Methods and Applications of Statistics in Clinical Trials

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.

Contributors xxiii

Preface xxix

1 Absolute Risk Reduction 1

1.1 Introduction 1

1.2 Preliminary Issues 1

1.3 Point and Interval Estimates for a Single Proportion 2

1.4 An Unpaired Difference of Proportions 5

1.5 Number Needed to Treat 8

1.6 A Paired Difference of Proportions 10

References 11

Further Reading 12

2 Accelerated Approval 14

2.1 Introduction 14

2.2 Accelerated Development Versus Expanded Access in the U.S.A 14

2.3 Sorting the Terminology—Which FDA Initiatives Do What? 15

2.4 Accelerated Approval Regulations: 21 C.F.R. 314.500, 314.520, 601.40 16

2.5 Stages of Drug Development and FDA Initiatives 16

2.6 Accelerated Approval Regulations: 21 CFR 314.500, 314.520, 601.40 17

2.7 Accelerated Approval with Surrogate Endpoints 18

2.8 Accelerated Approval with Restricted Distribution 20

2.9 Phase IV Studies/Post Marketing Surveillance 20

2.10 Benefit Analysis for Accelerated Approvals Versus Other Illnesses 21

2.11 Problems, Solutions, and Economic Incentives 22

2.12 Future Directions 24

References 25

Further Reading 26

3 AIDS Clinical Trials Group (ACTG) 27

3.1 Introduction 27

3.2 A Brief Primer on HIV/AIDS 27

3.3 ACTG Overview 28

3.4 ACTG Scientific Activities 29

3.5 Development of Potent Antiretroviral Therapy (ART) 29

3.6 Expert Systems and Infrastructure 36

References 37

4 Algorithm-Based Designs 40

4.1 Phase I Dose-Finding Studies 40

4.2 Accelerated Designs 43

4.3 Model-Based Approach in the Estimation of MTD 46

4.4 Exploring Algorithm-Based Designs With Prespecified Targeted Toxicity Levels 48

References 51

5 Alpha-Spending Function 53

5.1 Introduction 53

5.2 Alpha Spending Function Motivation 54

5.3 The Alpha Spending Function 56

5.4 Application of the Alpha Spending Function 57

5.5 Confidence Intervals and Estimation 59

5.6 Trial Design 59

5.7 Conclusions 61

References 61

Further Reading 63

6 Application of New Designs in Phase I Trials 65

6.1 Introduction 65

6.2 Objectives of a Phase I Trial 65

6.3 Standard Designs and Their Shortcomings 66

6.4 Some Novel Designs 67

6.5 Discussion 72

References 72

Further Reading 73

7 ASCOT Trial 74

7.1 Introduction 74

7.2 Objectives 74

7.3 Study Design 74

7.4 Results 75

7.5 Discussion and Conclusions 77

References 78

8 Benefit/Risk Assessment in Prevention Trials 80

8.1 Introduction 80

8.2 Types of B/RAs Performed in Prevention Trials 81

8.3 Alternative Structures of the Benefit/Risk Algorithm used in Prevention Trials 82

8.4 Methodological and Practical Issues with B/RA in Prevention Trials 84

References 87

9 Biased Coin Randomization 90

9.1 Randomization Strategies for Overall Treatment Balance 90

9.2 The Biased Coin Randomization Procedure 91

9.3 Properties 92

9.4 Extensions to the Biased Coin Randomization 92

9.5 Adaptive Biased Coin Randomization 94

9.6 Urn Models 99

9.7 Treatment Balance for Covariates 102

9.8 Application of Biased Coin Designs to Response-Adaptive Randomization 103

References 104

10 Biological Assay, Overview 106

10.1 Introduction 106

10.2 Direct Dilution Assays 108

10.3 Indirect Dilution Assays 109

10.4 Indirect Quantal Assays 113

10.5 Stochastic Approximation in Bioassay 116

10.6 Radioimmunoassay 117

10.7 Dosimetry and Bioassay 118

10.8 Semiparametrics in Bioassays 119

10.9 Nonparametrics in Bioassays 119

10.10 Bioavailability and Bioequivalence Models 120

10.11 Pharmacogenomics in Modern Bioassays 121

10.12 Complexities in Bioassay Modeling and Analysis 122

