Childhood exposure to ionizing radiation from computed tomography imaging in Nova Scotia

Abstract

BACKGROUND:

Examining radiation dose in the paediatric population is particularly important due to the vulnerability of paediatric patients (increased radiosensitive tissues and postexposure life-years) and risk for future radiogenic malignancy.

OBJECTIVES:

To evaluate trends in paediatric computed tomography (CT) use and ionizing radiation exposure using population-based data from Nova Scotia.

METHODS:

A retrospective, population-based cohort study of CT use in patients <20 years of age, from January 1, 2004 to December 31, 2011, was performed in Nova Scotia. CT examination data were retrieved from a provincial imaging repository. Trends in CT use were described, and both annual and cumulative effective dose exposures were calculated.

RESULTS:

In total, 29,452 CT events, involving up to 22,867 individuals were retrieved. Overall annual paediatric CT examination rates remained static (range 17.4 to 18.8 per 1000 per year). However, use in children <10 years of age decreased by >50% (P<0.001); this was counterbalanced by a steady increase among 15- to 19-year-olds (P<0.0001). Overall, 15.4% of scanned patients underwent ≥2 examinations, of which 58 patients (1.6%) exceeded 50 mSv of exposure.

CONCLUSIONS:

Despite a static rate in CT imaging among the entire cohort, children <15 years of age and, particularly, those <10 years of age displayed marked reductions in CT use. This may reflect increased awareness of campaigns emphasizing judicious CT use, revised clinical practice guidelines and increased availability of alternative modalities. A small subgroup demonstrated high-dose exposure (>50 mSv), and rates in individuals >15 years of age steadily increased, suggesting further exposure reduction efforts are necessary.

Computed tomography (CT) has widespread application and diagnostic benefits in paediatric medicine; however, concerns regarding the health risks for diagnostic imaging ionizing radiation exposure have led to concerted efforts to minimize paediatric CT examinations (1,2). Ionizing radiation exposure from medical imaging at doses comparable with paediatric CT may result in adverse long-term outcomes (3,4), including a risk for developing future malignancies (5,6). Predicated on the principle of ALARA (as low as reasonable achievable), international campaigns, in particular Image Gently, have sought to limit ionizing radiation exposure (1,7–9). However, patterns of use continue to vary on account of health care system factors, such as resource allocation and clinical practice culture.

Paediatric data from the 1990s to early 2000s from Canada, the United States (US) and internationally has demonstrated steadily increasing use for most diagnostic imaging procedures (10) including CT (6,11–14). Data emerging within the past decade suggests either static use patterns (15) or gradual reductions in CT use, particularly in patients <15 years of age (6,12,16). A recent study comparing US and Canadian (Ontario) patterns of CT use in the emergency department setting, demonstrated steadily decreasing rates for children <10 years of age in Ontario, yet static to slightly increasing rates within the US (14).

Changes in CT availability, the role of CT in clinical diagnosis, and shifts in the culture of medical practice have continued to evolve such that the current state of paediatric CT use is poorly described. Moreover, Canadian literature regarding this issue is limited. The objectives of the present study were to describe the current rates and patterns of CT use in the Nova Scotia paediatric population, and to provide estimates of cumulative effective dose (CED) exposure during an eight-year study period.

METHODS

Study design

The present study was a retrospective, population-based cohort analysis including all children and young adults <20 years of age in Nova Scotia, from January 1, 2004 to December 31, 2011, using CT data extracted from a provincial imaging repository. Ethics approval was obtained from the IWK Research Ethics Board (IWK Health Centre, Halifax, Nova Scotia). Given the enormity of the study sample and data volume, a waiver of individual consent was obtained. Approval from each of the nine local district health authorities (DHAs) was also obtained.

Study database and variables

Picture Archiving and Communication System – Nova Scotia (PACS-NS) is a CT examination repository that supports the hospitals within each of the nine provincial DHAs as well as the province’s sole paediatric centre. PACS-NS includes all CT examinations for Nova Scotia residents from approximately 15 operational scanners (year-to-year variance of one to two operational scanners). This database includes a limited number of CT examinations (2% to 8% per year) from out-of-province patients, which could not be excluded from analysis. Multi-slice CT technique comprised 80% to 95% of all examinations.

