Please wait while the transcript is being prepared...
0:00
Hello, everyone,
and welcome to this
Henry Stewart Talk
on the impact of randomization
on the evidence of
a clinical trial.
My name is Nicole Heussen.
I'm a biostatistician and professor
of evidence-based medicine
at Sigmund Freud University
in Vienna, Austria.
I'm also affiliated to the
Institute of Medical Statistics
at the RWTH Aachen
University in Germany.
0:31
After this talk,
you'll understand the link
between randomization, bias,
and the test decision in the
context of a clinical trial.
Furthermore, you will
be able to select
the appropriate randomization
procedure on a scientific basis,
guided by a
structured framework.
0:54
Let me start with an
outline of my talk.
I will start with a
short motivation,
including an illustrative
presentation
of some randomization procedures
that represent specific classes.
On the basis of
these procedures,
I will then present a template
which will guide us towards
a scientifically sound selection
of a randomization procedure.
Finally, I will show, by
means of a case study,
how to use the template.
1:32
Randomized controlled trials
are considered the gold
standard in clinical research.
Randomization describes the
random allocation of patients,
or study participants,
to treatment groups.
One question that
keeps coming up is,
why should we randomly assign
patients to treatments?
What is the main motivation?
Let's take a look at
the ICH E9 Guideline,
Statistical Principles
for Clinical Trials.
Here, it is stated that
randomization and blinding
are essential elements
of clinical trials
that are used to avoid bias.
Since bias can affect the
evidence of study results,
consideration of the influence
of bias is essential.
The practical implementation
of randomization
consists of two steps.
Firstly, the generation of an
unpredictable randomization list,
which results in the
allocation sequence.
Secondly, the concealment of this
list until the final allocation,
which is called
allocation concealment.
There are various
methods available
for generating a
randomization list,
some of which I would
like to present to you
and discuss their properties.
