Use of Medical Imaging Procedures With Ionizing Radiation in Children: A Population-Based Study

The rapid growth in use of medical diagnostic imaging procedures such as computed tomography (CT) has led to concern about low-dose ionizing radiation exposure in adults.^1-5 Despite widespread discussions about similar hazards for younger patients, contemporary data on the use of such imaging procedures in children are limited. Infants and children are at higher risk for future malignant neoplasms as compared with adults because their developing tissues are more sensitive to radiation and their longer expected life spans allow additional time for the emergence of detrimental effects.^6-10

Accordingly, we set out to examine patterns of the use of diagnostic imaging procedures with ionizing radiation among children. We used comprehensive inpatient and outpatient claims data sources from UnitedHealthcare, a large health care organization that administers benefits to millions of families across the United States. Because methods for accurately quantifying the effective dose of ionizing radiation exposure¹¹ (ie, its detrimental biological effect) in children are controversial, we focused primarily on describing the number and types of these procedures being used. As such, our goals were the following: (1) to determine overall population-based rates of the use of imaging procedures with ionizing radiation in individuals younger than 18 years; (2) to explore their use across age groups and sex; and (3) to identify the most frequently used procedures.

We conducted an investigator-initiated, retrospective cohort study using exhaustive inpatient and outpatient claims data from UnitedHealthcare. These data were collected between January 1, 2005, and December 31, 2007, from 5 large regional markets: Arizona; Dallas, Texas; Orlando, Florida; South Florida; and Wisconsin. These markets were specifically selected because of their size, the stability of their enrollment population, and the similarity of their insurance products as well as to provide a degree of geographic diversity. The study population included all individuals younger than 18 years at the beginning of the study period who were alive and continuously enrolled in one of the programs administered by UnitedHealthcare during the study period. To obtain a true denominator population, we included all individuals who were continuously enrolled regardless of whether they submitted a claim during the study period. After removing all personal identifiers, data were provided to the investigators for independent analysis and interpretation. The institutional review board of the University of Michigan approved this study protocol and waived the requirement for informed consent.

All claims from hospitals, outpatient facilities, and physician offices that were submitted during the study period were queried for Current Procedural Terminology codes that identified imaging procedures using radiation exposure (Radiology Schedule–Diagnostic Imaging and Nuclear Medicine codes 70010-76499 and 78000-79999; and Medicine Schedule–Cardiovascular and Noninvasive Vascular Diagnostic Studies codes 92950-93799 and 93875-94005).¹² Procedures were included regardless of whether they were performed for diagnostic or therapeutic indications (eg, interventional radiological procedures). However, claims related to the specific delivery of radiation for a therapeutic purpose and not for imaging (eg, total body irradiation prior to stem cell transplantation) were excluded. In cases where the Current Procedural Terminology code for a procedure changed during the study period, all versions of the code were included.

We obtained the following information from each claim: (1) age; (2) sex; (3) the market where the service was performed; and (4) the location of the service (hospital inpatient, hospital outpatient, and physician’s office). We categorized these procedures into mutually exclusive categories based on the technology used (plain radiography, CT, fluoroscopy and/or angiography, and nuclear medicine scans) and anatomical area of focus (chest [including cardiac imaging], abdomen, pelvis, extremity, head and neck [including brain imaging], multiple areas [including whole-body scans], and nonspecified). To be as conservative as possible and to avoid the potential of overestimating the number of procedures from duplicate claims, we limited individuals to 1 procedure per day for the same type of technology (eg, CT) performed on the same anatomical area (eg, chest).

We focused on describing the number and types of imaging procedures performed in the study population using simple descriptive statistics. Specifically, we calculated population-based rates of use where the numerators were the cumulative number of imaging procedures performed in an individual and the denominators were the total number of eligible children enrolled throughout the study period. For these analyses, children were categorized based on their age at the beginning of the study period (ages 0 to <2, 2 to <5, 5 to <10, 10 to <15, and 15 to <18 years) and sex. Imaging procedures were then categorized by the type of technology used. All statistical analyses were carried out with the use of SAS version 9.2 statistical software (SAS Institute, Inc, Cary, North Carolina).