References 122

Further Reading 124

11 Block Randomization 125

11.1 Introduction 125

11.2 Simple Randomization 125

11.3 Restricted Randomization Through the Use of Blocks 126

11.4 Schemes Using a Single Block for the Whole Trial 130

11.5 Use of Unequal and Variable Block Sizes 131

11.6 Inference and Analysis Following Blocked Randomization 134

11.7 Miscellaneous Topics Related to Blocked Randomization 135

References 136

Further Reading 138

12 Censored Data 139

12.1 Introduction 139

12.2 Independent Censoring 140

12.3 Likelihoods: Noninformative Censoring 143

12.4 Other Kinds of Incomplete Observation 143

References 141

13 Clinical Data Coordination 146

13.1 Introduction 146

13.2 Study Initiation 147

13.3 Study Conduct 151

13.4 Study Closure 158

13.5 Summary 161

References 162

14 Clinical Data Management 164

14.1 Introduction 164

14.2 How Has Clinical Data Management Evolved? 165

14.3 Electronic Data Capture 166

14.4 Regulatory Involvement with Clinical Data Management 167

14.5 Professional Societies 167

14.6 Look to the Future 168

14.7 Conclusion 169

References 169

15 Clinical Significance 170

15.1 Introduction 170

15.2 Historical Background 170

15.3 Article Outline 171

15.4 Design and Methodology 171

15.5 Examples 181

15.6 Recent Developments 181

15.7 Concluding Remarks 185

References 185

16 Clinical Trial Misconduct 191

16.1 The Scope of this Article 191

16.2 Why Does Research Misconduct Matter? 191

16.3 Early Cases 192

16.4 Definition 193

16.5 Intent 194

16.6 What Scientific Misconduct was Not 194

16.7 The Process 194

16.8 The Past Decade 195

16.9 Lessons from the U.S. Experience 196

16.10 Outside the United States 197

16.11 Scientific Misconduct During Clinical Trials 198

16.12 Audit 198

16.13 Causes 199

16.14 Prevalence 200

16.15 Peer Review and Misconduct 200

16.16 Retractions 201

16.17 Prevention 201

References 202

17 Clinical Trials, Early Cancer and Heart Disease 205

17.1 Introduction 205

17.2 Developments in Clinical Trials at the National Cancer Institute (NCI) 205

17.3 Developments in Clinical Trials at the National Heart, Lung, and Blood Institute (NHLBI) 209

References 213

18 Cluster Randomization 216

18.1 Introduction 216

18.2 Examples of Cluster Randomization Trials 217

18.3 Principles of Experimental Design 218

18.4 Experimental and Quasi-Experimental Designs 219

18.5 The Effect of Failing to Replicate 220

18.6 Sample Size Estimation 221

18.7 Cluster Level Analyses 222

18.8 Individual Level Analyses 223

18.9 Incorporating Repeated Assessments 225

18.10 Study Reporting 226

18.11 Meta-Analysis 227

References 228

19 Coherence in Phase I Clinical Trials 230

19.1 Introduction 230

19.2 Coherence: Definitions and Organization 230

19.3 Coherent Designs 232

19.4 Compatible Initial Design 233

19.5 Group Coherence 234

19.6 Real-Time Coherence 235

19.7 Discussion 238

References 238

20 Compliance and Survival Analysis 240

20.1 Compliance: Cause and Effect 240

20.2 All-or-Nothing Compliance 241

20.3 More General Exposure Patterns 242

20.4 Other Structural Modeling Options 242

References 244

21 Composite Endpoints in Clinical Trials 246

21.1 Introduction 246

21.2 The Rationale for Composite Endpoints 246

21.3 Formulation of Composite Endpoints 247

21.4 Examples 248

21.5 Interpreting Composite Endpoints 250

21.6 Conclusions 251

References 251

22 Confounding 252

22.1 Introduction 252

22.2 Confounding as a Bias in Effect Estimation 252

22.3 Confounding and Noncollapsibility 258

22.4 Confounding in Experimental Design 260

