Abstract: Predictors of Substance Use Latent Classes Among American Indian Youth on Reservations (Society for Prevention Research 26th Annual Meeting)

284 Predictors of Substance Use Latent Classes Among American Indian Youth on Reservations

Schedule:
Thursday, May 31, 2018
Congressional C (Hyatt Regency Washington, Washington, DC)
* noted as presenting author
Randall Craig Swaim, PhD, Senior Research Scientist, Colorado State University, Fort Collins, CO
Linda Stanley, PhD, Senior Research Scientist, Colorado State University, Fort Collins, CO
Introduction: Identical latent classes of substance use among reservation-dwelling American Indian (AI) adolescents were identified by Stanley and Swaim (2017) across middle and high school (MS/HS) youth, raising concerns that substance-using groups are established early in this population. It is not known, however, whether similar factors predict class membership across grade level. This paper examines a wide range of substance-related predictors of class membership. Variations across grade level may help identify effective targets for prevention for younger and older AI adolescents.

Methods: 1503 MS and 1995 HS students from 46 schools on or near reservations completed substance use surveys during the 2009-2013 school years. Potential latent class predictors included peer use, substance availability, perceived harm, family attitudes toward use, family communication about use, high risk substance behaviors, and early initiation. Predictor variables were included as auxiliary variables in four-class models as identified by Stanley and Swaim (2017). A 3-step approach was used to assess each auxiliary variable as described by Vermont (2010). For both age groups, the three substance-using classes were compared to the non-user class.

Results: Four latent class of use were identified: 1) nonusers, 2) marijuana & cigarettes, 3) alcohol, marijuana, & cigarettes, and 4) polysubstance users. Predictors differentiated between substance-using and nonuser class membership, but some were better predictors than others, and differences were found across grade level. Among MS students, the best predictors of user status were peer alcohol use, using substances when alone, and early initiation of alcohol and marijuana. Substance availability, family attitudes toward use, and family communication about use did not predict class membership. Among HS students, peer alcohol and marijuana use, substance availability, perceived harm of alcohol use, family attitudes toward use, using substances when alone, and early initiation of alcohol were the best predictors of user status.

Conclusions: While similar latent classes of substance use were found across MS and HS American Indian students, there was variation in the types of predictors that differentiated user and nonuser status. While peer use, using alone, and early initiation were predictive for both groups, lack of perceived availability, perceived harm, and family sanctions against use were protective against using among HS students. MS AI students establish substance-using groups similar to their HS peers, but differing predictors suggest that family attitudes and perceived harm may be important targets for prevention efforts to avert use among HS-age youth in this group.