Methods and Applications of Statistics in Clinical Trials, Volume 2 (eBook)

N. BALAKRISHNAN, PHD, is Professor in the Department of Mathematics and Statistics at McMaster University, Canada. He is the author of over twenty books and is the coeditor of Encyclopedia of Statistical Sciences, Second Edition, also published by Wiley.

Cover 1
Title Page 5
Copyright Page 6
Contents 7
Contributors 21
Preface 25
1 Analysis of Over- and Underdispersed Data 27
1.1 Introduction 27
1.2 Overdispersed Binomial and Count Models 28
1.2.1 Overdispersed Binomial Model 28
1.2.2 Overdispersed Poisson Model 28
1.2.3 Example 30
1.3 Other Approaches to Account for Overdispersion 30
1.3.1 Generalized Linear Mixed Model 30
1.3.2 Zero-Inflated Models 32
1.4 Underdispersion 32
1.5 Software Notes 33
References 33
2 Analysis of Variance (ANOVA) 36
2.1 Introduction 36
2.2 Factors, Levels, Effects, and Cells 37
2.3 Cell Means Model 38
2.4 One-Way Classification 38
2.4.1 Example 1 38
2.5 Parameter Estimation 39
2.6 The R(.) Notation—Partitioning Sum of Squares 39
2.7 ANOVA—Hypothesis of Equal Means 41
2.8 Multiple Comparisons 42
2.9 Two-Way Crossed Classification 43
2.10 Balanced and Unbalanced Data 43
2.11 Interaction Between Rows and Columns 46
2.12 Analysis of Variance Table 46
References 50
3 Assessment of Health-Related Quality of Life 52
3.1 Introduction 52
3.2 Choice of HRQOL Instruments 53
3.3 Establishment of Clear Objectives in HRQOL Assessments 53
3.4 Methods for HRQOL Assessment 55
3.5 HRQOL as the Primary End Point 57
3.6 Interpretation of HRQOL Results 58
3.7 Examples 58
3.7.1 HRQOL in Asthma 58
3.7.2 HRQOL in Seasonal Allergy Rhinitis 60
3.7,3 Symptom Relief for Late-Stage Cancers 61
3.8 Conclusion 62
Acknowledgment 62
References 62
Further Reading 65
4 Bandit Processes and Response-Adaptive Clinical Trials: The Art of Exploration Versus Exploitation 66
4.1 Introduction 66
4.2 Exploration Versus Exploitation with Complete Observations 67
4.2.1 The Model of Markov Decision Processes 68
4.2.2 The Bandit Processes Under the Bayesian Approach 68
4.2.3 The Response-Adaptive Clinical Trials 71
4.3 Exploration Versus Exploitation with Censored Observations 72
4.4 Conclusion 74
Acknowledgments. 75
References 75
5 Bayesian Dose-Finding Designs in Healthy Volunteers 77
5.1 Introduction 77
5.2 A Bayesian Decision-Theoretic Design 78
5.3 An Example of Dose Escalation in Healthy Volunteer Studies 80
5.4 Discussion 85
References 87
6 Bootstrap 88
6.1 Introduction 88
6.2 Plug-In Principle 90
6.2.1 How Useful is the Bootstrap Distribution? 90
6.2.2 Other Population Estimates 91
6.2.3 Other Sampling Procedures 91
6.2.4 Other Statistics 91
6.3 Monte Carlo Sampling—The "Second Bootstrap Principle" 92
6.4 Bias and Standard Error 92
6.5 Examples 93
6.5.1 Relative Risk 93
6.5.2 Linear Regression 94
6.6 Model Stability 98
6.6.1 Logistic Regression 101
6.7 Accuracy of Bootstrap Distributions 103
6.7.1 Systematic Errors in Bootstrap Distributions 108
6.7.2 Bootstrap Distributions Are Too Narrow 108
6.8 Bootstrap Confidence Intervals 109
6.8.1 t Intervals 109
6.8.2 Percentile Intervals 110
6.8.3 Bootstrap t 110
6.8.4 BCa Intervals 112
6.8.5 Bootstrap Tilting Intervals 113
6.8.6 Importance Sampling Implementation 113
6.8.7 Confidence Intervals for Mean Arsenic Concentration 115
6.8.8 Implications for Other Situations 116
6.8.9 Comparing Intervals 116
6.9 Hypothesis Testing 117
6.10 Planning Clinical Trials 118
6.10.1 "What If ' Analyses— Alternate Population Estimates 119
6.11 How Many Bootstrap Samples Are Needed 121
6.11.1 Assessing Monte Carlo Variation 124
6.11.2 Variance Reduction 124
6.12 Additional References 125
References 125
7 Conditional Power in Clinical Trial Monitoring 128
7.1 Introduction 128
7.2 Conditional Power 128
7.3 Weight-Averaged Conditional Power or Bayesian Predictive Power 131
7.4 Conditional Power of a Different Kind: Discordance Probability 132
7.5 Analysis of a Randomized Trial 133
7.6 Conditional Power: Pros and Cons 134
References 135
8 Cost-Effectiveness Analysis 137
8.1 Introduction 137
8.2 Definitions and Design Issues 137
8.2.1 The Various Types of Analysis used in Economic Evaluation 137
8.2.2 The Economic Perspective 138
8.2.3 Choice of Comparator 138
8.2.4 Setting and Timescale 139
8.2.5 Sample Size 139
8.2.6 Economic Analysis Plans 139
8.3 Cost and Effectiveness Data 140
8.3.1 The Measurement of Costs 140
8.3.2 The Measurement of Effectiveness 140
8.3.3 Quality-of-Life Scales 141
8.4 The Analysis of Costs and Outcomes 141
8.4.1 The Comparison of Costs 141
8.4.2 The Incremental Cost- Effectiveness Ratio 141
8.4.3 The Incremental Net Benefit 142
8.4.4 The Cost-Effectiveness Acceptability Curve 144