References 261

23 Control Groups 263

23.1 Introduction 263

23.2 History 263

23.3 Ethics 264

23.4 Types of Control Groups: Historical Controls 266

23.5 Types of Control Groups: Randomized Controls 268

23.6 Conclusion 271

References 271

24 Coronary Drug Project 273

24.1 Introduction 273

24.2 Objectives 273

24.3 Study Design and Methods 273

24.4 Results 275

24.5 Conclusions and Lessons Learned 281

References 282

Further Reading 284

25 Covariates 285

25.1 Universal Character of Covariates 285

25.2 Use of Covariates in Clinical Trials 286

25.3 Continuous Covariates: Categorization or Functional Form? 293

25.4 Reporting and Summary Assessment of Prognostic Markers 295

References 296

26 Crossover Design 300

26.1 Introduction 300

26.2 The Two-Period, Two-Treatment Design 301

26.3 Higher Order Designs 304

26.4 Model-Based Analyses 307

References 308

27 Crossover Trials 310

27.1 Introduction 310

27.2 2 x 2 Crossover Trial 312

27.3 Higher-Order Designs for Two Treatments 312

27.4 Designs for Three or More Treatments 312

27.5 Analysis of Continuous Data 314

27.6 Analysis of Discrete Data 315

27.7 Concluding Remarks 317

References 317

28 Diagnostic Studies 320

28.1 Introduction 320

28.2 Diagnostic Studies 320

28.3 Reliability 324

28.4 Validity 331

References 338

Further Reading 339

29 DNA Bank 340

29.1 Definition and Objectives of DNA Biobanks 340

29.2 Types of DNA Biobanks 343

29.3 Types of Samples Stored 344

29.4 Quality Assurance and Quality Control in DNA Biobanks 345

29.5 Ethical Issues 346

29.6 Current Biobank Initiatives 348

29.7 Conclusions 350

References 350

30 Up-and-Down and Escalation Designs 353

30.1 Introduction 353

30.2 Up-and-Down Designs 353

30.3 Escalation Designs 357

30.4 Comparing U&D, Escalation and Model-Based Designs 359

References 359

Further Reading 361

31 Dose Ranging Crossover Designs 362

31.1 Introduction 362

31.2 Titration Designs and Extension Studies 369

31.3 Randomized Designs 373

31.4 Discussion and Conclusion 376

References 379

Further Reading 382

32 Flexible Designs 383

32.1 Introduction 383

32.2 The General Framework 384

32.3 Conditional Power and Sample Size Reassessment 387

32.4 Extending the Flexibility to the Choice of the Number of Stages 392

32.5 Selection of the Test Statistic 393

32.6 More General Adaptations and Multiple Hypotheses Testing 393

32.7 An Example 395

32.8 Conclusion 395

References 396

33 Gene Therapy 399

33.1 Introduction 399

33.2 Requirements for Successful Therapeutic Intervention 399

33.3 Pre-Clinical Research 402

33.4 Translational Challenges of Gene Therapy Trials 404

33.5 Clinical Trials · 407

33.6 Lessons Learned 408

33.7 The Way Forward 411

References 411

Further Reading 422

34 Global Assessment Variables 423

34.1 Introduction 423

34.2 Scientific Questions for Multiple Outcomes 423

34.3 General Comments on the GST 424

34.4 Recoding Outcome Measures 424

34.5 Types of Global Statistical Tests (GSTs) 425

34.6 Other Considerations 428

34.7 Other Methods 430

34.8 Examples of the Application of GST 434

34.9 Conclusions 435

References 435

35 Good Clinical Practice (GCP) 438

35.1 Introduction 438

35.2 Human Rights and Protections 438

35.3 Informed Consent 439

35.4 Investigational Protocol 439

35.5 Investigator's Brochure 440

35.6 Investigational New Drug Application 440

35.7 Production of the Investigational Drug 440

35.8 Clinical Testing 441

35.9 Sponsors 442

35.10 Contract Research Organization 444

35.11 Monitors 444

35.12 Investigators 444

