Patterns of alcohol use among Argentinean adolescents: A latent class analysis of their drinking behaviors.

PILATTI, ANGELINA; GODOY, JUAN CARLOS; BRUSSINO, SILVINA; PAUTASSI, RICARDO

Congreso; 2010 ISBRA World Congress; 2010

Introduction: Any attempt to describe and comprehend alcohol use and abuse among adolescents starts with a reliable measure that reflects accurately adolescents’ drinking patterns. One of the biggest challenges researchers in this area have faced is related to the procedures that led to a valid and consistent measure of drinking behaviors. In this context, latent class analyses (LCA) offers the possibility to resolve some of these difficulties. LCA is a statistical tool that has been used in the literature to find groups among the data. Specifically, LCA allows to reduce the whole set of data through obtaining information regarding how multiple observed measures are related, and which behavior is underneath these relationships. This method starts with the assumption that co variation among a particular set of manifest variables is attributable to the association these variables exhibit with a latent variable, for instance, drinking pattern of alcohol use. Then, the goal is to find groups of people that show similar characteristics along the set of manifest variables used in the model. This method improves the traditional ones because LCA tolerates the simultaneously use of a larger number of variables than the typical questions regarding frequency and quantity of alcohol use. In this context, the present work was conducted in order to identify different drinking patterns among Argentinean adolescents. Method: Participants were 264 adolescents from 13 to 18-years old (56.1% female). Latent class analysis was performed with data from 217 (57.6% female) actual drinkers. The following set of observable variables, frequency of alcohol use, number of standard drinks, frequency of drinking until drunk, and probability of alcohol intake during the next week was used to recognize the patterns of alcohol use. Fit of the different models was evaluated through recommended indices (BIC, AIC, and AIC3).Results: According to the mention criteria, the three-class model was selected, and each of the three outcome classes was labeled as follow: social drinkers, binge drinkers, and heavy drinkers. Each category differs from the others as a function of the probability for various responses on each observed variable. A fourth class corresponding to those who reported they didn’t drink alcohol (47 participants) was added and labeled abstainers. Then, analyses of variance (ANOVA) was performed with the purpose of investigate differences concerning age and peer alcohol use among the latent classes. These results showed those adolescents who belonged to the social drinking category were younger and also, they reported to have fewer female and male friends that drink alcohol than those adolescents who were binge and heavy drinkers. Discussion: this procedure permits not only to achieve insight of the shared characteristics adolescents within a particular drinking pattern have, but also to characterize their drinking patterns in a more accurate way. Indeed, one of the most valuable characteristics of this type of analyses is the possibility of considering together a larger number of indices than the classical quantity and frequency questions, which allow to get a more complete and precise classification of drinking behaviors.