Một cuốn sách hay về các tiến bộ trong sinh thống kê dùng trong thử nghiệm lâm sàng. Sách gồm các phần. Part I METHODS FOR EARLY TRIALS 1. Bayesian Methods for Cancer Phase I Clinical Trials 1 James S. Babb and Andre´ Rogatko 2. Design of Early Trials in Stem Cell Transplantation: A Hybrid FrequentistBayesian Approach 41 Nancy L. Geller, Dean Follmann, Eric S. Leifer, and Shelly L. Carter Part II METHODS FOR RANDOMIZED TRIALS 3. Design and Analysis of Therapeutic Equivalence Trials 53 Richard M. Simon 4. Adaptive TwoStage Clinical Trials 67 Michael A. Proschan 5. Design and Analysis of Cluster Randomization Trials 83 David M. Zucker vii6. Design and Analysis of Clinical Trials with Multiple Endpoints 101 Nancy L. Geller 7. Subgroups and Interactions 121 Dean Follmann 8. A Class of Permutation Tests for Some TwoSample Survival Data Problems 141 Joanna H. Shih and Michael P. Fay 9. Bayesian Reporting of Clinical Trials 161 Simon Weeden, Laurence S. Freedman, and Mahesh Parmar Part III MORE COMPLEX PROBLEMS 10. Methods Incorporating Compliance in Treatment Evaluation 189 Juni Palmgren and Els Goetghebeur 11. Analysis of Longitudinal Data with Missingness 213 Paul S. Albert and Margaret C. Wu 12. Statistical Issues Emerging from Clinical Trials in HIV Infection 237
ADVANCES IN CLINICAL TRIAL BIOSTATISTICS edited by NANCY L GELLER National Heart, Lung, and Blood Institute National Institutes of Health Bethesda, Maryland, U.S.A MARCEL MARCEL DEKKER, INC DEKKER - NEWYORK BASEL This book was edited by Nancy L Geller in her private capacity The views expressed not necessarily represent the views of NIH, DHHS, or the United States Although great care has been taken to provide accurate and current information, neither the author(s) nor the publisher, nor anyone else associated with this publication, shall be liable for any loss, damage, or liability directly or indirectly caused or alleged to be caused by this book The material contained herein is not intended to provide specific advice or recommendations for any specific situation Trademark notice: Product or corporate names may be trademarks or registered trademarks and are used only for identification and explanation without intent to infringe Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBN: 0-8247-9032-4 This book is printed on acid-free paper Headquarters Marcel Dekker, Inc., 270 Madison Avenue, New York, NY 10016, U.S.A tel: 212-696-9000; fax: 212-685-4540 Distribution and Customer Service Marcel Dekker, Inc., Cimarron Road, Monticello, New York 12701, U.S.A tel: 800-228-1160; fax: 845-796-1772 Eastern Hemisphere Distribution Marcel Dekker AG, Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland tel: 41-61-260-6300; fax: 41-61-260-6333 World Wide Web http://www.dekker.com The publisher offers discounts on this book when ordered in bulk quantities For more information, write to Special Sales/Professional Marketing at the headquarters address above Copyright n 2004 by Marcel Dekker, Inc All Rights Reserved Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage and retrieval system, without permission in writing from the publisher Current printing (last digit): 10 PRINTED IN THE UNITED STATES OF AMERICA Biostatistics: A Series of References and Textbooks Series Editor Shein-Chung Chow Vice President, Clinical Biostatistics and Data Management Millennium Pharmaceuticals, Inc Cambridge, Massachusetts Adjunct Professor Temple University Philadelphia, Pennsylvania Design and Analysis of Animal Studies in Pharmaceutical Development, edited by Shein-Chung Chow and Jen-pei Liu Basic Statistics and Pharmaceutical Statistical Applications, James E De Muth Design and Analysis of Bioavailability and Bioequivalence Studies, Second Edition, Revised and Expanded, Shein-Chung Chow and Jen-pei Liu Meta-Analysis in Medicine and Health Policy, edited by Dalene K Stangl and Donald A Berry Generalized Linear Models: A Bayesian Perspective, edited by Dipak K Dey, Sujit K Ghosh, and Bani K Mallick Difference Equations with Public Health Applications, Lemuel A Moye and Asha Seth Kapadia Medical