Alcohol and injury: Multi-level analysis from the emergency room collaborative alcohol analysis project (ERCAAP)

Aims: To analyze the relationship between individual-level characteristics and site-level contextual variables on the association of acute alcohol use and injury. Methods: Blood alcohol concentration (BAC) and survey data collected (using similar methodology and instruments) at the time of the emergency department (ED) visit, between 1985 and 2003 on probability samples of injured and non-injured patients (n = 18 438) from 31 EDs in seven countries (Argentina, Canada, Italy, Mexico, Poland, Spain, USA) were analyzed using hierarchical linear modeling (HLM). Results: BAC and self-reported consumption were predictive of an injury (compared to a non-injury), controlling for gender and age, with odds ratios of 1.51 and 1.58, respectively. The likelihood of injury given a positive BAC and self-report was less for heavier drinkers (those reporting five or more drinks on an occasion) than for lighter drinkers, and was greater in those societies with greater detrimental drinking patterns than those with lower detrimental patterns. Conclusions: These data suggest a moderate, but robust association of a positive BAC and self-report with admission to the ED for an injury, which is modified by the patient's usual heavier drinking and by societal drinking patterns.