Evaluating the American Stop Smoking Intervention Study (ASSIST) and other tobacco control efforts across the U.S. is a complex endeavor that poses many challenges. Parsimonious models are needed to examine state-level tobacco-control outcomes that are potentially more powerful and flexible than existing models because data have not been available in a consistent fashion across states or over time. Developing composite indices are essential to represent environmental factors which may influence prevalence and consumption rates. This presentation will describe the methods that have been used to create such indices and demonstrate their use. For example, z-scores were used to create an “Initial Outcome” index and principal components analysis was used to create a “ Strength of Tobacco Control” index. We will also discuss alternative methods of expressing change in outcomes and the pros and cons of each method. These include a direct change measure (Time2 – Time1), relative change, and Time2 with regression adjustment for Time1. We also discuss approaches to adjusting these measures for confounding, as states differ from each other on many demographic and economic factors, and ignoring such differences could result in biased estimates. The methods we have developed for ASSIST have served both to identify the problems inherent in making state-level comparisons and to develop solutions that will have application in the next generation of tobacco control programs.
Learning Objectives: Participants will learn some analytical and statistical methods for evaluating state level tobacco control efforts across the U.S. These methods are generalizable to other settings (e.g., communities) and other topics or exposures where nonrandomized interventions and ongoing evaluations are conducted
Keywords: Tobacco Control, Methodology
Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I have a significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.
Relationship: Employee of NCI