8.5 Robustness and Generalizability in Cost-Effectiveness Analysis 146
8.5.1 Missing Data 146
8.5.2 Censored Data 146
8.5.3 Treatment Switches 146
8.5.4 Multicenter Trials and Pooling 147
8.5.5 Classic Sensitivity Analysis 147
8.5.6 Regression Models 147
8.5.7 Markov Models to Extrapolate Over Time 148
8.5.8 Examples of Modeling and Sensitivity Analysis 148
References 149
Further Reading 151
9 Cox-Type Proportional Hazards Models 152
9.1 Introduction 152
9.2 Cox Model for Univariate Failure Time Data Analysis 152
9.2.1 Hazard Rate Function 152
9.2.2 Model and Parameter Interpretation 153
9.2.3 Parameter Estimation and Inference 153
9.2.4 Stratified Population 154
9.3 Marginal Models for Multivariate Failure Time Data Analysis 155
9.3.1 Multiple Event Data 155
9.3.2 Clustered Failure Time Data 156
9.4 Practical Issues in Using the Cox Model 157
9.4.1 Tied Data 157
9.4.2 Time-Dependent Covariates 158
9.4.3 Censoring Mechanism 159
9.4.4 Assessing the Proportional Hazards Assumption 159
9.4.5 Sample Size Calculation for Time-to-Event Data 160
9.5 Examples 162
9.5.1 Application to Lung Study 162
9.5.2 Application to Framingham Heart Study 163
9.5.3 Application to Diabetes Study 163
9.6 Extensions 167
9.7 Softwares and Codes 167
9.7.1 R Code for the Sample Size Calculation 167
9.7.2 SAS and R Codes for Fitting Proportional Hazards Models 169
References 170
Further Reading 171
10 Empirical Likelihood Methods in Clinical Experiments 172
10.1 Introduction 172
10.2 Classical EL: Several Ingredients for Theoretical Evaluations 178
10.3 The Relationship Between Empirical Likelihood and Bootstrap Methodologies 180
10.4 Bayes Methods Based on Empirical Likelihoods 182
10.5 Mixtures of Likelihoods 182
10.6 An Example: ROC Curve Analyses Based on Empirical Likelihoods 183
10.7 Applications of Empirical Likelihood Methodology in Clinical Trials or Other Data Analyses 184
10.8 Concluding Remarks 184
Appendix 187
References 188
11 Frailty Models 192
11.1 Introduction 192
11.2 Univariate Frailty Models 193
11.3 Multivariate Frailty Models 196
11.3.1 Shared Frailty Model 196
11.3.2 Correlated Frailty Model 197
11.4 Software 197
References 198
12 Futility Analysis 200
12.1 Introduction 200
12.2 Common Statistical Approaches to Futility Monitoring 201
12.2.1 Statistical Background 201
12.2.2 Conditional Power and Stochastic Curtailment 201
12.2.3 Group Sequential Formulation of Futility Boundary 203
12.2.4 Other Statistical Approaches to Constructing Futility Boundaries 203
12.3 Examples 204
12.3.1 Optimal Duration of Tamoxifen in Breast Cancer 204
12.3.2 A Randomized Study of Antenatal Corticosteroids 205
12.4 Discussion 206
References 210
Further Reading 212
13 Imaging Science in Medicine I: Overview 213
13.1 Introduction 213
13.2 Advances in Medical Imaging 215
13.3 Evolutionary Developments in Imaging 216
13.3.1 Molecular Medicine 217
13.3.2 Human Vision 218
13.3.3 Image Quality 222
13.3.4 Image Display/ Processing 232
13.4 Conclusion 237
References 238
14 Imaging Science in Medicine, II: Basics of X-Ray Imaging 239
14.1 Introduction to Medical Imaging: Different Ways of Creating Visible Contrast Among Tissues 239
14.1.1 "On a New Kind of Ray" 240
14.1.2 Contrast from Differential Interaction of Imaging Probes with Tissues 241
14.1.3 Different Probes Interact with Different Tissues in Different Ways, Yielding Different Kinds of Contrast Information 242
14.1.4 X Rays, Light, and Other Forms of Electromagnetic Radiation 243
14.1.5 The Principal Concern in Medicine Is Usually "Good Enough" Contrast 246
14.2 What the Body Does to the X-Ray Beam: Subject Contrast From Differential Attenuation of the X-Ray Beam by Various Tissues 248
14.2.1 X-Ray Film of a Cracked Phalange 248
14.2.2 Generating the Beam at the Anode/Target of the X-Ray Tube 249
14.2.3 Exponential Attenuation of a Narrow Monochromatic Beam by a Homogeneous Medium 253
14.2.4 Both of the Two Principal Mechanisms for the Interaction of X-Ray Photons with Atomic Electrons, Photoelectric Absorption (PA) and Compton Scatter (CS), are Important, and for Different Reasons 255
14.2.5 What the Body Does to a Flat X-Ray Beam: Subject Contrast from Differential Attenuation 259
14.3 What the X-Ray Beam Does to the Body: Known Medical Benefits Versus Possible Radiogenic Risks 261
14.3.1 Dose of Ionizing Radiation, in Gray 262
14.3.2 Radiogenic Risk Comes from Damage to DNA 264
14.3.3 Deterministic and Teratogenic Radiation Health Effects 266
14.3.4 Stochastic Effects and the Ubiquitous Linear No-Threshold Dose- Risk Assumption 266
14.3.5 In Medicine, 1 Sievert = 1 Gray 267
14.3.6 Effective Dose Is Expressed in Sieverts, Even in Medicine 268
14.3.7 Special Considerations for Children 269
14.3.8 The Radiation Safety Component of the Quality Assurance Program 271
