Abstract

The standard UK alcohol unit is used to record alcohol consumption and, in health promotion, as a useful yardstick by which the public may be encouraged to monitor their own drinking levels. To investigate the correspondence between this standard unit and the actual amount contained in the ‘usual’ drink poured by a sample of the Scottish public, participants (n = 251) were recruited from three employers in a major city—a manufacturer, an academic and a financial institution. Following a brief questionnaire, participants were asked to pour their usual drink of wine, and then spirit, into a glass. Among drinkers (n = 238), the mean amount of alcohol in a drink of wine corresponded to not 1, but 1.92 UK units. For spirit, the corresponding figure was 2.3 UK units. For wine, 43% of the sample poured more than 2 units, for spirit, 55%. (Males poured significantly more spirit than females.) These findings may have important implications for individuals who wish to promote and to adopt sensible drinking practices when consuming wine and spirit at home. Also, the reliability of many consumption surveys, where there is often the implicit assumption that a ‘drink’ is equivalent to a ‘standard unit’, must be questioned.

Introduction

The social, personal and economic costs of chronic and inappropriate alcohol consumption are well recorded within the literature. Within Scotland, the overall annual cost of alcohol problems has been estimated recently at 1 billion pounds sterling or 1.5% of Scotland's GDP (Catalyst Health Economics Consultants, 2001). Public health initiatives within the UK prompted the publication of sensible drinking guidelines (Department of Health, 1995) that advise the restriction of daily alcohol consumption to 2–3 units for females and 3–4 units for males (1 UK unit is equivalent to 8 g or 10 ml of absolute alcohol). In addition it is recommended that there should be 2 alcohol-free days in the week The public is advised to equate 1 UK unit to one half pint of beer/lager/cider, one measure of spirits or one small glass of wine.

The power of the ‘Sensible Drinking’ message will in part be dependent on its successful dissemination and its interpretation. Lader and Meltzer (Lader and Meltzer, 2002) reported that 60% of UK adults were aware of the existence of daily consumption guidelines, but of those 40% did not know what they were. The accuracy of the standard unit as a measure of ‘usual drink’ was investigated in the UK by Wilson (Wilson, 1981) who asked study participants to pour an equivalent amount of water to their usual drink of spirit into their own glass. The average glass contained 1.44 UK units (SD = 0.72).

Studies conducted outside the UK report similar findings. Carruthers and Binns (Carruthers and Binns, 1992) asked Australian subjects (n = 340), using their own glasses and water, to pour samples of drinks. Results were presented as 95% confidence limits in terms of Australian standard alcohol units (10 g). For three groups of women within the age range 18–44 years, the 95% confidence limits for the mean all lay within the range 1.0–1.5 for red wine and white wine, and 1.0–1.9 for spirits. The ranges for men were 1.1–1.5 for wine and 1.1–1.8 for spirit.

Likewise, Lemmens (Lemmens, 1994) selected a sample of 863 members of the Dutch population and, similarly to the above report, subjects were asked to pour their usual drink (using water) into a typical beverage glass. The age range of the sample was 15–70 years. When results were compared to the standard unit of 10 g of alcohol, over-estimation was evident; mean errors of 23 and 34% were reported for males and females, respectively, when pouring one drink of spirit. For wine the errors were much lower: 8 and 2%, respectively.

Kaskutas and Graves (Kaskutas and Graves, 2000) asked a sample of pregnant women in the US to define their drink sizes not by pouring liquids, but by indicating drink levels on models and photographs. They reported median values of 1.3 and 2.0 times the standard drink (12 g alcohol) for wine and spirit, respectively.

Clearly the significance of these errors becomes relevant in situations where alcohol is not drunk as the standard measures served in licensed premises. In the UK, for example, recent research suggests that alcohol is just as likely to be purchased from supermarkets as licensed premises and that among people drinking weekly, home is the most frequently quoted drinking location. Among men, 47% had drunk in their own home last week. For women, the figure was 58%, while in both groups 10% had drunk in someone else's home (Lader and Meltzer, 2002).

