Оглавление
David Machin. Randomised Clinical Trials
Table of Contents
List of Tables
List of Illustrations
Guide
Pages
Randomised Clinical Trials. Design, Practice and Reporting
Preface
CHAPTER 1 Introduction
1.1 Introduction
1.2 Some completed trials
Example 1.1 Small parallel two‐group design – gastrointestinal function
Example 1.2 Parallel two‐group design – hepatitis B
Example 1.3 Unstructured three‐group design – newly diagnosed type 2 diabetes
Example 1.4 Small dose–response design – pain prevention following hand surgery
Example 1.5 Large dose–response design – HER2‐positive breast cancer
Example 1.6 Non‐inferiority trial – uncomplicated falciparum malaria
Example 1.7 Repeated measures – atopic eczema
Example 1.8 Cross‐over trial – known or suspected hypertension
Example 1.9 Paired design – glaucoma
Example 1.10 Split‐mouth design – implants for edentulous sites
Example 1.11 Cluster trial – hip protectors for the elderly
Example 1.12 Single arm – discrete de novo lesions in a coronary artery
1.3 Choice of design. 1.3.1 Biological variability
Example 1.13 Patient‐to‐patient variability – atopic eczema
1.3.2 Randomisation
1.3.3 Design hierarchy
1.3.3.1 Randomisation
1.3.3.2 Blinding or masking
1.3.3.3 Non‐randomised designs
Example 1.14 Glioblastoma in the elderly – non‐randomised design
1.3.3.4 Case series
1.4 Practical constraints
1.5 Influencing clinical practice
1.6 History
1.7 How do trials arise?
1.8 Ethical considerations
1.9 Regulatory requirements
1.10 Focus
1.11 Further reading
CHAPTER 2 Design Features
2.1 Introduction
Example 2.1 Two‐group parallel design – symptomatic oral lichen planus
2.2 The research question
2.3 Patient selection
Example 2.2 Trial eligibility – partial thickness burns
2.4 The consent process
2.5 Choice of interventions
2.5.1 Standard or control treatment
Example 2.3 Placebo controlled trial – advanced hepatocellular carcinoma
Example 2.4 Schizophrenia and schizophreniform disorder
2.5.2 Test treatment
2.6 Choice of design
2.7 Assigning the interventions
Example 2.5 Delay to start of treatment – inoperable non‐small‐cell lung cancer
2.8 Making the assessments
2.9 Analysis and reporting
2.9.1 Which patients to analyse
Example 2.6 Immediate start of treatment– children with fever
2.9.1.1 Intention‐to‐treat (ITT)
2.9.1.2 Per‐protocol
2.9.2 Trial publication
2.10 Technical details. 2.10.1 Statistical models
2.10.2 Fitting the model
2.10.3 Choice of design
2.10.4 Randomisation
2.11 Guidelines
2.12 Further reading
CHAPTER 3 The Trial Protocol
3.1 Introduction
3.2 Abstract
Example 3.1 Protocol SQGL02 (1999): Brimonidine as a neuroprotective agent in acute angle‐closure glaucoma (AACG)
Example 3.2 Protocol SQCP01 (2006): Comparing speech and growth outcomes between two different techniques and two different timings of surgery in the management of clefts of the secondary palate
3.3 Background
3.4 Research objectives
3.4.1 Objectives
Example 3.3 Protocol SQNP01 (1997): Standard radiotherapy versus concurrent chemo‐radiotherapy followed by adjuvant chemotherapy for locally advanced (non‐metastatic) nasopharyngeal cancer
