BACKGROUND AND OBJECTIVE: Diagnostic imaging procedures (DIPs) producing radiation exposure in children have been associated with increased cancer risk. To develop effective clinical practice interventions that reduce pediatric radiation exposure, a longitudinal assessment of DIP ordering practices in the current clinical climate was performed. Our objective was to determine the prevalence and characteristics of DIPs ordered by physicians in an independent physicians association treating a general pediatric population.
METHODS: DIP insurance claims from 2001 to 2009 were obtained from an independent physicians association in a large metropolitan area. Current Procedural Terminology codes, associated International Classification of Diseases, Ninth Revision codes, and patient demographics associated with DIPs were retrieved from insurance claim data and analyzed.
RESULTS: Overall, 214 538 DIPs were performed on 63 116 children from 2001 to 2009 at a rate of 1 DIP associated with ionizing radiation per 21 patient-months. Over the study period, performance of computed tomography, MRI, ultrasound, and radiography all significantly increased (P < .001 for all). Higher radiation DIPs were most frequently performed in the inpatient and emergency settings on older boy patients for gastrointestinal and congenital disorders. The 3 most common International Classification of Diseases, Ninth Revision codes associated with higher radiation DIPs were abdominal pain, headache, and head injury.
CONCLUSIONS: DIPs are frequently performed in children and higher radiation DIPs account for an increasing proportion of DIPs performed, especially among children evaluated in the inpatient and emergency department settings and those with gastrointestinal and neurologic symptoms, and congenital anomalies. Our findings may help guide development of clinical practice interventions to reduce DIP-related radiation exposure in youth.
- CT —
- computed tomography
- DIP —
- diagnostic imaging procedure
- ED —
- emergency department
- ICD-9 —
- International Classification of Diseases, Ninth Revision
- IPA —
- independent physicians association
What’s Known on This Subject:
Medical imaging that uses ionizing radiation provides notable benefits in the clinical setting. Controversy regarding increased cancer risk, particularly in children, dictates that ordering practices and use of such medical imaging be evaluated to reduce unnecessary exposure to imaging-related radiation.
What This Study Adds:
We evaluated the prevalence and characteristics of diagnostic imaging procedures in children. The proportion of higher radiation procedures is increasing, especially among children evaluated in the inpatient and emergency department settings and those with gastrointestinal and neurologic symptoms, and congenital anomalies.
Diagnostic imaging procedures (DIPs) are commonly performed for the assessment and monitoring of a variety of symptoms and diseases and have provided notable benefits in the clinical setting. Certain DIPs, such as computed tomography (CT), angiography, and fluoroscopy, are known to expose patients to significantly greater ionizing radiation than radiography.1,2 In contrast, MRI and ultrasound are not sources of ionizing radiation. There has been recent concern3,4 about the potential increased cancer risk with frequent performance of ionizing radiation-associated DIPs in the still developing pediatric patient. However, controversy exists regarding the actual contribution of medical imaging procedures to cancer incidence and deaths.5
To date, interventions such as the Image Gently6 campaign organized by the Alliance for Radiation Safety in Pediatric Imaging to increase awareness of the need for radiation protection for pediatric patients and to reduce radiation exposure in children receiving DIPs have primarily targeted the medical imaging community. However, this campaign remains to be fully implemented in adult settings that serve children (eg, adult hospital emergency departments [EDs]). Similarly, much work remains to be done in regards to changing how clinicians order DIPs to reduce ionizing radiation exposure to “as low as reasonably achievable”.7 This is likely because indications for performance of ionizing radiation-associated imaging procedures based on available evidence still need to be refined and defined.6
Opportunities may exist for interventions to reduce radiation exposure in children by modifying DIP-ordering practices by the clinician. To best develop such interventions, information regarding current DIP-ordering practices is needed. Although longitudinal data have been published in other countries8,9 and are available for limited pediatric cohorts10–12 in the United States, few studies have examined overall longitudinal rates of DIPs performed in the US pediatric population.13 Our objective in the current study was to determine the prevalence and characteristics of DIPs ordered by pediatric practitioners and performed in a general pediatric population cohort.
