Low-dose radiation exposure slightly increases leukemia risk: study


An international cancer research study has shown that low-dose radiation exposure over a long period slightly increases the risk of developing leukemia.

The study, funded in part by Japan’s health ministry, was reported in British medical journal The Lancet Haematology by a team of researchers, including from the Lyon, France-based International Agency for Research on Cancer.

The study analyzed data on the radiation exposure and health conditions of over 300,000 workers at nuclear facilities in Europe and the United States.

As little research on risks associated with protracted low-dose radiation exposure has been conducted, the latest findings may be useful for the health care of workers at nuclear power plants in Japan, and medical workers handling radiological equipment.

The team assembled data on 308,297 radiation-monitored workers employed for at least a year in France, Britain and the United States over about 60 years.

The study showed the relative risk of developing leukemia would increase by around three per thousand with a 1 millisievert radiation dose and would remain even with exposure to low doses of up to 100 millisieverts.

The workers’ average yearly radiation doses stood at 1.1 millisieverts and cumulative doses at 15.9 millisieverts. A total of 531 workers died of leukemia, according to the study.

According to the International Commission on Radiological Protection, the risk of cancer would increase with exposure to radiation doses above 100 millisieverts. Researchers are divided on the health impact of radiation exposure below 100 millisieverts.

As well as the healthy ministry, the research was also funded by the U.S. Department of Energy among other groups.

  • Sam Gilman

    I suppose one can blame Kyodo rather that the Japan Times, but this article seems to garble the risks. It says “The study showed the relative risk of developing leukemia would increase by around three per thousand with a 1 millisievert radiation dose and would remain even with exposure to low doses of up to 100 millisieverts.” but this appears to be confused about what relative risk means and the baseline doses the study refers to.

    Here’s the study:

    The study found a relative risk of 2.95 of contracting leukaemia per Gray – ie if we treat exposure dose and absorbed dose to be the same, per Sievert. That means for every Sievert of exposure, there is just under a threefold increase in the chances of getting leukaemia.

    As an example, 15 British men in 1000 will develop some form of leukaemia. If 1000 men were exposed to a long term dose of 1 Sievert, there would be 15×2.96 = 44 cases of leukaemia: 29 extra cases. The thing about the study is that, through force of great numbers, they found a linear relationship between increased relative risk and increased radiation going down to doses much below one Sievert. As far as I understand, that means, assuming a linear relationship between exposure and excess risk that the article suggests, that for a prolonged exposure of 1 millisievert, there would be, over a lifetime, 29/1000= roughly 0.03 extra cases per 1000 people over a lifetime. This is a relative risk of 1.002 per mSv in a condition that affects, for example 1.5% of British men in their lifetime and 1% of British women.

    The study is important because it gives weight to the idea that radiation much lower 100 mSv exposure can lead to increases in cancer, even though they may be small.

    However, in the same issue of the Lancet Haemotology, there is a commentary on the paper warning that it may overestimate risks, based on the higher possibility of statistical false positives that the methodology can result in, and goes over problems with a deficit of homogeneity (sameness) of the data collection methods in different countries That’s here:


    So we need to proceed with some caution.

    (Any corrections to errors in my calculations from someone more educated in health stats are welcome. In particular, I think using lifetime risks for males would overestimate the number of extra cases.)

