A Decision Tool for Causal Inference and Observational Data Analysis Methods in Comparative Effectiveness Research (DECODE CER)
Comparative effectiveness research can be thought of as evaluating which treatment works best for whom and under what circumstances. While observational studies have become increasingly popular for CER; these studies must be carefully designed and analyzed in a way that controls for self-selection bias, where patients and/or physicians select who receives which treatment. This talk describes one common set of approaches, based on propensity scores, and an overall framework for selecting the most appropriate observational analysis methods, with careful attention to the design and research question. We will also discuss potential applications of machine learning methods to propensity score-based models.