Course A: Mixed Effects Models for Longitudinal Data

Instructor

Dr. James Rochon, Research Professor at the Duke Clinical Research Institute

Course Description

This course is designed to update your knowledge and skills for analyzing repeated measures and longitudinal data in biomedical research. It is directed at Master's and Ph.D. biostatisticians who are heavily engaged in analyzing data on a day-to-day basis. We provide enough theory to bring you up to speed and allow you to follow trends in the literature. Practical computer examples using PROC MIXED in SAS are also presented. First, a brief review of biostatistical concepts in repeated measures designs and longitudinal data is presented . Then, we turn to statistical methods that have come into common use over the past 20 years. They include random-coefficient regression models, and mixed-effect models for longitudinal data. Fixed-effects vs. random effects are discussed with special emphasis on how to treat clinical sites in multi-center investigations. ML and REML estimation procedures are contrasted. The application of these techniques to group-randomized designs and hierarchical models (if time permits) is provided.

About the Instructor

Dr. James Rochon is a Research Professor at the Duke Clinical Research Institute in Durham, NC. He received his Ph.D. from the University of North Carolina at Chapel Hill in 1985. He is the Principal Investigator for the Coordinating Center for a number of multi-center, NIH studies including the CALERIE and DILIN networks. He has published methodological papers on repeated measures studies in a number of journals including Biometrics, Journal of the American Statistical Association, Journal of the Royal Statistical Society and elsewhere. They include methods for performing sample size calculations at the design stage and covariance structures at the analysis stage.. His current research interest lies in adjusting for confounders arising post-randomization in clinical trials.

Course B: Adaptive Design Methods in Clinical Trials

Instructor

Dr. Mark Chang, Millennium Pharmaceuticals, Inc.

Course Description

Objectives: Learn statistical methods for various adaptive designs; learn how to design adaptive trials - all the steps including how to use the SAS macros provided; explore the challenges in the implementations of adaptive designs. Ultimately, the students are expected to walk a way with the right knowledge and tools to start designing adaptive trials.

An adaptive design is a design that allows for modifications to the on-going trial based on either the observed data from the trial or external information. It is recognized that adaptive designs can improve the efficiency of a trial design and increase the probability of success. In this section of the workshop, we will study different adaptive designs including: (1) sample-size reestimation, (2) changes in the timing and the number of analyses, (3) drop-loser design, (4) response-adaptive randomization, (5) Biomarker-adaptive design, and (6) adaptive dose-escalation design. Different methods for these adaptive designs are discussed through trial examples. In addition to the theoretical discussions, we will use SAS macros and R functions provided to go through with you the various adaptive designs. Finally we will discuss the challenges of the design implementation and provide suggestions.

About the Instructor

Dr. Chang has years of experience as a statistician in the field of clinical trials in addition to years of teaching experience. Chang's most recent publications are mainly on adaptive designs and he co-authored the book "Adaptive Design Methods in Clinical Trials" in 2006. As the director of Biostatistics, his involvements in drug development at Millennium include both strategic and methodological aspects. As a technical resource in the biostatistics department at Millennium, he has lead and conducted numerous computer simulations for clinical trials and provided quantitative assessments for clinical development programs to assist in decision making.

Course C: New Regression Techniques in SAS

Instructor

Dr. Colin (Lin) Chen, SAS Institute Inc.

Course Description

This short course mainly introduces new developments on regression techniques in SAS. As a warm-up, we briefly go over ordinary least squares regression and general linear models by some examples using the SAS/REG and SAS/GLM procedures. Then, we will focus on the regression techniques developed recently in SAS, robust regression and quantile regression.

Robust regression, which is covered in the SAS/ROBUSTREG procedure, detects outliers and provides resistant results in the presence of outliers. Quantile regression, which is covered in the SAS/QUANTREG procedure, computes conditional quantile functions and conducts statistical inference on regression quantiles without any distributional assumptions.

Statisticians and data miners, who need to analyze messy data or heterogeneous data, will learn to select from the estimation methods available in these SAS procedures and interpret diagnostic results provided in the output. You will also learn to explore the heterogeneity of both cross-sectional and longitudinal data, and interpret the quantile effects and quantile processes. In addition, basic and advanced features of these procedures will be presented with examples, including the construction of growth charts for medical measurements.

This course requires basic knowledge of linear regression models.

About the Instructor

Dr. Colin (Lin) Chen received Ph.D. in Statistics and MS in Computer Science from Purdue University, West Lafayette, Indiana. He joined SAS Institute Inc., Cary, North Carolina in 1998 and is currently a Senior Research Statistician at SAS. He supports the QUANTREG, ROBUSTREG, LIFEREG, and PROBIT procedures in SAS/STAT. His research interests include Robust Statistics, Quantile Regression, Bayesian Computation, Survive Analysis, Growth Charts, Goodness of Fit, and Monte Carlo Simulation. He has published over a dozen papers.

