OUP user menu


Fran Ridout, Stuart Gould, Carlo Nunes, Ian Hindmarch
DOI: http://dx.doi.org/10.1093/alcalc/agg092 381-385 First published online: 1 July 2003


Aims: To assess the effects of carbon dioxide (CO2) in champagne on psychomotor performance and blood-alcohol concentration (BAC). Methods: Twelve subjects consumed ethanol (0.6 g/kg body weight) served as champagne or champagne with the CO2 removed, in a crossover study. Results: Champagne produced significantly greater BACs and significantly increased reaction times in a divided attention task, than degassed champagne. Conclusions: The CO2 in champagne may accelerate absorption of alcohol, leading to more rapid or severe intoxication.


The adverse effects of alcohol on cognitive and psychomotor skills are well documented, particularly with regard to those components of behaviour that may be related to car driving performance. While there is no international, scientific or legislative uniformity in blood-alcohol concentration (BAC) levels admissible for driving, there is substantial evidence to suggest that increases in reaction time and performance errors can be found at doses that are well within legally defined limits. Impairment of vigilance, perception, cognitive and visual skills, divided attention, as well as simulated driving performance, has been widely reported at BACs of 50 mg/dl or below (Starmer, 1989; Howat et al., 1991), while the risk of a single-vehicle crash for drivers with BACs of between 20 and 40 mg/dl has been estimated to be 1.4 times higher than that of drivers with BACs of zero (Zador, 1991). Interactions between ethanol and other psychotropic drugs, including codeine, benzodiazepines and antihistamines, have also been shown to have a deleterious effect on performance at BACs of less than 40 mg/dl (Doria, 1990; Kerr and Hindmarch, 1998).

Nevertheless, it is relatively common for people to drive having consumed a legally acceptable amount of ethanol. This has been confirmed in roadside surveys conducted in the UK between 1988 and 1990, in which 99.2% of drivers voluntarily took breath alcohol tests, revealed that 2.3% of those tested had breath alcohol results equivalent to BACs of between 40 and 80 mg/dl (the legal limit in the UK being 80 mg/dl), while only 1.2% exceeded this limit (Maycock, 1997). The concept of ‘units’ (10 ml or 8 g of ethanol) has been widely publicized as a means of monitoring alcohol consumption, and labelling on spirit, wine and beer bottles has been adopted on a voluntary basis as a means of informing people how many units of alcohol they contain. However, there is clearly a need to inform the public, particularly those who drink before undertaking potentially hazardous activities such as driving, of other factors that may impact on BAC levels.

It is widely believed that champagne is more intoxicating than wine and it has been suggested that carbonated beverages speed up the emptying of the stomach into the small intestine, where alcohol is absorbed faster (Lewy, 1995). However, a thorough literature search has failed to reveal any investigation performed under controlled conditions that might provide evidence to support or refute this belief. The purpose of this study was therefore to assess the effects of champagne and degassed champagne on psychometric performance and BAC.



Six male and six female subjects were recruited. All 12 participated in study 1, and six (four female and two male) from the same sample participated in study 2. Subjects were asked to have a standard breakfast on each assessment day, no less than 3 h before drinking commenced. All subjects gave written informed consent. Ethical approval was obtained from the Ethics Committee of the University of Surrey. The study was conducted in accordance with Good Clinical Practice guidelines.


Subjects were trained on all psychometric measures to ensure that learning effects were minimized (Parkin et al., 1997). A two-way counterbalanced crossover design was used for both studies. The minimum washout period was 7 days. A non-vintage brut champagne was used for both conditions. Champagne doses were measured out from freshly opened bottles and the wine (degassed champagne) condition consisted of the same champagne, from which the carbon dioxide (CO2) had been removed by whisking with an electric blender. Samples of both drinks were assessed with a blood gas analyser probe and were also tested for alcohol content and pH. All subjects received alcohol in a dose of 0.6 g/kg body weight, e.g. a 65 kg subject received 325 ml. The subjects were allowed 20 min to consume the drinks.

