AIMS
This study aims to explore the feasibility of using routine hospital discharge data, at the level of countries within Europe, to estimate the general population prevalence of alcohol dependence (AD).

METHODS
We utilised the European Core Health Indicators data tool to extract the annual rate of hospital discharges due to any wholly attributable alcohol condition as defined by the ICD-10. For those counties with data available, we systematically searched Medline, EMBASE, PsychINFO and Google for studies reporting an estimate of the prevalence of AD from national cross-sectional surveys. We compared these prevalence estimates with those developed from prediction models based on hospital discharge data.

RESULTS
The rate of hospital discharges due to any condition from the F10 diagnostic category (mental and behavioural disorders due to alcohol) was moderately correlated with AD prevalence (r = 0.56), while the rate due to any condition from the K70 diagnostic category (alcoholic liver disease) was weakly correlated with AD prevalence (r = 0.21). Two-thirds of the estimates from cross-sectional surveys were not significantly different to those generated using the F10 discharge rate prediction model.

CONCLUSIONS
Country-level AD prevalence estimates generated using annual F10 hospital discharge rates are likely to provide information of some utility, particularly when limited other sources of information are available or when examining relative trends over time or between regions. There is, however, currently insufficient evidence to make a definitive recommendation to use hospital discharge data to estimate the absolute prevalence of AD per country in Europe.