Digital radiography reject analysis of examinations with multiple rejects: an Australian emergency imaging department clinical audit

Introduction

Diagnostic X-ray examinations account for the largest source (over 50%) of manmade ionising radiation exposure to the public.¹ Recent literature has found patients who undergo multiple X-ray examinations have an increased risk of acquiring malignancy in the future.^1-4 Currently, there is no established threshold indicating how much radiation is thought to be completely safe.² This highlights the importance of dose optimisation by radiographers. Maintaining a high standard of image quality whilst minimising radiation exposure to As Low As Reasonably Achievable (ALARA) is a core principle of radiography.³

In digital radiography (DR), reject analysis (commonly referred to as retake or repeat analysis) is a method of quality assurance that aims to examine images rejected by radiographers, to determine how many and why particular images are being rejected.⁵ A rejected image is referred to as a radiograph that is discarded by the radiographer because it does not contribute any value to answering the clinical question.⁵ However, a radiographer may delete multiple (two or more) radiographs before acquiring the desired image. These deleted radiographs are referred to as multiple rejects. Some common reasons for rejection include inadequate patient positioning, anatomy cut-off, under or over exposure, patient motion or artefact.

Although more efficient, modern digital radiography systems have one major downfall. The era of conventional radiography almost saw the retirement of reject analysis, with the concept of digital radiography limiting the need for it.^{6, 7} Previously, the most significant reason for rejection in conventional radiography (CR) was under or over exposure.^6-8 Digital radiography manufacturers emphasised that the evolution X-ray equipment would significantly reduce the need for reject analysis since under or over exposure in DR is considerably more forgiving.⁶ Despite this, it has been proposed that a spike in reject rates after the changeover from CR to DR is attributed to it being easier than ever to obtain and discard radiographs.⁹ The departments that still routinely conduct reject analysis, often just skim the surface by determining the overall reject rate, and the reject rates for each body part.

In recent years, radiographers who perform quality assurance have advocated reject analysis as an essential tool in digital radiography.⁶ It has been reported that close to half (55%) of rejected images were not digitally rectifiable and suggested that these figures cannot be neglected.¹⁰ Almost all relevant studies pertaining to DR found positioning to be the primary reason for image rejection, followed by anatomy cut-off.^{6, 9, 11-13} Evidently, the need for reject analysis still exists.

Among the reviewed studies, a common trend of pelvis, hip, spine and knee examinations acquiring the highest reject rates was identified.^10-14 However, there was evidence to suggest that despite some noteworthy similarities between the literatures, the results cannot be applied across all radiography departments due to their incomparable factors. There are currently no guidelines for reject rates in radiography. This may be due to the fact that data can be skewed by a number of department-dependent factors.^{13, 15, 16} Several studies have concluded that the weekly or monthly reject rates varied significantly.^{13, 15, 16} It was suggested that this deviation is most likely attributed to irregularities in room utilisation.¹⁵ Despite this variability, the overarching purpose of reject analysis is to investigate and amend department-specific issues in order to correct staff weaknesses and thus, reject analysis is still viable.⁶Whilst several studies have investigated reject rates and common reasons for rejection, no study has explored multiple reject rates. Only two studies have reported on projection-specific reject rates, with most merely focused on whole body part examinations, limiting their practicality.^{13, 17} The purpose of this study was to analyse projection-specific reject rates and DR examinations with multiple rejects. This study will highlight specific projections that require targeted clinical education, thus decreasing unnecessary radiation exposure.

Methods

Results

Discussion

This research has established a previously unexplored area of reject analysis: radiographs rejected multiple times in one examination. Rejected images were categorised by body part and further categorised by projections. The reject rates and multiple reject rates for each were reported. It was identified that some projections yielded reject rates as high as 50.0% and multiple reject rates as high as 12.1%. The reject rates were used congruently with the reasons for rejection to highlight areas of clinical concern. Projections with high multiple reject rates were noteworthy as radiographers could not correct the issue after one attempt. The intrinsic effects of repeated radiographs go beyond the scope of unnecessary equipment usage and labour; avoidable exposures increase patient radiation dose which is known to have adverse effects on the human body.⁸ Focused clinical education on these projections may be beneficial in reducing reject rates and thus patient radiation exposure.

The overall reject rate (10.3%) identified in this study was slightly high compared to recent literature reporting on modern digital radiography systems, reporting rates of 9.0%, 4.9%, 11.0%, 8.0% and 8.9%.^{12, 14, 17-19} There are no current guidelines for reject rates in radiography, due to the fact that data can be skewed by a number of department-dependent factors such as variation in protocols and equipment.^{13, 15} Watkinson, Moores and Hill¹⁵ suggested that weekly reject rates varied by a factor of two and the reject rates for some examinations fluctuated up to five-fold. Similarly, Tzeng et al.¹³ suggested that generalisability is influenced by many factors including the types of examinations performed and the skill level of each radiographer in the department. Although there is some variability in reject analysis, the essence of reject analysis is to identify department-specific issues in order to plan for training needs to target staff weaknesses, as established by Foos et al.¹¹

Reject rates are subjective; their statistical significance is dependent on the relative frequency of occurrence. For example, if four ankle radiographs were performed and one of these was rejected, the reject rate would be 25.0%. On the other hand, if forty ankle radiographs were performed at another department and one of these was rejected, the reject rate would be 2.5%. For less common examinations, the reject rates were significantly higher, however due to there being fewer examinations, the statistical uncertainty was greater. Thus, reject rates as values may not necessarily be comparable between studies as they are not equal in significance and hold a high risk of bias. However, it is valuable to recognise which radiographs are commonly repeated by recognising both a high number of rejects and a high reject rate. Otherwise, they may be highly susceptible to misinterpretation.

