Positional Challenges of the Anteroposterior Pelvic X‐Ray: Comparison of Imaging Reject Rates Between Trauma Trolley and Table Bucky

ABSTRACT

Introduction

Pelvic x‐rays can be conducted on a trauma trolley or conventional table bucky. The aim of this study was to compare the positional challenges and reject rate between pelvic x‐ray images taken on a trauma trolley and a table bucky during a 12‐month period in an Australian public metropolitan hospital's emergency department and to determine the accuracy rate of anatomical inclusion via a qualitative assessment of pelvic x‐rays using a modified Visual Grading Scale (VGS).

Methods

A retrospective clinical audit of pelvic x‐ray image reject rates over a 12‐month period was conducted for an emergency department at an Australian hospital. Reject rate and anatomical cut‐off were compared between images taken on a trauma trolley and a table bucky using independent samples t‐test.

Results

A total of 1847 patients who underwent pelvic x‐ray examinations were included in the study. The mean reject rate and the first exposure accuracy of pelvis x‐rays taken on a trauma trolley were 35.5% and 56.7% respectively, while the mean reject rate and the first exposure accuracy for images taken on a table bucky were 18.8% and 81.8%, respectively (p < 0.01). The superior and lateral anatomy cut‐off were the major causes of image rejection for both techniques.

Conclusions

Pelvic x‐rays taken on a trauma trolley had a significantly higher reject rate and lower first exposure accuracy compared with those taken on an x‐ray table. Future studies could involve implementing strategies to reduce the reject rate of pelvic x‐rays taken on trauma trolleys.

Pelvic radiographs taken on a trauma trolley had a significantly higher reject rate and lower first exposure accuracy compared with those taken on an x‐ray table. Future studies could involve implementing strategies to reduce the reject rate of pelvic radiographs taken on trauma trolleys.

1. Introduction

The pelvic x‐ray is a diagnostic x‐ray image that demonstrates the bony anatomy of the pelvis. It encompasses the bones of the ilium, ischium and pubis, the sacral and coccygeal spine and the proximal femora. Clinically, a pelvis x‐ray is commonly acquired in the trauma x‐ray series to identify pelvic fractures. Pelvic fractures are serious orthopaedic injuries due to the role of the pelvis in supporting the body. They make up about 3% of all bone fractures and are found in 9.3% of blunt trauma hospital admissions, often caused by car accidents or falls [1, 2]. These injuries have a high risk of complications, with mortality rates ranging from 10% to 16% [1]. Open pelvic fractures are even more severe, with mortality rates between 18% and 30.2% [1], highlighting the need for urgent medical intervention. In the emergency department setting, it is commonly performed while a patient is on a trauma trolley due to the potential risk of injury exacerbation in transferring the trauma patient to the x‐ray tabletop [3]. In nontrauma cases, pelvic x‐rays are most commonly performed using the table bucky. Pelvic x‐rays can also be performed using the erect bucky to assess the weight‐bearing capability of the hip.

The image quality of a pelvic x‐ray taken on the trauma trolley can be influenced by unique features of the trauma trolley, including the mattress' construction and thickness, the image receptor (IR) holder, the object‐to‐IR distance, the source‐to‐image distance and the use of a stationary antiscatter grid. These distinct factors necessitate modified acquisition settings for pelvic x‐ray on trauma trolley, deviating from those employed in conventional table bucky imaging [4]. As for the patient positioning, patients rarely align perfectly along the central axis of the trauma trolley, frequently assuming an oblique position. Based on anecdotal observation, this misalignment often leads to repeated x‐ray examinations due to anatomical cut‐off.

When the image quality of the x‐ray is suboptimal for any reason, including anatomy cut‐off, radiographers may decide to reject the image and retake it. While the European Guidelines on Quality Criteria for Diagnostic Radiographic Images by the Commission of European Communities (CEC) in 1996 serve as a reference, the decision to reject an image can differ among radiographers due to the inherent subjectivity of the evaluation process [5]. Furthermore, no substantial evidence was presented with these CEC guidelines to validate their accuracy. Given that these guidelines originated in the film‐screen era, several criteria required updates to be compatible with the advancement of digital x‐ray technology. In an attempt to overcome the limitation of CEC guidelines, Mraity et al. utilised psychometric theory to create and validate a visual grading scale (VGS) that evaluated the visual perception of image quality for anteroposterior (AP) pelvis x‐ray [6]. They demonstrated that the VGS possesses both reliability and validity when evaluating the image quality of the AP pelvic x‐ray.

