Chair: Huiman Barnhart, PhD (Duke University)
Instructors:
Huiman Barnhart, PhD (Duke University)
Yuliya Lokhnygian, PhD (Duke University)
Roland Matsouaka, PhD (Duke University)
Frank W. Rockhold, PhD (Duke University)
Stuart Pocock, PhD (London School of Hygiene and Tropical Medicine)
Course Description:
Use of hierarchical endpoints has gained popularity in the last decade due to its intuitive appeal and its natural way of dealing with prioritization of multiple endpoints. Various methods for hierarchical endpoints have been increasingly available in the literature and more and more clinical investigators are interested in analyzing data or in designing trials with hierarchical endpoints. Due to its high demand, it is therefore crucial and beneficial to have a workshop on this important topic. The goal is to bring new comers up to speed on this area, explain to them the do’s and don’ts of hierarchical endpoints, and provide up-to-date information for researchers who wish to learn more.
This short course assumes that the audience have a great interest in hierarchical endpoints, but do not have the necessary background on the topic, the different approaches and the associated statistical methods. It is meant for both statisticians and non-statisticians. It is also designed for audience who wish to learn practical ways to do trial design with hierarchical endpoints. We will provide a tutorial starting from the introduction to hierarchical endpoints and existing population measures, i.e. win ratio, win odds, net benefit, probability index or win probability (i.e., desirability of outcome ranking), for hierarchical endpoints, starting with a single endpoint to hierarchical endpoints involving multiple mixed-type endpoints. We then give an overview of estimation and inference approaches for unmatched and matched scenarios. We will go over the existing R packages to illustrate the analysis approaches with hands-on examples. We then present practical considerations in trial design with hierarchical endpoints on how to select trial parameters that are realistic and defendable for sample size and power calculations. Programs or packages on sample size and power calculation will be used for illustration. Finally, we will share our perspectives with pros and cons on using these population measures for hierarchical endpoints in clinical trials. We provide the following outline of the tutorial:
- Introduction to hierarchical endpoints and population measures (45 minutes) – Presenter: Yuliya Lokhnygina
- Break (5 minutes)
- Estimation and inference in unmatched and matched scenarios (45 minutes hour) – Presenter: Roland Matsouaka.
- Break (10 minutes)
- Practical considerations in trial design (45 minutes) – Presenter: Huiman Barnhart
- Break (10 minutes)
- Perspective and experience on using hierarchical endpoints in Duke cardiovascular trials (10 minutes) – presenter: Frank Rockhold
- Systematic review of published win ratio trials during 2022-24 (10 minutes) – Presenter: Stuart Pocock
- Floor discussions and interaction with audience (15 minutes)
Learning Outcomes
- Learn the importance of why using hierarchical endpoints in clinical trials.
- Lean the various win measures used for hierarchical endpoints in assessing treatment effect.
- Learn the practical ways to do trial design with hierarchical endpoints.
- Learn how to use the R package and R-shiny tool for sample size and power calculation, and analyses.
Instructors:
Huiman Barnhart, PhD
James B. Duke Distinguished Professor
Department of Biostatistics & Bioinformatics
Duke University School of Medicine
Huiman Barnhart is a James B. Duke Distinguished Professor in Duke’s Department of Biostatistics and Bioinformatics. She is a Fellow of the American Statistical Association and a member of the Duke Clinical Research Institute (DCRI). Dr. Barnhart has led numerous multicenter clinical trials and registry studies at DCRI and has served as the principal investigator for several data coordinating centers. Her research focuses on outcomes, endpoints, and estimands, with interests in hierarchical endpoints, SMART trials, reliability/agreement of outcomes, and new medical diagnostic tests. She collaborates in areas such as cardiovascular, liver, kidney diseases, and pediatric rheumatoid arthritis.
