1、DESIGN AND ANALYSIS OFCLINICAL TRIALSConcepts and MethodologiesSecond EditionSHEIN-CHUNG CHOWMillennium Pharmaceuticals, Inc.Cambridge, MAJEN-PEI LIUNational Cheng-kung UniversityTainan, TaiwanNational Health Research InstitutesTaipei, TaiwanA John Wiley Editors Emeriti: Vic Barnett, J. Stuart Hunte
2、r, David G. KendallA complete list of the titles in this series appears at the end of this volume.ffirs.qxd 28.10.03 19:36 Page iiDESIGN AND ANALYSIS OFCLINICAL TRIALSConcepts and MethodologiesSecond EditionSHEIN-CHUNG CHOWMillennium Pharmaceuticals, Inc.Cambridge, MAJEN-PEI LIUNational Cheng-kung U
3、niversityTainan, TaiwanNational Health Research InstitutesTaipei, TaiwanA John Wiley goals of clinical trials and target population (Chapter 4); clustered randomized design,group sequential design, placebo-challenging design, and blinded reader design (Chapter5); superiority trials, active control a
4、nd equivalence/noninferiority trials, dose-responsetrials, bridging studies, and vaccine clinical trials (Chapter 7); sample size determinationon equivalence and noninferiority trials and comparing variabilities (Chapter 11); and useof genomic information for evaluation of ecacy (Chapter 12). The th
5、ree new chaptersinclude “Designs for Cancer Clinical Trials” (Chapter 6), “Preparation and Implementationof a Clinical Protocol” (Chapter 14), and “Clinical Data Management” (Chapter 15).ixfpref.qxd 27.10.03 15:12 Page ixFinally, the second edition includes more than 280 new references from clinical
6、-relatedliterature. We believe that this revised and expanded second edition will benefit clinicalscientists/researchers from the medical-pharmaceutical industry, regulatory agencies, andacademia by serving as an extremely useful reference source in clinical research.From John Wiley and Sons, I woul
7、d like to thank Steve Quigley for providing us theopportunity to work on this edition, and Susanne Steitz for her outstanding eorts inpreparing this edition. The first author would like to thank support from colleagues fromStarPlus, Inc. and Millennium Pharmaceuticals, Inc. during the preparation of
8、 this edition.The second author wishes to express his gratitude to his wife, Dr. Wei-Chu Chie, and theirdaughter Angela for their support, patience, and understanding during the preparation ofthis edition.Finally, the views expressed are those of the authors and not necessarily those ofMillennium Ph
9、armaceuticals, Inc., and National Cheng-Kung University and NationalHealth Research Institutes, Taiwan. We are solely responsible for the contents and errors ofthis edition. Any comments and suggestions will be very much appreciated.SHEIN-CHUNG CHOWJEN-PEI LIUCambridge, MassachusettsTainan, TaiwanSe
10、ptember, 2003x PREFACEfpref.qxd 27.10.03 15:12 Page xPREFACE TO THE FIRST EDITIONClinical trials are scientific investigations that examine and evaluate safety and efficacy ofdrug therapies in human subjects. Biostatistics has been recognized and extensivelyemployed as an indispensable tool for plan
11、ning, conduct, and interpretation of clinicaltrials. In clinical research and development, the bio-statistician plays an important role thatcontributes toward the success of the trial. An open and effective communication amongclinician, biostatistician, and other related clinical scientists will res
12、ult in a successfulclinical trial. The mutual communication, however, is a two-way street: not only (1) thebiostatistician must effectively deliver statistical concepts and methodologies to his/hercolleagues but also (2) the clinician must communicate thoroughly clinical and scientificprinciples emb
13、edded in clinical research to the biostatistician. The biostatistician can thenformulate these clinical and scientific principles into valid statistical hypotheses, models,and methodologies for data analyses. The integrity, quality, and success of a clinical trialdepend on the interaction, mutual re
14、spect, and understanding among the clinician, thebiostatistician, and other clinical scientists.There are many books on clinical trials already on the market. These books, however,emphasize either statistical or clinical aspects. None of these books provides a balancedview of statistical concepts an
15、d clinical issues. Therefore the purpose of this book is notonly to fill the gap between clinical and statistical disciplines but also to provide acomprehensive and unified presentation of clinical and scientific issues, statisticalconcepts, and methodologies. Moreover this book focuses on the inter
16、actions amongclinicians, biostatisticians, and other clinical scientists that often occur during the variousphases of clinical research and development. This book is intended to give a well-balancedoverview of current and emerging clinical issues and newly developed statisticalmethodologies. Althoug
17、h this book is written from a viewpoint of pharmaceutical researchand development, the principles and concepts presented in this book can be applied tononbiopharmaceutical settings.xiflast.qxd 27.10.03 15:13 Page xiIt is our goal to provide a concise and comprehensive reference book for physicians,c