The reject rate of x‐rays is defined as the ratio of rejected images to the total number of images taken [7]. The reject rate can vary based on the location of the imaging, the specific body part being imaged and the experience level of the radiographer. Owing to the inherent subjectivity and the variation in radiographic techniques employed, there is a notable discrepancy in reject rates across diverse studies as shown in Table 1. The comprehensive reject rate spans from approximately 0.02% to 15%. Specifically, the reject rate for pelvis x‐rays vary between 5% and 23%, which is higher than the overall reject rate in the majority of studies. This can be partially attributed to the intricacies inherent to this particular projection when performed on a trauma trolley.

TABLE 1.

Summary of previous studies exploring reject rate.

Research group	Country	Sampling (n)	Reject rate (%)	Major cause of reject
Alashban et al. [8]	Saudi Arabia	27,238	9%	Positioning, Anatomical cut‐off
			(Pelvis 20%)
Foos et al. [9]	USA	288,000	4.4%–4.9%	positioning, Anatomical cut‐off
			(Pelvis 8%)
Jones et al. [10]	USA	66,063	8%–10%	Positioning, exposure
			(Pelvis 5.2%)
Stephenson‐Smith et al. [11]	Australia	11,596	10.3%	Positioning, anatomical cut‐off
			(Pelvis 22.5%)
Atkinson et al. [5]	Australia	90,298	9%	Positioning, anatomical cut‐off
			(Pelvis 23%)
Parker et al. [12]	UK	1531	Pelvis 14.7%	—
Rastegar et al. [13]	Iran	14,022	8%	Positioning, patient preparation
Tzeng et al. [14]	Taiwan	4168	4.3%	Positioning, artefact
			(Pelvis 5.1%)
Bantas et al. [15]	New Zealand	76,325	5.9%–7.9%	Positioning
			(Pelvis 5%–14%)
Hofmann et al. [16]	Norway	5417	11%	Positioning, centring
			(Pelvis 8.2%)
Andersen et al. [17]	Norway	27,284	12%	Positioning
			(Pelvis 13%)

Abbreviations: UK, United Kingdom; USA, United States of America.

To assess the challenges of performing pelvic x‐ray on trauma trolleys in an emergency department, this study aimed to compare the reject rates of pelvic x‐rays taken on trauma trolleys and conventional table bucky systems over a 12‐month period. Additionally, the study aimed to determine the accuracy rate of anatomical inclusion via a qualitative assessment of pelvic x‐rays using a modified VGS.

2. Method

2.1. Study Design and Ethical Approval

A retrospective clinical audit was conducted through a review of data collected from January 1 to December 31, 2023. The Western Sydney Local Health District approved an ethical exemption for this study under a quality assurance protocol (reference number: 2401‐03 QA).

2.2. Setting and Equipment

The x‐ray equipment used in the hospital was a ceiling‐mounted PHILIPS 7000C—DigitalDiagnost C90 x‐ray console, featuring a fixed table bucky detector (43 cm × 43 cm) and a Philips Skyplate wireless detector (35 cm × 43 cm) compatible with a trauma trolley. Additionally, a Philips MobileDiagnost with SkyPlate wireless detector (35 cm × 43 cm) was utilised for imaging in the emergency resuscitation/trauma bays. All pelvis x‐rays included in the data were performed only in the emergency department. Data collection (Appendix S1) was carried out using reject analysis tools integrated within the Sectra Picture Archive and Communication System (PACS).

The Philips x‐ray systems include embedded software that produces DICOM data from all acquired images, including detector ID. These data were utilised during the review to determine the technique used in each examination. In practice, when an image is due for rejection, the radiographer is prompted to select a reason for rejection from a list of options, such as anatomy cut‐off. All rejected and clinically accepted images were sent to be stored on PACS, with rejected images cached separately. The procedure described in supplement I details the image data collection for this retrospective study.