Given that the Health of the Nation document in the UK aims to reduce the number of men and women in England and Wales exceeding sensible drinking guidelines (Department of Health, 1992), while in Scotland the Scottish Executive has published its plan to reduce the incidence of adults exceeding weekly sensible drinking guidelines to 29% for men and 11% of women by the year 2010 (Scottish Executive, 2002), it seems timely to repeat and extend the work of Wilson (Wilson, 1981) in the UK. An estimation of the measure of drink poured by this population in their own homes will, in very real terms, impact directly on the potential success of campaigns to reduce the levels of alcohol-induced harm. In this study the aim was to be the first not to use water but genuine alcoholic drinks and to investigate in the UK the volumes of wine as well as spirit poured. It could be argued that while beer, lager, etc., is usually sold and consumed in set volumes, the volume of spirit and wine poured at home is usually in the control of the drinker, and therefore likely to deviate from a standard measure. Spirits and wine are particularly popular drinks among women within the UK, and account for 61% of alcohol consumed by women, but only 30% of that consumed by men (Lader and Meltzer, 2002).

Results will be reported for three population samples, each drawn from a quite different employment location within a UK (Scottish) city.

Methodology

Three major employers within the Edinburgh region were approached and invited to participate in the study. These were an academic institution (A), a large financial institution (F) and a manufacturing plant (M). Information detailing the study protocol was then forwarded to the relevant human resources and/or occupational health departments. After discussion and explanation of the procedures involved, a survey date was agreed. [The human resources department of one firm (M) requested that the alcoholic drinks be substituted with coloured fruit juice and that the bottles be clearly labelled to state that they did not contain alcohol.]

Staff at each location were advised of the project in the week before the survey by either E-mail or poster although the nature of the ‘pouring test’ was not described at this time.

At each firm a small booth or room was made available and was located close to, or within, the staff canteen facilities. Staff were approached during their working day at break or meal times by one of two independent researchers and invited to participate in a research study related to alcohol drinking. After reading an information sheet and signing an ethical consent form, participants were asked to complete a small questionnaire which requested details of their age range, gender, frequency of alcohol consumption, frequency of consumption at home or in public houses/clubs and, lastly, whether they poured drinks for themselves or friends. Finally, each participant was asked to pour into a glass provided by the researcher ‘the drink of red wine you would pour at home’. This exercise was immediately repeated using whisky. In the case of whisky, the participant was given the choice of using a spirit tumbler or a tall glass. The researcher recorded the choice of glass.

At this stage participants were thanked for their time and allowed to leave. The volume of liquid poured into each glass was then measured using a glass measuring cylinder (tolerance 1 ml). The glasses were cleaned and prepared for the next interview. The entire procedure took around 5 min.

The style of wine glass was chosen by first obtaining a set of glasses normally supplied to the UK licensing trade, and displaying statutory etched line measures at 125, 175 and 250 ml. Two glasses identical in style to the middle volume (175 ml), but not showing the etched mark, were then purchased for use in the study.

To minimize the possibility that staff could have time to consider what was an appropriate volume of drink to pour, no advance information of the pouring test was given. In addition, each participant was asked to delay discussing details of the study with colleagues. It was for this same reason that each firm was visited on a single day only (although both day and night shift workers participated at Firm M). To encourage participation in the study, and to ensure confidentiality and the neutrality of the researchers, minimal demographic data relating to participants were recorded.

UK standard alcohol units contained within each drink were calculated assuming an alcohol content of 40% (v/v) for spirit and 12% (v/v) for wine. Data were analysed using SPSS.

Ethical approval for the study was obtained from the University College Research Ethics Sub-Committee.

Results

The numbers of participants at the various employment locations are summarized in Table I.

Table I.

Summary of characteristics of three participating firms


Firm
 

Total workforce
 

Description
 

No. of participants
 

Non-drinkers
 
Academic (A) 500 36% academic, 64% ancillary and administrative staff 67 
Manufacturing plant (M) 700 90% manual, 10% administrative and office staff 84 (day shift n = 54; night shift n = 30) 
Financial institution (F) 3700 3% ancillary and 97% office based 100 
Total
 

 

 
251
 
13
 

Firm
 

Total workforce
 

Description
 

No. of participants
 

Non-drinkers
 
Academic (A) 500 36% academic, 64% ancillary and administrative staff 67 
Manufacturing plant (M) 700 90% manual, 10% administrative and office staff 84 (day shift n = 54; night shift n = 30) 
Financial institution (F) 3700 3% ancillary and 97% office based 100 
Total
 

 

 
251
 
13
 

In the academic institution (Firm A), approximately two-thirds of staff were employed in administration or ancillary posts; in Firm F, approximately 90% of staff were employed in office-based posts, while a similar percentage in Firm M were employed in manual posts.