Example 3.4 Protocol PRESSURE (2000) Part Section 2: Pressure‐relieving support surfaces: a randomised evaluation. 2. RESEARCH OBJECTIVES. 2.1 Primary objective
2.2 Secondary objective
3.4.2 Outcome measures
Example 3.5 Protocol PRESSURE (2000) Part Section 9.1: Pressure‐relieving support surfaces: a randomised evaluation. 9.1 Primary endpoint. 9.1.1 Development of new pressure sores
9.2 Secondary endpoints. 9.2.1 Healing of existing pressure sores
9.2.2 Patient acceptability
Example 3.6 Protocol AHCC01 (1997): Randomised trial of tamoxifen versus placebo for the treatment of inoperable hepatocellular carcinoma
3.5 Design
Example 3.7 Protocol PRESSURE (2000): Pressure‐relieving support surfaces: a randomised evaluation. 3. DESIGN
3.6 Intervention details
Example 3.8 Protocol PRESSURE (2000): Pressure‐relieving support surfaces: a randomised evaluation
Example 3.9 Protocol SQNP01 (1997): Standard radiotherapy versus concurrent chemo‐radiotherapy followed by adjuvant chemotherapy for locally advanced (non‐metastatic) nasopharyngeal cancer
Example 3.10 Protocol SQOLP01 (1999): Comparison of steroid with cyclosporine for the topical treatment of oral lichen planus. Dose and duration. Steroid (S) – Triamcinolone acetonide 0.1% in oral base (Kenalog)
Cyclosporine (C) – Sandimmun Neoral solution 100 mg cyclosporine/ml
Application of topical medication
Dose modification
Example 3.11 Protocol ENSG5 (1990): Comparison of high dose rapid schedule with conventional schedule chemotherapy for stage 4 neuroblastoma over the age of 1 year. 12. MODIFICATION OF THERAPY DUE TO TOXICITY
3.7 Eligibility
Example 3.12 Protocol UKW3 (1992): Preoperative chemotherapy in Wilms’ tumour. 4. ELIGIBILITY
Example 3.13 Protocol PRESSURE (2000) Section 6.2: Pressure‐relieving support surfaces: a randomised evaluation. 6.2 Exclusion criteria
3.8 Randomisation
3.8.1 Design
Example 3.14 Protocol PRESSURE (2000) Section 7.2: Pressure‐relieving support surfaces: a randomised evaluation. 7.2 Mattress Allocation Method
3.8.2 Implementation
Example 3.15 Protocol SQOLP01 (1999, Section 7): Comparison of steroid with cyclosporine for the topical treatment of oral lichen planus. 7. Randomisation
Example 3.16 Protocol EXPEL (2016): Peritoneal lavage after curative gastrectomy. Enrolment and randomisation
Example 3.17 Protocol SQGL02 (1999): Brimonidine as a neuroprotective agent in acute angle‐closure glaucoma (AACG) Procedure for randomisation
3.9 Assessment and data collection. 3.9.1 Assessments
Example 3.18 Protocol SQCP01 (2006): Comparing speech and growth outcomes between two different techniques and two different timings of surgery in the management of clefts of the secondary palate. Method
Example 3.19 Protocol PRESSURE (2000): Pressure‐relieving support surfaces: a randomised evaluation
3.9.2 Data collection
3.10 Statistical considerations
3.10.1 Number of subjects
Example 3.20 Protocol SQGL02: Brimonidine as a neuroprotective agent in acute angle‐closure glaucoma (AACG) 9. STATISTICAL CONSIDERATIONS. Trial size
3.10.2 Analysis
