Radiation, Risk and the “Linear No-Threshold” Model

Earlier this month I posted about the longstanding debate over the ultimate death toll from the Chernobyl accident, and a new look at the data by a Union of Concerned Scientist physicist. Lisbeth Gronlund pored through scattered and hard-to-find data on the distribution of fallout from Chernobyl, crunched the numbers based on a statistical model of likely cancers at different exposure levels, and came up with an estimate of roughly 27,000 additional cancer deaths due to Chernobyl. This stands in stark contrast to a widely-quoted UN estimate of roughly 4,000, but also to estimates by Greenpeace and others of 90,000 or more cancer deaths.

This week, on the 25th anniversary of Chernobyl, we had Gronlund on the show to talk about her findings. She explained that the order of magnitude difference between her estimate and that of the UN study ultimately came down to how far afield they each looked.

“The official numbers have been lower, but it’s because they’ve only looked at a smaller number of people,” Gronlund said. “They looked at the people who were in the most highly contaminated areas. And what my number includes is people who would get cancer not just in Europe, but beyond Europe. Because the contamination from Chernobyl was quite widespread.”

A listener took Gronlund to task for her methodology. “Bill_Woods” commented on our website,

“As I expected, this study relies on the ‘Linear No-Threshold’ assumption; the idea that a given amount of radiation will produce the same number of fatalities, whether it’s concentrated on a few people or spread over billions. Which at very low levels is unsupported by evidence and pretty dubious in theory. No other poison works that way…”

Well, if the Linear No-Threshold model is dubious, it seems that most of the world’s scientific establishment has been duped into accepting it. It’s the scientific standard used by, among others, the US Nuclear Regulatory Commission, the US Environmental Protection Agency, the US National Academies, the United Nations and the International Atomic Energy Agency.

A little background here.

Establishing the effects of very low radiation exposures with a high degree of certainty has been a vexing challenge since studies of the first radiation victims — survivors of the Hiroshima and Nagasaki atomic bombs — began more than 60 years ago. That’s because it’s extremely difficult to tease out any impact of low doses from what likely would’ve happened anyway. Roughly 40 people in 100 will develop cancer at some point in their lives, and roughly half of them will die of it.

Meanwhile, according to a 2005 US National Academies report, a single radiation dose of 100 milisieverts will cause only one additional cancer among these same 100 people. That’s already a very small percentage, and most people exposed to man-made radioactivity receive many times less than 100 mSv, making any cancers from that exposure exceedingly difficult to identify.

That’s where the Linear No-Threshold (LNT) model comes in. In essence, it asserts two principles:

• that there is no safe dose, or exposure level, of radiation;

• that the likely number of cancers in a given exposed population can be extrapolated from the known impact of a known radiation exposure (e.g. 1 cancer per 100 people exposed to 100 mSv).

Here’s what the US Environmental Protection Agency says about the first of those, the “no safe level” assertion:

“Based on current scientific evidence, any exposure to radiation can be harmful (e.g., can increase the risk of cancer)…”

And here’s what the US National Academies says about the second, extrapolation from the known impacts of known exposures:

“A comprehensive review of available biological and biophysical data led the committee to conclude that the risk would continue in a linear fashion at lower doses without a threshold and that the smallest dose has the potential to cause a small increase in risk to humans.”

Like The World listener “Bill_Woods,” some scientists — and nuclear industry interest groups — take issue with this model. The American Nuclear Society, for instance, says “there is insufficient scientific evidence to support the use of the Linear No Threshold Hypothesis (LNTH) in the projection of the health effects of low-level radiation.”

Even some of the scientific institutes and regulatory agencies that use the model hedge their acceptance of it. Here’s what the International Atomic Energy Agency has to say:

“… at low doses of radiation, there is still considerable uncertainty about the overall effects. It is presumed that exposure to radiation, even at the levels of natural background, may involve some additional risk of cancer. However, this has yet to be established.”

So yes, there are questions and even doubts about the LNT model, and efforts to establish more certainty are ongoing, although given the complexity of the challenge, it may never be possible to establish much more certainty. In the meantime, in the absence of absolute certainty, policy decisions have to be made, so scientists and policy makers go with what they judge to be the best information available. And despite its limitations, the LNT model has been adopted as the best available by leading scientific and regulatory bodies.

Which brings us back to our interview with Gronlund.

As a journalist who is not a scientist, I have to try to establish the level of uncertainty surrounding any given scientific claim or model, and then make judgments about whether and how to report that in any given story. For me, its use by the above bodies indicates a pretty high standard of scientific acceptance for the Linear No-Threshold model. Yes, there are uncertainties, but it’s my judgment that they’re not significant enough to challenge Gronlund’s use of the model in a radio interview of less than four minutes. Especially since it was the same model used by the UN report that she was critiquing.

••••

Another listener comment on our website seems to dismiss our interview with Gronlund by suggesting we presented the figures of Chernobyl — related cancers out of context — without mentioning that whatever the actual number, they will be dwarfed by the overall cancer numbers in the same population.

Here’s what “FMCoNH” wrote:

”Interesingly and surprisingly Gronlund’s comments are generally logical and factual – atypical for the the UCS. However I find one very key point missing from both Gronlund and Lisa Mullins questions. That is regardless of the size of the population over which the radiation attributed cancer analysis is performed, the resulting number of radiation induced cancers is a very small fraction of the normal cancer mortality rate for the population analyzed. If it’s 4000 “excess cancers” as calculated by the World Health Organization or 27,000 cancers calculated by Gronlund, the important point is that those numbers are minute fractions of normal expectations. That’s why radiation epidemiologists say that any excess cancers (other than childhood thyroid) will be indistinguishable from the normal incidence in the affected populations.

“Any radiation cancer impacts are minor…”

Again, a four-minute interview didn’t allow time to get into this level of detail, but Gronlund does in her study.

But more to the point, how much does this “context” really affect the story of the harm caused by Chernobyl? Sure, statistically, “any radiation cancer impacts are minor.” But people aren’t statistics. Every one of the additional cancers caused by the disaster — whether it’s 4,000 as figured by the UN, 57,000 as figured by Gronlund or more as figured by Greenpeace — will affect real living, breathing people. And roughly half of these people will die sooner, and likely in a more agonizing way, than they would have otherwise. To suggest that these illnesses are somehow unimportant because they “will be indistinguishable from the normal incidence in the affected populations” reduces lives, illness and death to an abstraction. That’s not context, that’s callousness.

It’s true—statistically, the overall additional health and mortality risk from Chernobyl and from nuclear power accidents in general is very small. Listeners and readers can decide for themselves whether or not this added risk is an acceptable trade-off for the benefits they might get from nuclear power. But Gronlund’s point — which I agree with — is that we can’t really make that judgment without as accurate as possible an understanding of what the risk is.