The linear no-threshold theory: Readers weigh in

One of us is a radiologic educator, and the other a scientist. We do not support the linear-threshold (LT) radiation dose–response relationship that Jeffry Siegel and his coauthors do. The linear no-threshold (LNT) model has been extensively studied in numerous works that have established appropriate imaging with acceptably reduced patient radiation dose. Siegel and company correctly recognize that we do not know what the response is to medical radiation exposure below perhaps 100 mSv. However, many continue to subscribe to LNT, and we agree that it is the correct position.

To abandon LNT for LT would be another example of “normalization of deviance,” a term coined by Diane Vaughan following the 1986 Challenger disaster and first applied to medicine in anesthesiology ¹ in 2010. Normalization of deviance is the gradual shift in belief or behavior that strays from accepted safety standards because the belief or behavior has no adverse consequences—until it does. Andrew Woodward and Melissa Jackowski, in a presentation they gave at the Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, explored normalization of deviance as the reason for radiologic technologists taking shortcuts that violate the concept of ALARA (as low as reasonably achievable).

Although the true dose–response relationship may well be nonlinear at low doses, assuming a threshold would be irresponsible. In medical imaging, low doses of radiation are viewed as acceptable given the diagnostic benefits. But radiologists must always strive to minimize radiation exposure to their patients and to themselves.

Abandoning LNT, in medical imaging at least, will result in another example of normalization of deviance and in an unknown but large number of unnecessary deaths. Consider, for example, the 80 million CT imaging studies performed annually in the US. ² Estimates of lethality from radiation-induced cancer from such medical exposures approach 30 000 per year. ³

Of course, the difficulty with such predictions is that radiation-induced cancer has no tag of any kind to identify it as such. Furthermore, whatever the true rate of radiation-induced fatalities is, it is hidden by the 20% normal cancer lethality in our total population.

At least for medical imaging, we recommend continuing to use LNT while accepting that a patient radiation dose less than approximately 100 mSv is well worth the benefit of the imaging and should be accepted as safe.