Example 3.21 Protocol SQOLP01 (1999): Comparison of steroid with cyclosporine for the topical treatment of oral lichen planus. Analysis plan. Primary endpoints
Clinical response at 4 weeks
Pain score 4 weeks
Secondary analyses
Adverse events
Additional analysis
3.10.3 Interim analysis
Example 3.22 Protocol EXPEL (2016): Peritoneal lavage after curative gastrectomy
3.11 Ethical issues
3.11.1 General
Example 3.23 Protocol PRESSURE (2000): Pressure‐relieving support surfaces: a randomised evaluation
Example 3.24 Protocol SQGL02 (1999): Brimonidine as a neuroprotective agent in acute angle‐closure glaucoma
3.11.2 Informed consent
Example 3.25 Protocol SQGL02 (1999): Brimonidine as a neuroprotective agent in acute angle‐closure glaucoma
Example 3.26 eConsent – acute stroke
3.12 Organisational structure
3.13 Publication policy
Example 3.27 Protocol SQOLP01: Comparison of steroid with cyclosporine for the topical treatment of oral lichen planus. PUBLICATION POLICY
3.14 Trial forms
3.15 Appendices
3.16 Regulatory requirements
3.16.1 Protocol amendments
3.17 Guidelines
3.18 Protocols
CHAPTER 4 Measurement and Data Capture
4.1 Introduction
4.2 Types of measures. 4.2.1 Qualitative data. 4.2.1.1 Nominal or unordered categorical
4.2.1.2 Ordered categorical or ranked
4.2.2 Numerical or quantitative data
4.2.3 Time‐to‐event
4.2.4 Multiple sites per subject
4.2.5 Practice
4.3 Measures and endpoints
4.3.1 Assessments
4.3.2 Endpoints
4.3.2.1 Objective criteria
Example 4.1 Subjective skin assessment to identify bed sores
Example 4.2 VAS – pain assessment
4.3.2.2 Single measure – one data item
4.3.2.3 Single measure – several data items
Example 4.3 Time to develop a first (or new) pressure sore
4.3.2.4 Multiple measures
4.3.2.5 Repeated measures
Example 4.4 Multiple sclerosis
4.3.2.6 Multiple endpoints
4.3.2.7 Laboratory measures
Example 4.5 Laboratory assay – estrone levels
4.3.2.8 Surrogates
4.3.3 Patient self‐reported outcomes
4.3.4 Economic evaluation
4.4 Making the observations
4.4.1 Which observer
4.4.2 Precision
4.5 Baseline measures
4.6 Data recording. 4.6.1 Forms
4.6.1.1 Layout
Example 4.6 Form design – breast cancer
4.6.1.2 Closed questions
4.6.1.3 Open questions
4.6.1.4 Follow‐up or repeat forms
4.6.2 Questionnaires
4.6.2.1 Layout
Example 4.7 Question layout – sexual function after gynaecological cancer
4.6.2.2 Closed questions
Example 4.8 Closed question – sexual function after gynaecological cancer
Example 4.9 Likert scale – SF‐36
4.6.2.3 Open questions
4.6.2.4 Response bias
4.7 Technical notes
4.8 Guidelines
CHAPTER 5 Randomisation
5.1 Introduction
5.2 Rationale
5.3 Mechanics. 5.3.1 Simple randomisation. 5.3.1.1 Random numbers
5.3.1.2 Simple random allocation
Example 5.1 Simple randomisation – chronic gastro‐oesophageal reflux
5.3.2 Blocks
5.3.2.1 Randomised block designs (RBD)
5.3.2.2 Variable block size
5.3.2.3 Allocation ratio
Example 5.2 Unequal allocation – hepatocellular carcinoma
5.3.2.4 Stratified randomisation
Example 5.3 Stratified randomisation by recruiting centre – uncomplicated falciparum malaria
5.3.3 Dynamic allocation by minimisation
Example 5.4 Stratified randomisation – cranberry or apple juice for urinary symptoms
