Multi-level analysis of alcohol-related injury, societal drinking pattern and alcohol control policy: emergency department data from 28 countries.

Background And Aims: Societal-level volume and pattern of drinking and alcohol control policy have received little attention in the alcohol and injury literature. The aim of this study was to estimate the association between alcohol-related injury, individual-level drinking variables, country-level detrimental drinking pattern and alcohol policy.

Design: Probability samples of emergency department (ED) patients from the International Collaborative Alcohol and Injury Study (ICAIS), which includes four collaborative ED studies on alcohol and injury, all using a similar methodology, were analyzed with multi-level modeling of individual-level drinking variables and aggregate-level variables (country drinking pattern and alcohol policy) on alcohol-related injury in 33 ED studies.

Setting: Sixty-two emergency departments in 28 countries covering five regions.

Participants: A total of 14 390 injured patients arriving to the ED within 6 hours following injury.

Measures: Alcohol-related injuries (self-reported drinking prior to the event and causal attribution of injury to drinking) were analyzed in relation to individual-level volume and pattern of drinking, study-level alcohol volume, country detrimental drinking pattern (DDP) and an alcohol policy measure, the International Alcohol Policy and Injury Index (IAPII). The IAPII includes four regulatory domains: availability, vehicular, advertising and drinking context.

Findings: Controlling for demographic characteristics, individual-level drinking and study-level volume, the IAPII was associated significantly with the proportion of both self-reported drinking [confidence interval (CI) = 0.97-0.99; P < 0.001] and causal attribution (CI = 0.97-0.99; P < 0.01) and DDP had little effect on these associations. All four domains were significantly predictive of self-reported drinking [availability (CI = 0.93-0.98, P < 0.01); vehicular (CI = 0.91-0.97, P < 0.001); advertising CI = 0.82-0.94, P < 0.01); and context (CI = 0.93-0.99, P < 0.01], while only the vehicular domain was significantly predictive of causal attribution (CI = 0.92-0.99; P < 0.05).

Conclusions: The more restrictive the alcohol policy in a country, the lower the rate of alcohol-related injury, with country-level drinking pattern having little effect on this relationship.