Finding the Missing Units: Identifying Under-Reporting of Alcohol Consumption in England

Background: Under-reporting of alcohol consumption is a widespread and persistent problem in surveys worldwide. For Great Britain in 2010, 56% of alcohol sold was reported consumed. Under-reporting occurs for reasons, including, but not limited to: selective reporting, mode effects, recall bias, and under-estimation. Methods: Secondary analyses of Health Survey for England (HSE) 2008 (n=9,608) and General Lifestyle Survey (GLF) 2008 data (n=12,490) to estimate population level alcohol consumption with under-reporting taken into account. Secondary analyses of the HSE 2011 (n=3,774) to highlight possible risk factors for under-reporting of alcohol consumption in retrospective interview compared with prospective drinking diary. Qualitative semi-structured interviews (n=10) with HSE 2011 drinking diary participants to identify further potential risk factors. Self-completion questionnaire and pouring task with a convenience sample (n=283) to explore under-estimation of home drinking. Results: After accounting for under-reporting, 40% men and 30% women drank above weekly guidelines, an increase of over 10% points (GLF 2008). On the heaviest drinking day in the last week 75% men and 80% women drank above daily limits, compared with around half in the original survey (HSE 2008). Risk factors for prospective diary measures exceeding those of the retrospective interview were: weekly alcohol intake, number of drinking days, drinking a combination of drink types, and drinking exclusively in licensed premises (HSE 2011). Qualitative interviews identified having a non-routine drinking pattern and not using alcohol units to track consumption as linked to drinking more in the prospective diary than expected. The pouring task did not identify systematic under-estimation of a usual glass; however under-estimation was associated with increasing volume poured. Conclusions: Under-reporting of alcohol consumption has implications for public health research and policy. In three studies, alcohol-related factors were associated with under-reporting of alcohol consumption whereas demographic and social factors were not. Targeted interventions and policies may reduce under-reporting.