18、linical researchers, pharmaceutical scientists, clinical or medical research associates,clinical programmers or data coordinators, and biostatisticians in the areas of clinicalresearch and development, regulatory agencies, and academe. Hence this book is writtenfor readers with minimal mathematical
19、and statistical backgrounds. Although it is notrequired, an introductory statistics course that covers the concepts of probability, samplingdistribution, estimation, and hypothesis testing would be helpful. This book can also serveas a textbook for graduate courses in the areas of clinical and pharm
20、aceutical research anddevelopment. Readers are encouraged to pay attention to clinical issues and their statisticalinterpretations as illustrated through real examples from various phases of clinical researchand development.The issues covered in this book may occur during the various phases of clini
21、cal trials inpharmaceutical research and development, and their corresponding statistical interpreta-tions, concepts, designs, and analyses. All the important clinical issues are addressed interms of the concepts and methodologies of the design and analysis of clinical trials. Forthis reason this bo
22、ok is composed of clinical concepts and methodologies. Each chapterwith different topics is self-contained.Chapter 1 provides an overview of clinical development for pharmaceutical entities, theprocess of drug research and development in pharmaceutical industry, and regulatoryprocesses and requireme
23、nts. The aim and structure of the book is also discussed in thischapter. The concepts of design and analysis of clinical trials are covered from Chapters 2through 6. Basic statistical concepts such as uncertainty, bias, variability, confounding,interaction, and statistical versus clinical significan
24、ce are introduced in Chapter 2. Funda-mental considerations for the selection of a suitable design in achieving certain objectivesof a particular trial under various circumstances are provided in Chapter 3. Chapter 4illustrates the concepts and different methods of randomization and blinding that ar
25、eindispensable to the success and integrity of a clinical trial. Chapter 5 introduces differenttypes of statistical designs for clinical trials such as parallel, crossover, titration, andenrichment designs and discusses their relative advantages and drawbacks. Various typesof clinical trials, which
26、include multicenter, active control, combination, and equivalencetrials, are the subject of Chapter 6.Methodologies and the issues for clinical data analysis are addressed in Chapters 7through 12. Since clinical endpoints can generally be classified into three types,continuous, categorical, and cens
27、ored data, various statistical methods for analyses ofthese three types of clinical data and their advantages and limitations are provided inChapters 7, 8, and 9, respectively. In addition, group sequential procedures for interimanalysis are given in Chapter 9. Different procedures for sample size d
28、etermination areprovided in Chapter 10 for data under different designs. Statistical issues in analyzingefficacy data are discussed in Chapter 11. These issues include baseline comparisons,intention-to-treat analyses versus evaluable or per-protocol analyses, adjustment ofcovariates, multiplicity is
29、sues, and data monitoring. Chapter 12 focuses on the issues ofanalysis of safety data, including the extent of exposure, coding, and analysis of adverse.events, and analysis of laboratory data.For each chapter, whenever possible, real examples from clinical trials are included todemonstrate the clin
30、ical and statistical concepts, interpretations, and their relationships andinteractions. Comparisons of the relative merits and disadvantages of statistical methodo-logy for addressing different clinical issues in various therapeutic areas are discussed inappropriate chapters. In addition, if applic
31、able, topics for future development are provided.xii PREFACE TO THE FIRST EDITIONflast.qxd 27.10.03 15:13 Page xiiAll computations in this book were performed using SAS. Other statistical packages suchas SPSS, BMDP, or MINTAB may also be applied.At John Wiley, we would like to thank Acquisition Edit
32、or Steve Quigley for providingus with the opportunity to work on this book and for his outstanding effort in preparing thisbook for publication. We are greatly indebted to the Bristol-Myers Squibb Company andCovance, Inc. for their support, in particular, to S. A. Henry, L. Meinert, and H. Koffer. W
33、eare grateful for A. P. Pong, C. C. Hsieh, and G. Y. Han for their assistance in preparing themany charts, figures, graphs, and tables in this book. We are grateful to Y. C. Chi, F. Ki,and C. S. Lin for many helpful discussions and for reviewing the manuscript. We also wishto thank A. P. Pong, M. L.
