Analysis and results from a UK national dose audit of paediatric CT examinations

Abstract

Objective:

To present the results following a UK national patient dose audit of paediatric CT examinations, to propose updated UK national diagnostic reference levels (DRLs) and to analyse current practice to see if any recommendations can be made to assist with optimisation.

Methods:

A UK national dose audit was undertaken in 2019 focussing on paediatric CT examinations of the head, chest, abdomen/pelvis and cervical spine using the methods proposed by the International Commission on Radiological Protection. The audit pro-forma contained mandatory fields, of which the post-examination dosimetry (volume CT dose index and dose–length product) and the patient weight (for body examinations) were the most important.

Results:

Analysis of the data submitted indicates that it is appropriate to propose national DRLs for CT head examinations in the 0-<1, 1–<5, 5–<10 and 10–<15 year age ranges. This extends the number of age categories of national DRLs from those at present and revises the existing values downwards. For CT chest examinations, it is appropriate to propose national DRLs for the first time in the UK for the 5–<15, 15–<30, 30–<50 and 50–<80 kg weight ranges. There were insufficient data received to propose national DRLs for abdomen/pelvis or cervical spine examinations. Recommendations towards optimisation focus on the use of tube current (mA) modulation, iterative reconstruction and the selection of examination tube voltage (kV_p).

Conclusion:

Updated UK national DRLs are proposed for paediatric CT examinations of the head and chest.

Advances in knowledge:

A national patient dose audit of paediatric CT examinations has led to the proposal of updated national DRLs.

Introduction

The International Commission on Radiological Protection (ICRP) advise an ‘As Low As Reasonably Achievable’ (ALARA) approach to medical imaging undertaken using ionising radiation.¹ This process, of using the lowest amount of radiation consistent with achieving an image that fulfils the diagnostic purpose, is known as optimisation. The most recent recommendations of the ICRP¹ were adopted in the European Union’s 2013 Basic Safety Standards Directive (BSSD).² In the UK, the patient exposure aspects of the BSSD are addressed within the Ionising Radiation (Medical Exposure)³ Regulations 2017³ and the Ionising Radiation (Medical Exposure) Regulations (Northern Ireland) 2018⁴ (IR(ME)R), including the requirement that all patient exposures be optimised.

Diagnostic reference levels (DRLs) are a key tool in the optimisation process.⁵ The comparison of local patient dose data and DRL values with national DRL values, or international values as available, can indicate examinations in need of optimisation and highlight the examinations requiring the most urgent attention.

The Committee on the Biological Effects of Ionising Radiation (BEIR) have concluded that the risk of both cancer incidence and mortality is related to age at exposure and is highest for paediatric patients.⁶ The United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) also concluded in their 2013 review⁷ that the risks to paediatric patients following exposure were often higher than for adults. This being the case, the ICRP advised on the particular need for optimisation of examinations of paediatric patients. This recommendation was subsequently included in the BSSD, and it is therefore a requirement of IR(ME)R that special attention be paid to the optimisation of exposures of paediatric patients.^3,4

The ICRP advise that the use of adult protocols for paediatric patients will result in unnecessarily high doses.⁵ They recommend that the optimisation of paediatric examinations be considered separately, and that the process should make use of weight or size adjusted paediatric DRL values. The optimisation of paediatric CT examinations is complicated in the UK by a lack of national DRLs with which to compare local practice. At time of writing, national DRLs for paediatric CT examinations are only available for head examinations for the 0–1, 1–5 and >5 year age ranges.⁸ These national DRLs were proposed following analysis of data collected in 2011.⁹ The data collection exercise in that work only requested paediatric examination data for head examinations, believing these to be the only examinations undertaken often enough in general CT centres to merit inclusion. With many technological developments in the time since, most notably the widespread implementation of automated control of mA according to patient size and the increased use of iterative reconstruction, these national DRLs are now unlikely to be appropriate for current practice.

A working party of the Institute of Physics and Engineering in Medicine (IPEM) (the paediatric optimisation working party) began work in 2018 on national dose audits for CT, radiographic and fluoroscopic paediatric X-ray examinations in the UK with the intention of proposing updates to the current national DRLs for paediatric patients. This paper presents the analysis and results of the national dose audit for paediatric CT examinations in the UK.

Methods

Data collection

Workload data for paediatric CT examinations in NHS facilities across England is collected by NHS Digital as part of the Diagnostic Information Dataset. The most recent data available were for the 2019 calendar year. The most performed CT examinations on paediatric patients were the head followed by chest, cervical spine and then abdomen-pelvis. It is reasonable to assume this holds true for the other countries of the UK too. The working party made the decision to include the following examinations (with clinical indications) in the scope of the audit; head (trauma), chest (suspected tumour), cervical spine (trauma) and abdomen-pelvis (trauma). A comprehensive systematic review of paediatric CT DRLs by Satharasinghe et al¹⁰ confirms those as amongst the examinations of most interest across 15 countries. There was also scope for the user to define their own examination and clinical indication combination.