We conducted an investigator-initiated, retrospective cohort study evaluating patient claims data from an independent physicians association (IPA) comprising 160 pediatricians and 284 pediatric specialists linked with a tertiary-care pediatric hospital in the Greater San Diego area. After removing all personal identifiers, data were provided to the investigators for independent analysis and interpretation. The study protocol was reviewed and approved by the institutional review boards at the University of California, San Diego, and at Rady Children’s Hospital. The informed consent requirement was waived for this study. Data were collected between January 1, 2001, and December 31, 2009, and included CurrentProcedural Terminology codes, associated International Classification of Diseases, Ninth Revision (ICD-9) codes, performance site, and basic patient demographics of all DIPs performed on patients during the specified period as recorded from patient claims data. Radiology Current Procedural Terminology codes were further classified according to presence or absence of associated ionizing radiation. ICD-9 codes were categorized by organ or system related diseases or symptoms. The study population included all individuals covered by the IPA during the study period (2001–2009). Owing to differences in patient enrollment periods and variations in patient populations from year to year, DIP frequency data were presented in DIP per patient-months. The IPA-covered patient population’s patient-month total (denominator) was calculated by multiplying each enrolled patient by his or her enrollment period in months and summing the total.
All statistical analyses were performed with JMP statistical software (9th edition, SAS Institute, Inc, Cary, NC). After examining univariate and categorical descriptive frequencies for DIPs performed, diagnostic codes (ICD-9), patient demographics, and clinical performance sites, χ2 analyses for trend examined changes over time in proportions of specific types of DIPs performed adjusted for the growing insured patient population. The data were then stratified to look at specific types of radiologic diagnostics and to focus on the nonradiography, radiation-associated techniques. Univariate logistic analyses were also performed to test associations of patient characteristics, the most frequent ICD-9 diagnosis categories (trauma, gastrointestinal, congenital, genitourinary, musculoskeletal, pulmonary, and unspecified which in total accounted for 90% of all observations), and DIP performance settings with performance of higher radiation procedures. A saturated, restricted multivariate model was then tested controlling for these top 7 ICD-9 categories as well as select patient variables. Adjusted odds ratios and 95% confidence intervals were calculated.
Overall, 214 538 DIPs were performed on 63 116 unique children over 2001–2009. The study population consisted of 195 753 patients, had a mean age of 7 (6) years at enrollment, was 51% male, and was predominantly covered by health maintenance organizations (87%). By using the denominator of 3.98 million patient-months, 1 in 3 children received DIPs over the study period. One DIP was performed per 19 patient-months, and 1 radiation-associated DIP was performed per 21 patient-months. Among the 63 116 children who received at least 1 DIP during the study period, 39% had only 1 DIP with the majority (60%) having 2 or fewer DIPs (range 1–214 DIPs per patient). Of the 7969 who received higher radiation DIPs (CT, angiography, and/or fluoroscopy), the majority (62%) had only 1 procedure with a range of 1 to 83 higher radiation DIPs per patient.
Characteristics of the DIPs performed over the study period are listed in Table 1. The 3 most common DIPs performed were radiography, ultrasound, and CT, and the majority of DIPs were performed in the outpatient setting. Trauma, musculoskeletal symptoms and conditions, and pulmonary symptoms and conditions were the most common ICD-9 diagnostic categories for which DIPs were performed.
Absolute numbers of performed procedures are displayed according to type in Table 2 and Fig 1. Trend analyses of the various types of DIPs performed over the study period demonstrated that the frequencies of CT, MRI, ultrasound, and radiography in proportion to the increasing population of insured patients increased significantly (CT: χ2 = 295.2; P < .001; MRI: χ2 = 450.8; P < .001; ultrasound: χ2 = 67.4; P < .001; and radiography: χ2 = 702.6; P < .001). In particular, the likelihood of having a CT procedure increased by 34% over the 9-year observation period, whereas the likelihood of an MRI procedure increased by 84%.
Characteristics associated with higher radiation DIPs are displayed in Tables 3 and 4. Trauma and injuries, gastrointestinal symptoms and conditions, unspecified symptoms and conditions, and congenital/perinatal symptoms were the top diagnostic categories for which higher radiation DIPs were performed across the entire cohort, accounting for 51% of the procedures, but the rank order of these categories differed according to age. Between 36% and 41% of CT, fluoroscopy, and angiography procedures were ordered in the outpatient setting across all age groups.
Tables 5 and 6 display the results of univariate and multivariate logistic analyses performed to identify DIP and clinical variables associated with performance of higher radiation procedures. Univariate analyses demonstrated that performance of higher radiation DIPs was notably higher among inpatients and patients evaluated in the ED versus patients seen in other clinical settings, patients with gastrointestinal and congenital anomaly ICD-9 codes versus counterpart comparison groups, and associated with older versus younger patients. Boys and those managed under health maintenance organizations were also more likely to have higher radiation DIPs, as were patients whose conditions were coded as unspecified as compared with their respective comparison groups. These associations remained after the top 7 ICD-9 categories and select patient variables were entered in a saturated multivariate model.