Subjects were unaware that the ‘wine’ was degassed champagne. They were asked to abstain from alcohol for 24 h prior to each assessment period and were breathalysed to assess compliance. A validated psychometric test battery, comprising critical flicker fusion (CFF); a choice reaction time test (CRT) including recognition reaction time (RRT) and motor reaction time (MRT) components; a Sternberg memory task (STM); a compensatory tracking task (CTT); and rapid visual information processing (RVIP), were performed before the drinks were administered. The tests were repeated 20 and 60 min after the end of the 20-min drinking period.

Study 2 employed the same protocol (without psychometric testing) with the addition of serial blood samples taken 5, 10, 15, 20, 25, 30, 35 and 40 min after drinking, using an indwelling cannula. BAC was measured using gas–liquid chromatography (Tagliaro et al., 1992).


Critical flicker fusion. CFF assesses CNS arousal. Subjects are required to discriminate flicker from fusion, and vice versa, in a set of four light-emitting diodes arranged in a 1-cm square. Individual thresholds are determined by the psychophysical method of limits on four ascending and four descending scales (Woodworth and Schlosberg, 1958). The means of the eight scales give the threshold frequency in hertz (Hz). CFF has been shown to be sensitive to a variety of psychoactive compounds, e.g. antidepressants and anxiolytics (Hindmarch, 1975, 1994). A lowering of the threshold indicates a reduction in CNS arousal.

Choice reaction time. CRT (Hindmarch, 1980) is used as an indicator of sensorimotor performance, assessing the ability to attend and respond to a critical stimulus (Sherwood and Kerr, 1993). Subjects are required to extinguish one of six equidistant red lights by pressing the associated response button as quickly as possible. There are 50 trials per test. RRT (in ms) is the time between stimulus onset and the subject lifting their finger from the start button. MRT (in ms) indexes the movement component of this task.

Compensatory tracking test. CTT entails using a mouse to keep a cursor in alignment with a moving target on a computer screen. The mean difference between the centres of target and cursor in pixels is recorded, while a peripheral awareness task assesses speed of response (in ms) to a stimulus presented in the periphery of vision. CTT has been shown to be sensitive to the effects of psychotropic drugs (Subhan et al., 1986; Hindmarch and Shamsi, 2002).

Rapid visual information processing. RVIP assesses the performance of attention mechanisms in remaining vigilant to periodically occurring events. Subjects are required to monitor a series of single digits (0–9) appearing on the screen at a rate of 100 digits every minute, and respond to consecutive sequences of three odd or even digits by pressing a button (Wesnes and Warburton, 1983). The response measures are the mean reaction time (in ms) and the number of valid responses.

Sternberg memory task. High-speed scanning and retrieval from short-term memory are assessed using a reaction time method (Sternberg, 1966). Subjects memorize a random series of one, three or five digits, which are followed by a series of 12 single probe digits. Subjects indicate whether each probe digit is contained within the original stimulus set. The mean reaction time and number of correct/incorrect responses to 72 trials are recorded. Performance on the STM is sensitive to psychoactive compounds (Subhan and Hindmarch, 1984).

Statistical methods

SPSS version 10 (SPSS, Inc., Chicago, IL, USA) was used to analyse psychometric results. Owing to presence of non-uniform residuals in the results, a non-parametric test, Wilcoxon’s matched-pairs signed ranks test was employed. A between-treatment pairwise comparison was carried out on the maximum change from baseline for each variable, and in addition, within-treatment comparisons were used to examine the effects of time on performance for each drink. The areas under the plasma concentration time curves (AUC) of blood samples were analysed using SAS version 8.2 (SAS Institute, Cary, NC, USA) mixed procedure.


