Medical Radiation Exposure to the U.S. Population: The Turning Tide

See also the article by Mettler et al in this issue.

Andrew Einstein is a cardiologist at Columbia University Irving Medical Center and New York-Presbyterian Hospital, where he serves as tenured associate professor of medicine (in radiology), director of cardiac CT research, and co-director of cardiac CT and MRI. His clinical work focuses on cardiac imaging and clinical cardiology, and his research, which has been funded by numerous NIH grants and the IAEA, focuses on medical imaging and patient safety.

The spectacular growth in diagnostic and therapeutic applications of ionizing radiation in medicine during the past several decades has been associated with innumerable benefits. At the same time, however, the associated increase in exposure of patients to radiation from these procedures has generated considerable concern due to potential attributable risks of cancers (1).

The work of the congressionally chartered National Council on Radiation Protection and Measurements (NCRP) in characterizing radiation burden to the U.S. population has been comprehensive and widely accepted by policymakers, clinicians, and other stakeholders. NCRP Report 93 (2) characterized U.S. radiation burden in 1980–1982, whereas its successor, NCRP Report 160 (3), characterized this burden in 2006, finding a nearly sixfold increase in average per capita radiation effective dose from medical sources excluding radiation therapy. This dose increased from 0.53 mSv to approximately 3.0 mSv, whereby medical radiation population burden matched that from all other sources combined (4). Most of this increase was attributed to the markedly increased use of CT scans and nuclear medicine procedures. Given secular trends in the use of diagnostic imaging in the U.S. population, and in dosimetry of individual radiologic and nuclear medicine procedures, the NCRP convened in 2017 a subcommittee to evaluate changes in medical radiation exposure of patients. NCRP Report 184 (5), reflecting this work, was published in November 2019. The article by Mettler et al (6) in this issue of Radiology distills and summarizes this report.

Mettler et al paint a picture of a tide that is in many respects turning. The increasing number of CT examinations performed has flattened, with only minor fluctuations since 2010, and their per capita effective dose is unchanged from 2006 (1.46 mSv) to 2016 (1.45 mSv). With use of newer dosimetric methods, with tissue-weighting factors used in the calculation of effective dose updated in 2007 on the basis of more recent epidemiologic data, this dose is slightly lower, at 1.37 mSv. Similarly, the number of radiographic procedures has stabilized at around 280 million per year and the average per capita dose has decreased by 27%, from 0.30 to 0.22 mSv. Procedure volumes and per capita doses from interventional fluoroscopic procedures, cardiac and noncardiac alike, have all decreased. The most striking drop is noted in nuclear medicine procedures, which decreased from 17 million to 13.5 million per year, with a 44% drop in per capita effective dose from 0.73 to 0.41 mSv. With use of the updated definition of effective dose, the latter falls to 0.32 mSv. Overall, notwithstanding U.S. population growth of 24 million, the collective medical radiation dose to the population has decreased by nearly 20% and per capita medical dose is now 2.3 mSv (2.2 mSv using updated methodology), down from a recalculation of 2.9 mSv.

Methodologically, the study by Mettler et al is strong. They marshal an impressive array of data sources, including IMV Medical Information Division survey data, Medicare Part B claims data, Department of Veterans Affairs procedure counts obtained by using the Freedom of Information Act, Food and Drug Administration reports, professional society registries, and the published literature. These constitute virtually all available U.S. population-wide data sources for both individual procedure doses and procedure volumes. Then the authors rationally synthesize these data to provide estimates of procedure volumes and effective doses of radiation. For example, although some of the most comprehensive data on procedure volumes derive from the IMV surveys, which are generally excellent, there are areas where IMV data appear inaccurate and inconsistent with those from other sources. Here the authors have appropriately imputed volumes from Medicare data.

What accounts for the changes observed by Mettler et al? Why has the tide begun to turn? Undoubtedly, there are numerous factors that may have impacted procedure volumes and radiation doses. Three central themes are awareness, education, and economics. On the test performance side, factors include development and dissemination of appropriateness criteria in referral, campaigns aimed at avoiding unnecessary testing, such as the ABIM Foundation’s Choosing Wisely (7), economic barriers to unfettered testing (eg, insurance preauthorization and radiology benefits managers), and changes in practice cultures. On the radiation dose optimization front, these factors include education in a host of settings from professional society meetings to online, imaging protocol management efforts, campaigns of multiple organizations (eg, Image Wisely for adult patients and Image Gently for pediatric patients), and professional society guidelines addressing radiation dose. Other factors include research characterizing the increased radiation burden, development of and research on newer technologies such as iterative reconstruction in CT, and media coverage that, while at times alarmist, has raised public consciousness and increased demand for dose optimization.

The data, and hence the primary drivers of the declines observed, vary between modalities. For example, in CT, the number of procedures performed has stabilized, whereas the numbers of both cardiac and noncardiac/non-PET nuclear medicine procedures have decreased markedly, with PET procedures increasing.

Although the tide of medical radiation dose to the U.S. population seems, based on the data of Mettler et al, to have finally shown evidence of turning, numerous efforts remain to improve implementation of the fundamental radiologic protection principles of justification and optimization across the U.S. population. As an example of opportunities to improve test justification, Mettler et al report a total of 23.6 million brain or head and neck CT examinations performed in the United States in 2016, an increase from the 19.0 million such examinations performed in 2006 (3). But data suggest that most head CT scans obtained in emergency departments are unnecessary per Choosing Wisely criteria. For example, a Choosing Wisely guideline from the American College of Emergency Physicians suggests avoiding CT in patients with minor head injury who, based on validated decision rules, are at low risk. Nevertheless, nearly two-thirds of such patients underwent head CT in a series of more than 600 000 minor head injury cases (8). There are numerous examples of such opportunities to reduce inappropriate testing.

An illustration of opportunities to improve test optimization is provided by the Prospective Multicenter Registry on Radiation Dose Estimates of Cardiac CT Angiography in Daily Practice in 2017 (PROTECTION VI) study (9). PROTECTION VI is a 61-center, 32-country registry of radiation doses from cardiac CT angiography and included a disproportionate representation of expert centers. The PROTECTION VI investigators found 37-fold variation between sites in mean radiation dose. When using the same methodology as employed by Mettler et al, PROTECTION VI found an overall median effective dose of 5.1 mSv for cardiac CT angiography. The mean effective dose used by Mettler et al, based on U.S. national registry data from the American College of Radiology, was 50% higher at 7.6 mSv. (In fact, the mean effective dose is nearly twice this value when using updated methodology [10].) This considerably higher dose in typical U.S. clinical practice suggests that there are numerous opportunities remaining to implement best practices for radiation dose reduction in cardiac CT across the U.S. population. Such opportunities exist as well for other types of imaging.

The progress described by Mettler et al should thus be regarded as first steps in turning the tide. Nevertheless, sustainment of this trend is by no means assured, and in fact recently released 2019 IMV data suggest that the number of CT examinations performed in the United States may have begun to increase again, to an all-time high reported as 91 million (11). Redoubled efforts to implement the factors, enumerated above, that have begun to change this state of affairs are warranted, as are novel approaches, such as better alignment of reimbursement schemes with radiation-related best practices. The onus is on all imaging stakeholders to improve justification and optimization, so that when the NCRP next revisits the collective medical radiation burden to the U.S. population, continued progress will be assuredly seen.