Perioperative Medicine

ABSTRACT
Background: The physiology of diabetes mellitus can increase the risk of
perioperative aspiration, but there is limited and contradictory evidence on the

Baseline Gastric Volume in incidence of “full stomach” in fasting diabetic patients. The aim of this study
is to assess the baseline gastric content (using gastric ultrasound) in diabetic

Fasting Diabetic Patients

and nondiabetic patients scheduled for elective surgery who have followed
standard preoperative fasting instructions.

Is Not Higher than That Methods: This was a prospective, noninferiority study of 180 patients (84
diabetic and 96 nondiabetic patients). Bedside ultrasound was used for qual-

in Nondiabetic Patients: itative and quantitative assessment of the gastric antrum in the supine and

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

right lateral decubitus positions. Fasting gastric volume was estimated based

A Cross-sectional on the cross-sectional area of the gastric antrum and a validated model. The
hypothesis was that diabetic patients would not have a higher baseline fasting

Noninferiority Study gastric volume compared to nondiabetic patients, with a noninferiority margin
of 0.4 ml/kg. Secondary aims included the comparison of the incidence of
full stomach (solid content or more than 1.5 mL/kg of clear fluid), estimation
Anahi Perlas, M.D., F.R.C.P.C., Maggie Z. X. Xiao, B.Sc., of the 95th percentile of the gastric volume distribution in both groups, and
George Tomlinson, Ph.D., Binu Jacob, Ph.D., examination of the association between gastric volume, glycemic control, and
Sara Abdullah, M.D., Richelle Kruisselbrink, M.D., F.R.C.P.C., diabetic comorbidities.
Vincent W. S. Chan, M.D., F.R.C.P.C.
Results: The baseline gastric volume was not higher in diabetic patients
Anesthesiology 2024; 140:648–56 (0.81 ± 0.61 ml/kg) compared to nondiabetic patients (0.87 ± 0.53 ml/kg)
with a mean difference of −0.07 ml/kg (95% CI, −0.24 to 0.10 ml/kg). A total
of 13 (15.5%) diabetic and 11 (11.5%) nondiabetic patients presented more
EDITOR’S PERSPECTIVE than 1.5 ml/kg of gastric volume (95% CI for difference, −7.1 to 15.2%).
There was little correlation between the gastric volume and either the time
What We Already Know about This Topic since diagnosis or HbA1C.
• Aspiration of gastric contents during induction of anesthesia can Conclusions: The data suggest that the baseline gastric volume in diabetic
lead to significant morbidity or mortality. patients who have followed standard fasting instructions is not higher than
• Diabetes mellitus may lead to delayed gastric emptying due to auto- that in nondiabetic patients.
nomic dysfunction.
• Few data are available regarding the frequency of “full stomach” in (Anesthesiology 2024; 140:648–56)
diabetic versus nondiabetic patients following fasting guidelines for
elective surgery.

What This Article Tells Us That Is New • Current fasting guidelines by the American Society of Anesthesiologists
are similarly effective in diabetic and nondiabetic patients.
• In this prospective, noninferiority study of 180 fasted patients using
bedside ultrasound, gastric volume was not higher in diabetic ver-

A
sus nondiabetic patients. The frequency of full stomach was also spiration of gastric contents is the single most common
similar between the two groups.
cause of death in airway-related incidents.1 One of the

This article is featured in “This Month in Anesthesiology,” page A1. This article is accompanied by an editorial on p. 639. This article has a related Infographic on p. A16. This article
has an audio podcast. This article has a visual abstract available in the online version. Part of the work presented in this article has been presented as an abstract at the University
of Toronto Department of Anesthesiology and Pain Medicine Shield’s Research Day in Toronto, Ontario, Canada, May 5, 2023, and at the Canadian Anesthesiologists’ Society Annual
meeting in Quebec, Quebec, Canada, June 10, 2023.
Submitted for publication August 26, 2022. Accepted for publication October 13, 2023. Published online first on October 26, 2023.
Anahi Perlas, M.D., F.R.C.P.C.: Department of Anesthesiology and Pain Medicine, University of Toronto, Toronto, Ontario, Canada; and Department of Anesthesia and Pain
Management, Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada.
Maggie Z. X. Xiao, B.Sc.: University of Alberta, Edmonton, Alberta, Canada.
George Tomlinson, Ph.D.: Biostatistical Research Unit, University Health Network, Toronto, Ontario, Canada.
Binu Jacob, Ph.D.: Department of Anesthesia and Pain Management, Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada.
Sara Abdullah, M.D.: Department of Anesthesia and Pain Management, Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada.
Richelle Kruisselbrink, M.D., F.R.C.P.C.: Department of Anesthesia, McMaster University, Hamilton, Ontario, Canada.
Vincent W. S. Chan, M.D., F.R.C.P.C.: Department of Anesthesiology and Pain Medicine, University of Toronto, Toronto, Ontario, Canada.
Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Anesthesiology 2024; 140:648–56. DOI: 10.1097/ALN.0000000000004815

