Prediction of alcohol-related casualties among emergency room admissions

Logistic regression models are used to determine the predictive value on injury status of drinking-in-the-event variables, quantity and frequency of alcohol consumption, alcohol-related problems, and demographic characteristics among 1,896 emergency room patients interviewed and breathalyzed at San Francisco General Hospital. Being young, White compared to Black, a positive admission breathalyzer reading, drinking prior to the event, and binge drinking were predictive of admission to the emergency room for an injury compared to a non-injury problem. A positive breathalyzer reading, more frequent usual drinking, blackouts, and a previous accident while drinking were predictive of drinking prior to the presenting injury event.