Talk:Linear no-threshold model/Archive 2

Archive 1

Archive 2

Clarifications

Conrad, thank you your recent edits:

• [1] Some (plural) => One; suppressed => ignored
• I also prefer "extrapolate" to "project", since there's a line with a slope involved.

I wonder if we could get more information about little strips that workers carry around with them, to measure cumulative radiation exposure. Like, 10 hours times 10 (rate) = 100 (units) => worker has had all we dare allow; reassign him to other work. Same for 100 hours times 1 (rate), over three weeks or 100 hours times 0.1 rate over six months. Is there an industry standard? Or is this above the "threshold"?

Another thing to write about is background radiation. Have there been any studies about the cumulative effects of living for decades in areas of the world where there is a significant rate or level of natural radiation? I mean, has anyone tried to test the LNT hypotheses?

Is it just a model, used as a safety precaution by workers exposed to radiation hazards in their work, or what? --Uncle Ed (talk) 15:04, 1 February 2012 (UTC)

Glad you agreed with these edits. I feel one can argue pro or against the LNT later on in the article, but the introduction should stick to giving the most important definitions and background information to Wikipedia readers.

The problem with the safety threshold (formerly known as the tolerance dose) is that (1) no one can tell for sure whether there is one (2) no one can tell for sure where it would be. Statistic, long-term effects of radiation have been observed above 100 millisieverts (e.g. among the Hisroshima survivors), so it is almost certain that if there is a threshold, it must be lower than or equal to 100mSv. How much lower? Could be 100 mSv, could be 10 or 1, or the LNT could be right all along down to 0, there is no definitive answer yet.

As for the regulatory use of the LNT model, it actually is a 3-stage waterfall:

1. Scientists (more prominently epidemiologists) study the available data and propose risk models. The current reference for radiation risk is the National Research Council's Health Risks from Exposure to Low Levels of Ionizing Radiation:BEIR VII Phase 2 (available on-line).
2. Based on these data, some advisory boards (the National Council on Radiation Protection and Measurements in the US) or associations (the International Commission on Radiological Protection worldwide) propose radiation protection guidelines, such as the The 2007 Recommendations of the International Commission on Radiological Protection (extract available online).
3. Then each country can set up their own radiation safety regulations by using all or parts of these guidelines (see for example 10 CFR 835 amendment and Nuclear legislation in OECD countries).

Thus, each country sets limits on yearly dose intake for nuclear workers (and radiographers!), and these workers wear dose meters so that their dose intake can be compared against these dose limits:

• If I'm not mistaken, the current yearly dose limit for nuclear workers in the US currently is 50 millisieverts. It's lower (20 mSv) in Europe and the US have started moving toward that value as well.
• Based on the current scientific knowledge and on the currently accepted LNT model, this yearly 50mSv limit is not proven to be safe (no safety threshold) but it is assumed to be an acceptable health risk.
• If a nuclear worker were to work for 40 years and were to receive 50mSv each year, he then would receive 2000mSv = 2 Sv throughout his working lifetime. From the current BEIR VII's risk model, this worker would then have a 20% excess risk of having a cancer throughout his lifetime (compared to the US population-wide 40% lifetime cancer risk), leading to a 10% excess risk of dying from cancer (compared to the US population-wide 20% cancer death risk).
• Since a 10% excess cancer death risk is not negligible, the 50mSv yearly limit is no safety threshold, and radiation protection is all about keeping workers as far as possible from this limit: this objective is known as the ALARP principle (as low as reasonably practicable).
• BTW, the ALARP principle is a major reason for the opposition to the LNT model. When you aim at zero-risk scenarios, there is no end to the radiation protection effort, and radiation protection costs increase forever even when health risks have become negligible.

Regarding your "worker has had all we dare allow; reassign him to other work" assumption, I don't work in the nuclear industry so I'm not sure how they deal with the issue in normal conditions. I believe doses to permanent nuclear workers are minimized by rejecting away high risk situations on non-permanent workers (nuclear jumpers or nuclear gypsies). I'm pretty sure this is done here in France, and I believe it's the same in the US, but I can't give any guarantee.

However, this is definitely how it works for the clean-up at the Fukushima Nuclear Plant (which obviously is an extreme scenario): they hire workers, then have them work until they reach about 50mSv, then hire others. That's why the clean-up at the Fukushima Nuclear Plant has been using so much manpower (about 19000 workers have worked there in less than a year!).