Thirteen subjects identified themselves as non-drinkers and the characteristics of the remaining subjects (n = 238) are presented in Table II.

Table II.

Sample characteristics (%)



 

Total sample (n = 238)
 

Firm A (n = 64)
 

Firm F (n = 95)
 

Firm M (n = 79)
 
Gender (male) 50.4 26.6 48.4 72.2 
Age range (years)     
    <21 3.4 1.6 4.2 3.8 
    21–25 12.2 10.9 16.8 7.6 
    26–30 11.3 10.9 10.5 12.7 
    31–50 55.9 46.9 58.9 59.5 
    >50 17.2 29.7 9.5 16.5 
Reporting weekly consumption of alcohol 73 72 77 68 
Drinking at home     
    ‘often’ 40.3 46.9 37.9 38 
    ‘occasionally’ 52.9 46.9 56.8 53 
    ‘never’ 6.7 6.3 5.3 8.9 
Drinking in a pub or club     
    ‘often’ 29.0 17.2 33.7 32.9 
    ‘occasionally’ 66.8 75.0 62.1 65.8 
    ‘never’ 4.2 7.8 4.2 1.3 
‘Ever pour for friends, etc.’ (yes)
 
94
 
97
 
94
 
91
 


 

Total sample (n = 238)
 

Firm A (n = 64)
 

Firm F (n = 95)
 

Firm M (n = 79)
 
Gender (male) 50.4 26.6 48.4 72.2 
Age range (years)     
    <21 3.4 1.6 4.2 3.8 
    21–25 12.2 10.9 16.8 7.6 
    26–30 11.3 10.9 10.5 12.7 
    31–50 55.9 46.9 58.9 59.5 
    >50 17.2 29.7 9.5 16.5 
Reporting weekly consumption of alcohol 73 72 77 68 
Drinking at home     
    ‘often’ 40.3 46.9 37.9 38 
    ‘occasionally’ 52.9 46.9 56.8 53 
    ‘never’ 6.7 6.3 5.3 8.9 
Drinking in a pub or club     
    ‘often’ 29.0 17.2 33.7 32.9 
    ‘occasionally’ 66.8 75.0 62.1 65.8 
    ‘never’ 4.2 7.8 4.2 1.3 
‘Ever pour for friends, etc.’ (yes)
 
94
 
97
 
94
 
91
 

Approximately equal numbers of males and females took part in the study, and the majority (55.9%) were within the age group 30–50 years old. More people reported that they ‘often’ drank at home than in a public house or club (40.3 versus 29%) and the vast majority of the participants (94%) claimed to pour drinks for themselves or friends. Weekly consumption of alcohol was reported by 73% of the sample; monthly by 27%.

Volumes of wine poured

The mean volumes of wine poured by the total sample and various subgroups are shown below in Table III.

Table III.

Volume of red wine (in ml; mean and SD) poured by the total sample and various subgroups (for Firm M coloured fruit juice was employed)


Group
 

Mean volume (ml)
 

SD
 
Total sample (n = 238) 159.6 37.0 
Males (n = 120) 157.9 41.0 
Females (n = 118) 161.4 32.6 
Firm A (n = 64) 154.8 32.7 
Firm M (n = 79) 163.4 43.1 
Firm F (n = 95) 159.8 34.1 
Ever pour ‘yes’ (n = 223) 159.7 36.5 
Ever pour ‘no’ (n = 15)
 
158.2
 
45.1
 

Group
 

Mean volume (ml)
 

SD
 
Total sample (n = 238) 159.6 37.0 
Males (n = 120) 157.9 41.0 
Females (n = 118) 161.4 32.6 
Firm A (n = 64) 154.8 32.7 
Firm M (n = 79) 163.4 43.1 
Firm F (n = 95) 159.8 34.1 
Ever pour ‘yes’ (n = 223) 159.7 36.5 
Ever pour ‘no’ (n = 15)
 
158.2
 
45.1
 

No significant differences were detected between the various subgroups, between different age groups or day and night shift at Firm M. Assuming an alcohol content of wine of 12% (v/v), the mean volume of wine poured by the total sample corresponds to 1.92 UK units [95% confidence interval (CI) = 1.86–1.97]. Of the sample, 97.6% poured more than 1 UK unit; 42.9% poured more than 2 UK units.