Yuliya Lokhnygian, PhD
Associate Professor
Department of Biostatistics & Bioinformatics
Duke University School of Medicine
Yuliya Lokhnygina is an associate professor in the Duke Department of Biostatistics and Bioinformatics and a member of the Duke Clinical Research Institute. She has many years of experience leading statistical teams in designing and analyzing clinical trials. Her research interests include methods for novel hierarchical endpoints, cluster-randomized trials, and stepped-wedge trials. Her collaborations with clinical investigators mainly focus on projects in cardiovascular research, diabetes, kidney disease, and infectious diseases.
Roland Matsouaka, PhD
Associate Professor
Department of Biostatistics & Bioinformatics
Duke University School of Medicine
Roland Matsouaka is an associate professor in the Duke Department of Biostatistics and Bioinformatics and a member of the Duke Clinical Research Institute. His statistical interests include nonparametric, semiparametric, and causal inference methods for comparative effectiveness studies (for clinical trials affected by non-compliance and observational studies). He develops principled statistical methods that make the best use of the data, minimize bias, and ensure fair assessments. He has collaborated with clinical investigators on a wide range of projects to better understand and treat cardiovascular diseases while finding ways to improve racial equity in access to care. As a member of the DCRI’s Outcomes Research Group, he has led the statistical analyses of large registry data, including the Society of Thoracic Surgeons (STS) National Database, the STS and American College of Cardiology (ACC) Transcatheter Valve Therapy (TVTR) Registry, and the American Heart Association/American Stroke Association Get With The Guidelines (GTWG) Registry.
Frank W. Rockhold, PhD
Professor
Department of Biostatistics & Bioinformatics
Duke University School of Medicine
Frank Rockhold is Professor of Biostatistics and Bioinformatics and a member of the Duke Clinical Research Institute. He is an affiliate Professor of Biostatistics at Virginia Commonwealth University, and Strategic Consultant at Hunter Rockhold, Inc. His 40+-year career includes senior research positions at Lilly, Merck, and GlaxoSmithKline, where he retired as Chief Safety Officer and Senior Vice President of Global Clinical Safety and Pharmacovigilance. Dr. Rockhold served for 9 years on the board of directors of the non-profit CDISC and is past president of the Society for Clinical Trials and a past member of the PCORI Clinical Trials Advisory Panel. He is currently Chair of the Board of the Frontier Science and Technology Research Foundation and a technical advisor to EMA. Dr. Rockhold has diverse research interests and consulting experience in industry and academia including clinical trials design, data monitoring, benefit/risk, safety and pharmacovigilance and has been a leader in the scientific community in promoting data disclosure and transparency in clinical research. Frank is widely published in major scientific journals across a wide variety of research topics.
Stuart Pocock, PhD
Professor, Medical Statistics
London School of Hygiene and Tropical Medicine
Stuart Pocock has been a Professor of Medical Statistics since 1989 at the London School of Hygiene and Tropical Medicine, a premier European centre for biostatistical research and education. Professor Pocock oversees a statistical centre dedicated to the design, execution, analysis, and reporting of significant clinical trials, particularly in the field of cardiovascular diseases. He also provides consultancy services as a statistician for various clinical trials requiring expert statistical guidance and serves on numerous trial data monitoring and steering committees.
Professor Pocock collaborates internationally with institutions such as the Centro Nacional de Investigaciones Cardiovasculares in Madrid and the Cardiovascular Research Foundation along with the Mount Sinai School of Medicine in New York. A frequent lecturer on clinical trial issues, he has authored over 600 peer-reviewed articles and a widely used textbook titled “Clinical Trials: a Practical Approach.”
His primary research focus is on clinical trials, encompassing both methodological advancements and applied collaborations in major trials, predominantly within cardiology. His specific areas of expertise include standards for statistical reporting of trials and epidemiological studies, principles for data monitoring including ethical considerations and early stopping guidelines, the win ratio approach for hierarchical composite outcomes with clinical priorities, adaptive designs, handling non-proportional hazards, repeat-event analyses, appropriate use of propensity scores, evaluating non-inferiority trials, addressing multiplicity in trial reporting (such as subgroup analyses, multiple outcomes, and covariate adjustment), and the development and validation of prognostic risk scores.
