Secondary Logo

Journal Logo

Perioperative medicine

Black or white coffee before anaesthesia?

A randomised crossover trial

Larsen, Birgit; Larsen, Lars P.; Sivesgaard, Kim; Juul, Svend

Author Information
European Journal of Anaesthesiology: June 2016 - Volume 33 - Issue 6 - p 457-462
doi: 10.1097/EJA.0000000000000457



The aim of preoperative fasting is to minimise gastric content and reduce the risk of aspiration, which is a rare but serious complication of anaesthesia.1,2 Procedures requiring anaesthesia are in many cases cancelled or delayed if a patient has recently ingested coffee with added milk, and the current recommendations from the American Society of Anesthesiologists3 and European Society of Anaesthesiology (ESA)4 still categorise milk as solid food. A majority of the ESA guideline group considers adding up to 20% milk in the coffee to be acceptable.

The critical gastric volume constituting a risk of aspiration in humans is unknown. For example, Czarnetzki et al.5, while recognising the absence of a generally accepted definition, decided to consider a clear stomach as a residual volume of less than 40 ml of liquids and no solids. In the present study, we hypothesised that up to 50% of added milk to a normal cup of coffee has no or minor influence on the gastric volume 2 h later. We designed it as a crossover study and analysed it as a noninferiority study, testing the hypothesis that adding milk to coffee leads to no gastric volume increase or an increase of less than 12 ml, 2 h later.



Prior to the enrolment of participants, the trial was approved on 21 March 2013 by the Central Denmark Region Committee on Health Research Ethics, Committee II (Chairperson MD, PhD, Elisabeth Stenbøg), Skottenborg 26, 8800 Viborg (Approval no. 1-10-72-67-13), and on 12 March 2013, it was reported to the Danish Data Protection Agency, Borgergade 28, DK-1300 Copenhagen K (Journal no. 1-16-02-319-12). The trial was registered at the Clinical Trials Registry http:\\ (NCT 02361632).

Potential participants were recruited as a convenience sample by advertising, and a medical questionnaire and a clinical assessment were used to determine general physical and mental eligibility. In particular, study participants with present or former gastric disease or other conditions with a risk of gastric paresis, a BMI of more than 30 kg m−2, with metal implants, currently pregnant or having claustrophobia were excluded from the study. After submitting written informed consent, the participants were randomised to one of the three groups. Consecutively after inclusion, the allocations were performed by the main investigator at the Department of Day Surgery where the participants drew a random code number, which corresponded to a group number in a randomisation protocol. The participants received monetary compensation covering the expenses of coffee, milk and transportation.

Study design

The study was designed as a randomised, crossover trial where all participants were exposed to three coffee and milk mixtures. The 32 enrolled healthy volunteers were randomised to one of three parallel trial arms, each with a different sequence of the three milk exposure levels (Fig. 1).

Fig. 1:
Flow diagram.


The participants fasted 6 h for solid food, and 2 h before examination, each participant ingested one of three coffee mixtures: 175 ml coffee, including either 0 or 20 or 50% full fat milk. Mixture 1: 175 ml black coffee, Mixture 2: coffee with added 20% (37 ml) full fat milk and Mixture 3: coffee with added 50% (87 ml) full fat milk to a total of 175 ml (one standard cup). We used full fat milk containing 3.5% protein, 3.5% fat and 64 kcal 100 ml−1. At the time of enrolment, participants received three jugs, one for each examination day, with measures for the milk and coffee mixture according to the sequence of the mixtures predefined by the trial arm. The preparation of the coffee drink, including the temperature of coffee and milk before ingestion, was not specified in the instructions. The study was conducted between August 2013 and February 2014.


We used MRI to measure gastric volumes. MRI is a noninvasive technique with no exposure of the participants to ionising radiation and no effect on the gastric volume as would be the case for measurements based on an intragastric balloon-catheter. The technique has been validated both in vitro and in vivo and has been shown to reliably measure gastric volume.6 Furthermore, the technique shows good reproducibility for even small volumes with a coefficient of variation of 6 to 10% in the range from 1.6 to 86.5 ml.7 All examinations were performed at 7:30 a.m. at the Department of Radiology. The person performing the MRI and the radiologists who analysed the images were blinded to the coffee mixture ingested before the examination. The image analyses were performed independently by a consultant radiologist (L.P.L.) and a radiology resident (K.S.). To avoid an impact of the sequence of milk exposure levels, the interval between a participant's examinations was at least 2 days (see Supplemental Digital Content 1,, for a detailed description of the method of measurement).