14.3.9 Dose to the Image Receptor 274
14.4 Capturing the Visual Image: Analog (20th Century) X-Ray Image Receptors 274
14.4.1 Screen-Film Radiography: What the Primary X-Ray Image Does to the Image Receptor 274
14.4.2 Fluoroscopy with an Image Intensifier Tube and Electronic Optical Camera 279
14.4.3 Image Quality 279
14.4.4 The Image Quality Component of the QA Program 286
15 Imaging Science in Medicine, III: Digital (21st Century) X-Ray Imaging 290
15.1 The Computer in Medical Imaging 290
15.1.1 A Bit About Bytes, etc. 291
15.1.2 Digital Images 292
15.1.3 Digital Image Processing: Enhancing Tissue Contrast, Signal-to- Noise, Edge Sharpness, etc. 297
15.1.4 Picture Archiving and Communication Systems 300
15.1.5 Image Analysis and Interpretation: Computer-Assisted Detection 302
15.1.6 Computer and Computer-Network Security 303
15.1.7 Liquid Crystal and Other Digital Displays 304
15.1.8 The Joy of Digital 304
15.2 The Digital Planar X-Ray Modalities: Computed Radiography and Digital Radiography and Fluoroscopy 305
15.2.1 Computed Radiography 307
15.2.2 Digital Radiography with an Active Matrix Flat-Panel Imager 308
15.3 Digital Fluoroscopy and Digital Subtraction Angiography 313
15.4 Digital Tomosynthesis: Planar Imaging in Three Dimensions 316
15.5 Computed Tomography: Superior Contrast in Three-Dimensional X-Ray Attenuation Maps 318
15.5.1 Raw Data: The Ray- Projection/Ray-Sum Measures the Total Attenuation Along a Geometric Ray 319
15.5.2 Filtered Back- Projection Reconstruction 323
15.5.3 Generations of CT Devices 327
15.5.4 CT Dose and QA 334
16 Intention-to-Treat Analysis 339
16.1 Introduction 339
16.2 Missing Information 339
16.2.1 Background 339
16.2.2 Ignorable Missing Data 340
16.2.3 Conditionally Ignorable Missing Data 341
16.2.4 Potential for Bias 341
16.3 The Intention-to-Treat Design 342
16.3.1 Withdrawal from Treatment Versus Withdrawal from Follow-Up 343
16.3.2 Investigator and Subject Training/ Education 343
16.3.3 The Intent-to-Treat Analysis 344
16.3.4 Intent-to-Treat Subset Analysis 344
16.3.5 LOCF Analysis 344
16.3.6 Structurally Missing Data 345
16.3.7 Worst Rank Analyses 345
16.4 Efficiency of the Intent-to- Treat Analysis 345
16.4.1 Power 345
16.4.2 Sample Size 346
16.5 Compliance-Adjusted Analyses 346
16.6 Conclusion 346
References 347
Further Reading 347
17 Interim Analyses 349
17.1 Introduction 349
17.2 Opportunities and Dangers of Interim Analyses 350
17.3 The Development of Techniques for Conducting Interim Analyses 351
17.4 Methodology for Interim Analyses 351
17.4.1 The Treatment Effect Parameter 352
17.4.2 Test Statistics for Use at Interim Analyses 352
17.4.3 Stopping Rules at Interim Analyses 353
17.4.4 Analysis Following a Sequential Trial 354
17.5 An Example: Statistics for Lamivudine 354
17.6 Interim Analyses in Practice 355
17.7 Conclusions 357
References 357
18 Interrater Reliability 360
18.1 Definition 360
18.2 The Importance of Reliability in Clinical Trials 360
18.3 How Large a Reliability Coefficient Is Large Enough? 361
18.4 Design and Analysis of Reliability Studies 361
18.5 Estimate of the Reliability Coefficient—Parametric 362
18.6 Estimation of the Reliability Coefficient—Nonparametric 362
18.7 Estimation of the Reliability Coefficient—Binary 363
18.8 Estimation of the Reliability Coefficient—Categorical 363
18.9 Strategies to Increase Reliability (Spearman–Brown Projection) 363
18.10 Other Types of Reliabilities 364
References 364
19 Intrarater Reliability 366
19.1 Introduction 366
19.2 Intrarater Reliability for Continuous Scores 366
19.2.1 Defining Intrarater Reliability 367
19.2.2 Statistical Inference 368
19.2.3 Optimizing the Design of the Intrarater Reliability Study 370
19.3 Nominal Scale Score Data 374
19.3.1 Intrarater Reliability: Single Rater and Two Replications 375
19.3.2 Intrarater Reliability: Single Rater and Multiple Replications 377
19.3.3 Statistical Inference 378
19.4 Ordinal and Interval Score Data 379
19.5 Concluding Remarks 380
References 381
Further Reading 382
20 Kaplan–Meier Plot 383
20.1 Introduction 383
20.2 Estimation of Survival Function 384
20.2.1 An Example 385
20.2.2 Practical Notes 385
20.2.3 Median Survival Time 388
20.2.4 More Practical Notes 388
20.3 Additional Topics 389
References 390
21 Logistic Regression 391
21.1 Introduction 391
21.2 Fitting the Logistic Regression Model 392
21.3 The Multiple Logistic Regression Model 394
21.4 Fitting the Multiple Logistic Regression Model 395
21.5 Example 395
21.6 Testing for the Significance of the Model 397
21.7 Interpretation of the Coefficients of the Logistic Regression Model 399
21.8 Dichotomous Independent Variable 399
21.9 Polytomous Independent Variable 401
