OUP user menu

The influence of liberal alcohol consumption on glucose metabolism in patients with type 1 diabetes: a pilot study

D. Kerr, S. Penfold, S. Zouwail, P. Thomas, J. Begley
DOI: http://dx.doi.org/10.1093/qjmed/hcn163 169-174 First published online: 19 December 2008


Background: Little is known about the consequences of excessive alcohol ingestion in patients with type 1 diabetes.

Aim: To examine the metabolic effects of acute ingestion of liberal amounts of alcohol in patients with type 1 diabetes.

Design: A pilot study using a randomized, placebo controlled, double blind design in Hospital Clinical Research Unit.

Methods: The study included 10 patients with type 1 diabetes (seven male, age 43.9 ± 9.0 years, duration of diabetes 17.3 ± 13.8 years, HbA1c 8.0 ± 1.5%) who had a standard 600-calorie lunch on two separate occasions, together with either white wine (men eight units, women six units), or an equivalent volume of alcohol-free wine. Bloods were collected before lunch and hourly for 4 h for glucose, intermediary metabolites, counter-regulatory hormones and inflammatory markers.

Results: There were no significant differences between alcohol and alcohol-free days in levels of glucose, triglycerides, free fatty acids, glycerol, cortisol and growth hormone. In contrast, lactate levels rose in response to the meal but with alcohol the overall response was augmented (P = 0.014). β-Hydroxybutyrate levels were suppressed post prandially on the alcohol-free day but were significantly elevated with alcohol (P < 0.001).

Conclusions: A rise in ketones following alcohol ingestion occurred despite subjects being in a strictly controlled environment with no interruption in insulin administration. Such individuals might be at risk of significant ketosis in less-controlled circumstances where insulin administration might be more erratic. Patient education material should contain information to highlight these potential problems.


Although the health hazards of excessive alcohol consumption are well documented, there is epidemiological evidence that moderate alcohol consumption has demonstrable health benefits.1–3 For example, within the diabetic population, several studies have suggested a lowering of risk of premature vascular disease associated with modest alcohol consumption.4–7 Thus there are clinical reasons to encourage regular use of alcohol in patients with diabetes. Suggested limits for ‘safe’ levels of alcohol use are generally accepted to be 1–2 U of alcohol per day for women, 2–3 U/day for men (1 unit = 8–10 g of ethanol).8

Recent studies9,10 have also suggested that the pattern of alcohol intake may be important in that regular consumption of small quantities is safer than ‘binge drinking’. There is no internationally agreed definition of binge drinking, but in the UK, drinking surveys normally define binge drinkers as men consuming at least eight, and women six standard units of alcohol per week.11

One concern from the combination of type 1 diabetes and excessive alcohol is the risk of alcoholic ketoacidosis (AKA)12 seen in those individuals who drink excessively, while eating very little. AKA is often unsuspected as patients most often present some time after a drinking session so may not appear intoxicated nor have elevated blood alcohol levels. In the more common diabetic ketoacidosis, insulin deficiency results in excess mobilization of fatty acids from adipose tissue with the production of aceto-acetic acid and β-hydroxybutyric acid, resulting in a high anion gap acidosis.13 In alcoholic ketoacidosis, glucose levels may be only marginally elevated or even low and development of ketoacidosis results from a complex interaction of alcohol metabolism, decreased caloric intake, volume depletion and enhanced release of counter-regulatory hormones (e.g. cortisol, glucagon), the net result being a relative insulin deficiency and hyperglucagonaemia.12

Although similar hormonal changes underlie both conditions, in patients with AKA, the altered redox state, as a consequence of NADH+ formation during the metabolism of alcohol, favours the production of β-hydroxybutyrate and the β-hydroxybutyrate/acetoacetate ratio tends to be higher compared to DKA.14 However, β-hydroxybutyrate remains the main ketone produced15 and the ratio is an unreliable discriminator as similar values can result in DKA due to the presence of sepsis, vascular collapse or concomitant lactic acidosis.16

The potential for development of ketonaemia both in diabetes and with alcohol intake raises the possibility that the combination of excessive alcohol consumption together with lax diabetic control may conspire together to produce ketoacidosis. Yet, to our knowledge, no studies have examined the metabolic effects of significant alcohol ingestion in patients with type 1 diabetes.

The aim of this pilot study, therefore, was to examine the effect of acute ingestion of liberal amounts of alcohol, in a controlled environment, on glucose levels and intermediate metabolites in patients with type 1 diabetes.

