TY - JOUR
AU - Mathews, John, D
AB - Abstract Children undergoing computed tomography (CT) scans have an increased risk of cancer in subsequent years, but it is unclear how much of the excess risk is due to reverse causation bias or confounding, rather than to causal effects of ionising radiation. An examination of the relationship between excess cancer risk and organ dose can help to resolve these uncertainties. Accordingly, we have estimated doses to 33 different organs arising from over 900 000 CT scans between 1985 and 2005 in our previously described cohort of almost 12 million Australians aged 0–19 years. We used a multi-tiered approach, starting with Medicare billing details for government-funded scans. We reconstructed technical parameters from national surveys, clinical protocols, regulator databases and peer-reviewed literature to estimate almost 28 000 000 individual organ doses. Doses were age-dependent and tended to decrease over time due to technological improvements and optimisation. INTRODUCTION It was previously thought that epidemiological studies would be too small to detect any increase in cancer risks following exposure to ionising radiation from diagnostic medical imaging(1). However, recent large studies have shown increased cancer risks after computed tomography (CT) scans in children and young people(2–6). These studies have been made possible through data linkage of health records as well as increased data availability in recent years. It is pertinent to study the effects of CT exposure as it is a key modality within imaging practices. Diagnostic imaging using CT is highly beneficial for its low contrast capabilities and its ability to reconstruct either very thin planes at various angles (e.g. axial, coronal, sagittal) throughout the body or more complicated multiplanar reformats. This is enabled by isotropic resolution in all axes to allow 3D reconstructions. With technological advances, such as dual-energy imaging and iterative reconstruction, the diagnostic power of CT in medicine will continue to grow. Individual radiation doses from CT scans need to be linked to individual disease outcomes to assess disease causation and to characterise the dose-response relationship for cancer risk in organs exposed. This is challenging as radiation doses cannot be directly measured but must be estimated for the large cohorts needed to observe an effect. For such large datasets, it is rarely feasible to access archived clinical films or electronic radiological records to retrieve technical parameters from past years, meaning that dose information is essentially unavailable at the individual level(7). Several approaches have been used to retrospectively construct dosimetry depending on the time period being studied and the available sources of information. These include the use of earlier nationwide surveys in the UK(2,8), sampling old records from the film(9), extracting parameters from the Picture Archiving and Communication System (PACS) for more recent exposures(5,10–12), retrieving data from questionnaires, surveys, scientific publications and expert interviews (pre-PACS)(11,12), reconstructing from radiology department clinical CT protocols(13), or choosing not to use doses because of the individual variation and lack of data(4,6). We have used a multi-tiered approach to estimate organ doses from CT scans for the Australian Paediatric Exposure to Radiation Cohort (Aust-PERC). We previously established a cohort of over 11 million young Australians using reimbursement claims for many different types of health services funded under the federal government health scheme (Medicare)(3). Within this cohort, individuals exposed to radiation from CT scans have been identified. Medicare information on the funded CT services did not include technical parameter information but was limited to the billing description, date of service and age and sex of the individual scanned. We previously assessed the cancer risk in this cohort in a retrospective study based on generic whole-body dose estimates (effective dose) and used the number of CT scans as a proxy for dose(3). Here we describe the methodology that we developed by using national and state surveys, clinical protocols, CT manufacturer settings, Australian regulator databases of CT scanners and peer-reviewed literature for Australian CT use to estimate retrospective organ doses for over 900 000 CT scan exposures in the Australian cohort between 1985 and 2005. This organ dosimetry will allow a more detailed assessment of the dose-response relationship in this important cohort. METHODS Cohort The cohort was established with approval from the Human Research Ethics Committees of the University of Melbourne, the Australian Institute of Health and Welfare and other government agencies, and from data custodians for Medicare and all states and territories of Australia. De-identified electronic records from the universal Australian Medicare system were made available for all Australians aged 0–19 years between 1985 and 2005. These data included billing records for 902 031 CT scans, which we used to estimate individual organ doses in the exposure period (1 January 1985 to 31 December 2005). The Medicare records were linked electronically to the Australian Cancer Database and the National Death Index maintained by the Australian Institute of Health and Welfare, although these outcomes have not been considered here. Types of CT scans CT scan types were identified using the Medicare Benefits Schedule (MBS) item numbers and descriptions. The dataset contained 212 unique MBS items, which could be grouped by eight broad anatomical categories into 35 CT scan types (Table 1). Many identical or similar scans, allocated different MBS item numbers between 1985 and 2005, could be grouped together(14). Where possible, the same anatomical classifications as previously used for the UK CT cohort(2,8) were used to align with their methodology. Table 1 Anatomical and CT scan categories. Anatomical region . CT scan typea . Anatomical scan range and weightingsb (see Figure 1) . Head Brain Brain Facial bones and brain Facial bone + brain Orbits ± brain Partial brain Facial bones Facial bone Middle ear and temporal bone ± brain Partial brain Pituitary fossa ± brain Partial brain Petrous bones ± brain Partial brain Temporal bones with air Partial brain Neck Soft tissue neck Neck Spine Cervical spine C-spine Thoracic spine T-spine Lumbosacral spine L-spine Spine: 1 or more regionsc 24% c-spine 5% t-spine 71% l-spine Spine: 2 regions 25% c-spine + t-spine 75% t-spine + l-spine Spine: 3 regions Whole spine Chest Chest Chest Chest ± upper abdomen Chest 1% included abdomen Chest and upper abdomen Chest + abdomen Abdomen and/or Pelvis Upper abdomen Abdomen Pelvis Pelvis Pelvimetry Pelvis Upper abdomen and pelvis Abdomen and pelvis Upper abdomen or pelvis 50% abdomen 50% pelvis Colon Abdomen Combined Chest and upper abdomen and brain Chest + abdomen + brain Chest and brain Chest + brain Chest and brain ± upper abdomen Chest + brain 1% included abdomen Chest and abdomen and pelvis Chest + abdomen and pelvis Chest and abdomen and pelvis ± neck Chest + abdomen and pelvis 10% included neck Chest and abdomen and pelvis and neck Chest + abdomen and pelvis + neck Extremities Extremities: 1 or more regions Extremities Other Interventional Abdomen Spiral angiography 33% head 33% chest 33% abdomen Dynamicd 33% chest 33% abdomen 33% pelvis Body scan on a body scannere Weighted average of body scans in 1988 Anatomical region . CT scan typea . Anatomical scan range and weightingsb (see Figure 1) . Head Brain Brain Facial bones and brain Facial bone + brain Orbits ± brain Partial brain Facial bones Facial bone Middle ear and temporal bone ± brain Partial brain Pituitary fossa ± brain Partial brain Petrous bones ± brain Partial brain Temporal bones with air Partial brain Neck Soft tissue neck Neck Spine Cervical spine C-spine Thoracic spine T-spine Lumbosacral spine L-spine Spine: 1 or more regionsc 24% c-spine 5% t-spine 71% l-spine Spine: 2 regions 25% c-spine + t-spine 75% t-spine + l-spine Spine: 3 regions Whole spine Chest Chest Chest Chest ± upper abdomen Chest 1% included abdomen Chest and upper abdomen Chest + abdomen Abdomen and/or Pelvis Upper abdomen Abdomen Pelvis Pelvis Pelvimetry Pelvis Upper abdomen and pelvis Abdomen and pelvis Upper abdomen or pelvis 50% abdomen 50% pelvis Colon Abdomen Combined Chest and upper abdomen and brain Chest + abdomen + brain Chest and brain Chest + brain Chest and brain ± upper abdomen Chest + brain 1% included abdomen Chest and abdomen and pelvis Chest + abdomen and pelvis Chest and abdomen and pelvis ± neck Chest + abdomen and pelvis 10% included neck Chest and abdomen and pelvis and neck Chest + abdomen and pelvis + neck Extremities Extremities: 1 or more regions Extremities Other Interventional Abdomen Spiral angiography 33% head 33% chest 33% abdomen Dynamicd 33% chest 33% abdomen 33% pelvis Body scan on a body scannere Weighted average of body scans in 1988 a±means that a scan is optionally performed with the add on. For example, ‘Orbits ± Brain’ means a scan of the orbits, with a scan of the brain optionally performed in addition. bSome scan descriptions included optional imaging (e.g. scan of chest with or without scans of the upper abdomen) and in these cases a ‘weighting’ was used if one could be determined for the percentage of scans that would include the optional scan (e.g. upper abdomen). These were calculated as follows: • Orbits, middle ear and temporal bone, pituitary fossa, petrous bones: there was no way of estimating whether a brain scan had been performed with these. Therefore, it was allocated the scan range and dose for a partial brain scan. • Spine: prior to 2001 there was no information on the region of the spine that was scanned. Instead the descriptor was ‘one or more regions’. During 2001, this description was replaced specifically with cervical spine (c-spine), thoracic spine (t-spine) and lumbosacral spine (l-spine) items as well as an item for scans of two spinal regions and an item for all three regions. From 2002, 95% of scans are of a single region only. The scans in 2001 and earlier that were defined as ‘one or more regions’ were weighted according to c-spine (24%), t-spine (5%) and l-spine (71%) as these breakdowns remained consistent between 2002 and 2005. It was assumed that scans of two regions would include only adjacent regions (i.e. only cervico-thoracic (C-T) scans or thoraco-lumbar (TL)). Based on the single region breakdown, it was considered that 25% would be C-T and 75% TL. • Chest ± upper abdomen: the descriptor to include an optional upper abdomen scan with a chest scan was introduced in 1996. Prior to this, the items were clearly defined and only 1% of chest scans included the upper abdomen. This weighting was applied from 1996. This also applies to the chest and brain ± upper abdomen category. • Chest and abdomen and pelvis ± neck: the descriptor