The projections with both a high number of rejects and reject rate were AP chest (175, 18.1%), AP pelvis (78, 22.5%), horizontal beam hip (61, 33.5%) and horizontal beam knee (116, 30.5%). The study by Tzeng et al.¹³ yielded similar results by concluding that AP pelvis radiographs accumulated the highest reject rate among all radiographic examinations. The study by Hofmann, Rosanowsky, Jensen and Wah¹² indicated that knee and hip examinations were of the highest reject rates (20.6% and 18.5%, respectively). The clinical audit by Jabbari, Zeinali and Rahmatnezhad¹⁰ which investigated reject rates in three Iranian medical imaging departments similarly concluded that pelvis radiographs were of the highest repeat rate (14.0%). The clinical audit by Foos et al.¹¹ which explored the reject rates of two hospitals, agreed with these results by concluding that pelvis, hip and spine examinations were of the highest reject rates (above 8.0% of all these examinations were rejected).

Positioning errors (58.0%) and anatomy cut-off (18.3%) were the top reasons for rejection. This is consistent with recent literature that affirms that since DR was introduced, the overarching reason for rejection has shifted from exposure to positioning.^16-21

The current study was unique in that it investigated multiple reject rates in a single examination, which has not been explored in the existing literature. The projections with both a high frequency and multiple reject rate were horizontal beam knee (32, 8.4%) and horizontal beam hip (17, 9.3%). This finding suggests that the performing radiographers are not correcting the image on the second attempt, thus contributing to an increase in radiation dose to the patient. The reasons for such rejections should be noted. The top reasons for multiple rejects were positioning (112 images, 67.1%), anatomy cut-off (14 images, 8.4%) and incorrect detector selected (12 images, 7.2%). These results suggest that radiographers are unsure of how to correct the position after the initial image, with some attempting images up to five times. This figure cannot be compared with prior literature since multiple reject rates have not previously been explored.

The positional correction of horizontal beam knee images often consists of minute adjustments. However, a radiographer should be able to accurately correct the position on the second attempt from visual assessment of the initial image. Staff training on positioning patients for these projections is likely to be of benefit. An educational in-service reviewing pertinent radiographic knee anatomy and related positioning on X-ray is recommended to decrease multiple reject rates, particularly for patients undergoing these examinations in follow-up examinations. Horizontal beam hip radiographs can be difficult to achieve due to several factors including the physical limitations of equipment and patient cooperation. However, staff education surrounding the optimisation of positioning techniques could be advantageous. As a result, the amount of multiple rejects should decrease and subsequently minimise patient dose, adhering to the ‘As Low as Reasonably Achievable’ (ALARA) principle.

Several limitations are worthy of consideration in this study. Firstly, there was a high variability in the number of images acquired per radiographic projection within the sample, limiting the statistical significance. To address this in future studies, a larger sample size of more than six months would increase the viability of the study. Secondly, the department-specific differences in protocols and imaging systems limits the reproducibility of the study. Several innate confounders for reject rate included variation in projection complexity, radiographer experience, complex patient presentation, patient mobility, patient compliance, rooms/equipment utilised and patient body habitus. Thirdly, there was potential for reporting bias because radiographers must manually input the reason for rejection at the time of deletion. In times of urgency, accuracy here could be jeopardised. The study by Jabbari et al.¹⁰ stated that a number of images were labelled as ‘other’ but no more information was given, which resulted in invaluable data. Additionally, approximately 15–20% of cases in the study by Steffen et al.²², the reason for rejection was unavailable because they were not indicated by the radiographer. Within this study, ‘other’ was not an option for radiographers to select, so it is assumed that the correct reason for rejection was selected. However, it is possible that these manual inputs were not accurate, consequently introducing potential inconsistencies. Additionally, this study assumes that radiographers perform imaging under the correct projection label, that is lateral chests are imaged under ‘lateral chest’ rather than ‘PA chest’. Lastly, there was some ambiguity in the categories provided for reason for rejection. There was some crossover of terms that could be used interchangeably due to their subjective variability, that is anatomy cut-off could also be described as a positioning error. A degree of uncertainty in the inter-radiographer quality threshold was also acknowledged. It is unknown if low repeat rates are accredited to high quality work or low acceptance thresholds.

In the future, this study could be repeated with a larger sample size of at least six months to reduce the discrepancy in the number of images acquired per radiographic projection. Future research in this field could include investigating the relationship between radiographer experience and reject rates. Additionally, projections with high reject rates could be further explored by comparing images of the same projection acquired with the wall bucky to images acquired with the free detector or table; or comparing grid versus no grid chest X-ray reject rates. Another potential research study could involve re-auditing the department following the implementation of quality improvement strategies to decrease the reject rates.

Conclusion

This study achieved its intended aims by reporting the reject rates and multiple reject rates of each projection in order to provide evidenced-based research to assist departments in focusing on avoiding these multiple rejects. Projections that were frequently repeated in this study were horizontal beam knee and horizontal beam hip. The findings of this study can be used to lower radiation doses to patients presenting for X-rays by forming a base for designing future quality improvement initiatives. Additionally, this study has the potential to provide a benchmark for multiple reject rates of digital radiography systems alike. Future research could involve re-auditing the department following the implementation of recommended quality improvement strategies to reduce unnecessary radiation exposure.

Conflict of Interest

The authors declare no conflict of interest.