An innovative methodological approach to analyzing alcohol use transitions in the context of prescription drug misuse and other health behaviors

点击翻译按钮获取中文摘要

ABSTRACT This proposal describes a plan to study what drives transitions in alcohol misuse in youth, and how those transitions coalesce with other health behavior transitions. Of particular concern is the recent increase in the prescribing, and misuse, of prescription opioids and sedatives, which has facilitated the emergence of a new crisis related to alcohol misuse: its contribution to overdose risk when used in combination with prescription opioids and/or sedatives. Prior research has identified individual-, peer-, parental-, and community-level correlates of alcohol misuse, but little empirical research focuses on what drives transitions in alcohol use behavior. Filling this gap has great potential to inform intervention and prevention design. Specifically, factors that facilitate transitions into more problematic drinking patterns are important targets for prevention, while those associated with sustained problematic drinking over time are targets for intervention. To address this knowledge gap, we will analyze data collected during the NIDA-funded (R01) Flint Youth Injury study, a longitudinal study of 600 drug-using youth recruited from an Emergency Department in Flint, Michigan. Using an innovative analytic technique, we will model behaviors over the follow-up period (roughly 6-month follow-ups for two years, for a total of five measurements) as continuous-time Markov Chains with covariate-modulated transition probabilities. In addition to the direct modeling of how covariates affect transition rates, a key advantage of this modeling framework is the elegant handling of missing time points and variation in the exact follow-up schedule. The first, and primary, specific aim of this project will be to estimate the effects of variables at multiple levels (individual, peer, parental, community) on transitions between alcohol use states; covariate effects will be tested for state-dependence to explicitly differentiate between targets for prevention vs. intervention. In the second aim, we will carry out analogous analyses for transitions in other behaviors – HIV-risk behaviors, weapon aggression, and non-medical prescription drug use – over the study period. In the third aim, we will use 90-day substance use timeline follow-back calendar data gathered at each follow-up to ascertain the number of overdose risk days (i.e. days with concurrent alcohol use and prescription drug use) for each participant, and to model, using generalized linear models, how alcohol use transitions during the previous follow-up period map onto frequency of overdose risk days. These works will provide information about a) key targets for intervention and prevention programs; b) how alcohol use transitions coalesce with other health transitions; and c) what types of alcohol use patterns are predictive of prescription drug overdose risk.