to include an optional neck scan with a chest, abdomen and pelvis (CAP) scan was introduced in 1996. Prior to this, the items were clearly defined and 10% of CAP scans included the neck. This weighting was applied from 1996. Other body scans on a body scanner: in 1988 body scans were defined more definitively in terms of the anatomy being scanned. The number of scans performed in 1988 was used to weight the body scans performed in 1985–1987 according to chest (5%), chest and abdomen (1%), chest and abdomen and pelvis (1%), abdomen and pelvis (10%), abdomen or pelvis (11%), pelvis (3%), spine (42%) and extremities (27%). Note that head scans were a separate MBS item number. cScans of 1 or more regions may include 1, 2 or 3 regions. Although this appears to include the categories of 2 regions or 3 regions, this overlap is due to different MBS item numbers applying in different years. dDynamic scans were likely to be some type of angiography scan performed with another type of scan and were not billed to this MBS item after 1996. It was recognised that a proportion of the existing Medicare items were already rendered as dynamic scans and therefore the funding amount was increased for these scans while the item for dynamic scans was removed. eBody scans were performed between 1985 and 1987 when detailed MBS items were not yet implemented for CT services. Open in new tab Table 1 Anatomical and CT scan categories. Anatomical region . CT scan typea . Anatomical scan range and weightingsb (see Figure 1) . Head Brain Brain Facial bones and brain Facial bone + brain Orbits ± brain Partial brain Facial bones Facial bone Middle ear and temporal bone ± brain Partial brain Pituitary fossa ± brain Partial brain Petrous bones ± brain Partial brain Temporal bones with air Partial brain Neck Soft tissue neck Neck Spine Cervical spine C-spine Thoracic spine T-spine Lumbosacral spine L-spine Spine: 1 or more regionsc 24% c-spine 5% t-spine 71% l-spine Spine: 2 regions 25% c-spine + t-spine 75% t-spine + l-spine Spine: 3 regions Whole spine Chest Chest Chest Chest ± upper abdomen Chest 1% included abdomen Chest and upper abdomen Chest + abdomen Abdomen and/or Pelvis Upper abdomen Abdomen Pelvis Pelvis Pelvimetry Pelvis Upper abdomen and pelvis Abdomen and pelvis Upper abdomen or pelvis 50% abdomen 50% pelvis Colon Abdomen Combined Chest and upper abdomen and brain Chest + abdomen + brain Chest and brain Chest + brain Chest and brain ± upper abdomen Chest + brain 1% included abdomen Chest and abdomen and pelvis Chest + abdomen and pelvis Chest and abdomen and pelvis ± neck Chest + abdomen and pelvis 10% included neck Chest and abdomen and pelvis and neck Chest + abdomen and pelvis + neck Extremities Extremities: 1 or more regions Extremities Other Interventional Abdomen Spiral angiography 33% head 33% chest 33% abdomen Dynamicd 33% chest 33% abdomen 33% pelvis Body scan on a body scannere Weighted average of body scans in 1988 Anatomical region . CT scan typea . Anatomical scan range and weightingsb (see Figure 1) . Head Brain Brain Facial bones and brain Facial bone + brain Orbits ± brain Partial brain Facial bones Facial bone Middle ear and temporal bone ± brain Partial brain Pituitary fossa ± brain Partial brain Petrous bones ± brain Partial brain Temporal bones with air Partial brain Neck Soft tissue neck Neck Spine Cervical spine C-spine Thoracic spine T-spine Lumbosacral spine L-spine Spine: 1 or more regionsc 24% c-spine 5% t-spine 71% l-spine Spine: 2 regions 25% c-spine + t-spine 75% t-spine + l-spine Spine: 3 regions Whole spine Chest Chest Chest Chest ± upper abdomen Chest 1% included abdomen Chest and upper abdomen Chest + abdomen Abdomen and/or Pelvis Upper abdomen Abdomen Pelvis Pelvis Pelvimetry Pelvis Upper abdomen and pelvis Abdomen and pelvis Upper abdomen or pelvis 50% abdomen 50% pelvis Colon Abdomen Combined Chest and upper abdomen and brain Chest + abdomen + brain Chest and brain Chest + brain Chest and brain ± upper abdomen Chest + brain 1% included abdomen Chest and abdomen and pelvis Chest + abdomen and pelvis Chest and abdomen and pelvis ± neck Chest + abdomen and pelvis 10% included neck Chest and abdomen and pelvis and neck Chest + abdomen and pelvis + neck Extremities Extremities: 1 or more regions Extremities Other Interventional Abdomen Spiral angiography 33% head 33% chest 33% abdomen Dynamicd 33% chest 33% abdomen 33% pelvis Body scan on a body scannere Weighted average of body scans in 1988 a±means that a scan is optionally performed with the add on. For example, ‘Orbits ± Brain’ means a scan of the orbits, with a scan of the brain optionally performed in addition. bSome scan descriptions included optional imaging (e.g. scan of chest with or without scans of the upper abdomen) and in these cases a ‘weighting’ was used if one could be determined for the percentage of scans that would include the optional scan (e.g. upper abdomen). These were calculated as follows: • Orbits, middle ear and temporal bone, pituitary fossa, petrous bones: there was no way of estimating whether a brain scan had been performed with these. Therefore, it was allocated the scan range and dose for a partial brain scan. • Spine: prior to 2001 there was no information on the region of the spine that was scanned. Instead the descriptor was ‘one or more regions’. During 2001, this description was replaced specifically with cervical spine (c-spine), thoracic spine (t-spine) and lumbosacral spine (l-spine) items as well as an item for scans of two spinal regions and an item for all three regions. From 2002, 95% of scans are of a single region only. The scans in 2001 and earlier that were defined as ‘one or more regions’ were weighted according to c-spine (24%), t-spine (5%) and l-spine (71%) as these breakdowns remained consistent between 2002 and 2005. It was assumed that scans of two regions would include only adjacent regions (i.e. only cervico-thoracic (C-T) scans or thoraco-lumbar (TL)). Based on the single region breakdown, it was considered that 25% would be C-T and 75% TL. • Chest ± upper abdomen: the descriptor to include an optional upper abdomen scan with a chest scan was introduced in 1996. Prior to this, the items were clearly defined and only 1% of chest scans included the upper abdomen. This weighting was applied from 1996. This also applies to the chest and brain ± upper abdomen category. • Chest and abdomen and pelvis ± neck: the descriptor to include an optional neck scan with a chest, abdomen and pelvis (CAP) scan was introduced in 1996. Prior to this, the items were clearly defined and 10% of CAP scans included the neck. This weighting was applied from 1996. Other body scans on a body scanner: in 1988 body scans were defined more definitively in terms of the anatomy being scanned. The number of scans performed in 1988 was used to weight the body scans performed in 1985–1987 according to chest (5%), chest and abdomen (1%), chest and abdomen and pelvis (1%), abdomen and pelvis (10%), abdomen or pelvis (11%), pelvis (3%), spine (42%) and extremities (27%). Note that head scans were a separate MBS item number. cScans of 1 or more regions may include 1, 2 or 3 regions. Although this appears to include the categories of 2 regions or 3 regions, this overlap is due to different MBS item numbers applying in different years. dDynamic scans were likely to be some type of angiography scan performed with another type of scan and were not billed to this MBS item after 1996. It was recognised that a proportion of the existing Medicare items were already rendered as dynamic scans and therefore the funding amount was increased for these scans while the item for dynamic scans was removed. eBody scans were performed between 1985 and 1987 when detailed MBS items were not yet implemented for CT services. Open in new tab A typical item descriptor was ‘scan of brain with intravenous contrast medium and with any scans of the brain prior to intravenous contrast injection, when undertaken’ (MBS item 56007)(15). This was easily allocated to the ‘head’ anatomical category and then grouped with the ‘brain’ CT scan types. Subcategories were further divided according to whether a radio-dense contrast agent was used. CT scans can be performed with or without contrast, or involve both a pre- and post-contrast phase. The latter ‘dual-phase’ scans result in twice the radiation dose compared with a single-phase scan (with contrast or without contrast) when the same anatomy is exposed. The number of phases is therefore an important element to consider. A ‘phase factor’ was allocated to each scan type depending on the billing description. For most scans, the phase factor was one, as only a single scan was undertaken (either with contrast or without). For those that involved dual-phase scanning, the phase factor was 2. In some instances (such as MBS item 56007 above), it was not clear how many phases had been performed for a scan. This ambiguous contrast description was introduced into Medicare for some scan types in 1996. For these types of scans, a phase factor was determined based on the breakdown of single/dual-phase scans during earlier years when the scans were clearly defined. Two experienced radiographers were also consulted at a dedicated children’s hospital regarding how these scans were typically performed towards the end of the study period (a time at which both radiographers were working) and in the present day. This process of assessing the scan types using the clear MBS definitions, combined with clinical experience, gave a phase factor between one and two for those billing descriptions that were not well defined from 1996 (Table 2). Scans involving the abdomen and/or pelvis involved more dual-phase scanning in the earlier years and based on clinical experience it is now quite rare for these to be dual-phase. This necessitated two time-dependent phase factors after 1996 to reflect the more recent practice of single-phase scanning of this anatomy. Details for these contrast categories have previously been described(14). Table 2 Phase factor for each scan type. Years . Scan type/description . Phase factor . 1985–2005 With contrast 1 1985–2005 Without contrast 1 1985–1996 With and without contrast 2 1996–2005 With and/or without contrasta Head 2.0 Neck 1.0 Spine 1.2 Chest 1.6 Combined 1.5 Extremities 1.7 Other 1.1 1996–2000 With and/or without contrasta Abdomen and/or Pelvis 1.7 2001–2005 With and/or without contrasta Abdomen and/or Pelvis 1.2 Years . Scan type/description . Phase factor . 1985–2005 With contrast 1 1985–2005 Without contrast 1 1985–1996 With and without contrast 2 1996–2005 With and/or without contrasta Head 2.0 Neck 1.0 Spine 1.2 Chest 1.6 Combined 1.5 Extremities 1.7 Other 1.1 1996–2000 With and/or without contrasta Abdomen and/or Pelvis 1.7 2001–2005 With and/or without contrasta Abdomen and/or Pelvis 1.2 aWith and/or without contrast was a description added to the Medicare item numbers from 1996 and generally replaced the descriptor ‘with and without contrast’. However, it appears to have also been used for scans that were ‘with contrast’ only, even though a separate descriptor did exist for these single-phase scans, necessitating the calculation of the phase factors shown above. Open in new tab Table 2 Phase factor for each scan type. Years . Scan type/description . Phase factor . 