Tuesday, May 17, 2011

Limitations of BEIR VII Estimates of Radiation Risk

While I do love numbers, it’s important to understand their limitations. Numbers are essential for verifying predictions, but the limitations tell you how much you can rely on the numbers I gave in my previous post.

The biggest limitation of BEIR VII, or any other study of radiation effects, is the lack of data. The BEIR VII authors, after analyzing a large number of studies of various populations, conclude that the only studies that meet all their criteria are those of the Japanese atom-bomb survivors. And even those are not perfect: dose was not measured directly, but reconstructed as a function of distance from the blasts. And it’s a one-time dose, not continuing.

Other studies give slightly different results, but they have other problems. Studies of workers who may be exposed to radiation show lower cancer incidence; Chernobyl exposures seem to have resulted in much larger numbers of thyroid cancer than other exposures have. Doses are very well known for the radiation workers, less so for those exposed to Chernobyl fallout. Health studies of workers find that people who are working tend to be in better health than the general population, which may explain the lower incidence. The areas around Chernobyl are known to have a high incidence of goiter, indicating iodine deficiency, which could result in increased iodine-131 uptake and higher cancer incidence.

Probably the biggest limitation is not enough numbers of exposed people to investigate reliably the small increases in cancer that small exposures to radiation may cause. So the BEIR VII numbers have large associated uncertainties, and uncertainty in what the uncertainties might be.
Because of the various sources of uncertainty it is important to regard specific estimates of LAR with a healthy skepticism, placing more faith in a range of possible values. Although a confidence interval is the usual statistical device for doing so, the approach here also accounts for uncertainties external to the data, treating subjective probability distributions for these uncertainties as if they resulted from real data. The resulting range of plausible values for lifetime risk is consequently labeled a “subjective confidence interval” to emphasize its dependence on opinions in addition to direct numerical observation. [page 278]
LAR is “lifetime attributable risk,” the additional risk incurred by radiation exposure.

So, restating the numbers from my previous post on this subject,
For the public limit, the BEIR VII committee’s preferred estimate is in Table 12-6. 1 mSv per year throughout life, the expectation is that there will be 550 cases of cancer and 290 deaths per 100,000 males, 970 cases and 460 deaths per 100,000 females due to this incremental radiation exposure.
in terms of confidence limits gives

280 to 1100 cases of cancer and 140 to 580 deaths per 100,000 males and
510 to 1840 cases of cancer and 230 to 920 deaths per 100,000 females.

The committee has chosen the numbers it thinks are best justified, but the actual numbers might be twice or one-half those numbers. The committee’s estimates are the best we’ve got, but those ranges need to be kept in mind.

It’s important, as well, to put those cancer cases in context. BEIR VII does this in the Public Summary’s Figure PS-4.
In a lifetime, approximately 42 (solid circles) of 100 people will be diagnosed with cancer (calculated from Table 12-4 of this report). Calculations in this report suggest that approximately one cancer (star) per 100 people could result from a single exposure to 0.1 Sv of low-LET radiation above background.
That’s 100 milliSv, twice the yearly allowable dose for radiation workers.

The assumption in BEIR VII that radiation affects health in a linear way down to low doses has been criticized from both sides as giving too high and too low an estimate. The committee’s response is that the experimental evidence seems to be most supportive of the linear assumption [pages 8-9]

I came to BEIR VII fairly skeptical. It has seemed to me that the human body has evolved in a mildly radioactive world, more radioactive at earlier times than later, and that it must have evolved ways to deal with that radioactivity. Many chemical compounds and elements are necessary for life but toxic at higher concentrations; iron is only one example. Might it be the same for radiation?

Working through the report carefully has convinced me that it is the best information we have available. That is not the same as being precisely correct in all aspects. There are many holes in the data and in our understanding of the steps from cell damage to cancer. It is possible that some of the findings in BEIR VII will be overturned as our knowledge in these areas increases.

But that is how science is done: the best understanding is woven out of the available data and used in further investigation, the results of which are woven back into that understanding.

Cross-posted at BMJ Blog.


Karen Street said...

Iodine deficiency appears to be a factor. Also, this is probably the only cohort where children drank large amounts of milk with radioactive iodine. The systematic checking of thyroid health after Chernobyl, certainly better than before the accident, led to moving up the date at which tumors were noticed and included a number of problems which might never have been noticed.


Re small doses, I assume that a number of experiments are still going on to assess the model's power at low doses. I'm several years behind in reading the results at this site: http://lowdose.energy.gov/

opit said...

I have included some notes on the flawed linear scale of risk from radiation ( taken as if all radiation was the same ) in news posts. You might find some useful notes in http://opitslinkfest.blogspot.com/2009/08/uranium-mining-and-depleted.html
I haven't researched http://www.helencaldicott.com/ yet. It looks promising.