A comparison of different approaches to measure alcohol consumption

1. A comparison of different approaches to measure alcohol consumption 26 April 2006Ola Ekholm, Karina Christensen, Katrine Strandberg Larsen and Morten Grønbæk National Institute of Public Health

2. Background • Self-reported alcohol intake • Possible to link mortality and morbidity with drinking behaviour on an individual level • The recommended guidelines for alcohol intake are, to a great extent, based on epidemiological studies that have used self-reported information

3. More specific questions result in higher reported alcohol intake Beverage-specific questions Large time-variation in drinking

4. Aim of the study • To compare the reported intake obtained by beverage-specific questions and to compare how different reference periods and response formats influence the self-reported alcohol intake.

5. Danish Health Interview Survey 2003 Inclusion criteria: Danish citizen aged 18 years or more.

6. Characteristics of the respondents in each sample by sex, age and marital status. Percentage

7. Question included in all four samples • How many alcoholic drinks did you have during last weekday? (don’t include Friday, it is considered part of the weekend) • No. of drinks • Beer __ • Strong beer __ • Red & white wine __ • Liqueurs __ • Spirits __ • Ready to drink products __

8. 1 drink = 12 grams of alcohol

9. Alcohol intake on the most recent weekday

10. Four different assessment methods for alcohol consumption All questions were beverage-specific (beer; strong beer; red & white wine; liqueurs; spirits; ready to drink products)

11. The sex- and age-adjusted weekly mean number of alcoholic drinks

12. The age-adjusted weekly mean number of alcoholic drinks. Men

13. The age-adjusted weekly mean number of alcoholic drinks. Women

14. The Danish National Board of Health’s sensible drinking limits • Men: a maximum of 21 drinks per week • Women: a maximum of 14 drinks per week

15. Results from a multivariate logistic regression analysis showing the association between a high reported weekly alcohol intake and the different assessment methods

16. Conclusion A question concerning the intake in a typical week does not seem to underestimate the alcohol intake compared to the 7-day recall measure Estimating the intake for a typical week will take the large time-variation in drinking into account The average weekly intake measure was less specific than the other two measures and the lower reported intake was therefore as expected.

17. Questions concerning binge drinking

18. Close-ended response categories Never Less than once a month Approximately 1-3 times a month Approximately once a week More than once a week

19. Binged at least once a week

20. Binged at least once a month

21. Conslusion The open-ended response format is not advisable when the reference period is long The analyses showed that the reference period was important for the self-reported intake for the open-ended questions. However, this was not the case for the close-ended questions

22. Summary A question concerning typical alcohol intake is feasible to use in epidemiological studies It is more appropriate to use close-ended questions compared to open-ended questions when the reference period is long The length of the reference period seems to be ignorable for the self-reported frequency of binge occasions when using close-ended questions