Article Text

Download PDFPDF

Gastric emptying of a liquid nutrient meal in the critically ill: relationship between scintigraphic and carbon breath test measurement
  1. M J Chapman1,
  2. L K Besanko2,
  3. C M Burgstad3,
  4. R J Fraser2,4,
  5. M Bellon5,
  6. S O'Connor1,
  7. A Russo4,
  8. K L Jones4,
  9. K Lange4,
  10. N Q Nguyen3,
  11. F Bartholomeusz5,
  12. B Chatterton5,
  13. M Horowitz4
  1. 1Department of Intensive Care Medicine, Royal Adelaide Hospital, South Australia, Australia
  2. 2Investigation and Procedures Unit, Repatriation General Hospital, Daw Park, South Australia, Australia
  3. 3Department of Gastroenterology and Hepatology, Royal Adelaide Hospital, South Australia, Australia
  4. 4University of Adelaide, Division of Medicine, Royal Adelaide Hospital, South Australia, Australia
  5. 5Department of Nuclear Medicine, Royal Adelaide Hospital, South Australia, Australia
  1. Correspondence to A/Professor Marianne Chapman, Intensive Care Unit, Royal Adelaide Hospital, North Terrace, Adelaide, SA 5000, Australia; marianne.chapman{at}


  • Funding This study was supported by grants from the National Health and Medical Research Council of Australia.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Royal Adelaide Hospital.

  • Provenance and peer review Not commissioned; externally peer reviewed.

View Full Text

Statistics from

Significance of this study

What is already known about this subject?

  • Gastric emptying (GE) may be slow in critical illness.

  • Scintigraphy is considered the most accurate method for measurement of GE.

  • Breath test techniques for the measurement of GE have been validated in some patient groups and healthy subjects but not in critically ill patients.

What are the new findings?

  • Using scintigraphy, GE is demonstrated to be slow in about 50% of intensive care unit patients.

  • The magnitude of the delay in GE is frequently substantial.

  • The breath test appears to be an accurate alternative to scintigraphy for the measurement of GE in critically ill patients.

  • Gastric residual volume (GRV) measurement also correlates with GE.

How might it impact on clinical practice in the foreseeable future?

  • Breath tests can be used for research purposes to assess the prevalence and determinants of delayed GE and response to treatments including promotility agents.

  • A 24 h cumulative GRV of 150 ml is an indicator of slow GE and could be used as a trigger to initiate treatment with promotility agents. It may be unnecessary to slow the rate of nutrient delivery.


Continuous gastric administration of liquid nutrient is believed to be the optimal method to provide nutritional support to critically ill patients.1 However, nutritional goals are frequently not achieved using this approach.2–6 The most common cause for this failure is delayed gastric emptying (GE), evident clinically as large-volume gastric residues.3 Although GE rates have been reported to be reduced in a number of subgroups, the overall prevalence of delayed GE in critical illness remains uncertain.

When considering techniques for the measurement of physiological parameters in critically ill patients, accuracy and practicality are not always common features. Thus, although scintigraphy is believed to provide an accurate measurement of GE, it has rarely been used to measure GE in critically ill patients for practical reasons.7–9 The recent development of breath test methods has allowed non-invasive GE measurements in a variety of patient groups. These could potentially provide a practical method of quantifying GE in the intensive care unit (ICU), but have not been validated in this setting. Paracetamol absorption has been used in this group,10–19 but has also not been formally validated and accuracy may be affected by abnormalities in small intestinal absorption, hepatic metabolism and mesenteric blood flow. Its use is also limited by a clinical requirement to administer paracetamol for pain and temperature control. An alternative bedside technique uses a refractometer to determine changes in the concentrations of feed residue in gastric aspirates (Brix value) after the instillation of known volumes of feed or water.20–23 This technique has also not been validated in critically ill patients, but may be convenient in some centres. It also affords a means of measuring GE that accounts for gastric secretions. Gastric aspirate volumes are a convenient, albeit imprecise, indication of GE in critically ill patients19 24 as gastric aspiration is frequently incomplete due to tube position or blockage and is therefore likely to underestimate the severity of any delay in GE. There is, accordingly, a need for precise GE measurement in critical illness to facilitate an accurate assessment of the prevalence and determinants of delayed GE in this group, as well as to allow identification of patients who require promotility agents and provide a quantitative measure of the response to this therapy.

