Statistical Design and Analysis of Clinical Trials Principles and Methods Chapman & Hall/CRC Biostatistics Series

Statistical Design and Analysis of Clinical Trials: Principles and Methods concentrates on the biostatistics component of clinical trials. Developed from the authors? courses taught to public health and medical students, residents, and fellows during the past 15 years, the text shows how biostatistics in clinical trials is an integration of many fundamental scientific principles and statistical methods.

Teach Your Students How to Design, Monitor, and Analyze Clinical Trials

The book begins with ethical and safety principles, core trial design concepts, the principles and methods of sample size and power calculation, and analysis of covariance and stratified analysis. It then focuses on sequential designs and methods for two-stage Phase II cancer trials to Phase III group sequential trials, covering monitoring safety, futility, and efficacy. The authors also discuss the development of sample size reestimation and adaptive group sequential procedures, explain the concept of different missing data processes, and describe how to analyze incomplete data by proper multiple imputations.

Turn Your Students into Better Clinical Trial Investigators

This text reflects the academic research, commercial development, and public health aspects of clinical trials. It gives students a multidisciplinary understanding of the concepts and techniques involved in designing and analyzing various types of trials. The book?s balanced set of homework assignments and in-class exercises are appropriate for students in (bio)statistics, epidemiology, medicine, pharmacy, and public health.

Overview
What Is a Clinical Trial?
Requirements for a Good Experiment
Ethics and Safety First
Classifications of Clinical Trials
Multidisciplinary Teamwork in Clinical Trials
Appendix 1.1: Elements of Informed Consent

Concepts and Methods of Statistical Designs
External Validity
Internal Validity
Repeatability
The Phenomenon of Regression toward the Mean and Importance of a Concurrent Control Group
Random Samples and Randomization of Samples
Methods for Randomization
Minimization
Table of Patient Demographics and Baseline Characteristics

Efficiency with Trade-Offs and Crossover Designs
Statistical Efficiency of a Design
Crossover Designs
Analysis of 2 × 2 Crossover Designs
Appendix 3.1: Efficiency of the 1:1 Allocation Assuming Equal Variance
Appendix 3.2: Optimal Allocation under Unequal Variance
Appendix 3.3: Optimizing Number of Responders

Sample Size and Power Calculations
Fundamentals
Comparing Means for Continuous Outcomes
Comparing Proportions for Binary Outcomes
Comparing Time-to-Event (Survival) Endpoints
Clustered (or Correlated) Observations
Sample Size for Testing a Noninferiority or Equivalence Hypothesis
Comparing Ordinal Endpoints by Wilcoxon–Mann–Whitney Test
Sample Size Adjustments
Sample Size by Simulation and Bootstrap
Appendix 4.1: Fundamentals of Survival Data Analysis
Appendix 4.2: Exponential Distribution Model
Appendix 4.3: Survival with Independent Censoring
Appendix 4.4: MLE with Censoring under the Exponential Model

Analysis of Covariance and Stratified Analysis
Principles of Data Analysis
Continuous Response—ANOVA and ANCOVA
Variance Reduction by Covariates
Stratified Analysis
Appendix 5.1: Weekly Average Pain Score Data

Sequential Designs and Methods—Part I: Expected Sample Size and Two-Stage Phase II Trials in Oncology
Maximum Sample Size and Expected Sample Size
One-Stage versus Two-Stage Cancer Phase II Trials
Simon’s Two-Stage Designs
Discussion

Sequential Designs and Methods—Part II: Monitoring Safety and Futility
Monitoring Safety
Monitoring Futility with Conditional Probability
Appendix 7.1: R Function for Obtaining Parameters of Prior Distribution Beta(a, b) Based on Method A
Appendix 7.2: R Function for Obtaining Parameters of Prior Distribution Beta(a, b) Based on Method B
Appendix 7.3: Notes on the Two-Stage Monitoring Process

Sequential Designs and Methods—Part III: Classical Group Sequential Trials
Regulatory Requirements and Logistical Considerations for Trial Monitoring
Statistical Methods
Power, Information, and Drift Parameter
P-Value When Trial Is Stopped
Estimation of Treatment Effect
Appendix 8.1: R Function qfind for Calculating the Critical Value (Boundary) of the Second (Final) Analysis
Appendix 8.2: A Further Note on the Partial Sum Process with Independent Increments
Appendix 8.3: Information Time/Fraction and Maximum-Duration Trial versus Maximum-Information Trial

Monitoring the Maximum Information
Sample Size Reestimation
Monitoring Trial Duration for Studies with Survival Endpoints
Modification of the Classical GS Alpha-Spending Function Procedure
Adaptive GS Procedure—Change Not Dependent on Unblinded Interim Data
Appendix 9.1
Appendix 9.2

Missing Data
Introduction
Question to Answer: Causal Estimand
Missing Data Patterns and Mechanisms
Ignorability and Nonignorability of Missing Data
Analysis under the MAR Assumption by Multiple Imputation
Analysis of Longitudinal Data with Monotone Pattern Missing Values under MAR
Analysis under a Particular NMAR Model Assumption by MI
Use Reason for Withdrawal and Follow-Up Time to Form the Missing Data Pattern and Sensitivity Analyses
Other NMAR Approaches
Appendix 10.1: Sampling Distribution Inference
Appendix 10.2: Likelihood Inference
Appendix 10.3: Bayesian Inference
Appendix 10.4: Equivalence between Selection Model and Pattern-Mixture Model for MCAR and MAR
Appendix 10.5: NFD Missingness Mechanism as a Subclass of NMAR for Longitudinal Data with Monotone Missing Data Pattern
Appendix 10.6: Equivalence between the Selection Model NFD Missingness Condition (Equation 10A.10) and the Pattern-Mixture Model NFD Missingness Condition (Equation 10A.12)

Homework Problems and References are included at the end of each chapter.

Weichung Joe Shih, PhD, is professor and chair of the Department of Biostatistics in the Rutgers School of Public Health at Rutgers University, and director of the Biometrics Division at the Rutgers Cancer Institute of New Jersey. He is an elected fellow of the American Statistical Association and an elected member of the International Statistical Institute. He served on the advisory board of the U.S. FDA for reviewing new drug applications and was associate editor of professional journals, including Statistics in Medicine, Controlled Clinical Trials, Clinical Cancer Research, Statistics in Biopharmaceutical Research, and Statistics in Bioscience. His research interests include adaptive designs and missing data issues.

Joseph Aisner, MD, is a professor of medicine and a professor of environmental and occupational medicine at the Robert Wood Johnson Medical School of Rutgers University, director of the Medical Oncology Unit at the Robert Wood Johnson University Hospital, and co-leader of the Clinical Investigations Program at the Rutgers Cancer Institute of New Jersey. He is a fellow of the American College of Physicians and the American Society of Clinical Oncology. He serves on and chairs several National Data Monitoring Committees and has served on the editorial board of multiple journals, including Journal of Clinical Oncology, Cancer Therapeutics, Medical Oncology, Clinical Cancer Research, and Hematology-Oncology Today. His research interests include cancer clinical trials and evaluation of therapeutic interventions.