In keeping with the familiar methods used for UK national dose audit data collection in the past,^9,11,12 a Microsoft Excel^TM (Washington, USA) pro-forma was created for data collection. The pro-forma was of a similar format to the previous Public Health England (PHE) and IPEM run dose audits to maintain consistency. The pro-forma requested both mandatory data (which if uncompleted could result in the data not being used at the analysis phase) and optional data which were included to facilitate a more in-depth analysis that might allow for recommendations on protocol optimisation. A key mandatory request was for the weight of the paediatric patient for examinations of the body.

The mandatory data requested are shown in Table 1 and the optional data requested are shown in Table 2.

Table 1.

Mandatory data requested in the pro-forma, by category (*note that patient weight was mandatory for all examinations except CT head)

Administration	Scanner-specific	Protocol-specific	Patient-specific
Name of person completing pro-forma  Contact details	Hospital name  Local system unique identifier  Scanner manufacturer  Scanner model	Clinical indication  Protocol name  Number of scan acquisitions  Number of scout views  Scout view projection  CTDI phantom size  Is AEC used?  Is iterative reconstruction used?	Patient age  Patient weight*  CTDIvol  DLP

AEC, Automatic Exposure Control; CTDI, CT dose index; DLP, dose–length product.

Table 2.

Optional data requested in the pro-forma, by category

Scanner-specific	Protocol-specific	Patient-specific
No. of detector rows  Year of scanner manufacture  Software version  Error of indicated CTDIvol	Scout kVp & mA(s)  AEC system name & type  AEC setting value  AEC min & max mA  Manual mA  Iterative reconstruction type & setting  Collimated beam width  Number of slices & detector size  Is automatic tube voltage selection used?  Fixed examination kVp  Tube rotation time  Primary image slice thickness  Scan field of view  Reconstruction field of view  Axial or helical?  Pitch  Primary reconstruction algorithm or kernel  Is contrast used?  Anatomical landmarks for start & end points	Height  Scanned length  Start & end position  kVp (if different from protocol)  CTDI phantom (if different from protocol)  Scan field of view (if different from protocol)

AEC, Automatic Exposure Control; CTDI, CT dose index.

As the audit aimed to capture current practice, the pro-forma requested that retrospective data be from no earlier than January 2017, and that data should only be submitted from scanners that were still in use at the time of submission. The data submission ran from June to December 2019 to encourage a high response.

The pro-forma was circulated on the UK Medical Physics and Engineering Mailbase (hosted by JISCmail) which has over 3200 members at time of publication and was hosted on the CT users group website¹³ with a link circulated by email to their entire membership.

Data analysis

In keeping with national and international recommendations for paediatric patient dose audit,^5,14,15 the analysis of the data was undertaken within weight ranges for body examinations. As per the recommendations of ICRP report 135,⁵ these ranges are < 5 kg, 5–<15 kg, 15–<30 kg, 30–<50 kg and 50–<80 kg. Weight was included within the mandatory data requested in the pro-forma for all body examinations. For CT head examinations, it is recommended⁵ that age groupings be used when analysing patient dose audit data as there is a good correlation between age and head size. The recommendations of ICRP report 135⁵ and the International Atomic Energy Agency (IAEA) human health series report 24¹⁵ give these ranges as 0–<1, 1–<5, 5–<10, 10–<15 and 15–<18 years. Weight was not included in the mandatory data for CT head examinations.

All data were subject to quality assurance before analysis. Any data for body examinations that did not contain patient weight were rejected, as were data for head examinations that did not contain patient age. Data for patients older than 18 years, or in excess of 80 kg in weight, were rejected. Data that did not have the same clinical indication as in the pro-forma were rejected.

Once the data that did not meet these criteria had been removed, all the remaining data were compiled into weight (or age for head examinations) ranges for each scanner. Within each weight or age range for which there were 10 or more patients, the median values of volume CT dose index (CTDI_vol) and dose–length product (DLP) were calculated and the minimum and maximum values identified. This is in line with ICRP recommendations⁵ that DRLs are based on the medians rather than the means of the patient distribution. Where the ratio between the median/minimum or maximum/median value exceeded 10, the minimum/maximum value was investigated and discussed with the site that submitted the data. In all cases, transcription errors were identified. Once corrected or removed, the ratios of median/minimum and maximum/median were all less than eight across all weight or age ranges and all examinations.

The minimum, first quartile, median, third quartile and maximum values of this distribution of median values were then calculated for each examination, within each weight or age range.

National DRLs will be proposed at the third quartile of the range of these median values provided there are an adequate number of individual submissions within each weight or age range for each examination.

Results and discussion

Data received

Data were received from 29 hospitals and 40 CT scanners. Of the 29 hospitals, 5 were specialist paediatric hospitals. Table 3 shows the distribution of scanner manufacturer and model, with Automatic Exposure Control (AEC) and iterative reconstruction capability, for the data received. In total, 2.2% of the data submitted were rejected, with the absence of patient weight information the predominant reason.

Table 3.

