Measuring alcohol consumption

摘要

This article is an overview of different approaches to measuring alcohol consumption: self-reports and objective measures such as blood alcohol concentration (BAC) and aggregate level measures. These approaches are evaluated as regards their ability to capture quantity, frequency, volume and variability of drinking. This review focuses on self-report measures and on the current knowledge of undercoverage error when compared with sales data. In the comparative evaluation of measures, two analytical aims are examined: a) description and testing of differences across groups for which ordinal information is sufficient and b) establishment of cutoff points and risk relationships for which unbiased interval scale level is required. First, minimal differences were found between self-report measures when the recall period was sufficiently long enough. Second, prospective diaries appear to be stronger measures than retrospective recalls. However, prospective diaries commonly cover only short reporting periods and should be combined with simple retrospective measures to capture rare and infrequent drinking episodes. In regard to undercoverage, the discrepancy cannot be fully explained by non-response or concealment of consumption by drinkers. It is argued that undercoverage of sales data may be more related to sample frame defects-e.g., the non-inclusion of particular subpopulations such as the homeless or institutionalized.