Data retrieved from the PACS-NS database included examination date, patient age on examination date, sex, CT examination code and anatomical region scanned, DHA and performing institution. CT examination codes were categorized as CT head and related (eg, facial bones or sinuses), CT chest and related, CT abdomen/pelvis and related, CT spine or neck and related, and CT extremity and related. CT codes excluded from data analysis were limited (1.2%), but included second-opinion reports, postmortem examinations, three-dimensional reconstructions and incomplete/failed examinations. All PACS-NS examinations associated with a provincial patient identification number (76%) were recoded with a unique, nonidentifying study number that enabled linkage analysis of same individual repeat examinations. Records lacking a patient identification number (due to inadequate transmission of examination details) were coded individually (23%). Multianatomical region examinations, such as a CT chest/abdomen were coded as two unique examinations – CT chest and CT abdomen – with corresponding effective dose (ED) estimates. Multiphase examinations could not be uniquely identified and were analyzed as single phase.

Effective dosimetry

Ionizing radiation ED was estimated for each CT examination according to organ system. CT dose relies on human (age, sex, body habitus), examination (type of examination, coverage, precautions used) and technical (eg, tube current, voltage, pitch) factors. In the present study, proxy radiation dosimetry estimations (17) were used because PACS-NS, similar to other imaging repositories, does not record all technical factors of each examination (3,17,18). ED estimates for CT were obtained from a German nationwide study in 2005/2006, and are consistent across similar paediatric ED literature (17). Dosimetric calculations were performed for CT examinations that involved the most significant radiation exposures and long-term adverse outcome potential – head, face/sinuses, chest, abdomen/pelvis and spine. Inconsistencies within published dosimetric estimations for cervical spine CT and heterogeneity of exposures with extremity CT rendered these examinations inappropriate for dosimetric analysis.

Statistical analysis

CT imaging rates were calculated according to age (0 to <5, 5 to <10, 10 to <15 and 15 to <20 years of age), using annual population data as the rate denominator (www.statscan.gc.ca). Population annual mean ED (MED) exposures (mean mSv exposure per patient exposed, according to age) and CED exposures (total exposure, in mSv, for each patient who underwent ≥2 CT examinations during the study period) were calculated. Poisson regression was used to explore the trends associated with rates over time and differences in rates of CT use across age groups. General linear models were used to evaluate trends in annual MED and differences in MED across age groups using Tukey’s adjustment for multiple comparisons. Statistical analysis was performed using SAS version 9.2 (SAS Institute Inc, USA).

RESULTS

CT use patterns

From January 1, 2004 to December 31, 2011, 29,452 patients <20 years of age underwent a CT examination in Nova Scotia. There were 278 (1.2%) examinations excluded from analysis. An additional 6870 (23%) CT examinations lacked a patient identifying number and were coded as unique individuals with one CT event. There were 22,304 (76%) examinations suitable for linkage analysis, resulting in 15,997 individuals having undergone ≥1 CT examination. Based on linkage analysis and a denominator of 29,174 examinations, it was estimated that 20,875 to 22,867 individuals underwent ≥1 CT examination from 2004 to 2011. Males accounted for 54% of examinations. Approximately one-half (53%) of examinations occurred at the provincial paediatric or adult tertiary care hospital, with approximately 70% of CT examinations for children <16 years of age performed at the paediatric centre. CT of the head, including examinations targeted to paranasal sinuses, temporal bones and facial bones, accounted for nearly one-half of all examinations. CT examination and patient characteristics are summarized in Table 1.

TABLE 1.

Computed tomography (CT) use for patients <20 years of age in Nova Scotia, from January 1, 2004 to December 31, 2011