  • Mohan Doss

    Actually the article in Lancet is deeply flawed, and made a major mistake even a graduate student would not make, because they failed to take into consideration the large increase in per capita medical radiation dose during the long period of the study (1944-2005). The average annual medical radiation dose in 1960s was about 0.3 mSv per person, and it increased to 3 mSv per person in 2006 in USA. The medical radiation dose of the nuclear workers would have increased also during this time, even if not to the same extent. The occupational radiation dose, on the other hand used to be higher in the 1960s/1970s but was much lower in 1990s and 2000s. Whereas ignoring the medical radiation dose would not cause a significant error in 1960s, it would cause a major error in the 2000s, and so the dose-response curve they calculated would be wrong. I will demonstrate this with a simplified example.
    To simplify matters, for the sake of argument, let us say occupational dose was 1.7 mSv/y in 1960s, and 0.5 mSv/y in 2000s (averaging to 1.1 mSv/y). Let us assume annual per capita medical radiation dose of workers was 1/3 of USA population’s medical dose, i.e., 0.1 mSv/y in 1960s and 1 mSv/y in 2000s. The total annual dose would be 1.7+0.1 = 1.8 mSv/y in 1960s, and 0.5 + 1 = 1.5 mSv/y in 2000s. The shape of dose-response curve was calculated using 1.7 mSv/y and 0.5 mSv/y instead of the correct 1.8 mSv/y and 1.5 mSv/y for doses in 1960s and 2000s. The calculated dose-response shape would be obviously wrong. If a graduate student was given this information, and calculated
    the dose-response shape using the occupational dose only, he would get a failing grade. I am amused that such a simple mistake in calculation was made by the authors, and the journal editors and peer reviewers did not catch the mistake. I have asked the authors and the journal to withdraw the publication. I hope they will withdraw the publication.
    Publications which claim increased risk of cancer from low-dose radiation, whose faults get identified and conclusions get negated at a later date, do tremendous harm by perpetuating the myth of cancers caused by low-dose radiation, since the corrections do not get any publicity in comparison to the huge publicity to the original faulty publication. I hope Japan Times would publish a correction to this story in a prominent manner so they don’t contribute to radiophobia.

    • Chris Murray

      I am sure the authors of the report will respond to the specific criticisms above better than I can. Mohan Doss believes in hormesis, the idea, dismissed by many as nutty professor stuff, that radiation is actually good for you. In fairness, strictly speaking, Mohan may be right, but the evidence for hormesis has been found wanting even by BEIR, the ICRP and UNSCEAR, and Mohan is very much in a minority position scientifically.

      Mohan also believes that “CT scans are safe, and the present concerns regarding radiation dose from CT scans are not justified by any evidence.” However,
      even UNSCEAR, hailed by many pro-nukes as the “peak body”, acknowledges “Risk estimates vary
      with age, with younger people generally being more sensitive; studies of in utero radiation exposures show that the foetus is particularly sensitive, with elevated risk being detected at doses of 10 mSv and above.”

      10 mSv is roughly the dose from a whole body CT scan.

      • Mohan Doss

        Thanks for your comments, Chris. I was also skeptical about radiation hormesis, and used to believe in the LNT model in my younger days, as it seemed very plausible. Since then, I have been examining the evidence on both sides by reading publications. What I have found is that those articles claiming cancer risk from low-dose radiation all have major flaws, like the current Lancet one. They get lots of publicity in popular media, misleading the lay public as well as scientists and professionals. On the other hand, there are many publications that have shown evidence for reduced cancers following low-dose radiation. I have not found deep flaws in them. However, popular media ignore such publications, again deceiving the public. The idea of hormesis was indeed ridiculed before, but over the years, more evidence has accumulated supporting the idea of hormesis, and it is being given serious consideration. For example, NRC is considering three petitions (one of them being by me on behalf of “Scientists for Accurate Radiation Information”) that have asked NRC to stop using LNT model but start using radiation hormesis model for radiation safety regulations http://www.regulations.gov/#!docketDetail;D=NRC-2015-0057.

        Science is not a poll, and the final arbiter is not committees like ICRP but evidence. The evidence for radiation hormesis have not dissipated, unlike the evidence for the LNT model, including the most important evidence, the atomic bomb survivor data. This is not only my statement, but also the implicit statement by a prominent LNT model supporter, who did not cite these data to claim increased risk of cancers from low-dose radiation in the recent debate in Medical Physics journal http://scitation.aip.org/content/aapm/journal/medphys/41/7/10.1118/1.4881095 .

        The organizations you mention started backing the wrong side (LNT model) a long time ago, but have failed to change their views in spite of accumulating evidence for radiation hormesis and in spite of the evidences for the LNT model dissipating because of the major flaws in them. You have to ask them why they have followed this unscientific approach.

        UNSCEAR’s opinions you mentioned are completely wrong, because they extrapolated observations at high doses to low doses or used faulty studies. A reasonable person would not estimate the effect of taking one sleeping pill by studying what happens when people took 150, 100, 50 pills and extrapolating down to one pill. However, UNSCEAR and similar organizations have done similar extrapolation with regard to radiation. Their LNT model lacks credibility.