Course D: Pharmacogenomics/Pharmacogenetics and Their Utility in Clinical Studies

Instructor

Dr. Sue_Jane Wang, U.S. Food and Drug Administration

Course Description

Recently, there is an enthusiastic interest and desire to incorporate presumed genomic biomarker in early to late phases medical product development program. The entire medical product development program consists of exploratory and confirmatory phases. In this short course, exploration of pharmacogenomics/pharmacogenetics studies from drug discovery to drug development will be presented. This leads to the discussion of biomarker qualification. Prospective vs. retrospective study designs and rationales in exploratory versus confirmatory studies in the context of pharmacogenomics and pharmacogenetics will be illustrated. Pharmacogenomics clinical trials that use genomic/genetic composite biomarkers to adapt sensitive patient subpopulations will be elucidated using scenarios to highlight the often misused designs that should be avoided. The diagnostic tools that use individual genomic/genetic profiles may serve to identify responsive patients aiming at enrichment of study patients or personalized medicine. The goal and utility of therapeutic/diagnostic co-development in pharmacogenomics or pharmacogenetics clinical trials will be introduced. The statistics metrics for repeatability, reproducibility and standardization of a genomic/genetic diagnostic test will be discussed.

About the Instructor

Dr. Sue-Jane Wang is currently Associate Director, Office of Biostatistics, Office of Translational Sciences, Center for Drug Evaluation and Research, U.S. Food and Drug Administration. She has been with FDA for more than a decade. Dr. Wang works with statistical reviewers serving all medical divisions in CDER on adaptive clinical trial designs and pharmacogenomics/pharmacogenetics trials. Dr. Wang's research interest and publications in recent years have been focusing on adaptive/flexible clinical trial designs, noninferiority trials, pharmacogenomics/pharmacogenetics trials that make use of genomic/genetic composite biomarker classifier. Her collaborative research has resulted in more than 70 professional publications. She is an Editor in Chief of Pharmaceutical Statistics Journal.

Course E: Introductory Survival Analysis

Instructor

Prof. Jianwen Cai, University of North Carolina at Chapel Hill

Course Description

The course will introduce the basic concepts in the analysis of survival data. It will be oriented toward application and interpretation of various methodologies. Examples will be drawn mostly from medical and epidemiologic research. Course topics include censoring, Kaplan-Meier estimation, logrank test, Cox regression model, time-dependent covariates, stratification, and left truncation. No previous knowledge of survival analysis is required, although participants should have a good working knowledge of basic principles of statistical inference and linear regression. The course will emphasize applications with SAS, but lecture notes using Splus will also be available.

About the Instructor

Dr. Jianwen Cai is Professor and Associate Chair of Biostatistics at the University of North Carolina at Chapel Hill. She received her Ph.D. from the University of Washington in 1992. After graduation, she joined the Department of Biostatistics at the University of North Carolina at Chapel Hill, where she teaches graduate statistics courses, conducts statistical methodology research, and participates in collaborative research in public health. She was promoted to Associate Professor in 1999 and to Full Professor in 2004. Professor Cai is an elected fellow of the American Statistical Association. She currently serves as an Associate Editor for Biometrics and Lifetime Data Analysis.

Professor Cai's research interest includes survival analysis and regression models, design and analysis of clinical trials, analysis of correlated responses, cardiovascular disease research, and epidemiological models. Professor Cai has published over 90 scientific papers and made important contribution to statistical methodology, especially in the area of methods dealing with multivariate failure time data, and to obesity and cardiovascular disease research.

Course F: Analysis of Multivariate Failure Time Data

Instructor

Prof. Danyu Lin, University of North Carolina at Chapel Hill

Course Description

Multivariate failure time data arise when each study subject can potentially experience multiple events or when there exists clustering of subjects such that failure times within the same cluster are correlated. Major complications in analyzing such data include the dependence among related failure times and censoring due to limited follow-up or competing events. This short course presents a variety of statistical models and methods for the analysis of these data. We discuss both marginal and frailty models, paying primary attention to semiparametric regression methods. Relevant software will be described. A number of clinical and epidemiologic studies will be provided for illustrations.

About the Instructor

Dr. Danyu Lin received his Ph.D. from the University of Michigan in 1989. After one-year post-doc at Harvard, he joined the University of Washington Department of Biostatistics, where he was promoted to Associate Professor in 1994 and to Professor in 1998. Dr. Lin moved to the University of North Carolina at Chapel Hill in 2000 to become the first Dennis Gillings Distinguished Professor of Biostatistics. Dr. Lin has published more than 100 papers, most of which have appeared in leading statistical journals. He was identified by Thomson ISI as one of the world's most highly cited researchers in mathematics. Dr. Lin is an internationally recognized leader in survival analysis, especially the analysis of multivariate failure time data. He has received numerous awards and honors, including the Mortimer Spiegelgman gold medal from the American Public Health Association, Fellow of the ASA, Fellow of the IMS, JASA-Theory and Methods discussion papers and JRSS(B) discussion paper. He has served as a consultant to the FDA since 1997.