The aims of this study were to (i) determine the prevalence of delayed GE measured by scintigraphy in a heterogeneous cohort of critically ill patients and (ii) evaluate the relationships between GE assessed by scintigraphy and carbon breath tests.


This study was approved by the Royal Adelaide Hospital (RAH) Research Ethics Committee and carried out according to the National Health and Medical Research Committee guidelines for the conduct of research on unconscious patients.


Critically ill patients

Studies were performed on a convenience sample of 27 mechanically ventilated adult patients, in a tertiary referral, mixed ICU. Prior to each study, informed consent was obtained from the next of kin. Twenty patients had been fed nasogastrically prior to the study at a median rate of 72 ml/h (range 40–120 ml/h) and the remaining seven patients were considered suitable for but had not yet commenced nasogastric (NG) feeding. Studies were performed 9 (range 1–22) days after admission to the ICU. Fourteen patients were receiving morphine infusions at a rate of between 1 and 30 mg/h. Prokinetic drugs were withheld for the period of the study. Subjects were studied in the morning, after a fast of at least 4 h. A NG tube (12-Fr Flexiflo (Ross Laboratories, Columbus, Ohio, USA) or 14-Fr Levin tube (Pharma-Plast, Lynge, Denmark)) was inserted prior to the study. The position of the tube was verified radiologically, and by pH measurement of NG tube aspirates, prior to commencing the study.

Healthy subjects

Data were compared with 14 healthy subjects, all of whom had participated previously in studies involving nasal intubation. Healthy subjects were studied in the morning, after fasting overnight. A NG tube (12-Fr Flexiflo (Ross Laboratories)) was inserted for the purpose of the study and its position was verified by pH measurement of NG tube aspirates and by auscultation of air infusion.


Prior to the study, the gastric contents were aspirated and discarded. A test meal which consisted of 100 ml Ensure (1 kcal/ml—Abbott Laboratories BV, Zwolle, Holland) doped with 20 MBq 99mTechnetium sulphur colloid (radiation dose 0.5 mSv—RAH Radiopharmacy, Adelaide, South Australia, Australia) and 75 KBq octanoic acid, [1-14C] sodium salt (radiation dose 0.155 nSv—MP Biomedicals Australasia, Sydney, New South Wales, Australia) was infused into the stomach via the NG tube over 5 min. Blood glucose concentrations were measured using a bedside glucometer (MediSense Precision, Abbott Laboratories, MediSense Products, Bedford, Massachusetts, USA), using arterial (critically ill patients) or venous (healthy subjects) samples at 30 min intervals.


Scintigraphic measurement of GE

In patients, the scintigraphic measurement of GE was performed in the ICU using a mobile γ camera (General Electric Starcam 300 AM-GE (General Electric Starcam, Milwaukee, Wisconsin, USA) with 3 min dynamic frame acquisition). Healthy subjects were studied in the Department of Nuclear Medicine, PET & Bone Densitometry, RAH, using the same γ camera and technique. Reframed data were corrected for patient movement and radionuclide decay and scatter. All subjects were studied for 4 h supine, in the 20° left anterior oblique position.25 A region of interest that represented the stomach was identified. The isotopic counts within the region of interest were measured and used to derive GE curves (expressed as percentage of the maximum content of the total stomach). The intragastric contents at 60, 120, 180 and 240 min were determined.26 In both healthy subjects and critically ill patients, a scintigraphic half emptying time was also calculated.27

Breath test measurement of GE

End-expiratory breath samples were collected immediately prior to infusion of the test meal, every 10 min for the first hour and every 15 min thereafter for a further 3 h. In patients, breath samples were collected from the expiratory limb of the ventilator tubing. The healthy subjects were asked to expire into sample tubes. A proprietary non-toxic metallic hydroxide, CO2 trapping solution (RAH Nuclear Medicine, Adelaide, South Australia, Australia) was used to collect 0.5 mmol CO2. A colour change in the solution indicated the appropriate amount of CO2. Samples were solubilised with 10 ml Starscint liquid scintillation counting solution (Packard Instruments, Meriden, Connecticut, USA) and counted in a Packard 2100TR Tri-Carb liquid scintillation counter (Packard Instruments) to a 1% coefficient of variation. Resultant counts per minute were corrected to disintegrations per minute using an open window, full spectrum, curve stripping algorithm (Packard Direct DPM) with colour and chemical quench correction programs.