        If you are not familiar with my arguments above, you can view them in detail in a presentation I made in Tokyo, Japan a few months ago, available at: https://www.researchgate.net/publication/273984292_Coping_with_Low-Dose_Radiation_in_Fukushima

      • Chris Murray

        Thanks you, Mohan, for your polite reply. It makes a nice change from some other anti-LNTdebators. Obviously, I disagree with you on a number of points, as do UNSCEAR, the ICRP, BEIR etc., often claimed as the peak bodies by the pro-nuclear lobby, and as do the supporters of supralinearity. As a layman, there are limits to my ability to probe sources fully, and I wish this issue were resolved one way or the other, once and for all. Instead, this layman hears scientists – often apparently highly qualified scientists, from all sides of this debate – diametrically opposed to each other yet all claiming the scientific high ground. and accusing the other sides of sloppy work, lacking credibility, bias, conspiracy etc. This had gone on unabated for decades before I began studying the issue in the mid-1980’s. It does not inspire confidence in the scientific process and even undermines claims for the objectivity of science itself.

        But perhaps we can agree to disagree until I can at least read your work in more detail. Best wishes.

      • Sam Gilman

        I would disagree a little with your definition of “science”. I hope this doesn’t seem too pedantic.

        For those operating within a scientific field, then yes, evidence (that comes out from good methodology) is the key. Good methods is good science.

        However, when it comes to policy, we cannot rely directly on evidence. In general, people are not qualified to judge evidence in a specific field. Instead, we have to rely on properly operating scientific institutions (journals etc.) that can tell us about what consensus there is. It is entirely rational in a democratic society to rely on the ICRP and call it “scientific opinion” – unless you can point to political flaws in the process by which it comes to its conclusions. (I think there are issues with BEIR, for example where you might argue that.)

        So while I appreciate your belief in hormesis, your task is not to persuade me, but to persuade the generality of the scientific community. What’s important for the rest of us is learning about what genuine scientific consensus and dispute might look like.

  • Chris Murray

    ” The study showed the relative risk of developing leukemia would increase by around three per thousand with a 1 millisievert radiation dose and would remain even with exposure to low doses of up to 100 millisieverts.”? This is garbled indeed, as is any implication that the ICRP backs a 100 mSv threshold. It does not, and, for radiological protection, advocates, despite the many uncertainties (which can involve overestimation or underestimation of risk) the use of the Linear Non-threshold model.

    The ICRP and UNSCEAR caution against using collective dose because of uncertainties, but it is difficult to see why, unless it is to avoid embarrassing the nuclear industry with large body counts. It is difficult to understand why the evidence is sufficient for the ICRP to warn doctors “The higher dose diagnostic medical procedures (such a CT scan of the abdomen or pelvis) yield an effective dose of about 10 mSv. If there were a large population in which every person had 1 such scan, the theoretical lifetime risk of radiation induced fatal cancer would be about 1 in 2,000 (0.05%).” (this warning is in spite of uncertainties), but the same evidence is not sufficient, allegedly because of uncertainties, to assume that if one million people were exposed to that one in 2,000 risk that the theoretical death toll would be 500. This seems a political position, not a scientific one. Logic, reason and science are being seriously undermined.

    Using the (lower) older ICRP cancer (not leukaemia) risk estimates of one in 2,000 per 10 mSv, rather than the garbled ones in the article above, gives a risk of one in 20,000 per mSv. What would be the effect of 1 mSv per annum on, not a thousand people, but on one million people? Would not 50 people per million possibly get a fatal cancer? What is the permissible dose (above background) in Japan? 1 mSv? So if everyone of Japan’s 127 million people got the “permitted”/”allowed”/”safety” level of 1 mSv per annum you could expect an eventual cancer death toll of 6,350 per annum. How is this a “safety level”? Who “permitted” this? Why do we “allow” this? Do we really “allow” this? Decades ago, Dr. John Gofman called the “permitted” levels a “permit to commit random murder upon members of the population”, and he was right.