Speakers
Uwe Siebert, MPH, MSc, ScD, MD, UMIT TIROL - University for Health Sciences and Technology; Harvard Chan School of Public Health, Hall in Tirol, Austria; Felicitas Kuhne, MSc, PhD, Pfizer Pharma GmbH, Berlin, Germany; Nicholas Latimer, MSc, PhD, SCHARR, University of Sheffield, Nottingham, United Kingdom
Separate registration required.
In recent years, real-world evidence (RWE) has been increasingly used to inform regulatory, payer, and health technology assessment (HTA) decisions, as well as clinical guideline development. In addition, there is growing recognition that hypothetical estimands are needed in clinical trials when the standard intention-to-treat (ITT) analysis does not directly address the decision problem, particularly in the presence of treatment switching. An innovative framework integrating causal inference methods, target trial emulation, causal estimands and causal modeling guides the design and analysis of observational studies and clinical trials. This course will (1) introduce causal principles, causal diagrams (directed acyclic graphs; DAGs), and target trial emulation to avoid self-inflicted design biases (eg, time-zero bias, immortal time bias), (2) provide an overview of causal methods addressing baseline confounding (multivariable regression, propensity scores) and time-varying confounding (eg, g-formula, marginal structural models with inverse probability of treatment weighting, and rank-preserving structural failure-time models with g-estimation), (3) propose appropriate estimands to ensure that decision problems are directly addressed in analyses of observational data or clinical trial data affected by treatment switching, (4) present lessons learned from applied case examples in HTA, such as single-arm trials with external control arms and trials affected by treatment switching, (5) show examples of causal machine learning and causal AI, (6) provide recommendations regarding the use of causal inference methods and estimands and their application in causal modeling, and (7) discuss acceptance and barriers from an HTA agency perspective. The target audience includes stakeholders and researchers across all areas of health and healthcare.
PREREQUISITE: Basic knowledge in epidemiologic study designs, analytic methods and biases (including the concept of confounding) is helpful.
Topic
Real World Data & Information Systems