Vital Statistics Consulting Vital Statistics Consulting

Case Study: Health Policy Evaluation – Graphic Cigarette Warning Labels

Our CEO and President, Drs. Jessica Steier and Bill Gallo, partnered on an evaluation of The Family Smoking Prevention and Tobacco Control Act requiring the FDA to include new graphic warning labels on cigarette packages. Implementation of graphic labels in the US has been delayed indefinitely due to ongoing litigation, however, they have been in place internationally for over a decade. Using data from the International Tobacco Control (ITC) Project, we measured the impact of graphic cigarette warning labels on youth smoking intentions and adult smoking behaviors in Southeast Asia, specifically Thailand and Malaysia, based on data availability.

Research Question Generation We hypothesized that graphic labels would have a greater impact among youth, as visual communication of risk may be more effective among young smokers and would-be-smokers. Teens are more likely to systematically undervalue the future and long-term consequences of their behavior, making them appropriate targets for intervention. Further, level of addiction is expected to be lower among youth smokers who typically smoke less on average as compared to adults, making it easier to break the habit. We generated several research questions which were mapped to outcomes and variables available in the datatset.

Evaluation Design Plan We analyzed our data using two approaches. First, we utilized a quasi-experimental design capturing the time period after which Thailand had enacted the policy, but prior to policy implementation in Malaysia. Using this approach, outcomes were assessed for our Thai sample (as a proxy for the policy) using the Malaysian sample as a control, whereby country differences reflected the effects of the graphic warning label policy. The Malaysian sample was selected as a comparison group for the Thai sample because the two countries have important similarities such as per capita GDP and geographical proximity. Second, we limited the data to the post-policy implementation period in both countries to assess the effect of frequency of exposure to graphic warning labels on smoking outcomes among youth and adults. We employed measures of risk cognition and label salience as proxies for frequency of exposure, due to the potential collinearity between smoking status and exposure to warning labels which may have contaminated our analyses of smoking outcomes.

Analytic Approach and Statistical Analysis Theoretical constructs from public health and social psychology, and their practical applications in social marketing, were used to identify components of graphic labels would be likely to influence behavior and inform variable selection. Marginal modeling using generalized estimating equations (GEE) was used to examine the population-averaged effects of tobacco graphic warnings on smoking intention and behavior. Multiple wave-pairs of data were analyzed concurrently with the predictors measured at a baseline wave (wave t) predicting outcomes measured at the next wave, the outcome wave (wave t + 1). Thus, it was possible for participants to provide data for multiple wave-pairs which increased the power of the study. A secondary goal of our research was to explore purchasing loose cigarettes (loosies), as an effect modifier of the association between label salience and smoking outcomes. This was imperative as smokers may employ avoidance tactics such as covering cigarette packs or purchasing loosies to avoid graphic labels.