2. alcohol consumption in 1987

6
126 2. ALCOHOL CONSUMPTION IN 1987 Measurement of alcohol consumption This survey, like previous ad hoc surveys carried out by OPCS, asked drinkers about their alcohol consumption over a seven day period. This approach reflects the importance of covering a weekly cycle to represent the varying consumption patterns of regular drinkers: for example, people tend to drink more at weekends than during the week. Difficulties in measuring alcohol consump- tion in social surveys arise mainly because people tend to understate the amount they drink. In some cases, under-reporting is deliberate, because informants are just not prepared to admit how much they drink, perhaps because another member of the fami- ly is present, or because they think that the interviewer (usually a woman) will be dis- approving. Experience gained from previous surveys suggests that whereas deliberate under- reporting of alcohol consumption is probably quite rare (insofar as interviewers seldom suspect it), unintentional under-reporting is very common indeed. On one level, there is probably a sub conscious tendency to under- rather than over-estimate amounts drunk, and on another, people genuinely for get occa- sions on which they had a drink, and forget how much they had to drink. Furthermore, estimating amounts drunk is particularly dif- ficult for people drinking at home, where drinks are not dispensed in standard quan- tities, and are probably usually larger than those bought on licensed premises, It was therefore important on this survey, as on earlier ones, to use appropriate techni- ques to help informants recall their drinking as accurately as possible. The seven day peri- od began with “yesterday”, that is, the day before the interview, and informants were taken back, day by day, through the preceding week, and asked to remember each occasion on which they had had a drink. They were asked the time and location of each drinking occasion during the week, as well as exactly what they had drunk. Non-alcoholic drinks and those with a very low alcohol con- tent were excluded from this seven day retrospective drinking diary. To enable comparison between people who had consumed different types of alcoholic drink, amounts were converted to standard units containing a similar amount of pure al- cohol. One unit is equivalent to a half pint of beer, a single measure of spirits (l/6 gill) a glass of wine (about 4.5 fluid ounces) or a small glass of sherry or fortified wine (2 fluid ounces). This conversion method is somewhat rough and ready, because the alcoholic strength of drinks within the same group varies - for example, the alcohol content of beer ranges from under 3% to about 5% by volume. On average, however, one standard unit contains approximately 8.5 grams of alcohol. Accuracy of results Since the data presented in this report were obtained from a sample of the population, they are subject to sampling error, and this must be taken into account when considering the results. Any sample is only one of an almost infinite number that might have been selected, all producing somewhat different results. Sampling error is particularly high for data on alcohol consumption because of the wide variation in the amounts people drink (from nothing at all to over 100 units a week) and the uneven distribution of the amounts, as shown in Fig. 2.1. If everybody drank more or less the same amount, then any one sample would be likely to produce an estimate of average consumption that was fairly close to the true value in the population. However, if, as is the case, there is wide variation in the amounts people drink, it is more likely that

Post on 21-Jun-2016

224 views

Category:

Documents


2 download

TRANSCRIPT

126

2. ALCOHOL CONSUMPTION IN 1987

Measurement of alcohol consumption

This survey, like previous ad hoc surveys carried out by OPCS, asked drinkers about their alcohol consumption over a seven day period. This approach reflects the importance of covering a weekly cycle to represent the varying consumption patterns of regular drinkers: for example, people tend to drink more at weekends than during the week.

Difficulties in measuring alcohol consump- tion in social surveys arise mainly because people tend to understate the amount they drink. In some cases, under-reporting is deliberate, because informants are just not prepared to admit how much they drink, perhaps because another member of the fami- ly is present, or because they think that the interviewer (usually a woman) will be dis- approving.

Experience gained from previous surveys suggests that whereas deliberate under- reporting of alcohol consumption is probably quite rare (insofar as interviewers seldom suspect it), unintentional under-reporting is very common indeed. On one level, there is probably a sub conscious tendency to under- rather than over-estimate amounts drunk, and on another, people genuinely for get occa- sions on which they had a drink, and forget how much they had to drink. Furthermore, estimating amounts drunk is particularly dif- ficult for people drinking at home, where drinks are not dispensed in standard quan- tities, and are probably usually larger than those bought on licensed premises,

It was therefore important on this survey, as on earlier ones, to use appropriate techni- ques to help informants recall their drinking as accurately as possible. The seven day peri- od began with “yesterday”, that is, the day before the interview, and informants were taken back, day by day, through the preceding week, and asked to remember each

occasion on which they had had a drink. They were asked the time and location of each drinking occasion during the week, as well as exactly what they had drunk. Non-alcoholic drinks and those with a very low alcohol con- tent were excluded from this seven day retrospective drinking diary.