1985–2005 With contrast 1 1985–2005 Without contrast 1 1985–1996 With and without contrast 2 1996–2005 With and/or without contrasta Head 2.0 Neck 1.0 Spine 1.2 Chest 1.6 Combined 1.5 Extremities 1.7 Other 1.1 1996–2000 With and/or without contrasta Abdomen and/or Pelvis 1.7 2001–2005 With and/or without contrasta Abdomen and/or Pelvis 1.2 Years . Scan type/description . Phase factor . 1985–2005 With contrast 1 1985–2005 Without contrast 1 1985–1996 With and without contrast 2 1996–2005 With and/or without contrasta Head 2.0 Neck 1.0 Spine 1.2 Chest 1.6 Combined 1.5 Extremities 1.7 Other 1.1 1996–2000 With and/or without contrasta Abdomen and/or Pelvis 1.7 2001–2005 With and/or without contrasta Abdomen and/or Pelvis 1.2 aWith and/or without contrast was a description added to the Medicare item numbers from 1996 and generally replaced the descriptor ‘with and without contrast’. However, it appears to have also been used for scans that were ‘with contrast’ only, even though a separate descriptor did exist for these single-phase scans, necessitating the calculation of the phase factors shown above. Open in new tab After anatomical and contrast categorisation there were 79 unique CT scan types for the purpose of organ dose estimation. Organ dose estimation The National Cancer Institute dosimetry system for CT (NCICT) was used to calculate organ doses(16). This program is based on a Monte Carlo simulation of a reference CT scanner and the International Commission on Radiological Protection’s (ICRP) reference paediatric and adult voxel phantoms(17,18) to generate a database of organ dose coefficients. Calculating organ doses for specific CT exposure scenarios, requires inputs of age and sex, the CT slice locations of scan start and scan end, tube potential (kVp), volumetric computed tomography dose index (CTDIvol) and the size of the CTDI phantom (16 or 32 cm) used for deriving CTDIvol. Ideally, we would have used technical parameters from each scan as inputs, but these were unavailable. Accordingly, for each of the scan types, we worked to reconstruct average parameters from published and unpublished records relevant to Australian CT imaging practice for paediatric patients in the study period (Table 3). Table 3 Data sources for regression modelling of Australian CTDIvol estimates. Study . Type of study . Relevant yeara . Number of participantsb . Age of participants . CT scan type . 1. Henson(23) Clinical Protocol 1982 NA Adult Brain, orbit, abdomen 2. GE CT9800 Patient Protocol(30) Clinical Protocol 1985c NA 0-18 y Adult Brain, chest, abdomen 3. Sim and Case(24) Clinical Protocol 1986 NA Adult Brain 4. Thomson and Tingey(27) National Survey 1994 NA 0-18 y Brain, pituitary fossa, chest, cervical spine, thoracic spine, lumbar spine 5. Boal et al.(29) State Survey 2000 NA Adult Brain, chest, pelvis, abdomen 6. Moss and McLean(28) State Survey 2001d 26-98 (depending on scan type and age) 8 weeks 5–7 years Adult Brain, chest, abdomen 7. Brady(25) Hospital Survey 2009 220 0–18 y Brain, chest, abdomen/pelvis 8. Hayton and Wallace(26) National Survey 2009 1265 0–14 y Brain, chest, abdomen/pelvis Study . Type of study . Relevant yeara . Number of participantsb . Age of participants . CT scan type . 1. Henson(23) Clinical Protocol 1982 NA Adult Brain, orbit, abdomen 2. GE CT9800 Patient Protocol(30) Clinical Protocol 1985c NA 0-18 y Adult Brain, chest, abdomen 3. Sim and Case(24) Clinical Protocol 1986 NA Adult Brain 4. Thomson and Tingey(27) National Survey 1994 NA 0-18 y Brain, pituitary fossa, chest, cervical spine, thoracic spine, lumbar spine 5. Boal et al.(29) State Survey 2000 NA Adult Brain, chest, pelvis, abdomen 6. Moss and McLean(28) State Survey 2001d 26-98 (depending on scan type and age) 8 weeks 5–7 years Adult Brain, chest, abdomen 7. Brady(25) Hospital Survey 2009 220 0–18 y Brain, chest, abdomen/pelvis 8. Hayton and Wallace(26) National Survey 2009 1265 0–14 y Brain, chest, abdomen/pelvis aThe year relevant to when the survey data were collected, or the clinical protocol was used. bThe number of participants is not applicable (NA) if it was a single clinical protocol or a survey based on submission of protocol parameters rather than patient scans. cThis type of scanner was frequently installed in Australia and State CT registration records show this was installed from as early as 1985. dThe maximum (outlier) values found in this survey were greater than regression estimates based on other data inputs for 2001. This suggests that the outlier values correspond to radiological practice, which were slow to change from the 1990s. Open in new tab Table 3 Data sources for regression modelling of Australian CTDIvol estimates. Study . Type of study . Relevant yeara . Number of participantsb . Age of participants . CT scan type . 1. Henson(23) Clinical Protocol 1982 NA Adult Brain, orbit, abdomen 2. GE CT9800 Patient Protocol(30) Clinical Protocol 1985c NA 0-18 y Adult Brain, chest, abdomen 3. Sim and Case(24) Clinical Protocol 1986 NA Adult Brain 4. Thomson and Tingey(27) National Survey 1994 NA 0-18 y Brain, pituitary fossa, chest, cervical spine, thoracic spine, lumbar spine 5. Boal et al.(29) State Survey 2000 NA Adult Brain, chest, pelvis, abdomen 6. Moss and McLean(28) State Survey 2001d 26-98 (depending on scan type and age) 8 weeks 5–7 years Adult Brain, chest, abdomen 7. Brady(25) Hospital Survey 2009 220 0–18 y Brain, chest, abdomen/pelvis 8. Hayton and Wallace(26) National Survey 2009 1265 0–14 y Brain, chest, abdomen/pelvis Study . Type of study . Relevant yeara . Number of participantsb . Age of participants . CT scan type . 1. Henson(23) Clinical Protocol 1982 NA Adult Brain, orbit, abdomen 2. GE CT9800 Patient Protocol(30) Clinical Protocol 1985c NA 0-18 y Adult Brain, chest, abdomen 3. Sim and Case(24) Clinical Protocol 1986 NA Adult Brain 4. Thomson and Tingey(27) National Survey 1994 NA 0-18 y Brain, pituitary fossa, chest, cervical spine, thoracic spine, lumbar spine 5. Boal et al.(29) State Survey 2000 NA Adult Brain, chest, pelvis, abdomen 6. Moss and McLean(28) State Survey 2001d 26-98 (depending on scan type and age) 8 weeks 5–7 years Adult Brain, chest, abdomen 7. Brady(25) Hospital Survey 2009 220 0–18 y Brain, chest, abdomen/pelvis 8. Hayton and Wallace(26) National Survey 2009 1265 0–14 y Brain, chest, abdomen/pelvis aThe year relevant to when the survey data were collected, or the clinical protocol was used. bThe number of participants is not applicable (NA) if it was a single clinical protocol or a survey based on submission of protocol parameters rather than patient scans. cThis type of scanner was frequently installed in Australia and State CT registration records show this was installed from as early as 1985. dThe maximum (outlier) values found in this survey were greater than regression estimates based on other data inputs for 2001. This suggests that the outlier values correspond to radiological practice, which were slow to change from the 1990s. Open in new tab CTDIvol reconstruction We reconstructed CTDIvol values, which quantify the X-ray tube output from the CT scanner, summarising the effects of key scan parameters used for the scan (e.g. kVp, tube current (mA), rotation time, beam collimation and pitch in the case of helical scans). CTDIvol is derived from a weighted average of doses at the centre and periphery measured inside a cylindrical Perspex phantom in two reference sizes depending on the type of scan being performed (16 cm or 32 cm diameter). Since only paediatric patients were included in the current study, all CTDIvol values for the 32 cm phantoms were converted to the 16 cm reference phantom where necessary for consistent input. The ratio of CTDIvol values in terms of the 16 cm reference phantom to the 32 cm reference phantom is dependent on the beam quality (kVp and filtration)(19) and has been shown to be ~2(20,21). There are established CTDIvol libraries (e.g. the ImPACT CT Dosimetry Calculator(22) and NCICT(16)) and tools to interface with these so that direct measurement is not necessarily required to determine CTDIvol if the type of scanner and parameters used for the scan are known. This allowed retrospective calculation of CTDIvol values for a select number of body areas where enough information was available. A review of literature on Australian CT use was undertaken to either obtain direct protocol parameters(23,24) or measured and/or calculated CTDIvol values(25). In the cases where parameters were provided, the ImPACT CT Dosimetry Calculator was used to calculate the CTDIvol value for the type of scanner and parameters listed(22). Several surveys of Australian practice were also identified(26–28) and a set of regulatory guidance levels(29). The authors of most of these surveys and studies still practice in Australia and some were contacted to clarify methodologies when necessary. The Australian Radiation Protection and Nuclear Safety Agency (ARPANSA, previously the Australian Radiation Laboratory (ARL)) provided de-identified raw data for paediatric hospitals that participated in the 1994 ARL national survey(27), allowing detailed CTDIvol reconstruction for several protocol types. We used two sources of Australian information(25,26) that related to CT practice that was more recent than the exposure period being studied. The first of these(25) was based on work conducted by one of us using a CT scanner installed in a major children’s hospital since 2004, with parameters that had minimally changed over time. It was, therefore, included in the Australian dose estimates to reflect practice that would have been occurring in the years leading up to 2005. The second study(26) reported a survey of paediatric practice in Australia from 13 different facilities. This was a significant recent snapshot of representative practice across a range of sites, complementing the earlier nationwide survey in the mid-1990s(27). We also surveyed the types of CT scanners installed in Australia during the time period of the study using licence registers maintained by state and territory regulators(14). Mandatory registration/licensing was not in place nationally for the entire study period. By 2005, the licensing data were more comprehensive and across the eight jurisdictions in Australia, a total of 1017 eligible CT scanners were identified as being installed between 1985 and 2005. User and operation manuals were obtained for the types of scanners installed in Australia. The market was dominated by three main manufacturers; GE Healthcare (Milwaukee, WI), Toshiba (Tokyo, Japan) since incorporated into Canon Medical Systems (Tochigi, Japan), and Siemens (Erlangen, Germany). Detailed