Patients and methods

Twelve patients with type 1 diabetes (eight men) were recruited to the study. All the subjects were using multiple daily injections of analogue insulin (pre-meal Insulin Aspart or Insulin Lispro with basal Insulin Glargine) to control their diabetes and all reported modest use of alcohol (<14 U for females and <21 U for males each week). None were taking other drugs known to interfere with alcohol or glucose metabolism. All subjects gave written consent to participate in the study, which was approved by the Somerset Research Ethics Committee, Taunton.

Prior to the study day, subjects abstained from alcohol for 3 days. On the morning of the study, subjects had their usual insulin and breakfast prior to attending the Clinical Research Room in the Bournemouth Diabetes and Endocrine Centre. On arrival, an intravenous cannula with a three-way tap was inserted and a slow infusion of 0.9% saline commenced to maintain patency of the cannula. Venous blood glucose was monitored regularly throughout the morning and where necessary, small correction doses (2–4 U) of analogue insulin were given to attain pre-lunch blood glucose levels of 8–10 mmol/l.

Lunch was identical on both occasions and comprised a standard 600-calorie meal (83 g carbohydrate, 24 g fat, 19 g protein, 7.6 g fibre). Using a randomized double blind design (opaque, closed-envelope system with a randomized computer generated pattern), each subject also drank either white wine (men 8 units women 6 units) or an equivalent volume of alcohol-free wine, over a 90-min period (with the alternative consumed on a second visit 2 weeks later).

Venous blood samples were taken 45 min before (Baseline 1) and immediately prior to lunch (Baseline 2) and subsequently, at hourly intervals over 4 h for glucose, insulin, alcohol, triglycerides, β-hydroxybutyrate, free fatty acids, glycerol, lactate, cortisol, growth hormone and inflammatory markers—High Sensitivity CRP (hsCRP), Interleukin-6 (IL-6) and Tumour Necrosis Factor-α (TNF-α). Samples were either analysed immediately (glucose, alcohol, triglyerides, lactate, cortisol, growth hormone) or stored at −70°C prior to analysis. Heart rate and blood pressure were recorded hourly and participants were asked to complete a 100 mm Visual Analogue Score (VAS) to measure their perception of glucose levels and degree of alcohol intoxication. Extremes of the scales were ‘Sober’ and ‘Drunk’ for alcohol and ‘Low’ and ‘High’ for self-perception of glucose levels.

Following completion of blood sampling, a meal was provided and participants were accompanied home by a friend or relative who stayed with them during the evening and overnight. Each subject also received telephone follow-up the same evening, warning of the possible effects of hypoglycaemia that evening or the following morning.

Glucose, triglyerides and lactate were measured on an AU640 clinical chemistry analyser (Olympus UK Ltd, Watford); cortisol was measured by a solid-phase, competitive chemiluminescent immunoassay and growth hormone by a solid-phase, two-site chemiluminescent immunometric assay on a DPC 2500 automated immunoassay analyser (Siemens Diagnostics, Llanberis, Wales); alcohol was measured using a Phillips PU4500 gas chromatography system with flame ionization detection (Pye Unichem, Cambridge, UK); β-hydroxybutyrate and glycerol by kinetic enzymatic methods (Randox Laboratories Ltd, Crumlin, Northern Ireland); Non-Esterified Fatty Acids (NEFA) were measured by an enzymatic colorimetric method (Wako, NEFA-C, Alpha Laboratories, Eastleigh, UK); hsCRP was measured by a solid phase, chemiluminescent immunometric assay and IL-6 by a solid phase, enzyme labelled, chemiluminescent sequential immunometric assay, both on the Immulite 2500; TNF-α was measured using an ultrasensitive ELISA (Biosource, Invitrogen Ltd, Paisley, UK).

For the purposes of this pilot study, acetoacetate was not measured as practical issues in analysis preclude the ready availability of a commercial assay and17 and because of instability of acetoacetate on storage.18 Insulin was not measured due to the low cross reactivity of insulin analogues with available immunoassays19 while lack of radioimmunoassay facilities precluded the measurement of glucagon.

Statistical analysis

Data were analysed using Microsoft Excel. Difference scores between alcohol and non-alcohol arms of the study were assumed to be normally distributed and analysed using the paired t-test with a two-tailed 5% significance level. Four patients in the non-alcohol arm and three in the alcohol arm required supplemental insulin between −90 and −40 min, to adjust their blood glucose in accordance with protocol. To reduce the chance of types 1 and 2 statistical errors, the primary analysis has been based on an ‘area under the curve’ (AUC) analysis between t = 0 and t = 240, calculated using the trapezoidal rule, and using the mean of baselines 1 and 2 to represent t = 0.


