Please wait while the transcript is being prepared...
0:00
Welcome back to Part 2.
Now that we've covered
all the basic aspects
of trial design considerations,
I'm going to go over
the basic concepts and
applications of a number of
trial designs at
the macro level.
0:13
Specifically, I'm
going to cover what
I believe are the
most common designs
you'll come across in
the literature and may
be interested in running
depending on the situation.
I'll also touch on some more
unique and specific types
of trials that can be
applied outside of the
usual circumstances
of wanting to compare two
different treatments.
We'll go through individually
randomized trials
both parallel and
crossover followed by the
cluster randomized approach
and its crossover analog,
the stepped wedge trial.
We'll go over the N-of-1 and
factorial and adaptive trials,
which is an umbrella term for
a variety of designs that share
the commonality of being able to
change as a trial progresses.
Then we'll touch on some
less common designs.
0:52
The individual parallel trial is
the most basic and
most common form
that you'll find
in the literature.
In the design, participants are
randomized to receive either
the intervention or control.
They're usually with
just two treatments and
usually in a one-to-one
allocation ratio,
but the principles
of the individual
parallel design are also
applied in multi-arm trials
with three or more arms or
with different allocation
ratios so long as the unit of
randomization is the person and
they only receive one
of the treatments.
This design suits most
situations and, of course,
all the design
considerations that we
discussed in the last
part of this talk still
means there's an
incredible variation and
ability to adjust
the specifics of
the trial to fit
different circumstances
and different types
of interventions.
1:34
In a crossover trial,
all participants receive
all available treatments
and it is the order
that is randomized.
Using this design, each
participant serves
as their own control,
and the randomization to
order eliminates
potential biases in
non-controlled decisions
regarding which one
might come first.
This design reduces the
variability in the results if
the analyses properly account
for the within-person
variability.
This means fewer
participants are needed.
But this type of design
isn't always possible
because of treatment effects
that have to be considered.
Crossover trials need special
circumstances to be viable.
If the disease is progressive
or moves from one stage to
another in relatively short
time frames then it's not
appropriate because
you would have
an unfair comparison of the
effects of both interventions.
The intervention
can only relieve
symptoms or signs of the
condition and it can't
really have an expectation of
curing it because
then you wouldn't
be able to test
something else in
the second period
of the crossover.
That also relates to
carryover effects
between periods that also
need to be accounted for.
For example, if a drug stays
in the body for a while,
you don't want that still
to be in the system
and have some effect when you
introduce the other drug.
You need to consider appropriate
wash-up periods
durations of time
between the receipt of
the intervention so you
don't have a mixing of
effects in the second period.
