The FDA Center for Tobacco Products (FDACTP) has identified impact analyses of potential FDA regulatory action as an area of scientific interest. Randomized controlled trials (RCTs) are the gold-standard for understanding the impact of an intervention, but standard approaches to analyzing RCTs have a number of drawbacks which limit our understanding of the impact of a product standard. First, subjects may not adhere to their randomized treatment assignment. Second, an RCT that was designed to evaluate the effect of a product standard in the overall population may not be adequately powered to estimate the treatment effect within important sub-groups. Finally, the characteristics of the population enrolled in the RCT may not represent the target population. These limitations must be addressed to obtain a complete understanding of the impact of potential product standards on public health. The goal of this application is to develop novel statistical methodology that addresses these concerns.
In Aim 1, we will develop statistical methods to estimate causal effects (i.e., the effect if compliance were legally mandated) from multiple RCTs. Estimating causal effects is central to understanding the impact of an intervention as a regulatory policy and combining data from multiple trials in a principled manner will result in more efficient estimators of causal effects without introducing bias.
In Aim 2, we will develop a robust approach for estimating causal effects in vulnerable populations. A number of vulnerable populations are disproportionally burdened by tobacco, and the methodology developed in this aim will result in precise estimates of the impact of a product standard in these sub-populations, while providing a novel approach to elucidating population heterogeneity.
In Aim 3, we will develop methods to calibrate estimation of causal effects to a relevant target population. The enrolled population of a RCT may not be representative of the target population and the methodology developed in this aim will allow results of RCTs to be extrapolated to a target population in the presence of treatment effect heterogeneity. This application addresses FDACTP scientific interest ?Impact Analysis ? Understanding the impact of potential FDA regulatory actions?. Our application represents a significant contribution to the field of tobacco regulatory science through the development of innovative statistical methods that will result in more precise estimates of the impact of potential FDA regulatory action (including impact in vulnerable populations) by principally combining data from the many trials or product standards funded by FDACTP.
The Food and Drug Administration (FDA) Center for Tobacco Products has identified impact analyses of potential FDA regulatory action as an area of scientific interest. Tobacco regulatory scientists have completed, initiated, or proposed a number of randomized controlled trials (RCTs) to evaluate the impact of potential product standards, but existing statistical methods to analyze data from RCTs can be limited by non-adherence to randomized treatment assignment, limited power for evaluating vulnerable subpopulations, and lack of external validity. The goal of this project is to address these limitations by developing novel statistical methodology for estimating causal effects from multiple RCTs, for estimating causal effects in vulnerable populations, and for calibrating estimates of causal effects to a relevant target population.