Ventilator settings were not changed for 1 h prior to and during the study. Six patients received volume-cycled ventilation and 19 patients received pressure-cycled ventilation. CO2 production was assumed to be 300 mmol/m2 body surface area, assessed using the height/weight formula of Haycock et al.28 The results were expressed as percentage of 14CO2 recovery/hour and as cumulative values over the sampling period. Mathematical curve fitting using non-linear regression (GraphPad Prism 4, San Diego, California, USA), as previously described by Ghoos et al,29 allowed calculation of parameters of GE, half emptying time (BTt50) and GE coefficient.

Feeding protocol

Before and after the study, the gastric residual volume (GRV) was determined every 6 h as per the usual ICU feeding protocol and the total volumes were documented for the 24 h preceding the study. Patients were fed with Ensure (1 kcal/ml). After the study period, NG feeding and routine aspiration of the NG tube was recommenced in the patients as clinically indicated.

Statistical analysis

Data are presented as median and range unless otherwise specified and a p value of <0.05 was assumed to be significant. Statistical analysis was performed by a professional biostatistician (KL) using SPSS 17 (2009, SPSS Inc). The study was powered to detect a difference between healthy subjects and ICU patients in percentage retention of an effect size of 0.96, compared with an effect size of 1.7 in scintigraphic half emptying time reported by Spapen et al.9 Linear regression was performed to compare breath tests and scintigraphic data. GE data on (n=14) healthy volunteers were used to derive normal ranges of percentage retention as measured by scintigraphy at a range of time points (t=60–180 min). These data were analysed using the non-parametric bootstrap method to calculate reference intervals for the normal range of GE. Based on these results, delayed GE was defined as percentage retention greater than the upper limit of the reference interval at t=180 min of 13%. This time point was selected because it was a time at which there was a marked difference between patients and healthy subjects. This was then used to define normal and delayed GE. Limits of agreement analyses were performed for both healthy subjects and critically ill patients, according to Bland and Altman30 such that the differences between the scintigraphic and breath test half-emptying time (t50) values were plotted against the mean of each method (difference plot) and the limits of agreement were defined as acceptable if they were within the mean (±2SD) difference.30 The Mann–Whitney test was used to compare Acute Physiology and Chronic Health Evaluation (APACHE) II scores and age between the normal and delayed groups, and Fisher's exact test was used to compare all other demographic factors. The ability of the breath test and GRVs to diagnose delayed GE was assessed using receiver operating characteristic (ROC) curves and summarised by sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios.


The demographic variables of the study subjects are shown in table 1.

Table 1

Demographic information of study subjects

Complete data were obtained in the healthy cohort. Two patients did not complete the study because of regurgitation or vomiting. Technical difficulties resulted in loss or incomplete scintigraphic data in two patients. In an additional patient, breath test data could not be analysed because emptying was so slow that the curves could not be subjected to the usual mathematical modelling. GE data were available for analysis in 24 patients. Three patients had grossly delayed GE measured scintigraphically (retention at 240 min of 97%, 94% and 82%, respectively) precluding calculation of t50.

GE in critical illness

Comparisons of GE measurement between scintigraphic and breath testing techniques in both groups are shown in table 2. GE was slower in ICU patients using all parameters, except retention at 60 min and BTt50.

Table 2

Gastric emptying in patients and healthy subjects

The prevalence of delayed GE in ICU patients is shown in table 3 and figure 1.

Table 3

Prevalence of delayed gastric emptying using various parameters

Figure 1

Scintigraphic measurement of gastric emptying (GE): individual results for gastric meal retention over time in healthy subjects and ICU patients. The normal range (as indicated in the first graph) is shown in the shaded area in the second graph. At 180 min, 13 patients (out of 24) were outside the normal range. Eleven patients (46%) still had meal retention in the stomach at 240 min.

