Module 1: Introduction to Clinical Trials I: Design
Instructors: Dan Gillen and Susanne May
Module Description: This module provides an overview of the primary design issues that must be considered by the many disciplines that collaborate on a clinical trial. In this module, we consider the clinical, scientific, and regulatory setting of clinical trials, describing the phased approach to “treatment discovery” in which the safety, efficacy, and effectiveness of candidate treatments are investigated. This one day module focuses on the issues surrounding disease area and identification of the target population, definition of the treatment, choice of clinical outcomes, and choice of comparators. Throughout the module emphasis is placed on pre-specification of these key elements in a well-defined study protocol.
Module 2: Serving on a DSMB: What Every Member Should Know
Instructor: Scott Emerson
Module Description: During the conduct of a randomized clinical trial, it is commonplace for the accruing data to be regularly monitored by a committee of independent researchers in order to assure patient safety and ethical conduct. Typically such committees are comprised of 3-5 members, including clinicians expert in the disease under study, the treatment, and/or any expected toxicities, ethicists, and a biostatistician. It is incumbent upon all members of the data monitoring committee (DMC) to collaborate fully with each other in their highly interdisciplinary task. In this short course, we will elaborate on the common issues faced by DMC members. We consider the review of the DMC Charter, the investigator’s brochure, the protocol, and the interim monitoring plan from a DMC member’s viewpoint. We then discuss the elements of a DMC report that will be commonly encountered, and the organized approach to the review of the data. We discuss our experience with commonly encountered problems, including treatment toxicity, study dropout, treatment cross-over, and implementation of sequential stopping and/or adaptive modification of the protocol. The course is targeted to any individuals who might serve on a DMC, and an emphasis is placed on ensuring that the total DMC effort is greater than the sum of the parts contributed by the individual members.
Module 3: Introduction to Longitudinal Data Analysis
Instructors: Ben French and Colleen Sitlani
Module Description: Longitudinal studies follow individuals over time and repeatedly measure health status. This module will provide an overview of statistical methods for the analysis of longitudinal data. A brief introduction to pre-post data will be followed by an in-depth discussion of regression-based methods such as generalized linear mixed-effects models and generalized estimating equations. Relevant theoretical background will be provided. Examples from randomized controlled trials and observational studies will be used to illustrate key approaches and modeling strategies. Illustrative examples will be presented with R and STATA. This course is targeted toward individuals with little or no prior exposure to statistical methods for longitudinal data analysis. Attendees interested in more experience applying these methods should also consider “Case Studies in Longitudinal Data Analysis.”
Module 4: Evaluation of Biomarkers and Risk Prediction Models
Instructors: Holly Janes and Kathleen Kerr
Module Description: This short course will discuss the motivations for risk prediction models and the basic concepts underlying risk prediction for binary outcomes. Tools and metrics for evaluating risk prediction models will be covered, including ROC curves, calibration plots, and expected benefit. The course will also discuss methods to compare risk prediction models, paying particular attention to the case of two nested models and assessing the prediction increment of a new biomarker. Recent proposals for metrics that rely on the concept of reclassification to evaluate new biomarkers will be considered. Important concepts in using risk prediction models or biomarkers to guide treatment selection in the context of personalized medicine will be introduced.
Module 5: Introduction to Clinical Trials II: Conduct and Implementation
Instructors: Dan Gillen and Susanne May
Module Description: This module presents topics regarding the conduct and implementation of clinical trials that build upon the key concepts introduced in the module “Introduction to Clinical Trials – Day 1, Design”. This one day module focuses on choice of randomization strategies, blinding, specification of secondary endpoints, handling of missing data, conduct and monitoring of the study, the role of an independent data monitoring committee, and plans for reporting of the result. Throughout the module emphasis is placed on pre-specification of these elements in a well-defined study protocol.
Module 6: Advanced Topics in Clinical Trials: Adaptive Randomization (half day, 1:30-5:00 PM)
Instructor: Scott Emerson
Module Description: Recently much attention has been devoted to the re-design of ongoing randomized clinical trials based on interim results. One such area of interest is the use of interim trial results to determine the randomization of future subjects. In this short course we will review such methods as “covariate adaptive” randomization and “play-the-winner” randomization schemes. We will focus on the ability of these methods to better address the efficiency and ethical issues inherent in randomized clinical trials. We will also discuss the special considerations that must be made at the time of study design, during conduct of the clinical trial, and when reporting results, and how those special considerations might impinge on gaining regulatory approval for a new drug, biologic, or device.
Module 7: Case Studies in Longitudinal Data Analysis
Instructors: Ben French and Colleen Sitlani
Module Description: Longitudinal studies follow individuals over time and repeatedly measure health status. This module will focus on the practical application of statistical methods to the analysis of longitudinal data. A series of case studies will be used to discuss analysis strategies, the application of appropriate analysis methods, and the interpretation of results. Case studies will arise from randomized controlled trials and observational studies. Software implementation will be primarily in R and STATA; attendees are recommended to bring a laptop with relevant software installed. This course is targeted toward individuals who take “Introduction to Longitudinal Data Analysis” or have background knowledge of longitudinal data analysis methods.
