualitative examination).10 When the stomach contains clear fluid, however, an estimation of the volume is of great clinical interest. One needs to clarify whether the clear fluid seen is a small volume of gastric secretions, clinically inconsequential, or a significant volume of clear fluid that could pose an aspiration risk. Although there is no strict “volume threshold” over which aspiration risk increases, gastric fluid volumes of up to 1.5 mL/kg (about 100 mL for the average adult) are common in fasted individuals and are believed to be safe.11–15 Therefore, a method to estimate the volume of clear fluid in the stomach could help differentiate small negligible amounts from larger volumes that could place patients at risk of regurgitation and pulmonary aspiration. Two preliminary mathematical models have been reported to estimate gastric fluid volume based on sonographic assessment of the gastric antrum.5,6 However, important limitations in the design of both studies limit their widespread clinical applicability. The initial model reported by our group, although mathematically robust, was based on data obtained from fasted healthy volunteers randomized to ingest known volumes of water, and may have failed to account for baseline gastric secretions present at baseline before ingestion.5 Similarly, a subsequent model proposed by Bouvet et al. was built on data obtained by “blind” suctioning through a nasogastric tube under general anesthesia, and it is conceivable that the suctioned volumes may not have been the full gastric content.6 Therefore, both models were built on relatively imprecise volume controls. The main objective of the present study was to prospectively examine the performance of our previously published model by correlating the volumes predicted by our model (based on gastric antral cross-sectional area) with the “observed volumes” suctioned under gastroscopic vision, arguably the most precise method of gastric volume assessment.

A secondary objective of this study was to explore the correlation of an existing qualitative grading system (grades 0, 1, 2) with gastric fluid volume.7 This simple 3-point grading system classifies the gastric antrum as follows: A Grade 0 antrum appears completely empty in both supine and right lateral decubitus positions. A grade 1 antrum appears empty in the supine position but fluid is visible in the right lateral decubitus, suggesting a small volume of gastric content. Grade 2 antrum contains fluid that is visible in both patient positions, suggesting a higher volume state.7

Back to Top | Article Outline

METHODS

After receiving University Health Network Research Ethics Board approval, patients undergoing elective upper gastrointestinal endoscopy (UGE) in the Department of Gastroenterology at the Toronto Western Hospital were invited to participate in this study. Inclusion criteria were: age of 18–85 years old; weight of 45–110 kg, height more than 145 cm, ability to understand the study procedures and provide informed consent. Exclusion criteria were: recent upper gastrointestinal bleed (within 1 month), previous lower esophageal or gastric surgery, and known abnormal upper gastrointestinal anatomy including hiatus hernia and gastric tumors. After written documentation of informed consent, and after a minimum fasting period of 8 hours for both fluids and solids, patients were prepared as per standard institutional practice.

A baseline qualitative gastric ultrasound assessment was performed in both supine and right lateral positions to ensure an empty stomach at baseline.5,7

Patients were subsequently randomized to ingest one of 6 predetermined volumes of apple juice (0 mL, 50 mL, 100 mL, 200 mL, 300 mL or 400 mL), according to a computer-generated list of random numbers. Three minutes after the ingestion, patients underwent a second sonographic assessment following a previously described scanning protocol.5 The gastric antrum was identified in a sagittal to right parasagittal plane between the left lobe of the liver and the pancreas, at the level of the aorta, or inferior vena cava (Figures 1A and 1B). The probe was tilted as needed to ensure a proper transverse view of the antrum, avoiding oblique images that may have resulted in an overestimation of the cross-sectional area of the antrum (Right lat CSA) and ultimately gastric volume. Three consecutive still images were obtained, labeled and stored. All images were obtained with the antrum at rest and not during peristaltic contractions. One single certified sonographer experienced in abdominal ultrasound and blinded to the volume ingested, performed all sonographic examinations.