Cumulative GRVs were higher in the 24 h prior to the study in patients with delayed GE (20 (0–180) vs 240 (35–1050) ml; p=0.001).

Patients with delayed GE had greater severity of illness on admission to the ICU (APACHE II score on admission: 15 (8–34) vs 21 (13–35); p=0.03), and were more likely to have been admitted with trauma, sepsis and respiratory failure and less likely to have been admitted with intracranial haemorrhage (p=0.008). There were no differences in the following variables between patients with or without slow GE: gender, APACHE II score on day of study, past history of diabetes mellitus, mean blood glucose over the study period, hospital survival or administration of morphine, midazolam, inotropes or propofol.

Limits of agreement between scintigraphic and breath test measurement

The limits of agreement for the t50 values in healthy subjects (n=14) and critically ill patients (n=21), respectively, were: −95.0 and −14.4 min with a mean difference of −54.7 min versus −150.6 min and +183.2 min with a mean difference of +16.3 min. For both healthy subjects and critically ill patients, the range in the t50 difference fell within ±2SD, indicating acceptable limits of agreement.30

Relationship between scintigraphic and breath test measurement

There was a strong correlation between breath test measurements and intragastric retention in both ICU patients and healthy subjects (table 4).

Table 4

Relationship between scintigraphic and breath test measurements of gastric emptying

The ROC curves for breath test measurements of GE are shown in figure 2. The sensitivity and specificity and the positive and negative predictive values of the breath test compared with scintigraphy are shown in table 5. The breath test t50 of 99.5 min and GE coefficient of 3.03 were the values giving the best sensitivity and specificity for identification of normal and delayed GE when compared with scintigraphic intragastric retention at 180 min.

Figure 2

Receiver operating characteristic curves for breath test measures of gastric emptying in ICU patients. GEC, gastric emptying coefficient (area under the curve=0.833 (SE=0.090), p=0.011); t50, half emptying time (area under the curve=0.947 (SE=0.043), p<0.001).

Table 5

Diagnostic accuracy of breath test (A) t50≥99.5 and (B) GEC≤3.03

Relationship between scintigraphic measurement and GRV

There was a good correlation between GRV and scintigraphic retention (table 6).

Table 6

Relationship between GRV with gastric meal retention and breath test

The ROC curves for GRV measurements of GE are shown in figure 3. The positive and negative predictive values of the GRV measurement compared with scintigraphy are shown in table 7. GRVs of 150 and 250 ml were selected because these volumes are used commonly for identification of failed NG feeding in ICU feeding protocols. The value of 250 ml is used by the centre at which the study was performed.

Figure 3

Receiver operating characteristic curves for cumulative gastric residual volume in the 24 h prior to study in ICU patients. Area under the curve=0.909 (SE=0.062), p=0.001.

Table 7

Accuracy of cumulative GRV for the 24 h prior to the GE study of (A) 150 ml and (B) 250 ml as a prediction of delayed GE as defined by gastric retention at 180 min ≥13%

Effect of delayed GE on nutrient delivery

In patients with slow GE there was a trend for a reduced volume of nutrient delivery when the balance of volume delivered and gastric aspirate volumes discarded was calculated (normal GE 1920 (−150–2400) vs slow GE 680 (−220–2880) ml; p=0.072). Energy delivery was reduced in the patients with slow GE (normal GE 1920 (0–2400) vs slow GE 510 (0–2880) kcal; p=0.047).


The importance of this study lies in the determination of the prevalence of slow GE, which is likely to affect the success of nutrient delivery, and the accurate measurement of GE, in critically ill patients. Scintigraphy is considered the most accurate technique for measurement of GE. This study is the first to examine the prevalence of delayed GE in a heterogeneous critically ill population using scintigraphy, and establishes that about half of these patients have slow GE when compared with healthy subjects and that in some of these the magnitude of the delay is marked. Furthermore, the 14C-breath test gives an accurate measure of GE when compared with scintigraphy, suggesting that this technique can be used to measure GE in the critically ill population. We have also shown that there is a close relationship between 24 h cumulative GRVs and GE, although breath tests appear to have greater accuracy.