Module 8: Prognostic Biomarker Evaluation
Instructor: Patrick Heagerty
Module Description: Longitudinal studies allow investigators to correlate changes in time-dependent exposures or biomarkers with subsequent health outcomes. The use of baseline or time-dependent markers to predict a subsequent change in clinical status such as transition to a diseased state requires the formulation of appropriate classification and prediction error concepts. Similarly, the evaluation of markers that could be used to guide treatment require specification of operating characteristics associated with use of the marker. The first part of this course will introduce predictive accuracy concepts that allow evaluation of time-dependent sensitivity and specificity for prognosis of a subsequent event time. We will overview options that are appropriate for both baseline markers and for longitudinal markers. Methods will be illustrated using examples from HIV and cancer research. The second part of this course will involve a technology workshop that will introduce the R packages (survivalROC, risksetROC, compriskROC) that are currently available for predictive accuracy of survival model. This segment will include hands-on training and demonstration of how to use these R packages for answering research questions. Several real-data examples for analysis will be provided and the instructors will discuss implementation and interpretation.
Module 9: Introduction to the Design and Evaluation of Group Sequential Clinical Trials
Instructors: Dan Gillen and John Kittelson
Module Description: Increasingly, clinical trials are conducted using group sequential methods in order to address the ethical and efficiency requirements for performing experiments with human volunteers. The design, conduct, and analysis of a sequential clinical trial is necessarily more involved than that for a clinical trial in which the data would only be analyzed at the end of the study. In this module we provide an introduction to methodology for the design and evaluation of group sequential trials from both a frequentist and Bayesian perspective. Methodology will presented using case studies and examples of clinical trials with continuous, binary, and survival endpoints. The sequential methods will be illustrated using RCTdesign, an R module for the design, monitoring, and analysis of clinical trials. Emphasis will be placed on both the scientific and statistical impact of group sequential trial design.
Module 10: Randomized Clinical Trial Analysis in the Presence of Missing Data: Sensitivity Analysis
Instructor: Scott Emerson
Module Description: At the request of the U.S. Food and Drug Administration, the National Academy of Sciences convened the Panel on the Handling of Missing Data in Clinical Trials to prepare a report that would make recommendations that could be used to aid in the FDA’s eventual development of a Guidance for Industry on that topic. Chief among the findings and recommendations of the resultant report was the need to make all efforts to prevent the occurrence of missing data. However, when missing data is unavoidable, the report also stressed the need for the careful specification of data analysis models that would explicitly account for any missingness. Furthermore, because all such analysis methods are based on untestable assumptions, the report also stressed the need for carefully conducted sensitivity analyses that would quantify the robustness of trial results to departures from any assumed mechanism of missingness. In this module we explore an approach to defining sensitivity analyses that first defines a missing at random (MAR) mechanism, and then explores a spectrum of departures that encompass relevant missing not at random (MNAR) models.
Module 11: Introduction to Survival Analysis
Instructors: Barbara McKnight and Susanne May
Module Description: This one-day module provides an introduction to survival data and survival analysis methods. We will introduce important functions and quantities for summarizing a survival distribution including the survival function, the hazard function, and the median survival time. We will also describe how censoring occurs in data and how it can be accommodated in estimates of functions like the Kaplan-Meier curve, and in the comparison of two groups using the logrank test. The module will explain key concepts such as risk sets, informative censoring and competing risks. We will provide examples of studies where survival analysis is used and where it should not be used. The course will focus on application and understanding the concepts; mathematical details will be kept to a minimum.
Module 12: Developing Prognostic and Predictive Biomarkers with High Dimensional Data
Instructors: Noah Simon and Richard Simon
Module Description: Modern medicine has graduated from broad spectrum treatments to targeted therapeutics. With new advances in high throughput biotechnology, it is inexpensive to run whole genome assays. Genetic features from these assays can be developed into biomarkers used to target therapies. A major difficulty here lies in the fact that the number of potential features is generally much greater than the number of patients. The goal of this module will be to introduce ideas in high dimensional predictive modeling, to discuss model validation and testing, and to give hands-on experience with these tools to build predictive biomarkers on real, high-dimensional datasets. Through the course, participants will become familiar with various methods in penalized regression, applications of cross-validation, as well as ideas in multiplicity and selection bias. Participants will gain experience applying these methods in R.
Module 13: Adaptive Sample Size Re-estimation: Design and Inference
Instructors: Sarah Emerson
Module Description: Adaptive clinical trial designs have been proposed as a promising new approach that may help improve the drug discovery process. Recent statistical papers have introduced a variety of methods to allow unplanned interim modifications to the study design while preserving the type I error rate of the clinical trial. In particular, there is a large body of literature exploring designs with unplanned modifications to the sample size based on interim estimates of the treatment effect falling into a “promising zone”. In this module we present the impact such designs can have on the power of randomized clinical trials. We provide guidance on the adaptive rules that lead to the greatest benefit to study precision, and we present and compare methods for full statistical inference following the use of such designs.