Number of scanners by manufacturer and model, including the number with AEC and iterative reconstruction capability

Scanner manufacturer and model	Number (%)	Number with AEC	Number with iterative reconstruction
Canon Medical Systems Ltd	8 (20)	8	5
Aquilion CX	3 (7.5)	3	1
Aquilion One	2 (5)	2	2
Aquilion Prime	3 (7.5)	3	2
GE Healthcare	6 (15)	5	5
Discovery HD750	1 (2.5)	1	1
Lightspeed VCT	1 (2.5)	1	1
Optima CT 660	1 (2.5)	1	1
Revolution	1 (2.5)	0	0
Revolution WB	2 (5)	2	2
Philips Medical	6 (15)	6	3
Brilliance 64	3 (7.5)	3	0
iCT 256	2 (5)	2	2
Ingenuity	1 (2.5)	1	1
Siemens Healthineers	20 (50)	20	10
Definition AS	5 (12.5)	5	1
Definition Edge	6 (15)	6	6
Definition Flash	5 (12.5)	5	1
Definition Force	2 (5)	2	2
Sensation	2 (5)	2	0
Total	40 (100)	39	23

AEC, Automatic Exposure Control.

After removing rejected data, the total number by examination, and across weight ranges, for body examinations is shown in Table 4. Table 5 shows the total number across age ranges for head examinations.

Table 4.

Distribution of submitted data for body examinations across weight ranges after the removal of rejected data

			Number of patients
Examination	Hospitals	Scanners	<5 kg	5–<15 kg	15–<30 kg	30–<50 kg	50–<80 kg
Abdomen-pelvis	4	4	1	8	31	45	27
C-spine	4	4	0	10	73	54	26
Chest	9	9	48	348	547	258	100

Table 5.

Distribution of submitted data for head examinations across age ranges after the removal of rejected data

			Number of patients
Examination	Hospitals	Scanners	0-< 1 years	1–<5 years	5–<10 years	10–<15 years	15-< 18 years
Head	24	34	691	1111	877	1089	717

In considering whether the data that were received is representative enough to be able to recommend national DRLs, the workload data for paediatric CT examinations in NHS facilities across England in 2019 were used to estimate a UK workload using a simple correction for population. The total number of patient examinations shown in Tables 4 and 5 for each examination were compared to these estimated UK wide workload figures. The data received represent of the order of 14% of all paediatric head CT examinations undertaken in a year in the UK, 25% of all paediatric CT chest examinations, 7% of all paediatric c-spine examinations and 5% of all abdomen-pelvis CT examinations. The data received cannot be considered representative for c-spine or abdomen-pelvis examinations but is likely to be for head and chest examinations if the data meet certain criteria (i.e. normally distributed) in the analysis that follows.

Head examinations

Prior to analysis, all data were checked to ensure reference to the 16 cm diameter CTDI phantom. This was true for all examinations.

The minimum, first quartile, median, third quartile and maximum values of the distribution of scanner median values is shown in Table 6 for CTDI_vol and Table 7 for DLP, by age range.

Table 6.

Statistical analysis of the distribution of scanner median CTDI_vol values for head examinations by age range

Age range (yrs)	Number of scanners	Number of patients	CTDIvol (mGy)
			Minimum	First quartile	Median	Third quartile	Maximum
0–<1	17	503	10.2	16.5	17.8	19.1	22.6
1–<5	22	815	13.0	17.6	21.2	23.8	29.3
5–<10	20	605	15.1	25.2	31.4	35.6	43.8
10–<15	23	827	20.0	34.2	37.5	45.5	51.5
15–<18	16	526	23.7	37.4	44.5	49.5	60.5

CTDI, CT dose index.

Table 7.

Statistical analysis of the distribution of scanner median DLP values for head examinations by age range

Age range (yrs)	Number of scanners	Number of patients	DLP (mGy.cm)
			Minimum	First quartile	Median	Third quartile	Maximum
0–<1	19	625	163	220	258	292	364
1–<5	25	984	205	297	351	420	618
5–<10	23	751	314	420	505	574	711
10–<15	25	957	382	570	628	689	965
15–<18	18	596	379	638	759	834	1001

DLP, dose–length product.

All of the results follow the expected trend of increasing CTDI_vol and DLP with patient age range across all statistical analyses, with the exception of the minimum DLP for the 10–<15 and 15–<18 age ranges where the results are near equal. This is not considered significant as these patient cohorts can be considered of similar size for head examinations.⁹

International DRLs were proposed for paediatric CT examinations by Vassileva et al¹⁶ in 2015. These were proposed at the 75th percentile of the distribution of median values from CT scanners in 32 countries. International DRLs for CT head examinations of the 0–<1, 1–<5, 5–<10 and 10–<15 year age ranges were proposed. The third quartile values shown in Tables 6 and 7 are 18–34% lower than the corresponding proposed international DRL.

National DRLs for paediatric head examinations are currently available.⁸ These national DRLs were set at the third quartile of mean values, rather than the median values used for the analysis summarised in Tables 6 and 7. To demonstrate that any difference in the proposed national DRL values from this work is a true difference due to technological advancement or changes to protocol or technique and not the result of the change in calculation methodology, the third quartile of the mean values was also calculated. The current UK national DRLs for paediatric head CT examinations and the third quartile of the median and mean values from this survey are shown in Table 8 for CTDI_vol and Table 9 for DLP.