35.13 Documentation 444

35.14 Clinical Holds 445

35.15 Inspections/Audits 446

References 446

Further Reading 446

36 Group-Randomized Trials 448

36.1 Introduction 448

36.2 Group-Randomized Trials in Context 449

36.3 The Development of Group-Randomized Trials in Public Health 450

36.4 The Range of GRTs in Public Health 451

36.5 Current Design and Analytic Practices in GRTs in Public Health 452

36.6 The Future of Group-Randomized Trials 453

36.7 Planning a New Group-Randomized Trial 456

References 462

37 Group Sequential Designs 467

37.1 Introduction 467

37.2 Classical Designs 469

37.3 The á-Spending Function Approach 474

37.4 Point Estimates and Confidence Intervals 477

37.5 Supplements 478

References 479

38 Hazard Ratio 483

38.1 Introduction 483

38.2 Definitions 483

38.3 Illustration of Hazard Rate, Hazard Ratio and Risk Ratio 484

38.4 Example on the Use and Usefulness of Hazard Ratios 486

38.5 Ad-hoc Estimator of the Hazard Ratio 486

38.6 Confidence Interval of the Ad-hoc Estimator 487

38.7 Ad-hoc Estimator Stratified for the Covariate Renal Function 491

38.8 Properties of the Ad-hoc Estimator 493

38.9 Class of Generalized Rank Estimators of the Hazard Ratio 493

38.10 Estimation of the Hazard Ratio with Cox's Proportional Hazards Model 494

38.11 Discussion 497

Further Reading 499

References 499

39 Large Simple Trials 500

39.1 Large, Simple Trials 500

39.2 Small but Clinically Important Objective 500

39.3 Eligibility 502

39.4 Randomized Assignment 502

39.5 Outcome Measures 504

39.6 Conclusions 506

References 506

Further Reading 508

40 Longitudinal Data 510

40.1 Definition 510

40.2 Longitudinal Data from Clinical Trials 510

40.3 Advantages 512

40.4 Challenges 512

40.5 Analysis of Longitudinal Data 513

References 514

Further Reading 514

41 Maximum Duration and Information Trials 515

41.1 Introduction 515

41.2 Two Paradigms: Duration versus Information 516

41.3 Sequential Studies: Maximum Duration versus Information Trials 516

41.4 An Example of a Maximum Information Trial 519

References 521

42 Missing Data 522

42.1 Introduction 522

42.2 Methods in Common Use 524

42.3 An Alternative Approach to Incomplete Data 525

42.4 Illustration: Orthodontic Growth Data 527

42.5 Inverse Probability Weighting 531

42.6 Multiple Imputation 531

42.7 Sensitivity Analysis 532

42.8 Conclusion 533

References 533

43 Mother to Child Human Immunodeficiency Virus Transmission Trials 536

43.1 Introduction 536

43.2 The Pediatric Aids Clinical Trials Group 076 Trial 538

43.3 Results 538

43.4 The European Mode of Delivery Trial 540

43.5 The HIV Network for Prevention Trials 012 Trial 541

43.6 The Mashi Trial 544

References 545

Further Reading 549

44 Multiple Testing in Clinical Trials 550

44.1 Introduction 550

44.2 Concepts of Error Rates 551

44.3 Union-Intersection Testing 552

44.4 Closed Testing 553

44.5 Partition Testing 555

References 556

Further Reading 557

45 Multicenter Trials 558

45.1 Definitions 558

45.2 History 560

45.3 Examples 561

45.4 Organizational and Operational Features 563

45.5 Strengths 564

45.6 Counts 565

Readings 569

References 569

46 Multiple Endpoints 570

46.1 Introduction 570

46.2 Multiple Testing Methods 571

46.3 Multivariate Global Tests 573

46.4 Conclusions 574

References 575

47 Multiple Risk Factor Intervention Trial 577

47.1 Introduction 577

47.2 Trial Design 577

47.3 Trial Screening and Execution 579

47.4 Findings at the End of Intervention 580

47.5 Long-Term Follow-Up 581

47.6 Epidemiologie Findings from Long-Term Follow-up of 361,662 MRFIT Screenees 582