2.3. Data Collection and Inclusion Criteria

The images in this study were collected for a 12‐month period (January to December 2023). This sampling period was chosen to average out variability in the skill levels among different radiographers working across a large radiology department, ensuring adequate statistical power. Imaging was performed across the study period by approximately 50 radiographers and 32 student radiographers, data were not collected regarding radiographers' experience level. The inclusion criteria were all x‐ray examinations/images of patients who were clinically referred to the radiology department for pelvis x‐ray imaging as part of their diagnosis and treatment in the emergency department. If a patient is referred to the radiology department on a trauma trolley due to blunt trauma, the pelvic x‐ray is performed directly on the trolley. In all other cases, the examination is conducted on the table bucky. Included images were performed on a trauma trolley with wireless detectors and those performed on a table bucky with a fixed detector. Pelvis x‐rays performed as part of multiregion request codes, e.g., ‘Pelvis and Hip X‐ray’, were excluded from the study. Pelvis and hip x‐rays have different centring points. Bontrager suggests centring midway between the anterior superior iliac spine (ASIS) and the pubic symphysis for a pelvis x‐ray, whereas for a hip x‐ray, the pubic symphysis or greater trochanter is suggested as the centring point [18]. The hip x‐rays were excluded from this study because they do not strictly require inclusion of the iliac crests. For example, Mellor et al. reported that the difference in the centring point can be as large as 51 mm. The rejection decisions were made immediately after the examination by the radiographers who performed it. The assessment of anatomical cut‐off was categorised by a single observer. Examinations that did not specify the detector type in the DICOM were also excluded from the analysis.

2.4. Data Analysis

Descriptive analysis was undertaken, including frequency count and percentages for all variables [19, 20]. Means and standard deviation were calculated for variables with a normal distribution. Reject rate, first exposure accuracy and anatomy cut‐off were compared and analysed between trauma trolley and table bucky techniques. Additionally, the impact of other independent variables such as patient age, patient sex, presence of orthopaedic hardware and examination time were also investigated. Secondary data analysis was conducted using standard descriptive statistics with Microsoft Excel and IBM SPSS Statistics 29. The overall reject rate was calculated by dividing the total number of rejected images by the total number of images included in the study and expressing the result as a percentage.

$$\text{Reject Rate}\left(\%\right)=\frac{\text{number of rejected images}}{\text{total number of aquired images}}\times 100$$

Pelvic x‐rays can be rejected for various reasons, including anatomy cut‐off, artefacts and suboptimal exposure. However, this study focused on the reject rate due to anatomy cut‐off, which was identified as the leading cause of rejection in previous studies [15].

Additionally, the reject rate was calculated on a weekly basis to statistically compare the distribution of reject rates between the trauma trolley and table bucky techniques during the study period. The first exposure accuracy, defined as the ratio of the number of accepted images on the first exposure to the total number of first exposure images, was introduced to assess the positioning difficulty in pelvis x‐ray and compare the two techniques.

$$\text{First Exposure Accuracy}\left(\%\right)=\frac{\text{number of accepted images}\mathrm{at}\text{first exposure}}{\text{total number of first exposure images}}\times 100$$

When an image was rejected due to anatomy cut‐off, it was categorised into one of five groups: (1) superior cut‐off; (2) inferior cut‐off; (3) lateral cut‐off; (4) superior and lateral cut‐off; and (5) inferior and lateral cut‐off.

The reject rate was also analysed based on the patient's age, sex, presence of fractures or orthopaedic hardware and examination time to determine whether these factors could influence the reject rates. The independent samples t‐test was used to compare the results between trauma trolleys and the table bucky. The p value was set to 0.05.

3. Results

Across the 12‐month study period, a total of 2688 images from 1847 patients met the inclusion criteria and were included in the analysis. Of these patients, 67.4% (n = 1245) underwent pelvic x‐ray on a trauma trolley, while 32.6% (n = 602) were imaged using a table bucky. The total number of images and rejected images are summarised in Table 2.

TABLE 2.

Descriptive statistics of rejected images for trauma trolley and table bucky.