Year	Population	CTs, n	Total population	Male	Female	Head	Chest	Abdomen/pelvis	Neck/spine	Extremity/other
2004	219,697	4071	18.5 (18–19.1)	20 (19.2–20.8)	17 (16.2–17.8)	11.3 (10.8–11.7)	1.3 (1.1–1.4)	3.3 (3.1–3.6)	1.6 (1.4–1.7)	1 (0.9–1.2)
2005	215,498	3802	17.6 (17.1–18.2)	19.6 (18.8–20.5)	15.6 (14.9–16.4)	10.4 (10–10.8)	1.2 (1.1–1.4)	3.1 (2.9–3.3)	1.5 (1.3–1.7)	1.2 (1–1.3)
2006	211,810	3987	18.8 (18.2–19.4)	20.7 (19.9–21.6)	16.9 (16.1–17.7)	10.5 (10–10.9)	1.5 (1.3–1.7)	3.9 (3.6–4.1)	1.5 (1.4–1.7)	1.2 (1–1.3)
2007	208,589	3697	17.7 (17.2–18.3)	19.6 (18.8–20.5)	15.7 (15–16.5)	9.7 (9.3–10.1)	1.5 (1.4–1.7)	3.8 (3.6–4.1)	1.4 (1.3–1.6)	1.1 (0.9–1.2)
2008	206,130	3614	17.5 (17–18.1)	18.2 (17.4–19)	16.7 (16–17.6)	9.4 (9–9.9)	1.6 (1.4–1.8)	3.6 (3.4–3.9)	1.6 (1.5–1.8)	1.2 (1–1.3)
2009	203,366	3790	18.6 (18.1–19.2)	19.4 (18.6–20.3)	17.8 (17–18.7)	10.1 (9.6–10.5)	1.8 (1.6–2)	3.7 (3.4–3.9)	1.5 (1.3–1.7)	1.4 (1.3–1.6)
2010	200,484	3546	17.7 (17.1–18.3)	19.5 (18.7–20.4)	15.8 (15–16.6)	9 (8.6–9.5)	1.8 (1.6–2)	3.6 (3.4–3.9)	1.7 (1.5–1.9)	1.4 (1.2–1.5)
2011	197,347	3439	17.4 (16.9–18)	18.8 (17.9–19.6)	16 (15.2–16.8)	8.6 (8.2–9.1)	1.5 (1.4–1.7)	3.7 (3.4–3.9)	1.9 (1.7–2.1)	1.5 (1.4–1.7)

Data presented as rate per 1000 (95% CI). A single CT event may be encoded as multiple ‘anatomical regions’ (eg, CT chest/abdomen is coded as CT chest and CT abdomen)

Population rates (number of individuals scanned per 1000) revealed minimal, nonsignificant variation over time (P=0.12), peaking at 18.8 in 2006, and dropping to a low of 17.4 in 2011. However, annual rates according to age group demonstrated significant variability. Steady rate declines over the study interval were observed in the zero to four (P<0.0001), five to nine (P<0.001) and 10 to 14 (P=0.001) year-old age groups. The overall static CT rate in the entire <20 year-old cohort can be attributed to a parallel increase in the 15- to 19-year-old age group (P<0.001) (Figure 1).

Figure 1).

Computed tomography use according to age group from 2004 to 2011 in Nova Scotia

Among the cohort that underwent ≥1 CT examination(s) during the study period, 3541 individuals (15%) received ≥2 CT examinations. Nearly 35% (n=1253) of the repeat examination cohort received ≥3 examinations; 50.9% (638) of that cohort received ≥4 examinations. Within the repeat CT group, 43% were performed at the same institution and to the same anatomical region(s). Of repeat CT examinations for the same anatomical region at a different institution, 6.3% occurred within 24 h (10.1% within 48 h) of the index examination.

Effective dose exposures

ED estimates from CT examinations were derived based on the age at radiation exposure (Table 2). Annual MED ranged from 2.6 mSv to 6.1 mSv per year for all patients exposed to ≥1 CT examination during the study period. Annual MED increased slightly during the study interval (P<0.001) (Table 3). CEDs (calculated for patients who underwent ≥2 CT examinations) are presented in Figure 2. Of the 3541 patients who underwent ≥2 CT examinations, 2972 patients (84%) were included in dosimetric analysis (16% were not suitable for dosimetric analysis). In the ≥2 CT examinations cohort, high-risk exposures (>50 mSv) comprised 58 individuals (1.6%), 11 of whom were exposed to >100 mSv. This cohort was exposed, according to anatomical region, to a mean number of 12.9 examinations (95% CI 11.3 to 14.4 examinations). A greater percentage received their initial examination from 2004 to 2006 (48.3%), compared with those exposed to <50 mSv (37.8%). The mean age of first exposure in the high-risk (>50 mSv) and lower-risk (<50 mSv) groups were nearly identical (13.8 years versus 13.7 years). There was no trend toward a specific CT examination anatomical region for the high-risk cohort.

TABLE 2.

Radiation effective dose to children according to age at exposure from computed tomography (CT) examinations

Region	Age, years
	<1	1–5	6–10	11–15	Adult (>16)
Brain	2.2	1.9	2.0	2.2	1.9
Facial/sinuses	0.5	0.5	0.5	0.6	0.9
Chest	2.2	2.5	3.0	3.3	5.9
Entire abdomen	4.8	5.4	5.8	6.7	10.4
Spine	11.4	8.0	7.6	6.9	10.1

Data presented as dose, mSv. Dosimetry based on multislice CT. Adapted from Galanski et al (17)

TABLE 3.