21.10 Continuous Independent Variable 401
21.11 Multivariate Case 403
References 405
22 Metadata 406
22.1 Introduction 406
22.2 History/Background 406
22.2.1 A Metadata Example 406
22.2.2 Geospatial Data 407
22.2.3 Research Data and Statistical Software 408
22.2.4 Electronic Regulatory Submission 408
22.3 Data Set Metadata 409
22.3.1 Data Set-Level Metadata 409
22.3.2 Variable-Level Metadata 410
22.3.3 Value-Level Metadata 412
22.3.4 Item-Level Metadata 412
22.4 Analysis Results Metadata 414
22.5 Regulatory Submission Metadata 415
22.5.1 ICH Electronic Common Technical Document 415
22.5.2 FDA Guidance on eCTD Submissions 415
References 416
23 Microarray 418
23.1 Introduction 418
23.1.1 MammaPrint 419
23.2 What is a Microarray? 419
23.2.1 Types of Expression Microarrays 420
23.2.2 Microarrays Can Generate Reproducible Results 421
23.3 Other Array Technologies 421
23.3.1 Genotyping Using Expression Microarrays 421
23.3.2 Splicing Arrays 422
23.3.3 Exon Array 422
23.3.4 Tiling Array— Including Methylation Arrays 423
23.3.5 SNP Chip 423
23.3.6 ChlP-on-Chip 423
23.3.7 Protein Arrays 424
23.4 Define Objectives of the Study 424
23.5 Experimental Design for Microarray 425
23.5.1 Avoidance of Experimental Artifacts 425
23.5.2 Randomization, Blocking, and Blinding 425
23.5.3 Replication 425
23.5.4 Practice, Practice, Practice 426
23.5.5 Strict Experimental Practices 426
23.6 Data Extraction 427
23.6.1 Image Processing from cDNA and Long Oligo Arrays 427
23.6.2 Image Analysis of Affymetrix GeneChip Microarrays 427
23.6.3 Normalization of DNA Data 428
23.7 Microarray Informatics 428
23.8 Statistical Analysis 428
23.8.1 Class Prediction Analysis 428
23.8.2 Class Discovery Analysis 429
23.8.3 Class Differentiation Analysis 429
23.9 Annotation 430
23.10 Pathway, GO, and Class-Level Analysis Tools 430
23.11 Validation of Microarray Experiments 431
23.12 Conclusions 431
References 432
24 Multi-Armed Bandits, Gittins Index, and Its Calculation 442
24.1 Introduction 442
24.2 Mathematical Formulation of Multi-Armed Bandits 442
24.2.1 An Example 443
24.2.2 Solution Concept and the Gittins Index 444
24.2.3 Salient Features of the Model 444
24.3 Off-Line Algorithms for Computing Gittins Index 445
24.3.1 Largest-Remaining- Index Algorithm (Varaiya, Walrand, and Buyukkoc) 446
24.3.2 State-Elimination Algorithm (Sonin) 447
24.3.3 Triangularization Algorithm (Denardo, Park, and Rothblum) 449
24.3.4 Fast-Pivoting Algorithm (Niño-Mora) 451
24.3.5 An Efficient Method to Compute Gittins Index of Multiple Processes 453
24.4 On-Line Algorithms for Computing Gittins Index 454
24.4.1 Restart Formulation (Katehakis and Veinott) 455
24.4.2 Linear Programming Formulation (Chen and Katehakis) 455
24.5 Computing Gittins Index for the Bernoulli Sampling Process 456
24.5.1 Dynamic Programming Formulation (Gittins) 457
24.5.2 Restart Formulation (Katehakis and Derman) 457
24.5.3 Closed-Form Approximations 458
24.5.4 Qualitative Properties of Gittins Index 458
24.6 Conclusion 459
References 459
25 Multiple Comparisons 462
25.1 Introduction 462
25.2 Strong and Weak Control of the FWE 462
25.3 Criteria for Deciding Whether Adjustment is Necessary 463
25.4 Implicit Multiplicity: Two-Tailed Testing 464
25.5 Specific Multiple Comparison Procedures 465
25.5.1 Multiple Arms 465
25.5.2 Multiple End Points 466
25.5.3 Subgroup Analyses 468
25.5.4 Interim Monitoring 469
References 470
26 Multiple Evaluators 472
26.1 Introduction 472
26.2 Agreement for Continuous Data 473
26.2.1 Hypothesis Testing Approach 473
26.2.2 An Index Approach 473
26.2.3 An Interval Approach 475
26.3 Agreement for Categorical Data 475
26.3.1 Measuring Agreement between Two Evaluators 476
26.3.2 Extensions to Kappa and Other Approaches for Modeling Patterns of Agreement 477
26.3.3 Issues with Kappa 478
26.4 Summary and Discussion 479
Acknowledgments 479
References 479
27 Noncompartmental Analysis 483
27.1 Introduction 483
27.2 Terminology 484
27.2.1 Compartment 484
27.2.2 Parameter 484
27.2.3 Fixed Constant 484
27.2.4 Statistic 484
27.2.5 Comment 484
27.3 Objectives and Features of Noncompartmental Analysis 485
27.3.1 Advanced Noncompartmental Techniques 485
27.4 Comparison of Noncompartmental and Compartmental Models 486
27.5 Assumptions of NCA and Its Reported Descriptive Statistics 486
27.5.1 Assumption 1: Routes and Kinetics of Drug Absorption 487
27.5.2 Assumptions 2 to 4: Drug Distribution 487
27.5.3 Assumption 5: Routes of Drug Elimination 489
27.5.4 Assumption 6: Kinetics of Drug Elimination 489
27.5.5 Assumptions 7 and 8: Sampling Times and Monoexponential Decline 489
27.5.6 Assumption 9: Time Invariance of Disposition Parameters 490
27.5.7 Assumptions of Subsequent Descriptive Statistics 490
