Intro to Epidemiology

Before becoming an epidemiologist, I was a medical technologist. A medical technologist is someone who works in the clinical laboratory, performing tests on human samples like urine, blood, etcetera. My work in the lab exposed me to a lot of very interesting medical cases at the hospitals where I worked. Many of the health care providers I worked with would call the lab to ask about what tests would be the best to diagnose a given condition. As time went by, I did a lot of research into diseases and conditions and how to diagnose them. After a few years, my colleagues suggested that I study epidemiology. So I did, and I learned a lot more than I could ever have learned just googling things left and right. There is something to be said about a formal education, you know? Now I wish to share that knowledge, in my own special way.


Let me give you an introduction to epidemiology, as always, after the jump…

Bring with you the curiosity of a child

Simply stated, epidemiology is the study of anything that happens to the people. In popular culture, epidemiologists are seen as infectious disease experts, chasing down patient zeroes and placing entire cities on quarantine. In real life, epidemiologists study all sorts of things aside from infectious disease. Epidemiologists study everything from injuries to school absenteeism. I know an epidemiologist who works with a well-known telephone company, studying and understanding dropped cellular phone calls.

A good part of our training as epidemiologists is in biostatistics. This is because we will be conducting a lot of investigations in which events that seem associated are really not. If we fail to make that connection between a true cause and a true effect, we may miss the opportunity to put into place control measures to stop what is happening (whether that “happening” is an outbreak of respiratory disease or dropped calls). Another reason why we are trained in biostatistics is because we will be reading volumes of research studies about all sorts of things, and we need to know if the implied associations in those studies are due to chance or because there really is something there. One of the ways to do that, the most important way, is to look at the statistical analyses of the studies.

Previously, I wrote about Benjamin Franklin’s pamphlet on smallpox inoculation in the 1700’s. He presented data on how many people died from acquiring smallpox the “regular” way (person-to-person) and on how many people died from being inoculated artificially (a form of vaccination before smallpox vaccine came along). On its face, the numbers are compelling. Of those who were infected normally, 9.4% died. Of those who were inoculated with the virus in a controlled manner, 1.4% died. That should be enough to convince anyone that controlled inoculation is the way to go to get a grip of smallpox in the 1700’s. Right?

We live in a universe where there are a lot of coincidences. As human beings, we look for patterns all around us in order to explain what we experience. For example, almost every time it rains there is thunder and lightning. So we learned to seek shelter from rain when we hear thunder and see lightning. Now, we know that thunder and lightning do not always lead to rain, and that there isn't thunder and lightning every time it rains. It may be that the storm is going to miss us all together, or maybe the storm already passed. But we still associate thunder and lightning with rain and vice-versa. I once heard a loud crash outside my apartment and, because it was raining, assumed it was thunder. It wasn't. A car had just run into a telephone poll.

So how do we tell whether it is all in our heads or there is really something to thunder and lightning meaning rain?

This is where biostatistics comes into play. You can go over to the Franklin post and read the evidence that the numbers he reports were not so just out of chance. In fact, you can repeat the experiment of inoculating some and not others a thousand times and maybe once will you see more dying from the inoculation than from the natural disease. (And that’s a big maybe.) In short, biostatistics give us an objective view of what we are observing, without our inherent human bias.

It’s not all about the math, however. The data you get out of an experiment or a study is only as good as your design. Let’s say you’re looking into the ability of a pill to cure a disease. Let’s say that there already is a treatment for the disease, but your company is promoting the pill as a one-and-done cure. What would happen if you pick only people on the existing treatment to get into the study? Would you be able to definitively say that the pill cured them and not the existing treatment? What if you don’t try the pill on a control group? Designing the study properly is imperative to data that is meaningful.

After study design and proper use of biostatistics to analyze the data, presenting your findings is the third most important factor in public health, in my opinion. Let’s look at the following two graphs*:

It would seem from this example that the vaccine for disease X, which was introduced in 1983, really did wonders for our population until about 1997, when vaccination requirements were relaxed. The number of cases just skyrocketed, and this graph proves it. Right? Yes, if this is all the data that are presented. But let’s look at another graph from the same dataset*:

Ah, did you see that? The rate per 100,000 population decreased and remained steady, with a slight increase not after 1997, but closer to 2000. See, the number of cases increased after 1997, but it was more a factor of a jump in the population’s size than because of an outbreak or something similar.

See how presenting data can be manipulated? I hope you do. And this is what a lot of groups with special interests will do. Everyone from anti-vaccine groups to lobbyists will massage the data to get the effect they want. Look at what happens if I change the scale of the axis on the right:

Why the rate just plummeted for about eleven years, didn’t it? Again, it has to do with how you present the data. In all cases, you need to do it honestly or don’t do it at all, because someone is going to catch you, and it will be very embarrassing. Just ask Dr. Andrew Wakefield.

Yes, I will dive into some controversial topics while conducting this “night school.” And I know not all of you will agree on some of the conclusions we'll reach. I assure you, I will use nothing but science and math to explain these concepts. I'm not here to convince you of anything in these "lessons" but to give you the tools so you can see why we recommend vaccines, why we ask that you buckle up, why we tell you to get plenty of sleep, and why outbreaks are so hard to do. We'll have an honest discussion, you and I.

And that is the great thing about social media. You can discuss all these topics with me from the comfort of your home, and we can learn from each other. So with that introduction done, let’s go into lesson number one, descriptive epidemiology, next time. Thank you for your time.

*I made up the data for this example. In later lessons, we’ll use actual data sets.