Biostatistics, Abhaya lndrayan and Sanjeev B Sarrriukaddam Statistical Methods for Clinical Trials, Mark X Norleans Causal Analysis in Biomedicine and Epidemiology: Based on Minimal Sufficient Causation, Mike1Aickin 10 Statistics in Drug Research: Methodologies and Recent Developments, Shein-Chung Chow and Jun Shao 11 Sample Size Calculations in Clinical Research, Shein-Chung Chow, Jun Shao, and Hansheng Wang 12 Applied Statistical Designs for the Researcher, Daryl S Paulson 13 Advances in Clinical Trial Biostatistics, Nancy L Geller ADDITIONAL VOLUMES IN PREPARATION Series Introduction The primary objectives of the Biostatistics series are to provide useful reference books for researchers and scientists in academia, industry, and government, and also to offer textbooks for undergraduate and/or graduate courses in the area of biostatistics The series provides comprehensive and unified presentations of statistical designs and analyses of important applications in biostatistics, such as those in biopharmaceuticals A wellbalanced summary is given of current and recently developed statistical methods and interpretations for both statisticians and researchers/scientists with minimal statistical knowledge who are engaged in the field of applied biostatistics The series is committed to presenting easy-to-understand, state-of-the-art references and textbooks In each volume, statistical concepts and methodologies are illustrated through real-world examples whenever possible Clinical research is a lengthy and costly process that involves drug discovery, formulation, laboratory development, animal studies, clinical development, and regulatory submission This lengthy process is necessary not only for understanding the target disease but also for providing substantial evidence regarding efficacy and safety of the pharmaceutical compound under investigation prior to regulatory approval In addition, it provides assurance that the drug products under investigation will possess good characteristics such as identity, strength, quality, purity, and stability after regulatory approval For this purpose, biostatistics plays an imporiii iv Series Introduction tant role in clinical research not only to provide a valid and fair assessment of the drug product under investigation prior to regulatory approval but also to ensure that the drug product possesses good characteristics with the desired accuracy and reliability This volume provides a comprehensive summarization of recent developments regarding methodologies in design and analysis of studies conducted in clinical research It covers important topics in early-phase clinical development such as Bayesian methods for phase I cancer clinical trials and late-phase clinical development such as design and analysis of therapeutic equivalence trials, adaptive two-stage clinical trials, and cluster randomization trials The book also provides useful approaches to critical statistical issues that are commonly encountered in clinical research such as multiplicity, subgroup analysis, interaction, and analysis of longitudinal data with missing values It will be beneficial to biostatisticians, medical researchers, and pharmaceutical scientists who are engaged in the areas of clinical research and development Shein-Chung Chow Preface As the medical sciences rapidly advance, clinical trials biostatisticians and graduate students preparing for careers in clinical trials need to maintain knowledge of current methodology Because the literature is so vast and journals are published so frequently, it is difficult to keep up with the relevant literature The goal of this book is to summarize recent methodology for design and analysis of clinical trials arranged in standalone chapters The book surveys a number of aspects of contemporary clinical trials, ranging from early trials to complex modeling problems Each chapter contains enough references to allow those interested to delve more deeply into an area A basic knowledge of clinical trials is assumed, along with a good background in classical biostatistics The