Annual radiation limits are biological nonsense and murderously inefficient

I am not a big fan of AI. It's just a compute intensive way of regurgitating conventional wisdom. But there are two groups which I'm strongly in favor of replacing with AI: shrinks, and government bureaucrats. Weizenbaum's ELIZA showed that shrinks could be replaced with AI as early as the 1970's. If we need any more evidence that bureaucrats could and should go the same route, I offer the latest product of the Idaho National Lab, Wagner et al, Reevaluation of Radiation Protection Standards for Workers and the Public Based on Current Scientific Evidence.

The report like the title is written in early AI prose. It treats LNT in a determinedly even handed style: some people say it must stay, and some people say it must go. But despite claiming that the report was motivated by Trump's EO to "reconsider reliance on the linear no threshold model", INL ends up making no recommendation about LNT at all.

What happened to the Gold Standard?

But on the LNT must go side, there are some strange omissions. DNA repair is treated as a possibility. "Cellular repair mechanisms ... may reduce or eliminate harmful effects at low dose rates.[page ES-1] No, we know our bodies know how to repair DNA damage at rates thousands of times above background radiation damage. And we know why. It's not because background radiation was higher when life emerged, as the INL report implies[p 20]; but rather the need to repair the DNA damage from our oxygen based metabolism.¹ Without this repair system, humanity would be extinct in two generations. The INL report does not even mention metabolic DNA damage.

The radium dial painters are mentioned only to make the point that radiation can be harmful and then silence. The dial painters are the definitive example of how badly LNT handles spread out dose rate profiles. The Taipei apartment exposure, the second most compelling example, is not mentioned. The Karunagappally study is dismissed with:

Despite these elevated chronic exposures, researchers have not demonstrated conclusive evidence of increased cancer rates or other adverse health effects in these populations.[p 60]

That statement is totally bassackwards. When I look at Figure 2, the question I ask is: is this not conclusive evidence against LNT?

Figure 2. I think we can agree this is not "conclusive evidence of increased cancer rates".

But the strangest omission of all is the Gold Standard. The RERF Life Span Study of 85,000 bomb survivors on which tens of millions have dollars has been spent, is dismissed in a single paragraph, claiming that, since these were acute dose profiles, the results aren't really relevant to an examination of LNT. The pre-2012 RERF studies are referenced; but not the more recent analyses based on the much improved DS02R dosimetry, which clearly show the acute dose response is not linear. Here’s how the RERF itself describes the results:

The dose response, however, exhibited statistically significant (P ¼ 0.03) upward curvature in a linear-quadratic dose-response model that assumed common curvature for males and females\cite{grant-2017}[p 523].

The linear-quadratic model in men offered no statistical improvement over a purely quadratic model over the full dose range (P ¼ 0.11).\cite{grant-2017}[p 523]

The results of solid cancer mortality analyses in the 0–2 Gy dose range and over the entire follow-up (i.e., 1950–2009) were similar, revealing significant upward curvature in dose response for each sex.\cite{brenner-2022}[page 13]

In the last 10 years, the RERF Gold Standard has somehow gone from the strongest evidence for LNT to not relevant to the validity of LNT. LNT makes a prediction for both acute and chronic profiles. The RERF data show the acute response is non-linear, and the dial painters, Taipei, Karunagappally, etc, prove that LNT is totally bogus on chronic dose rate profiles. Given the biology, none of this should be surprising.

INWORKS Circularity

So how does INL support LNT? It's INWORKS and more INWORKS.[p 19, 23] There are at least 7 references to the INWORKS study of radiation workers in one form or another. INWORKS has become the new "Gold Standard" when it comes to defending LNT.

INWORKS suffers from the same problem as all the near background and below dose rate studies. Thanks to our remarkably effective DNA repair system, low dose rate radiation is such a weak carcinogen that, if you look only at such situations, it is statistically impossible to separate the effect of radiation on cancer incidence from the multitude of confounding factors. No semi-reasonable model is falsifiable.

This allows the "researcher" to play all sorts of games. INWORKS plays them all.

1) After looking at the data, INWORKS arbitrarily pruned their data base, rejecting 12 of the 15 countries' data, and the American shipyard study, the best controlled of the lot, and several other studies that showed no effect.

2) INWORKS only kept the three countries that had nuclear weapons programs. Unlike their earlier work, they then mixed Cold War bomb workers and nuclear power plant workers, two populations that could be expected to have rather different dose rate profiles.

3) INWORKS does not show us their dose rate profiles despite the fact that we can expect them to be rather spiky, at least for the higher dose, blue collar guys, and despite the fact that nuclear worker dosimetry allows the profiles to be estimated. INWORKS simply assures us with zero support that the doses were "protracted".

4) INWORKS is not reproducible. INWORKS will not show us their dosimeter readings, so we can do our own profile estimates, violating the most basic tenet of scientific research.