Sorry for this lengthy explanation. I have to run, so I'll get back to your other questions (high background radiation areas and LNT testing) some other time. Regards, ConradMayhew (talk) 08:02, 2 February 2012 (UTC)

Huh? Don't apologize. That answered my question. Now we just have to add all that information into the article and other related articles, like dose meter (or radiation dosimeter); and the article about cleaning up Fukushima after the tsunami. Some of it is there, but a consolidate section on "Radiation precautions for cleanup workers" would be good, instead of scattered comments. --Uncle Ed (talk) 18:35, 2 February 2012 (UTC)

Thx! Could be done someday... As a foreign English speaker, I rather shy away of large WP article edits! Anyway, back to the questions I left pending yesterday.

Regarding the testing of the LNT, I have a feeling it's not only part of the solution, it's also part of the problem.

• On the one hand, epidemiologists perform studies in the field of large human populations, for long time periods (up to several decades). However, they generally can't yield precise answers below 100mSv.
• On the other hand, radiobiologists can get results well below the 100mSv epidemiology limit. However their studies are done in the lab on animals or human cell cultures, and for limited time periods only (a few months at the most).

At the end of the day, epidemiologists and radiobiologists get results that can be at odds, and it's hard to reconciliate the two approaches. A good example is the recent publication by Neumaier et al (ref 24 at the end of the LNT article, "Evidence for formation of DNA repair centers and dose-response nonlinearity in human cells"). The authors claim to have found dose-response nonlinearity in human cells. I actually disagree with their conclusions, but that's another issue: let's assume they are right and the dose-response for isolated breast cells is non-linear. Are we sure that these findings scale up so that the dose-response curve is non-linear at the organ levels (whole breast) then at the individual level (whole body)? Would people bet their life (or their wife's or daughter's lives) on results obtained from a few cells? That's the whole dilemma of radiobiology results: they may disproof the LNT (and often do), but as long as there is not some sort of reconciliation with epidemiology studies, they can't be and won't be taken at face value.

Then we're back to epidemiology studies. There are plenty (see BEIR VII Phase 2), but they require large populations and rather large doses. Let us assume a population of 100,000 people receive a dose of 100 millisieverts each. Based on the current LNT model, they then have an extra 1% risk of getting a cancer at some point during their lifetime, so their risk of having a cancer someday just went from 40% (US population-wide lifetime cancer risk) to 41%... Not a big difference, is it? In total, this extra 1% risk means there should be an extra 1,000 cancers. Assuming an 80-year average lifespan (I don't know the true Us lifespan), that's an extra 12.5 cancers per year. The challenge then is to detect these extra 12.5 cancers per year among the normal cancer background, i.e. 500 cancers per year for a 100,000 people population. Needle in a haystack, drop in the ocean...

So we need a large population. Then we need a large enough dose. For a 10 millisievert dose, the extra cancer-risk predicted by the LNT model is only 0.1%. For the same 100,000 population, we must now detect one extra cancer per year among the normal 500 cancers per year! That's plain impossible, which is why epidemiology studies seldom go much below 100mSv.

Another issue is the time it takes to get definitive results. For the Hiroshima survivor study, I read somewhere that researchers in the seventies believed they had reached such definitive results. Then cancer figures within the survivor population started to rise, and rise again. Researchers started to get a much clearer picture around 1997, 50 years after the bombing. Assuming (again) an 80-year lifespan, I would imagine they'll have the definitive results somewhere around 2030 or 2040 (that's definitely a wild-guess). "Long-term effects of irradiation" mean just that: long term!

And to make matters worst, there are dozens of confounding factors. How to make sure of the individual dose? For Chernobyl cleaners, figures are very inaccurate for a variety of reasons, such as cleaners from the army being encouraged to show their patriotism by not wearing their dose-meters. People evacuated from the Fukushima area have lost their homes, their farms, their jobs; they could have lost relatives in the Tsunami; a lot of them are about to lose their unemployment benefits. People who have not been evacuated must cope with the anxiety of living with radiation: a thing you cannot see and cannot feel but that can harm you... They are often scared to death, and often feel guilty as parents for being unable to protect their children. So anxiety-related disorders are rising, and so are dangerous behaviors such as alcoholism. Because they are scared they may start a cancer, they go to the doctor more often, so that cancers detection-rate increases (screening bias). If one finds an increase of cancers among these populations, is it due to radiation or is it due to any other of these confounding effects? Your guess is just as good as anyone else's...