We identified 355 088 children continuously enrolled in a program administered by UnitedHealthcare during the study period. The mean (SD) age was 9.0 (4.9) years, and 181 795 children (51.2%) were boys. The largest proportion of the study population was located in the Dallas market area (124 079 children [34.9%]), while the smallest proportion came from the Orlando market area (45 466 children [12.8%]). Overall, the percentage of subjects who underwent at least 1 diagnostic imaging procedure using ionizing radiation ranged from 39.4% in the Orlando market area to 43.2% in the Dallas market area.

During the 3-year study period, a total of 436 711 imaging procedures were performed in 150 930 children (42.5%) (Table 1), resulting in an annual rate of 410 procedures per 1000 children. During the study period, 89 618 children (25.2%) underwent 2 or more of these procedures, while 56 754 (16.0%) underwent 3 or more. The highest annual rates of use were generally in children aged 10 years and older. Use of these procedures also was generally higher among boys than girls (80 638 [44.4%] vs 70 292 [40.6%], respectively; P < .001). Although the overall proportion of patients was smaller, similar patterns of use across age and sex groups were observed in children who underwent at least 2 imaging procedures and those who underwent 3 or more.

Table 1. 

View LargeDownload

Characteristics of Patients Undergoing All Procedures Using Ionizing Radiation

Procedure use was stratified by the type of imaging procedure across age and sex groups. Rates varied substantially based on the types of procedures, with 141 480 children (39.8%) receiving at least 1 plain radiographic examination (Table 2), 28 107 (7.9%) receiving at least 1 CT scan (Table 3), 7492 (2.1%) receiving at least 1 fluoroscopic or angiographic procedure (Table 4), and 2607 (0.7%) receiving at least 1 nuclear medicine scan (Table 5). Plain radiography also was the most commonly repeated of these procedures: 79 209 children (22.3%) underwent 2 or more of these studies. Similarly, CT scans were frequently repeated, with 12 494 children (3.5%) undergoing 2 or more of these studies. While use of CT was more frequent overall in boys, use in the group aged 15 to 17 years was higher in girls than in boys (93/1000 person-years vs 85/1000 person-years, respectively). Finally, overall patterns of plain radiography and nuclear medicine scans followed those described earlier for all imaging procedures, with use most frequent in infants younger than 2 years and children aged 10 years and older. However, CT scans were used less frequently in young children. In contrast, the largest proportion of children undergoing fluoroscopic and/or angiographic studies were younger than 2 years—especially among infant girls.

Table 2. 

View LargeDownload

Characteristics of Patients Undergoing Plain Radiography

Table 3. 

View LargeDownload

Characteristics of Patients Undergoing Computed Tomography

Table 4. 

View LargeDownload

Characteristics of Patients Undergoing Fluoroscopy

Table 5. 

View LargeDownload

Characteristics of Patients Undergoing Nuclear Medicine Scans

Overall, plain radiography accounted for 84.7% of the studies that were performed; CT scans were the next most commonly used modality, accounting for 11.9% of imaging procedures, followed by fluoroscopic and/or angiographic studies (2.5%) and then nuclear medicine scans (0.9%). However, this pattern of use varied across different age categories. For example, CT scans were used much more frequently in older groups, increasing to 17.6% of all imaging studies performed in children aged 15 years to younger than 18 years.

Finally, Table 6 shows specific data on the 10 most frequently used studies for the 4 different categories of imaging procedures. Chest radiography was the most common procedure performed overall, at an annual rate of 68 procedures per 1000 children. This was followed by plain radiography of extremity areas, the spine, and the abdomen. By far, the highest annual rate of use of CT scans involved studies of the head, followed by the abdomen and pelvis. Computed tomographic scans of the abdomen and pelvis, in particular, increased dramatically with age, from 5 procedures performed per 1000 children annually in patients aged 0 to 1 year to 41 procedures per 1000 children in patients aged 15 years to younger than 18 years. The use of fluoroscopic and/or angiographic studies and nuclear medicine scans was infrequent overall.

Table 6. 

