RT Journal Article SR Electronic T1 PS06 Finding the Missing Units: Identifying Under-Reporting of Alcohol Consumption in England JF Journal of Epidemiology and Community Health JO J Epidemiol Community Health FD BMJ Publishing Group Ltd SP A41 OP A41 DO 10.1136/jech-2012-201753.105 VO 66 IS Suppl 1 A1 Boniface, S YR 2012 UL http://jech.bmj.com/content/66/Suppl_1/A41.1.abstract AB Background Alcohol sales coverage (reported consumption as a percentage of total sales) is typically 40–60%. This study explores how accidental under-estimation of alcohol in drinks poured at home might contribute to low sales coverage. Methods 283 drinking adults (16+) completed a face-to-face survey comprising a questionnaire and pouring task on 12 occasions at six sites in and near London in summer 2011. Actual and perceived number of units poured in a ‘usual glass’ of wines and spirits is explored by demographic and social factors. Results There were 283 participants (54% women) who completed the questionnaire and pouring task. The 283 participants poured a total of 465 glasses of wine and spirits. 52% wine estimates and 42% spirit estimates were within 0.5 units of their actual value. More participants over-estimated the amount of alcohol they had poured than under-estimated. Multinomial logistic regression found significant risk factors for under-estimating wine were increasing volume of wine poured (RRR 1.02 95% CI 1.01–1.02, P<0.0001), belonging to a non-white ethnicity (RRR 3.88, 1.65–9.16, P=0.002), and being unemployed or retired (RRR 4.30, 1.08–17.07, P=0.038). Only the volume of spirits poured was a significant predictor of under-estimating spirits (RRR 1.04, 1.01–1.06, P=0.003). For both wines and spirits, not having a degree was significantly associated with an increased risk of over-estimating the amount of alcohol poured. Conclusion As participants were more likely to over-estimate than under-estimate, this survey is not able to explain low sales coverage as a product of accidental under-estimation of the amount of alcohol poured drinks drunk at home. A future qualitative study will explore how recall accuracy may contribute to low sales coverage.