Example 5.5 Trastuzumab for HER2‐positive breast cancer
5.4 Application
5.5 Carrying out randomisation. 5.5.1 Design
5.5.2 Preparing the list
5.5.3 Double‐blind trials
Example 5.6 Prevention of pain and bruising following hand surgery
5.5.4 Breaking the code
5.5.5 Computer‐based systems
5.6 Documentation
Example 5.7 Root caries repair
5.7 Unacceptable methods
5.8 Guidelines
CHAPTER 6 Trial Initiation
6.1 Introduction
6.2 Trial organisation. 6.2.1 Trial steering committee
6.2.2 Identifying the tasks
6.2.3 Trial office
6.2.4 Protocol and form production
6.2.5 Obtaining approval
6.2.6 Trial registration
6.2.7 Establishing the network
6.2.8 Mechanisms – supplies. 6.2.8.1 The protocol
6.2.8.2 Supplies
6.2.8.3 Training
6.2.9 Registering and randomisation of patients
6.3 Data collection and processing. 6.3.1 Data forms and their scheduling
6.3.2 Missing forms and data
6.3.3 Database
6.3.4 Checking
6.4 Internal data monitoring
6.5 Ethical and regulatory requirements. 6.5.1 Routine reporting
6.5.2 Serious adverse events
6.6 Launching the trial
6.7 Trial registries
6.8 Guidelines
CHAPTER 7 Trial Conduct and Completion
7.1 Introduction
7.2 Regular feedback. 7.2.1 Recruitment
7.2.2 Missing forms
7.2.3 Validation
7.2.4 Basics
7.3 Publicity. 7.3.1 Newsletters
7.3.2 Clinical journals
7.4 Protocol modifications
7.5 Preparing the publication(s)
7.5.1 Preliminaries
7.5.2 Writing committee
7.5.3 Practicalities
7.5.4 New developments
7.5.5 Choice of journal(s)
7.6 The next trial?
7.7 Protocol
CHAPTER 8 Basics for Analysis
8.1 Introduction
8.2 The standard Normal distribution
8.3 Confidence intervals
Example 8.1 Difference in means – atopic eczema
8.4 Statistical tests. 8.4.1 Significance tests
8.4.2 Null hypothesis
8.5 Examples of analysis. 8.5.1 Means
Example 8.2 Confidence interval – eczema
Example 8.3 Student t‐test – eczema
8.5.2 Medians
Example 8.4 Difference in medians – eczema
8.5.3 Proportions
Example 8.5 Difference in proportions – oral lichen planus
Example 8.6 Difference in rates – recurrent malaria
Example 8.7p‐value – recurrent malaria
8.5.4 Number needed to treat (NNT)
Example 8.8 NNT – cleft lip and palate, and oral lichen planus
8.5.5 Ordered categorical
Example 8.9 Ordered categorical data – eczema
8.5.6 Time‐to‐event
Example 8.10 Comparing groups – advanced neuroblastoma
Example 8.11 Logrank test – advanced neuroblastoma
8.6 Regression methods
8.6.1 Comparing means
Example 8.12 Regression model – eczema
8.6.2 Proportions and the odds ratio
Example 8.13 Odds ratio – recurrent malaria
Example 8.14 Logistic regression – oral lichen planus
8.6.3 Ordered categorical
8.6.4 Time‐to‐event
Example 8.15 Cox model – advanced neuroblastoma
8.6.5 Adjusting for baseline
Example 8.16 Cox model with covariate – advanced neuroblastoma
8.6.6 Dummy variables
8.6.7 Time‐varying covariates
8.6.8 Adjusting for strata
Example 8.17 Stratified by stage – advanced neuroblastoma
8.7 Other issues. 8.7.1 Missing values
Example 8.18 Missing data – eczema
8.7.2 Graphical methods
8.7.3 Multiple endpoints
8.7.4 Interim analyses
8.8 Practice
8.9 Technical details. 8.9.1 Recommended method for comparing proportions
8.9.2 Confidence interval for an odds ratio
CHAPTER 9 Trial Size
9.1 Introduction
9.1.1 Caution
9.2 Significance level and power. 9.2.1 Significance level
9.2.2 The alternative hypothesis
9.2.3 Power
9.3 The fundamental equation
9.4 Specific situations. 9.4.1 Comparing means
Example 9.1 Difference in means – disease activity
9.4.2 Comparing proportions
Example 9.2 Complete response rate – multiple myeloma
Example 9.3 High‐risk prostate cancer
9.4.3 Ordered categorical data
