Alcohol-attributable fraction of injury morbidities: A novel method combining emergency room injuries and population alcohol exposure data

The association of alcohol consumption with injury mortality and morbidity is unique as the injuries are generally caused by acute drinking and the effect from alcohol is transient, in comparison to the accumulative effect of alcohol found in chronic diseases such as liver cirrhosis. As a result, the alcohol-attributable fraction (AAF) of injury has been traditionally estimated using the direct method, combining the proportion of alcohol exposure among individuals in a case (e.g., injured) sample and the relative risk of injury related to acute drinking, an approach officially used by the US Center for Disease Control. This approach, however, relies heavily on the validity of the case sample exposure estimates. While the ER is an ideal place to collect injury cases, those attending a specific ER for an injury cannot be considered to be representative of the total injured population. A novel method to estimate AAF of injuries is presented here to address the representativeness issue of the ER data. Probability of acute alcohol consumption is first estimated using injury patients' usual drinking pattern. Population drinking distribution data are then used to re-estimate the acute alcohol exposure prevalence for AAF estimation. Using the ER data from 20 countries as well as their respective general population surveys, we estimate the AAF of injury morbidity and make comparisons to estimates using the old method without adjusting for population exposures.