The 130th Annual Meeting of APHA

Abstract #44935

Vincent Chen, PhD, Jean L. Forster, PhD, and Traci L. Toomey, MPH, PhD. Division of Epidemiology, University of Minnesota, 1300 South Second St., Suite 300, Minneapolis, MN 55454-1015, 612-625-4567, chen_v@epi.umn.edu

Minnesota statewide comprehensive tobacco-use prevention programs include a state-level intervention component, targeting the entire youth population in the state such as media campaigns, and state tobacco policies. These efforts will result in a time specific effect on tobacco-related behaviors beyond the effects of local programs. Each month, 600 Minnesota adolescents and 100 of their counterparts from other Midwest region are being surveyed based on our sampling plan to establish two representative monthly time-series of smoking prevalence over three years. This is a time series quasi-experimental design, which uses Minnesota as the intervention group and the upper Midwest region as the comparison group. In such state-level time series analyses, Box-Jenkins time-series analysis is employed to analyze the statewide blanket effects of specific statewide events and policies on adolescent smoking prevalence. Box-Jenkins is appropriate for three major reasons. First, it has the capacity to represent wide ranges of data generating processes with a parsimonious model. Second, it accommodates modeling in the presence of external events or multiple exogenous stochastic variables (i.e. transfer function models). Third, a well developed procedure for modeling has been developed. The estimation process for the analyses uses the SCA System to conduct intervention time-series analysis with variability in the dependent time series due to trend, seasonality, and other autocorrelation patterns. The presentation will examine the trend and difference between MN and Midwest control based on eighteen monthly observations with daily smoking prevalence as dependent variable, and media campaign coverage and expenditures, state smoking-related policies, etc. as predictors.

Learning Objectives:

• To describe state level predictors of adolescent tobacco use and introduce proper use of time series modeling in explaining the variation of smoking across states.

Keywords: Adolescents, Tobacco

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.