Randomized Clinical Trial Design
Home   Tutorials   Video Courses   Case Studies   Software   Research Presentations   Technical Reports

Short Course

Design, Monitoring, and Analysis of Clinical Trials

Scott S. Emerson, M.D., Ph.D.
Professor of Biostatistics
University of Washington


Increasingly, clinical trials are conducted using group sequential methods in order to address the ethical and efficiency issues that arise when performing experiments with human volunteers. The design, conduct, and analysis of a sequential clinical trial is necessarily more involved than that for a clinical trial in which the data would only be analyzed at the end of the study. However, the basic tenets of careful consideration of the scientific question and clinical setting remains the same. In this short course we present an overview of the issues that must be considered when conducting a group sequential trial.

Presentation MaterialsFormatSize (KB)
Presentation Handouts Day 1PDF362
Presentation Handouts Day 2PDF466
Case Studies, Part 1 HandoutsPDF238
Case Studies, Part 2 HandoutsPDF174
Example Protocol HandoutsPDF30

 Day One: Scientific Setting   Day Two: Statistical Design
SlidesAudio-Video (min:sec)  SlidesAudio-Video (min:sec)
1-29 Introduction and Overview: (24:28)    1-41 Sample Size: (73:45)
  Scientific and clinical setting    Standard formula; Application to common probability models; Adjustment for baseline; Crossover; Clustering
30-57Phases of Clinical Trials: (39:37)    42-58 Sequential Sampling: (29:05)
  Definition of an "indication"; Phase I, II, III, IV trials; Efficacy vs effectiveness    Control of type I error; Error spending
58-72Primary Endpoints: (36:47)    59-71 Sampling Distribution: (07:47)
  Multiple comparison problem; Criteria for primary clinical endpoint; Composite / co-primary endpoints    Effect on distribution of estimate, Z, P
73-101Surrogate Endpoints: (22:24)    72-101 Inferential Methods: (20:33)
  Historical examples; Potential mechanisms of treatment effect on clinical and surrogate endpoints; Validation of surrogate endpoints    Adjusted estimates, p values
102-114Refining Scientific Hypotheses: (13:36)    102-108 Nonbinding Futility Boundaries: (11:45)
  Causation versus association; Allowing for variation in response    Effect on power, inference
115-123Controls: (06:35)    109-123 Boundary Scales: (13:01)
 Historical controls; Concurrent controls    Correspondence with error spending, Bayesian, conditional power
124-134Randomization: (40:13)    124-136 Dangers of Statistical Scales: (13:18)
  Establishing cause and effect; Ensuring comparability of treatment arms; Analytic models    Potential for unethical stopping rules
135-142Blinding: (16:33)    137-147 Relative Advantages of Scales: (27:08)
  Advantages; Logistics; Blinding of DSMB    Use of estimates, Z scores, error spending scales
143-177Summarizing Treatment Effect: (34:02)    148-169 Stochastic Curtailment: (25:36)
  Statistical primary endpoint: Summary measures; Criteria for selection of summary measure; Use of the mean; Censored data    Issues with conditional power; predictive power
178-202Scientific Burden of Evidence: (15:51)    170-185 Evaluation of Designs: (15:50)
  Superiority, noninferiority, approximate equivalence; Clinical trials with active controls    Criteria for choosing among candidate designs
203-230Ethical Issues re Treatment and Eligibility: (22:41)    186-203 Adaptive Designs: (15:57)
  Definition of treatments: Planning for noncompliance; Definition of target population: Inclusion and exclusion criteria    Adaptive re-powering of study
231-254Statistical Criteria for Evidence: (15:30 )    204-208 Documentation of Design: (08:03)
  Frequentist and Bayesian frameworks; Parametric, semiparametric, and nonparametric probability models    Parameters to specify in protocol
255-297Criteria for Sample Size: (23:45)    37-72 Sepsis Case Study: (29:39)
  Reporting results; Precision of results   
298-300Scientific Review of Clinical Trial Design: (04:05)    73-126 Hodgkin's Case Study: (40:09)
  Scientific issues; Scientific review of statistical issues   
email RCTdesign.org