00:01
So let's now change gears a bit and talk about
terminology. It's important that we get our
words straight before we can talk more deeply
about epidemiology and how to use it. When
we're in mathematics versus clinical research,
versus laboratory research, versus epidemiology,
some common concepts often have different
names. So we have clinical research and mathematical
relationships, and in this domain we have
independent and dependent variables. So an
independent variable is free to be whatever
it wants, but it determines the value of dependent
variables. In epidemiologic research, a variable
that predicts an outcome is an exposure. You're
exposed to something which may lead to an
outcome. You can be exposed to a contaminant,
a pollutant, may be something in your food
and the outcome could be a disease, a cancer
or some other kind of behavior. So again traditionally
we have independent variables that lead to
dependent variables, in epidemiology these
are exposures leading to outcomes. The example
that I'm fond of making all the time is that
smoking is an independent variable, you are
free, you are independent to smoke if you
want to or not smoke if you want to, but the
smoking causes a certain outcome and that
outcome could be lung cancer. Now an exposure
that increases or decreases the likelihood
for developing a certain outcome or disorder
conditions or diseases, we call that a risk
factor. Now go back to John Snow again, John
Snow discovered that water from this pumping
station was likely associated with cholera.
01:41
He didn't know how it worked, he didn't understand
the biology, the mechanism by which water
caused the disease, he just knew that this
water was a risk factor for getting the disease
and this really is one of the foundations
for public health epidemiology. We can measure
statistically the relationship between risk
factors and outcomes and thus we can control
the risk factors and maybe then control the
outcomes without knowing the relationship,
without knowing the mechanism of how that
risk factor caused the outcome, or in fact
if it was indeed causal. For example, smoking
causes lung cancer, we know this because there
is a strong statistical association between
whether or not you smoke and whether or not
you are likely to get lung cancer. You don't
have to know the science or the biology or
the mechanism of how the smoking causes lung
cancer, it helps, it's useful, we recommend
that we figure this out, we don't have to
know in order to have a public health intervention
epidemiologically. I can control the risk
factor and thus reduce the likelihood of the
outcome.