	Trauma trolley	Table bucky	Total
Patients (n)	1245	602	1847
Images (n)	1930	758	2688
Images accepted at first exposure (n)	756	489	1245
Rejected images (n)	685	156	841
Reject rate (%)	35.5%	20.6%	31.3%
First exposure accuracy (%)	60.7%	81.2%	67.4%

The reject rate for pelvic x‐rays was generally higher on a trauma trolley compared to a table bucky across most data points, as shown in Figure 1. The mean reject rate was 35.5% for the trauma trolley and 18.8% for the table bucky.

FIGURE 1.

Reject rate (percentage) between pelvic x‐ray images taken on trauma trolley and table bucky during the study period (52 weeks).

The independent samples t‐test was conducted to find a statistically significant difference in reject rate between the trauma trolley and table bucky techniques (mean difference = 16.7; 95% CI: 13.4, 19.9; p < 0.001). The standard error (SE) for reject rate was 2.08, considerably smaller than the mean value. This low SE contributed to higher statistical power, reducing the likelihood of Type II error, where a false null hypothesis might otherwise not be rejected. The effect size was assessed using glass's delta because the independent variable has different variances in two groups. The effect size was medium (Glass's delta = 1.536, 95% CI: 1.045, 2.017). The distribution of reject rate for trauma trolley and table bucky resulted in median values of 35.2% and 18.5%, respectively.

3.1. First Exposure Accuracy

The first exposure accuracy per week across the study period for trauma trolley and table bucky are shown in Figure 2. The first exposure accuracy for pelvic x‐rays was generally lower on a trauma trolley compared with a table bucky across most data points. The mean first exposure accuracy was 56.7% for the trauma trolley and 81.8% for the table bucky.

FIGURE 2.

First exposure accuracy (percentage) of images taken on the trauma trolley and table bucky during the study period (52 weeks).

A statistically significant difference in first exposure accuracy between trauma trolley and table bucky technique was found (mean difference = −24.2; 95% CI: −28.3, −20.1; p < 0.001). The standard error (SE) for the reject rate was 2.08, considerably smaller than the mean value. The effect size was medium (Glass's delta = −2.09, 95% CI: −2.648, −1.531). The first exposure accuracy for trauma trolley and table bucky had median values of 56.7% and 81.8% respectively.

3.2. Anatomy Cut‐Off

The relative proportion of anatomy cut‐off is shown for the trauma trolley and table bucky techniques in Figure 3. Both techniques demonstrated a similar trend, with the highest proportions in the superior and lateral cut‐offs. Images taken in the trauma trolley showed the anatomy cut‐off proportion to be highest in superior direction (36%), followed by lateral (34%), inferior (13%), superior–lateral (12%) and inferior–lateral (5%) directions. In comparison, the table bucky had highest proportion of cut‐off in the lateral direction (42%), followed by superior (40%), inferior (12%), superior–lateral (3%) and inferior–lateral (3%) directions.

FIGURE 3.

Relative proportion (percentage) of anatomy cut‐off for images taken on the trauma trolley and table bucky, categorised into five cut‐off groups.

3.3. Impact of Patient Factors: Patient Age, Sex and Presence of Orthopaedic Hardware

Patients were divided into 10‐year age groups, and their reject rates were compared. As shown in Figure 4A, no noticeable relationship was observed between patients' age group and the reject rate. The box plot in Figure 4B illustrates the reject rate for male and female patients, showing median values of 33.8% and 28.6%, respectively. The impact of the patient's sex on the reject rate was investigated using the independent samples t‐test. The independent samples t‐test found a difference in reject rate between the trauma trolley and table bucky technique (mean difference = 5.46; 95% CI: 2.69, 8.23; p < 0.001). The standard error (SE) for reject rate was 1.4, considerably smaller than the mean value. The effect size was assessed using Cohen's d because the independent variable has equal variance in two groups in Levene's test. The effect size was large (Cohen's d = 0.770, 95% CI: 0.368, 1.168).

FIGURE 4.

Impact of patient‐specific factors on reject rate depending on (A) patient age, (B) patient sex and (C) presence of orthopaedic hardware.