chapters are at the level of journal articles in Biometrics or Statistics in Medicine and are meant to be read by second- or third-year biostatistics graduate students, as well as by practicing biostatisticians The book is arranged in three parts The first consists of two chapters on the first trials undertaken in humans in the course of drug development (Phase I and II trials) The second and largest part is on randomized clinical trials, covering a variety of design and analysis topics These include design of equivalence trials, adaptive schemes to change sample size during the course of a trial, design of clustered randomized trials, design and analysis of trials with multiple primary endpoints, a new method for survival analysis, and how to report a Bayesian randomized trial The third section deals v vi Preface with more complex problems: including compliance in the assessment of treatment effects, the analysis of longitudinal data with missingness, and the particular problems that have arisen in AIDS clinical trials Several of the chapters incorporate Bayesian methods, reflecting the recognition that these have become acceptable in what used to be a frequentist discipline The 20 authors of this volume represent five countries and 10 institutions Many of the authors are well known internationally for their methodological contributions and have extensive experience in clinical trials practice as well as being methodologists Each chapter gives real and relevant examples from the authors’ personal experiences, making use of a wide range of both treatment and prevention trials The examples reflect work in a variety of fields of medicine, such as cardiovascular diseases, neurological diseases, cancer, and AIDS While it was often the clinical trial itself that gave rise to a question that required new methodology to answer, it is likely that the methods will find applications in other medical fields In this sense, the contributions are examples of ‘‘ideal’’ biostatistics, transcending the boundary between statistical theory and clinical trials practice I wish to express my deep appreciation to all the authors for their patience and collegiality and for their fine contributions and outstanding expositions I also thank my husband for his constant encouragement and Marcel Dekker, Inc., for their continuing interest in this project Nancy L Geller Contents Series Introduction Preface Contributors iii v ix Part I METHODS FOR EARLY TRIALS Bayesian Methods for Cancer Phase I Clinical Trials James S Babb and Andre´ Rogatko Design of Early Trials in Stem Cell Transplantation: A Hybrid Frequentist-Bayesian Approach Nancy L Geller, Dean Follmann, Eric S Leifer, and Shelly L Carter 41 Part II METHODS FOR RANDOMIZED TRIALS Design and Analysis of Therapeutic Equivalence Trials Richard M Simon 53 Adaptive Two-Stage Clinical Trials Michael A Proschan 67 Design and Analysis of Cluster Randomization Trials David M Zucker 83 vii viii Contents Design and Analysis of Clinical Trials with Multiple Endpoints Nancy L Geller Subgroups and Interactions Dean Follmann A Class of Permutation Tests for Some Two-Sample Survival Data Problems Joanna H Shih and Michael P Fay Bayesian Reporting of Clinical Trials Simon Weeden, Laurence S Freedman, and Mahesh Parmar Part III 10 101 121 141 161 MORE COMPLEX PROBLEMS Methods Incorporating Compliance in Treatment Evaluation Juni Palmgren and Els Goetghebeur 11 Analysis of Longitudinal Data with Missingness Paul S Albert and Margaret C Wu 12 Statistical Issues Emerging from Clinical Trials in HIV Infection Abdel G Babiker and Ann Sarah Walker Index of Abbreviations Index of Clinical Trials Used as Examples Subject Index 189 213 237 273 277 279 Index of Abbreviations* AAD = antiarrhythmic drug, 135 AIDS = acquired immunodeficiency syndrome, 237 ALR = approximate likelihood ratio, 107 ARMA = autoregressive moving average, 217 AUC = area under the curve, 3, 255 BCD = bleomycin, cyclophophamide and dactinomycin (combination chemotherapy), 165 CABG = coronary artery bypass graft, 129 CAD = coronary artery disease, 135 CDDP = cisplatin, 163 CDF = cumulative distribution function, 20 CE = conditional error, 74 CI = confidence interval, 163, 252 CI = credible interval, 168 CLC = centered linear combination, 105 CRM = continual reassessment method, 20 CRML = maximum likelihood based version of CRM, 23 DBP = diastolic blood pressure, 122 ddC = zalcitabine, 244 ddl = didanosine, 244 DiM = difference in means, 143 DLT = dose-limiting toxicity, DOX = doxorubicin, 163 DPT = Distribution Permutation Test, 141 DSMB = Data and Safety Monitoring Board, 130 *The page reference gives the first use of an abbreviation in a chapter 273 274 EM Algorithm = expectation maximization algorithm, 145, 249 EOI = European Osteosarcoma Intergroup, 163 EORTC = European Organization for Research on the Treatment of Cancer, 163 EWOC = efficient dose escalation with overdose control, 23 GEE = generalized estimating equations, 92, 217, 247 GLS = generalized least squares, 105 HAART = highly active antiretroviral therapies, 238 HDMTX = high dose methotrexate, 165 HIV = human immunodeficiency virus, 237 HR = hazard ratio, 163 ICC = intraclass correlation coefficient, 84 ICD = implantable cardioverter defibrillator, 135 IF = variance inflation factor, 84 IPS = Internal Pilot Study, 68 ITT = intention to treat, 260 IWD = integrated weighted difference, 145 LATE = local average treatment effect, 191 LD10 = lethal dose producing 10% mortality, LHR = log hazard ratio, 167, 181 LOCF = last observation carried forward, 256 LV = left ventricular, 129 LVEF = left ventricular ejection fraction, 136 Index of Abbreviations MAR = missing at random, 220 MCAR = missing completely at random, 219 MC = Mantel-Ciminera (test), 152 MCRD = minimally clinically relevant difference, 69 MCRTD = minimally clinically relevant treatment difference, 67–68 MEMS = medical events monitoring system, 191 MRC = (British) Medical Research Council, 163, 237 MSC = cluster mean square, 89 MSE = error mean square, 89 MW = Mann-Whitney, 143 MTD = maximum tolerated dose, NIH SS = National Institutes of Health Stroke Scale, 113 NINDS = National Institute of Neurological Diseases and Stroke, 113 NPMLE = non-parametric maximum likelihood estimator, 145 NRTI = nucleoside analog reverse transcriptase inhibitors, 240 OLS = ordinary least squares, 104, 204 PBSC = peripheral blood stem cell, 42 PFS = progression-free survival, 163 PPW = Peto-Prentice-Wilcoxon (test), 154 PTCA = percutaneous transluminal coronary angioplasty, 129 PTE = proportion of treatment effect explained, 253 Index of Abbreviations RE = relative efficiency, 95 RR = relative risk, 245 SIOP = Socie´te´ Internationale d’Oncologie Paedriatrique, 163 t-PA = tissue plasminogen activator, 113 UKCCSG = United Kingdom Children’s Cancer Study Group, 163 275 VCR = vincristine, 153 VF = ventricular fibrillation, 135 VT = ventricular tachycardia, 135 WBC = white blood cell count, 165 WKM = weighted Kaplan-Meier, 145 WMW = weighted Mann-Whitney, 143 ZDV = zidovudine, 237 Index of Clinical Trials Used as Examples Active control trial of a new antiestrogen, 65 AVID (Antiarrhythmics Versus Implantable Defibrillator), 135 BARI (Bypass Angioplasty Revascularization Investigation), 129 Blood Pressure Reduction Trial (example of partial compliance), 206 Buprenorphine versus methodone for the treatment of addiction, 214 Breast Cosmesis Study, 148 CATCH (Child and Adolescent Trial for Cardiovascular Health), 83 COMMIT (Community Health Trial for Smoking Cessation), 83 CONCORDE (randomized doubleblind controlled trial of immediate and deferred zidovudine in symptom-free HIV infection), 242 DASH (Dietary Alterations to Stop Hypertension), 70 DELTA (randomized double-blind controlled trial comparing combinations of zidovudine plus didanosine or zalcitabine with zidovudine alone in HIV-infected individuals), 240 Diabetic Retinopathy Study, 154 Felbmate versus placebo for intractable partial epilepsy, 214 GI Tumor Study, 147 HLA matching and skin graft survival, 156 Intermittent Positive Pressure Breathing (IPPB) Trial, 214 Jerusalem Study of Hygiene Education in Kindergartens, 85 