Just how dangerous is radiation? Scientists have been debating this question for decades yet, despite extensive studies, there is still controversy. The working assumption, which is currently accepted as the basis for regulation and legislation, is that radiation raises the risk of cancer at a rate that is directly proportional to dose at all dose levels. A consequence of this “linear no threshold” (LNT) model is that it assumes that there is no safe level of radiation dose. The other possibility is that below a certain threshold level, radiation is essentially harmless: any damage done by ionization and the consequent radiochemical and radiobiological effects is effectively and quickly repaired by the human body, with neither lasting harm nor elevated risk of cancer.

Conclusive evidence in favour of one model or the other would be of enormous interest. Scientists who design and operate nuclear power plants and radioactive-waste repositories would benefit from greater clarity. Medical physicists, who routinely weigh up the benefits of diagnostic tests and radiation treatments against the risks to patient health, would be on firmer ground – as would the politicians who approve the necessary regulations. But any changes in policy or clinical practice must be driven by data. Bold claims that radiation-protection regulations are a factor of 1000 too cautious may be appealing, but they should be dismissed out of hand unless they are supported by both a reasoned argument and unequivocal data.

In Radiation and Reason: The Impact of Science on a Culture of Fear, Wade Allison, a physicist at the University of Oxford, sets out a reasoned argument in favour of the threshold model, and against the LNT assumption outlined above. To support this argument, Allison provides examples from engineering and biology where there are indeed thresholds for irreparable effects. For example, an individual who suffers a bruise or laceration will recover completely from such a minor injury, but beyond a certain threshold, laceration is irreparable and possibly life-threatening. Why, Allison asks, should radiation carcinogenesis be different? After all, we know that damaged DNA can be repaired, and that in some cases irreparably damaged cells can be eliminated by apoptosis, or programmed cell death. Surely this is evidence that the LNT model is flawed?

In the course of researching this self-published book, Allison clearly became convinced that the radiobiological processes underlying carcinogenesis are well enough understood that the LNT assumption can be dismissed. However, the reader should be aware that the data he uses to support this argument have also been reported and discussed extensively by researchers in the field, and their conclusions were rather different. Notably, the L H Gray Conference in June 2008, which brought together international experts in radiobiology, epidemiology and risk assessment, concluded that “at the present time, although the possibility of a low-dose threshold cannot be ruled out, current thinking on radiation protection suggests it is likely that low doses of radiation will carry some risk”.

The threshold hypothesis set out in Radiation and Reason is based on observations from human populations. In particular, data on survivors of the Hiroshima and Nagasaki atomic bombings and those exposed to radiation in the aftermath of the Chernobyl incident show that the number of “excess” cancers is lower than would be expected from the LNT model. However, the evidence from radiotherapy patients, who Allison claims are safely exposed to doses many orders of magnitude higher than radiological-protection dose constraints, is not completely appropriate in this context, nor is it complete.

In radiotherapy, the volume of tissue irradiated to very high doses is typically less than 1% of the whole body. A much higher volume of tissue receives dose levels that can produce functional side effects (such as damage to the integrity of the skin or blood vessels, or reduced saliva production) rather than carcinogenesis. Allison correctly cites the repair processes for these side effects as being the mechanism whereby therapeutically effective doses can be delivered to tumours without doing irreparable damage to normal tissue.

Repairing functional damage is, however, very different from the repair at the molecular level that is necessary to reverse the genetic damage that leads to cancer. Moreover, there is also an unavoidable whole-body dose associated with radiation therapies – typically 4 mSv per day from leakage and scattered radiation. (In comparison, the average annual dose from background radiation is approximately 2.4 mSv.) This scattered radiation is known to induce “second cancers”, which can occur far away from the regions of high dose. For example, a large study of men with prostate cancer demonstrated an increased subsequent risk of lung cancer for those treated using radiotherapy compared with a matched cohort treated by surgery. Although the appearance of such cancers does not preclude the existence of a threshold, it does undermine the grounds for rejecting the LNT model on the basis that radiotherapy is risk-free.

The bottom line is that the scientific debate on the existence of a threshold cannot be resolved by population studies alone, simply because the data are so sparse (thankfully, since they stem largely from nuclear wars and accidents). As the old saying goes, “absence of evidence is not evidence of absence”. Resolution of the threshold question, if it is possible, will be indirect and will depend on quantitative basic radio_biology rather than epidemiology.

While the book draws on data from many applications of radiation, it is the nuclear-power industry that would, its author believes, benefit most from relaxed regulation. Yet Allison acknowledges that most of the vast expense involved in designing safe reactors and appropriate storage systems is directed at avoiding or reducing the risk of major incidents; as such, these costs do not depend on the existence of a threshold dose. The necessary storage time for fission products and other medium-half-life radioactive waste would be influenced by the level of a threshold, and by public and scientific acceptance of it. However, the costs of constructing a waste-storage facility will not be very sensitive to the threshold dose, nor to the timescale required. A facility built to last 500 years is unlikely to cost three times as much as one designed to last 150 years – another example of nonlinearity.

Radiation and Reason also poses questions of a sociological and political nature. Why, Allison asks, is radiation perceived as being particularly harmful? Can that perception be changed to ensure that nuclear power can be made more affordable and available, leading to worldwide societal benefits? In exploring these questions, the author suggests that overzealous regulation has persuaded the public to believe that radiation is more dangerous than it actually is. He argues that this has produced an ever-tightening spiral of constraints, which others have described as the “ratchet of radiation protection”. Perhaps, he says, the time has come to release it. In this respect, Allison may have a point: a relaxation of constraints may indeed be in order, and it should certainly be on the agenda.

However, such sensible thinking is undermined by Allison’s statements regarding would-be nuclear terrorists. In particular, his suggestion that terrorists will be deterred if regulations on the storage and use of radioactive materials are relaxed, in response to evidence of reduced risk, strikes me as fanciful. For one thing, it wrongly assumes that terrorism is based on rational behaviour. It also ignores the fact that if radiation were known to carry a lower risk than current thinking suggests, then terrorists would simply need to steal a bigger flask of radioactive material to cause the same effect.

So is this a book about science, the public understanding of science, or politics? Perhaps all three, but the author’s emotive language in stating that “the public need to know the truth” implies that in the past they have been told lies. This puts the matter squarely in the political domain. For scientists, the threshold debate is not about truth or lies; rather, it is about how to deal with facts in a world of uncertainty, where decisions have to be made on the basis of the balance of probabilities. Allison is acerbic in his criticism of international bodies such as the International Commission on Radiological Protection, but their conservatism is not an abrogation of scientific responsibility. Rather, it is a recognition that scientists have a responsibility to make judgments as well as reporting their results. Until the radiobiological and radiological-protection communities reach a consensus, it would be unreasonable to expect legislators to relax regulation and undertake an experiment that will take generations to mature.