47.7 Conclusions 583

References 583

Further Reading 586

48 N-of-1 Randomized Trials 587

48.1 Introduction 587

48.2 Goal of N-of-1 Studies 587

48.3 Requirements 588

48.4 Design Choices and Details for N-of-1 Studies 589

48.5 Statistical Issues 592

48.6 Other Issues 593

48.7 Conclusions 596

References 596

49 Noninferiority Trial 598

49.1 Introduction 598

49.2 Essential Elements of Noninferiority Trial Design 598

49.3 Objectives of Noninferiority Trials 600

49.4 Measure of Treatment Effect 600

49.5 Noninferiority Margin 601

49.6 Statistical Testing for Noninferiority 603

49.7 Medication Nonadherence and Misclassificat ion/Measurement Error 604

49.8 Testing Superiority and Noninferiority 605

49.9 Conclusion 605

References 605

50 Nonrandomized Trials 609

50.1 Introduction 609

50.2 Randomized vs. Nonrandomized Clinical Trials 609

50.3 Control Groups in Nonrandomized Trials 611

50.4 Statistical Methods in Design and Analyses 613

50.5 Conclusion and Discussion 616

References 617

51 Open-Labeled Trials 619

51.1 Introduction 619

51.2 The Importance of Blinding 619

51.3 Reasons Why Trials Might Have to be Open-Label 622

51.4 When Open-Label Trials Might be Desirable 623

51.5 Concluding Comments 623

References 623

Further Reading 624

52 Optimizing Schedule of Administration in Phase I Clinical Trials 625

52.1 Introduction 625

52.2 Motivating Example 627

52.3 Design Issues 627

52.4 Trial Conduct 631

52.5 Extensions and Related Research 632

References 632

53 Partially Balanced Designs 635

53.1 Introduction 635

53.2 Association Schemes 635

53.3 Partially Balanced Incomplete Block Designs 641

53.4 Generalizations of PBIBDs and Related Ideas 648

References 655

54 Phase I/II Clinical Trials 658

54.1 Introduction 658

54.2 Traditional Approach 659

54.3 Recent Developments 660

54.4 Illustrations 663

References 665

55 Phase II/III Trials 667

55.1 Introduction 667

55.2 Description and Legal Basis 668

55.3 Better Dose-Response Studies with Phase 2/3 Designs 672

55.4 Principles of Phase 2/3 Designs 673

55.5 Inferential Difficulties 676

55.6 Summary 678

References 679

Further Reading 680

56 Phase I Trials 682

56.1 Introduction 682

56.2 Phase I in Healthy Volunteers 683

56.3 Phase I in Cancer Patients 684

56.4 Perspectives in the Future of Cancer Phase I Trials 687

56.5 Discussion 688

References 688

57 Phase II Trials 692

57.1 Introduction 692

57.2 Proof-of-Concept (Phase Ha) Trials 693

57.3 Dose-Ranging (Phase lib) Trials 695

57.4 Efficacy Endpoints 697

57.5 Oncology Phase II Trials 697

References 697

Further Reading 699

58 Phase III Trials 700

58.1 Introduction 700

58.2 Research Methodology in Phase III 700

58.3 Type of Design 706

58.4 Discussion 708

References 709

59 Phase IV Trials 711

59.1 Introduction 711

59.2 Definitions and Context 711

59.3 Different Purposes for Phase IV Trials 712

59.4 Essential and Desirable Features of Phase IV Trials 715

59.5 Examples of Phase IV Studies 715

59.6 Conclusion 717

References 717

Further Reading 718

60 Phase I Trials in Oncology 719

60.1 Introduction 719

60.2 Dose-Limiting Toxicity 719

60.3 Starting Dose 720

60.4 Dose Level Selection 720

60.5 Study Design and General Considerations 720

60.6 Traditional, Standard, or 3 + 3 Design 721

60.7 Continual Reassessment Method and Other Designs that Target the MTD722

60.8 Start-Up Rule 722

60.9 Phase I Trials with Long Follow-Up 722

60.10 Phase I Trials with Multiple Agents 723

60.11 Phase I Trials with the MTD Defined using Toxicity Grades 723

