Comment by acidburnNSA | Hacker News Reader

Comment by acidburnNSA | Hacker News Reader

Parent Full thread

Oh yeah I didn't mean to imply that the fringe group numbers are not published. There are a few groups who have published very high numbers. They are very fringe. For these big-science things you have to look at meta-analyses. An idea with a few publications should not be weighted evenly alongside one with thousands. That's my beef with the show.

The assumptions you have to believe to believe those numbers run into big trouble when you compare low natural background radiation areas with high background ones. For example, Ramsar, Iran has ~100 mSv/year background, compared to 3 mSv/yr in the USA. And Guarapari, Brazil has up to 30 mSv/yr. And smaller variations exist throughout the world. Yet we don't find statistically significant differences in cancer rates as a function of dose rate at any of these levels. So taking a dose rate far lower than these values, using a unjustifiably conservative dose-to-death conversion factor, and multiplying by large numbers of people is not a great method. This is what the UNSCEAR people came out and said in their recent reports. See section 3.1.f in the 2012 UNSCEAR report [1].

The UNSCEAR team also stopped saying their own 4000 and 9000 numbers because they also relied upon using the unjustifiable method of multiplying low doses by large populations too.

[1] https://www.unscear.org/docs/publications/2012/UNSCEAR_2012_...

EDIT: Adjusted dose rates in Iran below absolute peak (200 mSv/yr) via input in comment below. And correction section reference in UNSCEAR report

> In the published estimates shown, studies have utilised a methodology termed the ‘linear no-threshold model’ (LNT).

LNT is neither a method, nor is it a model. It is an assumption, which happens to be convenient for regulatory purposes. Let me quickly explain how it works.

Caffeine is poisonous with an LD50 of around 15g or about 100 cups of coffee. If two people each drink 100 cups of coffee, one of them is expected to die. If we now apply the "method" of LNT, it follows that if 200 people each drink one cup of coffee, one of them is expected to die. Moreover, since 85% of adult Americans consume on average 164mg of caffeine daily, which is roughly 30t in total, we can conclude that caffeine kills one million Americans every single day.

Now you can evaluate for yourself whether LNT makes any sense.

This is a ridiculously bad analogy. It’s so bad, I would expect it only from right-wing disinformation efforts who are self-parodies of incorrectness.

There is no correspondence between an acute chemical poisoning mechanism and the mechanism of ionizing radiation induced illness for longer-term effects.

A more correct analogy for LNT, indeed, a “model”, would be a rifle fired at random in to a room. If there is one person in the room, there is say a 1 in 5,000 chance the bullet will hit them. If they are hit, there is a 1 in 10 chance they will die.

If you fire the rifle a million times in to a million rooms, you get a linear scale-up and expect 200 people to be hit, and 20 of them to die. It scales linearly, with no threshold below which the mortality rate declines.

If you pack the room with more people the probability of hitting someone goes up, etc.

Now there are indeed well-based arguments to say that there are indeed thresholds of effects of ionizing radiation, including studies where low ambient levels are present in populations, to studies attempting to show radiation hormesis effects. Those studies have not displaced LNT thus far because they are not clearly conclusive in understanding what they are saying.

LNT has been the standard for 70 years not because it was promulgated by idiots, but rather because it is a conservative protective measure designed by the best minds of the time, using the best information that was reliably known.

So please, discuss hormesis, show alternative models, etc., but please do not try to impugn the model as stemming from idiots - the Manhattan-era physicists were certainly miles smarter than anyone reading HN today - and has not been displaced yet because no one understands a better model of how to protect people from injury and death.

The way LNT works with radiation is that we knew a death-dose curve from the atomic bombs in Japan at extremely high dose. This was acute radiation poisoning (comparable to chemical poisoning). This response curve is precisely the thing that has been extrapolated down to miniscule doses just above background. The analogy you are critiquing with acute chemical dose is right on.

The rifle fired at random in a room analogy is vastly more flawed and does not take into account how the body deals with radiation. There are cell damage events all the time, and there are repair mechanisms that fix them in the body at a baseline rate. These repair mechanisms make radioactive cell damage look a lot like chemical damage at low levels.

Radiation has been a part of the universe forever. 50% of the internal heat of the earth is from radiation. Space is radioactive. Cosmic rays rain down on us all day and all night. The body can and does deal with this, up to a point, after which it becomes inundated.

Thus, a small boost in radiation, far below natural background, cannot be expected to increase deaths.

> It’s so bad, I would expect it only from right-wing disinformation efforts who are self-parodies of incorrectness.

That's not very helpful. It would be helful, if you explained how it is a bad analogy for ionizing radiation. (By the way, that wasn't even the point. I merely explained LNT in a context where everyone immediately recognizes it as ridiculously wrong.)

> do not try to impugn the model as stemming from idiots

I would never! The guy who came up with LNT ten years after the conclusion of the Manhattan project, Herman Muller, was not an idiot!

He was a fraud.

My apologies, but your analogy is flawed. An adverse effect of ionizing radiation is that it could cause a mutation that replicates (before it's caught by the immune system) - cancer. It's therefore intrinsically probabilistic: if you're exceedingly unlucky, a small amount of ionizing radiation could get you there. That does not apply to your caffeine analogy, which actually does have a threshold.

