Please wait while the transcript is being prepared...
0:00
Hello. My name is Brennan Kahan.
I'm based at the MRC
Clinical Trials Unit at UCL
and I'll be talking
about a practical guide
to the estimands framework.
0:14
I'll be giving some
of the background and
motivation to why
estimands are important.
I'll be covering
what an estimand is.
Going through the
estimand framework,
I'm giving an
example of applying
the framework to a real trial
and offering some conclusions.
0:32
So, we know that
randomized trials
are used to answer questions
about the safety
and effectiveness
of different interventions,
but trials can actually be
used to answer
different types of
questions about how safe and
effective an intervention is.
So, if we had a trial where
some patients stopped their
assigned treatment early,
you might be interested
in understanding
the effective treatment if
all patients have actually
taken treatment for
its full course
or else we might be
interested in the effect of
the intervention regardless of
whether the fact some
patients had to stop early.
These different questions
might actually lead to
quite different
answers about how
safe and effective
an intervention is.
So, if we're someone who's using
trial results to inform
decision making,
it's important to
understand which of
these questions a trial has
been designed to answer.
1:26
However, historically,
most randomized trials have
not reported this information.
So, they don't say
exactly which question
or which treatment effect
they're trying to estimate.
So, if we're someone who's
using results from this trial
to inform decision making,
we need to instead
try to infer what
the question is from
the study methods.
So, how they've designed the
trial and done the analysis.
Now, the issue is that this is
actually really
challenging doom,
and I'll give you
an example of that.