View LargeDownload

Most Commonly Performed Procedures by Imaging Modality

To our knowledge, we report the first large, population-based study examining the use of diagnostic imaging procedures with low-dose ionizing radiation specifically in a pediatric population. Among 355 088 children across 5 large health care markets in the United States, we found that use of these procedures during a 3-year study period was frequent, with at least 1 of these procedures being performed in 42.5% of children. Importantly, many children underwent more than 1 procedure. Based on these data, the average child in this study population will have received more than 7 procedures by the time he or she reaches age 18 years. While plain radiography—which is associated with much lower levels of radiation exposure—was responsible for most procedures, the use of other types of studies such as CT was not rare.

The National Research Council's Biological Effects of Ionizing Radiation VII Phase 2 report¹³ cautions that there is no lower threshold of exposure to radiation that has been identified as without risk and that repeated exposure increases risk in a linear fashion. Furthermore, studies suggest that the hazards of radiation may be greater in children than in adults.^6-8,14-16 For example, Brenner et al⁷ estimated that the risk of fatal malignant neoplasms from radiation exposure with abdominal CT was 8-fold higher in the first year of life as compared with the risk in a 50-year-old adult. This results primarily from children having greater sensitivity to the effects of radiation owing to growing tissues and having a longer life expectancy than adults, allowing more time for the latent effects of radiation to emerge.

The risks of radiation exposure in children are not restricted to the development of cancer. For example, repeated head CT that includes imaging of the lens of the eye may increase the risk of later cataract formation. Concern for the possibility of later developmental and other nonmalignant problems has also been raised. More than 3000 children in this study population—1.0% of the subjects—received 2 or more head CTs during the 3-year study period.

A key finding from our analysis was that most imaging procedures performed in this study population were plain radiography, which typically delivers a low dose of radiation in comparison with other techniques. Yet, while the use of CT scans, nuclear medicine scans, and fluoroscopic and/or angiographic procedures was much less common, it was not rare. In particular, CT scans were used in 7.9% of the study population and their use increased dramatically in older groups. As such, the overall contribution of these imaging procedures to the long-term risks of radiation should not be overlooked. This is particularly true since their use appears to be concentrated in a minority of individuals who may undergo repeated studies.

Of the imaging procedures we examined, CT scans may be the most important from the standpoint of radiation exposure. Nationally, the use of CT has rapidly grown over time owing to an increased availability of CT scanners and a lower threshold for ordering these studies in routine clinical practice.^6,17 Although there is evidence that CT volumes in pediatric hospitals have decreased since 2003 relative to the total number of cross-sectional imaging examinations, the absolute trend in use is less clear.¹⁸ If one were to extrapolate our findings to the pediatric population of the United States,¹⁹ 5.8 million children younger than 18 years would be expected to undergo at least 1 CT scan during a 3-year period. Furthermore, nearly 2.6 million would undergo 2 or more CT scans. The average child in this study population received 0.86 CT scans by the time they reached age 18 years. Importantly, these findings extend results recently reported by the congressionally chartered National Council on Radiation Protection and Measurements, which estimated that 8% to 10% of CT scans in the United States are performed in children but did not examine rates of use longitudinally over time within the same child or describe the specific procedures used.²⁰

These data reinforce the importance of judicious use of imaging procedures with ionizing radiation, particularly in children. Numerous regulatory groups and publications have highlighted the importance of the concept of ALARA—as low as reasonably achievable—in the application of radiation for imaging procedures.⁶ However, recent studies in the literature suggest that this practice is not being used broadly or cohesively in adult imaging.^3,21 We suspect that the same is true for pediatric imaging. For many types of imaging procedures, the development of age- and weight-dose protocols has been lacking, although this may be less of a problem in pediatric hospitals. In the case of CT scans, there is concern that these protocols are not yet widely applied. The Image Gently and Step Lightly programs²² are important educational campaigns that are enlisting physicians and parents to reduce radiation exposure in children.

Appropriate use of these procedures requires balancing the long-term risks inherent in radiation exposure with the necessity for making clinical decisions at the bedside. Developing better guidelines for these procedures may help guide clinicians struggling to determine the best role for these studies. For example, CT scans have revolutionized the management of head trauma in children. Having either too high or too low a threshold for ordering a CT scan in this clinical situation may be problematic. Several studies have recently examined the use of head CT in this context^23-29 and have derived decision rules for its use based on identifying clinically important traumatic brain injuries, suggesting that CT can be avoided in many children presenting to the emergency department with head trauma.

