Alcohol Consumption across the Lifecourse
One of the strengths of the British Birth cohort studies is the ability to examine
trajectories of drinking behaviour over the lifecourse. As Jefferis et al (2007) have
shown the prevalence of non drinking remains similar throughout adulthood for both
men and women, while the prevalence of binge drinking declines somewhat between
early adulthood (age 23) and mid-adulthood (age 42). Whereas this previous paper
focused only on non-drinking and binge drinking, here we present data on reported
levels of weekly alcohol consumption from age 23 to age 46. The mean number of
units of alcohol consumed by members of the 1958 cohort at age 23, 33, 42 and 46
are therefore shown in Table 6a below. This suggests that for both men and women
alcohol consumption declines between age 23 and 33 but then increases at age 42
and declines again at age 46. However, these results should be interpreted with care.
As was discussed above, in 2004 a telephone interview rather than a face-to-face
interview was used for the first time to collect data on alcohol consumption from the
1958 birth cohort. In order to minimise the length of the telephone interview, cohort
members were not asked separately about the amounts of different types of alcohol
they had consumed in the previous week but were simply asked for an overall
summary of their alcohol consumption. As the figures below suggest, this seems to
have resulted in a sizeable under-reporting of the amount of alcohol consumed.
In addition, the figures for average weekly levels of reported alcohol consumption
appear surprisingly high at age 42 in 2000, particularly for men. Further investigation
suggested that there was a strong possibility that beer consumption had been
recorded inconsistently by interviewers in the 2000 survey. This was due to some
ambiguity in the instructions in the computer aided personal Interview schedule. This
is likely to have resulted in some interviewers recording beer consumption in pints of
beer and the majority recording consumption in terms of units of beer (i.e. half-pints).
(The data is labelled as measured in pints of beer and this figure is therefore doubled
before adding it to reported consumption of other types of alcohol).This would have a
greater impact on men’s reported alcohol consumption than on women’s alcohol
consumption as we have seen above that a much greater proportion of men’s alcohol
consumption is beer whereas women are more likely to report drinking wine. In order
to investigate this further, Table 6b provides the data on non-beer alcohol
consumption at age 23, 33 and 42 and Table 6c and 6d provide data on reported
alcohol consumption for the separate sweeps of the 1970 cohort. The data for the
1970 cohort in particular is suggestive of a problem with recording beer drinking in the
2000 sweep (and note that exactly the same interview protocol was used for both the
1958 and 1970 cohort in 2000). Appendix B presents the frequencies of beer drinking
for men in the 1958 cohort at age 33 and 42 and for men in the 1970 cohort at age 30
and 34, which provides further evidence that beer drinking in 2000 has been inflated
due to an ambiguity in the interview protocol. Further work is being carried out to
discover whether specific interviewers can be identified who have recorded beer
consumption in units rather than pints so that the data can be corrected or the values
which are most likely to be incorrect can be flagged.
18