Statistical analysis

The results were analysed as a noninferiority study to assess whether we could rule out an effect of milk larger than a predefined noninferiority margin. A gastric volume increase of less than 12 ml when exposed to milk was considered clinically unimportant (the noninferiority margin). Assuming no effect of added milk and assuming a SD of the difference of 20 ml, we needed to include 30 participants in the study to obtain a statistical power of 90% to exclude an effect larger than the noninferiority margin.

We estimated the association between milk exposure and gastric volume in a mixed-effects linear regression model, adjusting for period (examination number) as fixed effect and for participant as random effect to allow for multiple examinations per participant. We estimated the gastric volume difference between examinations with and without milk exposure. If the upper 95% confidence limit of the gastric volume difference was less than the chosen noninferiority margin of 12 ml, we would consider the result noninferior. The agreement between the two examiners’ measurements of the gastric volumes were analysed with paired t-test and Pearson correlation. Stata version 13.1 (StataCorp, College Station, Texas, USA) was used for the analyses.


A total of 32 healthy participants were included in the study. Table 1 shows characteristics of the participants.

Table 1:
Characteristics of participants

The agreement between the two examiners’ MRI measurements of the gastric volumes was high with a Pearson correlation of 0.94 and a mean difference of 1.0 ml [95% confidence interval (CI), −0.5, 2.5, P = 0.18], and the average of the measurements was used in the analysis.

The mean gastric volume for black coffee was 27.8 ml, for coffee with 20% milk 17.9 ml and for coffee with 50% milk 20.6 ml. Figure 2 illustrates the gastric volume for each participant in relation to the milk exposure. For the majority of participants, the gastric volume was low for all milk exposure levels. Two participants had a much higher gastric volume after exposure to coffee without milk than with milk. They were included in the primary analysis, but excluded from a supplementary analysis. One participant with a high gastric volume for all milk exposure levels was included in the analysis. Two participants who did not complete all three examinations were included in the analysis with the information available.

Fig. 2:
Individual gastric volume measurements among all participants.

Figure 3 shows the gastric volume distribution after each of the three different amounts of milk in the coffee. Table 2 shows the gastric volume change after milk exposure compared with exposure to coffee without milk. Compared with black coffee, the gastric volume for 20% milk was significantly decreased with a difference of −10.0 ml (95% CI, −18.2, −1.8), and for 50% milk it was insignificantly decreased, −7.2 ml (95% CI, −17.4, +2.9). The upper CI for the difference in gastric volume between the ‘no milk added’ group and each ‘milk added’ group did not reach the noninferiority limit of 12 ml. In a supplementary analysis, excluding two participants with very high gastric volume after coffee without milk, there was little and insignificant difference in gastric volume between the exposure levels. There was no evidence of an effect of the sequence of interventions (period effect) (P = 0.80).

Fig. 3:
Gastric volume distribution at three milk exposure levels.
Table 2:
Difference in gastric volume (ml) in relation to milk exposure level


The study compares the gastric volume 2 h after ingestion of coffee and milk with the volume after ingesting black coffee. The findings confirmed the hypothesis that up to 50% full fat milk (87 ml) in a standard cup of coffee caused no clinically important increase of the gastric volume compared with black coffee.

The main strengths of the study are the crossover design, which reduced noise from interindividual variations in gastric volume, the thorough MRI measurements, and the statistical noninferiority analysis, which addresses the relevant question of whether we can exclude any major effect of the exposure.

The randomisation should have led to a balanced design with approximately equally sized groups and varying sequence of milk exposure levels, but because of a communication error, this was not done as intended. Thus, the power of the study to detect and adjust for a period effect was less than optimal. We have, however, no prior reason to suspect a period effect, and with the data at hand, we found no evidence of a period effect. We therefore conclude that this flaw does not substantially affect the main result of the study.

Two participants had a surprising outcome pattern with very high gastric volumes when not exposed to milk, and they had quite a strong effect on the result in the primary analysis. We excluded them in a supplementary analysis, and the main conclusion remained unchanged.

A well tolerated volume for gastric fluids to prevent aspiration has never been empirically established, and our assumption that an increase of 12 ml of the residual gastric volume to be of no clinical importance was not based on experimental evidence, but on residual volumes reported from clinical studies. Experimental trials of critical volumes cannot be performed on humans for ethical reasons. A study in monkeys from 1990, where Raidoo and colleagues8 instilled gastric aspirate directly in the trachea of monkeys and produced various grades of pneumonitis, concluded that a volume of 0.4 ml kg−1 would serve as a critical volume. Raidoo and others also suggested 0.8 ml kg−1 as a critical volume when the results from the monkey experiment were extrapolated to humans, but the validity of such an extrapolation can be questioned.9