Now, there's indications that at least the linearity of the LNT model is off in that there IS any effect of the immune system on the outcome, but that at high doses, it may be suppressed thus making the response non linear. I'm sure there's other effects of that nature and I don't think this is particularly well understood. But I dare say the linear no threshold model is actually somewhat more supported by common sense than you imply. Your distinction between the words "model" and "assumption" is moot by the way. A model is an attempt to (typically mathematically) describe a complex process. That's always an assumption in that sense. :)

(Former high energy physicist. Not an MD. Not a medical physicist.)

> An adverse effect of ionizing radiation is that it could cause a mutation that replicates - cancer. It's therefore intrinsically probabilistic.

This statement betrays a very naive understanding of cancer and complete ignorance of DNA repair pathways. Don't take it personally - nobody understands cancer completely. However, we know that a single mutation will not cause cancer. Cells have "defense in depth" against adverse mutations. There are multiple repair mechanisms, DNA exists in four copies usable for reconstruction, and there are crowbar mechanisms that kill a cell when it detects damage, and there is an immune system which can kill rogue cells. Therefore, to cause cancer, a number of mutations is needed (probably around ten). Clearly, the risk of cancer isn't linear in the dose, it grows approximately with the tenth power.

There are other mechanism absent from your naive view. Ionizing radiation typically doesn't damage DNA directly (alpha radiation can cause double strand breaks, but beta and gamma don't). Rather, it creates reactive oxygen radicals which can oxidize DNA, causing miscoding lesions. (These are repairable, to some extent.) But cells contain compounds that scavenge such radicals (the most important one is glutathione). If glutathione has be exhausted before oxidative damage occurs, that creates a threshold effect.

So response to ionizing radiation is likely nonlinear and has a threshold. LNT is twice wrong.

> Your distinction between the words "model" and "assumption" is moot by the way.

Then I will make it more clear. A model is something that is tested and discarded if it turns out to be no good. An assumption is something that is kept despite failing the tests. LNT was based on nothing (Herman Muller made it up, claiming knowledge about how DNA is affected; and that was before the structure of DNA, let alone the existence of DNA repair was known), which may be forgivable. But it should have been discarded after its predictions didn't match the reality of the Nuclear Navy dock workers study, or the Taiwanese radioactive high rise buildings, or how low dose full body irradiation suppresses cancer instead of causing it.

I shouldn't have said assumption. Let's be honest and call it a dogma.

> Don't take it personally

No offense taken! https://xkcd.com/386/ :)

Thank you for the clear example that drives the point home.

LNT for radiation made sense in 1950 before we had sufficient data on how the body responds to radiation. Today, after carefully studying irradiated populations for 70 years, we know enough to conclude that it no longer makes sense.

> Oh yeah I didn't mean to imply that the fringe group numbers are not published

"they disagree with me so they must be fringe"

> Ramsar, Iran has ~200 mSv/year background

from wikipedia https://en.wikipedia.org/wiki/Ramsar%2C_Mazandaran "Record levels were found in a house where the effective radiation dose due to external radiation was 131 mSv/a, and the committed dose from radon was 72 mSv/a.[8] This unique case..." Note the word unique.

So your claim for that area is in fact higher than the highest single location they found (one house), as I read it.

> See recommendation 1e

I would if I could find it. "1e" doesn't appear in that document. Could you quote it for us? Thanks.

> The UNSCEAR team also stopped saying...

Specifically, where?

> "they disagree with me so they must be fringe"

That is not what I am saying. I am saying that they disagree with vast institutions of scientists who explain quite clearly in the above-linked report why they are wrong.

Regarding different background levels, I'm sorry if my numbers were not precise (I was reading them off of this DOE chart [1]), but the point remains that there are variation across at least 1 and almost 2 orders of magnitude in the world's natural radiation at dose rates far higher than what Greenpeace has used to make shocking Chernobyl estimates that can be used to robustly throw out the high-number estimates anti-nuclear institutions have used for years to scare people.

Does that make sense? Greenpeace is literally taking dose rates lower than US average background and saying that any increment even below that level will cause more Chernobyl deaths. This is unjustifiable by any stretch of any imagination.

Happy to quote. Sorry about not being specific enough about where it was. I even got the section wrong. The quote is from page 10 in the report (pg 18 in the pdf). Section 3.1.f:

"In general, increases in the incidence of health effects in populations cannot be attributed reliably to chronic exposure to radiation at levels that are typical of the global average background levels of radiation. This is because of the uncertainties associated with the assessment of risks at low doses, the current absence of radiation-specific biomarkers for health effects and the insufficient statistical power of epidemiological studies. Therefore, the Scientific Committee does not recommend multiplying very low doses by large numbers of individuals to estimate numbers of radiation-induced health effects within a population exposed to incremental doses at levels equivalent to or lower than natural background levels; "

[1] https://i.imgur.com/TDaZI08.jpg Dose chart (sourced from https://www.energy.gov/sites/prod/files/2018/01/f46/doe-ioni...)