648 April 2024 Anesthesiology, V 140 • NO 4

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.
 Gastric Volume and Diabetes

main risk factors for pulmonary aspiration is the presence of a university-based tertiary care center in Toronto, Ontario,
a “full stomach” at the time of induction, which may be due Canada, between September 2016 and February 2022.
to delayed gastric emptying or noncompliance with fasting Approval was obtained by the University Health Network
instructions. While the absolute incidence of aspiration is Research Ethics Board (approval No. 15-8808-AE), and the
low in elective surgical patients (less than 1:3,000), diabetic study protocol was registered in the clinicaltrials.gov registry
patients are often considered to be at increased risk due to system on September 5, 2016 (NCT02888951). This study
the potential for delayed gastric emptying related to auto- follows the Strengthening the Reporting of Observational
nomic dysfunction.2 However, there is limited and conflict- Studies in Epidemiology (STROBE) reporting guidelines.
ing evidence concerning the incidence of a full stomach
in fasting diabetic patients, and there are no epidemiologic Inclusion and Exclusion Criteria
studies of aspiration in this particular patient population.3
Diabetic and nondiabetic patients scheduled for elective sur-

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

Moreover, although gastroparesis may present with nonspe-
gery were screened to participate in the study. Inclusion cri-
cific symptoms such as postprandial fullness, early satiety,
teria were age 18 to 85 yr, ASA Physical Status I to III, and
bloating, nausea, and vomiting, many are asymptomatic,
body mass index less than 40 kg/m2. Patients were excluded
which makes preoperative risk stratification difficult.4
if they had a history of gastrointestinal surgery, were preg-
As there are no consistent specific clinical features that
nant (currently or within the past 3 months), or had doc-
define gastroparesis, the primary risk reduction strategy is to
umented abnormalities of the upper gastrointestinal tract
ensure an “empty” stomach at the time of anesthetic induc-
such as achalasia. Eligible diabetic and nondiabetic subjects
tion. The latest 2017 fasting guidelines by the American
were enrolled in a 1:1 ratio. Participants were consecutively
Society of Anesthesiologists (ASA; Schaumburg, Illinois)
enrolled as permitted by surgical lists and availability of
aim to limit residual gastric volume by recommending a
research team members, and no effort was made to maintain
time frame in which to abstain from food and drinks before
balance between the two groups throughout. All patients
elective surgery: a minimum of 2 h of fasting for clear fluids,
provided written, informed consent before enrollment.
6 h for a light meal, and 8 h of fasting for a full meal contain-
ing fatty food.5 However, these guidelines are only appli-
cable to low-risk, otherwise healthy patients undergoing Study Design
elective surgery, and there is no consensus on whether the All patients (diabetic and nondiabetic) followed the stan-
standard fasting intervals are adequate in patients with dia- dard ASA preoperative fasting guidelines: a minimum fasting
betes.5 The guidelines state that patients with diabetes may duration of 2 h after ingesting clear fluid, 6 h after a light
require longer fasting intervals to ensure an empty stomach. meal, and 8 h for a full meal, although we note that the latest
As such, it remains equivocal whether diabetic patients fol- iteration of the guidelines published in 2017 were established
lowing the same ASA fasting guidelines have similar gastric after enrollment started in 2016.The following demographic
residual volumes to nondiabetic patients.3 In addition, the and clinical characteristics were collected prospectively: age,
relationship between the duration of diabetic disease, glyce- sex, height, weight, ASA Physical Status classification (I to
mic control, and gastric residual volume is poorly defined. IV), type of surgery, and anesthetic technique.
Bedside gastric ultrasound is a noninvasive, accurate, and Other relevant data, including medication history (use of
reliable tool to assess gastric content and inform aspiration antacids and motility or prokinetic agents), comorbid med-
risk assessment at the bedside before anesthesia and surgery.6 ical conditions (gastroesophageal reflux disease, obstructive
A bedside ultrasound exam can readily distinguish an empty sleep apnea, and liver dysfunction), and laboratory values
stomach from one with clear fluid or solid content.6 When (hemoglobin, fasting blood glucose, serum creatinine, and
clear fluid is present, a gastric volume greater than 1.5 ml/kg estimated glomerular filtration rate) were also documented.
is the most accepted critical volume threshold of an at-risk Diabetic patients were also queried for diabetes mellitus
stomach.7 The primary aim of the study is to determine class (type I or II), years since diagnosis, type of treatment
whether current preoperative fasting guidelines consistently (diet-controlled, oral hypoglycemics, or insulin), glyco-
ensure an empty stomach in diabetic patients. Secondarily, we sylated hemoglobin concentrations (HbA1c), gastroparesis
sought to evaluate the effects of diabetic severity and dura- symptoms (anorexia, early satiety, or postprandial fullness),8
tion on preoperative residual gastric volume. Our hypothesis and evidence of end-organ complications (retinopathy,
is that diabetic patients who have followed standard fasting nephropathy, neuropathy, autonomic dysfunction, coronary
instructions before elective surgery have a baseline gastric artery disease, stroke or transient ischemic attack, peripheral
volume that is not higher than that in nondiabetic patients vascular disease, and diabetic ulcers). Episodes of aspiration
with a noninferiority margin of 0.4 ml/kg. were noted, if any.