Table 8.

Current UK CTDI_vol national DRLs for paediatric head CT examinations, derived from mean values, compared with the third quartile of median and mean values from this survey

UK NDRLs		This survey	This survey; third quartile of median values		This survey; third quartile of mean values
Age range (yrs)	CTDIvol (mGy)	Age range (yrs)	CTDIvol (mGy)	Reduction from NDRLs	CTDIvol (mGy)	Reduction from NDRLs
0–<1	25	0–<1	19.1	24%	19.6	22%
1–<5	40	1–<5	23.8	41%	25.3	37%
5+	60	5–<10	35.6	41%	34.1	43%
		10–<15	45.5	24%	41.7	31%
		15–<18	49.5	18%	48.3	20%

CTDI, CT dose index; DRL, diagnostic reference level.

Table 9.

Current UK DLP national DRLs for paediatric head CT examinations, derived from mean values, compared with the third quartile of median and mean values from this survey

UK NDRLs		This survey	This survey; third quartile of median values		This survey; third quartile of mean values
Age range (yrs)	DLP (mGy.cm)	Age range (yrs)	DLP (mGy.cm)	Reduction from NDRLs	DLP (mGy.cm)	Reduction from NDRLs
0–<1	350	0–<1	292	17%	292	17%
1–<5	650	1–<5	420	35%	433	33%
5+	860	5–<10	574	33%	589	32%
		10–<15	689	20%	708	18%
		15–<18	834	3%	828	4%

DLP, dose–length product; DRL, diagnostic reference level.

The third quartile values for CTDI_vol and DLP for head examinations are lower than the current national DRLs across all age ranges using both the median and mean values. These are consistently sufficiently close to each other (with a median absolute difference of 2.6% for any individual age range) to confirm the reduction in third quartile value is not the result of the change in calculation methodology. This is expected as the sample sizes in the 2011 review⁹ on which the current national DRLs are based were generally high enough to reduce the likelihood of the mean being disproportionately affected by a few high values.

These reduced values are expected for a dose audit undertaken with patient data collected around 8 years after the data used to propose the current national DRLs. With improvements in CT detector technology, the widespread implementation of automated control of mA according to patient size and the increased use of iterative reconstruction, it is expected that there would be a reduction in patient dose for an equivalent image quality. The existence of the national DRLs alone may also account for some of the reduction, with these providing a level against which departments can gauge and improve upon their local performance.

The significant increase in the third quartile of median CTDI_vol and DLP values with each age range suggests it is appropriate to propose national DRLs that use each of the age ranges identified, which would mean an extension of those available at present. However, a further consideration is that a national dose audit for CT examinations of adults ran concurrently with this audit, with the data collected, analysed and to be published by PHE. Preliminary analysis of these data indicate that the proposed national DRL for adult head examinations will be a little lower than would be suggested for the 15–<18 year age range in this work (hough they can be considered broadly similar).¹⁷ This apparent anomaly can be explained by a much larger sample size in the PHE adult dose audit and a far higher contribution from scanners using iterative reconstruction (75% in the adult dose audit vs 58% in this work). As it will clearly not be helpful to propose a national DRL for the 15–<18 year age range that is higher than that proposed for an adult, the working party is not recommending a national DRL for the 15–<18 year age range and will instead recommend the use of the updated national DRL for adult examinations when available.

Effect of tube voltage (kV_p)

The examination kV_p was part of the optional information on the pro-forma but was well responded to. Of the data for 4485 individual patient examinations, 3285 of these (73%) contained the examination kV_p. The number of examinations undertaken within each age range at each kV_p is shown in Figure 1.

Figure 1.

Number of head examinations undertaken at each kV_p within each patient age range.

The median CTDI_vol for examinations undertaken at each kV_p are shown in Table 10 for each age range where there were at least 10 examinations in the sample. The equivalent for DLP is shown in Table 11.

Table 10.

Median CTDI_vol (and number of examinations) by kV_p for head examinations by age range

Age range (yrs)	Median CTDIvol (mGy) (number of exams) by kVp
	80 kVp	100 kVp	120 kVp
0 –<1	–	17.5 (350)	18.2 (90)
1–<5	12.8 (16)	21.5 (388)	24.7 (285)
5–<10	–	23.0 (97)	31.6 (436)
10–<15	–	35.2 (74)	37.3 (663)
15–<18	–	40.7 (63)	41.0 (390)

CTDI, CT dose index.

Table 11.

Median DLP (and number of examinations) by kV_p for head examinations by age range

Age range (yrs)	Median DLP (mGy.cm) (number of exams) by kVp
	80 kVp	100 kVp	120 kVp
0–<1	–	255 (351)	297 (151)
1–<5	202 (16)	358 (391)	436 (365)
5–<10	–	391 (97)	527 (512)
10–<15	–	646 (74)	620 (711)
15–<18	–	694 (63)	729 (444)

DLP, dose–length product.