Module 15: Survival Analysis in Clinical Trials
Instructors: Barbara McKnight and Susanne May
Module Description: This one-day module will cover more advanced topics in survival analysis, with an emphasis on applications in clinical trials. Knowledge of material in Module 11 will be assumed. We will introduce methods suitable for testing hypotheses about a variety of quantitative summaries of the discrepancy between treatment groups’ survival. The choice of outcome event for the primary analysis will be considered. We will also discuss adjustment for prognostic variables using stratification and regression methods, and estimation of power and sample size. The course will focus on application and understanding the concepts; mathematical details will be kept to a minimum.
Module 16: Introduction to Causal Models and Potential Outcomes
Instructor: Thomas Richardson
Module Description: In this module we will introduce the Potential Outcome framework, that is the basis of many causal models. We first consider the simple setting of a binary outcome and treatment. Using this as our starting point we will then describe adjustment for confounding, and the analysis of direct effects (aka mediation). We will then link the potential outcome framework to causal graphs, which provides a basis for determining whether a set of covariates is sufficient to control for confounding. Lastly we will consider the analysis of studies with non-compliance. The topics will be illustrated with worked examples in R. The course will focus on conceptual understanding; mathematical details will be kept to a minimum.
Module 17: Special Topics in Clinical Trials
Instructor: Thomas Fleming
Module Description: This module provides detailed discussion of a number of topics that are currently receiving special emphasis in the community of clinical trialists. As such, this module presumes the material of Module 1. Included in this module are presentations on the important distinctions between confirmatory and exploratory studies; the use of biomarkers and surrogate endpoints in clinical trials; the handling of missing data in clinical trials and issues surrounding the conduct of noninferiority trials.
Module 18: Adaptive Enrichment Designs: Methods and Software
Instructor: Michael Rosenblum
Module Description: Adaptive enrichment designs involve preplanned rules for modifying enrollment criteria based on accrued data. They involve multiple populations of interest, e.g., defined in terms of a biomarker or risk score measured at baseline. This module will present methods and software for the design and analysis of adaptive enrichment designs. It will also present methods for improving precision of estimators of the average treatment effect, by leveraging information in baseline variables; these methods can be used in adaptive designs as well as non-adaptive trial designs.
Module 19: Sequential and Adaptive Analysis with Time-to-Event Endpoints
Instructors: Scott Emerson and Dan Gillen
Module Description: A great many confirmatory phase 3 clinical trials have as their primary endpoint a comparison of the distribution of time to some event (e.g., time to death or progression free survival). The most common statistical analysis models include the logrank test (usually unweighted, but possibly weighted) and/or the proportional hazards regression model. Just as commonly, the true distributions do not satisfy a proportional hazards assumption. Providing users are aware of the nuances of those methods, such departures need not preclude the use of those analytic techniques any more than violations of the location shift hypothesis precludes the use of the t test. However, with the increasing interest in the use of adaptive sample size re-estimation, adaptive enrichment, response-adaptive randomization, and adaptive selection of doses and/or treatments, there are many issues (scientific, ethical, statistical, and logistical) that need to be considered. In fact, when considering references to “less well understood” methods in the draft FDA guidance on adaptive designs, it is likely the case that many of the difficulties in adaptive time to event analyses can relate as much to aspects of survival analysis that are “less well understood” as to aspects of the adaptive methodology that has not been fully vetted. In this course we discuss some aspects of the analysis of censored time to event data that must be carefully considered in sequential and adaptive sampling. In particular, we discuss how the changing censoring distribution during a sequential trial affects the analysis of distributions with crossing hazards and crossing survival curves, as well as issues that arise owing to the ancillary information about eventual event times that might be available on subjects who are censored at an adaptive analysis.
Module 20: Propensity Scores: Methods, Models, and Adjustment
Instructor: Dave Stephens
Module Description: The propensity score is a key component of many causal inference procedures. After establishing the basic causal inference framework, we will outline the key methods of construction of propensity score functions, and study their core mathematical properties. We will detail the use of the propensity score in matching, inverse weighting and regression adjustments that allow the unconfounded effect of an exposure or treatment of interest to be estimated consistently. Using the framework of semiparametric inference, we will contrast the statistical properties of estimators derived using each method. We will investigate issues of model selection for the propensity score, and demonstrate the utility of judicious choice of predictors that enter into the propensity function: this will be illustrated in standard problems and also in the case of high-dimensional predictors. Longitudinal data will also be studied in the causal setting. Finally, we will develop the Bayesian framework for handling causal inference, and investigate how propensity function construction and usage translates to the new setting. All methods will be illustrated using examples from biostatistics, health research and econometrics. Computation will be performed in R.