Example 9.4 Moderate‐to‐severe eczema
9.4.4 Time‐to‐event
Example 9.5 Gastric cancer
9.5 Practical considerations. 9.5.1 Trial objectives
9.5.2 The anticipated effect size
Example 9.6 Anticipated effect size
9.5.3 Significance level and power
9.5.4 Allocation ratio
9.5.5 Limited resources
Example 9.7 Homeopathic arnica
9.5.6 Subject withdrawals
Example 9.8 Adjusting for withdrawals
9.6 Further topics
9.6.1 Several primary outcomes
Example 9.9 Two major endpoints – disease activity
9.6.2 Competing risks outcomes
Example 9.10 Competing risks – nasopharyngeal cancer
9.6.3 Revising trial size
Example 9.11 Internal pilot to modify trial size – gastric emptying time
9.6.4 1‐sided alternative hypotheses
9.7 Guideline
9.8 Software
CHAPTER 10 Data and Safety Monitoring
10.1 Introduction
10.2 The DSMB. 10.2.1 Remit and terms of reference
10.2.2 Logistics
10.2.3 Composition
10.2.4 Frequency of meetings
10.2.5 Stopping rules
10.3 Early reviews
Example 10.1 Protocol UKW3 (1992) – preoperative chemotherapy in Wilms' tumour
10.3.1 Safety
Example 10.2 Peritoneal lavage after curative gastrectomy
Example 10.3 Preterm infants
10.3.2 Trial size
Example 10.4 Review of trial size – nasopharyngeal cancer
10.3.3 Recruitment issues
Example 10.5 Cleft palate
10.4 Interim reviews. 10.4.1 Rationale
10.4.2 Efficacy
Example 10.6 Heroin dependence
10.4.2.1 Stopping for efficacy
Example 10.7 Children undergoing general anaesthesia
Example 10.8 Resectable hepatocellular carcinoma
10.4.3 Futility
Example 10.9 Children with severe infection
Example 10.10 Advanced cancer and cachexia
10.4.3.1 Stopping for futility
10.4.3.2 Conditional power test
Example 10.11 Acute ischaemic stroke
10.5 Protocols
CHAPTER 11 Reporting
11.1 Introduction. 11.1.1 Preliminaries
11.1.2 When to publish
11.2 Publication
11.2.1 Evidence‐based medicine
11.2.2 Plain language
11.3 Responsibilities. 11.3.1 Authorship
11.3.2 Registration
11.3.3 Funding sources
11.4 Background
11.5 Methods
11.5.1 Participants
11.5.2 Procedures
11.5.3 Assessments
11.5.4 Consent
11.5.5 Monitoring
11.5.6 Randomisation
11.5.7 Statistical considerations. 11.5.7.1 Justification of trial size
11.5.7.2 Interim analysis for data monitoring
11.5.7.3 Intended final analysis
11.6 Findings
11.6.1 CONSORT
11.6.2 Participant characteristics
11.6.3 Endpoints
11.6.4 Adverse events
11.6.5 Graphics
11.7 When things go wrong
11.8 Conclusions
11.9 Guidelines. 11.9.1 General
11.9.2 Editorial
11.9.3 CONSORT
11.9.4 Statistical standards
CHAPTER 12 More Than Two Interventions
12.1 Introduction
12.2 Unstructured comparisons. 12.2.1 Hypotheses
12.2.2 Analysis
Example 12.1 Postmenopausal women with early breast cancer
12.2.3 Selection design
Example 12.2 Selection design – advanced non–small cell lung cancer
12.2.4 Trial size
12.3 Comparisons with placebo (or standard) 12.3.1 Design
Example 12.3 Comparison with a standard – schizophrenia
12.3.2 Trial size
Example 12.4 Comparison with placebo – heroin dependence
12.3.3 Analysis
Example 12.5 Comparison with a placebo – intellectual disability
12.4 Dose–response designs. 12.4.1 Design
Example 12.6 Dose–response – rheumatoid arthritis
12.4.2 Trial size
Example 12.7 Rheumatoid arthritis
12.4.3 Analysis
Example 12.8 Dose–response – plaque psoriasis
12.4.4 Reporting
12.5 Factorial trials. 12.5.1 Design
Example 12.9 Low back pain
Example 12.10 23 factorial design – cleft lip and palate
12.5.2 Randomisation
Example 12.11 Cleft palate repair
12.5.3 Trial size
Example 12.12 Chronic obstructive pulmonary disease
12.5.4 Analysis. 12.5.4.1 Comparing means
Example 12.13 Chronic obstructive pulmonary disease
12.5.4.2 Modelling approach