References 723

Further Reading 724

61 Placebos 725

61.1 History of Placebo 725

61.2 Definitions 725

61.3 Magnitude of the Placebo Effect 726

61.4 Influences on the Placebo Effect 727

61.5 Ethics of Employing Placebo in Research 728

61.6 Guidelines for the Use of Placebos in Research 729

61.7 Innovations to Improve Research Involving Placebo 731

61.8 Summary 732

References 732

62 Planning a Group-Randomized Trial 736

62.1 Introduction 736

62.2 The Research Question 736

62.3 The Research Team 737

62.4 The Research Design 737

62.5 Potential Design Problems and Methods to Avoid Them 738

62.6 Potential Analytic Problems and Methods to Avoid Them 739

62.7 Variables of Interest and Their Measures 739

62.8 The Intervention 740

62.9 Power 742

62.10 Summary 742

References 743

63 Postmenopausal Estrogen/Progestin Interventions Trial (PEPI) 744

63.1 Introduction 744

63.2 Design and Objectives 744

63.3 Study Design 746

63.4 Outcomes 747

63.5 Results 749

63.6 Conclusions 753

References 754

Further Reading 756

64 Preference Trials 759

64.1 Introduction 759

64.2 Potential Effects of Preference 759

64.3 The Patient Preference Design 761

64.4 Advantages and Disadvantages of the Patient Preference Design 761

64.5 Alternative Designs 764

64.6 Discussion 767

References 768

Further Reading 769

65 Prevention Trials 770

65.1 Introduction 770

65.2 Role Among Possible Research Strategies 771

65.3 Prevention Trial Planning and Design 773

65.4 Conduct, Monitoring, and Analysis 775

References 776

66 Primary Efficacy Endpoint 779

66.1 Defining the Primary Endpoint 779

66.2 Fairness of Endpoints 780

66.3 Specificity of the Primary Endpoint 782

66.4 Composite Primary Endpoints 782

66.5 Missing Primary Endpoint Data 784

66.6 Censored Primary Endpoints 784

66.7 Surrogate Primary Endpoints 785

66.8 Multiple Primary Endpoints 786

66.9 Secondary Endpoints 786

References 786

Further Reading 788

67 Prognostic Variables in Clinical Trials 789

67.1 Introduction 789

67.2 A General Theory of Prognostic Variables 791

67.3 Valid Covariates and Recognizable Subsets 792

67.4 Stratified Randomization and Analysis 793

67.5 Statistical Importance of Prognostic Factors 795

References 797

68 Randomization Procedures 799

68.1 Basics 799

68.2 General Classes of Randomization: Complete Versus Imbalance-Restricted Procedures 800

68.3 Procedures for Imbalance-Restricted Randomization 801

68.4 Randomization-Based Analysis and the Validation Transformation 809

68.5 Conclusions 810

References 810

69 Randomization Schedule 813

69.1 Introduction 813

69.2 Preparing the Schedule 814

69.3 Schedules for Open-Label Trials 817

69.4 Schedules to Mitigate Loss of Balance in Treatment Assignments Because of Incomplete Blocks 818

69.5 Issues Related to the use of Randomization Schedule 822

69.6 Summary 824

References 825

Further Reading 826

70 Repeated Measurements 827

70.1 Introduction and Case Study 827

70.2 Linear Models for Gaussian Data 828

70.3 Models for Discrete Outcomes 831

70.4 Design Considerations 836

70.5 Concluding Remarks 837

References 838

71 Simple Randomization 841

71.1 Introduction 841

71.2 Concept of Randomization 841

71.3 Why is Randomization Needed? 842

71.4 Methods: Simple Randomization 842

71.5 Advantages and Disadvantages of Randomization 845

71.6 Other Randomization Methods 846

71.7 Stratified Randomization 846

References 849

Further Reading 849

72 Subgroups 850

72.1 Introduction 850

72.2 The General Problem 851

72.3 Definitions 851