Annual mean effective dose (MED) for all patients <20 years of age who underwent ≥1 computed tomography examination from January 1, 2004 to December 31, 2011

Year	Age, years	Overall
		Patients, n	MED, mSv (95% CI)	Median, mSv (range)
2004	0–4	449	2.7 (2.5–3)	2.2 (0.5–35.2)
	5–9	400	2.6 (2.3–2.8)	2 (0.5–20.8)
	10–14	781	3.3 (3.1–3.5)	2.2 (0.5–33.2)
	15–19	1551	5.2 (5–5.5)	1.9 (0.6–62.1)
2005	0–4	316	2.8 (2.5–3)	2.2 (0.5–13.3)
	5–9	395	2.6 (2.4–2.9)	2 (0.5–32)
	10–14	641	3.1 (2.9–3.3)	2.2 (0.5–20.1)
	15–19	1566	5.4 (5.1–5.6)	2.2 (0.6–36.6)
2006	0–4	315	2.8 (2.6–3.1)	2.2 (0.5–19.7)
	5–9	382	2.8 (2.5–3.2)	2 (0.5–32.9)
	10–14	629	3.3 (3.1–3.5)	2.2 (0.5–42.5)
	15–19	1682	5.9 (5.6–6.2)	2.2 (0.6–65.2)
2007	0–4	246	2.7 (2.4–3.1)	2.2 (0.5–31.6)
	5–9	294	3.1 (2.6–3.5)	2 (0.5–41)
	10–14	593	3.4 (3.1–3.7)	2.2 (0.5–48.9)
	15–19	1674	5.9 (5.6–6.2)	2.2 (0.6–67.1)
2008	0–4	243	2.8 2.5–3.2	2.2 (0.5–25.6)
	5–9	253	3.1 (2.6–3.5)	2 (0.5–35.2)
	10–14	489	3.7 (3.4–4.1)	2.2 (0.5–26.9)
	15–19	1699	5.8 (5.4–6.1)	2.2 (0.6–84.8)
2009	0–4	251	3.3 (2.8–3.7)	1.9 (0.5–34.1)
	5–9	251	2.9 (2.6–3.2)	2 (0.5–17.4)
	10–14	503	3.7 (3.3–4)	2.2 (0.5–42.1)
	15–19	1695	5.8 (5.5–6.1)	1.9 (0.6–79.3)
2010	0–4	234	3.2 (2.8–3.7)	2.2 (0.5–25.6)
	5–9	208	3.2 (2.7–3.6)	2 (0.5–25.9)
	10–14	473	3.5 (3.3–3.8)	2.2 (0.5–22.3)
	15–19	1651	6.1 (5.8–6.5)	2.2 (0.6–169.2)
2011	0–4	167	3.3 (2.7–3.9)	2.2 (0.5–36.4)
	5–9	172	3.6 (2.9–4.3)	2 (0.5–35.2)
	10–14	453	3.9 3.6–4.2	2.2 (0.5–19.1)
	15–19	1601	6.1 (5.8–6.4)	2.2 (0.6–77.2)

Figure 2).

Cumulative effective dose (CED) exposures for patients having received ≥2 computed tomography examinations from 2004 to 2011 (log₁₀ scale)

DISCUSSION

To our knowledge, the present study was the first Canadian, paediatric, population-based analysis of CT use and CED exposures. From 2004 to 2011, the overall rate of CT use for the Nova Scotia paediatric population (individuals <20 years of age) was static. However, there was a marked decrease in CT use in children <10 years fo age, and a smaller yet significant reduction in those 10 to 14 years of age; in those 15 to 19 years of age, there was a steady increase in use. Decreasing rates of use in the paediatric population, specifically those <15 years of age, is consistent with data from other countries (12,16), as well as from Canadian emergency department CT use in children (14). US data are more inconsistent, although most recent data suggest a similar decline in use (6). While annual MED increased slightly over the study period, CT examinations, according to anatomical region, varied over this interval – a decrease in head CT examinations resulted in a proportionately increased dose contribution of abdominal/pelvic and chest CT.

Rate estimates of CT use from the present study are consistent with data from other jurisdictions, which demonstrate wide variability, from 3.54 per 1000 person years (<22 years) (19) to 46 per 1000 person years (<24 years) (20). Trends of use in paediatric patients <15 years of age may reflect the impact of campaigns, such as Image Gently, which aim to reduce unnecessary CT use and seek to achieve safer practice (eg, paediatric-specific dosing protocols). Development of clearer guidelines for CT as a diagnostic tool, particularly in head injury, may also contribute to current trends (21,22). Finally, increasing availability of other imaging modalities may enable equivalent or superior imaging without radiation exposure. During the study period, there was increased availability and use of both magnetic resonance imaging and ultrasound.