We assessed an increase of 12 ml as clinically acceptable. Residual gastric volumes in healthy persons have been measured over time after different fasting periods and by various methods. Median gastric volumes from these studies were reported to be about 10 to 35 ml, and with a range between 0 and 100 ml.2,10–13

Most studies found outliers with higher residual volume, and in a clinical setting we do not know the actual gastric volume of a patient presenting for anaesthesia. Aspiration of gastric contents is a rare complication, and physical characteristics of the patient, the nature of the procedure and skills of the anaesthetist may contribute to the risk of pulmonary aspiration.2,9

The current recommendations from the American Society of Anesthesiologists are for a fasting period of 6 h for all nonhuman milk products, whereas all but one in the ESA guideline group accepts up to 20% milk in coffee or tea. Various interpretations of these guidelines have resulted in different practices concerning delays and cancellations, and the frequency of these negative consequences is unknown.

The difficulties in estimating gastric volume are well known. The most common methods for assessment are aspiration with gastric tubes, ultrasonography and MRI. These different methods are not directly comparable. Gastric aspiration provides an unreliable measure that usually underestimates the volume, and the method is not convenient when the study participants are awake. Ultrasound is a well tested method and well suited to bedside estimation of gastric fluid volume, but problems of diagnostic accuracy and reproducibility have been described.14–16

In this study, we used MRI to measure gastric volume. MRI is a noninvasive procedure causing minor discomfort to the person examined. MRI allows an accurate direct measure of the gastric volume in real time, and it is well suited to provide precise measures of very small quantities of gastric fluid.7,17 The method has shown good correlation with scintigraphy, which serves as the gold standard for estimating gastric volume.18 In addition, MRI has a good reproducibility when measuring gastric volume.19

In a randomised crossover study, Hillyard and colleagues20 compared gastric emptying time in 10 healthy participants after intake of 300 ml tea without and with a modest amount of milk. They found no evidence that adding 50 ml (17%) of milk to the tea increased the gastric emptying time and suggested that the current preoperative guidelines could safely be altered. In a crossover study, including eight healthy volunteers, Okabe and others21 compared the emptying rate after intake of 500 ml of various fluids, including undiluted and diluted milk. They found that the emptying rate was slower for diluted and especially undiluted milk than for water, but similar to other fluids with the same caloric content. They concluded that the gastric emptying of liquids depends on the total caloric content rather than the composition.

The volunteers received both oral and written information and were reminded by e-mail or text message to comply with instructions. We did not observe the participants, and there was no registration of the exact time for the ingestion of the three different amounts of milk in the coffee. Thus, there was a risk of noncompliance, but this is not different from everyday clinical practice, which we attempt to reflect as closely as possible. We did not instruct the participants to ingest their coffee drink at a specified temperature and preparation method, but the crossover design reduces the risk of bias from this. A recent study found that the temperature has only limited effect on gastric emptying after ingestion of 60 ml juice at 50, 37 or 4°C.22 Gastric emptying of a liquid meal or small bowel transit is not affected by coffee itself.23 The number of participants (n = 32) and the variety in the participants’ age (18 to 71 years) contribute to the generalisability of the results, but the study included healthy persons only, and the findings cannot be applied to other groups of patients without further studies.

The results of the present study are in agreement with results from previous studies on gastric volumes after black coffee ingested according to the current recommended fasting periods. It complements the findings from Hillyard and colleagues20 that a moderate amount of milk added to tea would not delay gastric emptying time.


The study provides new knowledge about the impact on gastric volume of adding milk to coffee, and it contributes to the current discussion of more liberal fasting recommendations for milk in modest amounts before anaesthesia. The results are consistent with the hypothesis that a cup of coffee with up to 50% full fat milk has no effect or only minor effect on the gastric volume 2 h later, and they are not consistent with the hypothesis that it increases the gastric volume by 12 sml or more. The results support a liberalisation of policy on the addition of milk to hot drinks before planned anaesthesia.

Acknowledgements related to this article

Assistance with the study: we would like to thank Associate Professor M. Frydenberg, Department of Public Health, Section for Biostatistics, Aarhus University, Denmark, for statistical advice, and Language Coach MG Hansen, Aarhus University Hospital, Denmark, for language editing support.

Financial support and sponsorship: this study was supported by the Department of Anaesthesiology and Intensive Care, Aarhus University Hospital, Aarhus, Denmark.

Conflicts of interest: none.

Presentation: preliminary data from this study were presented at the Danish Society of Day Surgery (DSDK) annual meeting, 25 April 2015, Fredericia, Denmark.