34、 Lee, and E. Nordbrock for their constant support andencouragement. The first author also wishes to express his appreciation to his wife, Yueh-Ji,and their daughters, Emily and Lilly, for their patience and understanding during thepreparation of this book. Finally, we are fully responsible for any e
35、rrors remaining in the book. The viewsexpressed are those of the authors and are not necessarily those of Covance, Inc. and theNational Cheng-Kung University.SHEIN-CHUNG CHOWJEN-PEI LIUPrinceton, New JerseyTainan, TaiwanOctober 1997PREFACE TO THE FIRST EDITION xiiiflast.qxd 27.10.03 15:13 Page xiiif
36、last.qxd 27.10.03 15:13 Page xiv11INTRODUCTION1.1 WHAT ARE CLINICAL TRIALS?Clinical trials are clinical investigations. They have evolved with different meanings bydifferent individuals and organizations at different times. For example, Meinert (1986)indicates that a clinical trial is a research act
37、ivity that involves administration of a testtreatment to some experimental unit in order to evaluate the treatment. Meinert (1986) alsodefines a clinical trial as a planned experiment designed to assess the efficacy of a treat-ment in humans by comparing the outcomes in a group of patients treated w
38、ith the testtreatment with those observed in a comparable group of patients receiving a control treat-ment, where patients in both groups are enrolled, treated, and followed over the same timeperiod. This definition indicates that a clinical trial is used to evaluate the effectiveness ofa treatment.
39、 Piantadosi (1997) simply defined a clinical trial as an experimental testingmedical treatment on human subject. On the other hand, Spilker (1991) considers clinicaltrials as a subset of clinical studies that evaluate investigational medicines in phases I, II,and III, the clinical studies being the
40、class of all scientific approaches to evaluate medicaldisease preventions, diagnostic techniques, and treatments. This definition is somewhatnarrow in the sense that it restricts to the clinical investigation conducted by pharmaceuti-cal companies during various stages of clinical development of pha
41、rmaceutical entitieswhich are intended for marketing approval. The Code of Federal Regulations (CFR)defines a clinical trial as the clinical investigation of a drug that is administered or dis-pensed to, or used involving one or more human subjects (21 CFR 312.3). Three importantkey words in these d
42、efinitions of clinical trials are experimental unit, treatment, and evalu-ation of the treatment.Design and Analysis of Clinical Trials: Concepts and Methodologies, Second Edition By Shein-Chung Chow and Jen-pei LiuISBN 0-471-24985-8 Copyright 2004 John Wiley magnetic resonanceimaging (MRI) with a c
43、ontrast imaging agent as a combination of diagnostic test and adrug for enhancement of diagnostic enhancement; or a class III antiarrhythmic agent andan implanted cardioverter defibrillator as a combination of a drug and a medical device fortreatment of patients with ventricular arrhythmia. As a res
44、ult, a treatment is any interven-tion to be evaluated in human subjects regardless that it is a new intervention to be testedor serves as a referenced control group for comparison.EvaluationIn his definition of clinical trials, Meinert (1986) emphasizes the evaluation of efficacy ofa test treatment.
45、 It, however, should be noted that the assessment of safety of an interven-tion such as adverse experiences, elevation of certain laboratory parameters, or change infindings of physical examination after administration of the treatment is at least as impor-tant as that of efficacy. Recently, in addi
46、tion to the traditional evaluation of effectivenessand safety of a test treatment, clinical trials are also designed to assess quality of life, phar-macogenomics, and pharmacoeconomics such as cost-minimization, cost-effectiveness,and cost-benefit analyses to human subjects associated with the treat
47、ment under study. It istherefore recommended that clinical trials should not only evaluate the effectiveness andsafety of the treatment but also assess quality of life, impact of genetic factors, pharma-coeconomics, and outcomes research associated with the treatment.Throughout this book we will def
48、ine a clinical trial as a clinical investigation in which treat-ments are administered, dispensed, or used involving one or more human subjects for evalua-tion of the treatment. By this definition, the experimental units are human subjects either witha pre-existing disease under study or healthy. Un
49、less otherwise specified, clinical trials in thisbook are referred to as all clinical investigations in human subjects that may be conducted by2 INTRODUCTIONc01.qxd 27.10.03 12:41 Page 2pharmaceutical companies, clinical research organizations such as the U.S. National Institutesof Health (NIH), university hospitals, or any other medical research centers.1.2 HISTORY OF CLINICAL TRIALSWe humans since our early days on earth have been seeking or trying to identify some inter-ventions, whether they be a procedure or a drug, to remedy ailments that inflict ourselvesand our loved ones. In this