The distribution of kV_p for head examinations indicates a strong preference for 100 kV_p over 120 kV_p for the 0–<1 year age range (with a 69–30% selection frequency). The median CTDI_vol and DLP is lower for the 100 kV_p selection. In the 1–<5 year age range there is only a slight preference for 100 kV_p over 120 kV_p (with a 51–47% selection frequency). Again, the median CTDI_vol and DLP is lower for the 100 kV_p selection, and by a more substantial margin than for the 0–<1 year age range.

In older age ranges, the selection frequency shifts definitively towards 120 kV_p (16–84% selection frequency for the 5–<10 year age range, 9–91% selection frequency for the 10–<15 year age range and 12–88% selection frequency for the 15–<18 year age range). The median CTDI_vol and DLP are lower for the 100 kV_p selection for the 5–<10 year age range by a significant margin. For the 10–<15 and 15–<18 year age ranges, there is very little between the median CTDI_vol and DLP at 100 and 120 kV_p.

There are limitations in analysing the data in this manner. In grouping all individual examinations together in one distribution, there will be bias introduced by either very well or poorly optimised sites, particularly where the sample size for a single combination of kV_p and examination is low. The differences as a result of scanner technology and other protocol choices are potentially confounded. However, this analysis is not intended to propose reference levels, instead it is intended to identify potential routes to optimisation worth exploring.

The analysis of the kV_p selected for head examinations indicates that a shift to 100kV_p is worth investigating for the 0–<1, 1–<5 and 5–<10 year age ranges for those sites not already using this as standard practice as there is evidence suggesting a consistently lower patient dose. In line with the recommendations of the Committee on Medical Aspects of Radiation in the Environment (COMARE) in their 16th report,¹⁸ optimisation of CT examinations should be undertaken by a multidisciplinary team who should consider whether a move to a lower kV_p would suit their image quality requirements for paediatric head CT examinations. There is no strong evidence in support of a reduction in examination kV_p for the 10–<15 or 15–<18 year age ranges from this work.

Effect of reconstruction type

The data were analysed in respect of the reconstruction technique. The median and third quartile CTDI_vol and DLP are shown in Table 12 for examinations using filtered back projection and iterative reconstruction.

Table 12.

Median and third quartile CTDI_vol and DLP by reconstruction type for head examinations by age range

	Filtered back projection						Iterative reconstruction
	CTDIvol (mGy)			DLP (mGy.cm)			CTDIvol (mGy)			DLP (mGy.cm)
Age range (yrs)	n	Median	Third quartile	n	Median	Third quartile	n	Median	Third quartile	n	Median	Third quartile
0 –<1	137	17.1	21.3	194	278	343	385	17.4	19.9	453	250	299
1–<5	216	21.6	28.4	271	394	486	615	21.5	25.5	697	361	451
5–<10	149	33.8	38.7	187	519	620	466	28.0	34.0	580	463	596
10–<15	245	43.8	51.5	257	646	740	581	35.1	41.3	706	585	722
15–<18	231	49.5	56.7	238	811	988	357	39.0	45.2	428	691	806

CTDI, CT dose index; DLP, dose–length product.

This analysis has the same limitations as for the effect of examination kV_p as discussed in the section looking at the effect of tube voltage. As is expected having been reported elsewhere in the literature,¹⁹ the median and third quartile of the distribution of median patient doses are consistently lower with the use of iterative reconstruction. This can assist with local optimisation; e.g. where a site has more than one CT scanner and only some are capable of iterative reconstruction, paediatric patients should undergo their examination on a scanner capable of iterative reconstruction wherever possible. Care is required in implementing iterative reconstruction, however, as it can change the visual impression of the images.

Effect of automatic exposure control

The data were analysed in respect of whether the AEC was used. The median and third quartile CTDI_vol and DLP are shown in Table 13 for examinations with and without the AEC.

Table 13.

Median and third quartile CTDI_vol and DLP with and without the AEC for head examinations by age range

	With AEC						Without AEC
	CTDIvol (mGy)			DLP (mGy.cm)			CTDIvol (mGy)			DLP (mGy.cm)
Age range (yrs)	n	Median	Third quartile	n	Median	Third quartile	n	Median	Third quartile	n	Median	Third quartile
0–<1	439	18.1	20.8	563	261	321	102	17.1	19.7	103	248	295
1–<5	732	21.5	25.8	863	365	446	116	28.3	28.4	153	372	538
5–<10	596	29.8	35.9	744	482	597	35	33.9	35.5	42	626	678
10–<15	805	36.7	44.4	942	596	722	42	46.6	46.8	42	882	944
15–<18	539	42.3	49.8	617	719	836	59	49.5	49.5	59	983	1024

AEC, automatic exposure control; CTDI, CT dose index; DLP, dose–length product.

