Observational Studies of Exposure to Tobacco and Nicotine Products: Best Practices for Maximizing Statistical Precision and Accuracy

This review explores methodological considerations for maximizing the precision and accuracy of observational cohort studies assessing the risk profiles of tobacco and nicotine products. These considerations, informed by the ROBINS framework for minimizing statistical bias, are anchored in a comprehensive characterization of exposure to all tobacco products currently or formerly used, with corroboration of dose-response relationships. Best-practice considerations for non-randomized observational studies of exposure are enumerated, to most precisely and accurately inform future waves of databases, studies and meta-analyses. These principles are then considered within the context of a recent comprehensive meta-analysis, representing 107 observational studies, which evaluated the health effects of using electronic cigarettes (EC) compared to non-use, use of combusted cigarettes (CC) and dual-use of both products (DU). The meta-analysis had previously reported the relative risk from all-sources, including tobacco use and non-tobacco use. We now report the magnitude of risk reduction associated with displacing CC use with EC use indicated from the primary references, along with considerations regarding the precision of characterization of dose, duration, and timing of CC and EC use in the cited studies.