Materials and Methods Ultrasound Examination

This single-center, prospective, noninferiority cross- A bedside gastric ultrasound was performed preoperatively,
sectional study was conducted at Toronto Western Hospital, on the morning of surgery by an experienced anesthesiologist

Perlas et al. A nesthesiology 2024; 140:648–56 649

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.
 Perioperative Medicine

blinded to the patient’s diabetes mellitus status using a stan-

dardized gastric scanning protocol, previously described.9
The ultrasound examinations were performed or directly
supervised by one of two investigators (A.P. or R.K.), both
of whom have several years of experience in gastric ultra-
sound and have been involved in multiple related stud-
ies using gastric ultrasound. These investigators were not
involved in patient screening and enrollment, which was
performed by an independent research assistant to ensure
blinding. Briefly, the qualitative and quantitative assessment
of the gastric antrum was performed in the supine posi-

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

tion and then in the right lateral decubitus position using
a portable ultrasound unit (Sonosite Edge system, Sonosite
Inc., USA) and a curved array low-frequency (2 to 5 MHz)
transducer. The gastric antrum was located between the left
lobe of the liver and the pancreas at the level of the abdom-
inal aorta.10,11
The sonographic appearance of the antrum was used to Fig. 1. Sonographic image of the gastric antrum in the epi-
determine the nature of gastric content: empty (flat antrum), gastric area. Note the yellow dotted line indicating how cross-
clear fluid (hypoechoic content), or solid (heterogeneous, sectional area of the antrum is measured by tracing the outer
hyperechoic content).9–11 The semiquantitative three-point surface of the gastric antrum at the level of the aorta. A, antrum;
Ao, aorta; L, liver; P, pancreas; Rm, rectus muscle.
Perlas grading system was used to discriminate between low
and high gastric fluid volumes.10 Grade 0 refers to an empty
antrum in both supine and right lateral decubitus positions, reliable in a study comparing the measurements of three
suggesting no gastric content. A grade 1 antrum has clear independent blinded observers with intraclass correlation
fluid that is appreciable in the right lateral decubitus position coefficients greater than 0.8, which are considered to be
only and corresponds with a low-volume state, commonly “almost perfect” reliability in the literature.14
observed in healthy fasting patients.6,10 A grade 2 antrum An empty stomach was defined as a grade 0 antrum (no
contains clear fluid in both supine and right lateral decubitus fluids or solid content appreciable) or a gastric residual vol-
positions and correlates with higher gastric volumes.6 ume of less than 1.5 ml/kg of clear fluid, consistent with
For the quantitative component, the cross-sectional baseline gastric secretions. A full stomach was defined as the
area of the gastric antrum was measured in the right lateral presence of solid content or greater than 1.5 ml/kg of clear
decubitus position, including the full thickness of the gas- fluid, suggesting greater than normal baseline volume.6,10
tric wall, using the free-tracing function of the ultrasound The examinations in which the gastric antrum was not
unit (fig. 1).6 The gastric fluid volume was estimated using visualized well enough for a meaningful assessment were
the following previously validated model by Perlas et al.11,12: considered “inconclusive,” and those patients were excluded
gastric volume (ml) = 27.0 + 14.6 × right lateral decubi- from the final analysis. The results of the ultrasound exam-
tus cross-sectional area (cm2) – 1.28 × age (yr), where the ination were used for research purposes only. As it is not
right lateral decubitus cross-sectional area is the antral cross- standard practice to perform gastric ultrasound in elective
sectional area measured in the right lateral decubitus surgical patients who have followed fasting instructions, the
position. This model originated in a study that obtained results of this examination were not shared with the clinical
measurements of gastric antral area in the right lateral decu- team, and neither did they inform anesthetic care.
bitus performed by a blinded sonographer on 108 patients
who had ingested 0 to 400 ml of clear fluid (apple juice)
assigned at random immediately before gastroscopy. These Outcomes
measurements were then correlated to the total gastric vol- The primary aim of the study was to determine whether
ume suctioned during the gastroscopic exam performed current preoperative fasting guidelines consistently ensure
just a few minutes later by a blinded gastroenterologist. an empty stomach in diabetic patients. To that end, we
From the data set, 50% of values were used for model devel- compared residual gastric volume (estimated in ml/kg) in
opment, and the other 50% were used for model valida- diabetic and nondiabetic patients scheduled for elective
tion. The mean bias or systematic error of the model was surgery with a noninferiority study design. The nonin-
reported to be 6 ml.12 This model was further validated in feriority margin for the mean difference was predefined
a separate study using a similar methodology on patients as a 0.4 ml/kg higher gastric volume in diabetic patients.
with body mass index greater than 40 kg/m2.13 This method Secondarily, we sought to (1) describe the distribution of
for gastric volume evaluation has been found to be highly baseline gastric volume in fasting diabetic and nondiabetic

650 A nesthesiology 2024; 140:648–56 Perlas et al.

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.
 Gastric Volume and Diabetes

patients; (2) compare the incidence of full stomach (defined endpoint of baseline gastric volume (in ml/kg) used the
as the presence of solid content or greater than 1.5 ml/kg unpaired Student’s t test to calculate a two-sided 95% CI
of clear fluid) in fasted diabetic versus nondiabetic subjects; for the mean difference in volume between diabetic and
(3) estimate the upper limit of normal fasting gastric vol- nondiabetic patients. If the upper limit of the interval was
ume (defined as the 95th percentile) in each cohort; (4) less than 0.4 ml/kg, noninferiority was declared. To check
evaluate the association between gastric fluid volume, the the sensitivity of the assumptions of normality made by the
duration of disease, degrees of glycemic control (dichoto- t test, a percentile-based CI was calculated using the boot-
mized into HbA1c less than 7% vs. HbA1C 7% or more), strap (with 2,000 replications) and a CI for the difference in
and the type of therapy required for diabetes (diet only, oral medians was calculated using the Wilcoxon rank-sum test.
hypoglycemic agents, or insulin); and (5) report the inci- For secondary objectives, (1) the 95% CI for the difference
dence of episodes of intraoperative regurgitation, vomiting, in the incidence of full stomach (solid content or greater

