00:01
Hello and welcome to epidemiology. Today we're
going to look at what I think is your favorite
study design. I think so because it's probably
the most famous study design, the most important
one and definitely the most expensive one.
It's the foundation of drug delivery. It's
the randomized controlled trial or randomized
clinical trial or RCT. It's a kind of experiment.
00:21
In fact we'll be talking about all kinds of
experiments in this lecture. After this lecture,
you're going to know the characteristics of
a randomized controlled trial or RCT. You're
going to know why some RCTs are blinded. You'll
know what that means after today as well.
00:36
You're going to know the characteristics of
a quasi-experiment, which is like an experiment
but not quite. You're going to know the characteristics
of what's called a natural experiment and
about something called an ecological study,
which may or may not be an experiment, but
we'd like to talk about it in the same breath
as experiments.
00:55
So a randomized controlled trial is a kind
of a experiment, it's also the gold standard
for evidence. We like to use that term a lot,
gold standard. Iit's our best test for evidence
in terms of causality, does this cause
that, does an exposure cause an outcome, does
a drug actually cause a change in the patient.
It's the highest form of proof that one thing
causes another and pretty much every approved
drug or treatment or therapy or technology
on the market has gone through at least one
rigorous round of RCT testing before the government
allows it onto the marketplace. So today in this
lecture, we're going to cover where experiments
fit onto the big map of study designs that
we keep going back to. We're going to look
at the basic design of an RCT, which is of the
best kind of experiment for epidemiological
purposes. We're going to look at how an RCT
differs from a case-control study, even though
they use the same words. We're going to look
at the advantages and disadvantages of an RCT,
because even though it's pretty good as a
study design, it has got some bad parts to it
as well. And we're going to talk about why
it is the 'R' exists in RCT, why you randomize,
have you given this some thought? Why do we
prefer this thing called blinding, I want
you to think about that as well and we're
going to find some answers together. And we're
going to look at the design of this thing
called a quasi-experiment which resembles
our RCT, but is lacking one important characteristic,
the 'R', the randomization part. We're also
going to spend some time to talk about the
difference between an internal and an external
validity, both kinds of validity, but distinctly
different from each other. And we're going to
look at something called a natural experiment
as well as, as I mentioned, an ecological
study.
02:43
So where does an RCT fit onto our grand map
of study designs? As you recall, the universe
of research is divided into qualitative and
quantitative research. Qualitative are things
like interviews and focus groups. Quantitative
is when we use numbers and statistics.
03:01
The world of quantitative research is divided
into descriptive and analytical research.
03:06
Descriptive research is when we describe something,
the who, what, where, when, never the why,
of something going on, usually with one variable
in mind. When we do an analytical study, we're
comparing or drawing associations between
two or more variables. And there are two kinds
of analytical studies, observational and experimental,
do you remember what other kinds of observational
studies there are? I'll help you, there's
a case control, there's a cohort and sometimes
there's a cross-sectional as well. There are
other types as well, but those are our three
main ones. In the world of experimental studies
though, we have the RCT, that's our favorite
go to experiment in. It's an interventional
clinical trial. So, the RCT, as I mentioned,
is the only method that we understand that
offers undeniable proof that something causes
something else. It is the gold standard for
determining causality. We use it to measure
usually the efficacy of new drugs, that's
why RCTs are mostly conducted by drug companies,
probably because they have a vested interest
in measuring the efficacy of their products,
also because RCTs are enormously expensive
to do well and only drug companies and governments
tend to be wealthy enough to do good ones
and large ones well. So RCTs are study designs
that can be used for evaluating a variety
of different things, usually approaches to
treatment and prevention. Is this treatment
better than nothing, or better than the existing
standard, or better than a placebo, or better
than some other kind of therapy? We use it
in the clinical setting, or in the community
setting. A lot of people haven't thought about
that. Can you think of some examples when
an RCT is appropriate for a non-clinical setting,
a non-drug trial, a non-biotech trial, a non-intervention?
It can be used in a community; perhaps you're
testing the differences in educational interventions
or government policies. The design is applicable
to a variety of scenarios, not just medical
scenarios. So RCTs are used to evaluate new
drugs and treatments, they're used to assess
new programs and they're used to determine
new ways of organizing and delivering healthcare
even. Again, we can just randomize people under
a variety of different paths and compare them
after randomization. So let's go back to the
question that I'd like to bring all my study
designs back to, is there an association between
smoking and lung cancer? Smoking is exposure,
lung cancer is the outcome. In an RCT design,
this is how we'd approach this. We have a
defined population, maybe adults living in
this country and we choose some of them and
we randomize that sample into two groups,
some will be smokers and some will be non-smokers.
They don't get to choose. I get to choose.
05:54
I'm the researcher. I'm the investigator,
I've randomized them. I make this motion because
that's a coin flip. That's the most basic
kind of randomization, flip a coin, heads
you smoke, tails you don't smoke. So I'm deciding
the exposure that you get. After you're exposed,
we wait some time and we see which group gets
what outcome. If it's a drug trial, we're
seeing which group has more people improve
from the disease versus the other group, if
it's smoking and lung cancer, we're looking
to see the prevalence or incidents rather,
of lung cancer in both of the groups and comparing
the two. The key there is the randomization.
06:31
Now it sounds similar to a case-control study
right, because we have cases with people who
get diseases or the people that we're going
to get the intervention, they have controls
of people who don't, case-controls uses the
same words, but in RCT it is quite distinct
from a case-control study. A case-control
you may recall from our lecture on observational
studies, begins by ascertaining the outcome,
deciding who got a disease and who didn't
and then looking backwards in time to see
what the exposure was. An RCT is quite different,
we look forwards in time, first of all its
prospective and also I, the researcher, decide
who gets what exposure. Now because it's prospective
it also resembles a cohort design, but it
is distinct from a cohort design because cohort
designs, the world unfolds as it will, we
simply observe the outcome, we don't get to
choose who gets what exposure, they choose
themselves. In an RCT we, the investigators,
we choose who gets what. So the basic design
again, we have our study population, individuals
we've recruited to be a part of our study,
we randomize them into two groups. Study A,
gets the treatment, maybe it's a drug,
usually is. Study B, doesn't get the treatment,
they are the control group. They're either
getting nothing, they're getting the current
therapy that's different from the therapy
I'm testing, or they're getting something
called a placebo, which we will talk about
in a bit. You know what a placebo is already
I bet, but we're going to talk about it a
bit more anyway.