Effect of drink alcohol content adjustments to alcohol volume on the risk for alcohol problems

Studies in the US and elsewhere have determined that standard drink assumptions made in the interpretation of survey response are typically inaccurate for individuals and on average. Questions included in the 2000 and 2005 National Alcohol Surveys (NAS), along with detailed information from methodologic follow-up studies to these surveys and from a study of on-premise drinks, enable individual estimates of drink alcohol content. These estimates are then used to adjust alcohol consumption volume variables in each survey. Because drink alcohol content estimates are on average larger than the US standard drink these adjustments increase the average volume of consumption and improve coverage of sales data by each survey. To determine whether these adjustments also improve the prediction of alcohol-related consequences and alcohol dependence symptoms, the present study estimates parallel models of these outcome measures using un-adjusted and adjusted alcohol volume variables in the 2000 and 2005 NAS surveys. These surveys include 8,650 current drinkers who are the focus of these analyses. Results indicate better model fit in terms of the R-squared statistic when adjusted volume is used. Additional models with a variable representing the difference in volume after adjustment find that this is a positive and significant predictor, controlling for unadjusted volume. These results indicate the importance of drink alcohol content estimates for surveys. While the improvements in models fit are small, and do not indicate a major limitation for surveys that do not include these adjustments, they represent an additional refinement contributing to more accurate survey-based alcohol consumption estimates.