A comparative analysis of different methods for obtaining estimates of alcohol consumption in a Danish population...

The aim of the present study was to compare different measures of alcohol consumption used in a Danish survey.

A stratified random sample was extracted from the Civil Registration System. From January 1997 to January 1998 approximately 8 telephone interviews were conducted every day, including Sundays and holidays, ending with a total of 3,050 interviews after 1 year. Two main approaches to measure alcohol consumption were used: the quantity-frequency and the recent occasion approaches; the latter is subdivided into previous week and previous day approaches.

The overall estimated number of units (= 12 g of pure alcohol) per week was 6.8 (95% confidence intervals (CI): 6.5-7.1), 6.7 (95% CI: 6.4-7.1) and 8.5 (95% CI: 7.8-9.1) for the quantity-frequency, previous week and previous day approaches, respectively. A total of 50% of the men and 70% of the women did not drink alcohol the previous day. Among people classified as high consumers in the previous week and previous day approaches, less than 60% and 30%, respectively, were similarly classified in the quantity-frequency approach.

There was agreement on the level of alcohol consumption between the quantity-frequency and previous week approaches, but higher estimates when using the previous day approach. The previous day approach varied more in relation to the interview day and season compared with the quantity-frequency approach and the previous week approach. The recent occasions approach showed some difficulties in classifying the individuals. If the alcohol consumption is included in a model as a risk indicator or a confounder, the quantity-frequency approach would be more preferable than the recent occasion approach.