Samples taken during testing revealed that the ethanol content of the degassed champagne was 11.4 g/dl, whereas that in natural champagne was 11.6 g/dl. Analyses of the gas content of the drinks confirmed the very low CO2 level in degassed champagne, but the equipment was designed to measure dissolved gas (upper limit: 150 mmHg ± 1.1 SD) and the CO2 content of champagne was too high to be measured and was therefore not determined. The pH levels of samples measured at 4°C were 2.91 for champagne and 2.96 for degassed champagne.

BAC. There were significant differences between champagne and degassed champagne in AUC0–5 (P < 0.01), AUC0–10 (P < 0.01), AUC0–15 (P < 0.05) and AUC0–20 (P = 0.0274), when champagne produced a plasma concentration time curve of 203.08 compared with 167.33 mg•min/100 ml with degassed champagne. The between-drink difference in AUC0–25 was not significant (P = 0.0668), and there were no other significant treatment differences (Fig. 1).

Fig. 1.

Mean blood-alcohol concentration (BAC) time curves. Values are means ± SEM in samples taken 5–40 min after the end of the 20 min drinking period with treatments (champagne and degassed champagne). Number of subjects = 6. *P < 0.5, **P < 0.01 (two-tailed).


CFF. There was no significant effect of treatment on CFF. There were also no significant within-treatment differences with degassed champagne. Champagne did not affect performance at +20 min, but significant impairment was revealed with champagne at +60 min (P = 0.045) (Fig. 2).

Fig. 2.

Mean scores (baseline and post-treatment) for psychometric assessments. Values are means ± SEM. *P < 0.5, **P < 0.01 (two-tailed) comparisons with baseline score. For abbreviations, see text.

CRT. There were no significant differences between treatments in either component of the CRT test, but there were significant within-treatment effects. There was a significant increase in RRT at +20 min (P = 0.023) and +60 min after administration (P = 0.05) of degassed champagne. Champagne also significantly impaired reaction times at both post-dose evaluations (P = 0.002) (Fig. 2). There were no significant differences between the baseline assessments of MRT and any post-dose assessment with either drink.

CTT. There was a significant between-treatment difference in reaction times to peripheral stimuli. Subjects were significantly slower to respond after drinking champagne (P = 0.019) than they were after drinking degassed champagne (Table 1). There was also a significant within-treatment time effect with champagne at both +20 min (P = 0.012) and +60 min (P = 0.004), but not with degassed champagne (Fig. 2). There was no significant treatment effect on the tracking performance component of this task, but there were significant within-treatment effects. With champagne, tracking performance was significantly impaired at +60 min (P = 0.01), but not at +20 min (P = 0.06). There were no significant within-treatment effects with degassed champagne (Fig. 2).

View this table:
Table 1.

Mean maximum change from baseline (changes were recorded 20 or 60 min post-treatment)

Psychometric testDegassed champagne [mean (SEM)]Champagne [mean (SEM)]
CFF, critical flicker fusion; RRT, recognition reaction time; MRT, motor reaction time; CTT, compensatory tracking task; RVIP, rapid visual information processing; STM, Sternberg memory task.
*Significant between-treatment difference, P < 0.05 (two-tailed). Non-parametric paired comparisons.
CFF (Hz)–1.0 (0.8)–1.7 (0.9)
RRT (ms)44.6 (16.8)45.9 (17.0)
MRT (ms)–27.1 (24.2)22.8 (12.8)
CTT reaction time (ms)46.3 (37.0)196.5 (62.9)*
CTT error (pixels)8.1 (7.3)5.7 (1.7)
RVIP reaction time (ms)–24.5 (25.3)33.8 (9.7)
RVIP valid responses (n)–2.8 (6.1)–5.08 (3.2)
STM reaction time (ms)–59.9 (61.2)–10.1 (33.8)
STM correct responses (n)–2.3 (2.3)–1.1 (1.2)

RVIP. There was no significant between-treatment effect on RVIP performance. There were no significant within-treatment effects with degassed champagne. Champagne significantly increased reaction times at +20 min (P = 0.004), but not at +60 min (P = 0.071), and reduced the number of valid responses to target stimuli at +20 min (P = 0.021) (Fig. 2).

