We use radiation for everything from generating electricity, to gaining a look at the inside of our bodies without resorting to a scalpel, to rescuing patients from the ravages of cancer, to sterilizing medical equipment and food, to numerous industrial applications. Modern society benefits greatly from technologies that harness the beneficial uses of radiation. Nuclear power alone has saved 1.8 million lives worldwide by avoiding air pollution, and medical imaging has been recognized as one of the top 15 medical advances of the past 170 years.

The use of radiation is regulated by a legal framework that establishes radiation dose limits for workers and the public. The framework calls for ever‐​lower exposures that are “as low as reasonably achievable” (ALARA), taking social and economic factors into account.

Over the decades since the discovery of radiation, this legal framework has unquestionably protected workers and others who once received radiation doses that today we recognize as alarmingly high. However, today further reductions that are compelled by ALARA often provide only trivial benefits or simply demonstrate novel technologies. There is no serious debate that enormous radiation exposures—for example, what a person might receive within several hundred meters of ground zero of a nuclear detonation—measurably increase the risk of cancer. But with very rare exceptions, people today are not receiving doses anywhere near that high. Adverse health effects from the extremely low radiation doses typical of modern occupational and environmental exposure scenarios are so rare that they cannot be directly observed, if they exist at all.

Ultra‐​low exposure limits do not just impose a cost in the form of more and more radiation‐​reduction measures, but also in the suppression of beneficial uses of radiation. Another cost comes from the promotion of unjustified levels of fear of radiation among the public. Moreover, in practice, the “reasonability” of dose‐​reduction efforts explicitly embedded in the very definition of ALARA is frequently overlooked, and doses are instead driven as low as technologically possible, ignoring the costs and consequences of doing so. In the United States, this has led to excessive regulatory burden, which is a major driver of the premature closing of an increasing number of nuclear power plants that provide more than half of the country’s emission‐​free energy.

Linear dose / Underlying ALARA is the assumption that increases in cancer risk observed after enormous radiation doses can be extrapolated down to doses that are less than a hundredth or a thousandth of the size—like those typical of modern occupational and environmental doses. This idea, known as the linear no‐​threshold (LNT) model, predicts that any radiation dose, no matter how small, increases cancer risk in linear proportion to the reference dose all the way down to zero exposure. It discounts the possibility of there being a safe exposure level. The LNT assumption provides the justification for driving already trivial doses even lower.

How did this assumption arise? A seminal event was the National Academy of Sciences’ formal recommendation in 1956 that the threshold model of radiation effects be rejected in favor of the LNT model. Careful examination of historical records has revealed that the studies upon which the 1956 adoption of the LNT model was based—the work of Nobel‐​winning geneticist Hermann Muller—did not in fact support the LNT model. Muller examined gene mutations in fruit flies exposed to radiation doses so high (that is, about 100 million times greater than background) that they would be lethal to humans, and concluded that mutations increased linearly with radiation dose, even at doses close to zero. But later research by his team showed that there was a dose‐​rate threshold. Modern fruit fly experiments show that not only do low radiation doses fail to increase the number of mutations, they actually reduce mutations below the normal background level of mutations in nonexposed flies. (See “The Troubled History of Cancer Risk Assessment,” Spring 2019.)

This phenomenon is known as “hormesis”: low radiation doses stimulate natural biological defense mechanisms to provide a mild protective effect, while high radiation doses overwhelm these defenses and cause negative effects. Thousands of experiments in cells and whole organisms ranging from bacteria to humans, using diverse measures of biological effects, confirm that hormesis is a general phenomenon that has been conserved over the course of the evolution of life on Earth.

Rather than systematically evaluating and employing this vast body of scientific literature, regulatory agencies and advisory groups steadfastly ignore it even as the evidence continues to accumulate. Biological studies using modern techniques to determine the effects of radiation in genes and chromosomes provide compelling evidence consistent with hormesis, not the LNT model. Yet, modern radiation regulations are based almost exclusively on epidemiological studies that have limited statistical power and significant uncertainty at low doses that prevent them from illuminating the effects of those doses.