To enable comparison between people who had consumed different types of alcoholic drink, amounts were converted to standard units containing a similar amount of pure al- cohol. One unit is equivalent to a half pint of beer, a single measure of spirits (l/6 gill) a glass of wine (about 4.5 fluid ounces) or a small glass of sherry or fortified wine (2 fluid ounces). This conversion method is somewhat rough and ready, because the alcoholic strength of drinks within the same group varies - for example, the alcohol content of beer ranges from under 3% to about 5% by volume. On average, however, one standard unit contains approximately 8.5 grams of alcohol.

Accuracy of results

Since the data presented in this report were obtained from a sample of the population, they are subject to sampling error, and this must be taken into account when considering the results. Any sample is only one of an almost infinite number that might have been selected, all producing somewhat different results.

Sampling error is particularly high for data on alcohol consumption because of the wide variation in the amounts people drink (from nothing at all to over 100 units a week) and the uneven distribution of the amounts, as shown in Fig. 2.1. If everybody drank more or less the same amount, then any one sample would be likely to produce an estimate of average consumption that was fairly close to the true value in the population. However, if, as is the case, there is wide variation in the amounts people drink, it is more likely that

127

Fig. 2.1. Percentage of the sample who drank more than a given number of units of alcohol in the seven days before the interview.

70%,

m-

50% -

40% -

30% -

205t-

IO%-

r---v-- -7 ---7 0 10 20 30 40 50 60 70

Weekly akohol eonrumphoo (uniu)

any one sample may produce a result that is very different from the true value.

The other main factor which affects sampl- ing error is sample size; other things being equal, the larger the sample, the less is the likelihood of it being an unrepresentative one, so the sampling error is smaller. Thus sampl- ing errors for subgroups of the sample are greater than for the sample as a whole. Dif- ferences between groups are generally only commented on if they are statistically signili- cant at the 95% level; that is, there is a 95% probability that the difference in question reflects a real one in the population and is not due to the chances of sampling.

Drinking in the seven day reference period

Proportions drinking during the seven days before interview

Overall, 69% of those interviewed had had an alcoholic drink during the previous week,

*See original source for Tables 2.1-2.13.

and a further 25% had not done so, but said that they did sometimes drink. Only 7% of respondents said that they never drank at all nowadays, not even on special occasions such as Christmas or birthdays, or for medicinal reasons.

Men were considerably more likely than women to have had a drink in the previous week - overall, 78% of men, compared with 62% of women, had done so. There was also an association between age and this aspect of drinking behaviour: among both men and women, the proportion who had drunk something alcoholic in the last week was highest among those aged 18-24, and then de- clined with increasing age.

As this is the first time they have been in- cluded in a survey of adult drinking, those ag- ed 16 and 17 are of particular interest. The proportion of 16 and 17 year olds who had drunk something alcoholic in the previous week was lower than among those aged 18-24 - presumably this would be likely, if for no other reason than that their opportunities for drinking may be more limited, as they cannot legally buy alcohol or drink in licensed bars. Although for all other age groups, men were more likely than women to have had a drink in the previous week, among those aged 16 and 17 this was reversed - 72% of women compared with 63% of men having done so. It should be emphasised that this difference is not statistically significant (that is, there is. more than a 5% probability that it occurred by chance sampling variation). Furthermore, it does not accord with other recent survey findings [l], which have tended to show slightly higher proportions of drinkers among young men than among young women of that age (Table 2.1*).

Reasons for not drinking Various studies have suggested that total

abstention from drinking, as well as heavy al- cohol consumption, is associated with in- creased ill health and that drinking in moderate amounts may possibly be beneficial

128

to health [2]. The report on the 1982 General Household Survey showed, for example, that 52% of non-drinkers reported a longstanding illness, compared with 35% of all other respondents [3]. However, there is doubt about the extent to which the association of abstention with ill health arises because peo- ple give up drinking following the onset of il- lness, rather than the illness being in some way a consequence of not drinking. Questions about respondents health were beyond the scope of this survey, but those classified as non-drinkers were asked why they didn’t drink. The most common reason, given by two fifths of those asked, was simply dislike of the taste or effects of alcohol. About a fifth of non-drinkers said that they did not drink because of health problems.

Assuming this proportion applies also to the GHS data, then the prevalence of longstanding illness among the remaining abs- tainers identified on the GHS would have been about 45%. This is still somewhat higher than the rate for other respondents, sug- gesting that only part of the raised morbidity among abstainers is accounted for by those whose abstention from alcohol is due to ill health (Table 2.2).