This analysis has the same limitations as for the effect of examination kV_p as discussed in the sections looking at the effect of tube voltage and the effect of reconstruction type. As is expected, having been reported elsewhere in the literature,¹⁹ the median and third quartile of the distribution of median patient doses are consistently lower for examinations that use the AEC except for the 0–<1 year age range where the median and third quartile values are lower when using a fixed mA. This matches the findings of Worrall et al²⁰ who performed CT head examinations on three different sized paediatric anthropomorphic phantoms on a large number of CT scanners across Scotland and found the lowest doses were associated with protocols using a fixed mA for the 1-year-old phantom but using AEC for the 5- and 10-year-old phantoms. Worrall et al concluded that the reason for the higher doses when using AEC for the 1-year-old phantom was inappropriately set minimum mA levels in the AEC setup. For these examinations, the mA was not observed to vary as it was not able to go below the pre-set minimum value. This is a potential explanation for the trends with AEC seen in this work.

Whilst the majority of sites are using AEC across all age ranges for CT head examinations, some are not (15% for the 0–<1 and 1–<5 year age ranges, <10% for all other age ranges). The results of this analysis suggest that optimised protocols for the 1–<5 year age range use AEC. Optimisation of protocols for the 0–<1 year age range should consider the minimum mA that is set where AEC is used.

Chest examinations

Prior to analysis, all data were checked to ensure reference to the 32 cm diameter CTDI phantom. This was true for all data.

The minimum, first quartile, median, third quartile and maximum values of the distribution of scanner median values is shown in Table 14 for CTDI_vol and Table 15 for DLP, by weight range.

Table 14.

Statistical analysis of the distribution of scanner median CTDI_vol values for chest examinations by weight range

Weight range (kg)	Number of scanners	Number of patients	CTDIvol (mGy)
			Minimum	First quartile	Median	Third quartile	Maximum
0–<5	2	22	0.31	0.35	0.40	0.44	0.48
5–<15	6	316	0.24	0.52	0.68	0.86	3.6
15–<30	7	505	0.4	0.9	1.2	2.0	3.3
30–<50	7	228	0.8	1.2	1.5	2.6	4.2
50–<80	3	62	1.2	1.8	2.5	3.8	5.2

CTDI, CT dose index.

^aNote that where there are less than five scanner median values, quartile values are calculated by linear interpolation between the minimum and maximum values.

Table 15.

Statistical analysis of the distribution of scanner median DLP values for chest examinations by weight range

Weight range (kg)	Number of scanners	Number of patients	DLP (mGy.cm)
			Minimum	First quartile	Median	Third quartile	Maximum
0–<5	2	22	4.9	6.0	7.0	8.0	9.1
5–<15	7	342	5.1	11.1	14.5	17.8	72.5
15–<30	8	542	9.8	18.2	26.9	40.2	91.0
30–<50	8	255	22.4	37.7	48.6	62.0	121.5
50–<80	4	86	38.6	73.4	90.1	113.1	167.0

DLP, dose–length product.

^aNote that where there are less than five scanner median values, quartile values are calculated by linear interpolation between the minimum and maximum values.

In general, the results follow the expected trend of increasing CTDI_vol and DLP with patient weight range across all statistical analysis, with some exceptions for minimum and maximum CTDI_vol values between adjacent weight ranges.

There are no national DRLs for chest examinations in the UK, but there are European DRLs as proposed by the European Society of Radiology in 2018.²¹ These are shown in Table 16 along with the percentage reduction of the third quartile values from Tables 14 and 15 by comparison. The third quartile values from this work are consistently significantly lower than the current European DRLs.

Table 16.

European DRLs for paediatric CT chest examinations²¹ and the percentage by which the third quartile values of this survey are lower

Weight range (kg)	European DRLs		Percentage by which the third quartile values in this survey are lower
	CTDIvol (mGy)	DLP (mGy.cm)	CTDIvol	DLP
<5	1.4	35	69%	77%
5–<15	1.8	50	52%	64%
15–<30	2.7	70	26%	43%
30–<50	3.7	115	30%	46%
50–<80	5.4	200	30%	43%

DRL, diagnostic reference level.

Effect of examination tube voltage (kV_p)

Of the data for 1301 individual patient examinations, 1171 of these (90%) contained the examination kV_p. The number of examinations undertaken within each weight range at each kV_p is shown in Figure 2.

Figure 2.

Number of chest examinations undertaken at each kV_p within each patient weight range

The median CTDI_vol for examinations undertaken at each kV_p are shown in Table 17 for each weight range where there were at least 10 examinations in the sample. The equivalent for DLP is shown in Table 18.

Table 17.

Median CTDI_vol (and number of examinations) by kV_p for chest examinations by weight range

Weight range (kg)	Median CTDIvol (mGy) (number of exams) by kVp
	70 kVp	80 kVp	90 kVp	100 kVp	120 kVp
0–<5	0.15 (13)	0.42 (22)	––	–	–
5–<15	0.24 (34)	0.45 (201)	–	0.78 (62)	–
15–<30	0.32 (23)	0.76 (372)	0.89 (10)	1.17 (52)	2.55 (28)
30–<50	–	1.05 (117)	1.05 (15)	1.74 (47)	3.12 (25)
50–<80	–	1.19 (25)	–	2.61 (21)	–

CTDI, CT dose index.