27.6 Calculation Formulas for NCA 490
27.6.1 NCA of Plasma or Serum Concentrations by Numerical Integration 491
27.6.2 NCA of Plasma or Serum Concentrations by Curve Fitting 496
27.6.3 NCA with Plasma or Serum Concentrations and Amounts in Urine 497
27.6.4 Superposition Methods and Deconvolution 497
27.7 Guidelines for Performance of NCA Based on Numerical Integration 498
27.7.1 How to Select Concentration Data Points for Estimation of Terminal Half-Life 498
27.7.2 How to Handle Samples Below the Quantification Limit 500
27.7.3 NCA for Sparse Concentration-Time Data 501
27.7.4 Reporting the Results of an NCA 501
27.7.5 How to Design a Clinical Trial That Is to Be Analyzed by NCA 502
27.8 Conclusions and Perspectives 503
Acknowledgments 503
References 503
Further Reading 508
28 Nonparametric ROC Analysis for Diagnostic Trials 509
28.1 Introduction 509
28.2 Different Aspects of Study Design 510
28.3 Nonparametric Models and Hypotheses 512
28.4 Point Estimator 513
28.5 Asymptotic Distribution and Variance Estimator 514
28.6 Derivation of the Confidence Interval 516
28.7 Statistical Tests 516
28.8 Adaptations for Cluster Data 516
28.9 Results of a Diagnostic Study 517
28.10 Summary and Final Remarks 520
Acknowledgments. 520
References 520
29 Optimal Biological Dose for Molecularly Targeted Therapies 522
29.1 Introduction 522
29.2 Phase I Dose-Finding Designs for Cytotoxic Agents 523
29.3 Phase I Dose-Finding Designs for Molecularly Targeted Agents 523
29.3.1 Dynamic De-escalating Designs 524
29.3.2 Dose Determination through Simultaneous Investigation of Efficacy and Toxicity 525
29.3.3 Individualized Maximum Repeatable Dose 526
29.3.4 Proportion Designs and Slope Designs 527
29.3.5 Generalized Proportion Designs 527
29.4 Discussion 528
References 529
Further Reading 531
30 Over- and Underdispersion Models 532
30.1 Introduction 532
30.2 Count Dispersion Models 534
30.2.1 Mixed Poisson 534
30.2.2 Compound Poisson 535
30.2.3 Weighted Poisson 536
30.2.4 Lagrangian Poisson 538
30.2.5 Semiparametric Poisson 539
30.3 Count Explanatory Models 540
30.3.1 Generalized Linear Models 541
30.3.2 Count Time Series Models 542
30.3.3 Nonparametric Models 544
30.4 Summary and Final Remarks 545
References 546
31 Permutation Tests in Clinical Trials 553
31.1 Randomization Inference—Introduction 553
31.2 Permutation Tests—How They Work 554
31.3 Normal Approximation to Permutation Tests 557
31.4 Analyze as You Randomize 558
31.5 Interpretation of Permutation Analysis Results 559
31.6 Summary 560
References 560
32 Pharmacoepidemiology, Overview 562
32.1 Introduction 562
32.2 The Case–Crossover Design 563
32.3 Confounding Bias 565
32.3.1 Missing Confounder Data 566
32.3.2 The Case–Time– Control Design 567
32.4 Risk Functions Over Time 569
32.5 Probabilistic Approach for Causality Assessment 571
32.6 Methods Based on Prescription Data 572
Acknowledgments 573
References 574
33 Population Pharmacokinetic and Pharmacodynamic Methods 577
33.1 Introduction 577
33.2 Terminology 578
33.2.1 Variable 578
33.2.2 Parameter 578
33.2.3 Variability 578
33.2.4 Uncertainty 578
33.2.5 Covariate (also Called Covariable) 579
33.2.6 Covariance 579
33.2.7 Mixed Effects Model 579
33.2.8 Subject Variability 579
33.3 Fixed Effects Models 579
33.3.1 Input-Output Models 579
33.3.2 Group Models 580
33.4 Random Effects Models 581
33.4.1 Parameter Variability Model 581
33.4.2 Residual Error Model 582
33.5 Model Building and Parameter Estimation 582
33.5.1 Hypothesis Testing 583
33.5.2 Objective Function 584
33.5.3 Bootstrap 584
33.5.4 Bayesian Estimation 585
33.5.5 Mixture Models 587
33.6 Software 587
33.6.1 NONMEM 588
33.6.2 Other Programs 588
33.7 Model Evaluation 588
33.7.1 Residuals 589
33.7.2 Predictive Checks 589
33.7.3 Prediction Discrepancy 589
33.7.4 Cross-Validation and External Validation 591
33.8 Stochastic Simulation 591
33.9 Experimental Design 591
33.10 Applications 592
33.10.1 Drug Development 592
33.10.2 Regulatory Science 592
33.10.3 Human and Disease Biology 592
Acknowledgment 593
References 593
Further Reading 594
34 Proportions: Inferences and Comparisons 596
34.1 Introduction 596
34.2 One-Sample Case 597
34.2.1 Point Estimation 597
34.2.2 Interval Estimation 597
34.2.3 Hypothesis Testing 601
34.2.4 Power and Sample Size Determination 603
34.2.5 One-Sample Summary 604
34.3 Two Independent Samples 604
34.3.1 Asymptotic Methods 605
34.3.2 Exact Methods 607
34.3.3 Bayesian Methods 610
34.3.4 Study Design and Power 611
34.3.5 Noninferiority and Equivalence Tests 612
34.3.6 Two-Sample Summary 613
34.4 Note on Software 614
References 615
35 Publication Bias 621
35.1 Publication Bias and the Validity of Research Reviews 621
35.2 Research on Publication Bias 622
