OBJECTIVE: Heavy drinking among college students continues to be a substantial problem on campuses across the United States. Attempts to predict these drinking events have been restricted to assessments of the correlates of heavy drinking (measured at 4 or 5 drinks) and have not examined the peak drinking levels that can be fatal to students. This article presents a theoretical analysis of college drinking patterns that provides a basis for estimating peak drinking levels and predicts future risks related to peak drinking events. METHOD: Survey data were collected on sociodemographics and drinking patterns of 2,102 college students from two college campuses in California. A mathematical model of drinking patterns was used to characterize the stochastic distribution of drinking events among 1,273 students who drank five or more times and consumed more than one drink on some occasion since the beginning of the school year. An application of extreme value theory enabled the estimation of peak drinking levels for every college drinker. These estimates were related to self-reported maximum drinking levels and sociodemographic characteristics of respondents. RESULTS: Among these drinkers, the distribution of self-reported maximum drinking levels ranged from 2 to 43 drinks per occasion. Estimated peak drinking levels ranged from 3 to 49. Maximum drinking levels were well characterized by peak drinking estimates (R2 = 0.503). Variations in peak drinking levels were large and specifically related to particular sociodemographic groups (i.e., white male freshmen). CONCLUSIONS: The theoretical model of peak drinking events effectively characterizes maximum drinking levels among college students. High levels of peak drinking are to be expected among specific sociodemographic subgroups. These risks can be assessed on an individual basis. At the population level, risks for harm related to peak drinking events are predictable.
OBJECTIVE: Heavy drinking among college students continues to be a substantial problem on campuses across the United States. Attempts to predict these drinking events have been restricted to assessments of the correlates of heavy drinking (measured at 4 or 5 drinks) and have not examined the peak drinking levels that can be fatal to students. This article presents a theoretical analysis of college drinking patterns that provides a basis for estimating peak drinking levels and predicts future risks related to peak drinking events. METHOD: Survey data were collected on sociodemographics and drinking patterns of 2,102 college students from two college campuses in California. A mathematical model of drinking patterns was used to characterize the stochastic distribution of drinking events among 1,273 students who drank five or more times and consumed more than one drink on some occasion since the beginning of the school year. An application of extreme value theory enabled the estimation of peak drinking levels for every college drinker. These estimates were related to self-reported maximum drinking levels and sociodemographic characteristics of respondents. RESULTS: Among these drinkers, the distribution of self-reported maximum drinking levels ranged from 2 to 43 drinks per occasion. Estimated peak drinking levels ranged from 3 to 49. Maximum drinking levels were well characterized by peak drinking estimates (R2 = 0.503). Variations in peak drinking levels were large and specifically related to particular sociodemographic groups (i.e., white male freshmen). CONCLUSIONS: The theoretical model of peak drinking events effectively characterizes maximum drinking levels among college students. High levels of peak drinking are to be expected among specific sociodemographic subgroups. These risks can be assessed on an individual basis. At the population level, risks for harm related to peak drinking events are predictable.
Authors: Dennis L Thombs; R Scott Olds; Susan J Bondy; Janice Winchell; Dolly Baliunas; Jürgen Rehm Journal: J Stud Alcohol Drugs Date: 2009-09 Impact factor: 2.582