277 278 LRC-CPPT (Lipid Research Clinics Coronary Primary Prevention Trial), 190 NINDS (National Institute of Neurological Diseases and Stroke) t-PA (Tissue Plasminogen Activator) Trial, 113 PATHS (Prevention and Treatment of Hypertension Study), 123 Phase I trial of 5-FU in combination with leucovorin and topotecan, Index of Clinical Trials Used as Examples Phase I trial of PNU, Phase I/II study in peripheral blood stem cell transplantation, 43, 48 Prostate Cancer Clinical Trial, 153 Randomized trial of two regimens of chemotherapy in operable osteosarcoma, 162 Tamoxifen Prevention Trial, 65 Subject Index Accelerated titration designs, 25 Accepting the null hypothesis, 55 Active control clinical trials, 53–65 limitations, 65 reversed decision structure for, 56 Adaptive design, 67–81 Additive cluster effect, 85 Adherence, 190 AIDS-defining event, 240 Alpha inflation, 73 Approximate likelihood ratio test, 107, 110 Area under the curve, Autoregressive moving average correlation structure, 217 Baseline covariate(s) 137, 138 Bayes’ theorem, 7, 10, 16 Bayesian analysis, 164 approach to therapeutic equivalence, 60 design for active control trials, 60 [Bayesian] factor, 186 meta-analysis, 254 methods, 1–36, 45–51, 161–188 feasible, 20 optimal, 22 randomized trial(s), 161–188 stopping rules for safety 42, 45–51 Bias, 68 in estimation, 131 Bioequivalence, 54 Bivariate response, 49 Bonferroni, 103 128 Causal framework, 190 Censored data, 241 interval-, 142, 146, 260, 264 left-, 256 matched-pair-, 142 related to treatment, 156–157 related to treatment and stratum, 157 right-, 141, 146 Centered linear combination, 105 279 280 Clinical trial(s) Bayesian, 1–36, 45–51, 166–186 noninformative, 58 phase I, 1, 19, 32, 41, 43 phase II, 32, 41, 43 Cluster mean, 90 mean square, 89 randomization, 83 Compliance, 189–211, 238 always takers, 195 conditional on covariates, 192 -covariate interaction, 191 defiers, 195 full, 193, 201, 202 noncompliance, 243, 260 nonrandom noncompliance, 190, 191, 197 null, 193, 201 numerical scores, 203 partial, 201, 202 pattern, 189 potential, 191 prediction(s), 200 Composite endpoint, 246 Conditional error function, 74 circular, 76–78 linear, 75, 76 Conditional power, 74, 78 Confidence interval for difference between treatments, 57 Constrained escalation, 25 Continual reassessment method, 13, 20, 23, 26, 27 Correlated endpoints, 102, 109 estimates, 132 Counting process, 251 Covariate(s) adjustment, 91 cluster level, 91 individual level, 5, 91 Subject Index Coverage distribution, 23 Cox proportional hazard model [see Model(s)] Credible interval, 35, 168 Crossover design, 54 Curve-free model [see Model(s)] Data and Safety Monitoring Board, 45, 130 Delta method, 253 Design accelerated titration, 25 adaptive (see Adaptive design) combining frequentist and Bayesian, 51 crossover, 54 optimal Bayesian feasible, 22 Difference in means functional, 143 Differential treatment effect, 122, 123 Distribution coverage, 23 Dirichlet, 16 marginal posterior, 17, 20 marginal prior, 10 permutation test(s), 141–160 posterior, 7–8, 16–19, 23, 28, 35, 46, 60–63 prior, 6, 8, 10–17, 19, 26, 31–35, 46, 47, 60–63, 167 Dose escalation, 12, 45 finding, 44 with drug combinations, 4, 28 limiting toxicity, 16, 23, 30, 33 -toxicity relationship, 6, 8, 15, 29, 32 Dropout, 214, 219 informative, 224, 241, 256 nonignorable, 220, 222–224 Effectiveness, 64, 189, 244 Efficacy, 32, 41, 167, 185, 189, 244 EM algorithm, 145, 249 Subject Index Empirical estimation of treatment effect, 78 Equivalence trials (see also Bayesian methods), 53–66 based on confidence intervals, 58– 59 need for large sample size, 56 Error mean square, 89 Estimating equations generalized, 220, 226 linear, 205 Estimation of survival nonparametric maximum likelihood, 145 self-consistent, 149 Exchangeable, 144 Exclusion restriction, 194, 196 Factorial design(s), 238 Failure time methods, 256, 257, 259 Failure to reject null hypothesis, 55, 56 Familywise Type I error control, 115 Feasibility bound, 20, 23 Feasible Bayesian methods (see Bayesian methods) Feiller’s theorem, 253 Fibonacci