Perfect Study?

Okay, this has been bothering me like crazy. ONE STUDY MEANS NOTHING!!!! One study that says high fat diets don't hurt you means nothing.. nothing. Look at the hundreds of studies that say it is bad, not the one (poorly analyzed in my opinion). Why would one study show no effect if there really is one? Well there are tons of reasons, here's one.

One of the first things I learned while at Berkeley is that every single Epidemiological Study is wrong. All of them are wrong (and please read the rest of the blog, don't take this out of context).

First, a basic defninition of an Epidemiological study - a study done on humans. Simple as that . Not rats, or mice, or water, but humans. These types of studies have told us that smoking is bad(so is second hand smoking), seat-belts are good, hormone replacement therapy has problems, and thylidamide is really bad for fetuses.

So, how can I say that every single Epi. study is wrong when we know all these things, plus more, to be true?? Well, first I am not saying that we draw the wrong conclusions from the studies, but that each study only gives us a little glimmer of the truth and one study doesn't tell us much.

Here's why... these are studies on humans, and yes, that is a problem. Suppose I thought that a chemical might keep a fetus from developing correctly. The best way to test this is to take a bunch of pregnant women and expose some to the chemical, make sure that others aren't exposed, and compare the babies when they are born. But here's the catch.. how many pregnant women are going to let me give them something that I think might hurt their child? Well, um, not many. (Do you know any?) And personally, I would feel really guilty running an experiment like that. This is the type of situation is what makes Epi. fairly difficult. We want to see what hurts people so we can stop it but due to that ethics and things like that, experiments are rare.

Epidemiologists have come up with several ways to see what goes in the real world, but they all have problems. One of the best ways is to look at a whole bunch of pregnant women (let's use the previous example again), see which ones are exposed to the chemical, which aren't, and compare their babies. But here's the problem with that. What if all the women that are exposed to the chemical also live in the same neighborhood. Well.... is it the chemical that is causing the problem or something else in their neighborhood? How do narrow down what the cause of the problem?

This makes life pretty hard. Epidemiologists have ways to deal with these things (and believe me, things get much more complex than my example), but they aren't perfect. One study might say the chemical is twice as likely to hurt babies, another 10 times, another might say it's good for the baby. None of these results are perfect. Things are going to depend on who you are looking at, where, when, for how long, their memory of things they have eaten etc. Each study has different people, has looked at different things (such as where they live) and try to take into account different things. But no study is perfect and there are bound to be studies that show one thing when another is true.

So if 100 studies show that high fat is bad, and 1 that says it's good, are you really, truely going to trust that one study? Or are you just paying attention to it because the media and the big ol food lobbyists want you to eat as much food as possible (don't get me started on the food lobbyists. That's for another blog).

I am so sick of people looking at the one study that says the thing that they want and then they ignore the rest of the evidence or, worse, call it junk science because they don't like the results. Can they find faults with the studies? Well yes, there will always be problems. But I can find fault in their's. Let's look at the preponderance of evidence, not one study.

AHHHHHHHHHHHHHHHHHHHHHHHHH. I feel better.