USING TEMPORAL VARIATION IN RISK BEHAVIOR TO UNDERSTAND TRENDS IN ADOLESCENT ALCOHOL MISUSE

Washington University, Saint Louis MO

Investigators

Linked publications, trials & patents

Abstract

Project Summary Reductions in the prevalence of multiple adolescent risk behaviors such as binge drinking, use of most illicit drugs, early sexual debut, crime, and violence have occurred in recent decades. It is unknown whether these are separate trends or if they reflect a broader reduction in the propensity to engage in risk behavior. This is an essential question because the potential set of causes for behavior-specific trends is very different from those that might explain a unitary trend related to risk-behavior propensity. Separate trends imply separate, behavior- specific causes (e.g. alcohol policies, sexual health education). But if these concurrent trends reflect a reduction in risk for multiple outcomes, the explanation likely involves individual-level factors such as predisposition to externalizing behaviors. Using 25 years of data from 3 national monitoring surveys, this project will be the first to address this question as it will be the first epidemiological study to examine multiple trends as a behavioral syndrome rather than as separate phenomena. This will involve methodological innovation as it requires multivariate methods that are not typically used in epidemiological trend studies. Informed by Problem Behavior Theory, the overarching purpose of the proposed project is to identify whether reductions in risk behaviors reflect reductions in an underlying trait or ?Risk Behavior Syndrome.? The project is significant because characterizing the phenomena is a first step toward understanding its causes. We will also examine changes in theoretically relevant psychosocial risk and protective factors that correlate with trends in risk behaviors. This will shift the focus of our research question from ?What?? to ?Why?? A broader understanding of individual-level risk and protective factors associated these trends can help refine existing hypotheses about causal factors and may lead to new hypotheses. We will utilize Item Response Theory modeling and tests of measurement invariance over time to examine the degree to which trends in risk behaviors reflect trends in an underlying Risk Behavior Syndrome. We will construct factors related to 4 canonical domains of risk behavior: alcohol and drug use, delinquency, early substance initiation, and sexual behaviors. We will model those as a function of a higher-order factor and incorporate risk and protective factors into the model. This will be the first-ever study of population trends in multiple risk behaviors and the relations among them. We hypothesize that the trends reflect changes in liability for Risk Behavior Syndrome and that this broader trend accounts for much of the decline in binge drinking. Successful accomplishment of these aims will lead to a more comprehensive understanding of secular trends of interest to multiple disciplines and identify salient risk and protective factors. By developing an integrated model of change, we will establish a firm scientific groundwork for formal studies of causality.