Another piece that requires some background, at a minimum, LNT is Nonsense. In fact, this post is really an addendum to the LNT is Nonsense piece. Inevitably there is some repetition.
One of the mainstays of the defenders of the Linear No Threshold(LNT) radiation harm model is the claim that the cancer incidence data we have is not inconsistent with LNT. So let's stick with it. In the NRC's rejection of the Marcus petition, asking the NRC to discontinue use of LNT as the basis for its radiation regulation, this argument is invoked at least three times.1
Here's how the National Council on Radiation Protection, the official defender of LNT, puts it.
However, few experimental studies, and essentially no human data, can be said to prove or even to provide direct support for the concept of collective dose with its implicit uncertainties of nonthreshold, linearity, and dose-rate independence with respect to risk. The best that can be said is that most studies do not provide quantitative data that, with statistical significance, contradicts the concept of collective dose. ... It is conceptually possible, but with a vanishingly small probability, that any of these effects [leading to cancer] could result from the passage of a single charged particle. ... It is the result of this type of reasoning that a linear nonthreshold dose-response relationship cannot be excluded. 2 [page 45]
Cannot be excluded is all we need. This is not a rule by which science works. Normally we require statistically significant rejection of the null hypothesis to accept a model. But since LNT is so well entrenched, let's accept the very weak not-inconsistent test.
LNT's linearity implies that the only thing that counts is the cumulative dose. This is not only accepted by LNT proponents; but is offered as one of its selling points. If the only thing that counts is cumulative dose, we don't have to worry about how rapidly or how slowly that dose was received. All our calculations become much simpler. As BEIR VII puts it, LNT is a ``computationally convenient starting point"3[p 7]
Unfortunately, the data does not match this convenient model. For example, atom bomb survivors who received a 1000 mSv over a period of seconds showed a 10% increase in cancer. But in high background radiation areas such as Kerala, we see no increase in cancer in people who receive 500 mSv over 15 years.4 LNT says we should have seen a 5% increase.
Figure 1. Cancer incidence in Kerala, India as a function of dose rate
The not inconsistent argument is often coupled with the lack of statistical power claim. LNTers often argue that the reason we can't see the elevated cancer rates at low doses that LNT calls for, is the sample size is not large enough. In other words, ignorance requires us to accept LNT.
Brenner, a strong supporter of LNT, argues that to be statistically confident of the impact of a 5 mSv difference in dose, we would need to study a population of ten million people for their entire life.5[Figure 1] Not inconsistent is enough and no way to show inconsistency means we are left with LNT. But Brenner is done in by LNT's cumulative assumption. Brenner using the same argument calculates that to see the impact of 500 mSv difference we would need 1000 people. The Kerala study of 70,000 people easily meets his requirement.
The NRC in its rejection of the Marcus petition tries to get around this by claiming ``no direct inferences about radiation effects can be drawn from studies where background radiation levels are higher than normal". This is because ``these studies were ecologic in design and utilized population based measurement of exposure rather than individual estimates of radiation dose." This is a possible objection unless you are an LNTer. But LNT is the one model which explicitly allows you to combine individual doses into collective doses. A non-LNTer can make this argument; an LNTer cannot. In invoking the ecologic fallacy, the NRC is rejecting LNT.
And we have a great deal of other than natural background data. Table 1 compares groups which received high doses over periods shorter than the repair period, very roughly a day, with groups who received the same or higher dose over periods of years. For the former groups, we start to see an increase in cancer somewhere around 100 mSv. But we see no detectable increase in people who received more than 10,000 mSv over 10 years. LNT, for which exposure period is irrelevant, says the latter group should have one hundred times higher increase in cancer incidence. The data is not only inconsistent with LNT; the data most emphatically rejects LNT.
Table 1. 100 mSv in less than a day, detectable harm. 10,000+ mSv over 10 years, no detectable harm. LNT says latter group should have 100+ times higher increase in cancer incidence.
The reason why exposure period is crucial is repair. Nature has equipped us with extraordinarily effective processes for repairing the damage radiation can do to our DNA. As Table 1 indicates, these processes can be overwhelmed if the dose rate is high enough. But a model such as LNT that does not recognize our ability to repair radiation damage is not just quantitatively wrong. It is qualitatively wrong.
NRC, Linear No Threshold Model and Standards for Protection Against Radiation, 10 CFR Part 20, August, 2021.
Kathren, R., Principles and Application of Collective Dose in Radiation Protection, National Council on Radiation Protections and Measurements, NCRP Report No 121, November, 1995.
Committee to Assess Risk of Low Level Radiation, Health Risks from Exposure to Low Levels of Ionizing Radiation, National Research Council, BEIR VII Phase 2, 2006.
Nair, M. et al, Background radiation and cancer incidence in Kerala India, Karanagappally cohort study, Health Physics, Vol 96, January, 2009.
Brenner, D. et al. Cancer Risks attributable to low doses of ionizing radiation: Assessing what we really know, PNAS, Volume 100, No 24, November 2003.
Nothing like an unfalsifiable argument that changes over time. That is what passes for “science” now in almost every field.
The NRC already have the answer, and their jobs depend on that answer. Therefore they can point to any tiny data gap in your model even though theirs is wholly inconsistent from the start.
Thank you for this excellent piece. I have one question. You don’t mention the possibility of health benefits from radiation hormesis which seem to be backed up by a significant amount of evidence. I was wondering why: (1) You don’t believe in it, (2) you don’t have an opinion, (3) you believe in it but would rather not go into the debate for tactical reasons, or (4) others ?