protocol settings from the GE CT9800 Manual(30) (representing up to 45% of registered scanners in 1985–1995) were used to reconstruct CTDIvol values. We assumed that the protocols were used in clinical practice as specified in the manual and programmed on the scanner with little or no adjustment in the earliest years(9). From the review of past records, publications and unpublished sources documenting past CT use in Australia(23–30) (Table 3), we compiled a list of 61 separate CTDIvol estimates corresponding to at least 3,390 individual scans, cross-indexed by the region scanned, the age of the patient, and the year of the scan. We regressed the logarithm of the CTDIvol on scan region, year of age (0–19 years) and calendar year (1985–2005), to effectively smooth the data and fill-in the gaps for individual years where primary datapoints were missing and used the antilog of the regression estimate as the estimate of mean CTDIvol. Later data-points within a few years of the end of the exposure period in 2005 were also included to help stabilise the regression estimates. We considered whether to weight the regression model by the number of individual patient scans on which the input datapoints were based. We decided to use the unweighted estimates as the CTDIvol estimates based on clinical protocols represented an unknown number of scans compared with survey data based on a known number of scans (however, weighted and unweighted models were very consistent and robust to minor modifications of the input data). Based on work undertaken for the UK CT cohort, we decided to keep CTDIvol values constant prior to the year 1990 as there was little evidence of adjustment of scanning protocols in these early years(9). We found that the regression model was more stable if exposures prior to 1990 were coded together with exposures in 1990. From our regression model, CTDIvol values for 2520 strata covering the main anatomical regions (Table 1), each year of age of exposure (0–19 years) and each year in the study period (1985–2005) were obtained and cross-indexed with the detailed scan types within each main region. Due to a lack of Australian data, UK CTDIvol values were added for the small number of neck and whole-body scans(8). Organ dose calculation The NCICT program also required sex, kVp and scan start and end locations for input. We assumed that the CTDIvol values were the same for males and females. Although CTDIvol incorporates kVp, the kVp is still required as a separate input parameter for NCICT as the beam quality (determined by the kVp and beam filtration) defines the energy spectrum of the X-ray photons and therefore affects dose deposition in tissue. Most CTDIvol values used for the regression model were at 120 kVp. However, some kVp values were not known. Work undertaken for the UK CT cohort has also shown that 120 kVp is used in a majority of scans for both head and body(7,9). We used a fixed value of 120 kVp for input into NCICT. The scan start and end locations were determined from typical scan lengths used for the UK CT cohort(8) and verified by Australian clinical staff performing these types of CT scans over several decades. The anatomical ranges are shown in Figure 1. Table 1 includes information on how these scan ranges were then applied to the categories for this cohort, particularly in cases where the MBS item descriptors incorporated optional scan types (e.g. scan of the chest with or without scans of the upper abdomen). Fortunately, the MBS item descriptions were unambiguously defined for specific time periods and the patterns of use from those periods were used to inform the dosimetry for years in which the corresponding descriptors were ambiguously defined. Figure 1 Open in new tabDownload slide Anatomical ranges shown on the NCICT phantom representing a 5-year-old child. Figure 1 Open in new tabDownload slide Anatomical ranges shown on the NCICT phantom representing a 5-year-old child. Using the Australian input parameters, NCICT was used to generate an output dose matrix of absorbed doses in milligray (mGy) by scan type (for single-phase scans), age and year for 33 different tissues and organs (Table 4) and whole-body effective doses in millisieverts (mSv) calculated according to the ICRP Publication 103 definition(31). The absorbed dose to the (active) red bone marrow is challenging to determine because of changes in the distribution of red and (inactive) yellow bone marrow at different ages(32) and partial irradiation resulting from CT scans. At birth, all bone marrow is predominantly red marrow. With age and depending on the site of the bone, red marrow gradually transforms into yellow marrow with only about a third of marrow being active by adulthood(33). The modelling used in the NCICT phantoms is one of the most comprehensive and detailed available and based on up-to-date fluence-to-dose response functions reported by the ICRP(16,34,35). Table 4 List of tissues and organs included in the NCICT dosimetry system(16). NCICT Tissues and organs Brain, pituitary gland, lens, eyeballs, salivary glands, oral cavity, spinal cord, thyroid, oesophagus, trachea, thymus, lungs, breast, heart wall, stomach wall, liver, gall bladder, adrenals, spleen, pancreas, kidney, small intestine, colon, rectosigmoid, bladder, prostate, uterus, testes, ovaries, skin, muscle, active marrow, shallow marrow NCICT Tissues and organs Brain, pituitary gland, lens, eyeballs, salivary glands, oral cavity, spinal cord, thyroid, oesophagus, trachea, thymus, lungs, breast, heart wall, stomach wall, liver, gall bladder, adrenals, spleen, pancreas, kidney, small intestine, colon, rectosigmoid, bladder, prostate, uterus, testes, ovaries, skin, muscle, active marrow, shallow marrow Open in new tab Table 4 List of tissues and organs included in the NCICT dosimetry system(16). NCICT Tissues and organs Brain, pituitary gland, lens, eyeballs, salivary glands, oral cavity, spinal cord, thyroid, oesophagus, trachea, thymus, lungs, breast, heart wall, stomach wall, liver, gall bladder, adrenals, spleen, pancreas, kidney, small intestine, colon, rectosigmoid, bladder, prostate, uterus, testes, ovaries, skin, muscle, active marrow, shallow marrow NCICT Tissues and organs Brain, pituitary gland, lens, eyeballs, salivary glands, oral cavity, spinal cord, thyroid, oesophagus, trachea, thymus, lungs, breast, heart wall, stomach wall, liver, gall bladder, adrenals, spleen, pancreas, kidney, small intestine, colon, rectosigmoid, bladder, prostate, uterus, testes, ovaries, skin, muscle, active marrow, shallow marrow Open in new tab The output from NCICT gave us basic organ doses. These did not yet take into account scans that were billed under one MBS code (e.g. where the chest and brain were both scanned and billed as a single item) or multiple phases of the one scan type (e.g. a brain scan with and without contrast). Furthermore, some of the MBS descriptions included optional imaging (e.g. scan of the chest with or without scans of the upper abdomen), which also needed to be accounted for. Therefore, the estimated organ doses for an individual having a scan were calculated from the NCICT basic organ doses by (1) adding organ doses together for scans of multiple regions (e.g. chest plus brain); (2) applying weights for optional descriptor categories where the exact type of scan could not be ascertained (Table 1); and (3) applying the relevant phase factor(s) to account for multiphase scanning (Table 2). Collective doses were then calculated by summing across the cohort or relevant sub-cohort. In this methodological paper, we do not report cancer outcomes but simply report mean and collective doses by organ, by type of CT scan, by age and by calendar year. No lagging of exposure has been applied. RESULTS Almost 28 000 000 individual organ doses have been calculated for 902 031 CT scans performed on 692 879 Australians younger than 20 years of age in the years 1985–2005. The cohort and types of CT scans are described in Tables 5 and 6. More scans were performed on males (53%) and the vast majority of scans were of the head (70%). In this Medicare dataset, more scans were recorded in the oldest age group of 15–19 years (51%), with very few in the youngest infants under five years (6%). More scans (40%) were performed in the later years (2000–2005) than any of the earlier time periods (1985–1999). Table 5 Number of people undergoing medicare funded CT scans in Australia by sex and the number of CT scans per person in a cohort of people <20 years of age undergoing scans between 1985 and 2005, inclusive. Characteristic . Number (%) . Sex  Male 364 030 (53)  Female 328 849 (47) Number of CT scans per person  1 564 556 (81)  2 88 188 (13)  3 24 799 (4)  4 7470 (1)  ≥5 7866 (1) Total no. of people 692 879 (100) Characteristic . Number (%) . Sex  Male 364 030 (53)  Female 328 849 (47) Number of CT scans per person  1 564 556 (81)  2 88 188 (13)  3 24 799 (4)  4 7470 (1)  ≥5 7866 (1) Total no. of people 692 879 (100) Open in new tab Table 5 Number of people undergoing medicare funded CT scans in Australia by sex and the number of CT scans per person in a cohort of people <20 years of age undergoing scans between 1985 and 2005, inclusive. Characteristic . Number (%) . Sex  Male 364 030 (53)  Female 328 849 (47) Number of CT scans per person  1 564 556 (81)  2 88 188 (13)  3 24 799 (4)  4 7470 (1)  ≥5 7866 (1) Total no. of people 692 879 (100) Characteristic . Number (%) . Sex  Male 364 030 (53)  Female 328 849 (47) Number of CT scans per person  1 564 556 (81)  2 88 188 (13)  3 24 799 (4)  4 7470 (1)  ≥5 7866 (1) Total no. of people 692 879 (100) Open in new tab Table 6 Number of medicare funded CT scans in Australia by anatomical region, year the CT scan was performed and age at CT scan in a cohort of people <20 years of age undergoing scans between 1985 and 2005, inclusive. Characteristic . Number (%) . Anatomical region  Head 628 093 (70)  Neck 8654 (1)  Spine 73 482 (8)  Chest 21 964 (2)  Abdomen and/or pelvis 49 564 (5)  Combined 7208 (1)  Extremities 86 284 (10)  Other 26 782 (3) Year of CT scan  1985–1989 115 702 (13)  1990–1994 173 646 (19)  1995–1999 246 874 (27)  2000–2005 365 809 (41) Age at CT scan (years)  0–4 57 549 (6)  5–9 128 861 (14)  10-14 258 081 (29)  15–19 457 540 (51) Total number of CT scans 902 031 (100) Characteristic . Number (%) . Anatomical region  Head 628 093 (70)  Neck 8654 (1)  Spine 73 482 (8)  Chest 21 964 (2)  Abdomen and/or pelvis 49 564 (5)  Combined 7208 (1)  Extremities 86 284 (10)  Other 26 782 (3) Year of CT scan  1985–1989 115 702 (13)  1990–1994 173 646 (19)  1995–1999 246 874 (27)  2000–2005 365 809 (41) Age at CT scan (years)  0–4 57 549 (6)  5–9 128 861 (14)  10-14 258 081 (29)  15–19 457 540 (51) Total number of CT scans 902 031 (100) Open in new tab Table 6 Number of medicare funded CT scans in Australia by anatomical region, year the CT scan was performed and age at CT scan in a cohort of people <20 years of age undergoing scans between 