|
Abstract:
|
The US FDA can authorize the marketing of a new Modified Risk Tobacco Product (MRTP) only if the evidence submitted in the application shows the product is expected to benefit population health, and it encourages the use of statistical models to project population health effects. A modeling challenge is to account for large uncertainties in input parameters, such as transition rates to and from the new product, and the Excess Relative Risk (ERR) for product users on a scale from 0 for never-users to 1 for cigarette smokers. These uncertainties can be incorporated simultaneously by replacing fixed inputs with probability distributions and then calculating the resulting output distributions. Methods for calculating output distributions include probability trees for representative cases from the input distributions, Monte Carlo simulation of all the input distributions, and hybrid methods incorporating both. An illustrative analysis suggests that each approach has strengths and that key uncertainties include ERR and reduction in cigarettes per day (CPD) by dual users, as well as transition rates between MRTP-only use and dual use or cigarette-only smoking.
|