12.5.5 Practical issues
Example 12.14 Malignant pleural mesothelioma
12.5.6 Reporting
Example 12.15 Unilateral cleft lip and palate
12.6 Complex structure comparisons. 12.6.1 Design
12.6.2 Trial size. Example 12.16 Newly diagnosed type 2 diabetes
12.6.3 Analysis
Example 12.17 Newly diagnosed type 2 diabetes
CHAPTER 13 Paired and Matched Designs
13.1 Matched‐pair trials. 13.1.1 Design
Example 13.1 Anaesthesia for deformed eyelid surgery
Example 13.2 Sputum samples from patients with cystic fibrosis
13.1.2 Analysis. 13.1.2.1 Difference in means
Example 13.3 Anaesthesia for deformed eyelid surgery
Example 13.4 Anaesthesia for deformed eyelid surgery
13.1.2.2 Difference in proportions
Example 13.5 Culture of P. aeruginosa in cystic fibrosis
13.1.3 Trial size. 13.1.3.1 Continuous outcome
Example 13.6 Anaesthesia for bilateral blepharoplasty of upper eyelids
13.1.3.2 Binary outcome
Example 13.7 Cystic fibrosis
13.2 Cross‐over trials. 13.2.1 Design
13.2.1.1 Two‐period – two‐treatment
13.2.1.2 Washout to reduce carry‐over
Example 13.8 Cross‐over trial – Anacetrapib and blood pressure
13.2.1.3 Run‐in
13.2.2 Difficulties
Example 13.9 Ambulatory blood pressure
13.2.3 Analysis
Example 13.10 Fuel metabolism during exercise
Example 13.11 Anaesthesia for bilateral blepharoplasty of upper eyelids
13.2.4 Trial Size
Example 13.12 Cross‐over trial size – ambulatory blood pressure
13.3 Split‐mouth designs. 13.3.1 Design
13.3.2 Difficulties
Example 13.13 Restoratives for dental caries
13.3.3 Analysis
Example 13.14 Split‐mouth design – caries prevention
13.3.4 Trial size. 13.3.4.1 Continuous outcome
Example 13.15 Split‐mouth design – restorative dental treatment
13.3.4.2 Binary outcome
Example 13.16 Split‐mouth design – Caries prevention
13.4 Guidelines
CHAPTER 14 Repeated Measures Design
14.1 Introduction
Example 14.1 Heart failure
Example 14.2 Moderate‐to‐severe eczema
Example 14.3 Adult patients with atopic eczema
14.2 Simplified analysis
14.2.1 Change from baseline. 14.2.1.1 One baseline – one post‐randomisation assessment
Example 14.4 Heart failure
14.2.1.2 Mean baseline – mean post‐randomisation assessments
14.2.1.3 Adjusting for baseline
Example 14.5 Heart failure
Example 14.6 Patients with phenylketonuria
14.3 Regression models
14.3.1 Fixed effects
14.3.2 Trends over time
14.3.3 Random effects
14.4 Auto‐correlation
14.4.1 Auto‐correlation coefficient
14.4.2 Patterns of auto‐correlation
14.4.2.1 Independent
14.4.2.2 Exchangeable
14.4.2.3 Autoregressive
14.4.2.4 Unstructured
Example 14.7 6MWT in patients with heart failure
14.5 Accounting for auto‐correlation
14.5.1 Linear profiles
Example 14.8 Heart failure – influence of time on 6MWT
Example 14.9 Heart failure – comparing Placebo with FCM
14.5.2 Non‐linear profiles
14.6 The design effect (DE)
14.6.1 Change from baseline
Example 14.10 Design effect – heart failure
14.6.2 Adjusting for baseline
Example 14.11 Design effect – phenylketonuria – atopic eczema
14.6.3 Post‐intervention trends
Example 14.12 Design effect – atopic eczema
14.6.4 Selected contrasts
Example 14.13 DE for a specific contrast
14.6.5 Comparing design effects
14.7 Trial size
14.7.1 Continuous outcomes
Example 14.14 Primary Sjögren's syndrome
14.7.2 Binary outcomes
Example 14.15 Sample size – primary Sjögren's syndrome
14.8 Practicalities. 14.8.1 Number of repeat assessments
14.8.2 Making the assessments
14.8.3 Choice of auto‐correlation structure
14.8.4 Strength of auto‐correlation
14.8.5 Withdrawals and missing values
14.8.6 Protocol description of intended analysis
14.9 Reporting
14.10 Matched organs receiving the same intervention. 14.10.1 Design
14.10.2 Single outcome per patient