1985 and 2005, inclusive. Characteristic . Number (%) . Anatomical region  Head 628 093 (70)  Neck 8654 (1)  Spine 73 482 (8)  Chest 21 964 (2)  Abdomen and/or pelvis 49 564 (5)  Combined 7208 (1)  Extremities 86 284 (10)  Other 26 782 (3) Year of CT scan  1985–1989 115 702 (13)  1990–1994 173 646 (19)  1995–1999 246 874 (27)  2000–2005 365 809 (41) Age at CT scan (years)  0–4 57 549 (6)  5–9 128 861 (14)  10-14 258 081 (29)  15–19 457 540 (51) Total number of CT scans 902 031 (100) Characteristic . Number (%) . Anatomical region  Head 628 093 (70)  Neck 8654 (1)  Spine 73 482 (8)  Chest 21 964 (2)  Abdomen and/or pelvis 49 564 (5)  Combined 7208 (1)  Extremities 86 284 (10)  Other 26 782 (3) Year of CT scan  1985–1989 115 702 (13)  1990–1994 173 646 (19)  1995–1999 246 874 (27)  2000–2005 365 809 (41) Age at CT scan (years)  0–4 57 549 (6)  5–9 128 861 (14)  10-14 258 081 (29)  15–19 457 540 (51) Total number of CT scans 902 031 (100) Open in new tab The main inputs for the dose calculation are the CTDIvol values, summarised in Table 7. Time-related changes in CTDIvol values for head, chest and abdomen/pelvis scans are shown in Figure 2. The CTDIvol values are held constant until 1990 (see Methods section). Scan doses increase in direct proportion to the number of phases per scan. The mean number of phases per scan for different anatomical regions are summarised in Table 8. The number of phases per CT head scan has decreased over time, while for body imaging the average number of phases per scan is 1.2 for the chest and 1.5 for the abdomen and/or pelvis. Scans of the spine and extremities are routinely performed without contrast. Table 7 Mean volumetric CT dose index (CTDIvol) for different anatomical regions, age groups and selected years in terms of the 16 cm reference phantom for single-phase scans. Anatomical region . Age (years) . Mean CTDIvol (mGy) Year of CT scan . 1985 . 1995 . 2005 . Head All ages 60 49 33 0–4 40 33 22 5–9 52 42 28 10–14 66 54 36 15–19 84 68 46 Partial head All ages 42 34 23 0–4 28 23 15 5–9 36 29 19 10–14 45 37 25 15–19 58 47 32 Neck All ages 26 18 9 0–4 22 15 8 5–9 24 17 9 10–14 27 19 10 15–19 29 22 11 Spine All ages 49 40 27 0–4 33 27 18 5–9 42 34 23 10–14 54 44 29 15–19 68 56 37 Chest All ages 25 21 14 0–4 17 14 9 5–9 22 18 12 10–14 28 23 15 15–19 35 29 19 Abdomen and/or Pelvis All ages 35 29 19 0–4 24 19 13 5–9 30 25 16 10–14 38 31 21 15–19 49 40 27 Extremities All ages 29 24 16 0–4 19 16 11 5–9 25 20 13 10–14 32 26 17 15–19 40 33 22 Anatomical region . Age (years) . Mean CTDIvol (mGy) Year of CT scan . 1985 . 1995 . 2005 . Head All ages 60 49 33 0–4 40 33 22 5–9 52 42 28 10–14 66 54 36 15–19 84 68 46 Partial head All ages 42 34 23 0–4 28 23 15 5–9 36 29 19 10–14 45 37 25 15–19 58 47 32 Neck All ages 26 18 9 0–4 22 15 8 5–9 24 17 9 10–14 27 19 10 15–19 29 22 11 Spine All ages 49 40 27 0–4 33 27 18 5–9 42 34 23 10–14 54 44 29 15–19 68 56 37 Chest All ages 25 21 14 0–4 17 14 9 5–9 22 18 12 10–14 28 23 15 15–19 35 29 19 Abdomen and/or Pelvis All ages 35 29 19 0–4 24 19 13 5–9 30 25 16 10–14 38 31 21 15–19 49 40 27 Extremities All ages 29 24 16 0–4 19 16 11 5–9 25 20 13 10–14 32 26 17 15–19 40 33 22 Open in new tab Table 7 Mean volumetric CT dose index (CTDIvol) for different anatomical regions, age groups and selected years in terms of the 16 cm reference phantom for single-phase scans. Anatomical region . Age (years) . Mean CTDIvol (mGy) Year of CT scan . 1985 . 1995 . 2005 . Head All ages 60 49 33 0–4 40 33 22 5–9 52 42 28 10–14 66 54 36 15–19 84 68 46 Partial head All ages 42 34 23 0–4 28 23 15 5–9 36 29 19 10–14 45 37 25 15–19 58 47 32 Neck All ages 26 18 9 0–4 22 15 8 5–9 24 17 9 10–14 27 19 10 15–19 29 22 11 Spine All ages 49 40 27 0–4 33 27 18 5–9 42 34 23 10–14 54 44 29 15–19 68 56 37 Chest All ages 25 21 14 0–4 17 14 9 5–9 22 18 12 10–14 28 23 15 15–19 35 29 19 Abdomen and/or Pelvis All ages 35 29 19 0–4 24 19 13 5–9 30 25 16 10–14 38 31 21 15–19 49 40 27 Extremities All ages 29 24 16 0–4 19 16 11 5–9 25 20 13 10–14 32 26 17 15–19 40 33 22 Anatomical region . Age (years) . Mean CTDIvol (mGy) Year of CT scan . 1985 . 1995 . 2005 . Head All ages 60 49 33 0–4 40 33 22 5–9 52 42 28 10–14 66 54 36 15–19 84 68 46 Partial head All ages 42 34 23 0–4 28 23 15 5–9 36 29 19 10–14 45 37 25 15–19 58 47 32 Neck All ages 26 18 9 0–4 22 15 8 5–9 24 17 9 10–14 27 19 10 15–19 29 22 11 Spine All ages 49 40 27 0–4 33 27 18 5–9 42 34 23 10–14 54 44 29 15–19 68 56 37 Chest All ages 25 21 14 0–4 17 14 9 5–9 22 18 12 10–14 28 23 15 15–19 35 29 19 Abdomen and/or Pelvis All ages 35 29 19 0–4 24 19 13 5–9 30 25 16 10–14 38 31 21 15–19 49 40 27 Extremities All ages 29 24 16 0–4 19 16 11 5–9 25 20 13 10–14 32 26 17 15–19 40 33 22 Open in new tab Figure 2 Open in new tabDownload slide Temporal changes by age group for CTDIvol in terms of the 16 cm reference phantom for (a) CT brain scans, (b) CT chest scans and (c) CT abdomen/pelvis scans (all scans are single phase). Figure 2 Open in new tabDownload slide Temporal changes by age group for CTDIvol in terms of the 16 cm reference phantom for (a) CT brain scans, (b) CT chest scans and (c) CT abdomen/pelvis scans (all scans are single phase). Table 8 Mean number of phases per scan for the Australian CT cohort for two time periods. Anatomical region . Mean number of phases per scan . 1985–1995 . 1996–2005 . Head 1.5 1.2 Neck 1.3 1.4 Spine 1.0 1.0 Chest 1.2 1.2 Abdomen and/or pelvis 1.5 1.5 Combined 1.4 1.4 Extremities 1.0 1.0 Anatomical region . Mean number of phases per scan . 1985–1995 . 1996–2005 . Head 1.5 1.2 Neck 1.3 1.4 Spine 1.0 1.0 Chest 1.2 1.2 Abdomen and/or pelvis 1.5 1.5 Combined 1.4 1.4 Extremities 1.0 1.0 Open in new tab Table 8 Mean number of phases per scan for the Australian CT cohort for two time periods. Anatomical region . Mean number of phases per scan . 1985–1995 . 1996–2005 . Head 1.5 1.2 Neck 1.3 1.4 Spine 1.0 1.0 Chest 1.2 1.2 Abdomen and/or pelvis 1.5 1.5 Combined 1.4 1.4 Extremities 1.0 1.0 Anatomical region . Mean number of phases per scan . 1985–1995 . 1996–2005 . Head 1.5 1.2 Neck 1.3 1.4 Spine 1.0 1.0 Chest 1.2 1.2 Abdomen and/or pelvis 1.5 1.5 Combined 1.4 1.4 Extremities 1.0 1.0 Open in new tab Organ absorbed doses averaged over the entire cohort (both sexes, all ages and all years) are shown in Table 9 for different types of CT scans. As expected, CT scans of the head lead to the highest mean organ absorbed dose for the brain (47 mGy), CT chest scans resulted in 17 mGy mean dose to the breast and 19 mGy to the lungs. The mean dose to the red bone marrow was highest after an abdomen and/or pelvis scan (9.2 mGy), followed by a scan of the head (7.4 mGy). Organ absorbed doses were higher when CT scans were performed in the earlier years of the study period for all age groups. This can be seen in Table 10 that provides the mean organ dose by age group and the year the scan was performed, but for single-phase scans only (see the Supplementary Tables for minimum and maximum values also). For a child aged in the range 5–9 years, a single-phase of a CT brain scan performed in 1985–1990 resulted in a brain dose of 42 mGy on average and this reduced to 25 mGy by the time these scans were performed between 2000 and 2005. The range of doses (including all phases) across different age groups is narrower in more recent years. For example, the dose to the brain after a CT head scan (including both full and partial brain scans) ranged from 52 to 127 mGy in 1985, compared with 16 to 32 mGy in 2005 (Figure 3). A detailed breakdown of red bone marrow doses after different types of CT scan types is shown in Figure 4 and by age group. Table 9 Mean organ and tissue absorbed doses for different types of CT scans (including multiple phases), averaged over both sexes, all age groups and all years for the Australian CT cohort. Anatomical region of CT scan performed . Mean absorbed dose (mGy) . Brain . Breast . Lungs . Liver . RBM . Head 47 <1 <1 <1 7.4 Neck 3.2 <1 1.3 <1 1.8 Spine 1.7 2.5 3.9 9.9 5.8 Chest <1 17 19 12 4.4 Abdomen and/or pelvis <1 9.4 5.8 24 9.2 Combined 4.2 34 30 42 17 Extremities <1 <1 <1 <1 <1 Other 2.0 8.2 7.4 15 6.0 Anatomical region of CT scan performed . Mean absorbed dose (mGy) . Brain . Breast . Lungs . Liver . RBM . Head 47 <1 <1 <1 7.4 Neck 3.2 <1 1.3 <1 1.8 Spine 1.7 2.5 3.9 9.9 5.8 Chest <1 17 19 12 4.4 Abdomen and/or pelvis <1 9.4 5.8 24 9.2 Combined 4.2 34 30 42 17 Extremities <1 <1 <1 <1 <1 Other 2.0 8.2 7.4 15 6.0 Open in new tab Table 9 Mean organ and tissue absorbed doses for different types of CT scans (including multiple phases), averaged over both sexes, all age groups and all years for the Australian CT cohort. Anatomical region of CT scan performed . Mean absorbed dose (mGy) . Brain . Breast . Lungs . Liver . RBM . Head 47 <1 <1 <1 7.4 Neck 3.2 <1 1.3 <1 1.8 Spine 1.7 2.5 3.9 9.9 5.8 Chest <1 17 19 12 4.4 Abdomen and/or pelvis <1 9.4 5.8 24 9.2 Combined 4.2 34 30 42 17 Extremities <1 <1 <1 <1 <1 Other 2.0 8.2 7.4 15 6.0 Anatomical region of CT scan performed . Mean absorbed dose (mGy) . Brain . Breast . Lungs . Liver . RBM . Head 47 <1 <1 <1 7.4 Neck 3.2 <1 1.3 <1 1.8 Spine 1.7 2.5 3.9 9.9 5.8 Chest <1 17 19 12 4.4 Abdomen and/or pelvis <1 9.4 5.8 24 9.2 Combined 4.2 34 30 42 17 Extremities <1 <1 <1 <1 <1 Other 2.0 8.2 7.4 15 6.0 Open in new tab Table 10 Mean absorbed dose to the brain and red bone marrow for CT single-phase scans of the brain by time periods and age groups (see also Supplementary Table 1). Age (years) . Year of CT scan . Mean absorbed dose (mGy) . Brain . Red bone marrow . 0–4 1985–1989 36 13 1990–1994 33 12 1995–1999 27 9 2000–2005 22 8 5–9 1985–1989 42 11 1990–1994 38 11 1995–1999 31 9 2000–2005 25 7 10–14 1985–1989 51 8 1990–1994 47 8 1995–1999 39 6 2000–2005 31 5 15–19 1985–1989 63 6 1990–1994 58 6 1995–1999 47 5 2000–2005 38 4 All ages 1985–1989 48 10 1990–1994 44 9 1995–1999 36 7 2000–2005 29 6 Age (years) . Year of CT scan . Mean absorbed dose (mGy) . Brain . Red bone marrow . 0–4 1985–1989 36 13 1990–1994 33 12 1995–1999 27 9 2000–2005 22 8 5–9 1985–1989 42 11 1990–1994 38 11 1995–1999 31 9 2000–2005 25 7 10–14 1985–1989 51 8 1990–1994 47 8 1995–1999 39 6 2000–2005 31 5 15–19 1985–1989 63 6 1990–1994 58 6 1995–1999 47 5 2000–2005 38 4 All ages 1985–1989 48 10 1990–1994 44 9 1995–1999 36 7 2000–2005 29 6 Open in new tab Table 10 Mean absorbed dose to the brain and red bone marrow for CT single-phase scans of the brain by time periods and age groups (see also Supplementary Table 1). Age (years) . Year of CT scan . Mean absorbed dose (mGy) . Brain . Red bone marrow . 0–4 1985–1989 36 13 1990–1994 33 12 1995–1999 27 9 2000–2005 22 8 5–9 1985–1989 42 11 1990–1994 38 11 1995–1999 31 9 2000–2005 25 7 10–14 1985–1989 51 8 1990–1994 47 8 1995–1999 39 6 2000–2005 31 5 15–19 1985–1989 63 6 1990–1994 58 6 1995–1999 47 5 2000–2005 38 4 All ages 1985–1989 48 10 1990–1994 44 9 1995–1999 36 7 2000–2005 29 6 Age (years) . Year of CT scan . Mean absorbed dose (mGy) . Brain . Red bone marrow . 