Aims To describe and model the sources of the variation and trends in the meaning of subjective drunkenness.
Design Trend analyses of three cross-sectional surveys.
Setting US general population.
Participants Those who report being drunk in the past year among those in the 1979, 1995 and 2000 National Alcohol Surveys.
Measurements Number of drinks reported to feel drunk (dependent variable), past-year alcohol consumption measures, beverage preference, state drunk driving blood alcohol concentration (BAC) limit and demographics.
Findings The mean reported number of drinks to feel drunk declined significantly between each survey and was significantly lower for women. Considerable variation was also found within surveys and was explained partially by available variables. Volume of alcohol and heavy drinking occasions were associated positively with the number of drinks to feel drunk. Higher educational attainment was associated negatively as was being a wine drinker, of older age, of African American ethnicity and of becoming drunk more frequently than once per month. Living in a state with a per se BAC limit of 0.08% was associated negatively in models for men.
Conclusions A substantial shift downward in the meaning of drunkenness occurred in the US between 1979 and 2000. This may be explained partly by the increase in educational attainment, the ageing of the population, the decline in per capita alcohol consumption and changes in alcohol policy towards lower BAC limits for drunk driving along with greater penalties, enforcement and awareness.