• Abstention;
  • age–period–cohort modelling;
  • alcohol;
  • time–series;
  • trends


Background and aims

British alcohol consumption and abstinence rates have increased substantially in the last 3 decades. This study aims to disentangle age, period and birth cohort effects to improve our understanding of these trends and suggest groups for targeted interventions to reduce resultant harms.


Age, period, cohort analysis of repeated cross-sectional surveys using separate logistic and negative binomial models for each gender.


Great Britain 1984–2009.


Annual nationally representative samples of approximately 20 000 adults (16+) within 13 000 households.


Age (eight groups: 16–17 to 75+ years), period (six groups: 1980–84 to 2005–09) and birth cohorts (19 groups: 1900–04 to 1990–94). Outcome measures were abstinence and average weekly alcohol consumption. Controls were income, education, ethnicity and country.


After accounting for period and cohort trends, 18–24-year-olds have the highest consumption levels (incident rate ratio = 1.18–1.15) and lower abstention rates (odds ratio = 0.67–0.87). Consumption generally decreases and abstention rates increase in later life. Until recently, successive birth cohorts' consumption levels were also increasing. However, for those born post-1985, abstention rates are increasing and male consumption is falling relative to preceding cohorts. In contrast, female drinking behaviours have polarized over the study period, with increasing abstention rates accompanying increases in drinkers' consumption levels.


Rising female consumption of alcohol and progression of higher-consuming birth cohorts through the life course are key drivers of increased per capita alcohol consumption in the United Kingdom. Recent declines in alcohol consumption appear to be attributable to reduced consumption and increased abstinence rates among the most recent birth cohorts, especially males, and general increased rates of abstention across the study period.