This study shows that GE was significantly slower in patients than in healthy controls, which is consistent with previous, limited, scintigraphic measurements of GE in critically ill patients. In the healthy group, all subjects had complete emptying at 4 h, while there was a median of 14% retention at 4 h in the critically ill group. Spapen et al reported that GE was significantly delayed in a small mixed ICU group compared with healthy subjects.9 Although the scintigraphic t50 was prolonged, gastric retention and the prevalence of delayed GE were not observed. In our study, scintigraphic t50 could not be calculated in three patients because GE was extremely slow (intragastric retention >82%). GE measured by scintigraphy has been reported to be delayed in 50–80% of patients with severe head injury.7 8 However, in all these studies, there was a difference in the method of meal ingestion in patients and controls which may have confounded the results. The healthy subjects swallowed the test meal, while in the patient group the test meal was infused via a NG tube. In the current study, the test meal was infused through a NG tube in both groups. This allows a more valid comparison, as the act of swallowing may accelerate GE.32 However, when measuring GE using a NG tube in a healthy cohort, it is important that the subjects are accustomed to the positioning of a NG tube prior to the study. The rate of GE is exquisitely sensitive to stress or discomfort,33 34 which could result in false slow GE measurements in the control group. Accordingly, all healthy subjects in this study had undergone previous NG tube placement and appeared to be unaffected by this process. The use of scintigraphy for GE measurement, and the careful duplication of technique in both groups in this study, provides the most comprehensive assessment of delayed GE in critically ill patients to date. However, although this is the largest study performed in a heterogeneous group of critically ill patients, it is still limited in size, reflecting the difficulties in performing these studies in this patient group. The prevalence of delayed GE in certain diagnostic subgroups cannot, therefore, be made with any certainty.

Previous studies have examined GE in critically ill patients using techniques other than scintigraphy or breath tests. However, no method has been specifically validated in critically ill patients and the unique environment of the ICU can limit the applicability of some techniques. Four studies have used paracetamol absorption to compare GE in critically ill patients with healthy subjects.19 35–37 The study by Tarling et al reported an incidence of slow GE of 60% in a mixed group of ICU patients,19 which is similar to the 54% reported in this study. Goldhill et al reported that patients on the day after cardiac surgery had markedly delayed GE compared with before surgery (area under curve at 60 min (AUC60) 892 (SEM 57) vs 131 mg min/l (SEM 25)).35 By contrast, normal GE was reported by Heyland et al in a mixed group of ICU patients compared with controls (AUC120 9301 (7343) vs 11 644 (1336) μmol/min/l (p=0.28))36, and by Hu et al in a group of patients with burns.37 All studies demonstrated variability in GE in the patients, with some patients exhibiting marked slowing. Paracetamol is subject to a variable ‘first pass’ effect due to unpredictable hepatic function in critical illness and is, therefore, an unproven and possibly invalid method for the measurement of GE in critical illness.

Factors that have previously been associated with slow GE in critically ill patients include raised intracranial pressure,14 16 reduced Glasgow coma score,7 age,36 38 administration of dopamine,19 time from traumatic brain injury7 and the height of a spinal cord lesion.38 There are conflicting results on the effects of the administration of opiates14 36 39 and gender.7 36 In this study, slow GE was related to illness severity at admission, but not on the day of study. This suggests that neurohumoral disturbance at the time when the patient is at their sickest is responsible for disturbances in gut function which do not then resolve as quickly as other parameters. GE was also more delayed in certain diagnostic groups, principally trauma and respiratory failure. Trauma is a known risk factor for difficulties with enteral nutrient administration.40 Other factors that have been reported to affect GE were not found to do so in this study, including gender, age and opiate and catecholamine administration. However, this study may well be underpowered to show a difference in these subgroups. For example, all three patients receiving inotropes had slow GE, but as the group was small this finding was not significant.

