1. Design of Clinical Trials
Oye Gureje

2. The nature of clinical trials
Clinical trials are intervention studies
Therapeutic intervention (secondary prevention)
Preventive intervention (primary prevention)
Set up to explore the superiority of one type of therapeutic intervention or preventive intervention over another
Experimental in nature
Exposure is introduced by investigator
Why experiment?
Observation is prospective rather than retrospective

3. Selecting the study population
Reference population (to which the outcomes may apply)
Nonparticipants
Experimental population (considerations: size, outcome events, follow-up)
Study population (fulfilling criteria, consenting, etc.)

4. Defining cases and controls in clinical trials
Criteria for inclusion and exclusion must be rigorously defined
Controls must be similar to cases in every relevant respect other than in the feature of interest
Adds strength to association by reducing uncontrolled confounding
Is matching necessary?
Cases and controls must be representative of the population
Sampling

5. The “cross-over” design
Cases are their own controls
“cases”: when exposed to intervention of interest
“controls”: when not exposed to intervention of interest
Observation may be affected by:
changes in ambient environmental factors
natural course of illness

6. Special features of clinical trials
Human experimentation
Full disclosure
Informed consent
Protection of human subjects
Explicit criteria for withdrawal
Analytic strategies for drop-outs and unequal attrition
Unique problems
Ethical considerations
Feasibility
Cost

7. Outcome measures in clinical trials
Efficacy
Side effects and interactions
Time course
Pharmacoeconomics
Prevention

8. “Blinding” in clinical trials
Necessary to reduce bias (e.g. observer bias) and control for placebo effect (e.g. Hawthorne effect)
Types
Single (simpler but with observation bias)
Double (eliminates observation bias; but complex)
Requirements
Identical-looking interventions/preparations
Multiple researchers
Difficult with unique and obvious effects of interventions

9. Analyzing clinical trials
Experiment designed to test the Null Hypothesis
Sample size determination and statistical power
Dependent on not just number of participants! (but also number of end points and differential attrition)
Who to include in analysis
Effects of multiple comparisons
Type I and Type II errors