So one of the important qualities of epidemiology is through the observational process,
we can determine risk factors and likely causes of disease.
I say likely causes because I use that word cause very carefully.
We’ll talk later on about how we define what a causal factor is
but in absence of knowing for sure, that something causes something else,
we call that risk factor - when we talk about associations rather than causations.
So many times we don’t know the cause of a disease
but we can associate it with various exposures
for example, streptococcal infection is often followed by rheumatic fever
and sometimes rheumatic heart disease,
so we can prevent rheumatic heart disease by preventing streptococcal infection,
even if we’re not entirely sure what the causal mechanism might be.
We know that rheumatic fever is more frequent amongst army recruits than the school children
so now we know the population to focus on and when an intervention is most likely.
Lung cancer and smoking is are the classic example.
It was fought in courts for many years but whether or not tobacco really was the cause of lung cancer
and we’re pretty sure that it is but in absence of solid laboratory evidence
we had mountains of epidemiological observational evidence showing that people
who smoke tended to have lung cancer more so than people who didn’t smoke.
The power observation in many cases overcomes the need for solid specific causal information.
Observational epidemiology is also useful for understanding mobility and mortality
from diseases in the population as a whole.
We can associate lifestyle factors like driving without a seatbelt or eating too much fat
or having too many calories in our diet or being too immobile
and not moving enough with other kinds of negative health outcomes -
we don’t need to know the mechanism in order to be able to control the outcome
by controlling lifestyle choices and risk factors.
Let’s talk about some of the important task that epidemiologist are engaged in.
Well, the first and most important thing that a population epidemiologist cares about is disease surveillance.
Most modern countries have several complicated surveillance programs
and action all the time including something called a notifiable disease registry.
That’s when a list of key diseases are made so that any time a health professional encounters one,
they must by law inform the government or whoever’s in-charge of that surveillance program
that they saw a case, in this way we have a solid idea of
whether or not our country or population has a particular disease,
some of the key ones include tuberculosis or HIV AIDS or even Ebola
now has made the list in recent months in most countries.
Disease surveillance allows us to detect whether or not an epidemic is happening.
It allows us to detect whether or not a disease is changing its profile in a way,
it allows us to predict or detect anytime the population is changing its behavior
with respect to certain diseases as well.
Epidemiologist are also involved in diagnostic tests.
We’re going to go into greater detail on diagnostic tests in a further lecture
but know right now that these test involved computing things like sensitivity and specificity,
deciding whether or not we can use this test in this context
or whether or not a test is viable as a screening tool to identify individuals
who are good candidates for further investigation further on.
Epidemiologists are also useful in trend analysis
and we will talk a little bit more about trend analysis in a second.
This is only a look at the changes in diseases over time or over populations
and try to ascertain some wisdom from looking at the changes in the numbers
without actually investigating individual cases.
And also one of the important thing that Clinical Epidemiologist and Population Epidemiologist do is designing studies.
Very often, a researcher will contract an epidemiologist to go over their study design
and their protocols to make sure that everything is methologically sound.
Let’s look at some of the issues in trend analysis. Here’s an example from US data.
This is the changes in mortality rates of white women and lung cancer.
As you can see from 1973 to 1995 mortality rates were going up dramatically,
this is a white woman in America dying of lung cancer.
If you’ll look at black women, well, the rates are kind of the same
so there’s no reason to expect white women and black women
to be physiologically different so this is not surprising.
If we look at breast cancer, white women, the rates have come down slightly from 1973 to 1995
with black women they’ve gone up - that’s interesting.
Here’s all the data in one slide, we see that there are no changes or no differences
between white and black women except with respect to breast cancer,
that’s very interesting and there could be a host of reasons for this observation
including something we called detection bias, that’s when we’re looking for more cases
so we find more cases and again we’re talking more about detection bias in a future lecture about biases;
but the point that I’m trying to make here,
is that trend analysis allows us to know what questions to ask,
allows us to ascertain that there's probably something happening here that I need to investigate further.