Breath test measurement of GE, first described by Maes and Ghoos in 1993,29 has been validated against scintigraphy in healthy subjects for both solid and non-nutrient liquid emptying29 41–47 and also in patients with diabetes, dyspepsia, respiratory disease and gastroparesis.42 43 47–51 However, the emptying of liquid nutrient is more relevant in a critically ill population and this is only the second study validating the use of breath tests for the measurement of liquid nutrient emptying, and the first in an adult population. In the previous study, breath tests were validated to measure GE of milk in a paediatric group.52 Although breath tests have been used to measure GE in critically ill patients,53 the rate of excretion of markers into expiratory gases could possibly be affected by factors other than GE, including small intestinal absorption and hepatic metabolism. In the current study, we demonstrated a strong relationship between breath test and scintigraphic measurement of GE, in both patients (r2=0.44–0.71) and healthy subjects (r2=0.37–0.64), with acceptable limits of agreement between the two techniques. This is consistent with studies reported by other groups.45 52 54 These data, therefore, support a potential role for breath testing in critically ill patients for epidemiological, physiological and drug studies on GE. We have also recently demonstrated that the 13C octanoate breath test is reproducible in this patient group. The 13C and 14C breath tests have been shown to be interchangeable for the measurement of solid emptying.45 55

Regular measurement of GRV during the infusion of enteral nutrition is considered a convenient clinical tool to indicate GE, success of feeding and potential risk of aspiration. Despite acceptance of GRV in feeding protocols by the majority of ICUs, the utility and significance of this measurement is controversial, as it is dependent on a number of factors. These include the position of the tube, tube characteristics (such as tube type and number of openings), the volume of the syringe used56 and the operator performing the test.57 There are conflicting data on the relationship between GRV and GE, which has lead to a lack of consensus on an acceptable value for GRV during enteral feeding.8 19 53 58 59 Currently, the majority of ICUs have protocols for feeding that consider a change in delivery rate or site if the GRV is between 150 and 400 ml/s, a large range that reflects the uncertainty of the technique. In this study, we have demonstrated a close relationship between GRV and scintigraphic measurement of GE; however, the accuracy of GRVs to detect slow GE will depend on the GRV chosen. Our data show that a cumulative GRV over 24 h of >250 ml will detect all subjects with delayed GE (as defined in this study by scintigraphic retention at 180 min of ≥13%). However, sensitivity is compromised. A cumulative GRV of 150 ml or greater over a 24 h period will more reliably predict slow GE and, in such patients, feeding may be improved by the administration of a combination of metoclopramide and erythromycin.60

There are several technical issues that should be considered when interpreting these data. Although scintigraphy is currently considered to be the most accurate method for measurement of GE, it is difficult to perform in critically ill patients and three-dimensional pictures cannot be generated due to the necessity for the patient to remain supine. To overcome this, in this study the patients and healthy subjects were studied in the left anterior oblique position.25 61 62 In addition, identification of the stomach can be affected by radiation from the overlying bowel that may hamper interpretation. Furthermore, scintigraphic and breath test techniques do not measure total GE, as neither takes into account gastric secretion, the importance of which is unknown in critically ill patients. Also, these two techniques measure slightly different parameters. The breath test technique requires GE, duodenal absorption, liver metabolism and respiratory excretion to occur. GE is thought to be the rate-limiting step in this process, but it is possible that in certain critically ill patients marked derangements in the other steps may occur, although the strong relationship demonstrated in this study between breath tests and scintigraphy suggests otherwise.

There are additional limitations of this study. The normal ranges were calculated using a sample of only 14 healthy subjects and it is possible that some patients were incorrectly classified as delayed/normal using the definition of delayed GE proposed in this study. Normal ranges calculated for a larger sample would be required to establish robust normal ranges of GE using this technique. In two patients recruited into this study, data could not be collected because of regurgitation or vomiting. It is highly likely that these two patients had delayed GE and the prevalence of abnormally slow GE may, accordingly, be higher than that suggested.


The current study suggests that (i) GE is delayed in about half of mixed critically ill patients and markedly delayed in about 20%; (ii) breath tests can be used as a convenient method for the measurement of GE in critically ill patients for research purposes to quantify the effects of promotility agents; (iii) a cumulative GRV as low as 150 ml measured over 24 h is indicative of slow GE and administration of promotility agents without a reduction in feed administration should be considered.


View Abstract


  • Funding This study was supported by grants from the National Health and Medical Research Council of Australia.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Royal Adelaide Hospital.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.