Using circular logic, in the decades since the adoption of the LNT model in 1956, reviews conducted by advisory groups and regulators start with the assumption that the LNT model is accurate. They do this for no other reason than that is what they have assumed in the past. They will only abandon this assumption if there is incontrovertible epidemiological proof that the LNT is incorrect—a practically impossible threshold. The uncertainty in epidemiological data, combined with the exclusion of biological data supporting thresholds and hormesis, and most importantly the a priori assumption that the LNT is true, combine to provide the perfect environment for the continued survival of this discredited theory. This subtle and improper shift in the burden of proof has consequences in the form of irrational public fear of radiation and the public health consequences born of that fear.

Nuclear power / Major nuclear accidents are exceedingly rare but they do happen, as evidenced by Three Mile Island, Chernobyl, and Fukushima. The harm from those incidents was made far worse by bad decisions. That makes it imperative that we learn the lessons those tragic events have to offer.

One mistake made at both Chernobyl and Fukushima was based on the LNT model and offers a cautionary tale of the consequences of irrational fear. In both incidents, the public health responses caused more harm than good, and indeed more harm than the accidents themselves. In the aftermath of Chernobyl, tens of thousands of terrified expectant mothers across Europe succumbed to the fear of having babies deformed by radiation and chose to have elective abortions. This was in spite of the lack of evidence that radiation causes heritable genetic mutations in humans. Likewise, the highest plausible doses received by these women were less than a thousandth of the level observed to cause birth defects. As at Chernobyl, LNT‐​induced fear of radiation caused the mass evacuation of Fukushima’s residents and their multi‐​year exile from their homes. This caused over 2,000 avoidable deaths during the evacuation and widespread mental health effects among displaced Fukushima residents. Arguments have been raised that these effects were caused by misunderstanding and misapplication of the LNT model rather than the LNT model itself, but honest policy analysis has to account for the real‐​world consequences of regulations, whether intended or not.

Need for reform / Resistance to reforming radiation regulations away from reliance on the LNT model often include three dubious arguments:

  • The LNT model is easy to use.
  • Even if the LNT model overestimates risks, this protects public health.
  • There isn’t a better, more plausible model.

Not one of these arguments is compelling; one of them is debatable and two of them are false.

Even advocates of the LNT model admit that it overestimates risks of low doses of radiation delivered at low dose‐​rates. Some of them propose adjustments to LNT predictions to account for this inaccuracy. The appropriate magnitude of such an adjustment is itself the subject of debate and introduces complexity that mitigates the LNT model’s simplicity.

The assertion that overestimating the risks from low‐​dose and low‐​dose‐​rate radiation exposure protects public health by building in a safety margin has been shattered by the real‐​world experiences of Chernobyl and Fukushima. In both cases the message, based on the LNT model, that there is no safe dose of radiation has permeated the public consciousness and led to irrational responses that unequivocally damaged public health.

In the strictest sense, the efficacy of the LNT model does not depend on whether or not a “more plausible” model exists. The LNT model either accurately describes the observed data on low‐​dose radiation effects or it doesn’t. However, from a public health policy perspective, it is unlikely that regulators will finally abandon their misplaced reliance on the LNT model unless an alternative model gains favor.

Contrary to LNT advocates’ assertions, several regulatory improvements have been proposed. These include melding the LNT and hormesis models to account for the uncertainty in the effects of low radiation doses, establishing a low‐​dose “stopping point” for ALARA, and the related idea of establishing a “de minimis” dose below which there would be no regulation. Unfortunately, so far regulators have steadfastly resisted these improvements.

All of these ideas have their merits and deserve honest, unbiased consideration and debate. Each alternative should be considered from a policy perspective that includes objective benefit–cost analysis and an assessment of public health outcomes.

It is well past time for our radiation regulation philosophy to evolve. The first steps are to unshackle ourselves from the LNT model that traps us in the past, and to begin developing a radiation regulatory strategy that reflects modern knowledge of low‐​dose radiation effects.