Box plots indicating the distribution of the volumes of wine poured at each of the three test locations are illustrated in Figure 1. Findings for each of the three groups are presented as medians and boxes represent the upper and lower quartiles of the distribution. All four outliers in the Firm M poured 0.68 UK units or less and were male. The one outlier in Firm F was female and poured 0.77 UK units.

Fig. 1.

Number of UK units poured in a ‘drink’ of wine by the sample groups from each of the three firms. Boxes represent the upper and lower quartiles of the distribution, with the dark bar showing the median. Circles show outliers. Whiskers delineate the maximum and minimum of remaining data that lie outside the quartiles.

Fig. 1.

Number of UK units poured in a ‘drink’ of wine by the sample groups from each of the three firms. Boxes represent the upper and lower quartiles of the distribution, with the dark bar showing the median. Circles show outliers. Whiskers delineate the maximum and minimum of remaining data that lie outside the quartiles.

Volumes of spirit poured

The mean volumes of spirit poured by the total sample and various subgroups are shown in Table IV.

Table IV.

Volume of whisky spirit (in ml; mean and SD) poured by the total sample and various subgroups (for Firm M coloured fruit juice was employed)


Group
 

Mean volume (ml)
 

SD
 
Total sample (n = 238) 57.1 24.0 
Males (n = 120) 63.0 25.0 
Females (n = 118) 51.2 21.5 
Firm   
    A (n = 64) 54.2 22.4 
    M (n = 79) 66.2 27.9 
    F (n = 95) 51.6 19.2 
Age (years)   
    <20 (n = 8) 61.4 18.0 
    21–25 (n = 29) 74.1 36.9 
    26–30 (n = 27) 59.9 22.6 
    31–50 (n = 133) 53.2 20.0 
    >50 (n = 41) 55.3 22.2 
Ever pour   
    ‘yes’ (n = 223) 57.7 24.2 
    ‘no’ (n = 15) 49.3 21.2 
Glass   
    tall glass (n = 30) 56.5 21.9 
    small (n = 205)
 
57.1
 
24.5
 

Group
 

Mean volume (ml)
 

SD
 
Total sample (n = 238) 57.1 24.0 
Males (n = 120) 63.0 25.0 
Females (n = 118) 51.2 21.5 
Firm   
    A (n = 64) 54.2 22.4 
    M (n = 79) 66.2 27.9 
    F (n = 95) 51.6 19.2 
Age (years)   
    <20 (n = 8) 61.4 18.0 
    21–25 (n = 29) 74.1 36.9 
    26–30 (n = 27) 59.9 22.6 
    31–50 (n = 133) 53.2 20.0 
    >50 (n = 41) 55.3 22.2 
Ever pour   
    ‘yes’ (n = 223) 57.7 24.2 
    ‘no’ (n = 15) 49.3 21.2 
Glass   
    tall glass (n = 30) 56.5 21.9 
    small (n = 205)
 
57.1
 
24.5
 

Using the Mann–Whitney statistical test, significant differences were noted between the volumes of spirit poured by males compared to females (P < 0.01), Firm M compared to Firm A (P < 0.01), Firm M compared to Firm F (P < 0.01), and between the 21–25 year age group and the over 50 age group (P < 0.05) and over 30 age group (P < 0.01).

Assuming an alcohol content in spirit of 40% (v/v), the mean volume of spirit poured by the total sample corresponds to 2.3 UK units (95% CI = 2.16–2.41) (97.5% of the sample poured in excess of 1 UK unit, while 55% poured more than 2 units). Box plots are presented in Figure 2. All three outliers in Firm A poured in excess of 4.3 UK units and were female. The three outliers in Firm F poured 4.0 UK units or more, reported weekly consumption of alcohol and were all male. The outlier in Firm M was male, and reported only occasional drinking (monthly) at home and in public houses.