Table 18.

Median DLP (and number of examinations) by kV_p for chest examinations by weight range

Weight range (kg)	Median DLP (mGy.cm) (number of exams) by kVp
	70 kVp	80 kVp	90 kVp	100 kVp	120 kVp
0–<5	2.06 (13)	7.25 (22)	–	–	–
5–<15	5.02 (34)	11.0 (202)	–	15.0 (62)	–
15–<30	7.66 (23)	18.8 (372)	21.0 (10)	28.0 (52)	60.2 (28)
30–<50	–	31.6 (117)	28.6 (15)	47.0 (47)	95.1 (25)
50–<80	–	38.6 (25)	–	85.0 (21)	–

DLP, dose–length product.

This analysis has the same limitations as for the effect of examination kV_p as discussed for head examinations. The distribution of kV_p for chest examinations indicates a preference for 80 kV_p across all weight ranges (with a selection frequency of 52, 66, 76, 56 and 44% in each weight range respectively).

The median CTDI_vol and DLP is generally lower for the 80 kV_p selection than for higher kV_p (except for one case in the 30–<50 kg weight range, where the median DLP for examinations undertaken at 90 kV_p is lower than that for 80 kV_p). The median CTDI_vol and DLP is consistently lower for examinations undertaken at 70 kV_p than for examinations undertaken at 80 kV_p but there are few scanners that offer this option.

The analysis of the kV_p selected for chest examinations indicates that a shift to 80 kV_p is worth investigating across all weight ranges for those sites not already using this as standard practice, as there is evidence suggesting a consistently lower patient dose. In line with the recommendations of COMARE in their 16th report,¹⁸ optimisation of CT examinations should be undertaken by a multidisciplinary team who should consider whether a move to a lower kV_p would suit their image quality requirements for paediatric chest CT examinations.

Effect of reconstruction type

The data were analysed in respect of the reconstruction technique. The median and third quartile CTDI_vol and DLP are shown in Table 19 for examinations using filtered back projection and iterative reconstruction.

Table 19.

Median and third quartile CTDI_vol and DLP by reconstruction type for chest examinations by weight range

	Filtered back projection						Iterative reconstruction
	CTDIvol (mGy)			DLP (mGy.cm)			CTDIvol (mGy)			DLP (mGy.cm)
Weight range (kg)	n	Median	Third quartile	n	Median	Third quartile	n	Median	Third quartile	n	Median	Third quartile
0–<5	19	0.35	0.46	19	5.00	7.47	25	0.31	0.48	29	6.05	10.4
5–<15	249	0.59	0.81	257	12.6	18.0	77	0.46	0.78	91	11.3	16.5
15–<30	337	0.75	1.11	340	18.9	27.8	180	1.25	1.50	207	28.8	35.2
30–<50	177	1.40	3.12	179	40.2	95.6	62	1.32	1.60	79	41.3	53.4
50–<80	67	3.44	5.12	70	122	167	17	2.61	2.83	30	90.9	113

CTDI, CT dose index; DLP, dose–length product.

This analysis has the same limitations as for the effect of examination kV_p as discussed for head examinations. In particular, 90% of the data for the 15–<30 kg weight range with filtered back projection has come from a single site which makes it likely that there are confounding factors. The effect of reconstruction type is less clear for chest examinations than for head examinations. Lower median and third quartile CTDI_vol and DLP values are generally associated with iterative reconstruction as might be expected,¹⁹ but this is not consistent across all weight ranges and the difference between the two is often small. This is suggestive that there is a lot of scope for optimisation of these examinations across the UK.

Abdomen-pelvis and cervical spine examinations

As shown in Table 4, there was a low response to the request for data for both abdomen-pelvis and cervical spine examinations. For abdomen-pelvis examinations, data were only received from 4 hospitals and 112 patient examinations in total. For cervical spine examinations, data were only received from 4 hospitals and 163 patient examinations in total.

There is not enough data for either examination to propose national DRLs, however analysis of the data is reproduced here to provide some useful comparative information for sites looking to optimise their protocols.

Prior to analysis, all data were checked to ensure reference to the 32 cm diameter CTDI_vol phantom. This was true for all data.

For this analysis, all of the patients for each examination were included as one sample rather than analysed separately per site. The analysis is therefore not comparable to that undertaken for head or chest examinations and does not follow the recommendations of ICRP report 135⁵ in proposing national DRLs. The minimum, first quartile, median, third quartile and maximum values of the combined patient distribution is shown in Table 20 for CTDI_vol and Table 21 for DLP for each weight range for abdomen-pelvis examinations. Tables 22 and 23 show the equivalent for cervical spine examinations.

Table 20.