We present DIP performance data among children evaluated by physicians of an IPA over a span of 9 years. Although 1 in 3 children in our cohort received DIPs over the study period, performance of DIPs occurred at a frequency of 1 procedure per 19 patient-months. In regards to radiation-associated DIPs, utilization was more infrequent at a frequency of 1 procedure per 21 patient-months; however, CT procedures overall significantly increased over the study period. The proportionate use of CT, MRI, ultrasound, and radiography increased over time, controlling for the increased patient base. The associations of use of higher radiation DIPs with studied variables differed significantly according to age and diagnosis.
Few pediatric data depicting DIP epidemiology among general pediatric cohorts are available. Dorfman et al13 evaluated 2 years of DIP data from >350 000 children from 1 health plan encompassing 5 large US regions. Other studies have published national data from the United Kingdom and Australia.8,9 Our study population is smaller than Dorfman et al13 but is able to address patterns of DIP performance over a longer period and permits analysis of trends in radiology practice and performance in the Southern California region.
In our cohort, we demonstrate performance of radiation-associated DIPs at a rate of 1 procedure per 21 patient-months and a significant increase in higher radiation DIPs over time. In comparison with data performed in other studies in the United States,13 our data demonstrate a slightly higher prevalence of radiation-associated DIP per patient-month (1 DIP per 29 patient-months in Dorfman et al13) within a comparable cohort but longer time frame. Pearce et al9 reported a prevalence rate of 3.54 CT scans per 1000 persons <22 years in Northern England, whereas Brady et al8 reported a rate of 46 CT scans per 1000 persons <24 years in Australia. We also demonstrate an increase in DIPs performed across all modalities, including CT, over the 9-year study period. Although a number of studies support our finding of an increase in CT performance in children over time,14,15 Townsend et al16 evaluated data from 39 children’s hospitals over a similar period (2003–2007) and demonstrated a reduction in performance of CTs (as a proportion of all cross-sectional imaging studies) over the 4-year study period. Differences in observed performance rates may reflect differences in base population characteristics, comparison variables (ie, CT as a proportion of cross-sectional imaging versus all DIPs), data collection methods, site performance (ED versus nonemergency locations), and our inclusion of fluoroscopy as a radiation-associated procedure. Although our study demonstrates some difference in rates compared with others published in the literature, the cohort size and length of study period reinforces our statistically significant increase in the proportion of DIPs, particularly in radiation-associated DIPs (specifically CT scans) performed over time and is further validated by other available data gathered both here in the United States (among adults) and abroad.8,14,15,17
We identify several clinical risk factors for increased likelihood of higher radiation DIP performance. In particular, older patients with gastrointestinal and congenital complaints evaluated in the inpatient and ED settings were more likely to undergo such procedures. The increased performance of higher radiation DIPs among older as compared with younger children is consistent with previously published data in children.9,13 In addition, although we are the first to explicitly demonstrate the relationship between specific ICD-9 categories and performance of higher radiation DIPs among children, the higher risk for higher radiation DIPs among patients with gastrointestinal and congenital complaints is well recognized in the literature.18–20 In contrast, although the literature confirms our findings that the ED should be a primary area of focus for reduction of performance of higher radiation DIPs, our data suggest that the greatest risk for higher radiation DIP performance occurs in the inpatient setting. Interventions aimed at reducing orders of higher radiation DIPs should thus target both emergency and inpatient clinical practices.
One issue not evaluated in this study is that of cumulative childhood radiation exposure. We have previously evaluated cumulative childhood radiation exposure in a longitudinal cohort of children followed for inflammatory bowel disease.12 However, we did not repeat this evaluation in the current cohort owing to lack of ability to account for all DIPs performed in a given individual’s lifetime (as restricted by the period of study), restrictions of billing data limited to physicians covered by the designated IPA, and Health Insurance Portability and Accountability Act requirements of patient data without identification. Nevertheless, we do report the prevalence of repeat higher radiation procedures in the current cohort, which may provide an estimate of the risk for cumulative radiation exposure.
Indications for performance of radiation-associated imaging procedures based on available evidence still need to be refined and defined.3 In our evaluation, we demonstrate performance of higher radiation DIPs, such as CT, angiography, and fluoroscopy, for specific ICD-9 codes for which such DIPs are not routinely indicated. In particular, abdominal pain and headaches/head injuries were the most common complaints for which CT was performed even though clinical guidelines for evaluation of these complaints do not currently require routine performance of CT or other higher radiation DIPs for adequate and accurate assessment.21–23 In contrast, various studies have revealed the benefit of diagnostic imaging (including higher radiation modalities) in patient assessment and triage of such complaints, as well as in reducing associated health care costs.24,25 The American College of Radiology has determined appropriateness criteria for performance of DIPs in various clinical scenarios, which state that higher radiation DIPs are “usually” appropriate in the setting of abdominal pain and fever; sudden onset of severe or unilateral headache, headache with sinusitis, and/or new headache with suspected meningitis or encephalitis; and in patients with moderate or high risk for intracranial injury.26 Despite such published guidelines, clinical practices diverge from these recommendations and suggest that improved conversations are needed between professional bodies authoring guidelines and practitioners performing care to reach the goal to perform only necessary DIPs for accurate and adequate assessment of patient complaints.