35.2.1 Direct Evidence of Publication Bias 622
35.2.2 Indirect Evidence of Publication Bias 623
35.2.3 Clinical Significance of the Evidence 623
35.3 Data Suppression Mechanisms Related to Publication Bias 623
35.4 Prevention of Publication Bias 624
35.4.1 Trials Registries 625
35.4.2 Prospective Meta-Analysis 625
35.4.3 Thorough Literature Search 625
35.5 Assessment of Publication Bias 625
35.5.1 File Drawer Analysis (Failsafe N Approach) 626
35.5.2 Funnel Plots 626
35.5.3 Statistical Tests 627
35.5.4 Selection Models 629
35.5.5 Comparing the Results of the Different Methods 630
35.6 Impact of Publication Bias 631
References 631
Further Reading 633
36 Quality of Life 634
36.1 Background 634
36.1.1 Health-Related Quality of Life 634
36.2 Measuring Health-Related Quality of Life 635
36.2.1 Health Status Versus Patient Preferences 635
36.2.2 Objective Versus Subjective 637
36.2.3 Generic Versus Disease-Specific Instruments 637
36.2.4 Global Index Versus Profile of Domain- Specific Measures 637
36.2.5 Response Format 638
36.2.6 Period of Recall 638
36.2.7 Scoring 639
36.3 Development and Validation of HRQoL Measures 639
36.3.1 Development 639
36.3.2 Validation 639
36.3.3 Translation / Cross- Cultural Validation 641
36.3.4 Item Banking and Computer-Adaptive Testing 641
36.4 Use in Research Studies 641
36.4.1 Instrument Selection 641
36.4.2 Multiple End Points 642
36.4.3 Missing Data 642
36.5 Interpretation/Clinical Significance 643
36.5.1 Distributional Methods 643
36.5.2 Anchor-Based Methods 643
36.6 Conclusions 644
References 645
37 Relative Risk Modeling 648
37.1 Introduction 648
37.2 Why Model Relative Risks? 648
37.3 Data Structures and Likelihoods 649
37.4 Approaches to Model Specification 650
37.4.1 Empiric Models 650
37.4.2 Models for Extended Exposure Histories 653
37.4.3 Nonparametric Models 654
37.5 Mechanistic Models 655
References 656
38 Sample Size Considerations for Morbidity/Mortality Trials 659
38.1 Introduction 659
38.2 General Framework for Sample Size Calculation 659
38.3 Choice of Test Statistics 660
38.3.1 Parametric Tests 660
38.3.2 Nonparametric Tests 661
38.4 Adjustment of Treatment Effect 662
38.4.1 The Discrete Nonstationary Markov Process Model 663
38.4.2 Implementation 664
38.4.3 Illustration 665
38.5 Informative Noncompliance 665
References 666
39 Sample Size for Comparing Means 668
39.1 Introduction 668
39.2 One-Sample Design 669
39.2.1 Test for Equality 669
39.2.2 Test for Noninferiority / Superiority 669
39.2.3 Test for Equivalence 670
39.2.4 An Example 670
39.2.4 An Example 670
39.3 Two-Sample Parallel Design 671
39.3.1 Test for Equality 671
39.3.2 Test for Noninferiority / Superiority 671
39.3.3 Test for Equivalence 672
39.3.4 An Example 672
39.4 Two-Sample Crossover Design 672
39.4.1 Test for Equality 673
39.4.2 Test for Noninferiority / Superiority 673
39.4.3 Test for Equivalence 674
39.4.4 An Example 674
39.5 Multiple-Sample One-Way ANOVA 674
39.5.1 Pairwise Comparison 674
39.5.2 Simultaneous Comparison 675
39.5.3 An Example 675
39.6 Multiple-Sample Williams Design 676
39.6.1 Test for Equality 676
39.6.2 Test for Noninferiority / Superiority 677
39.6.3 Test for Equivalence 677
39.6.4 An Example 677
39.7 Discussion 677
References 678
40 Sample Size for Comparing Proportions 679
40.1 Introduction 679
40.2 One-Sample Design 680
40.2.1 Test for Equality 680
40.2.2 Test for Non-Inferiority / Superiority 680
40,2.3 Test for Equivalence 681
40.2.4 An Example 681
40.3 Two-Sample Parallel Design 681
40.3.1 Test for Equality 682
40.3.2 Test for Non-Inferiority / Superiority 682
40.3.3 Test for Equivalence 682
40.3.4 An Example 683
40.4 Two-Sample Crossover Design 683
40.4.1 Test for Equality 684
40.4.2 Test for Non-Inferiority / Superiority 684
40.4.3 Test for Equivalence 685
40.4.4 An Example 685
40.5 Relative Risk—Parallel Design 685
40.5.1 Test for Equality 686
40.5.2 Test for Non-Inferiority / Superiority 686
40.5.3 Test for Equivalence 686
40.5.4 An Example 687
40.6 Relative Risk—Crossover Design 687
40.6.1 Test for Equality 688
40.6.2 Test for Non-Inferiority / Superiority 688
40.6.3 Test for Equivalence 688
40.6.4 An Example 689
40.7 Discussion 689
References 689
41 Sample Size for Comparing Time-to-Event Data 690
41.1 Introduction 690
41.2 Exponential Model 690
41.2.1 Test for Equality 691
41.2.2 Test for Noninferiority / Superiority 692
41.2.3 Test for Equivalence 692
41.2.4 An Example 693
41.3 Cox's Proportional Hazards Model 693
41.3.1 Test for Equality 693
41.3.2 Test for Noninferiority / Superiority 694
41.3.3 Test for Equivalence 694
41.3.4 An Example 695
41.4 Log-Rank Test 695
41.4.1 An Example 696
41.5 Discussion 696
References 696
42 Sample Size for Comparing Variabilities 698
42.1 Introduction 698