Thank you for clarifying.

People have a ~50% chance of developing cancer and a ~20% chance of dying from cancer in their lifetimes. This means there are vast numbers of people very close to getting cancer at any time and there is no population wide safe exposure level.

To actually disprove the theory you need an experiment that can detect changes in cancer rate on the order of 40k per billion or 0.004% which have simply not been preformed.

Saying ‘we can’t measure it therefore it does not exist’ might be acceptable if it where the oldest theory. Except it’s not, people want to overthrow the existing idea with zero direct supporting evidence which is simply unscientific. If you want to supplant a theory then it’s on you to provide evidence that the existing theory is wrong, otherwise I can say low exposure levels are more dangerous because high levels result in multiple indecent cancers at the same time in some people. Therefore the risks increase at lower levels do to probability theory and we already agreed no evidence was needed.

PS: I am not saying my example is correct, just the argument your making is faulty.

> To actually disprove the theory you need an experiment that can detect changes in cancer rate on the order of 40k per billion or 0.004% which have simply not been preformed.

If the LNT model was correct then you wouldn't need such a high level of precision, only a sample group with more radiation exposure (i.e. a medium level rather than low or high), so that the LNT model would predict that enough people would get cancer that it could actually be measured. If the people in those areas don't have an incidence of cancer which is linear with respect to the very high doses currently being used as the baseline then the LNT model is falsified.

> Saying ‘we can’t measure it therefore it does not exist’ might be acceptable if it where the oldest theory. Except it’s not, people want to overthrow the existing idea with zero direct supporting evidence which is simply unscientific.

By this logic we should be operating under the assumption that the Greek Gods actually exist because the theory is very old and it is difficult to measure.

If the LNT model was correct then you wouldn't need such a high level of precision, only a sample group with more radiation exposure (i.e. a medium level rather than low or high), so that the LNT model would predict that enough people would get cancer that it could actually be measured. If the people in those areas don't have an incidence of cancer which is linear with respect to the very high doses currently being used as the baseline then the LNT model is falsified.

The problem with this assumption is you can’t keep everything else identical. Population X and Y don’t grow up live and die in exactly identical areas expect for different radiation levels.

What you can do is control for as many factors as possible or use a proxy. It turns out skin cancer is a solid proxy, * and populations like Australia and skin cancer support the linear model down to the limits of our ability to detect it. But, then people want to suggest the non linear model kicks in as soon as the signal falls below the noise floor without evidence, even though that’s the same behavior in both models.

* As in you can detect population differences and work out whole body radiation dose equivalents.

PS: By this logic we should be operating under the assumption that the Greek Gods actually exist because the theory is very old and it is difficult to measure. People are promoting a new idea, not just trying to disprove an old one. Aka, Greek gods and the flying spaghetti monster have equal support. That’s absolutely fine, it’s saying some other untested theory is supported without evidence that’s at issue.

> The problem with this assumption is you can’t keep everything else identical. Population X and Y don’t grow up live and die in exactly identical areas expect for different radiation levels.

For most other things you would do a placebo controlled trial, but it might be hard to get people to sign up for a trial where you're giving them enough radiation to expect to cause a significant number of them to get cancer.

So controlling for confounders is really the only thing left if you want an answer.

And it works pretty well -- at least better than nothing -- if you actually want the answer. The problem with it is that it unfortunately can also produce whatever answer you want from a political perspective by reverse engineering which confounders to control for based on which ones shift toward the desired outcome, so you can't trust somebody else to do it for you without checking their work at a level of detail that basically amounts to redoing the analysis yourself.

> It turns out skin cancer is a solid proxy

Wait, what are you using it as a proxy for?

It’s a proxy for low level radiation exposure over time.

It’s not great, but skin cancer is more closely tied to radiation exposure rather than everything that can cause lung cancer etc.

Doesn't that require assuming the conclusion? If LNT is wrong then skin cancer shouldn't be linearly proportional to radiation exposure either.

Plus, you still have the same problem with confounders. Expect more skin cancer in a population that spends more time outside, but also potentially a very different lifestyle in that case.

And problems with radiation type as well, since many types of radiation don't penetrate the skin, so they could cause skin cancer but not other cancers. Which has implications for the fatality rate because skin cancer has a much higher survival rate than most other types.

You’re always stuck building a model and testing it in any case. The point is you don’t need to adjust for many common causes of cancer like smoking, so it’s a simpler model.

https://www.sciencedirect.com/science/article/abs/pii/S05315...

In the german wikipedia edition 814 mSv/a is mentioned (based on the linked study) for Ramsar, Iran. To be fair, a comment has been added stating:

"Allerdings dürfte bei etwa 33.000 Einwohnern Ramsars die Kohortengröße von weit unter 5 % für die über 100 mSv/a Exponierten wahrscheinlich zu gering sein, um statistisch signifikante Gesundheitsfolgen festzustellen."

Translated: The size of the measured cohort of 33.000 inhabitants of Ramsar of less than 5% for >100 mSv/a exposed participants is probably too small to allow to derive statistically significant health-related conclusions.

Reply on news.ycombinator.com