Several limitations should be noted regarding our analysis. First, our study analyzed a pediatric medical center IPA’s billing data. Although the data reflect only the properties of a specific population, the studied IPA is the largest IPA associated with the only pediatric medical center in the region (with a catchment of ∼1 million persons) and thus its DIP data are likely to be typical of the pediatric-focused care region although it does not incorporate pediatric DIP performance in adult-focused settings. Second, it is well known that reported ICD-9 codes are not always representative of a patient’s overall complaint or situation and do not readily capture whether a patient has a chronic condition that would predispose one to more DIPs. However, these codes do represent the immediate cause and indication for ordering the DIP and can therefore be considered informative in identifying risk factors for DIP performance and planning future targeted clinical interventions aimed at modifying DIP performance. Third, our analysis of the database was somewhat restricted by the prevalence of ICD-9 codes without a clearly associated category. We labeled codes (780–799) “unspecified symptoms” because they correspond to “symptoms, signs, abnormal results of laboratory or other investigative procedures, and ill-defined conditions regarding which no diagnosis classifiable elsewhere is recorded.”27 One in 10 DIPs in our database was linked with this unspecified ICD-9 category, which likely reflects either imprecise coding by the ordering physician or issues/problems with the current coding system. Either improved use of these diagnosis codes in the future or additional clinical data may therefore enhance our understanding of the ordering patterns for DIPs that use ionizing radiation in medical imaging.
DIPs are performed frequently during childhood. Clinical risk factors associated with higher exposure to ionizing radiation from DIPs include older age, evaluation in the ED and inpatient settings, and gastrointestinal, neurologic, or congenital complaints. Improved understanding of DIP-ordering practices of clinicians will enable development of interventions to reduce ionizing radiation exposure from DIPs.
We thank Dr. Nathaniel Chuang for his help in reviewing this article for concepts related to pediatric radiology practice and medical imaging radiation.
- Accepted August 21, 2012.
- Address correspondence to Jeannie S. Huang, MD, MPH, Associate Professor, Department of Pediatrics, 9500 Gilman Dr, MC 0984, La Jolla, CA 92093-0984. E-mail:
Mr Tompane drafted the initial article, performed the statistical analyses in conjunction with Drs Huang and Bush, and approved the final article as submitted; Dr Bush oversaw and helped perform the statistical analyses with Mr Tompane and Dr Huang, reviewed and revised the article, and approved the final article as submitted; Dr Dansky helped conceptualize the study, supervised and coordinated retrieval of the data, critically reviewed the article, and approved the final article as submitted; and Dr Huang conceptualized and designed the study, oversaw and performed the statistical analyses with Mr Tompane and Dr Bush, reviewed and revised the article, and approved the final article as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
- Furlow B
- Hall EJ,
- Brenner DJ
- ↵American Association of Physicists in Medicine. AAPM Position Statement on radiation risks from medical imaging procedures. American Association of Physicists in Medicine; 2011. Available at: www.aapm.org/org/policies/details.asp?id=318&type=PP. Accessed October 30, 2012
- Ait-Ali L,
- Andreassi MG,
- Foffa I,
- Spadoni I,
- Vano E,
- Picano E
- Lewis DW,
- Ashwal S,
- Dahl G,
- et al.,
- Quality Standards Subcommittee of the American Academy of Neurology,
- Practice Committee of the Child Neurology Society
- Di Lorenzo C,
- Colletti RB,
- Lehmann HP,
- et al.,
- AAP Subcommittee,
- NASPGHAN Committee on Chronic Abdominal Pain
- Kuppermann N,
- Holmes JF,
- Dayan PS,
- et al.,
- Pediatric Emergency Care Applied Research Network (PECARN)
- Peña BM,
- Taylor GA,
- Fishman SJ,
- Mandl KD
- ↵American College of Radiology. ACR appropriateness criteria. American College of Radiology; 2011. Available at: www.acr.org/Quality-Safety/Appropriateness-Criteria. Accessed October 30, 2012
- Copyright © 2013 by the American Academy of Pediatrics