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

or aspiration. than 1.5 ml/kg of clear gastric content) between the two
groups was computed using the normal approximation to
Sample Size and Data Analysis the difference in two proportions; (2) the distributions of
the values of antral grades and gastric volume was visu-
The statistical plan was written and filed with a private ally inspected with density and quantile–quantile diagnostic
entity (institutional research ethics board) before any data plots and Kolmogorov–Smirnov two-sample test; (3) the
were accessed. The initial plan was to do a superiority anal- CIs for the 95th percentiles of antral cross-sectional area
ysis, which corresponded to a sample size of 260. In late and gastric volume were calculated using a binomial-based
2019, the study was presented at internal departmental method; (4) the associations between gastric volume and
research rounds, before accessing any data, and during the duration and type (I and II) of diabetes were summarized
enrollment period. At that time, it was recommended that by Pearson correlation coefficient; and (5) the gastric vol-
a noninferiority analysis would be more consistent with ume was summarized by type of therapy (diet only, oral
our hypothesis that diabetic patients would not have higher hypoglycemic agents, or insulin). Except for those for the
baseline volume than nondiabetic patients. An amendment noninferiority analyses, all statistical tests were two-tailed,
was submitted to the institutional research ethics board in and P values less than 0.05 were taken to indicate statistical
2020 requesting the change in analysis plan which resulted significance.
in a changed sample size to 190. This was approved by the
board in 2020 before accessing any data and before enroll-
ment was complete. Results
The sample size calculation was then based on a non- A total of 208 patients were screened for eligibility between
inferiority design. The primary outcome measure was gas- September 2016 and February 2022. Of these patients, 191
tric volume per unit of weight (ml/kg). Based on previous (91 diabetic and 100 nondiabetic) met the inclusion criteria
studies, we assumed a mean value in the control group of and provided written consent for enrollment; 11 patients
0.8 ml/kg and a SD of 0.8 ml/kg. We hypothesized that had inconclusive ultrasound examinations with missing
baseline gastric volume would not be higher in diabetic data that precluded a meaningful analysis (drop-out rate,
patients compared to nondiabetic patients with a noninfe- 6.8%). After inconclusive exams were excluded, 180 sub-
riority margin of 0.4 ml/kg. This noninferiority margin is jects (84 diabetic and 96 nondiabetic) were included in the
half of the anticipated within-group SD, a value that is gen- final analysis (fig. 2).
erally considered a small effect. This value was also judged Patient characteristics are summarized in table 1.
to be a clinically unimportant difference. To achieve 90% Compared with control subjects, diabetic patients were
power to reject this margin and claim noninferiority with a older by almost 10 yr, had a higher body mass index and a
one-sided type I error of 0.025, a minimum of 85 subjects higher ASA classification score, and reported higher rates of
per group was required. To account for 10% of participants obstructive sleep apnea, lipid dysfunction, and use of antac-
being withdrawn or nonevaluable, the total number was ids.The intended inclusion criteria was ASA Physical Status
increased to 190 subjects. I to III. Although all patients in the study were classified as
Data analysis was completed using R software version ASA Physical Status I to III during the screening process,
4.2.0. Categorical data were summarized as counts and per- two patients were classified as ASA Physical Status IV at
centages (%) within the diabetes and nondiabetes groups. the time of surgery by the attending anesthesiologist, so the
The standardized differences in baseline characteristics classification was changed for those two patients.
between the groups were calculated; these quantify the dif- The most common surgical procedures were orthope-
ference between groups in units of standard deviations and dics (60.2%), ophthalmology (13.6%), plastics (7.3%), and
allow judgment of which variables differ the most and least. urology (6.3%). Among diabetic patients, the mean glyco-
Continuous data were summarized by the mean and SD or sylated hemoglobin (HbA1c) was 7.2% (1.5%), and 41%
median and interquartile range as appropriate.The noninfe- had HbA1C levels above 7%. The mean time since diag-
riority comparison between the two groups for the primary nosis of diabetes mellitus was 12.2 (11.6) years. The most

Perlas et al. A nesthesiology 2024; 140:648–56 651

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.
 Perioperative Medicine

was −0.22 to 0.10, much the same as in the analysis with

complete case data.
(1) Worst case: assigning the 95th percentile within-group
to missing values. The 95% CI from this procedure was
−0.18 to 0.16, with an upper limit of the CI still sub-
stantially below the noninferiority margin of 0.4.
(2) Best case: assigning the 5th percentile within-group to
missing values. The 95% CI from this procedure was
−0.26 to 0.07, with an upper limit of the CI below that
from our primary analysis
(3) Worst case/best case: in the diabetic group, assign the

