Sequential sampling in prospective observational studies
Many factors must be taken into account when designing an observational study. Unlike randomized or experimental studies, observational studies cannot mitigate the effects of confounding through randomization, and such factors should be incorporated into the study design as well as the study analysis. Unfortunately, there is often little data available on most of these factors at the design stage, making it difficult to reliably predict the impact of these factors on the outcome, which may impact our ability to precisely estimate the treatment effect. In this talk we will first introduce the idea of using sequential testing to efficiently monitor prospective observational studies where data collection is expensive or burdensome, allowing for possible early study termination. Next, we will demonstrate how failure to account for adjustment covariates in the design stage of such observational studies negatively affects the study’s observed power. Finally, we’ll outline a constrained boundaries procedure that uses data collected at interim analyses to improve variance estimation over initial design-stage assumptions and to correct stopping boundaries; this will allow us to maintain power and type I error while still allowing for possible early study termination.