Average amounts drunk Alcohol consumption data can be pre-

sented as averages based either on the whole sample or on recent drinkers-those who had a drink in the week before the interview. The first method enables the estimation of overall consumption among groups of the population (for example, young women, retired men, and so on) whereas the second gives more infor- mation about individuals who do drink. Whichever method is used, similar relation- ships and associations are generally apparent in the data, since groups with a high propor- tion of recent drinkers tend also to have a high average consumption per drinker. In this chapter, consumption averages are based on all persons.

Overall, those in the sample had drunk an

average of 9.4 units of alcohol in the previous week - that is, just over 4.5 pints of beer, or the equivalent in other types of drink. This overall figure conceals a threefold difference in consumption between men and women - men had drunk an average of 14.5 units com- pared with an average consumption of 4.8 units for women.

Beer, lager and cider accounted for two thirds of the total amount of alcohol consum- ed by the sample, but was a much more popu- lar type of drink with men than with women, among whom no one type of drink predominated. In fact, if beer is excluded, the total amount of alcohol consumed, 3.2 units per person, is the same for men and for women (Table 2.3).

Average consumption by region There was considerable regional variation

in average alcohol consumption, but because of the small sample sizes in most regions, this was to be expected. Overall weekly consump- tion ranged from 6.9 units in East Anglia to 10.7 units in the Yorkshire and Humberside region, and was generally higher to the north and west of a line drawn from the Severn to the Wash. However, when men’s and women’s consumption were examined separately, it was evident that this geographical variation really only occurred among men; among women, no pattern was apparent. These findings are broadly consis- tent with those from the 1984 General Household Survey [4].

Although long drinks - beers, lagers and eiders - accounted for over half of all alcohol consumed in all regions, there was some regional variation in the popularity of dif- ferent types of drink. Long drinks accounted for about three quarters of the alcohol drunk in the north, the midlands and in Wales, com- pared with only just over half of the alcohol drunk in London and the South East. Con- versely, wines and spirits accounted for a higher proportion of alcohol consumed in the south and east of England than in the rest of

the country. For example, respondents in London and the South East had drunk roughly twice as much wine per head during the reference week as had respondents in Wales and the West Midlands (Tables 2.4-2.6).

Consumption in relation to age and marital status

Among both men and women, consump- tion was highest among those aged 18-24 and decreased among the older age groups; those aged 18-24 drank on average about three times as much as those aged 65 and over. Those aged 16 and 17 had drunk appreciably less than those aged 18-24 (5.6 units com- pared with 14.3 units respectively). It was noted above that, overall, men drank much more than women: the smallest difference be- tween the sexes occurred among this youngest age group, where average consumption in the previous week was 6.5 units among the young men, and 4.6 units among the young women. This is not simply because, as noted above, more women than men in this age group had had a drink in the previous week - the con- sumption averages for recent drinkers show the same result.

Analysis of alcohol consumption in relation to marital status is complicated by the strong association between marital status and age, but the expected finding that consumption is higher among single than among married peo- ple persists even when age is held constant; in each age group single people drink more on average than do married people. However, consumption was highest of all among the relatively small numbers of divorced and separated respondents; men in this category had drunk 26.0 units in the previous week, and women 6.8 units, compared with the average of 14.5 and 4.8 units respectively for all men and women (Tables 2.7,2.8.

Consumption in relation to employment status and social class

Among both men and women, those who

129

were economically active (working or unemployed) drank more on average than those who were not. Although the highest consumption averages for both sexes were among the unemployed, because of the small number of unemployed people interviewed the differences compared with those who were in paid employment were not statistically sig- nificant.

Analysis of alcohol consumption in relation to social class is presented here in two ways. In the first, the social class of the head of the household is used as an indication of socioeconomic status of all members of that household. In the second, for informants who were in paid employment at the time of the survey, their own social class is used as an in- dividual characteristic deriving specifically from their occupation.

Among men, consumption did vary accor- ding to house hold social class. There was no clear pattern in the variation, but Table 2.10 shows that men in Social Class V (unskilled manual) households had somewhat higher consumption than average. Two groups of men, those in Social Classes I (professional) and IV (semi-skilled manual) households had lower than average consumption. Among women, apart from those in Social Class I households, whose consumption was about average, those in non-manual households tended to have higher than average consump- tion; for example, women in Social Class II households (managerial and junior profes- sional) had drunk an average of 6.1 units in the previous week, compared with 4.8 units for all women.

These results persisted when allowance was made for the difference in age composition be- tween social classes. For each social class, the table also shows the alcohol consumption measured by the survey as a percentage of the consumption that would be expected if the age distribution of respondents in that social class was the same as that of the whole sample.

