Guest Editorial: Primum non nocere: A call for a re-evaluation of radiation doses used in CT

By Eliot Siegel, MD
pdf path

Dr. Siegel is a Professor of Radiology and a Vice Chairman, Department of Diagnostic Radiology, University of Maryland Medical Center, and Chief of Imaging,VA Maryland Healthcare System, Baltimore, MD.He is also a member of the editorial board of this journal.

"As to diseases, make a habit of two things- to help, or at least, to do no harm."
Hippocrates, Epidemics, Bk. I, Sect. XI

Every medical student is taught the phrase, primum non nocere , which roughly translates: "First, do no harm." This oft-cited and somewhat wry admonition from Hippocrates reminds us that even the best intentions to help our patients may result in unwanted consequences. Radiology residents are taught that they have a responsibility to "understand and employ examination methods for producing diagnostic image quality with minimal radiation dose"-the "as low as reasonably achievable" (ALARA) approach. 1 We teach them that it is their responsibility to protect patients from the potentially harmful effects of radiation--particularly the most vulnerable groups, including pediatric and pregnant patients.

The combination of increased utilization of computed tomography (CT) and the widespread adoption of multidetector CT scanners has resulted in substantial increases in patient radiation doses. Wall and Hart 2 reported a 30% reduction in doses of radiation from common radiologic procedures over the past decade but an increase in radiation doses of approximately 35% for CT of the abdomen and pelvis during the same period. Although CT studies account for only 13% of imaging procedures in the United States, they generate >70% of the total medical radiation burden.

The debate about the harmful effects of radiation has continued without abatement since before 1900, when the first damaging effects from repeated occupational exposure were noted in pioneer roentgenographers. Today, the debate centers on the effects of so-called "low-level" radiation exposures, in the range of 0.1 to 0.2 Sv (10 to 20 rem). The long-dominant linear no-threshold (LNT) model was reinforced by the International Commission on Radiological Protection in 2005 in its "Biological Effects of Ionizing Radiation VII" report. 3

Brenner and Elliston 4 estimated that the radiation dose from full-body CT scans results in a mean effective radiation dose of 12 mSv. According to their estimates, "a 45-year-old adult who plans to undergo 30 annual full-body CT examinations would potentially accrue an estimated lifetime cancer mortality risk of 1.9% (almost 1 in 50). …Correspondingly, a 60-year-old who plans to undergo 15 annual full-body CT examinations would potentially accrue an estimated lifetime cancer mortality risk of 1 in 220." They noted that, by comparison, "the lifetime odds that an individual born in the United States in 1999 will die in a traffic accident are estimated to be 1 in 77." 4 In another article, Brenner and others 5 suggested that the lifetime cancer mortality risk attributable to the radiation exposure from a single abdominal CT examination in a 1-year-old child is approximately 1 in 550 and approximately 1 in 1500 for a head CT examination.

A number of credible experts have strongly objected to the LNT model, however. For example, the 6000-member Health Physics Society has issued a white paper stating that "Below 10 rads, the risks of health effects are either too small to be observed or are non-existent." 6 Bernard Cohen from the Department of Physics, University of Pittsburgh, states that the LNT model "fails badly in the low-dose [<10-20 rem] region because it grossly overestimates the risk from low level radiation." 7 Others, such as Kenneth Mossman, have suggested that some cancers, such as breast and thyroid, may have a linear threshold at low doses, while other cancers, such as leukemia, may have a threshold of radiation exposure before they are induced. 8

Despite a strong curiosity and research interest in low-dose CT, I do not feel at all qualified to weigh in on either side of the debate over the LNT model. However, I believe strongly that the ALARA principle should be followed whenever possible and that the vast majority of radiologists and imaging departments have not done all they can to minimize radiation dose in CT. Radiologists cite their subspecialty training in radiation biology and physics when questions arise about whether other subspecialists (most often cardiologists and vascular surgeons) should be operating CT scanners, but I'm afraid that we do not apply this expertise in the regulation of CT doses in our daily practice. We radiologists tend to pay much more attention to the equally complex issue of contrast-induced nephropathy, which, like radiation, has been demonstrated to be harmful at higher doses but with limited research on the impact of lower doses.

The radiation dose used in CT varies widely from practice to practice. We have found radiation doses for thoracic CT, for example, that vary by a factor of more than 3-fold, from 80 to 300 mAs. Most radiologists are not aware of the technical settings used for their own CT studies. These are usually set by vendor applications specialists or, in some cases, simply copied over from previous generations of scanners. A phenomenon first noted in computed radiography as "dose creep" also occurs in CT, with higher doses yielding equal or better-looking images, and doses that are too low negatively affecting image quality. Technologists who are able to change settings in such circumstances have a tendency to increase, rather than decrease, doses. Most CT scanners do not measure the total amount of radiation delivered to the patient and do not store this information in a database for analysis and tracking. Vendors differ substantially in the ways in which they measure radiation doses.

To their credit, however, the major CT vendors have implemented various forms of dynamic dose reduction that adjust the delivered dose according to the thickness of a body part during scanning or by using a preliminary overview "scout" image. Unfortunately, no baseline doses have been established for various studies, nor have these been optimized for patients within a certain weight range or according to indication. For example, a 6-month follow-up study for a lung nodule might use a considerably lower dose than a study performed for a pulmonary embolism or to detect mediastinal lymphadenopathy. My colleagues and I have a database of thoracic CT studies acquired using 180 and 11 mAs, a reduction of 94%. Despite the wide variation in dose, the difference in the ability to visualize lung nodules and the thorax in general at "lung" settings is minimal (images available at the Applied Radiology Web site [www.] for a side-by-side demonstration). This seems to support the potential for modifying dose according to indication, in addition to age, weight, and anatomic region.

We have been working with and performing research using a computational approach called the visual discrimination model, which can predict the likelihood that a radiologist will be able to discern a difference between 2 images. It is possible to use this technique, along with the introduction of image noise, to determine for a given study whether dose reduction would result in a perceptibly different image. Incorporating this approach into practice would be the equivalent of using a sliding control that would allow the radiologist to decrease the dose to the point at which the lowest possible dose yielded a clinically acceptable image. More sophisticated and intelligent techniques should be explored to help create a scientifically based rationale for radiation settings in CT. In addition to computational-driven approaches, it is also possible to achieve clinical dose reduction by simply noting from experience what dose level is acceptable to answer a specific clinical question, a process by which the radiologist may choose to accept an image that is less "pretty" but adequate for a clinical determination.

The American College of Radiology has indicated that a review of radiation doses used in CT may soon be included in its departmental quality assessment program, a move that might lead to recommendations for improved tracking of patient doses. I believe that we radiologists need to work much harder to increase our awareness of the potential to decrease these doses and to help educate and work with our referring colleagues to track and reduce total cumulative radiation burden. Hippocrates would have expected no less from us.

Back To Top

Guest Editorial: Primum non nocere: A call for a re-evaluation of radiation doses used in CT.  Appl Radiol. 

April 20, 2006

Copyright © Anderson Publishing 2019