Optimisation as the incentive for this study
The three pillars of the radiation protection model of the International Commission on Radiological Protection (ICRP) are justification, optimisation and dose limits [32]. To comply with the second obligation, optimisation, X-ray procedures should be re-evaluated from time to time. We performed such an evaluation of pelvic radiography in children, using data already available in our Picture Archiving and Communication System (PACS) (Kodak Carestream Health, Rochester, NY). Organ and effective doses, and the effect of gonad shielding on these doses, were calculated using PCXMC, a Monte Carlo program for estimating patient doses in medical X-ray examinations (STUK—Radiation and Nuclear Safety Authority, Helsinki, Finland). These calculations showed that gonad shielding (1 mm Pbeq., Dr. Goos-Suprema, Heidelberg, Germany) gave a trivial reduction of the already very small radiation risk. At the same time, we often observed a serious loss of diagnostic information caused by improper gonad shielding. On the basis of these findings the radiological staff unanimously decided to discontinue gonad shielding in paediatric pelvic imaging using X-rays. Realising that our results might be of wider interest and appropriate for publication, we extended our initial investigation with more data, obtained both before and after discontinuation of gonad shielding. This extended study is the subject of this article; with respect to its execution the Medical Ethics Committee of the Maastricht University Medical Center formally acknowledged that no medical ethical approval was required.
Patients and X-ray projection
In the first retrospective study we investigated gonad shielding and patient dose in children aged 0–15 years. The children had a standard anterior-posterior (AP) pelvic radiograph in 2007 or 2008, mostly in relation to congenital hip dysplasia. The projection covered the region from the iliac crests to 2 cm below the trochanter minor on the femur. Following the majority of other investigators [22–26, 33, 34], we distinguished age groups of 0–1 years, 1–5 years, 5–10 years and 10–15 years. Additional radiographs were evaluated in the second retrospective study in 2010 after we had stopped the application of gonad shields.
Image acquisition and storage
The pelvic radiographs had been acquired in three rooms, each equipped with an X-ray system using a digital flat panel detector (AXIOM Aristos FX Plus; Siemens, Erlangen, Germany). These systems have pre-programmed protocols, which set the high voltage (kVP), additional filter, focus to image-detector distance, and either a fixed tube current-exposure time product (mAs value) or the automatic exposure control (AEC). The protocol indicates whether the bucky anti-scatter grid has to be present or absent, but grid insertion or removal has to be done manually. We have no special team of technicians for paediatric radiology within our department.
For the younger patients, age-dependent protocols with fixed technique parameters are used. For the 10–15 age group different protocols are applied because the children of this age vary strongly in size. For the smaller patients no grid and no AEC are used, for the group of more adult sized patients they often are.
After acquisition, the images are “shuttered”, i.e. irrelevant edges of the images are removed digitally. Subsequently the images are sent to our PACS in Digital Imaging and Communications in Medicine (DICOM) format. The DICOM header contains retrievable information on the exposure [e.g. kVP, filter, mAs value, use of the grid and the air kerma-area product (KAP)]. Information on the use of the AEC is not stored, neither is the size of the unshuttered field of view.
Evaluation of gonad shielding
Because the quality of gonad shielding potentially has a large impact on the diagnostic information in the image and the dose reduction as well, the shielding was visually evaluated (by M.J.F.). For boys, the gonad shield had to fully cover the testes, while all bony pelvic parts remained uncovered. For girls, the central section of the pelvis had to be covered, again without shielding parts of the pelvic bones, with the exception of the sacrum to just below the sacro-iliac joints. The radiograph made during the patient’s very first visit to the department was obtained without shielding.
Judging from the presence of poor images, retakes had been limited to the absolute minimum. In the evaluation, we therefore identified also all images in which important diagnostic landmarks were not visible due to shielding. Specifically we considered assessability of Shenton’s line, Perkin’s lines and the acetabular angle (see Fawcett et al. [20], their Fig. 3); these studies were labelled as “retake required”, although no retake was obtained in reality. Unfortunately, the real number of retakes could not be determined.
We also estimated the fraction of the testes that was actually shielded (percent coverage) by visual inspection of the radiographs. The corresponding testes dose reduction was calculated with PCXMC as the dose to the testes due to an exposure with the size of the gonad shield, with the field overlaying “percent coverage” of the testes in the cranial-caudal direction. For girls, we assumed that protection is not very sensitive to positioning of the shield (in a statistical sense at least), because of the large spread in the position of the ovaries [9].