Two subjects attended on only one occasion and subsequently withdrew consent. Results, therefore, are for the 10 subjects (seven male, age 43.9 ± 9.0 years, duration of diabetes 17.3 ± 13.8 years, HbA1c 8.0 ± 1.5%) who completed both arms of the study. Participants (7 of 10) received alcohol on the first visit.

Their reported usual alcohol consumption was 13 ± 5 U/week. Four patients in the non-alcohol arm and three in the alcohol arm required supplemental insulin (between 90 and 40 min before the meal) to adjust their blood glucose in accordance with protocol. For all measurements, the mean of the two baseline levels was taken for time zero levels with the exception of ketone measurements for those patients requiring supplemental insulin because of the influence of insulin on blood ketones.

Table 1 shows changes in glucose, triglycerides, free fatty acids, glycerol, cortisol and growth hormone, measured at hourly intervals after the meal on both study days, together with the alcohol levels when alcohol was taken (there was no detectable alcohol in any subject on the alcohol-free days). There was no significant difference between the two study days for the AUC analysis for any of these measurements.

View this table:
Table 1

Baseline observations and responses to a standard meal on alcohol (A) and alcohol free (NA) study days.

TimeAlcoho (mmol/L)β-OH-Butyrate (mmol/L)Lactate (mmol/L)Glucose (mmol/L)Triglycerides (mmol/L)Glycerol (mmol/L)NEFA (mmol/L)Cortisol (pmol/L)Growth hormone (IU/L)
00.00.09 ± 0.110.09 ± 0.120.93 ± 0.270.94 ± 0.329.3 ± 2.59.8 ± 2.61.1 ± 0.481.3 ± 0.6956.5 ± 54.143.0 ± 26.00.34 ± 0.290.24 ± 0.23216 ± 104227.3 ± 1030.96 ± 1.080.80 ± 1.00
+153.2 ± 29.40.12 ± 0.030.03 ± 0.011.84 ± 0.371.96 ± 0.459.6 ± 4.211.3 ± 2.81.3 ± 0.561.5 ± 0.7259.7 ± 43.037.5 ± 16.10.18 ± 0.100.13 ± 0.10232 ± 104225 ± 900.46 ± 0.360.27 ± 0.20
+2106.0 ± 31.70.11 ± 0.030.03 ± 0.012.31 ± 0.921.75 ± 0.3013.6 ± 3.413.4 ± 3.81.4 ± 0.571.5 ± 0.6849.8 ± 19.035.3 ± 20.40.16 ± 0.060.08 ± 0.05129 ± 71178 ± 701.61 ± 2.620.42 ± 0.39
+3104.5 ± 22.40.12 ± 0.040.03 ± 0.011.80 ± 0.571.26 ± 0.3014.7 ± 3.614.2 ± 3.41.5 ± 0.531.5 ± 0.7037.1 ± 14.527.4 ± 14.10.08 ± 0.030.08 ± 0.04156 ± 146157 ± 590.82 ± 1.441.06 ± 1.73
+484.9 ± 21.60.10 ± 0.060.03 ± 0.011.58 ± 0.471.12 ± 0.3214.5 ± 3.312.9 ± 2.61.7 ± 0.631.6 ± 0.8434.7 ± 11.330.7 ± 15.80.09 ± 0.030.10 ± 0.06160 ± 148.1171 ± 750.37 ± 0.670.62 ± 0.66
AUC (p)N/A<0.0010.010.920.570.070.220.530.46
  • All observations are mean ± SD

On the alcohol-free day, levels of β-hydroxybutyrate were suppressed post prandially whereas when alcohol was ingested levels rose and remained elevated 4 h after the meal. There was a significant difference between levels at all time points from 60 to 240 min (Figure 1, all P < 0.001, AUC P < 0.001).

Figure 1.

Mean ± SE of β-hydroxybutyrate levels on alcohol (full squares, unbroken line) and alcohol-free (open squares, broken line) study days. There were highly significant differences at each time point from 60 to 240 min (all P < 0.01, AUC P < 0.001).

Lactate rose in response to the meal but with alcohol the overall response was augmented (AUC P = 0.014) with peaks at 120–240 min (Figure 2). In contrast, there were no differences in any of the inflammatory markers either in AUC or at individual time points on respective study days (Table 2).

View this table:
Table 2

Comparison of mean (SD) levels of inflammatory markers after drinking alcoholic (A) and non-alcoholic (NA) wine.