72.4 Subgroup Effects and Interactions 852

72.5 Tests of Interactions and the Problem of Power 853

72.6 Subgroups and the Problem of Multiple Comparisons 856

72.7 Demographic Subgroups 858

72.8 Physiological Subgroups 861

72.9 Target Subgroups 861

72.10 Improper Subgroups 863

72.11 Summary 865

References 865

73 Superiority Trials 867

73.1 Introduction 867

73.2 Clinicians Ask One-Sided Questions, and Want Immediate Answers 867

73.3 But Traditional Statistics Is Two-Sided 867

73.4 The Consequences of Two-Sided Answers to One-Sided Questions 868

73.5 The Fallacy of the "Negative" Trial 868

73.6 The Solution Lies in Employing One-Sided Statistics 868

73.7 Examples of Employing One-Sided Statistics 868

73.8 One-Sided Statistical Analyses Need to be Specified Ahead of Time 869

73.9 A Graphic Demonstration of Superiority and Noninferiority 869

73.10 How to Think about and Incorporate Minimally Important Differences 870

73.11 Incorporating Confidence Intervals for Treatment Effects 871

73.12 Why We Should Never Label an "Indeterminate" Trial Result as "Negative" or as Showing "No Effect" 871

73.13 How Does a Treatment Become "Established Effective Therapy"? 872

73.14 Most Trials are Too Small to Declare a Treatment "Established Effective Therapy" 872

73.15 How Do We Achieve a Superiority Result? 872

73.16 Superiority and Noninferiority Trials when Established Effective Therapy Already Exists 872

73.17 Exceptions to the Rule that It Is Always Unethical to Substitute Placebos for Established Effective Therapy 873

73.18 When a Promising New Treatment Might be Added to Established Effective Therapy 873

73.19 Using Placebos in a Trial Should Not Mean the Absence of Treatment 874

73.20 Demonstrating Trials of Promising New Treatments Against (or in Addition to) Established Effective Therapy 874

73.21 Why We Almost Never Find, and Rarely Seek, True "Equivalence" 874

73.22 The Graphical Demonstration of "Superiority" and "Noninferiority" 876

73.23 Completing the Circle: Converting One-Sided Clinical Thinking into One-Sided Statistical Analysis 876

73.24 A Final Note on Superiority and Noninferiority Trials of "Me-Too" Drugs 877

References 877

Further Reading 877

74 Surrogate Endpoints 878

74.1 Introduction 878

74.2 Illustrations 879

74.3 Validation of Surrogates 880

74.4 Auxiliary Variables 883

74.5 Conclusions 884

References 885

75 TNT Trial 887

75.1 Introduction 887

75.2 Objectives 887

75.3 Study Design 887

75.4 Results 888

75.5 Conclusions 892

References 892

Further Reading 893

76 UGDP Trial 894

76.1 Introduction 894

76.2 Design and Chronology 895

76.3 Results 906

76.4 Conclusion and Discussion 909

References 914

77 Women's Health Initiative Hormone Therapy Trials 918

77.1 Introduction 918

77.2 Objectives 918

77.3 Study Design 918

77.4 Results 919

77.5 Conclusions 927

References 928

78 Women's Health Initiative Dietary Modification Trial 931

78.1 Rationale for Biomarker Calibration of Self-Report Measures of Diet 931

78.2 Nutrient Biomarker Study Energy and Protein Calibration 932

78.3 Measurement Error Properties of 4DFR, 24HR, and FFQ 933

78.4 Calibration of Self-Report Measures of Physical Activity 933

78.5 Psychosocial Measures and Biomarker-Calibrated Intake 936

78.6 Calibrated Energy, Protein, Protein Density, and Cardiovascular Disease Incidence 937

78.7 Diabetes and Calibrated Consumption 938

78.8 Cancer and Calibrated Intake 940

78.9 Associations Between Protein Intake, Frailty, and Renal Function 940

78.10 Summary and Future Directions 941

References 943

Index 945