Determination of patient dose
In the first dosimetric study (data from 2007 and 2008) gonad shielding was standard, in the second study (data from 2010) shielding was absent.
Together with the discontinuation of the gonad shielding, we continued to optimise our protocols. In the first place, more guidance for the imaging of larger children (10–15 years) was introduced because of large differences in exposure within this group. Secondly, following the European Guidelines [35], we added an additional copper filter of 0.1 mm for the two intermediate age groups. For the youngest patients (0–1 years) the decrease in image quality was considered too large by the paediatric radiologist; for the 10–15 age group, the filter was already used.
The basic filtration of the X-ray beam (i.e. in the absence of 0.1 mm Cu) was determined from the measured half-value layer in aluminium. Since most authors had reported the entrance dose as a dose-in-air including backscatter [22, 23, 25–31], we followed this convention. We estimated backscatter factors for the conversion of incident air kerma to entrance dose for each age group and radiation quality using data from the NRPB report R279 [36]. These factors were 1.26, 1.35, 1.39, 1.45 and 1.48 for the age groups 0–1 years, 1–5 years, 5–10 years, 10–15 years small childen and 10–15 years large children, respectively. The incident air kerma was derived by PCXMC from the KAP and the estimated unshuttered image size. Note that the dose in tissue, used in two publications [24, 33], is about 5% higher than the entrance dose as used by most other investigators and us.
Input information for PCXMC comprises imaging technique settings (kVp, filtration), KAP, image field size and projection. Note that these exposure parameters are generally independent of the use of gonad shielding, unless the shield was in front of an active AEC chamber. The effect of shielding is taken into account by subtracting a fraction of the gonad dose calculated for the no-shield exposure; under optimal circumstances this subtracted fraction amounts to 95% in male patients and 50% in female patients [7].
PCXMC has the option to adjust the size of the phantoms of all ages. The dose estimation in a specific age group of children will be illustrated using the 5–10 age group as an example. Rather than treating these children as a single group in the dose calculation, we considered 5- to 7.5-year-olds and 7.5- to 10-year-olds separately to reduce the potential effects of spread in patient size. By interpolation between data for child phantoms of different age present in PCXMC, we determined the weight and length of children of 6.25 and 8.75 years. Using the exposure parameters and interpolated patient size and weight for each subgroup, organ doses and effective dose were calculated with PCXMC for the 6.25- and 8.75-year-olds. The average dose for the 5–10 age group was then calculated by averaging the results for these two subgroups. Doses for other age groups were estimated in a similar fashion (always using two subgroups).
The problem of the collimator settings not being stored in the DICOM header (necessary as input on field size) was addressed by determining the average difference in width and height of the original (unshuttered) and shuttered images for a limited set of images for which both types of data were retained. Considerable variations in measured differences existed, but on average 1 cm had to be added in all four directions to compensate for the shuttering. Although this correction for shuttering is necessarily approximate, its application was preferred over leaving it out as this systematically would overestimate the entrance dose. To derive the effective dose from organ doses, the tissue-weighting factors published in ICRP 103 were chosen in PCXMC.
As gonad shielding is applied to limit hereditary effects, the detriment adjusted risk for heritable disease was estimated (the ‘detriment adjusted risk’ takes the various potential, negative consequences of the exposure into account, weighting for the seriousness of the corresponding harms; for the official definition see ICRP 103 [32]).
Cost-benefit calculation
Because we saw no solution to the problem of translating the loss of diagnostic information into a numerical risk that might be compared with the risk of radiation, we limited the quantitative risk analysis to radiation effects only.
Shortcomings of gonad shielding, which manifest themselves inevitably in daily practice, and which have an impact on the patient dose, are: (1) in males, an incomplete coverage of the testes, (2) the need for retakes because vital landmarks are obscured and (3) an increase in dose when an active AEC-chamber is (partly) shielded by the lead supposed to shield the gonads. Only point 1 was taken into account, the other two were not, because of missing information.
The detriment adjusted hereditary risk of a pelvic radiograph can be calculated using the nominal coefficient of 5.4 × 10-3 Sv-1 for persons of reproductive age [32]. For the computation of the total detriment adjusted risk, the risk coefficient for cancer (5.5 × 10-2 Sv-1 for the whole population [32]) with an age-dependent correction for the higher cancer risk for children [37] is needed as well. These calculations further require the effective dose, the testes dose and the dose to the ovaries. The reduction of hereditary risk achieved by gonad shielding was also expressed as a percentage of the total risk of a radiograph obtained without shielding.