Why Does FDA Tolerate More Radiation Than EPA?

[UPDATED with comment from FDA, at bottom, and EPA's MCL for Cesium-137]
Since the Environmental Protection Agency began detecting radiation in rainwater and milk at levels above its maximum contaminant level, government officials have been downplaying the importance of EPA's maximum contaminant level.
They would much prefer us to speak in terms of the Food and Drug Administration's "Derived Intervention Level."
The two levels could hardly be more different:
EPA does not allow drinking water to contain more than 3 picoCuries per liter of iodine-131 and 200 pCi/l of cesium-137.
FDA allows up to 4,700 picoCuries of iodine-131 in a liter of milk and up to 33,000 picoCuries of cesium-137.
Officials from both agencies—as well as many state governments—explain the difference in terms of time: EPA assumes long-term exposure over 70 years. FDA assumes you're encountering the radiation all at once.
But time isn't the only difference between these two standards:
FDA tolerates a higher mortality rate.
In Hawaii, where milk from Hilo contained the highest levels seen so far, Environmental Health administrator Lynn Nakasone suggested the EPA's standard is irrelevant to milk contamination.
"It's like drinking two liters of water for 70 years to get (the EPA's) limit," Nakasone told the Honolulu Star-Advertiser. "So if you extrapolated to milk, you'd have to drink two liters of milk for 70 years to get that limit."
Nakasone prefers the FDA's standard. But here's what Nakasone isn't telling Hawaiians:
The EPA's level is calculated so that in a population of one million people, the radiation will result in no more than one additional cancer fatality.
The FDA standard, on the other hand, accepts two extra cancer fatalities in a population of 10,000.
Why does the FDA tolerate more radiation, and more mortality, than the EPA? I posed a question Wednesday morning to FDA spokesman Siobhan Delancey, who said:
Let me check with my experts and get back to you, okay?"
Okay. When she does get back to me, I'll add her answer to this post, so stay posted. Meanwhile, I'll give you the answers we found in documents from both agencies.
First, I have some people to credit and thank. I owe this post to some of the other participants on this page who have become diligent researchers on this topic. Chargirl in particular dug up pertinent documents from the FDA. Mothra, and rickcromack have been dogged in their pursuit of facts. And daviddelosangeles has chipped in too.
As Chargirl pointed out in a comment yesterday, FDA's Derived Intervention Limits are not radioactive exposure limits. In the FDA's own words:
FDA has set Derived Intervention Levels for foods prepared for consumption. These levels do not define a safe or unsafe level of exposure, but instead a level at which protective measures would be recommended to ensure that no one receives a significant dose.
In other words, the FDA's DIL is set at the point at which a single liter of milk is so radioactive, you should take protective action.
The number itself is conservatively estimated, with children and the elderly and our most vulnerable citizens in mind—but in practice, the DIL is more a commercial level than an exposure-safety level: DILs are recognized internationally as the level above which foods are unfit for sale or trade.
The EPA's MCL Goal, by contrast, is "the level of a contaminant in drinking water below which there is no known or expected risk to health."
And that's not just over a 70-year period. EPA's annual MCL for iodine-131 is equivalent to 700 picoCuries per liter, according to this EPA document.
That means FDA's 4,700 picoCurie limit for one liter of milk is almost seven times higher than EPA's exposure maximum for a year.
To arrive at that level of tolerance, FDA has to accept a higher mortality rate. But why would it?
I suspect it has something to do with the cost/benefit analysis that some regulatory agencies are required to conduct when they set standards.
EPA's mandate is to protect public health while avoiding a "significant economic impact" to industry. If EPA finds high levels of radionuclides in a municipal drinking water system, the water can be cleaned relatively cheaply. Depending on the specific contaminant, the water can be treated with reverse osmosis, activated carbon, ion exchange, or better: all three.
If FDA finds high levels of radionuclides in milk, that milk can't go to market. That cow can't be implemented with a treatment system. And that dairy farmer faces a significant economic impact.
So the FDA observes a much more tolerant standard, and the impact is transferred to those theoretical two people in 10,ooo.
If FDA has another explanation, as I said earlier, I'll add it to this post.
I should point out, once again, that the administrators of both agencies agree that the radiation contamination levels in the U.S. are far below levels of concern.
EPA Administrator Lisa P. Jackson, in testimony before a Senate committee yesterday:
EPA has not seen and does not expect to see radiation in our air or water reaching harmful levels in the United States. While radiation levels are slightly elevated in some places, they are significantly below harmful levels.
From FDA's Radiation Safety FAQ:
At this time, there is no public health threat in the U.S. related to radiation exposure. FDA, together with other agencies, is carefully monitoring any possibility for distribution of radiation to the United States. At this time, theoretical models do not indicate that significant amounts of radiation will reach the U.S. coast or affect U.S. fishing waters.
The EPA's MCL is due for review in 2015. There have already been allegations that EPA plans to relax radiation standards. In the wake of this conflict of agencies, expect someone to try to relax the MCL for radionuclides.
Comment from Siobhan DeLancey of FDA Office of Public Affairs, sent 4/19/2011:
The EPA MCLs are based on consumption of water every day for a period of 70 years under “normal” circumstances where little to no radioactivity is expected. The FDA DILs are conceived for emergency conditions only and provide a level of contamination where protective actions should be considered to avert further dose from accidental (or terrorist) contamination of food. This averted dose is referred to as the PAG or Protective Action Guide and for food is defined as 500 mrem (5 mSv) whole body (CEDE) or 5 rem (50 mSv) to a single organ (CDE). It is not intended, under the FDA paradigm, that an individual would continually consume contaminated food for a full year. However, for risk estimation purposes only, we have determined that, if someone were to consume contaminated food for a year, he/she would receive a dose estimated at 500 mrem (5 mSv) committed effective dose equivalent (CEDE), which corresponds to an excess risk of cancer mortality of approximately 1 in 4400 above the baseline of 1 in 5 for all people before any excess radiation exposure.
The terminology “mortality tolerance” is not used in practice and should not be used to imply that FDA is willing to allow consumption of radioactive food based on an “acceptable” level of mortality in the population. Risk coefficients (one in a million, two in ten thousand) are statistically based population estimates of risk. As such they cannot be used to predict individual risk and there is likely to be variation around those numbers. Thus we cannot say precisely that “one in a million people will die of cancer from drinking water at the EPA MCL” or that “two in ten thousand people will die of cancer from consuming food at the level of an FDA DIL.” These are estimates only and apply to populations as a whole. Our protective action guides and derived intervention levels are designed to avoid excessive dose and limit the risk to individuals from contaminated food. Further, our values have such a degree of conservatism that even if one were to consume food at the DIL, it is not conceivable that he/she would actually receive the PAG of 500 mrem/5 mSv. FDA would implement and recommend protective actions/interdiction long before anyone received a significant dose.