14.10.3 Several (single‐time) outcomes per patient
CHAPTER 15 Non‐Inferiority and Equivalence Trials
15.1 Introduction
15.2 Non‐inferiority
15.2.1 Limit of non‐inferiority
15.2.2 Hypothesis
15.2.3 Establishing non‐inferiority
15.2.3.1 T less effective than S anticipated
15.2.3.2 T more adverse than S anticipated
Example 15.1 Paired design – methicillin‐resistant Staphylococcus aureus (MRSA)
15.3 Analysis
15.3.1 Continuous endpoint. 15.3.1.1 Independent groups
Example 15.2 Dosing of warfarin
15.3.1.2 Matched pairs
Example 15.3 Non‐inferiority cross‐over trial analgesia in preterm infants
15.3.2 Binary endpoint. 15.3.2.1 Independent groups
Example 15.4 Femoropopliteal periphery artery disease
15.3.2.2 Matched pairs
Example 15.5 Non‐inferiority cross‐over trial analgesia in preterm infants
15.4 Trial size. 15.4.1 Continuous endpoint. 15.4.1.1 Independent groups
Example 15.6 Clinical dosing of warfarin
15.4.1.2 Matched pairs
Example 15.7 Cross‐over trial – chronic non‐cancer pain
Example 15.8 Split‐mouth design – oral soft tissue augmentation
15.4.2 Binary endpoint. 15.4.2.1 Independent groups
Example 15.9 Rheumatoid arthritis
15.4.2.2 Matched pairs
Example 15.10 Analgesia in preterm infants
15.5 Equivalence
15.5.1 Hypotheses
15.5.2 Analysis
Example 15.11 Parallel‐group design – head louse infestation
15.5.3 Trial Size
Example 15.12 Cross‐over design – type 2 diabetes
15.6 Reporting
15.7 Practical Issues
15.8 Guidelines
CHAPTER 16 Cluster Designs
16.1 Design features
Example 16.1 Cluster design – enhanced diabetes care
16.2 Procedures. 16.2.1 Consent
Example 16.2 Facemask protection for health care workers
Example 16.3 Exercise for depression in elderly residents in care homes
16.2.2 Randomisation
Example 16.4 Number of clusters
16.3 Regression models
16.4 Intra‐class correlation
16.5 Trial size
16.5.1 Design effect (DE)
16.5.2 Continuous endpoint
Example 16.5 Trial size – systolic blood pressure levels
Example 16.6 Trial size – systolic blood pressure levels
16.5.3 Binary endpoint
Example 16.7 Comparing proportions – hypertension and hypercholesterolemia
Example 16.8 Facemask protection for health care workers
16.5.4 Varying cluster size
Example 16.9 Varying cluster size
16.6 Analysis
Example 16.10 Systolic blood pressure
16.7 Practicalities
16.8 Reporting
16.9 Further reading
CHAPTER 17 Stepped Wedge Designs
17.1 Introduction
17.2 Notation
Example 17.1 Trial design – ischaemic stroke
17.3 Basic structure
17.3.1 Full design
Example 17.2 Design matrix – ischaemic stroke
17.3.2 Incomplete and other designs
17.4 Randomisation
Example 17.3 Randomisation
17.5 Cross‐sectional design
17.5.1 Analysis
17.5.1.1 Before‐and‐after
Example 17.4 Simplified analysis – ischaemic stroke
17.5.1.2 Modelling approach
Example 17.5 Simple logistic regression – ischaemic stroke
Example 17.6 Accounting for cluster and time – ischaemic stroke
17.5.2 Design effect
17.5.3 Trial size
Example 17.7 Cross‐sectional – acute chest pain
17.6 Closed cohort design
Example 17.8 Closed cohort – free breakfast
Example 17.9 Closed cohort – nurse‐led medicines' monitoring
17.6.1 Analysis
17.6.2 Design effect
17.6.3 Trial size
Example 17.10 Closed cohort – provision of school breakfast
17.7 Practicalities
CHAPTER 18 Genomic Targets
18.1 Introduction
18.2 Predictive markers. 18.2.1 Definition
Example 18.1 Predictive classifier – epidermal growth factor receptor (EGFR)
Example 18.2 Potential biomarker – EGFR amplification and KRAS mutations
18.2.2 Implications for trial size
18.2.3 Role of strata
18.3 Enrichment design. 18.3.1 Design
18.3.2 Trial size
Example 18.3 Enrichment design – renal cell carcinoma