1. Beck-Schimmer BB, Bonvini JM. Bronchoaspiration: incidence, consequences and management. Eur J Anaesthesiol 2011; 28:78–84.
2. Søreide E, Eriksson LI, Hirlekar G, et al. Preoperative fasting guidelines: an update. Acta Anaesthesiol Scand 2005; 49:1041–1047.
3. American Society of Anesthesiologists Committee. Practice guidelines for preoperative fasting and the use of pharmacologic agents to reduce the risk of pulmonary aspiration: application to healthy patients undergoing elective procedures: an updated report by the American Society of Anesthesiologists Committee on Standards and Practice Parameters. Anesthesiology 2011; 114:495–511.
4. Smith I, Kranke P, Murat I, et al. European Society of Anaesthesiology. Perioperative fasting in adults and children: guidelines from the European Society of Anaesthesiology. Eur J Anaesthesiol 2011; 28:556–569.
5. Czarnetzki C, Elia N, Frossard JL, et al. Erythromycin for gastric emptying in patients undergoing general anesthesia for emergency surgery: a randomized clinical trial. JAMA Surg 2015; 150:730–737.
6. Haans JJ, de Zwart IM, Eilers PH, et al. Gastric volume changes in response to a meal: validation of magnetic resonance imaging versus the barostat. J Magn Reson Imaging 2011; 34:685–690.
7. Schmitz A, Kellenberger CJ, Neuhaus D, et al. Fasting times and gastric contents volume in children undergoing deep propofol sedation--an assessment using magnetic resonance imaging. Paediatr Anaesth 2011; 21:685–690.
8. Raidoo DM, Rocke DA, Brock-Utne JG, et al. Critical volume for pulmonary acid aspiration: reappraisal in a primate model. Br J Anaesth 1990; 65:248–250.
9. Schreiner MS. Gastric fluid volume: is it really a risk factor for pulmonary aspiration? Anesth Analg 1998; 87:754–756.
10. Soreide E, Holst-Larsen H, Veel T, Steen PA. The effects of chewing gum on gastric content prior to induction of general anesthesia. Anesth Analg 1995; 80:985–989.
11. Brady Marian C, Kinn S, Stuart P, Ness V. Preoperative fasting for adults to prevent perioperative complications. Cochrane Database Syst Rev 2003; 4:CD004423.
12. Lobo DN, Hendry PO, Rodrigues G, et al. Gastric emptying of three liquid oral preoperative metabolic preconditioning regimens measured by magnetic resonance imaging in healthy adult volunteers: A randomised double-blind, crossover study. Clin Nutr 2009; 28:636–641.
13. Hausel J, Nygren J, Lagerkranser M, et al. A carbohydrate-rich drink reduces preoperative discomfort in elective surgery patients. Anesth Analg 2001; 93:1344–1350.
14. Soreide E, Hausken T, Soreide JA, Steen PA. Gastric emptying of a light hospital breakfast: a study using real time ultrasonography. Acta Anaesthesiol Scand 1996; 40:549–553.
15. Van de Putte P, Perlas A. Ultrasound assessment of gastric content and volume. Br J Anaesth 2014; 113:12–22.
16. Perlas A, Davis L, Khan M, et al. Gastric sonography in the fasted surgical patient: a prospective descriptive study. Anesth Analg 2011; 113:93–97.
17. Schmitz A, Kellenberger CJ, Lochbuehler N, et al. Effect of different quantities of a sugared clear fluid on gastric emptying and residual volume in children: a crossover study using magnetic resonance imaging. Br J Anaesth 2012; 108:644–647.
18. Feinle C, Kunz P, Boesiger P, et al. Scintigraphic validation of a magnetic resonance imaging method to study gastric emptying of a solid meal in humans. Gut 1999; 44:106–111.
19. Fruehauf H, Menne D, Kwiatek MA, et al. Inter-observer reproducibility and analysis of gastric volume measurements and gastric emptying assessed with magnetic resonance imaging. Neurogastroenterol Motil 2011; 23:854–861.
20. Hillyard S, Cowman S, Ramasundaram R, et al. Does adding milk to tea delay gastric emptying? Br J Anaesth 2014; 112:66–71.
21. Okabe T, Terashima H, Sakamoto A. Determinants of liquid gastric emptying: comparisons between milk and isocalorically adjusted clear fluids. Br J Anaesth 2015; 114:77–82.
22. Sun WM, Houghton LA, Read NW, et al. Effect of meal temperature on gastric emptying of liquids in man. Gut 1988; 29:302–305.
23. Boekema PJ, Lo B, Samsom M, et al. The effect of coffee on gastric emptying and oro-caecal transit time. Eur J Clin Invest 2000; 30:129–134.

Supplemental Digital Content

© 2016 European Society of Anaesthesiology