07/25/2011 - 09:25

Should we be giving up on low-dose radiation research?

David J. Brenner

David J. Brenner

Brenner is the Higgins Professor of Radiation Biophysics, and the Director of the Center for Radiological Research, at Columbia...


In the United States, just one government-sponsored funding program focuses on the health effects of low doses of ionizing radiation. That is the Energy Department's Low-Dose Program, which supports biomedical radiation research at academic institutions throughout the United States.  In the most recent presidential budget for fiscal 2012, funding for this program is slated to decrease from $25 million to $14 million. Is this reasonable? Do scientists and policymakers already know enough about the health risks of low doses of ionizing radiation to make quality decisions?

To illustrate what isn't known about the health effects of low doses of radiation, let us ask how many fatal cancers were induced as a result of the Chernobyl accident in 1986. What is known (the numbers of people exposed and the average radiation doses) is given in the two middle columns in the table below. The final column gives estimated numbers of radiation-induced fatal cancers for each group, based on a standard radiation risk coefficient from the International Commission on Radiological Protection.

Population Number of people* Average radiation
dose (milliSievert)*
Estimate of the number of radiation-induced
fatal cancers**

Residents of more
contaminated areas




Cleanup workers




Residents of less
contaminated areas




Other residents of
Belarus, Ukraine, and
parts of Russia




Rest of Europe




Rest of world






* Main sources: United Nations Scientific Committee on the Effects of Atomic Radiation (2008), Journal of Radiological Protection, 2006
** Estimated using a standard radiation-induced fatal cancer risk factor of five percent per Sievert (International Commission on Radiological Protection, 2007).

The most commonly quoted estimate, from IAEA and WHO, of the total number of fatal cancers potentially attributable to the radiation exposure from Chernobyl is 4,000. It is clear from the table above how this number is obtained: It comes from considering only the people from the top two rows, i.e., just the more highly exposed groups -- the implicit assumption being that the radiation-induced cancer risk for the large number of people receiving lower doses is zero.

It is clear from this example that any estimate of the cancer mortality resulting from a large-scale radiological event depends entirely on whether or not one chooses to ignore the typically large populations that received only very low radiation doses. For Chernobyl, this is illustrated in the following table:

If we ignore populations exposed to a
mean radiation dose less than …

… then the estimated number of radiation-induced cancer fatalities due to Chernobyl is

600 milliSievert (mSv)


120 mSv


10 mSv


1 mSv


0.3 mSv


0.03 mSv


Is it reasonable to assume that there is some radiation dose below which cancer risks are actually zero? And if so, what is that dose? Researchers really do not know the answers to these questions. It follows that neither they, nor the policymaking community, can really understand or predict the long-term consequences of Chernobyl, or of Fukushima, or of any future accidental or terrorist-based large-scale radiological event.

That science has established so little about cancer risks caused by very low doses of radiation is, prima facie, quite surprising. After all, the health effects of ionizing radiation have been studied for a long time, and quite intensively since atomic bombs were used in World War II. But it is tough to understand cancer risks at very low radiation doses because epidemiological studies at such low levels of exposure are essentially impossible to conduct. The fundamental difficulty: About 25 percent of any population will die of cancer anyway. So, for example, an epidemiological study of people in Europe exposed to an extra radiation dose of 0.3 mSv (see the "Rest of Europe" row in the first table) would need to be designed to detect an increase in cancer mortality from (say) 25 percent to (say) 25.001 percent. This is an essentially impossible task because one would need to track many millions of people over many decades, as well as a corresponding "control" population of people who were not exposed to this tiny extra radiation dose.

But the inability to conduct an epidemiological study of low-dose radiation exposure doesn't say anything, one way or another, about whether the risks of tiny radiation doses are real. Absence of evidence is not evidence of absence.

So if low-dose radiation epidemiological studies are not feasible, what is to be done? The question becomes: How can researchers extrapolate radiation risks measured in epidemiological studies that can actually be done (i.e., of people who were exposed to higher doses), down to the much lower doses that are of societal interest but cannot be studied in the same way? The answer, inevitably, involves a better understanding of the basic science. What are the molecular processes whereby radiation damage to DNA results in the production of pre-malignant cells and, years later, in a frank cancer? Some of the mechanisms are likely to be similar to those for carcinogenesis in general, but because of the unique nature of radiation energy deposition, many of the mechanisms will be unique to radiation.

Which brings us back full circle to the planned budget cuts in the Energy Department's Low-Dose Program. If researchers were already in a position to quantify low-dose risks, even approximately, the cuts, though major, might not be unreasonable. But science simply has not established the significance of very small increases in radiation to very large populations. The radiation risk for any given individual exposed to a very low dose of radiation will no doubt be miniscule, but what is the significance of tiny increases in individual risks, when many millions or even billions of people are exposed? In most other fields, risks to whole populations are estimated by multiplying the average individual risks by the number of people exposed, no matter how small the risk. Is this the right methodology to estimate the population risks when huge populations are exposed to very small radiation risks? There are some clues (see, for example, Health Physics 2009), but we really do not know, and basic science research is the only avenue forward here.

Drastically cutting low-dose radiation health research in the United States will significantly impair the country's ability to perform risk-benefit analyses regarding the future of nuclear power, or to assess the impact of a radiological terrorist event, or even to find the optimal benefit-risk balance in medical imaging. The dollar amounts for this type of research are quite small, but the bang for the buck is large in terms of our ability to make optimal policy decisions in such a wide range of energy and medically related areas.