Innovative Statistical Methods for Estimating the Impact of Tobacco Product Standards - SUMMARY Evaluating the impact of a potential tobacco product standard requires a comprehensive understanding of the direct impact of the intervention, as well as an understanding of the unintended consequences and other down-stream effects to capture the full effect of the regulation on public health. To support FDA regulatory action, impact analyses must be supported by rigorously designed studies and innovative statistical methods. We propose to develop novel statistical methods for evaluating the impact of a tobacco product standard. We motivate our development by gaps in the literature related to nicotine reduction, but many of the barriers to evaluating the impact of nicotine reduction are challenges for other tobacco product standards, as well. Since 2010, the FDACTP and NIH have funded many randomized controlled trials (RCTs) to understand the impact of a low nicotine standard for cigarettes, yet, critical unanswered questions remain related to 1) the impact on cessation, 2) understanding the population-level impact calibrated to the United States smoking population, 3) understanding the heterogeneity in the effect of a low nicotine standard to achieve an equitable impact across the population, and 4) estimating the real-world impact of nicotine reduction. Addressing these questions is challenging due to no single data source providing sufficient power or precision (i.e. cessation), lack of a sufficiently diverse and representative population (i.e. calibration and heterogeneity), or fundamental differences between implementing the intervention in practice and in a RCT (i.e. real-world impact). To understand the full impact of potential regulatory action, we must integrate information from multiple data sources, but existing approaches to data integration are insufficient. For the last seven years, we have developed statistical methods for tobacco regulatory science (TRS), providing the statistical tools needed to answer the questions most relevant to FDACTP. In this application, we address limitations in the existing methods for data integration that prevent the TRS community from answering key questions through the following specific aims: 1) Develop causal meta-analytic techniques for estimating the regulatory effect of an intervention; 2) Develop a Bayesian approach to dynamic borrowing that leverages the underlying latent structure of the data; 3) Develop methods to estimate the individual treatment effects to understand disparities in the impact of regulations; 4) Develop approaches to sensitivity analysis to understand the effect of unmeasured factors on the estimated regulatory effect of tobacco product regulation. This application addresses FDACTP scientific interest “Impact Analysis – Understanding the impact of potential FDA regulatory actions”. Our proposal represents a significant contribution to TRS and will result in statistical methods that provide more precise estimates of the impact of FDA regulatory action. The innovative methods proposed in this application will provide powerful tools for leveraging the extensive data resources for evaluating the impact of tobacco product standards.