I'll stop there for the LNT test issue. This was only a quick tour. At the end of the day, there are plenty of attempts, but also plenty of difficulties! I'll try to answer to the high background question next. ConradMayhew (talk) 08:23, 3 February 2012 (UTC)

Again, most of what you're "commenting on" could go into this or related articles. Where's that article about longitudinal studies on cancer-causing factors? I wonder if we have several scattered articles like:

• asbestos and cancer - I'll look in Asbestos
• cell phones and cancer => Mobile phone radiation and health
• Ultraviolet light and cancer
• Tobacco and cancer => health effects of tobacco

One problem with "studies" like this is asking people to remember what they did years ago. Someone with a tumor on the left side of the brain may remember putting his cell phone up to his left ear, better than someone who didn't get a tumor (selective memory). This isn't dishonesty, it's a normal bias. But it's still a problem for science. --Uncle Ed (talk) 15:39, 3 February 2012 (UTC)

Thanks for your comments. Again, I'll try to put this content for real into WP someday, but time is of the essence. On a talk page, I can say things from memory, use an informal tutorial-like writing style, and I don't worry too much about my English. On a regular article page, my writing time goes double or triple! So if there is any content you feel might be useful anywhere, please feel free to use it and rewrite it at will. I swear I won't sue you (and I'll browse through my literature database for the sources).

Re "Where's that article about longitudinal studies on cancer-causing factors", sorry I'm not sure I know what you're referring to. Is it a WP article? If it's a scientific reference, I believe the reference for radiation-induced cancer risks still is BEIR VII Phase 2. The book itself obviously is quite a piece of work, but they have done an excellent leaflet. BTW, I didn't provide any reference for my claims on non-radiation-related health issues in Fukushima (material is scattered though my literature database, and there are many publications I haven't collected yet). Still, these issues were well documented for Chernobyl, so the Chernobyl Forum' report probably is a good starting point, despite its weaknesses.

Regarding the other cancer-related pages, well... My field of work is (or was) radiology physics, so I can't claim any expertise on these topics. Yet... God, is the Ultraviolet light and cancer article a joke? Anyway, if I'm not mistaken, LNT currently is used almost only for radiation toxicology (a debate seems to be going on extending it BTW). I think the model currently used in chemical toxicology is a linear-threshold one. Yet, that's definitely out of my league, so I won't comment further.

I didn't know about the selective memory bias! Yeah, it does make perfect sense, actually. Thx for the piece of intel.

Back to the LNT questions, I realized I overly simplified some things last time, so I need to go just a bit deeper.

Firstly, the risk models established by BEIR VII are not always LNT-based. The BEIR position could be summarized as a rule of thumb: if you can fit your data with a straight-line, and if the fit is not much better with a more complex model, stick to simplicity and to the straight-line fit! Thus, "for solid cancer incidence the linear-quadratic model did not offer a statistically significant improvement in fit, so the linear model was used. For leukemia, a linear-quadratic model was used since it fitted the data significantly better than the linear model."(BEIR VII, page 15).

Secondly, contrary to what I wrote (oversimplification!), there are cases where "long term effects" of radiation may appear rather quickly shortly after radiation exposition:

• For childhood leukemia, the onset can be pretty quick following radiation exposure[2]: after a minimum latent period of 2 years, risk rises and reaches a peak some 7 years after exposure, then slowly goes down again. That's partly why so many studies focus on leukemia: it's sort of an early marker!
• Childhood thyroid cancer in the USSR following Chernobyl was an other exception. A rise in the figures for this cancer started as early as 1990, but it seems no one believed it at first, as no one expected such an early onset for a solid cancer. So, it was first believed to be a screening bias. Around 1995, it was obvious the effect was real: the Chernobyl fallout had reached non-evacuated areas where there was chronic iodine deprivation, so when the kids drank contaminated milk, their thyroid took in the radioactive iodine like hell. If I remember well, Chernobyl-induced thyroid cancers in the ex-USSR should have peaked between 2005 and 2010 and should now be on the decline. But even then, the story is far from over: these early thyroid cancers were aggressive but rather easily cured, and there were only 15 deaths. Thyroid cancers with a later onset may be mode difficult to cure, and the death toll could rise in the years to come. Time will tell...