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

95th within-group percentile to missing values, and
in the nondiabetic group, assign the 5th within-group
percentile. This extreme (and unrealistic) case assumes
that all missing values were high for diabetic patients
and low for nondiabetic patients. The 95% CI from this
procedure was −0.13 to 0.23, a finding that still satisfies
the noninferiority criterion.
The gastric ultrasound findings are presented in table 2.
The incidence of full stomach, defined as clear fluid volume
in excess of 1.5 ml/kg, was not higher in the diabetic group.
There were no significant differences in the distribution of
the antral grades in both groups (table 2). Of the diabetic
patients, 31 (36.5%) had a grade 0 antrum versus 30 (31.2%)
control patients; 42 (49.4%) presented a grade 1 antrum ver-
sus 50 (52.1%) control patients; and 12 (14.1%) had a grade
2 antrum versus 16 (16.7%) control patients. No patient was
found to have solid content.
Fig. 2. Flow chart describing patient flow and the results of the
ultrasound examination.
The mean antral cross-sectional area in the right lat-
eral decubitus position was no different in the two groups
(8.3 ± 3.6 cm2 in diabetic patients and 7.8 ± 3.0 cm2 in con-
trol subjects, P = 0.37), and the distribution of gastric fluid
prevalent diabetic comorbidities and complications were volumes was not higher in diabetic compared to nondia-
retinopathy (38.5%), neuropathy (27.5%), and nephrop- betic patients (fig. 4). Further, the 95th percentile for gastric
athy (16.5%). Gastroparesis symptoms such as postpran- volume was similar between the two groups: 1.94 (95% CI,
dial fullness, early satiety, and anorexia were reported in 1.73 to 2.48) ml/kg in the diabetes cohort and 1.73 (95%
34.1, 28.6, and 12.1% of diabetic patients, respectively. CI, 1.49 to 2.20) ml/kg in controls.
Approximately 10% of diabetic patients had documented There were no significant associations between gastric
autonomic dysfunction. volume and the type of diabetes (I or II; mean difference
The mean fasting gastric volume was not higher in dia- II vs. I, 0.09 ml/kg (95% CI, −0.28 to 0.47), duration of
betic patients (0.81 ± 0.61 ml/kg) compared to nondiabetic disease (0.00 ml/kg change in volume for each 5 yr of dis-
controls (0.87 ± 0.53 ml/kg). The mean difference in gas- ease; 95% CI, −0.06 to 0.06), or HbA1c level (0.03 ml/kg
tric volume was −0.07 ml/kg (95% CI, −0.24 to 0.10 ml/ decrease in gastric volume for each 1% increase in HbA1c;
kg; fig. 3). Noninferiority was therefore confirmed, as the 95% CI, −0.14 to 0.09). There were no episodes of regur-
upper limit of the 95% CI was below the a priori margin of gitation, vomiting, or aspiration noted upon induction or
0.4 ml/kg. emergence of anesthesia in either group.
We have performed a sensitivity analysis to explore the
possible effect of missing values on the results of the pri-
mary outcome with multiple imputation of the 11 missing Discussion
cases. This model included the patient group, age, height, The current study suggests that baseline fasting gastric volume
weight, body mass index, and use of antacids. We used the is not higher in diabetic compared to nondiabetic patients
mice package in R and generated 10 complete data sets. after following standard preoperative fasting instructions. The
The bootstrap analyses were carried out in each one, and small difference of −0.07 ml/kg with a trend toward slightly
the 95% CI was computed on the entire set of 20,000 boot- higher volumes in control subjects was neither clinically nor
strapped mean differences.The 95% CI from this procedure statistically significant. The 95% CI of this difference of 0.24

652 A nesthesiology 2024; 140:648–56 Perlas et al.

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.
 Gastric Volume and Diabetes

Table 1. Patient Characteristics

Characteristics Nondiabetic (n = 100) Diabetic (n = 91) Standardized Mean Difference P Value

Age, yr, mean ± SD 53.9 ± 14.5 63.4 ± 11.2 0.74 < 0.001
Female sex 41 (41.4%) 37 (40.7%) 0.02 > 0.999
Weight, kg, mean ± SD 82.8 ± 17.6 88.0 ± 18.7 0.28 0.051
Height, m, mean ± SD 1.72 ± 0.10 1.69 ± 0.10 0.26 0.073
Body mass index, mean ± SD 28.0 ± 5.2 30.8 ± 5.5 0.51 0.001
ASA Physical Status, % (nN = 97, nD = 90)
 I 36 (37.1) 0 (0.0) 1.57 < 0.001
 II 39 (40.2) 18 (20.0)

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

 III 22 (22.7) 70 (77.8)
 IV 0 (0.0) 2 (2.2)
Type of surgery, % (nN = 99, nD = 91)
 Hands/plastics 12 (12.1) 2 (2.2) 0.73 0.001
 General surgery 0 (0.0) 8 (8.8)
 Orthopedic surgery 65 (65.7) 50 (54.9)
 Neurosurgery 3 (3.0) 3 (3.3)
 Spine surgery 5 (5.1) 4 (4.4)
 Ophthalmology 7 (7.1) 19 (20.9)
 Urology 7 (7.1) 5 (5.5)
Diabetes-related medical history
 Type II diabetes, % — 77 (84.6) —
 Time since diagnosis, yr
  Mean ± SD — 12.18 ± 11.57 —
  Median (interquartile range) — 10 (3 to 17) —
 Retinopathy, % — 35 (38.5%) —
 Nephropathy, % — 15 (16.5%) —
 Neuropathy, % — 25 (27.5%) —
 Autonomic dysfunction, % — 9 (9.9%) —
 Coronary artery disease, % — 13 (14.3%) —
 Stroke or transient ischemic attack, % — 7 (7.7%) —
 Peripheral vascular disease, % — 5 (5.5%) —
 Diabetic ulcers, % (nD = 90) — 5 (5.6%) —
Gastroparesis symptoms
 Anorexia, % — 11 (12.1%) —
 Early satiety, % — 26 (28.6%) —
 Postprandial fullness, % — 31 (34.1%) —
Other comorbidities
 Gastroesophageal reflux disease, % (nN = 99, nD = 90) 23 (23.2%) 32 (35.6%) 0.27 0.089
 OSA, % (nN = 99) 12 (12.1%) 23 (25.3%) 0.34 0.032
 Lipid dysfunction, % (nN = 99) 1 (1.0%) 11 (12.1%) 0.46 0.005
Medication history
 Diet controlled, % — 25 (27.5%) —
 Oral hypoglycemic agents, % (nD = 90) — 67 (74.4%) —
 Insulin, % — 27 (29.7%) —
 Antacids, % 16 (16.0%) 28 (30.8%) 0.35 0.025
 Motility agent, % 1 (1.0%) 0 (0.0%) 0.14 > 0.999
Laboratory values
 Hemoglobin, mean ± SD 145.1 ± 11.3 137.1 ± 18.2 0.53 0.003
 Fasting blood sugar, mean ± SD 5.8 ± 1.1 7.9 ± 2.1 1.28 < 0.001
 Creatinine, median (interquartile range) 75.0 (67.5 to 84.0) 77.0 (70.0 to 98.0) 0.42 0.139
Estimated glomerular filtration rate, median (interquar- 96.4 (78.5 to 120.3) 93 (67.0 to 108.8) 0.06 0.089
tile range)
 HbA1c, mean ± SD — 7.2 ± 1.5 — —
 HbA1c > 7%, % — 25 (41.0%) — —