Baseline 11.64 (1.49)1.75 (1.64)2.63 (2.63)2.08 (2.15)1.11 (0.70)0.85 (0.69)
Baseline 21.72 (1.12)1.89 (1.91)1.99 (2.07)2.13 (2.26)1.88 (1.43)1.36 (0.88)
±1 h1.79 (1.89)2.04 (2.12)1.99 (2.09)2.14 (2.31)2.53 (0)1.10 (0.98)
±2 h1.77 (2.11)1.78 (1.90)1.71 (1.93)2.13 (2.27)1.86 (1.34)1.42 (1.04)
±3 h1.68 (1.93)1.56 (1.51)1.88 (1.97)2.05 (2.24)1.61 (1.21)1.74 (1.73)
±4 h1.66 (1.77)2.24 (2.01)1.83 (1.86)1.98 (2.06)1.16 (0.72)1.98 (1.54)
AUC P-value0.380.670.18
Figure 2.

Mean ± SE lactate levels on alcohol (full squares, unbroken line) and alcohol-free (open squares, broken line) study days. There were significant differences at each time point from 120 to 240 min (P = 0.03 at 120 min, P < 0.01 at 180 and 240 min, AUC = 0.014).

Subjects showed no differences in the perception of prevailing glucose levels after wine compared to the non-alcoholic wine (AUC P = 0.98, Figure 3a) but correctly reported feelings of intoxication after wine (AUC P < 0.001, Figure 3b).

Figure 3.

(a) Change from baseline in Visual Analogue Scores for perception of blood glucose level (unbroken lines) on alcohol days (full triangles) and non-alcohol days (open triangles) compared with actual blood glucose level (broken lines, triangles as before). (b) Change from baseline in Visual Analogue Scores for perception of blood alcohol level (unbroken lines) on alcohol days (full triangles) and non-alcohol days (open triangles) compared with actual blood alcohol level (unbroken lines, black triangles—alcohol day only represented as no change in alcohol level on non-alcohol days).


There is no evidence that individuals with type 1 diabetes have a different approach to alcohol compared to the background population. A recent systematic review reported that consumption of a moderate amount of alcohol may be associated with a small decrease in plasma glucose concentrations but there are no studies examining diabetes self-care behaviours including medication adherence, glucose monitoring diet or exercise.20 Therefore, it is not surprising that insulin-treated patients often receive conflicting advice about how to deal with alcohol in relation to their diabetes. In this study, liberal lunchtime ingestion of alcohol was associated with failure to suppress post-prandial levels of β-hydroxybutyrate and elevated levels of lactate, with no differences in triglycerides NEFA or glycerol. The latter would be consistent with good diabetes control and adequate insulin levels preventing lipolysis and indicate that the increased β-hydroxybutyrate and lactate level were attributable to alcohol metabolism. No acute changes in levels of counter-regulatory hormones were noted.

Alcohol can also have further important implications for patients treated with insulin. Ingestion of even small amounts may impair the ability of the individual to detect the onset of hypoglycaemia at a stage when they are still able to take appropriate action, i.e. eat some carbohydrate.21 Also, hypoglycaemia per se may be mistaken for intoxication by third parties with legal as well as health consequences. Alcohol has also been shown to directly impair the usual hormonal counter-regulatory responses to low blood glucose levels22 and even small amounts can augment the cognitive deficits associated with hypoglycaemia in individuals with type 1 diabetes.23 In a laboratory-based study, Turner and colleagues reported that ingestion of alcohol with an evening meal increased the risk of hypoglycaemia the next morning in patients with type 1 diabetes.24 Here, subjects correctly identified feelings of intoxication but did not ‘feel’ more hypoglycaemic after alcohol.

Whether the type of alcohol intake, wine, beer or spirits and the potential differences between red and white wine, or whether only the total amount of alcohol independent of the medium is of importance for any health benefit of alcohol, has been the subject of many studies, without a definitive conclusion.25 Systemic markers of inflammation have been associated with a higher prevalence of cardiovascular disease and moderate alcohol consumption appears to have beneficial effects on inflammation26 notably with red wine or sparkling white Cava.27,28 Here, we found no effect of acute ingestion of white wine on markers of inflammation in subjects with type 1 diabetes. It may be noteworthy that our subjects were already regular users of alcohol and therefore any effect of alcohol on inflammation may already have been present. Alternatively the effect of the type 1 diabetic state per se on inflammatory responses remains unclear.

In summary, liberal lunchtime ingestion of alcohol was associated with failure to suppress post-prandial levels of β-hydroxybutyrate and elevated levels of lactate without changes in levels of counter-regulatory hormones, free fatty acids, glycerol and triglyceride levels. Patient education material should contain information to highlight the potential problems associated with ‘binge’ drinking which may be of more significance individuals with diabetes in addition to the established concerns for the general population.


Wessex Medical Trust.

Conflict of interest: None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
View Abstract