STM. There were no significant between-treatment effects on reaction time or the number of correct responses. There were also no significant within-treatment effects with either degassed champagne or champagne (Table 1).


The BAC analysis suggests that the high CO2 content of champagne may increase the rate of absorption of ethanol, as a significant treatment difference was revealed during the first 20 min after ingestion. The results of the psychometric testing support this finding as there was a significant between-treatment difference in reaction times to CTT peripheral stimuli, with a mean maximum change from baseline of 196.54 ms following administration of champagne, compared with 46.3 ms with degassed champagne. In addition, the results of the within-treatment analyses showed that champagne significantly increased reaction times in the RRT, RVIP and CTT tests, as well as reducing CFF thresholds (impairment) and the number of correct responses in the RVIP test. Champagne also significantly impaired tracking performance in the CTT, while degassed champagne had a significant detrimental effect on only one performance variable: the RRT component of the CRT test. Neither drink impaired short-term memory, as assessed by STM, and neither drink had a significant effect on MRT.

The significant treatment difference revealed in the CTT test is of interest, as divided attention tasks are particularly sensitive to the effects of ethanol, with impairment regularly being reported at BACs at 20 mg/dl or below (Hindmarch et al., 1992; Roehrs et al., 1994). Typically, ethanol impairs performance, by increasing concentration, on the main component of the task (in this case tracking), to the detriment of the secondary task (reacting to peripheral stimuli). Our results corroborate this, as the between-treatment difference was seen in the reaction time component of this task and degassed champagne had no effect on tracking error or reaction time. In addition, champagne impaired reaction time at both post-treatment assessments, but only affected tracking performance at +60 min. These findings and the results of the other tests employed are also in accordance with the results of a recent review on the effects of low BACs on performance related to car driving skills. STM and tests of psychomotor performance such as MRT, neither of which was significantly affected by either drink, are typically less sensitive to the impairing effects of ethanol than the other measures employed here, impairment rarely being detected at BACs of <60 mg/dl (Moskowitz and Fiorentino, 2000).

Although the results of the BAC analysis show that both drinks appear to reach a similar plateau after the first 20 min, the psychometric results provide additional support for the view that champagne may produce more rapid or severe intoxication than degassed champagne or wine. The analysis of the samples of degassed champagne and champagne show that a small amount of ethanol (0.2 g/dl) may have been lost in the preparation of the degassed champagne condition and the pH value of degassed champagne was found to be 0.05 higher than that of champagne. However, these differences between the treatments are relatively minor and seem unlikely to provide a sufficient explanation of these results. It therefore appears that the high CO2 content of champagne may alter gastric emptying in a way that enhances the absorption of alcohol. Tests of gastric emptying would cast light on a possible mechanism. One possible limitation of this study is that BAC measurements were only taken for 40 min post-treatment. Further studies would be enhanced by the inclusion of additional measurements corresponding to the timing of the 60 min test battery to embrace the average peak BAC levels that occur between 45 and 90 min following ingestion (Ideström and Cadenius, 1968). The absence of a placebo arm in this study may, however, be justifiable, because the difficulty of blinding treatments could have confounded the results, if subjects had been aware of whether or not they had consumed alcohol.


Although, this was very much a pilot study, employing a small number of subjects, the results support the popular belief that champagne may be more intoxicating than wine, although the mechanism remains unclear. Further research into the effects of CO2 in alcoholic drinks could be of relevance to car drivers and others who wish to monitor their BACs.


This project was partly sponsored by Lay and Wheeler.


  • * Author to whom correspondence should be addressed at: HPRU Medical Research Centre, University of Surrey, Egerton Road, Guildford GU2 7XP, UK.


View Abstract