The numbers in the nondiabetic and diabetic groups are 100 and 91 unless specified otherwise, where nD is the number in the diabetic group, and nN is the number in the nondiabetic
group.
—, not applicable (no data in nondiabetic group); ASA, American Society of Anesthesiologists; OSA, obstructive sleep apnea.

to 0.10 ml/kg fell well below the noninferiority margin. different between those with diabetes and those without.This
Furthermore, all other markers of gastric “fullness” (overall dis- suggests that the current fasting guidelines by the ASA are just
tribution of antral grades and volume, values of antral area and as effective in preventing a full stomach in diabetic patients as
rates of volume greater than 1.5 ml/kg) were not significantly in nondiabetic subjects before elective surgery.

Perlas et al. A nesthesiology 2024; 140:648–56 653

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.
 Perioperative Medicine

et al.16 studied 100 patients and also found similar fasting

volumes within the normal range for patients with (57 ml)
and without diabetes (51 ml). Likewise, Reed and Haas17
conducted a study on 40 patients with type 2 diabetes and
reported a mean fasting volume of 23.75 ml, which is well
within the normal range for all fasting individuals.
In contrast, a study by Sabry et al.18 estimated higher
gastric volumes in the diabetic cohort compared to non-
diabetic patients (177 vs. 83 ml). However, without weight
Fig. 3. Difference in baseline gastric volume between diabetic corrections, it is difficult to conclude whether these vol-
and nondiabetic patients. Squares represent mean values of the
umes fall within the range that is currently considered safe.
difference, and the lines represent the 95% CI of the difference.

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

Note the 95% CI values all fall below the noninferiority margin Finally, two studies reported a higher incidence of full stom-
of 0.4 ml/kg. ach in diabetic patients.19,20 Zhou et al.19 studied 52 diabetic
patients and reported a higher incidence of full stomach
(48%) compared to nondiabetic patients (8%). However, full
At present, there is no consensus on what constitutes an stomach in this study was defined based only on qualitative
adequate fasting interval to minimize aspiration events in sonographic findings (e.g., Perlas grade 2 antrum), which
diabetic patients. No study specifically reports the incidence is a good screening method but not as accurate as doing
of pulmonary aspiration in patients with diabetes, likely an actual volume assessment. Similarly, a recent study by
because aspiration is a relatively rare event. A recent scop- Rousset et al.20 on 42 diabetic patients observed a higher
ing review of the literature suggested that more prospective incidence of full stomach, which the authors defined as
studies are required to evaluate the effectiveness of fasting either Perlas grade 2 antrum or antral cross-sectional area
guidelines in diabetic patients.3 greater than 3.40 cm2 (which correlates with a gastric vol-
To date, six published studies have evaluated fasting gas- ume of 0.8 ml/kg. However, there is growing evidence to
tric content in diabetic patients as a surrogate marker of suggest that these thresholds for at-risk stomach of 3.4 cm2
aspiration risk.15–20 The findings were mixed, and researchers or 0.8 ml/kg used by Rousset et al.20 and based on ani-
reached different conclusions based on what baseline con- mal models are overly conservative for humans and may
tents and volumes they perceive to constitute an “at-risk” result in many “false-positive” full stomachs, being just at or
stomach. below the mean values of normal baseline gastric volume
Three of the six studies reported negligible fasting vol- in fasting individuals.6,7 In the current study, for example,
umes in both diabetes mellitus patients and controls.15–17 the mean antral cross-sectional area was 7.8 (3.0) cm2, and
Using a standardized scanning protocol, Garg et al.15 assessed the mean volume was 0.9 (0.5) ml/kg in the nondiabetic
53 diabetes mellitus subjects and reported a higher mean control group, which is consistent with previous litera-
gastric volume in patients with diabetes mellitus (9 mL) ture.10,12,21,22 In summary, we suggest that our study offers
compared with non–diabetes mellitus patients (2 mL). more definitive results compared to previous studies as (1)
However, gastric volumes up to 1.5 ml/kg are normal in it has a larger sample size, (2) it uses a noninferiority analysis
healthy fasting patients, and this difference reported by Garg that more conclusively shows the baseline volume is not
et al.,15 while statistically significant, is clinically negligible. higher in diabetic patients, (3) it reports both qualitative
In addition to our findings in the current study, Sharma and quantitative findings of ultrasound, and (4) ultrasound