With two independent variables, technique and patient sex, affecting the same dependent variable, reject rate, a multiple linear regression was conducted to assess the sensitivity of each variable on the reject rate, based on the model:

$$\text{Reject rate}=a+b\times \text{technique}+c\times \text{patient}\mathrm{sex}$$

The multiple linear regression analysis yielded values of b = −0.696 and c = 0.071, indicating that technique has a much more sensitive impact on the reject rate compared to patient sex.

The box plot in Figure 4C shows the reject rate for patients with and without the presence of orthopaedic hardware, showing median values of 33.3% and 31.4%, respectively. The independent samples t‐test in this study did not find a statistically significant difference in the reject rate between samples (mean difference = 0.99; 95% CI: −3.41, 5.40; p = 0.303). The standard error (SE) for the reject rate was 2.15, considerably smaller than the mean value.

3.4. Time of Examination

The box plot in Figure 5A illustrates the reject rate for daytime (08:00–19:59 h) and night time (20:00–07:59 h), with median values of 30.2% and 32.8%, respectively. The independent samples t‐test did not find a difference in reject rate between the pelvic x‐rays performed during the daytime and night time (mean difference = −1.41; 95% CI: −4.12, 1.29; p = 0.653). The standard error (SE) for reject rate was 2.2, considerably smaller than the mean value. Figure 5B shows the reject rate by the examination time of the day, indicating no significant correlation between the time of examination and the reject rate (R ² = 0.02).

FIGURE 5.

Impact of examination time on reject rate (percentage). (A) Daytime: 08:00–19:59 h; night time: 20:00–07:59 h, (B) examination hours.

4. Discussion

The findings from this study showed that pelvic x‐rays taken on a trauma trolley had a consistently higher reject rate compared with those taken on a table bucky across a 12‐month period. This result confirms that the trauma trolley technique presents greater challenges in achieving optimal image quality on the first attempt. This may have been due to a range of factors, including the additional patient positioning challenges of performing a pelvic x‐ray on a trauma trolley, where the radiographer must align the x‐ray tube, detector and patient independently, although this can only be inferred from the data collected in this study. One key difference between the images acquired on the trauma trolley and table bucky in this study were the size of detector used, where larger free detectors (43 cm × 43 cm) were used in the table bucky, compared with the smaller detectors (35 cm × 43 cm) used in trauma beds. It is possible that use of larger detectors, with larger field of view on the trauma trolley may decrease repeat exposures to patients due to anatomy cut‐off. Hence, results from this study may be used to guide equipment configurations in future procurement decisions for imaging in emergency departments. Standardising the positioning protocol with consistent use of anatomical landmarks for accurate centring of x‐ray tube, peer double‐check system and conducting routine audits with individual feedback can support ongoing reduction in reject rate.

An important finding from this study was that the reject rate was higher than that of pelvic x‐rays in most previous studies, which range from 5.2% to 23% (Table 1). The reason for this difference may stem from the variations in sample size, inclusion criteria and clinical environment from where data were collected in previous studies. Of note, data in the current study were collected exclusively from the emergency department, where patient presentation may affect imaging outcomes. In contrast, previous studies often included collated data from across various areas of the hospital setting. For example, Foos et al. analysed 288,000 samples from two hospitals and found an 8% reject rate for pelvic images [7].

With respect to the anatomy cut‐off, this study categorised anatomical cut‐offs into five distinct groups and analysed the frequency of each cut‐off, comparing the trauma trolley and table bucky techniques. Results showed that both the trauma trolley and table bucky images had the highest cut‐off rates at the superior and lateral margins, which represented 70% of cut‐off anatomy for the trauma trolley and 82% of cut‐off anatomy for the table bucky. For each of the other cut‐off directions evaluated in this study (inferior, superior–lateral and inferior–lateral), the trauma trolley had a higher proportion of anatomy cut‐off compared with images acquired in the table bucky. These findings indicate that while both techniques present challenges in achieving full anatomical coverage, the trauma trolley introduces more variability in positioning errors. Unlike most prior studies that did not specify missing anatomical details [6, 7, 8, 9, 10, 11, 12, 13, 14, 15], this study highlights that the superior and lateral sections are more frequently missing.