18.3.3 Analysis
18.4 Biomarker‐Stratified Designs. 18.4.1 Design
Example 18.4 Biomarker‐Stratified Design – acute myeloid leukaemia
18.4.2 Prognostic influence of the biomarker
18.4.3 Analysis
18.4.3.1 Anticipate T effective in B− and B+ patients
Example 18.5 Biomarker‐Stratified Design – breast cancer
18.4.3.2 Anticipate T effective in B− only if effective in B+
Example 18.6 Biomarker‐Stratified Design – renal cell carcinoma
18.4.4 Trial size
18.4.4.1 Anticipate T effective in B− and B+ patients
Example 18.7 Biomarker‐Stratified Design – survival time endpoint
18.4.4.2 Anticipate T effective in B− only if effective in B+
Example 18.8 Biomarker‐Stratified Design – survival time endpoint
18.4.5 Practical considerations
18.5 Adaptive threshold designs. 18.5.1 Design
Example 18.9 Adaptive threshold design – unconfirmed predictive classifier
18.5.2 Analysis
18.5.3 Trial size
Example 18.10 Adaptive threshold design – non‐small‐cell lung cancer
18.5.4 Practicalities
CHAPTER 19 Feasibility and Pilot Studies
19.1 Introduction
19.2 Feasibility studies
Example 19.1 Feasibility – patient informed consent
Example 19.2 Feasibility – refining the interventions
19.3 External‐pilot studies. 19.3.1 Aims
19.3.2 Single‐arm design
Example 19.3 Single‐group external‐pilot – varicose veins
19.3.3 Randomised design
Example 19.4 Determine the SD – major depressive disorder
19.3.4 Sample size. 19.3.4.1 General considerations
19.3.4.2 Flat rule‐of‐thumb
19.3.5 Stepped rule‐of‐thumb
Example 19.5 Music therapy in hospice care
19.3.6 Specific objectives. 19.3.6.1 Revising the effect size
19.3.6.2 Revising the SD
Example 19.6 Revising the SD – major depressive disorder
19.3.7 Impact on the main trial size
19.3.7.1 Upper confidence limit (UCL) inflation. Example 19.7 Revising the main trial size – music therapy in hospice care
19.3.7.2 Noncentral t‐distribution (NCT) inflation
Example 19.8 Non‐central t‐distribution – major depressive disorder
19.4 Considerations across external‐pilot and main trial
19.4.1 Optimising overall sample size
Example 19.9 External‐pilot and main trial
19.5 Internal‐pilot studies
19.5.1 Continuous outcomes
Example 19.10 Internal‐pilot – children with non‐anaemic iron deficiency
19.5.2 Binary and other outcomes
19.6 Other preliminary studies
Example 19.11 Treatment of tungiasis
19.7 Reporting
CHAPTER 20 Further Topics
20.1 Introduction
20.2 Adaptive approaches
20.2.1 Sequential designs
Example 20.1 Mattress types for operative procedures
Example 20.2 Pressure sore prevention trial
Example 20.3 Amytrophic lateral sclerosis
20.2.2 Other adaptive designs
Example 20.4 Acute myeloid leukaemia
Example 20.5 Prostate cancer
20.3 Large simple trials
Example 20.6 Chronic heart failure
20.4 Bayesian methods
20.4.1 Mechanics
Example 20.7 Prior, likelihood and posterior distributions
20.4.2 Constructing the priors
Example 20.8 Nasopharyngeal cancer
Example 20.9 Hepatocellular carcinoma
20.5 Interim analyses. 20.5.1 Alpha‐spending function
20.5.2 Updated prior
20.5.3 Predictive probability and predictive power tests
Example 20.10 Gastric cancer
20.5.3.1 Final analysis
20.5.3.2 Very small trials
20.5.3.3 Comment
20.6 Zelen randomised consent designs
Example 20.11 Vaginal cuff infections following hysterectomy
Example 20.12 Patients in acute psychiatric crisis
20.7 Systematic overviews
20.7.1 The protocol
20.7.1.1 Literature search
20.7.1.2 Assessing quality
20.7.2 Combining trial results
20.7.3 Forest plot
Example 20.13 Ovarian cancer
20.7.4 Heterogeneity
Example 20.14 Corneal astigmatism
Statistical Tables
Glossary
References
Index
WILEY END USER LICENSE AGREEMENT
