Premium
The collectivity of British alcohol consumption trends across different temporal processes: a quantile age–period–cohort analysis
Author(s) -
Holmes John,
Ally Abdallah K.,
Meier Petra S.,
Pryce Robert
Publication year - 2019
Publication title -
addiction
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.424
H-Index - 193
eISSN - 1360-0443
pISSN - 0965-2140
DOI - 10.1111/add.14754
Subject(s) - demography , medicine , quantile , per capita , confidence interval , cohort , population , quantile regression , cohort study , consumption (sociology) , cohort effect , gerontology , environmental health , statistics , mathematics , social science , sociology
Background and aims UK alcohol consumption per capita has fallen by 18% since 2004, while the alcohol‐specific death rate has risen by 6%. Inconsistent consumption trends across the population may explain this. Drawing on the theory of the collectivity of drinking cultures and age–period–cohort analyses, we tested whether consumption trends are consistent across lighter and heavier drinkers for three temporal processes: (i) the life‐course, (ii) calendar time and (iii) successive birth cohorts. Design Sex‐specific quantile age–period–cohort regressions using repeat cross‐sectional survey data. Setting Great Britain, 1984–2011. Participants Adult (18+) drinkers responding to 17 waves of the General Lifestyle Survey (total n = 175 986). Measurements Dependent variables: the 10th, 25th, 50th, 75th, 90th, 95th and 99th quantiles of the logged weekly alcohol consumption distribution (excluding abstainers). Independent variables: seven age groups (18–24, 25–34… 65–74, 75+ years), five time‐periods (1984–88… 2002–06, 2008–11) and 16 five‐year birth cohorts (1915–19… 1990–94). Additional control variables: ethnicity and UK country. Findings Within age, period and cohort trends, changes in consumption were not consistently in the same direction at different quantiles of the consumption distribution. When they were, the scale of change sometimes differed between quantiles. For example, between 1996‐2000 and 2008‐2011, consumption among women decreased by 18% [95% confidence interval (CI) = −32 to –2%] at the 10th quantile but increased by 11% (95% CI = 2–21%) at the median and by 28% (95% CI = 19–38%) at the 99th quantile, implying that consumption fell among lighter drinkers and rose among heavier drinkers. This type of polarized trend also occurred between 1984–88 and 1996–2000 for men and women. Age trends showed collectivity, but cohort trends showed a mixture of collectivity and polarization. Conclusions Countervailing alcohol consumption and alcohol‐related harm trends in the United Kingdom may be explained by lighter and heavier drinkers having different period and cohort trends, as well as by the presence of cohort trends that mean consumption may rise in some age groups while falling in others.