Table 2. Results

Nondiabetic Patients Diabetic Patients Difference (95% CI) Diabetic

Results (n = 96) (n = 84) versus Nondiabetic P Value

Qualitative findings, n (%)

 Empty – grade 0 30 (31.2) 30 (35.7) 4.5 (−10.5%, 19.4%,) 0.791
 Clear fluid – grade 1 50 (52.1) 42 (50.0) −2.1% (−17.8%, 13.7%)
 Clear fluid – grade 2 16 (16.7) 12 (14.3) −2.4% (−14.1%,9.3%)
 Solid 0 (0) 0 (0)
Quantitative findings
 Right lateral cross-sectional area, cm2, mean ± SD 7.8 ± 3.0 8.3 ± 3.6 0.5 (−0.5, 1.5) 0.369
 Gastric volume, ml/kg, mean ± SD 0.9 ± 0.5 0.8 ± 0.6 −0.1 (−0.2, 0.1) 0.224
 95th percentile of gastric volume, ml/kg, value 1.7 (1.5 to 2.2) 1.9 (1.7 to 2.5) 0.2 (−0.2, 0.5) 0.248
(95% CI)
 Patients with > 1.5 ml/kg, n (%) 11 (11.5%) 13 (15.5%) 4.0% (−7.1%, 15.2%) 0.568

654 A nesthesiology 2024; 140:648–56 Perlas et al.

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.
 Gastric Volume and Diabetes

However, rather than assuming a full or an empty

stomach based on patient history alone, the true value of
bedside gastric ultrasound lies in the possibility of assess-
ing gastric content at the bedside and individualizing care
based on sonographic findings. While this is still an evolv-
ing tool, it is being increasingly incorporated into clinical
practice. The European Society of Anesthesiologists, for
example, was the first society to incorporate gastric ultra-
sound into their fasting guidelines and recommend the
use of gastric ultrasound in children in situations in which
fasting status is not clear or for urgent or emergency sur-

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

gery.23 In conclusion, our data suggest that the baseline
gastric volume in diabetic patients is not higher than that
in nondiabetic patients when standard preoperative fasting
intervals are followed.

Research Support
Fig. 4. Distribution of baseline gastric volumes in diabetic and
nondiabetic patients. A conventional threshold of 1.5 ml/kg is Supported by the Physician Services Incorporated
used as a marker of a full stomach. Foundation and the Canadian Society of Anesthesiologists,
Toronto, Ontario, Canada. Dr. Perlas receives sup-
port for nonclinical time through a merit award from
assessments were performed by blinded observers. At least the Department of Anesthesiology and Pain Medicine,
one of these qualities were missing from the previous stud- University of Toronto, Toronto, Ontario, Canada, and the
ies, which limits their internal validity. Department of Anesthesia and Pain Management, Toronto
There are some limitations to this study. First, Western Hospital, University Health Network, Toronto,
although all subjects followed standard preoperative Ontario, Canada.
fasting instructions, we do not know exactly how many
hours of fasting they presented with. It is possible that Competing Interests
the fasting intervals may have been actually longer that Dr. Perlas is executive editor of the journal Regional
the minimum recommended intervals. However, the Anesthesia and Pain Medicine. She does consulting work for
conditions of the study represent normal clinical cir- FujiFilm SonoSite (Bothell,Washington).The other authors
cumstances. Second, we found a higher-than-expected declare no competing interests.
rate of full stomach (11 to 15% compared to 3 to 5% in
previous studies). However, any sonographic definition
Correspondence
of full stomach based on a threshold value of clear fluid
content is somewhat arbitrary. The selected threshold of Address correspondence to Dr. Perlas: University of Toronto,
1.5 ml/kg, while highly quoted and often used in the 399 Bathurst Street, McLaughlin Pavilion 2-405, Toronto,
literature, is likely still rather conservative and oversen- Ontario M5T 2S8, Canada. [email protected]
sitive to detect full stomach. Third, our study popula-
tion consisted of mostly orthopedic, ophthalmology, and References
plastic surgery populations, with ASA Physical Status I
to III and a body mass index of less than 40 kg/m2. It is 1. Cook TM, MacDougall-Davis SR: Complications
unknown whether our results are generalizable to other and failure of airway management. Br J Anaesth 2012;
surgical populations such as those undergoing abdominal 109:i68–85
surgery, with greater comorbidities or morbid obesity, 2. Bharucha AE, Kudva YC, Prichard DO: Diabetic gast-
which could be at increased risk. roparesis. Endocr Rev 2019; 40:1318–52
The main clinical implication of this study is that at 3. Xiao MZX, Englesakis M, Perlas A: Gastric content
least in the absence of a clear diagnosis of gastroparesis, the and perioperative pulmonary aspiration in patients
majority of diabetic patients present with an empty stom- with diabetes mellitus: A scoping review. Br J Anaesth
ach after following standard fasting instructions. One caveat 2021; 127:224–35
to this statement is that the enrollment period in this study 4. Camilleri M, Bharucha AE, Farrugia G: Epidemiology,
was completed before the introduction of glucagon-like mechanisms, and management of diabetic gastroparesis.
peptide-1 receptor agonists, a common class of drugs now Clin Gastroenterol Hepatol 2011; 9:5–12
indicated for diabetes and weight loss whose mechanism of 5. Practice guidelines for preoperative fasting and the
action includes a delay in gastric emptying. use of pharmacologic agents to reduce the risk of

