Oh that chemical? I wouldn't worry about it.

It seems that today one of the favorite topics on shows like 20-20 and 60 minutes is "chemicals". The shows examine things like why BPA will kill you and why nuclear waste will kill everybody.

There are a few reasons why these shows work so well. Firstly, the average person doesn't know much about chemistry. Most people didn't like chemistry when they took it. It's a fairly difficult subject, and there is less logic in it than mathematics and it is less intuitive than the physics you usually learn in high school. Another reason is that unnatural chemicals are all around us. They're relevant.

I'm definitely not going to say that all of these reports are blown way out of control, I just want to point out that, as I have mentioned before, you shouldn't believe everything you hear, even if it's from "experts." As a few examples, consider the books Silent Spring and Slow Death by Rubber Duck. (As a disclaimer, I haven't read either book in it's entirety, only selections from each and both laudatory and critical commentary.)

Silent Spring was written largely as an attack on the use of DDT, which the author claimed was resulting in the near extinction of many animal species. Her book was a very large part of the decision by many countries to ban the use of DDT. In the developing world, the abrupt halt in DDT application resulted in thousands of malaria related deaths that would have been easily prevented had DDT continued to be used. Even if DDT was just as bad as they claimed it was, this would have been an awful decision, but as it turns out DDT doesn't seem to be nearly as bad for you as was originally claimed. Still, it sounds scary and people don't understand it, so, as usual, political policy trumps good public health policies.

Slow Death by Rubber Duck is a similar type of book. It discusses all of the things that you encounter everyday that contain chemicals that can kill you. Although several selections from the book bother me a bit (it bothered another group of people so badly they named a "bad science" award after the book), one is the section where the authors talk about the amount of BPA they find in their bloodstream after a week of self experimentation. I think this article does an excellent job of pointing out the key problem: yeah, levels of the chemical went up, but the levels were so small that it is very unrealistic to expect it to be causing a problem. In addition, just as with DDT, BPA doesn't seem to be as bad as it's made out to be. The results of studies on the health effects of BPA are so inconclusive that even the EFSA (the European version of the EPA) thinks it's silly to regulate it.

It seems that these types of problems in accounting for health effects pop up all over the place. In a recent article on one of my favorite websites, STATS.org, there is a discussion about attempts by government policy makers to use models relating vice-taxes and health to decide how much of a health benefit would come from a certain tax on alcohol. The article does an excellent job of pointing out the main flaws in the model. I thought this example was particularly interesting because in a recent posting I presented an article that discussed why alcohol may be good for you. Combining this and the criticism of the STATS.org author, the model used to predict health effects could likely be qualitatively wrong, meaning that the actual result would be overall worse health instead of improved health for certain levels of tax.

To wrap up, I want to present on last argument, and this is the subject of using LD-50 data to predict fatality rates from a chemical release. I think Alan Waltar did an excellent job of covering the problem with this in his book on nuclear energy, so I will use a modified version of his example here.

Let's assume there is a major chemical spill and arsenic is released. Arsenic is one of the most deadly poisons in the world. The LD50 for arsenic (the amount it takes to kill 50% of subjects exposed) is only 13 mg/kg body weight, so for a fairly average person weighing 80 kg it takes only 1040 mg to have a 50% shot at killing you. Let's say the 80 kg you is exposed to 104 mg of arsenic. What are your chances for survival. A quick bit of math says that it should be around 5%. What about if you were secretly given that 104 mg dose of arsenic for 10 days? It looks like the chance of death is about 50% again.

Well, it looks like, but it's not. To demonstrate how preposterous this is, consider water. Water is, believe it or not, toxic if you have too much of it. As a matter of fact, it's LD-50 has been measured to be 90 ml/kg (for rats, the unfortunate species on which most of these tests are performed). So if you still weigh 80 kg after your encounters with arsenic, it would take 7200 ml to kill you half of the time. Using the same analysis we used for arsenic, you would die 5% of the time whenever you drank 720 ml (about 3 cups) of water. If you kept this up for 10 days, you would have a 50% chance of dying. Actually, you'd probably be pretty dehydrated.

The point of this is to demonstrate why assuming the effects of chemical exposure scale linearly with the exposure amount is not just overly simplified, it's just plain wrong. Things are much more complicated than this, and using accurate measurements that don't apply will result in incorrect conclusions about the danger of certain scenarios. This is one of the reasons why practical options like dumping nuclear waste into deep ocean trenches will always be rejected. It's too easy to come up with a practical sounding and scary argument (and also because of monster movies, which always seem to involve a situation like this...)

To conclude: in the future, remember to be careful, but it's probably not as bad as they say it is.