Biostatistics seminar
Rebecca Hubbard will give a talk titled "The Collision of Real-World Data and Causal Inference."
About
Rebecca Hubbard, PhD is the Carl Kawaja and Wendy Holcombe Professor of Public Health and the Director of the School of Public Health at Brown University.
Lunch will be provided during the seminar.
Abstract
Real-world data (RWD) are ubiquitous, creating exciting opportunities to generate evidence about the effectiveness and safety of health interventions in populations and settings that are difficult to study in randomized trials. The growing availability of RWD has coincided with a transformation in how we think about causal inference from observational data. Modern causal inference frameworks bring much-needed rigor to the design and analysis of RWD studies, but their application also highlights challenges created by data that were generated for healthcare delivery, administration, and billing rather than for research. In this talk, I will explore the consequences of causal inference colliding with the messy realities of RWD. I will discuss how principles such as target trial emulation can prevent common biases, while also examining problems that arise when the target trial framework is superimposed on imperfectly observed healthcare data. Examples will illustrate how measurement error, missing data, and the complex process of RWD data provenance can undermine otherwise rigorous study designs. I will also highlight emerging statistical and design-based approaches for addressing these challenges. Illustrations from cancer epidemiology will demonstrate both the potential of RWD to transform evidence generation and the importance of accounting for information bias when applying causal inference tools. Ultimately, the collision of causal inference and RWD leaves neither unchanged, revealing important challenges but also creating opportunities for methodological innovation and more credible real-world evidence.
Contact
Andy Ni