A curvilinear array, low frequency (2 to 5 MHz) transducer and a Philips HD11XE system (Philips Healthcare, Markham, Ontario, Canada) or General Electric Logiq E unit (GE Healthcare Beijing, China) with image compounding technologies were used. The sonographer classified each patient’s antrum as grade 0, 1 or 2.7 After the examination, Right lat CSA was calculated based on the anteroposterior and craniocaudal antral diameters as previously described.16 The two antral diameters were measured from serosa to serosa as is current standard practice. Therefore, an antrum that is totally empty is expected to have a Right lat CSA that is more than 0, corresponding to the thickness of the gastric wall (usually in the order of 2–5 cm2).5–7 Three measurements from three consecutive images were obtained, and the mean of the three measurements used as the Right lat CSA value. A “predicted volume” was calculated based on the Right lat CSA and our previously described mathematical model.5

Once the ultrasound scan was completed, patients underwent UGE under light IV sedation as per current standard practice. This consisted of IV midazolam 1 mg increments (up to a maximum of 2 mg) and IV fentanyl 50 ug increments (up to a maximum of 100 ug). The UGE was performed by an attending gastroenterologist per current standard institutional practice using an Olympus fiberoptic endoscope. Gastroenterologists were blinded to the volume ingested by the subjects and unaware of the ultrasonographic findings. All gastric fluid was thoroughly suctioned through an endoscope side port, measured to the nearest mL and recorded. This volume was called the “observed volume.”

Back to Top | Article Outline

Sample Size Calculation and Statistical Analysis

The mathematical model being tested is expressed as

Log (Right – latCSA) = 1.318 + 0.00354* predicted volume + 0.00594* weight (1)

To validate this model, a linear regression model was fitted, having as outcome the logarithm of the observed volume and as unique predictor the predicted volume.17 The fitted regression model can be expressed as

observed volume = a + b* predicted volume (2)

If a = 0 and b = 1, the original model will be proven adequate to describe the new data. If, on the other hand, there is a significant deviation from those values, an update of the original prediction model would be recommended. For the sample size calculation we tested the null hypothesis that H0: b = 1, against the alternative H1: b = 0.75 using estimates of the standard deviations of the observed volume and right lateral CSA from previous studies.5,7 We estimated that 110 patients are needed for a power of 80% to detect the alternative hypothesis H1: b = 0.75, using a two-sided test with alpha level = 0.05. For the calculation we use PASS software version 08.05, for Windows XP. In addition a Bland-Altman analysis was used to determine the limits of agreement between the predicted volume and the observed volume. If a new model needed to be fitted, the R2, Akaiki Information Criterion and Root Mean Squared Error model fit criteria would be used. All patient demographics reported were checked in their ability to improve the accuracy of the model.

Back to Top | Article Outline

RESULTS

One hundred ten patients were enrolled in the study. One patient was withdrawn from the study due to a long delay between ultrasound assessment and gastroscopic examination (more than 25 minutes) due to logistical issues. One further patient was withdrawn from the study because a significant amount of air in the stomach impaired proper evaluation of the antrum. The remaining 108 patients (39 males, 69 females) completed all study assessments according to protocol and were included in the final analysis. Patient demographics are summarized in Table 1. The regression model using the observed volume as outcome and the predicted volume as predictor, shows a large deviation from the ideal parameters of intercept equal to 0 (-14.3, SE 17.1, P= 0.41) and a beta coefficient equal to 1 (0.78 ± 0.06, P = 0.0006). In other words, observed volume = 0.78 predicted volume – 14. Pearson’s correlation coefficient for this model (r2) is 0.656.

Table 1

Table 1

A Bland–Altman analysis was performed (Figure 2). This type of analysis plots the difference between the estimated volume based on the model being tested and the observed volume (suctioned during gastroscopy) for each subject, against the mean difference of the two values and allows us to place the differences in a clinical context. The 95% limits of agreement calculated by this method represent differences likely to arise between the two measurements with a 95% probability. This analysis suggests that the existing mathematical model tends to overestimate gastric volume with a mean “bias” or systematic error of 70 mL (Figure 2). Additionally, the upper limit of a 95% agreement band is over 230 mL. This difference is clinically relevant, especially at low volume states, and suggests that the existing model may not be accurate enough for clinical use. We therefore decided to fit a new model based on the new dataset considered to be a more accurate “gold standard.”

Figure 1

Figure 1

Figure 2

Figure 2

Back to Top | Article Outline

Fitting a New Model

The raw values of observed volume are presented on Figure 3. Based on these data, we fitted a number of linear regression models of the observed (suctioned) volume (the dependent variable) predicted by the measured Right lat CSA (independent variable), weight, height, body mass index (BMI), age and gender. We also investigated whether the fit of the models was improved if Right lat CSA was log-transformed. Using R2, Akaiki Information Criterion and Root Mean Squared Error model fit criteria, we selected the model containing as predictors Right lat CSA and age as the best fit as follows:

Figure 3

Figure 3

Volume = 27.0 + 14.6 *Right – latCSA – 1.28* age (3)

