The Novel Toolbox Quartet for Using Real-World-Data in Health Technology Assessment: Causal Diagrams, Target Trial Emulation, G-Methods, and Causal Modeling
Author(s)
Uwe Siebert, MD, MPH, MSc, ScD, UMIT TIROL - University for Health Sciences and Technology, Hall in Tirol, Austria. ONCOTYROL - Center for Personalized Cancer Medicine, Innsbruck, Austria. Harvard T.H. Chan School of Public Health and Harvard Medical School, Hall in Tirol, 7, Austria
PURPOSE: The objectives of this educational, practice-oriented and interactive lecture are to (1) introduce the modern toolbox needed to derive answers from real-world evidence (RWE) for decision making authorities, (2) provide applied case examples, and (3) discuss current challenges and solutions. DESCRIPTION: One of the most important tasks of health sciences researchers is to support decision makers with causal interpretations derived from real-world data (RWD). This lecture has four parts. First (30 min), an educational introduction of the four causal toolbox elements: Causal Diagrams, Target Trial Emulation, g-Methods, and Causal Modeling. Causal diagrams are used to assess the causal interpretability of intervention effects. Target trial emulations are used to avoid self-inflicted biases (e.g., immortal time bias) and to transport study results to target populations of interest. G-Methods are used to control for time-varying confounders (i.e., confounders simultaneously affected by treatment), which are present in most RWD concerning chronic disease treatments. Causal modeling includes structuring decision models causally, and carefully selecting causal model input parameters. This lecture part includes real-time polling for knowledge self-check. Second (15 min), real case examples from oncology, cardiovascular and infectious disease are used to demonstrate challenges and applied solutions, and to understand biases that arise when missing one of the toolbox elements. Examples include the application of the novel counterfactual “data cloning” approach. The lecturer will compare HTA bodies from different countries regarding the use of causal RWE approaches. Third (15 min), a structured discussion/survey with the audience using a unified terminology along the four toolbox elements will provide participants the opportunity to translate the lecture content to their own problems and to contribute with their own experiences. This lecture may guide researchers on RWE generation, RWD analysis, and causal modeling supporting approval and HTA decisions and provide didactical skills to educate their peers.
Conference/Value in Health Info
2023-11, ISPOR Europe 2023, Copenhagen, Denmark
Code
252
Topic
Methodological & Statistical Research