Perlas et al. A nesthesiology 2024; 140:648–56 655

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.
 Perioperative Medicine

pulmonary aspiration: Application to healthy patients 16. Sharma G, Jacob R, Mahankali S, Ravindra MN:
undergoing elective procedures: An updated report by Preoperative assessment of gastric contents and volume
the American Society of Anesthesiologists Task Force using bedside ultrasound in adult patients: A prospec-
on Preoperative Fasting and the Use of Pharmacologic tive, observational, correlation study. Indian J Anaesth
Agents to Reduce the Risk of Pulmonary Aspiration. 2018; 62:753–8
Anesthesiology 2017; 126:376–93 17. Reed AM, Haas RE: Type 2 diabetes mellitus:
6. Perlas A, Arzola C,Van de Putte P: Point-of-care gastric Relationships between preoperative physiologic stress,
ultrasound and aspiration risk assessment: A narrative gastric content volume and quality, and risk of pulmo-
review. Can J Anaesth 2018; 65:437–48 nary aspiration. AANA J 2020; 88:465–71
7. Van de Putte P, Perlas A: The link between gastric vol- 18. Sabry R, Hasanin A, Refaat S, Abdel Raouf S, Abdallah
ume and aspiration risk: In search of the Holy Grail? AS, Helmy N: Evaluation of gastric residual volume in

Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/140/4/648/701930/20240400.0-00010.pdf by Flavio Vicente on 11 April 2024

Anaesthesia 2018; 73:274–9 fasting diabetic patients using gastric ultrasound. Acta
8. Tokumine J, Sugahara K, Fuchigami T, Teruya K, Nitta Anaesthesiol Scand 2019; 63:615–9
K, Satou K: Unanticipated full stomach at anesthesia 19. Zhou L, Yang Y, Yang L, Cao W, Jing H, Xu Y, Jiang X,
induction in a type I diabetic patient with asymptom- Xu D, Xiao Q, Jiang C, Bo L: Point-of-care ultrasound
atic gastroparesis. J Anesth 2005; 19:247–8 defines gastric content in elective surgical patients with
9. Cubillos J, Tse C, Chan VWS, Perlas A: Bedside ultra- type 2 diabetes mellitus: A prospective cohort study.
sound assessment of gastric content: An observational BMC Anesthesiol 2019; 19:179
study. Can J Anaesth 2012; 59:416–23 20. Rousset J, Coppere Z,Vallee A, Ma S, Clariot S, Burey J,
10. Perlas A, Davis L, Khan M, Mitsakakis N, Chan VWS: Adjavon S, Devys JM, Quesnel C, Fischler M, Bonnet F,
Gastric sonography in the fasted surgical patient: A pro- LE Guen M: Ultrasound assessment of the gastric con-
spective descriptive study. Anesth Analg 2011; 113:93–7 tent among diabetic and non-diabetic patients before
11. Perlas A, Chan VWS, Lupu CM, Mitsakakis N, elective surgery: A prospective multicenter study.
Hanbidge A: Ultrasound assessment of gastric content Minerva Anestesiol 2022; 88:23–31
and volume. Anesthesiology 2009; 111:82–9 21. Van de Putte P, Vernieuwe L, Perlas A: Term pregnant
12. Perlas A, Mitsakakis N, Liu L, Cino M, Haldipur N, patients have similar gastric volume to non-pregnant
Davis L, Cubillos J, Chan V: Validation of a mathemat- females: A single-centre cohort study. Br J Anaesth
ical model for ultrasound assessment of gastric vol- 2019; 122:79–85
ume by gastroscopic examination. Anesth Analg 2013; 22. Valencia JA, Cubillos J, Romero D, Amaya W, Moreno J,
116:357–63 Ferrer L, Pabón S, Perlas A: Chewing gum for 1 h does
13. Kruisselbrink R, Arzola C, Jackson T, Okrainec A, not change gastric volume in healthy fasting subjects:
Chan VWS, Perlas A: Ultrasound assessment of gas- A prospective observational study. J Clin Anesth 2019;
tric volume in severely obese individuals: A validation 56:100–5
study. Br J Anaesth 2017; 118:77–82 23. Frykholm P, Disma N, Andersson H, Beck C, Bouvet
14. Kruisselbrink R, Arzola C, Endersby R, Tse C, Chan V, L, Cercueil E, Elliott E, Hofmann J, Isserman
Perlas A: Intra and inter-rater reliability of ultrasound R, Klaucane A, Kuhn F, de Queiroz Siqueira M,
assessment of gastric volume. Anesthesiology 2014; Rosen D, Rudolph D, Schmidt AR, Schmitz A,
121:46–51 Stocki D, Sümpelmann R, Stricker PA, Thomas
15. Garg H, Podder S, Bala I, Gulati A: Comparison of M, Veyckemans F, Afshari A: Pre-operative fast-
fasting gastric volume using ultrasound in diabetic and ing in children: A guideline from the European
non-diabetic patients in elective surgery: An observa- Society of Anaesthesiology and Intensive Care. Eur J
tional study. Indian J Anaesth 2020; 64:391–6 Anaesthesiol 2022; 39:4–25

656 A nesthesiology 2024; 140:648–56 Perlas et al.

Copyright © 2023 American Society of Anesthesiologists. All Rights Reserved. Unauthorized reproduction of this article is prohibited.