sequence, 24 Fisher information, 19 Fisher’s product of independent p-values, 78 Fleming’s phase II design, 45 Frailty [see Model(s), frailty] Frequentist, 42 Functional, 145–147 Gaussian quadrature, 229 Generalized estimating equations, 247 Generalized linear models [see Model(s)] Group randomization, 83 Group sequential methods, 110, 116 (see also Interim analysis; Monitoring) 281 Heterogeneity, 121, 125, 127 HIV, 237–271 Homogeneity, 126 Hotelling’s T2, 107 Hyperbolic tangent, 8, 9, 11 Hyperparameters, 13, 15 Imputation, 221 Inflation factor, 239 Inmigration, 86 Instrumental variable(s), 197, 205 Integrated weighted difference(s), 145, 147 Intention to treat, 86, 87, 189, 197, 256, 261 Interaction, 204, 250 (see also Subgroups) estimation of, 133 qualitative, 122–123, 131–133, 137 quantitative, 122–123, 131, 137 test(s), 123, 125–127, 130–131, 133– 134, 137 likelihood ratio, 132 permutation, 130 power of, 126, 127 range, 133 Interim analysis, 71, 111 (see also Monitoring; Group sequential methods) Internal pilot study, 68 Interpretation section in clinical trial reports, 162, 180 Intraclass correlation, 89, 90, 94 Irregularly spaced observations, 217 Iterative convex majorant algorithm, 145 Kaplan-Meier estimate, 48, 148 weighted, 145, 157 Kappa coefficient, 90 Latent counts, 199 282 [Latent] subgroup, 196 variable(s), 192, 204, 208 Likelihood approximate, 106, 110 conditional, 231 function, 15, 16 marginal, 223 profile, 249 ratio test(s), 106 Linear combinations of endpoints, 104, 221 Link function, 217 logarithm, 258 logistic, 230 probit, 224 Local average treatment effect, 191, 194, 197 Logistic, logit model [see Model(s)] Logrank test, 141, 146 permutation form, 147, 152 stratified, 150, 152, 157 Longitudinal data, 213–236, 242, 255 continuous, 215 discrete, 216 Gaussian, 216, 226 Loss function, 7, 17, 35 Mann-Whitney functional, 143, 144, 146 scores, 146 weighted functional, 143, 146 Markov dependence, 219 Maximum likelihood, 23, 89 tolerated dose, 1, 2, 6, 14, 27, 30 Minimally clinically relevant treatment difference, 67–68 Minimal difference in efficacy, 54 Minimally worthwhile benefit, 183 Missing (see also Missingness) completely at random, 220 data, 87, 213 Subject Index [Missing] at random, 219–220 Missingness, 219, 226 (see also Missing) ignorable, 220 informative, 220, 223 monotonically, 219 nonignorable, 224, 230 non-monotonic, 219 nonrandom, 220 Model(s) accelerated failure time, 207, 261–263 causal, 261 conditional, 203, 206, 247, 250, 251 Cox proportional hazards, 62, 248, 258 curve-free, 10, 15, 29 diagnostic tools for, 208 discrete, 259 frailty, 247, 249, 251 generalized linear, 218 latent variable, 208 logistic, 9, 13 logit, marginal, 217, 247 misspecification, 250, 259 mixture, 92, 93, 97, 218 multistate, 250–251 overparameterized, 202 parsimonious, 201 pattern mixture, 222, 225 probit, random effects, 90, 216, 217 saturated, 199 selection, 222, 224, 225 shared random parameter, 220, 223, 225, 231 structural, 191, 197, 203, 205, 207 transition, 217, 219, 223, 230 Moment-based method, 90 Subject Index Monitoring, 116 (see also Group sequential methods; Interim analysis) Multiple comparisons, 108 Multiple endpoints, 101–119, 238 Multiple outcomes (see Multiple endpoints) Multiplicity, 127–129 Multivariate (see also Multiple endpoints) approach(es) to interaction, 133– 135 failure time data, 242 Newton-Raphson, 259 Noncompliance, 186, 187, 243, 260 nonrandom, 190, 191, 197 partial, 201 Noninformative clinical trial, 58 Nonparametric (see also Rank) maximum likelihood estimate, 145 mixture(s), 229 tests, 108 Nonproportional hazards, 141–142 Nonselection assumption, 190 Nonsequential design, Nuisance parameter(s), 10, 67, 73, 80 Null hypothesis global, 109, 112 strong, 130 One-sided test, 73, 102 bootstrap, 108 One-tailed test (see One-sided test) On-treatment analysis, 87, 260 Opportunistic infections, 240 Optimal antitumor activity, Bayesian feasible design, 22 dose, spacing, 29 Optimality, 44 Outmigration, 86 283 Paired Wilcoxon test, 155 Permissible dose set, 34 Permutation test(s), 85, 91, 130 (see