As shown in Figure 3, superior cut‐off was the most common anatomical cut‐off in trauma trolley technique. This result can be attributed to a range of factors, in addition to the inherent difficulty associated with the trauma trolley technique. Firstly, the trauma trolley technique uses a rectangular detector with a shorter length in the vertical direction, making it easier to miss the superior or inferior parts of the pelvis during examination. The centring point for the projection is another potential contributing factor. A study conducted by Snaith et al. reported that one of the reasons for the higher reject rate in pelvic x‐rays compared with other body parts is the lack of a standardised centring point between different hospitals and radiographers [21]. A similar explanation applies to this study, as it is likely that individual radiographers use different centring points, which can potentially increase the likelihood of anatomy cut‐off in the vertical direction. For instance, some radiographers use 5 cm below the anterior superior iliac spine (ASIS) as their centring point, while others do not prioritise a specific centring point, instead focusing on the inclusion of the iliac crest on the detector. Another contributing factor for anatomy cut‐off in a trauma trolley may be due to magnification factors. The source‐to‐image distance (SID) is not standardised during trauma trolley imaging, and hence SID is highly variable compared with the fixed settings used in the table bucky. Also, object‐to‐image distance (OID) is greater in the trauma trolley, causing further magnification. However, none of these factors justify why the inferior cut‐off is less common than the superior cut‐off in this study.

In previous studies, it has been reported that the reject rate was higher when x‐ray examinations were performed by less experienced radiographers [8] or outside regular working hours [12]. This is partly due to the possibility that examinations conducted outside working hours may be handled by a single radiographer. While the current study did not evaluate radiographers' experience, data from the study found no significant relationship between examinations conducted during different times and reject rate (Figure 5). Furthermore, the hospital protocol of involving more than two radiographers in each examination is intended to minimise the impact of any individual radiographer's experience on the reject rate.

4.1. Limitations

A notable limitation in comparing the two techniques in this study is due to the retrospective study design, where independent variables could not be controlled. For instance, the images used in this study for the trauma trolley technique were acquired using a smaller detector compared with the trauma trolley technique, yielding a smaller area that the anatomy can be projected onto. In addition, results from this study are limited to techniques undertaken within a single clinical setting, which aligned with the aims of this study, yet may limit generalisability to other hospital areas. It is also important to note that this study lacks inter‐rater and intra‐rater reliability testing, which represents a limitation in methodology of this study. This is because the decision regarding anatomy cut‐off was made by a single observer, which may introduce bias and affect the consistency and generalisability of the findings. However, given the lack of available evidence in this area, this study's findings provide significant and relevant insights into pelvic imaging using two techniques in the emergency department.

5. Conclusion

This study retrospectively compared reject rate, first exposure accuracy and anatomy cut‐off between pelvis x‐rays performed on trauma trolley and table bucky techniques. The results of this study underscore the technical challenges associated with performing pelvic x‐rays on a trauma trolley. The increased reject rate and lower first exposure accuracy suggest that additional training or modifications to workflow and equipment may be necessary to improve image quality in emergency departments.

The clinical implication of this study highlights the need for improved imaging practices in trauma settings. High reject rates can compromise diagnostic accuracy, increase patient radiation exposure and delay treatment. The findings may support the development of better positioning protocols, staff training and investment in more compatible equipment to reduce image rejection. For example, the radiology department clearly distinguishes between the examination codes and centring points for pelvic and hip x‐rays to ensure clarity in the imaging process. Given the more frequent superior anatomy cut‐off observed in this study, the centring point for pelvic x‐rays should be positioned at or above the midpoint between the ASIS and the pubic symphysis. In terms of hardware, larger detectors and positioning aids can be introduced to improve imaging accuracy and reduce the risk of anatomical cut‐off. By addressing this issue, radiology departments can enhance workflow efficiency, improve patient safety, and ensure higher quality of care in emergency situations. Future studies could explore strategies to enhance image acquisition and reduce variability in trauma imaging environments.

Ethics Statement

The Western Sydney Local Health District approved an ethical exemption for this study under a quality assurance protocol (reference number: 2401‐03 QA).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Appendix S1.

Data Availability Statement

The data sets used and/or analysed during the current study are available from the corresponding author upon reasonable request.