Several limitations to this study should be noted. First, we used administrative claims data. While this meant that we were able to exhaustively capture information on the use of imaging procedures in the study population, we do not know the clinical context under which these procedures were ordered. We therefore cannot comment on their appropriateness. However, the intent of our study was to examine contemporary use of these procedures—both appropriate and inappropriate. Second, we did not attempt to estimate effective doses of radiation from these imaging procedures. There is a paucity of available data on radiation dosimetry in the pediatric population. Extrapolating effective dose estimates for children from estimates available in the literature could be misleading.^11,22 Instead, we focused on describing the patterns of use across various types of imaging procedures, which we believe adds critical new information for policy makers and health care providers. Third, data in the study population were gathered from 5 specific market areas. While these areas represent large and diverse communities, there may be aspects of these populations that differ from the greater United States as a whole. This could potentially limit the generalizability of these findings to the broader United States. Similarly, these data include only children who were insured and do not represent use in uninsured settings. Fourth, our results likely underestimated the total number of studies performed in children by restricting individuals to 1 procedure per day of the same technology on the same anatomical region and excluding patients who died during the study period. We did this to be as conservative as possible in reporting our findings and to focus explicitly on survivors—the group in whom the long-term risks of radiation are most concerning. Finally, these results represent a snapshot in time. Imaging procedures are constantly evolving and their use has been in flux during recent years.

In conclusion, our study describes patterns of use of diagnostic and therapeutic imaging procedures with ionizing radiation in a large pediatric population. We found that the use of these procedures is common and that studies associated with high doses of radiation are not infrequent and are performed repeatedly in a smaller group of children. These results highlight the importance of generating databased guidelines to aid clinicians in determining the appropriateness of performing imaging procedures in children.

Correspondence: Adam L. Dorfman, MD, Department of Pediatrics, Division of Pediatric Cardiology, University of Michigan Health Systems, 1500 E Medical Center Dr, Women's L1242 SPC 5204, Ann Arbor, MI 48109 (adamdorf@med.umich.edu).

Accepted for Publication: November 3, 2010.

Published Online: January 3, 2011. doi:10.1001/archpediatrics.2010.270

Author Contributions: Dr Dorfman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Dorfman, Fazel, Einstein, Applegate, Christodoulou, Sanchez, and Nallamothu. Acquisition of data: Fazel and Nallamothu. Analysis and interpretation of data: Dorfman, Fazel, Einstein, Applegate, Krumholz, Wang, Christodoulou, Chen, and Nallamothu. Drafting of the manuscript: Dorfman, Applegate, Christodoulou, and Nallamothu. Critical revision of the manuscript for important intellectual content: Dorfman, Fazel, Einstein, Applegate, Krumholz, Wang, Christodoulou, Chen, Sanchez, and Nallamothu. Statistical analysis: Fazel, Applegate, and Wang. Obtained funding: Einstein. Administrative, technical, and material support: Applegate. Study supervision: Nallamothu.

Financial Disclosure: Dr Einstein has served as a consultant for the International Atomic Energy Agency and for GE Healthcare, has received support for other research from Spectrum Dynamics and a Nuclear Cardiology Foundation grant funded by Covidien, and has received travel funding from GE Healthcare, INVIA, Philips Medical Systems, and Toshiba America Medical Systems. Dr Applegate has a textbook contract for Evidence-Based Imaging in Pediatrics with Springer. Dr Krumholz is the chairman of a scientific advisory board for UnitedHealthcare.

Funding/Support: Dr Einstein was supported by K12 institutional career development award KL2 RR024157 from the National Institutes of Health, by the Louis V. Gerstner Jr Scholars Program, and by the Lewis Katz Cardiovascular Research Prize for a Young Investigator. Dr Chen was supported in part by American Heart Association Clinical Research Program Award 10CRP2640075 and Agency for Healthcare Research and Quality Career Development Award 1K08HS018781-01.