0–4 1985–1989 36 13 1990–1994 33 12 1995–1999 27 9 2000–2005 22 8 5–9 1985–1989 42 11 1990–1994 38 11 1995–1999 31 9 2000–2005 25 7 10–14 1985–1989 51 8 1990–1994 47 8 1995–1999 39 6 2000–2005 31 5 15–19 1985–1989 63 6 1990–1994 58 6 1995–1999 47 5 2000–2005 38 4 All ages 1985–1989 48 10 1990–1994 44 9 1995–1999 36 7 2000–2005 29 6 Open in new tab Figure 3 Open in new tabDownload slide Mean brain absorbed dose after a CT head scan (including brain and partial brain scans) by sex and age at exposure for two different years (1985 and 2005). Figure 3 Open in new tabDownload slide Mean brain absorbed dose after a CT head scan (including brain and partial brain scans) by sex and age at exposure for two different years (1985 and 2005). Figure 4 Open in new tabDownload slide Range of red bone marrow absorbed doses after CT scans of different anatomical regions by age group, averaged over both sexes and all years. Values >1.5 times the interquartile-range are classified as outliers and have been excluded. Figure 4 Open in new tabDownload slide Range of red bone marrow absorbed doses after CT scans of different anatomical regions by age group, averaged over both sexes and all years. Values >1.5 times the interquartile-range are classified as outliers and have been excluded. The collective brain absorbed dose after CT head scans across the entire cohort was 29 784 Gy and the collective red bone marrow dose for all CT scans was 5933 Gy (Table 11). Effective dose is not a quantity applicable to individual risk but is used here to describe the cohort dose. The collective effective dose across all types of CT scans was 3673 Sv, with a mean effective dose of 4.1 mSv per scan and 5.3 mSv per person. Table 12 illustrates the frequency of multiple scans in individual patients. For example, 19 502 males had one head scan in the age range 0–4 years compared with 14 083 females. The total person-years of exposure for each category provides a denominator, allowing approximate rates of multiple exposures to be calculated as required. Multiple scans of the head were more likely to be performed for a single individual than of any other anatomical area for children aged under 15 years. In the oldest age bracket, 15–19 years, when five or more scans were performed in an individual, they were more likely to be of a body area other than the head. Table 11 Summary of collective effective and absorbed doses with comparison to our earlier study(3) on cancer outcomes and dose-response. Summary of collective dosesa . . This study . Mathews et al.b,(3) . All CT scans  Collective effective dose (Sv) 3673 3900  Mean effective dose per CT scan (mSv) 4.1 4.5 Mean effective dose per person (mSv) 5.3 5.7 CT head scansc  Collective brain absorbed dose (Gy) 29 784 19 800  Mean brain absorbed dose per CT scan (mGy) 33 40  Mean brain absorbed dose per person (mGy) 43 49 All CT scans  Collective red bone marrow absorbed dose (Gy) 5933 4000  Mean red bone marrow absorbed dose per CT scan (mGy) 6.6 4.6  Mean red bone marrow absorbed dose per person (mGy) 8.6 5.9 Summary of collective dosesa . . This study . Mathews et al.b,(3) . All CT scans  Collective effective dose (Sv) 3673 3900  Mean effective dose per CT scan (mSv) 4.1 4.5 Mean effective dose per person (mSv) 5.3 5.7 CT head scansc  Collective brain absorbed dose (Gy) 29 784 19 800  Mean brain absorbed dose per CT scan (mGy) 33 40  Mean brain absorbed dose per person (mGy) 43 49 All CT scans  Collective red bone marrow absorbed dose (Gy) 5933 4000  Mean red bone marrow absorbed dose per CT scan (mGy) 6.6 4.6  Mean red bone marrow absorbed dose per person (mGy) 8.6 5.9 aMultiphase scanning included in doses. bThese values include a 1 year lag, where doses within the first year after a CT scan are excluded. cThese CT head scans include all scans for the head (see Table 1). Open in new tab Table 11 Summary of collective effective and absorbed doses with comparison to our earlier study(3) on cancer outcomes and dose-response. Summary of collective dosesa . . This study . Mathews et al.b,(3) . All CT scans  Collective effective dose (Sv) 3673 3900  Mean effective dose per CT scan (mSv) 4.1 4.5 Mean effective dose per person (mSv) 5.3 5.7 CT head scansc  Collective brain absorbed dose (Gy) 29 784 19 800  Mean brain absorbed dose per CT scan (mGy) 33 40  Mean brain absorbed dose per person (mGy) 43 49 All CT scans  Collective red bone marrow absorbed dose (Gy) 5933 4000  Mean red bone marrow absorbed dose per CT scan (mGy) 6.6 4.6  Mean red bone marrow absorbed dose per person (mGy) 8.6 5.9 Summary of collective dosesa . . This study . Mathews et al.b,(3) . All CT scans  Collective effective dose (Sv) 3673 3900  Mean effective dose per CT scan (mSv) 4.1 4.5 Mean effective dose per person (mSv) 5.3 5.7 CT head scansc  Collective brain absorbed dose (Gy) 29 784 19 800  Mean brain absorbed dose per CT scan (mGy) 33 40  Mean brain absorbed dose per person (mGy) 43 49 All CT scans  Collective red bone marrow absorbed dose (Gy) 5933 4000  Mean red bone marrow absorbed dose per CT scan (mGy) 6.6 4.6  Mean red bone marrow absorbed dose per person (mGy) 8.6 5.9 aMultiphase scanning included in doses. bThese values include a 1 year lag, where doses within the first year after a CT scan are excluded. cThese CT head scans include all scans for the head (see Table 1). Open in new tab Table 12 Scan frequency per person by anatomical region of the scan, age group and sex. Age (years) . Number of CT scans . Male . Female . CT head scan . All Other CT scan . CT head scan . All other CT scan . 0–4 1 19 502 3052 14 083 2623 2 1612 308 1127 308 3 682 139 481 110 4 199 61 172 55 ≥5 309 99 242 90 Total person-yearsa 14 962 187 14 176 104 5–9 1 50 339 8136 35 975 6780 2 3659 719 2586 539 3 935 235 609 179 4 190 90 138 54 ≥5 277 120 177 118 Total person-yearsa 18 620 860 17 842 144 10–14 1 81 878 28 023 66 237 24 743 2 6913 2718 5415 2179 3 1497 753 1171 583 4 337 220 245 165 ≥5 325 242 224 207 Total person-yearsa 19 770 630 19 144 834 15–19 1 102 608 63 920 118 369 55 814 2 9275 7517 10 780 5546 3 2318 2283 2576 1669 4 502 735 497 430 ≥5 432 826 396 567 Total person-yearsa 20 927 603 20 556 356 Age (years) . Number of CT scans . Male . Female . CT head scan . All Other CT scan . CT head scan . All other CT scan . 0–4 1 19 502 3052 14 083 2623 2 1612 308 1127 308 3 682 139 481 110 4 199 61 172 55 ≥5 309 99 242 90 Total person-yearsa 14 962 187 14 176 104 5–9 1 50 339 8136 35 975 6780 2 3659 719 2586 539 3 935 235 609 179 4 190 90 138 54 ≥5 277 120 177 118 Total person-yearsa 18 620 860 17 842 144 10–14 1 81 878 28 023 66 237 24 743 2 6913 2718 5415 2179 3 1497 753 1171 583 4 337 220 245 165 ≥5 325 242 224 207 Total person-yearsa 19 770 630 19 144 834 15–19 1 102 608 63 920 118 369 55 814 2 9275 7517 10 780 5546 3 2318 2283 2576 1669 4 502 735 497 430 ≥5 432 826 396 567 Total person-yearsa 20 927 603 20 556 356 aPerson-years are included so that the incidence of scans at different ages can be compared using a suitable denominator. Open in new tab Table 12 Scan frequency per person by anatomical region of the scan, age group and sex. Age (years) . Number of CT scans . Male . Female . CT head scan . All Other CT scan . CT head scan . All other CT scan . 0–4 1 19 502 3052 14 083 2623 2 1612 308 1127 308 3 682 139 481 110 4 199 61 172 55 ≥5 309 99 242 90 Total person-yearsa 14 962 187 14 176 104 5–9 1 50 339 8136 35 975 6780 2 3659 719 2586 539 3 935 235 609 179 4 190 90 138 54 ≥5 277 120 177 118 Total person-yearsa 18 620 860 17 842 144 10–14 1 81 878 28 023 66 237 24 743 2 6913 2718 5415 2179 3 1497 753 1171 583 4 337 220 245 165 ≥5 325 242 224 207 Total person-yearsa 19 770 630 19 144 834 15–19 1 102 608 63 920 118 369 55 814 2 9275 7517 10 780 5546 3 2318 2283 2576 1669 4 502 735 497 430 ≥5 432 826 396 567 Total person-yearsa 20 927 603 20 556 356 Age (years) . Number of CT scans . Male . Female . CT head scan . All Other CT scan . CT head scan . All other CT scan . 0–4 1 19 502 3052 14 083 2623 2 1612 308 1127 308 3 682 139 481 110 4 199 61 172 55 ≥5 309 99 242 90 Total person-yearsa 14 962 187 14 176 104 5–9 1 50 339 8136 35 975 6780 2 3659 719 2586 539 3 935 235 609 179 4 190 90 138 54 ≥5 277 120 177 118 Total person-yearsa 18 620 860 17 842 144 10–14 1 81 878 28 023 66 237 24 743 2 6913 2718 5415 2179 3 1497 753 1171 583 4 337 220 245 165 ≥5 325 242 224 207 Total person-yearsa 19 770 630 19 144 834 15–19 1 102 608 63 920 118 369 55 814 2 9275 7517 10 780 5546 3 2318 2283 2576 1669 4 502 735 497 430 ≥5 432 826 396 567 Total person-yearsa 20 927 603 20 556 356 aPerson-years are included so that the incidence of scans at different ages can be compared using a suitable denominator. Open in new tab DISCUSSION We have undertaken comprehensive, retrospective, individual organ dosimetry for one of the largest cohorts of CT-exposed persons available internationally. As individual CT scan parameters were not available, this involved reconstructing CTDIvol values from surveys and protocols to reflect historical CT usage and using the NCICT algorithms to estimate absorbed organ doses. We found that doses decreased over time and that the range of doses across age groups became narrower by 2005. This is consistent with greater awareness of radiation risks and efforts to reduce dose, particularly in young people(36). Retrospective dosimetry for the UK CT cohort has similarly used CTDIvol values and the NCICT program to estimate individual organ doses(2,8,9). In their study, CTDIvol values were derived from nationwide surveys of practice(2,8) and retrieved from CT films(9). Table 13 provides a comparison between the Australian and UK values and demonstrates a generally broad agreement. We also compared values with Australian national diagnostic reference levels (DRLs)(26). The Australian national survey represents data on paediatric practice collected between 2009 and 2011, slightly after the end of our study period. National DRLs are set at the 75th percentile of site median values. Our mean CTDIvol for CT head scans in 2005 was smaller by 27% and larger by 17% for 0–4 years and 5–14 years, respectively, than those for national DRLs (CTDIvol DRL for 0–4 years is 30 mGy and 5–14 years is 35 mGy). Our mean CTDIvol for CT chest scans in 2005 was greater by 125% and by 70% for 0–4 years and 5–14 years, respectively, than those for national DRLs (CTDIvol DRL for 0–4 years is 4 mGy and 5–14 years is 10 mGy after applying a conversion to express in terms of the 16 cm reference phantom rather than the 32 cm phantom). For abdomen and/or pelvis scans, our mean CTDIvol in 2005 was smaller by 7% and larger by 20% for 0–4 years and 5–14 years, respectively, than those for national DRLs (CTDIvol DRL for 0–4 years is 14 mGy and 5–14 years is 20 mGy after applying a conversion to express in terms of the 16 cm reference phantom rather than the 32 cm phantom). Interestingly, for those age and scan categories where our values exceeded the national DRLs, in some instances considerably (e.g. chest scans), it was found that the Australian national DRLs were less than any other international DRL set for the same quantity(26). This suggests that the DRLs are based on very well optimised scans. The type of facilities contributing data to the survey was not stated, although it is expected that these were dedicated paediatric practices in children’s hospitals and most likely would not represent the private imaging practices that offered the Medicare funded scans that our cohort have received. In Australia, infants are more likely to present to a hospital where CT services are funded by state governments, which means that many CT services in state-funded paediatric hospitals are not captured by Medicare. A further comparison (Table 14) was made with the work undertaken for the UK cohort(7) where they accounted for shared and unshared uncertainties in the UK dosimetry caused by individual parameter settings by using retrieved CT films(9). We found that the median brain absorbed dose from our Aust-PERC study was lower by 12% before 1990, the same between 1990–1999 and ranged from 8% lower to 31% higher from the year 2000 when categorising by age group. This is due to the different CTDIvol values used to determine the organ doses. We incorporated a dose reduction from 1990 into our CTDIvol values on the basis of the results of the Lee et al.