Transcript Blinding/randomization reduces bias

Clinical Trials in PKU
Georgianne Arnold, MD
Professor of Pediatrics
University of Pittsburgh Medical Center
Pittsburgh Children’s Hospital
Disclosure
• I am a member of the PKUDOS advisory
board
• I am involved in Kuvan and PEG-PAL
clinical research funded by Biomarin
• I have been an advisor for Biomarin
Why a Clinical Trial?
• The effect of a new drug or treatment can
be influenced by the power of suggestion,
random chance, and other factors
• A clinical trial attempts to minimize these
influences and give us a scientific way to
determine if the drug or treatment
really works
Why can’t I just have the treatment?
• Blinding/randomization reduces bias
– Investigator bias
– Subject bias
– Chance bias
• The power of the human mind and
expectation are not to be
underestimated!
What do they mean by…?
• Blind – one party does not know
what treatment the subject is getting
• Double blind – neither the subject
nor the researcher know what
treatment the subject is getting
• Randomized: the patient is
assigned to a study arm….
randomly
Why can’t I just have the treatment?
• Blinding/randomization reduces bias
– Investigator bias
– Subject bias
– Chance bias
• The power of the human mind and
expectation are not to be
underestimated!
What does “arm” mean?
• Subjects are divided into various treatment
groups. Measures of outcome are
planned (in this case changes in brain
structure on MRI and changes in
neurocognitive performance).
• Randomization
Diet
Kuvan
Responder
Nonresponder
Outcome - MRI/Cognitive changes
Sample study schemes
Treatment A
Outcome
Treatment B
Outcome
How do you decide how many
study subjects are needed?
• Short answer – ask a statistician
• Long answer – depends on what size
effect you are looking for
– The larger the change, the smaller the sample
size needed to see it
– When sample size is small, there may
be some constraints to your conclusions
How do you decide if it worked?
• There is always the possibility that the
results were obtained by a fluke
• Researchers use statistics to determine
the chance that the result is real
– Generally accepted standard is the likelihood
that the result is real is ≥ 95%
– So, if the chance of error is ≤ 5%
(0.05 or less), then we trust the result
– Try to make the chance that you missed
a real effect ≤ 80%
Who are all these people?
• Study overseen, and often developed, by
the PI (Principal Investigator)
• PI is assisted by study coordinators,
nurses, technicians, and many others
• Other people you may meet include
psychologists, medical specialists, etc.
• Statisticians, data entry assistants,
student assistants and others may
also see your data, but also undergo
confidentiality training
Can I Learn More?
• More basic information on clinical trials is
available in a later session
• Check out www.ClinicalTrials.gov
Barriers to Clinical Trial
Participation
Barriers to Clinical Trial
Participation
• Time!
• Travel!
• Executive Functions!
What Makes Studies Tick
• Study Design
• Study Funding
• Study Execution
• Study Subjects
Clinical Trials Central
Additional information on clinical trials is
available from the National Institutes of Health
Clinicaltrials.gov
Medlineplus.gov