Missed and inconsistent classification of current drinkers: results from the 2005 US National Alcohol Survey

Abstract

This study compares current 12-month drinkers who do not report drinking in the last 30 days with current drinkers who drank in the last 30 days and assesses possible misclassification errors from use of a 30-day consumption measure. Data are from the 2005 US National Alcohol Survey (n = 6919), a national household probability survey. Telephone interviews were used to measure alcohol use and alcohol-related problems. This study compared 1300 current drinkers who did not drink in the last 30 days with 2956 current drinkers who drank in the last 30 days. Volume was measured by quantity/frequency scales (12-month and 30-day) and a graduated frequency scale (12-month). Both groups were compared by demographic, alcohol volume, days of five or more drinks, social consequences and dependence measures. Results indicate a significantly lower prevalence rate of current drinking for 30-day measures-47.3% (45.8%, 48.8%) versus 67.3% (66.0%, 68.7%) with 12-month measures. Further, 385 non-30-day drinkers reported 12-month drinking frequencies of once a month or more often, suggesting possible inconsistent reporting of their alcohol use. When this group of 'inconsistent' respondents is compared with the 915 non-30-day current drinkers who reported less than monthly drinking, they reported significantly higher yearly volume, days of five or more drinks, mean social consequences and proportion reporting alcohol dependence. In population surveys assessing alcohol use, asking about the previous 12 months rather than the past 30 days provides higher estimates of current use, including more days of heavy episodic use.