Fig. 2.

Number of UK units poured in a ‘drink’ of spirit by each of the three sample groups (see also Figure 1 legend).

Fig. 2.

Number of UK units poured in a ‘drink’ of spirit by each of the three sample groups (see also Figure 1 legend).

Discussion

The present data suggest two key characteristics of the drink of wine or spirit poured by members of the Scottish population for their own consumption: (1) there is considerable variation in the amount of drink that they pour and (2) on average they routinely pour not 1, but approximately 2 UK standard units (1.9 units of wine and 2.3 units of spirit). More than 2 units of wine were poured by 43% of the sample. For spirit the figure was 55%. This result, if applicable to the general population, may have important implications for those individuals who wish to regulate their personal consumption in line with health guidelines. In addition, those who seek to quantify population consumption levels and relate them to risk must be aware that the simple assumption that one ‘drink’ translates to 1 UK unit may have an associated error of 100% if that drink was wine or spirit, poured and consumed outside of licensed premises. [The alcohol content of the average poured glass of wine may in fact be greater. This study assumed an alcohol content of wine of 12% (v/v). Wine typically sold in the UK varies between 10.5 and 14% (v/v). If the latter wine had been used in the present study, then the average glass of wine would have contained 2.2 UK units.] If substantiated by future research, the present findings impact directly on the potential success of government sponsored health initiatives in this field.

For wine, no significant differences were noted between the volumes of wine poured by the samples representing each of the three firms or between males and females. However, significant differences were noted in the quantity of spirit poured; males pouring more than females, and Firm M (manufacturing) pouring more than Firms A (academic) and F (financial). The gender difference in volume poured may well explain the latter findings for Firm M; the sample from Firm M was 72% male.

The limitations of these findings must be acknowledged. The sample is one of convenience. The method of selection offers no certainty that the sample characteristics reflect those of the participating firms. In addition, little is known about the drinking experience of the participants. In order to comply with the requests made by the various human resources departments, minimal data relating to weekly consumption levels were recorded. It may be that the advance notice of the study ensured that the heavy drinkers or those concerned about their drinking did not take part, and the findings refer to that part of the population who are drinking well within guidelines and are therefore potentially less experienced, i.e. less accurate, pourers of drinks. This study may have failed to reveal the pouring behaviour of those who are most at risk from their drinking. However, it is known that 73% of the sample reported weekly consumption of alcohol, 40% ‘often’ drank at home and 94% poured drinks for themselves or friends. Further work is required to establish whether these findings are replicated in other parts of the UK.

It is also important to stress that an estimate of the ‘usual drink’ poured has been obtained. There is no evidence that the participants incorrectly assumed that this volume equalled 1 standard UK unit. This awareness, which impacts on the relevance of the present findings to the effectiveness of present health education initiatives, needs to be tested by future research.

In this study it was proposed to improve on previous work in this area by being the first study to use alcoholic beverages rather than water as the test liquid. It seemed reasonable to believe that the smell and colour of wine and whisky might have an influence on the volume poured. [However due to the conditions made by one Human Resources Department, this change could not be adopted at one test site (Firm M), although a colour-matched fruit juice solution was used.]

The significantly higher volume of spirit poured by males, it has been suggested, may explain the higher volume of spirit poured by members of Firm M (Firm M sample was 72% male). Alternatively the use of alcohol substitute solutions and therefore the lack of the associated smell of the spirit, may have encouraged individuals in Firm M to be less restrained as they poured. However, this argument would suggest that a similar difference should have been detected when wine was employed, no such effect was seen. It may be that a colour-matched liquid is an appropriate substitute for alcohol and has minimally affected the results.