(9) work and the Australian sources of information (Table 3) and hence our results similarly reflect a reduction in dose between the time periods. This is not reflected in the original CT dosimetry for the UK cohort(8), which was based on nationwide surveys and this incremental change was not captured. Table 13 Comparison of Australian mean volumetric CT dose index (CTDIvol) values for CT brain scans (full brain, single phase only) with those from the UK CT cohort study(8,9). Year of CT scan . Age (years) . CT brain scan . This study . Kim et al. (2012)(8) . Lee et al (2016)(9) . <1990 All ages 60 51a 71 1990–1999 All ages 51 57 ≥2000 0–4 24 25b 33 5–9 31 34c 39 10–14 40 44d 47 15–19 51 56e 43 Year of CT scan . Age (years) . CT brain scan . This study . Kim et al. (2012)(8) . Lee et al (2016)(9) . <1990 All ages 60 51a 71 1990–1999 All ages 51 57 ≥2000 0–4 24 25b 33 5–9 31 34c 39 10–14 40 44d 47 15–19 51 56e 43 All CTDIvol values are in terms of the 16 cm dosimetry phantom. aValue is for scans pre-2001. bValue is for 0–1 years. cValue is for 5 years. dValue is for 10 years. eValue is for 15 years to adult. Open in new tab Table 13 Comparison of Australian mean volumetric CT dose index (CTDIvol) values for CT brain scans (full brain, single phase only) with those from the UK CT cohort study(8,9). Year of CT scan . Age (years) . CT brain scan . This study . Kim et al. (2012)(8) . Lee et al (2016)(9) . <1990 All ages 60 51a 71 1990–1999 All ages 51 57 ≥2000 0–4 24 25b 33 5–9 31 34c 39 10–14 40 44d 47 15–19 51 56e 43 Year of CT scan . Age (years) . CT brain scan . This study . Kim et al. (2012)(8) . Lee et al (2016)(9) . <1990 All ages 60 51a 71 1990–1999 All ages 51 57 ≥2000 0–4 24 25b 33 5–9 31 34c 39 10–14 40 44d 47 15–19 51 56e 43 All CTDIvol values are in terms of the 16 cm dosimetry phantom. aValue is for scans pre-2001. bValue is for 0–1 years. cValue is for 5 years. dValue is for 10 years. eValue is for 15 years to adult. Open in new tab Table 14 Absorbed dose to the brain after a CT brain scan (full brain, single phase only) by time period and age group. Year of CT Scan . Age (years) . Brain absorbed dose (mGy) . Lee et al.(7) including . uncertainties . This study . Median . 2.5th . 97.5th . Median . 2.5th . 97.5th . <1990 All ages 52 15 173 46 34 67 1990–1999 All ages 38 11 132 38 25 61 ≥2000 0–4 24 7 80 22 19 25 5–9 24 7 80 25 21 30 10–14 26 8 87 31 26 36 15–19 29 9 98 38 32 44 Year of CT Scan . Age (years) . Brain absorbed dose (mGy) . Lee et al.(7) including . uncertainties . This study . Median . 2.5th . 97.5th . Median . 2.5th . 97.5th . <1990 All ages 52 15 173 46 34 67 1990–1999 All ages 38 11 132 38 25 61 ≥2000 0–4 24 7 80 22 19 25 5–9 24 7 80 25 21 30 10–14 26 8 87 31 26 36 15–19 29 9 98 38 32 44 The Lee et al.(7) values are from 500 dose realisations simulating uncertainties in values directly extracted from CT films in the relevant time period. Open in new tab Table 14 Absorbed dose to the brain after a CT brain scan (full brain, single phase only) by time period and age group. Year of CT Scan . Age (years) . Brain absorbed dose (mGy) . Lee et al.(7) including . uncertainties . This study . Median . 2.5th . 97.5th . Median . 2.5th . 97.5th . <1990 All ages 52 15 173 46 34 67 1990–1999 All ages 38 11 132 38 25 61 ≥2000 0–4 24 7 80 22 19 25 5–9 24 7 80 25 21 30 10–14 26 8 87 31 26 36 15–19 29 9 98 38 32 44 Year of CT Scan . Age (years) . Brain absorbed dose (mGy) . Lee et al.(7) including . uncertainties . This study . Median . 2.5th . 97.5th . Median . 2.5th . 97.5th . <1990 All ages 52 15 173 46 34 67 1990–1999 All ages 38 11 132 38 25 61 ≥2000 0–4 24 7 80 22 19 25 5–9 24 7 80 25 21 30 10–14 26 8 87 31 26 36 15–19 29 9 98 38 32 44 The Lee et al.(7) values are from 500 dose realisations simulating uncertainties in values directly extracted from CT films in the relevant time period. Open in new tab In our earlier publication(3) on cancer outcomes from the Australian CT cohort, we used either the number of CT scans as a surrogate for dose or the effective dose to assess indicative dose-responses. Organ absorbed doses were only reported for the brain and red bone marrow based on previously published data(8,25). Now we have performed detailed individual organ dosimetry and a comparison is made with the values used in our earlier study and is included in the collective doses summarised in Table 11. There is excellent agreement between the effective doses per scan used in the earlier study with those now calculated from the individual organ doses. Overall, our individual organ dosimetry shows that the mean absorbed dose to the brain is lower than previously estimated and the red bone marrow dose is slightly higher. For the youngest children, the greatest percentage of active marrow is in the cranium(32). As a result, the red bone marrow dose after a scan of the head is highest in the youngest age range (0–4 years). The os coxae (pelvic bone), sacral and lumbar spine have a greater proportion of red bone marrow at older ages. This is observed in higher doses to the red bone marrow in older individuals after CT scans of these regions (Figure 4). More detailed analyses would be required to describe how multiple exposures are related to factors, such as cancer diagnosis. For example, we already know that, as CT scans are often used for the monitoring or management of cancer, the frequency of multiple scans tends to increase in the period following such a diagnosis. In documenting dose-response relationships between exposure and cancer, multiple exposures are best dealt with by cumulating individual doses over time and by invoking exclusion or lag periods, and other measures to adjust for biases arising from reverse causation(37) or confounding(38). It is acknowledged that there are significant limitations in undertaking retrospective dosimetry where the actual scan parameters and characteristics of the individual are not available. In our dose estimates, every child of the same age and sex with the same scan in the same year will be allocated the same organ doses. Due to the lack of specificity in the data associated with the cohort, assumptions had to be made to facilitate the organ dosimetry. Furthermore, the scan descriptions used for Medicare funding were not always clear and some ambiguities contributed to uncertainty in dose estimation. Information on the number of phases was not always available and in the later years, we made assumptions on the number of phases performed in a service based on earlier usage patterns. We assumed that the age of the patient correlated with their size and consequently the ICRP reference phantoms used in NCICT to simulate dose deposition. The phantom organ size, depth and location relative to the scan range were considered to be representative for the cohort we were investigating, while limited in their ability to closely simulate any one individual. We also assumed that the CTDIvol values (i.e. the scan parameters) were set the same for males and females and that all scans were performed at 120 kVp. The scan start and end locations were fixed on the NCICT reference phantoms and we did not allow for variation. In particular, small organs on the edge of the scan range (e.g. thyroid for a chest scan or testes for an abdomen/pelvis scan) may have large dose variations depending on whether they are fully within the scan volume, partially exposed or situated outside of it. Several authors(16,39–40) have discussed the considerable variations possible when undertaking paediatric dosimetry with computational software solutions. In a later paper, we will assess the uncertainty in dose estimates arising from the assumptions made and from other sources of measurement error in deriving these dose estimates. The dose reconstructions undertaken here lead to Berkson type errors(41), while other sources of variation (e.g. missed doses) lead to classical errors. In future work, we will examine effects of both classical and Berkson type errors. Despite the limitations of retrospective dosimetry and the methodological differences between Australia and the UK approaches, there is surprisingly good agreement between the mean dose estimates for the UK and Australian cohorts. CONCLUSION We have used a multi-tiered approach to reconstruct individual organ doses for a large cohort of young people exposed to CT scans in Australia between 1985 and 2005. We started with the detailed billing descriptions and reconstructed the CTDIvol values using: (1) scanner-specific protocols from the review of scientific literature relevant to past Australian CT usage; (2) Australian regulator databases of CT scanners registered for the relevant years; and (3) manufacturer manuals to obtain protocol parameters. Additionally, data were retrieved from national, state and local surveys and through expert interviews. Unweighted regression modelling was then used to predict a CTDIvol matrix for input to the NCICT program. The doses provided here are the first summary of Australian CT paediatric organ doses over a significant proportion of CT usage in this country. Although limited by uncertainties, this dosimetry allows more detailed assessment of the radiation-induced cancer risk and dose-response in this important cohort for the Aust-PERC study. In later publications, we will assess how the errors in dose estimation may have affected our dose-response estimates. FUNDING This work was supported by project grants from the National Health and Medical Research Council, Australia [grant numbers 509190, 1027368]. ACKNOWLEDGEMENTS We thank state and territory radiation regulators for data on CT scanner registrations, the Department of Health for the provision of detailed Medicare data sets and the Medical Branch of the Australian Radiation Protection and Nuclear Safety Agency (ARPANSA) for supplying historical Australian CT survey data. We also thank Mita Pederson, Chief Medical Imaging Technologist (MIT) and Fiona Ramanauskas, CT Supervisor MIT, at The Royal Children’s Hospital Melbourne for expert clinical input, and Brian Moroz at NCI (Washington) for NCICT calculations. References 1. Brenner , D. J. et al. Cancer risks attributable to low doses of ionizing radiation: assessing what we really know . Proc. Natl. Acad. Sci. USA 100 ( 24 ), 13761 – 13766 ( 2003 ). Google Scholar Crossref Search ADS WorldCat 2. Pearce , M. S. et al. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study . Lancet 380 ( 9840 ), 499 – 505 ( 2012 ). Google Scholar Crossref Search ADS PubMed WorldCat 3. Mathews , J. D. et al. Cancer risk in 680,000 people exposed to computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians . Br. Med. J. 346 , f2360 ( 2013 ). Google Scholar Crossref Search ADS WorldCat 4. Huang , W.-Y. , Muo , C.-H., Lin , C.-Y., Jen , Y.-M., Yang , M.-H., Lin , J.-C., Sung , F.-C. and Kao , C.-H. Paediatric head CT scan and subsequent risk of malignancy and benign brain tumour: a nation-wide population-based cohort study . Br. J. Cancer 110 ( 9 ), 1 – 7 ( 2014 ). Google Scholar Crossref Search ADS PubMed WorldCat 5. Meulepas , J. M. et al. Radiation exposure from pediatric CT scans and subsequent cancer risk in the Netherlands . J. Natl. Cancer Inst. 111 ( 3 ), 256 – 263 ( 2019 ). Google Scholar Crossref Search ADS PubMed WorldCat 6. Hong , J.-Y. , Han , K., Jung , J.-H. and Kim , J. S. Association of exposure to diagnostic low-dose ionizing radiation with risk of cancer among youths in South Korea . JAMA Netw. Open 2 ( 9 ), e1910584 ( 2019 ). Google Scholar OpenURL Placeholder Text WorldCat 7. Lee , C. , Journy , N., Moroz , B. E., Little , M., Harbron , R., McHugh , K., Pearce , M. and Berrington de González , A. Organ dose estimation accounting for uncertainty for pediatric and young adult CT scans in the United Kingdom . Radiat. Prot. Dosimetry 184 ( 1 ), 44 – 53 ( 2019 ). Google Scholar Crossref Search ADS PubMed WorldCat 8. Kim , K. P. , Berrington de González , A., Pearce , M. S., Salotti , J. A., Parker , L., McHugh , K., Craft , A. W. and Lee , C. Development of a database of organ doses for paediatric and young adult CT scans in the United Kingdom . Radiat. Prot. Dosimetry 150 ( 4 ), 415 – 426 ( 2012 ). Google Scholar Crossref Search ADS PubMed WorldCat 9. Lee , C. , Pearce , M. S., Salotti , J. A., Harbron , R. W., Little , M. P., McHugh , K., Chapple , C.-L. and Berrington de González , A. Reduction in radiation doses from paediatric CT scans in great Britain . Br. J. Radiol. 89 ( 1060 ), 20150305 – 20150308 ( 2016 ). Google Scholar Crossref Search ADS PubMed WorldCat 10. Meulepas , J. M. et al. Leukemia and brain tumors among children after radiation exposure from CT scans: design and methodological opportunities of the Dutch pediatric CT study . Eur. J. Epidemiol. 29 , 293 – 301 ( 2014 ). Google Scholar Crossref Search ADS PubMed WorldCat 11. Thierry-Chef , I. et al. Assessing organ doses from paediatric CT scans—a novel approach for an epidemiology study (the EPI-CT study) . Int. J. Environ. Res. Public Health 10 , 717 – 728 ( 2013 ). Google Scholar Crossref Search ADS PubMed WorldCat 12. Bosch de Basea , M. et al. EPI-CT: design, challenges and epidemiological methods of an international study on cancer risk after paediatric and young adult CT . J. Radiol. Prot. 35 , 611 – 628 ( 2015 ). Google Scholar Crossref Search ADS PubMed WorldCat 13. Bernier , M.-O. , Rehel , J.-L., Brisse , H. J., Wu-Zhou , X., Caer-Lorho , S., Jacob , S., Chateil , J. F., Aubert , B. and Laurier , D. Radiation exposure from CT in early childhood: a French large-scale multicentre study . Br. J. Radiol. 85 , 53 – 60 ( 2012 ). Google Scholar Crossref Search ADS PubMed WorldCat 14. Brady , Z. , Forsythe , A. V. and Mathews , J. D. The changing use of pediatric CT in Australia . Pediatr. Radiol. 46 , 1199 – 1208 ( 2016 ). Google Scholar Crossref Search ADS PubMed WorldCat 15. Australian Government Department of Health . Medicare Benefits Schedule (MBS) Handbook . ( Canberra, Australia : Australian Government Department of Health ) ( 2001 ). Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 16. Lee , C. , Kim , K. P., Bolch , W. E., Moroz , B. E. and Folio , L. NCICT: a computational solution to estimate organ doses for pediatric and adult patients undergoing CT scans . J. Radiol. Prot. 35 ( 4 ), 891 – 909 ( 2015 ). Google Scholar Crossref Search ADS PubMed WorldCat 17. International Commission on Radiological Protection (ICRP) . Adult Reference Computational Phantoms. ICRP publication 110 . Ann. ICRP 39 , 1 – 166 ( 2009 ). OpenURL Placeholder Text WorldCat 18. Lee , C. , Lodwick , D., Hurtado , J., Pafundi , D., Williams , J. L. and Bolch , W. E. The UF family of reference hybrid phantoms for computational radiation dosimetry . Phys. Med. Biol. 55 ( 2 ), 339 – 363 ( 2010 ). Google Scholar Crossref Search ADS PubMed WorldCat 19. Huda , W. , Sterzik , A. and Tipnis , S. X-ray beam filtration, dosimetry phantom size and CT patient dose conversion factors . Phys. Med. Biol. 55 , 551 – 561 ( 2010 ). Google Scholar Crossref Search ADS PubMed WorldCat 20. Shrimpton , P. C. , Hillier , M. C., Lewis , M. A. and Dunn , M. Doses from computed tomography (CT) examinations in the UK—2003 Review Report NRPB-W67 . ( Chilton, UK : National Radiological Protection Board ) ( 2005 ). Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 21. Brady , Z. , Ramanauskas , F., Cain , T. M. and Johnston , P. N. Assessment of paediatric CT dose indicators for the purpose of optimisation . Brit. J. Radiol. 85 , 1488 – 1498 ( 2012 ). Google Scholar Crossref Search ADS WorldCat 22. ImPACT CT Dosimetry Calculator, Version 1.0.4 . London, UK : St. George’s Healthcare NHS Trust , ( 2011 ). http://www.impactscan.org/ctdosimetry.htm Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 23. Henson , P. W. Acceptance tests and patient dose measurements on a Siemens Somatom 2 CT scanner . Australas. Phys. Eng. Sci. Med. 5 ( 3 ), 102 – 112 ( 1982 ). Google Scholar OpenURL Placeholder Text WorldCat 24. Sim , L. H. and Case , C. C. A comparison of the radiation dose to the lens of the eye from four modern C.T. scanners . Australas. Phys. Eng. Sci. Med. 11 ( 2 ), 76 – 80 ( 1988 ). Google Scholar PubMed OpenURL Placeholder Text WorldCat 25. Brady , Z. Radiation doses and risks from paediatric computed tomography . PhD Thesis, Royal Melbourne Institute of Technology University, Melbourne, Australia , ( 2012 ). 26. Hayton , A. and Wallace , A. Derivation of Australian diagnostic reference levels for paediatric multi detector computed tomography . Australas. Phys. Eng. Sci. Med. 39 , 615 – 626 ( 2016 ). Google Scholar Crossref Search ADS PubMed WorldCat 27. Thomson , J. E. M. and Tingey , D. R. C. Radiation doses from computed tomography in Australia . ( Yallambie, Australia : Australian Radiation Laboratory, Department of Health and Family Services, Commonwealth of Australia ) ( 1997 ). Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 28. Moss , M. and McLean , D. Paediatric and adult computed tomography practice and patient dose in Australia . Australas. Radiol. 50 ( 1 ), 33 – 40 ( 2006 ). Google Scholar Crossref Search ADS PubMed WorldCat 29. Boal , T. J. , Einsiedel , P. F. and Cardillo , I. Guidance levels for diagnostic radiology in Victoria . Australas. Phys. Eng. Sci. Med. 23 ( 1 ), 7 – 14 ( 2000 ). Google Scholar PubMed OpenURL Placeholder Text WorldCat 30. GE Medical Systems . CT9800 Patient Protocol , 46-237099P1, Rev 2. (US) . ( 1991 ). 31. International Commission on Radiological Protection (ICRP) . The 2007 Recommendations of the International Commission on Radiological Protection. ICRP publication 103 . Ann. ICRP 37 , 1 – 332 ( 2007 ). OpenURL Placeholder Text WorldCat 32. Cristy , M. Active bone marrow distribution as a function of age in humans . Phys. Med. Biol. 26 ( 3 ), 389 – 400 ( 1981 ). Google Scholar Crossref Search ADS PubMed WorldCat 33. International Commission on Radiological Protection (ICRP) . Basic anatomical and physiological data for use in radiological protection: reference values. ICRP publication 89 . Ann. ICRP 32 ( 3-4 ), 1 – 277 ( 2002 ). Crossref Search ADS WorldCat 34. Hough , M. , Johnson , P., Rajon , D., Jokisch , D., Lee , C. and Bolch , W. An image-based skeletal dosimetry model for the ICRP reference adult male: internal electron sources . Phys. Med. Biol. 56 ( 8 ), 2309 – 2346 ( 2011 ). Google Scholar Crossref Search ADS PubMed WorldCat 35. Johnson , P. B. , Bahadori , A. A., Eckerman , K. F., Lee , C. and Bolch , W. E. Response functions for computing absorbed dose to skeletal tissues from photon irradiation—an update . Phys. Med. Biol. 56 ( 8 ), 2347 – 2365 ( 2011 ). Google Scholar Crossref Search ADS PubMed WorldCat 36. Brenner , D. J. , Elliston , C. D., Hall , E. J. and Berdon , W. E. Estimated risks of radiation-induced fatal cancer from pediatric CT . Am. J. Roentgenol. 176 , 289 – 296 ( 2001 ). Google Scholar Crossref Search ADS WorldCat 37. Smoll , N. R. , Mathews , J. D. and Scurrah , K. J. CT scans in childhood predict subsequent brain cancer: finite mixture modelling can help separate reverse causation scans from those that may be causal . Cancer Epidemiol. 67 , 101732 ( 2020 ). Google Scholar Crossref Search ADS PubMed WorldCat 38. McBain-Miller , J. , Mathews , J. D. and Scurrah , K. J. Using propensity scores to account for confounding by indication . Joint International Society for Clinical Biostatistics, Melbourne, Australia: Melbourne, Australia , ( 2018 ). 39. Brady , Z. , Cain , T. M. and Johnston , P. N. Comparison of organ dosimetry methods and effective dose calculation methods for paediatric CT . Australas. Phys. Eng. Sci. Med. 35 , 117 – 134 ( 2012 ). Google Scholar Crossref Search ADS PubMed WorldCat 40. Gao , Y. , Quinn , B., Mahmood , U., Long , D., Erdi , Y., St. Germain , J., Pandit-Taskar , N., Xu , X. G., Bolch , W. E. and Dauer , L. T. A comparison of pediatric and adult CT organ dose estimation methods . BMC Med. Imaging 17 , 28 ( 2017 ). Google Scholar Crossref Search ADS PubMed WorldCat 41. Masiuk , S. V. , Shklyar , S. V., Kukush , A. G., Carroll , R. J., Kovgan , L. N. and Likhtarov , I. A. Estimation of radiation risk in presence of classical additive and Berkson multiplicative errors in exposure doses . Biostatistics 17 ( 3 ), 422 – 436 ( 2016 ). Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2020. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - CT DOSIMETRY FOR THE AUSTRALIAN COHORT DATA LINKAGE STUDY
JF - Radiation Protection Dosimetry
DO - 10.1093/rpd/ncaa175
DA - 2020-12-17
UR - https://www.deepdyve.com/lp/oxford-university-press/ct-dosimetry-for-the-australian-cohort-data-linkage-study-J2v4Emo3kR
SP - 423
EP - 438
VL - 191
IS - 4
DP - DeepDyve
ER -