This model shows a significant correlation to the new dataset as assessed by Pearson’s correlation coefficient (r2 = 0.731, Tables 2 and 3). The remaining demographic variables (weight, height and gender) were not found to be independent predictors of gastric volume. The residual plots for both independent predictors (Right lat CSA and age) are shown in Figure 4. These plots suggest that the linear regression assumption is valid, and that no polynomial transformation is required. Of note, this model is only applicable when the result is a positive value. Thus, when the stomach is empty, small values of Right lat CSA will yield a negative volume value, which only indicates an empty state. In addition, the best fit regression model proposed was validated with the use of a procedure involving partitioning of the data into two sets, one for “training” and one for validating the model, each set comprising 50% of the original data. The Average Squared Error (ASE, equal to the squared difference between the true and estimated volume values averaged over all data points) was calculated. A smaller ASE indicates a better fit. Two possible models were assessed: one with the age as a covariate, and one without the age. This validation procedure showed that the model with age covariate had a smaller ASE value for the validation set (6849.9) than the model without age (7179.1), and therefore was preferred. A Bland-Altman analysis was performed using the “validating” dataset (50% of the data set aside for validating the model). The results are shown in Figure 5. This analysis suggests that the new model predicts gastric volume more accurately, or with a higher level of agreement than the previous model. The new “bias” or systematic error is only 6 mL, and the upper and lower limits of agreement of a 95% agreement band are considerably smaller.

Table 3

Table 3

Table 2

Table 2

Figure 4

Figure 4

Figure 5

Figure 5

In addition, we also evaluated a previously described 3-point grading system of the gastric antrum and its correlation with gastric fluid volume.7 The gastric antrum is classified as grade 0 when it appears empty in both supine and right lateral decubitus positions. The antrum is classified as grade 1 when clear fluid content is visible only in the right lateral decubitus position suggesting a small fluid volume. Finally, a grade 2 antrum is that in which clear fluid is apparent in both supine and right lateral decubitus positions, suggesting a larger gastric fluid volume. To this end we performed a graphical exploratory analysis by plotting estimates for the densities of the gastric fluid volume for the patients classified as having Grade 1 and Grade 2 antrum (Figures 6 and 7). No subjects in this study were classified as grade 0. Figure 7 illustrates that a Grade 1 antrum is associated with lower gastric volumes than a grade 2 antrum. In the sample studied, only 23% of subjects with a grade 1 antrum had a gastric volume larger than 100 mL, and none of them had a volume larger than 250 mL. In contrast, 75% of subjects with a grade 2 antrum had a volume higher than 100 mL, and more than 50% had volumes larger than 250 mL. In other words, based on binomial proportions with continuity corrections, our data suggest that the absolute risk difference for having more than 100 mL is 52 percentage points (23% for grade 1 to 75% for grade 2). The absolute risk increase is 52%, 95% CI 31–73%, Wald test P< 0.0001).18 This conventional volume threshold of 100 mL is clinically relevant since it is the upper limit of what is considered normal baseline gastric secretions posing no significant aspiration risk in fasted surgical patients.11–15 This suggests that this 3-point grading system based on qualitative sonography alone can serve as a screening tool to differentiate between low and high volume states.

Figure 6

Figure 6

Figure 7

Figure 7

Back to Top | Article Outline

DISCUSSION

The present experimental study of 108 patients tested the performance of our previously reported mathematical model by comparing the predicted volume based on antral sonography to the observed volume measured by suctioning all gastric contents under gastroscopic examination, arguably the gold standard for gastric volume assessment. Based on our results, we conclude that the previously proposed model tends to overestimate gastric volume to a degree that is clinically significant (mean of 72 mL, but 95% CI up to 200 mL). Therefore we propose a revised mathematical model that better describes the distribution of the new dataset, considered to be more accurate. We believe this new revised model has several advantages over both existing models. First, it is built on data obtained with a more rigorous method of gastric fluid measurement (suctioning under gastroscopic examination). Second, it is statistically robust with a high correlation coefficient (Rsq 0.731). Third, the new revised model is applicable to a wider range of volumes (up to 500 mL) than previous models (up to 250 mL in the model by Bouvet et al and up to 300 mL in the model by Perlas et al). Finally, unlike the model proposed by Bouvet et al, our revised model is only influenced by patient age and is independent of all other patient demographic variables within the demographic range studied (148 to 188 cm tall, 45–110 kg, BMI 17 to 39 kg/m2). This makes the model simpler to apply in clinical practice. Table 3 results from solving the Best fit model equation in 2 cm2 intervals for values of Right-lat CSA for every decade of life. The shaded cells in the table represent low volume states that are well accepted in the literature to carry negligible aspiration risk.