130

A similar picture of alcohol consumption in relation to social class emerges when the anal- ysis is restricted to those who were in paid employment and based on their own social class rather than that of the head of the household. Unfortunately there are too few working women in Social Class I for the estimate of their consumption to be reliable (Tables 2.9-2.11).

Alcohol consumption levels

Although average alcohol consumption is a useful summary measure, it throws no light on the proportion of men and women drinking different amounts, and it has already been shown that the average conceals wide varia- tion in the amounts people drink. Table 2.12 shows the proportions of men and women drinking different amounts during the seven day reference period. As would be expected from the tables of average consumption presented earlier, men and younger respondents were more likely than women and older people to have drunk larger amounts (Table 2.12).

Sensible drinking levels In recent years there has been increased

publicity about the dangers of drinking too much, although advice to drinkers as to what constitutes sensible drinking has varied. Ex- perts do agree, however, that “safe” or “sensi- ble” levels are lower for women than for men. (Alcohol affects women more than men, part- ly because they generally weigh less, but also because their total body water is significantly less than that of men.)

Clearly, the amount that is safe to drink is bound to differ according to the particular hazardous consequences that are being con- sidered. The first report on drinking by the Royal College of Psychiatrists [5] concentra- ted on the physical consequences of excessive alcohol use, and suggested that consumption below about 50 units a week for men and 35 for women was associated with little increased risk either of physical or psychological depen-

dence on alcohol or of alcohol-related disease such as cyrrhosis of the liver. These levels are still generally considered appropriate in rela- tion to severe long term physical or psycho- logical damage [6].

Recently, however, consideration has been given to a wider range of social and physical consequences of drinking too much, such as alcohol-related violence and crime, and mild damage to the heart, liver, brain and immune system. Although it was agreed that lower levels of drinking were safe in relation to these types of damage, opinions of experts differed on the actual amounts. This caused some con- fusion, so in 1986, medical and health educa- tion bodies agreed to promote sensible maximum levels of 21 units a week for men, and 14 units for women [2].

In this report, consumption is classified into categories which are based on current thinking about sensible drinking levels as described above. It might be thought that given the under-reporting of alcohol con- sumption acknowledged earlier, analysis of survey data by suggested sensible drinking levels would be inappropriate. However, it should be noted that the sensible levels themselves are based on alcohol consumption reported to doctors, data which are probably subject to the same limitations as survey data.

In each category, the consumption levels for women are about 70% of those for men, and are intended to be roughly equivalent in physiological terms. For convenience, the categories are labelled as shown below.

Men Women

Light drinkers Moderate drinkers Fairly heavy drinkers Very heavy drinkers

l-21 units I-14 units 22-35 units 15-25 units 36-50 units 26-35 units 51 + units 36 + units

Because of the numbers of respondents in the “very heavy” drinking category, the two highest categories are often combined in the analysis and referred to as “heavy” drinkers.

It can be seen from Table 2.13 that overall, 6% of men and 1% of women had drunk more than the recommended maximum amounts of 50 units and 35 units respectively in the previous week. Those aged 18-24 were more likely than those in other age groups to have exceeded the safe levels - 11% of men and 4% of women in that age group had done so.

The table also shows that one in four men, and about one in twelve women, had drunk more than the much lower sensible levels of 2 1 units for men and 14 units for women. Again, the proportions exceeding these levels were highest among younger respondents; in the 18-24 age group, 38% of men and 15% of women had drunk more than the sensible levels.

Although people drinking more than these “sensible” amounts may suffer some long term damage to their health, they would not necessarily be aware of problems resulting from their drinking now. Although overall level of consumption is clearly one factor

131

which might indicate the presence of current drinking problems, other factors, such as per- sistent drunkenness, drinking and driving and physical and psychological dependence on al- cohol, are also important. Problem drinking, taking into account some of these other fac- tors, is discussed in more detail in the next chapter, together with patterns of drinking behaviour associated with high consumption (Table 2.13).

References

Marsh A, Dobbs J, White A: Adolescent Drinking Lon- don: HMSO, 1986. A Great and Growing Evil; The Medical Consequences of Alcohol Abuse. London: Royal College of Physicians, 1986. General Household Survey 1982. London: HMSO. 1984. Chap. 1 I. General Household Survey 1984. London: HMSO. 19X6, Chap. I I. Alcohol and Alcoholism. The report of a special commit- tee of the Royal College of Psychiatrists, Tavistock, 1979. Alcohol: Our Favourite Drug. Royal College of Psychiatrists, 1986.