Experimental%20Study%20Designs

1
Experimental Study Designs

• Dr. Birgit Greiner
• Dep. of Epidemiology and Public Health

2
Learning objectivesAt the end of this lecture
you

• Are familiar with areas of application of
  experimental studies
• Know different types of experimental studies
• Understand the essential components of the design
• Appreciate potential limitations
• Are familiar with ethical issues in trials

3
Experimental Studies
Type of Study
Alternative Name
Unit of Study
Randomised Controlled Trials
Clinical Trials
Patients
Field Trials
Healthy People
Community Intervention Studies
Community Trials
Communities
4
Trials are useful for evaluating

• New drug or other treatment for disease
• New medical/health care technology
• Methods of prevention
• Methods of health promotion
• New health protection policies
• Programs for screening and diagnosis
• Methods of providing health care
• New health care policies

5
Example of an RCT
A randomised controlled trial investigated
whether preoperative smoking intervention reduced
postoperative complications in patients
undergoing hip or knee replacement Moeller AM,
Villebro N, Pedersen T. Toennesen H. The
Lancet, Vol 359 Jan 12, 2002
6
Experimental Research Requirements
Intervention
Randomisation
Control
Blinding
7
Randomised Controlled Trial
new treatment
group 1
Outcome
population
Outcome
group 2
control treatment
8
Randomised controlled trial
POC
Preoperative Smoking intervention
No POC
Patients undergoing hip or knee replacement
POC
Standard care
No POC
9
Fundamental Design Issues
Overview
Chance - Sample Size
Bias - Randomisation/Blinding/ Standardization
Confounding - Randomisation
10
Chance Sample size

• Large enough to have high probability (power) of
  detecting a clinically important difference of a
  given size
• Size of effect
• Error level (usually 5)
• Statistical power (usually 80)
• Variation of outcome

Use StatCalc in the EpiInfo Software
http//www.cdc.gov/epiinfo/EI2000.htm
11
Bias
Systematic deviation of results from the truth
Measurement bias Sytematic error arising from
inaccurate measurement
Selection bias Error due to systematic
differences between those selected for the study
and those who are not
12
Confounding
Effects of 2 exposures are not seperated
13
(No Transcript)
14
Validity issues
Internal validity Degree to which the
investigators conclusions correctly describe
what happened in the study
External validity Degree to which the
investigators conclusions are appropriate when
applied to the universe outside the study
15
Advantages of RCTs

• Strong ability to prove causal relationships
• Risk of confounding factors minimised by
  randomisation
• Efficient and powerful statistical manipulation
  possible

16
Disadvantages of RCTs

• Loss-to-follow up of patients and non-compliance
  might constitute major biases
• Often based on volunteer samples, results may not
  be generalizable to the wider population
• Not feasible/ethical for major public health
  questions

17
Basic Ethical Principles in Biomedical Research
Involving Human Subjects

• Experimental procedure clearly formulated in
  protocol and reviewed by independent committee
• Importance of objective is in proportion to risk
  to the subject
• Assessment of predictable risks in comparison
  with forseeable benefits to subject
• Cease intervention if hazards outweigh benefits
• Informed consent from subject
• Subject can withdraw at any time

Adapted from Helsinki Declaration
www.mrc.ac.uk/pdf-ctg.pdf
18
Voluntary consent of human subject is absolutely
essential.. free power of choice, without the
intervention of . ulterior form of constraint or
coercion. Helsinki Declaration 2000
19
Ethical Considerations in RCTs

• Is proposed treatment safe?
• For the sake of trial, can a treatment ethically
  be withheld?
• What patients may be brought into trial and
  allocated randomly to treatments?
• Is it ethical to use a placebo or dummy
  treatment?
• Is it proper for the trial to be in any way
  masked?

Adapted from Hill (1977)
20
Meta-analysis
Pooling of data from multiple studies addressing
the same hypothesis. Provide a more precise
estimate
!
Watch out for publication bias
21
Cochrane Collaborationwww.cochrane.org
Database of Systematic Reviews Database of
Abstracts of Reviews Cochrane Collaboration
Trials Registry Cochrane Review
Methodology