5) INWORKS does not have to show us the dose rate profiles because INWORKS assumes that the only thing that counts is total dose. But if the only thing that counts is total dose, you have already assumed LNT. The entire INWORKS argument is circular.

6) INWORKS subtracted out the background dose which for most of their population was larger than the "occupational" dose that they ended up with. INWORKS's results are not based on total dose which they claim is the only thing that counts.

7) For their control group, INWORKS uses the near-zero-occupational-dose sub-group. These were desk job, lots of education years, white collar suits and dresses. The higher occupational dose people tended to be blue collar guys, who had all sort of differences in life style from the control group: eating, drinking, etc. Any one of these differences would overwhelm the effect of radiation for all but the tallest, most spiky dose rate profiles in the population. No serious attempt was made to control for these differences, not even smoking.

Despite all this manipulation and obsfucation, INWORKS can only come up with weak statements based on fallaciously making LNT the null hypothesis, such as "the current status of scientific knowledge does not contradict the use of LNT"[p 23] INWORKS, the new Gold Standard, is as worthless as pyrite.²

Annual limits are biological nonsense and murderously inefficient.

Following standard protocol, INL mischaracterizes LNT as the No Threshold model.

However, it is important to stay focused on the central issue with LNT: that all radiation exposure, without threshold, will create deleterious health effects.p[3]

No, No, No! The central issue with LNT is its foundational premise: which is radiation damage cannot be repaired. INL at least admits that DNA repair could be important, and that this repair is dose rate sensitive,[p 22]. But in their report, there is no mention at all of the all important time to repair. The concept of a dose rate profile shows up nowhere in the report. Instead all INL talks about is total dose or total dose per year. A year is a period that has no biological meaning.³ Since LNT is the only harm model which allows you to ignore the repair period, INL accepts LNT without saying so.

Given that acceptance, INL comes up with some strange regulatory recommendations. It recommends a 50 mSv/y limit for nuclear workers, and proposes that there be no ALARA requirement to reduce the exposure below that level. The LNT lifetime cancer incidence for a 50 year career at this exposure level is about 20%. That's 2000 times the 10 in a million Maximum Individual Risk (MIR) that the EPA deems "reasonably acceptable". Since under INL's reg our most exposed worker could get his annual 50 mSv in a single day, perhaps during refueling, the SNT Most Exposed Person(MEP) lifetime incidence is about 10%. I don't regard myself as a radiophobe, but these numbers are reckless.

INL also recommends a 5 mSv/y limit for the public; but inconsistently does not repeat the prohibition against ALARA reducing the exposure further. Let's assume (almost certainly incorrectly) this was an unintentional omission. The LNT 80 year life time risk associated with 5 mSv/year is 4% or 400 times the EPA MIR. The SNT MEP cancer incidence under this requirement is 1.25% or 125 times the EPA MIR. Annual limits are nonsense and dangerous.⁴

If we change the 5 mSv/y limit to 5/365 = 0.014 mSv per day, then the MEP SNT lifetime cancer incidence becomes 0.54e-6, about 200 times less than the EPA MIR. For the record, the daily SNT limit implied by the EPA lifetime MIR is 0.11 mSv/d.

We must go back to daily dose rate limits, as we had prior to 1950. If an annual limit is set to a reasonable daily limit times 365, it will be recklessly dangerous. If an annual limit is set such that a reasonable daily limit can not be exceeded, it will be auto-genocidally expensive. Either way annual limits kill people.

Enact the NRA

INL's basic approach, implicitly accepting LNT and using annual limits, is both inconsistent with our biology and murderously inefficient. The proper approach is:

1) Decide on a radiation harm model. The model must be consistent with established biology and all the data: low dose, high dose, low dose rate, high dose rate, and all the combinations thereof. This decision should in no way depend on the model's implications for regulation.

2) Decide on acceptable risks, which will depend on the situation under analysis. This decision should in no way depend on the chosen harm model.

3) Run the model in each situation to determine the dose rate profiles and the daily limits that comply with the acceptable risk.

The INL study is just the latest example that government bureaucrats are unable to undertake these three simple steps. The survey of "Current Scientific Evidence" in the INL report could have been done far better by AI. I doubt if AI would have so carefully avoided the exposures that confirm that LNT's denial of our DNA repair ability leads to many order of magnitude errors for the sort of dose rate profiles incurred in a nuclear power plant release.

Elected officials must take on the tasks of choosing a harm model, and choosing the acceptable risks. The The Nuclear Reorganization Act shows how. Congress must enact the NRA.

1. It is likely that early O2 burners, borrowed the rad protection systems of their cyano forebears, and then drastically improved them. See Why are we so good at repairing radiation damage?

2. For a far more detailed critique of INWORKS, check out Circling the LNT Wagons.

3. Equivalently, you are assuming the repair period is a year, which means DNA repair is so slow, it can be neglected. See LNT and Space Travel for more on assuming long repair times.

4. Once again for the SNT Maximum Exposed Person, I have assumed the worst case which he gets his 5 mSv in a single day.