42.2 Comparing Intrasubject Variabilities 698
42.2.1 Parallel Design with Replicates 699
42.2.2 Replicated Crossover Design 700
42.3 Comparing Intersubject Variabilities 702
42.3.1 Parallel Design with Replicates 703
42.3.2 Replicated Crossover Design 704
42.4 Comparing Total Variabilities 706
42.4.1 Parallel Designs Without Replicates 706
42.4.2 Parallel Design with Replicates 708
42.4.3 The Standard 2 x 2 Crossover Design 710
42.4.4 Replicated 2 x 2m Crossover Design 711
42.5 Discussion 713
References 713
43 Screening, Models of 715
43.1 Introduction 715
43.2 What is Screening? 715
43.3 Why Use Modeling? 717
43.4 Characteristics of Screening Models 718
43.4.1 Types of Model 718
43.4.2 Markov Framework for Modeling 718
43.5 A Simple Disease and Screening Model 719
43.6 Analytic Models for Cancer 721
43.7 Simulation Models for Cancer 730
43.8 Model Fitting and Validation 734
43.8.1 An Example of Model Fitting 737
43.8.2 An Application of the Model to Breast Cancer Screening 739
43.8.3 A Comparison of Two Models for Breast Cancer 739
43.9 Models for Other Diseases 741
43.10 Current State and Future Directions 742
References 743
44 Screening Trials 747
44.1 Introduction 747
44.2 Design Issues 747
44.3 Sample Size 748
44.4 Study Designs 749
44.4.1 Classic Two-Arm Trial That Addresses a Single Question 749
44.4.2 Designs for Investigating More than One Question in the Same Study 749
44.5 Analysis 751
44.5.1 Primary Analysis 751
44.5.2 Overall Mortality Analysis 751
44.5.3 Limited Mortality Analysis 752
44.5.4 Comparability 752
44.5.5 Secondary Analyses 753
44.6 Trial Monitoring 754
References 754
45 Secondary Efficacy End Points 757
45.1 Introduction 757
45.2 Literature Review 760
45.3 Review of Methodology for Multiplicity Adjustment and Gatekeeping Strategies for Secondary End Points 762
45.4 Summary 764
References 764
Further Reading 765
46 Sensitivity, Specificity, and Receiver Operator Characteristic (ROC) Methods 766
46.1 Evaluating a Single Binary Test Against a Binary Criterion 766
46.2 Evaluation of a Single Binary Test: ROC Methods 769
46.3 Evaluation of a Test Response Measured on an Ordinal Scale: ROC Methods 771
46.4 Evaluation of Multiple Different Tests 773
46.4.1 A Family of Tests 773
46.5 The Optimal Sequence of Tests 773
46.6 Sampling and Measurement Issues 775
46.6.1 Naturalistic Sampling 775
46.6.2 Prospective or Two- Stage Sampling 775
46,6.3 Retrospective (Case- Control) Sampling 776
46.7 Summary 776
References 777
47 Software for Genetics/Genomics 778
47.1 Introduction 778
47.2 Data Management 778
47.2.1 Data Storage and Retrieval 780
47.2.2 Data Visualization 780
47.2.3 Data Processors 780
47.2.4 Error Detection for Family and Genotype Data 787
47.2.5 Error Detection for Genetic Maps 788
47.3 Genetic Analysis 788
47.3.1 Summary Statistics 788
47.3.2 Familial Aggregation, Commingling, and Segregation Analysis 789
47.3.3 Linkage Analysis 789
47.3.4 Association Analysis 790
47.3.5 Linkage Disequilibrium and Transmission Disequilibrium Tests 791
47.3.6 Admixture Mapping 792
47.3.7 Haplotype Analysis 793
47.3.8 Software Suites 794
47.4 Genomic Analysis 794
47.5 Other 796
47.5.1 Power Calculations/ Simulation 796
47.5.2 Meta-analysis 796
References 796
Further Reading 802
48 Stability Study Designs 804
48.1 Introduction 804
48.2 Stability Study Designs 805
48.2.1 Full Design 805
48.2.2 Reduced Design 806
48.2.3 Other Fractional Factorial Designs 808
48.3 Criteria for Design Comparison 808
48.4 Stability Protocol 814
48.5 Basic Design Considerations 814
48.5.1 Impact of Design Factors on Shelf Life Estimation 814
48.5.2 Sample Size and Sampling Considerations 815
48.5.3 Other Issues 815
48.6 Conclusions 816
Acknowledgments 816
References 816
49 Subgroup Analysis 819
49.1 Introduction 819
49.2 The Dilemma of Subgroup Analysis 819
49.3 Planned Versus Unplanned Subgroup Analysis 820
49.4 Frequentist Methods 821
49.4.1 Significance Testing Within Subgroups 821
49.5 Testing Treatment by Subgroup Interactions 822
49.6 Subgroup Analyses in Positive Clinical Trials 823
49.7 Confidence Intervals for Treatment Effects within Subgroups 824
49.8 Bayesian Methods 825
References 826
50 Survival Analysis, Overview 828
50.1 Introduction 828
50.2 History 828
50.2.1 The Prehistory of Survival Analysis in Demography and Actuarial Science 828
50.2.2 The "Actuarial" Life Table and the Kaplan– Meier Estimator 829
50.2.3 Parametric Survival Models 830
50.2.4 Multistate Models 830
50.3 Survival Analysis Concepts 830
50.4 Nonparametric Estimation and Testing 831
50.5 Parametric Inference 833
50.6 Comparison with Expected Survival 833
50.7 The Cox Regression Model 833
50.8 Other Regression Models for Survival Data 835
50.9 Multistate Models 835