The cohort exposed to >50 mSv averaged nearly 13 CT exposures (according to anatomical region) and were exposed to their initial examination earlier in the study period (a bias that may result in a greater likelihood of increased exposures over time). A 2010 Canadian study (3), demonstrated CEDs exceeding 100 mSv in 41% of a cohort of paediatric cancer patients, and the majority of radiation stemmed from CT use. At the population level, a United Kingdom study reviewed ionizing radiation exposure from CT in children and young adults from 1985 to 2002. Follow-up of >175,000 individuals provided evidence of associations between radiation doses from CT and brain tumours or leukemia (5). Similarly, an American study projected malignancy outcomes exceeding 4000 children per year due to ionizing radiation (6); these projections remained substantial even if strict paediatric dosing protocols were presumed (23).

While absolute thresholds for exposure risk are not discernable, there is consensus that patients who are younger at the time of exposure, and those with multiple exposures over time, are at an increased risk for future radiogenic cancer (24,25). For individuals exposed to repeat CT examinations with CED exceeding 40 mSv to 50 mSv, the increased risk for future radiogenic malignancy becomes significant (24,26,27). Australian data demonstrated a 24% increased risk for cancer (any subtype) in children and adolescents exposed to a single CT examination, compared with those unexposed (28). A dose-response relationship of increased cancer risk was observed with the highest response (cancer risk) correlating with earlier exposure (28). Therefore, identifying opportunities to limit CT exposure is the prudent strategy, not only in children likely to experience multiple CT examinations, but to all paediatric patients. Furthermore, minimizing repeat examinations due to technical failures – particularly where index examinations are deemed inadequate or unacceptable – is of great importance.

In the context of increasing CT use during the 1990s, international campaigns emerged to champion changes in paediatric radiological practice, targeting strategies to minimize exposures. Driven by the ALARA principle, weight- and age-based CT protocols have been developed, or are underway, in most paediatric centres. The widespread implementation of Image Gently has not only targeted physicians, but has enlisted public support, particularly parents, to advocate for improved paediatric protocols, development of evidence-based guidelines for CT use, and seek increased availability of alternative modalities (8,29). Through Image Gently, parental health literacy regarding ionizing radiation has been fostered, and parents have become empowered to advocate for their children (30).

Finally, the present study highlights that further investigation is required to understand the rationale for increased CT use in individuals >15 years of age, and to seek exposure-limiting strategies for this cohort. Possible reasons for this upward trend include: higher rates of traumatic injury, presentation and treatment in adult centres with increased CT availability and the presence of a clinical practice culture promoting CT use within this cohort.

Strengths and limitations

PACS-NS represents a highly inclusive, population repository, capturing province-wide CT exposures over an eight-year interval. However, the database lacks unique patient identifiers for 23% of examinations, precluding them from linkage analysis. This resulted in an underestimation of CED, but not MED, exposures. Furthermore, the present study does not account for potential improvements in paediatric protocols over the study duration, but rather affixes constant dose estimates. In most diagnostic imaging departments, radiation doses are not logged for individual examinations; therefore, estimates are used in proxy and may underestimate actual practice (6). It is possible that our ED estimates do not accurately reflect current practice (CT protocols used, type of CT scanner) in Canada. Our estimates do not account for variability in CT slicing (single- versus multislice) or for the use of multi-phase techniques. A lack of well validated dose exposure estimates in published literature precluded cervical spine and extremity CT from dosimetric analysis. Finally, the present study analyzed CT exclusively, which is a subset (albeit likely the largest) of medical ionizing radiation exposure.

CONCLUSION

The present study described the current pattern of CT use and cumulative ionizing radiation exposure in a population-based paediatric cohort in Nova Scotia. Despite a relatively static rate of CT use in the <20 years of age cohort, significant reduction in use was observed in children <10 years of age and in those 10 to 14 years of age. A subset of individuals was exposed to a substantial CED, exceeding 50 mSv, and may be at risk for radiogenic malignancy. Patients >15 years of age require further study to discern why rates of CT use are increasing, and to develop strategies that could reverse this trend. In general, future efforts should be directed toward the clarification of CT diagnostic indications, ensuring paediatric specific CT dosing protocols are used, and increasing the availability of alternative modalities.