I have to stop there for today... but I definitely hope to answer the high background radiation area question soon enough! Regards, ConradMayhew (talk) 15:16, 5 February 2012 (UTC)

Some believe that if radiation is distributed evenly enough, so that the levels are comparable to the natural levels, it has no harmful health effects

I removed this sentence from the start of the Controversy section because:

• It gives too much to emphasis to one of the publications listed in the section (Health Physics Society's position statement) while the other publications are in apparent disagreement.

• The Health Physics Society's position statement does _not_ say that doses comparable to the natural background have no harmful effects. Rather the "risks of health effects are either too small to be observed or are nonexistent."

• The statement as it was could cause confusion by implying that there is something special about the natural background level from a health physics point of view. This is not true and in fact it is believed that natural background radiation causes cancers, hence the work to remove radon gas (part of the natural background) from buildings. Stephen David Williams (talk) 20:08, 25 August 2012 (UTC)

The effects of natural variation in background radioactivity on humans, animals and other organisms

Natural levels of radioactivity on the Earth vary by more than a thousand-fold; this spatial heterogeneity may suffice to create heterogeneous effects on physiology, mutation and selection. We review the literature on the relationship between variation in natural levels of radioactivity and evolution. First, we consider the effects of natural levels of radiation on mutations, DNA repair and genetics. A total of 46 studies with 373 effect size estimates revealed a small, but highly significant mean effect that was independent of adjustment for publication bias. Second, we found different mean effect sizes when studies were based on broad categories like physiology, immunology and disease frequency; mean weighted effect sizes were larger for studies of plants than animals, and larger in studies conducted in areas with higher levels of radiation. Third, these negative effects of radiation on mutations, immunology and life history are inconsistent with a general role of hormetic positive effects of radiation on living organisms. Fourth, we reviewed studies of radiation resistance among taxa. These studies suggest that current levels of natural radioactivity may affect mutational input and thereby the genetic constitution and composition of natural populations. Susceptibility to radiation varied among taxa, and several studies provided evidence of differences in susceptibility among populations or strains. Crucially, however, these studies are few and scattered, suggesting that a concerted effort to address this lack of research should be made.

http://onlinelibrary.wiley.com/doi/10.1111/j.1469-185X.2012.00249.x/abstract;jsessionid=5BC4E92A9C9CC4CE2ED064299EA8C07B.d03t03

Jpritikin (talk) 12:50, 19 November 2012 (UTC)

2011 in vitro study

"In 2011 an in vitro time-lapse study of the cellular response to low doses of radiation showed a strongly non-linear response of certain cellular repair mechanisms called radiation-induced foci (RIF). "

In the study the low dose used was 0.1 Gy, the high dose was 2 Gy. Given that humans exposed to doses above 0.1 Gy are likely to develop radiation sickness, maybe the article should include the definition for "low dose" used in the paper?

• "We typically refer to “low dose” or “high dose” as doses below or equal to 0.1 Gy or larger than 1 Gy, respectively."
  

Also, the non-linearity was observed at the high doses: "we observe an absolute RIF yield that is surprisingly much smaller at higher doses" Ssscienccce (talk) 01:05, 5 October 2013 (UTC)
Mentioning the same study a second time as "Recent fundamental research", using a second reference (that points to the same source) is misleading. The quote: "casts considerable doubt on the general assumption that risk to ionizing radiation is proportional to dose" requires context, in their conclusion, they write: "extrapolating risk linearly from high dose as done with the LNT could lead to overestimation of cancer risk at low doses."
May I also suggest that the LNT part of the study was simply added for "publicity"? It didn't follow directly from their results, for that you would have to measure misrejoining frequencies. But it was safe for them to "hypothesize", since it had been proven a decade ago by others. Löbrich et al. reported in 2000 that DSB misrejoining frequencies due to X-ray irradiation strongly decreased with increasing fractionation (spreading the radiation over time in several smaller doses). That already proved conclusively what this 2011 study only suggests. So please spare us the "recent fundamental research" talk. Ssscienccce (talk) 02:59, 5 October 2013 (UTC)
A decent overview of what is known: http://radiology.rsna.org/content/251/1/6.full Ssscienccce (talk) 03:03, 5 October 2013 (UTC)