This will be a series of posts on the underlying science supporting BEIR VII. There have been a number of comments reflecting various degrees of misunderstanding (and outright ignorance) of science and epidemiology.
First, no science is perfect. We don't even fully understand gravity. Science is about finding the best explanation that fits the facts. If you want to play the "science isn't perfect therefore X", read no further. If you have evidence for X, then that becomes a part of science. We can test evidence.
Second, epidemiology by its nature is a science of estimation. So it's clearly more imperfect than a science in which highly precise measurements can be made, like in chemistry where you pour 1.0 ml into a beaker. Or mass spectroscopy where you measure minute masses.
In epidemiology, we're trying to look at the relationship between an exposure (usually called a dose, but it's different than a radiation dose. Frequently a dose is a weight or concentration of something...concentration of radon, mg of arsenic, etc.) and an effect (cancer, blindness, redness, etc.).
We know before we do any study, there is going to be a 0 dose/0 effect point. And if there is any discernible effect at all it will be above that point (+ dose, + effect). There will always be a "hole" between 0 (the origin of the axes) and the lowest discernible effect point. You might be able to lower the dose&effect down towards zero, but at some point (ug, fCi/L, uSv) the statistics requires a jump to 0.
We don't really care much about this low dose/low effect point, we assume the effect continues to 0. Unless, of course we have some evidence to think otherwise. We'll get into that in a bit.
There are many types of epidemiological studies. It would be ideal to have a large number of people, where the group is composed of many clones of identical people. Let's say I could clone 5 more of you. And clone 5 of someone else, and on and on. Then I could take you and a single clone from everyone else and form a group of unique individuals. Then I could do that again and again until I have 6 groups of unique individuals. But in each group there is an individual who is exactly the same as an individual in each of the 5 other groups. Then, one of these six groups would be the unexposed group, and I could expose the other 5 groups to a unique, uniform, well characterized radiation field. I could achieve 5 various doses all below 0.1 Sv, but > 0. This would be ideal radiation epidemiology to continue the existing LNT curve, but obviously it's not going to happen.
But even with such epidemiology, there would be a +dose, +effect point above the 0 point. There would still be a gap. And since real humans are exposed to a wide range of radiation levels, there will always be a gap of knowledge associated with the dose some real human beings actually receive. That doesn't mean any underlying mechanisms of cancer induction changes at that very low dose level. It just means we can't statistically discern it from epidemiology, which in the example above is the best possible epidemiology.
It's a detection threshold, not an effect threshold.
The "perfect" radiation study described above is similar to how a drug study is run (without the clones, but with close matches of people). But before running a drug study, the developer has some sense from pharmacology how the drug should perform. And then the developer will run animal studies. Assuming those studies show efficacy (the drug works) and no significant toxicity, the developer will move on to human studies.
(With radiation our experience from radiobiology is that radiation is deleterious. Animal studies show that radiation induces cancer.)
Assuming the drug developer's human studies go well (show efficacy, little or no toxicity), there will be an LNT (assuming that is the outcome of that particular drug) relationship and it will have the detection gap discussed above.
With a drug (intended to help people), we can call that gap a threshold and adjust the LNT curve (see graph from earlier blog). We can do this because it would be unethical to sell someone a drug with a dose which isn't likely to work. And we want drugs to work in short time frames, not over decades. So we take the position that someone has to prove there is some toxicity which we haven't observed, when we take the threshold position with a drug.
And guess what? Once the drug is sold to millions of people, toxic effects can start to appear that couldn't be statistically discerned before. That's just statistics, not "evul" drug companies.
Since all doses above the detection gap for radiation are carcinogenic, it's up to someone to prove there is an actual threshold or hormesis in the detection gap. It would be unethical to expose people to what is likely a small carcinogenic dose, without ample reason. Because we would probably be causing more cancers than would otherwise occur.
Next: Radiation Study Options
No comments:
Post a Comment