The effect of age on predicted volume is intriguing and somewhat unexpected. For any given gastric fluid volume, there is a trend for older patients to have a higher Right lat antral CSA than their younger counterparts (Table 3). Or, in other words, a given Right lat antral CSA corresponds to a lower gastric fluid volume in older versus younger patients. For example, a CSA of 10 cm2 corresponds to a 147 mL of gastric fluid in a 20 year old patient but only 71 mL of gastric fluid in an 80 year old patient. This could possibly be explained by a more compliant gastric wall in older versus younger patients.

The present study is not without limitations. First, although the method to measure gastric volume is likely more precise than those used in previous studies, no method is infallible. Despite our best efforts to minimize the transition times between study interventions, there was a mean lag of 9 minutes between ultrasound examination and gastric suctioning, during which time some gastric emptying may have occurred. To minimize this possibility, patients were placed in the left lateral decubitus position during this time interval, and a clear fluid with high caloric content (apple juice) was used which is emptied more slowly than noncaloric clear fluids such as water. Second, the model proposed herein is applicable to a wide range of adult individuals but not to all subjects. In particular, it is not applicable to children or adolescents, pregnant women, adults with a BMI over 40 kg/m2, or with underlying anatomical abnormalities of the upper gastrointestinal tract.

Finally, this prediction model is only applicable when a similar scanning protocol to that used in this study is followed. The gastric antrum needs to be identified in cross-section in a right parasagittal plane with the subject placed in the right lateral decubitus. We have previously demonstrated that for any given fluid volume, antral CSA is larger in the right lateral versus the supine position due to a gravitational fluid shift towards the antrum, and a model based on Right lat CSA is more sensitive than its supine counterpart to detect volume changes particularly in low volume states.5 This may explain some of the differences between our model and that proposed by Bouvet et al. based on supine assessments.

All measurements need to be taken with the antrum at rest, between peristaltic contractions. Measuring the antrum during a peristaltic contraction would yield a lower Right lat CSA value and would underestimate gastric volume. Finally, for this model, antral CSA is measured from serosa to serosa (including the full thickness of the gastric wall).

The use of sedative medications immediately before UGE on patients who have ingested up to 400 mL of apple juice is potentially controversial. The following precautions were taken to minimize aspiration risk: doses were titrated to achieve anxiolysis only, while maintaining the subjects awake; sedatives were administered immediately before the procedure; patients were placed in the left lateral decubitus position for the gastroscopic examination, and the stomach was thoroughly suctioned as soon as the gastroscope entered the gastric cavity. The study protocol was developed in collaboration with the participating gastroeneterologists and approved by the institutional research ethics board. No cases of aspiration occurred during the study.

Many questions remain to be answered in the future regarding the clinical applicability of this diagnostic tool. Questions regarding the cost effectiveness of this type of assessment, the reliability of measurements, as well as the level of skill and training required to achieve competence are outside the scope of this study and need to be addressed by future studies.

Back to Top | Article Outline

CONCLUSION

Using direct suctioning of gastric fluid under gastroscopic examination on 108 patients, we report a new prediction model (Volume = 27.0 + 14.6 *Right-lat CSA – 1.28*age) to assess gastric volume noninvasively at the bedside based on sonographic measurements of Right lat CSA. This model can predict volumes from 0 to 500 mL and is applicable to nonpregnant adult patients with BMI < 40 kg/m2. In addition, our data also suggest that a simple 3-point grading system based on qualitative antral assessment, as previously described, could help differentiate low volume from higher volume states. Further research is needed to evaluate the impact of both qualitative and quantitative gastric ultrasound assessment on bedside aspiration risk assessment and patient management

Back to Top | Article Outline

DISCLOSURES:

Name: Anahi Perlas, MD, FRCPC

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript

Attestation: Anahi Perlas has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Nicholas Mitsakakis, M.Sc, PhD

Contribution: This author helped analyze the data and write the manuscript

Attestation: Nicholas Mitsakakis has seen the original study data, reviewed the analysis of the data, and approved the final manuscript

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Louis Liu, MD, FRCPC

Contribution: This author helped conduct the study, analyze the data, and write the manuscript

Attestation: Louis Liu reviewed the analysis of the data and approved the final manuscript