Statistical analysis of the distribution of individual patient CTDI_vol values for abdomen-pelvis examinations by weight range

Weight range (kg)	Number of scanners	Number of patients	CTDIvol (mGy)
			Minimum	First quartile	Median	Third quartile	Maximum
0–<5	1	1	–	–	0.43	–	–
5–<15	2	5	0.59	0.73	2.09	3.05	3.35
15–<30	4	26	0.71	0.86	1.88	3.28	6.05
30–<50	4	41	0.83	1.70	2.30	3.10	6.70
50–<80	4	19	2.41	3.27	4.00	4.66	10.3

CTDI, CT dose index.

Table 21.

Statistical analysis of the distribution of individual patient DLP values for abdomen-pelvis examinations by weight range

Weight range (kg)	Number of scanners	Number of patients	DLP (mGy.cm)
			Minimum	First quartile	Median	Third quartile	Maximum
0–<5	1	1	–	–	7.89	–	–
5–<15	3	8	16.6	28.7	41.6	61.2	82.0
15–<30	4	31	22.3	34.0	61.5	107	204
30–<50	4	45	17.3	72.2	100	136	332
50–<80	4	27	82.8	122	196	256	577

DLP, dose–length product.

Table 22.

Statistical analysis of the distribution of individual patient CTDI_vol values for cervical spine examinations by weight range

Weight range (kg)	Number of scanners	Number of patients	CTDIvol (mGy)
			Minimum	Frst quartile	Median	Third quartile	Maximum
0–<5	0	0	–	–	–	–	–
5–<15	3	8	1.07	1.73	1.85	2.10	2.62
15–<30	4	75	0.50	0.90	1.30	2.17	10.1
30–<50	4	52	0.50	1.68	2.50	2.95	6.46
50–<80	4	27	1.06	2.60	3.30	3.60	13.8

CTDI, CT dose index.

Table 23.

Statistical analysis of the distribution of individual patient DLP values for cervical spine examinations by weight range

Weight range (kg)	Number of scanners	Number of patients	DLP (mGy.cm)
			Minimum	First quartile	Median	Third quartile	Maximum
0–<5	0	0	–	–	–	–	–
5–<15	3	11	14.4	17.0	39.2	43.7	49.7
15–<30	4	78	3.0	10.5	17.3	45.5	124
30–<50	4	55	5.3	31.2	52.1	84.2	112
50–<80	4	27	21.5	61.4	89.4	103	213

DLP, dose–length product.

Whilst the sample sizes are low, these data may prove useful to medical physics experts (MPE) and multi disciplinary teams undertaking local examination protocol optimisation in providing a broad range of dosimetric values for examinations performed across the UK.

Other methods may be worth considering where the examination frequency is low, such as the successful implementation of reference curves described by Almén et al.²²

Proposed National Diagnostic Reference Levels

The working party propose adoption of the following national DRLs for head examinations (Table 24) and chest examinations (Table 25). In line with the recommendations of the ICRP,⁵ these are proposed at the third quartile of the distribution of median values. For ease of use, values for head examinations are rounded to the nearest five for CTDI_vol and nearest 10 for DLP and values for chest examinations are rounded to the nearest 0.1 for CTDI_vol and nearest one for DLP for the 5–<15 kg weight range and the nearest 0.5 for CTDI_vol and nearest five for DLP for all other weight ranges.

Table 24.

Proposed national DRLs for paediatric head CT examinations

Age range (yrs)	CTDIvol (mGy)	DLP (mGy.cm)
0–<1	20	290
1–<5	25	420
5–<10	35	570
10–<15	45	690
15–<18	Use updated adult National DRLs

CTDI, CT dose index; DLP, dose–length product; DRL, diagnostic reference level.

Table 25.

Proposed national DRLs for paediatric chest CT examinations

Weight range (kg)	CTDIvol (mGy)	DLP (mGy.cm)
5–<15	0.9	18
15–<30	2.0	40
30–<50	2.5	60
50–<80	4.0	115

CTDI, CT dose index; DLP, dose–length product; DRL, diagnostic reference level.

Note that no national DRLs are proposed for the 0–<5 kg weight range for CT chest examinations; this is because the data in this survey has come from too few CT scanners to be considered representative. The data presented for the 0–<5 kg weight range in Tables 14 and 15 are a useful comparator however.

Conclusions

A national dose audit of paediatric CT head, chest, abdomen-pelvis and cervical spine examinations throughout the UK has been undertaken by an IPEM working party in collaboration with PHE. The working party has been able to recommend updated national DRLs for head examinations based on patient age ranges and, first of a kind, national DRLs for paediatric chest examinations based on patient weight ranges. Once adopted, the availability of these national DRLs will assist with CT protocol optimisation and, with the practice of routinely recording patient weight for paediatric CT chest examinations, to allow local performance to be compared with national DRLs.

Although there were insufficient data received to propose national DRLs for abdomen-pelvis and cervical spine examinations, the data that were received have been presented to offer some comparator for sites looking to optimise these examinations.

In addition to the proposed national DRLs, analysis that may assist with local protocol optimisation has been presented. By looking at the relationship between examination kV_p, reconstruction method and (for head examinations) the use of automatic exposure control, with median and third quartile CTDI_vol and DLP, sites have information about the general preferences across the UK and the resultant dosimetric effect against which to review their local protocols.