50.10 Other Kinds of Incomplete Observation 837
50.11 Multivariate Survival Analysis 837
50.12 Concluding Remarks 837
References 838
51 The FDA and Regulatory Issues 841
51.1 Caveat 841
51.2 Introduction 841
51.3 Chronology of Drug Regulation in the United States 842
51.4 FDA Basic Structure 846
51.5 IND Application Process 846
51.5.1 Types of IND 847
51.5.2 Parallel Track 848
51.5.3 Resources for Preparation of IND Applications 848
51.5.4 The First Step, the Phase I IND Application 849
51.5.5 Meetings with the FDA (http://www. fda. gov / cder / guidance / 2125fnl.pdf) 854
51.6 Drug Development and Approval Time Frame 855
51.6.1 Accelerated Development /Review 855
51.6.2 Fast Track Programs 856
51.6.3 Safety of Clinical Trials 856
51.7 NDA Process 857
51.7.1 Review Priority Classification 859
51.7.2 P—Priority Review 859
51.7.3 S—Standard Review 859
51.7.4 Accelerated Approval (21 CFR Subpart H Sec. 314.510) 860
51.8 U.S. Pharmacopeia and FDA 860
51.9 CDER Freedom of Information Electronic Reading Room 861
51.10 Conclusion 861
52 The Kappa Index 862
52.1 Introduction 862
52.2 The Kappa Index 862
52.2.1 Reliability in Two Applications of a Dichotomous Test 863
52.2.2 Reliability for Tests with Multiple Categorical Outcomes 864
52.3 Inference for Kappa via Generalized Estimating Equations 866
52.4 The Dependence of Kappa on Marginal Rates 868
52.5 General Remarks 869
References 869
53 Treatment Interruption 872
53.1 Introduction 872
53.2 Therapeutic TI Studies in HIV/AIDS 872
53.2.1 Overview 872
53.2.2 Design Issues in Therapeutic TI studies 876
53.3 Management of Chronic Disease 879
53.4 Analytic Treatment Interruption in Therapeutic Vaccine Trials 880
53.5 Randomized Discontinuation Designs 881
53.6 Final Comments 882
References 882
54 Trial Reports: Improving Reporting, Minimizing Bias, and Producing Better Evidence-Based Practice 886
54.1 Introduction 886
54.2 Reporting Issues in Clinical Trials 886
54.2.1 Incomplete Reporting of Trial Methods 886
54.2.2 Selective Reporting of Outcomes 887
54.2.3 Failure to Report Trials 887
54.2.4 Spin in the Interpretation of Trials 888
54.3 Moral Obligation to Improve the Reporting of Trials 889
54.4 Consequences of Poor Reporting of Trials 889
54.4.1 Impact on Knowledge Syntheses 889
54.5 Distinguishing Between Methodological and Reporting Issues 890
54.5.1 Cochrane Risk of Bias Tool 891
54.6 One Solution to Poor Reporting: CONSORT 2010 and CONSORT Extensions 892
54.7 Impact of CONSORT 892
54.8 Guidance for Reporting Randomized Trial Protocols: SPIRIT 896
54.9 Trial Registration 896
54.10 Final Thoughts 897
References 898
55 U.S. Department of Veterans Affairs Cooperative Studies Program 902
55.1 Introduction 902
55.2 History of the Cooperative Studies Program (CSP) 902
55.3 Organization and Functioning of the CSP 904
55.3.1 Planning Request 904
55.3.2 Planning Phase 906
55.3.3 Evaluation Phase 906
55.3.4 Implementation of the Trial 907
55.3.5 Final Analysis and Publication Phase 910
55.4 Roles of the Biostatistician and Pharmacist in the CSP 911
55.5 Ongoing and Completed Cooperative Studies (1972–2000) 913
55.6 Current Challenges and Opportunities 913
55.6.1 Changes in the VA Health Care System 913
55.6.2 Concerns about Patients' Rights in Research 917
55.6.3 Efficiency and Interdependence of the CSPCCs 918
55.6.4 Ensuring the Adequacy of Flow of Ideas and Training of Investigators 918
55.6.5 Partnering with Outside Organizations 919
55.7 Concluding Remarks 921
Acknowledgments 921
References 923
56 Women's Health Initiative: Statistical Aspects and Selected Early Results 927
56.1 Introduction 927
56.2 WHI Clinical Trial and Observational Study 927
56.3 Study Organization 929
56.4 Principal Clinical Trial Comparisons, Power Calculations, and Safety and Data Monitoring 929
56.5 Biomarkers and Intermediate Outcomes 934
56.6 Data Management and Computing Infrastructure 934
56.7 Quality Assurance Program Overview 936
56.8 Early Results from the WHI Clinical Trial 937
56.9 Summary and Discussion 938
Acknowledgments. 938
References 938
57 World Health Organization (WHO): Global Health Situation 940
57.1 Introduction 940
57.2 Program Activities to the End of the Twentieth Century 941
57.2.1 African Region 942
57.2.2 Region of the Americas 942
57.2.3 Eastern Mediterranean Region 943
57.2.4 European Region 943
57.2.5 Southeast Asia Region 943
57.2.6 Western Pacific Region 943
57.2.7 Main Activities at Global and Regional Levels 944
57.3 Vision for the Use and Generation of Data in the First Quarter of the Twenty-First Century 945
Reference 949
Further Reading 949
Index 951
EULA 963

"This book provides a good overview on most relevant topics for clinical trials." (Biometrical Journal, 1 October 2015)