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Maria Cino, MD, FRCPC

Contribution: This author helped conduct the study and write the manuscript

Attestation: Maria Cino reviewed the analysis of the data and approved the final manuscript

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Nidhi Haldipur, BSc

Contribution: This author helped conduct the study and write the manuscript

Attestation: Nidhi Haldipur reviewed the analysis of the data and approved the final manuscript

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Liisa Davis, BSc, RDMS

Contribution: This author helped conduct the study

Attestation: Liisa Davis approved the final manuscript

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Javier Cubillos, MD

Contribution: This author helped conduct the study

Attestation: Javier Cubillos approved the final manuscript

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Vincent Chan, MD, FRCPC, FRCA

Contribution: This author helped design the study and write the manuscript

Attestation: Vincent Chan reviewed the analysis of the data and approved the final manuscript

Conflicts of Interest: Vincent Chan reported a conflict of interest with Philips, reported a conflict of interest with SonoSite, and reported a conflict of interest with General Electric Equipment and support for educational activities (workshops and conferences)

This manuscript was handled by: Dwayne R. Westenskow, PhD

Back to Top | Article Outline

REFERENCES

1. Lockey DJ, Coats T, Parr MJA. Aspiration in severe trauma: a prospective study. Anesthesia. 1999;54:1097–1099
2. Warner MA, Warner ME, Weber JG. Clinical significance of pulmonary aspiration during the perioperative period. Anesthesiology. 1993;78:56–62
3. Lienhart A, Auroy Y, Péquignot F, Benhamou D, Warszawski J, Bovet M, Jougla E. Survey of anesthesia-related mortality in France. Anesthesiology. 2006;105:1087–97
4. Cotton BR, Smith G. The lower oesophageal sphincter and anaesthesia. Br J Anaesth. 1984;56:37–46
5. Perlas A, Chan VW, Lupu CM, Mitsakakis N, Hanbidge A. Ultrasound assessment of gastric content and volume. Anesthesiology. 2009;111:82–9
6. Bouvet L, Mazoit JX, Chassard D, Allaouchiche B, Boselli E, Benhamou D. Clinical assessment of the ultrasonographic measurement of antral area for estimating preoperative gastric content and volume. Anesthesiology. 2011;114:1086–92
7. Perlas A, Davis L, Khan M, Mitsakakis N, Chan VW. Gastric sonography in the fasted surgical patient: a prospective descriptive study. Anesth Analg. 2011;113:93–7
8. Carp H, Jayaram A, Stoll M. Ultrasound examination of the stomach contents of parturients. Anesth Analg. 1992;74:683–7
9. Koening S, Lakticova V, Mayo P. Utility of ultrasonography for detection of gastric fluid during urgent endotracheal intubation. Intensive Care Med. 2011;134:2125–9
10. Cubillos J, Tse C, Chan VW, Perlas A. Bedside ultrasound assessment of gastric content: an observational study. Can J Anaesth. 2012;59:416–23
11. Agarwal A, Chari P, Singh H. Fluid deprivation before operation. The effect of a small drink. Anaesthesia. 1989;44:632–4
12. Read MS, Vaughan RS. Allowing pre-operative patients to drink: effects on patients’ safety and comfort of unlimited oral water until 2 hours before anaesthesia. Acta Anaesthesiol Scand. 1991;35:591–5
13. Phillips S, Hutchinson S, Davidson T. Preoperative drinking does not affect gastric contents. Br J Anaesth. 1993;70:6–9
14. Harter RL, Kelly WB, Kramer MG, Perez CE, Dzwonczyk RR. A comparison of the volume and pH of gastric contents of obese and lean surgical patients. Anesth Analg. 1998;86:147–52
15. Hausel J, Nygren J, Lagerkranser M, Hellström PM, Hammarqvist F, Almström C, Lindh A, Thorell A, Ljungqvist O. A carbohydrate-rich drink reduces preoperative discomfort in elective surgery patients. Anesth Analg. 2001;93:1344–50
16. Bolondi L, Bortolotti M, Santi V, Calletti T, Gaiani S, Labò G. Measurement of gastric emptying time by real-time ultrasonography. Gastroenterology. 1985;89:752–9
17. Draper NR, Smith H Applied regression analysis. 19812nd ed New York John Wiley and Sons Inc
18. Fleiss JL, Levin B, Paik MC Statistical methods for rates and proportions. 20033rd ed New York John Wiley and Sons Inc
© 2013 International Anesthesia Research Society