The values reported here for mean units of spirit poured are approximately 60% higher than the mean reported by Wilson (Wilson, 1981) in the UK more than 20 years ago. Similarly the errors reported by Lemmins (Lemmins, 1994) studying a sample of the Dutch population [23 and 34% for males and females respectively when pouring 1 unit (10 g) of spirit, and 8 and 2% when pouring wine] are much lower than present findings. Carruthers and Binns (Carruthers and Binns, 1992) study presented their findings as 95% confidence limits in terms of Australian standard alcohol units (10 g). For women aged 18–44 years, the confidence limits all lay within the range 1.0–1.5 units for red wine and white wine (here 1.87–2.01 UK units) and 1.0–1.9 units for spirits (here 1.9–2.21). The ranges for men were 1.1–1.5 units for wine (here 1.81–1.98) and 1.1–1.8 units for spirit (here 2.34–2.70). The Australian standard unit contains 25% more alcohol than the UK unit; however, even after correction by this factor, this small sample of the UK population still appear to pour drinks containing more alcohol than their Australian counterparts. Part of the discrepancy between the present findings and the studies reported above may be due to their exclusive use of water as a test drink.

It seems reasonable to expect that the size of glass will affect the amount of alcohol poured. Stockwell et al. (Stockwell et al., 1991) reported that the larger the size of the wineglass, the more people deviated from the standard volume. In the present study the researcher provided a standard wineglass and a choice of spirit glasses. In the latter case, no significant difference was noted between the amounts of spirit poured into these two glasses. Many participants commented to the researchers that the study wine glass was much smaller than the one they would use at home.

Another factor which may influence the results is the effect of being observed while pouring alcohol, a substance known to be associated with various health education messages. However, it seems plausible, as suggested by Carruthers and Binns (Carruthers and Binns, 1992), that observation might have an inhibitory effect and would tend to decrease the amount of alcohol poured. If true, the present study may have underestimated the pouring error.

It remains to be determined whether the number of drinks an individual has already consumed in a session will influence the amount poured. In addition, it is possible that only the first drink at home is poured into an empty glass, later drinks may simply ‘top up’ the level of alcohol and be very difficult to quantify.

In conclusion, provisional evidence suggests that individuals within Scotland who consume a major proportion of their alcohol away from licensed premises, predominantly in the form of wine and spirit, may habitually pour drinks which contain much greater amounts of alcohol than they, and many health educators and researchers assume. Given the popularity of these drinks with women, the message may be most relevant to this section of the population (Gill, 2000; Lader and Meltzer, 2002) and underscores the need to accelerate the trend to adopt standard unit content labelling of drink bottles (Stockwell and Single, 1997).

The assistance of the following researchers is gratefully acknowledged: Sarah Henderson and Suzanne Whiteford. This research was supported by a grant from the Portman Group, London, UK.

References

Carruthers, S.J. and Binns, C.W. (
1992
) The standard drink and alcohol consumption.
Drug and Alcohol Review
 ,
11
,
363
–370.
Catalyst Health Economics Consultants (
2001
) Alcohol Misuse in Scotland: Trends and Costs. Catalyst Health Economics Consultants UK.
Department of Health (
1992
) Health of the Nation: A Strategy for Health in England. HMSO, London.
Department of Health (
1995
) Sensible drinking: report of an interdepartmental working group. HMSO, London.
Gill, J. (
2000
) The effects of moderate alcohol consumption on female hormone levels and reproductive function.
Alcohol and Alcoholism
 ,
35
,
417
–423.
Kaskutas, L and Graves, K. (
2000
) An alternative to standard drinks as a measure of alcohol consumption.
Journal of Substance Abuse
 ,
12
,
67
–78.
Lader, D. and Meltzer, H. (
2002
) Drinking: Adults' Behaviour and Knowledge in 2002. Office for National Statistics, London.
Lemmens, P.H. (
1994
) The alcohol content of self-report and ‘standard drinks.
Addiction
 ,
89
,
593
–601.
Scottish Executive (
2002
) Plan for Action on Alcohol. Scottish Executive Health Department, Edinburgh.
Stockwell, T. and Single, E. (
1997
) Standard unit labelling of alcohol containers. In Plant, M., Single, E and Stockwell, T. (eds), Alcohol; Minimising the Harm, What Works? Free Association Books, London, pp. 85–104.
Stockwell, T., Blaze-Temple, D. and Walker, C. (
1991
) The effect of ‘standard drink’ labelling on the ability of drinkers to pour a ‘standard drink’.
Australian Journal of Public Health
 ,
15
,
56
–63.
Wilson, P. (
1981
) Improving the methodology of drinking surveys.
The Statistician
 ,
30
,
159
–167.

Comments

0 Comments