also Distribution permutation tests) Peto-Prentice-Wilcoxon test, 154 Phase I clinical trials, 1–39, 41, 43 Phase II clinical trials, 32, 41–52 Pitman efficiency, 141 Placebo group, 65 synthetic, 59 Positive quadrant alternative, 76 Posterior (see also Distribution, posterior) enthusiastic, 185 expected loss, 7, 18 marginal distribution, 19, 20, 61 skeptical, 185 Preclinical studies, 4, 5, 8, 10, 36 Primary study population, 86 Primary unit of analysis, 98 Prior distribution (see also Distributions, prior) beliefs, 168 beta, 46, 48 correlated, 61 enthusiastic, 171 joint, 13, 31 noninformative, 11, 61, 63, 168 normal, 12, 60, 61 skeptical, 171 subjective, 11 uniform, 12, 34 vague, 11 Prioritized hypotheses, 50 Probit model [see Model(s)] Profile analysis, 215, 216 Proportion of treatment effect explained, 253 (see also Surrogate) Proportional hazards model [see Model(s), Cox proportional hazards] 284 Proportional odds regression (see Regression, proportional odds) Randomization cluster, 83 test(s), 84, 85 Rank (see also Nonparametric) invariant, 151 multivariate, 221 test(s), 144, 154 Ranking outcomes, 246 (see also Severity score) Range of equivalence, 168, 171 Recensoring, 262, 264 Reestimation of sample size, 68, 73, 75 Regression normal interval, 256 Poisson, 259 proportional odds, 33 Repeated measures, 88, 213–236, 241 Reporting of clinical trial results, 164 Resampling methods, 103, 104 Response, 177 Restricted design, 71 maximum likelihood, 89 test, 70 Safety and efficacy, 33 Sample size for cluster randomization trials, 94 for multiple endpoint trials, 105 for therapeutic equivalence trials, 58–59, 63–64 reestimation of, 68, 73, 75 Sandwich estimator, 205 Score tests, 141 Selection bias, 261 Self-consistent estimator, 149 Subject Index Semiparametric estimation, 204 [see also Model(s), Cox proportional hazards] Sensitivity, 162, 185 Sequential (see also Group sequential methods) design, search procedures, 21 Sequentially rejective procedure, 103, 128 Severity score, 133 (see also Ranking outcomes) Shrinkage estimator, 150, 151 Sign test, 154 Significance tests, over-reliance on, 57 Simon’s optimal phase II design, 41, 43 Simulation of power, 152, 155 studies, 34, 47, 49 Spending function, 115 Stable unit treatment value assumption, 194 Stein’s procedure, 69–71 Step-down closed procedure(s), 108, 110, 116 Stopping boundary, 47, 113–115 for futility, 77, 79 rule, 42, 47 Strata, stratum, 142, 149, 150 analysis by, 90 variance between, 150 variance within, 150 Strong monotonicity, 195 Subgroup(s), 121–138 (see also Interaction) analyses within separate, 122–125 misleading results, 125 treatment effect in, 132 Subjectivity, 162 Summary measures, 215 Surrogate, 239, 241, 252 Subject Index Survey of clinical opinion, 167 Survival, 179 nonparametric maximum likelihood, 145 progression-free, 178 self-consistent estimate, 149 Target dose, Therapeutic equivalence trials (see Equivalence trials) Time-dependent treatment effect, 245 Toxicity, 1, 16 binary, continuous, 25 multinomial, 26 multiple grades, 26 Transplantation, 42 Treatment changes, 248 -free response, 191 transient effect, 263 285 Two-sample tests, 110 for multiple endpoints, 101–119 Two-stage analysis, 91, 227 design, 68, 77 Type I error inflation, 68, 85 strong control, 109, 111, 115 weak control, 109 Up-down design, 24 Versatile test(s), 155 Wald-type test, 105, 110, 113 Weighted estimating equations, 226 least squares, 224 Mann-Whitney functional, 143 Schoenfeld residuals, 245 ... and analysis of clinical trials arranged in standalone chapters The book surveys a number of aspects of contemporary clinical trials, ranging from early trials to complex modeling problems Each... Longitudinal Data with Missingness Paul S Albert and Margaret C Wu 12 Statistical Issues Emerging from Clinical Trials in HIV Infection Abdel G Babiker and Ann Sarah Walker Index of Abbreviations Index... phase I study of PNU214565 (PNU) involving patients with advanced adenocarcinomas of gastrointestinal origin (Babb and Rogatko, 2001) Previous clinical and preclinical studies demonstrated that