Structured reporting of x-rays for atraumatic shoulder pain: advantages over free text?

Clear and unambiguous X-ray reports are a prerequisite for interdisciplinary patient management and enable the radiologist to add value to the clinical process of proper patient care. Comprehensive and precise reports are crucial to avoid misunderstandings and miscommunication between radiologists and referring physicians [1] which might even affect patient management negatively [2].

There have been statements and initiatives by several radiological societies on the potential and the preferable use of structured reports (SR) and standardized terminology such as RadLex, the reporting initiative of the Radiological Society of Northern America (RSNA) [3‐5] and the reporting guidelines issued by the European Society of Radiology [6].

There has been increasing evidence more recently that SR are preferred over free text reports (FTR) by both radiologists and referring physicians [7‐13]. Several previous studies evaluating the use of SR in different imaging modalities, including radiography, sonography, computed tomography (CT) and magnetic resonance imaging (MRI) [12‐16], provide evidence on the potential of SR with respect to completeness, accuracy, perceptions and satisfaction of the referring clinicians. Moreover, a survey among North American radiologists from academic teaching hospitals found out that approximately half of them use SR at least for some reports [17]. Thus, the existing studies point to the great potential of the implementation of SR in clinical settings. However, there are also some possible disadvantages of SR such as the risk of over-simplification, distraction by the introduction of additional SR tools, cumbersome decision trees and the possibility of reduced detection rates [18‐20]. Furthermore, the implementation of SR might lead to prolonged reporting times, especially in the beginning and among radiologists who are used to report by free speech dictation or free text entry [21].

Consequently, the use of SR up to now has not yet been widely established as clear evidence of the superiority of SR over FTR is lacking. Therefore, studies comparing template-based SR with FTR might provide further evidence for the advantage and potential of SR and facilitate the implementation process.

Chronic shoulder pain is common in the elderly population and X-ray examination is the first step in the diagnostic imaging process. To the best of our knowledge, no study has yet evaluated template-based SR compared to conventional FTR for the radiographic examination of degenerative diseases of the shoulder. Therefore, the aim of the current study was to compare the quality of SR and FTR regarding the radiographic examination of degenerative diseases of the shoulder.

We included X-ray exams of the shoulders of 31 patients. There was an anonymized FTR and SR (N = 62 reports in total) for each patient that were evaluated by two orthopaedic surgeons using the online survey tool (N = 124 completed questionnaires). The results of each item are reported as medians and interquartile ranges as the ratings are not normally distributed within the groups.

In our study, SR received significantly better ratings for all items compared to FTR. The SR were considered to be either of high or very high quality whereas FTR obtained significantly lower ratings for the overall quality of the reports. Furthermore, there was a great heterogeneity among the FTR, as indicated by the large interquartile range whereas the SR had consistently high ratings with very little variation.

As far as we are aware, this is the first study evaluating template-based SR compared to FTR in patients receiving an X-ray examination for the assessment of painful degenerative shoulder joint alterations. The results of a higher homogeneity among SR are in line with a previous study evaluating SR in CT examinations of patients suspected of pancreatic cancer [13]. Another study which aimed to compare the quality of SR in staging of rectal cancer with MRI to FTR displayed an improved quality when using SR [12]. According to our results, the template-based SR for the X-ray examination of shoulder pain also lead to an improved overall quality as perceived by the referring surgeons. As the ratings of SR and FTR differed significantly for all items, the results clearly implicate that the referring physicians who rated the reports in our study prefer SR for the radiographic evaluation of atraumatic shoulder pain. This is in line with previous studies that underline the preference for SR by referring physicians [9, 15].

Nevertheless, it cannot be simply assumed that the use of SR leads uniformly to an increased completeness, as illustrated by a study by Johnson et al.: The authors did not find a significant difference regarding completeness and accuracy scores for radiology residents’ SR and FTR of cranial MRI studies in patients suspected of having a stroke [10]. Therefore, it is crucial to validate each newly designed structured template.

Furthermore, the SR were rated as being more complete, especially with regard to content-related items. As this is crucial for further patient management, our template might assist with an improved radiologic report to enhance the clinical decision-making process and treatment of patients. Our results on the improved completeness of reports when using SR compared to FTR is in accordance with previous findings from a study evaluating SR versus FTR of chest X-ray exams and the study on reports of pancreatic cancer CT exams mentioned above [12, 13, 16]. However, some radiologists argue that an improved clarity might be at the expense of a comprehensive report that is able to adequately address the complexity of imaging findings [19].

A further advantage of providing SR is the simplified extraction of relevant information. Our template follows a standardized structure and order of the elements included, designed to make the reports comparable to each other with a high recognition value. Thus, one could speculate that, after a short implementation phase of getting used to the transition from FTR to SR, the referring physician would able to identify the relevant findings, resulting in time-saving without missing relevant information. Another point that deserves attention is the potential advantage for less experienced radiology residents for whom the templates of SR may serve as a kind of checklist, which has been shown to improve patient care in various medical fields [25‐28]. Whether the use of a clickable decision tree in SR might even lead to a decrease of reporting times for the radiologist, especially those more experienced, is currently being debated controversially. One study found short reporting times [29], whereas others found prolonged reporting times [10, 18] underlining the need for further studies.

Moreover, the use of an online-based template, such as in our case, can be seen an important prerequisite for vendor-neutral use by copying and pasting the SR generated into the current established radiologic reporting software. However, the use of an online-based template is not a prerequisite for the generation of SR and should only be seen as one means amongst others. To allow a smooth integration of structured reporting into the daily clinical routine, an integration of SR tools in the existing reporting software would be favourable.

Our study has a few limitations due to the study design. One limitation is that the reports were only evaluated by two orthopaedic surgeons (representing only one subspeciality) from two different university hospitals with a similar level of experience. It is, therefore not possible to generalize our findings towards a preference for SR of radiographic exams of the shoulder for all referring physicians. A review of the reports by a more varied group of providers would have been beneficial. There might be differences in the preference of SR compared to FTR depending on the specialty, the clinical setting (e.g. outpatient care) and level of expertise of the referring physician. Although the referring physicians evaluated all cases in one session in a randomized order, we do not believe that this possibly influenced the ratings, since only the report examinations were evaluated.

Secondly, we created a template that was tailored for patients with atraumatic shoulder pain who commonly present at our MSK outpatient clinic. Our aim was to create a template that is easy to use and to avoid that using the template becomes a cumbersome process by including several sublevels to accommodate all possible findings in X-rays of the shoulder. However, it is likely that the use of a more generalized template would further improve its potential for use in clinical practice. While our template is tailored for the evaluation of atraumatic shoulder pain it is not limited to patients with those complaints only and allows to report additional pathologies such as fractures. If there is no according clickable element predefined, the radiologist always has the possibility to make a free text entry. It has previously been pointed out that the introduction of several specialized templates might distract radiologist from the actual reporting [18]. On the other hand, there is evidence that specific templates (e.g. CT abdomen for suspected pancreatic cancer [13]) lead to a higher completeness of reports. We hypothesize that the solution may be a compromise between decision trees with many sublevels which cover a broad spectrum and tailored templates like the one we evaluated in our present study. Feasibility studies on the implementation of such templates into clinical routine will be necessary to further evaluate this aspect.

Furthermore, there is inherent bias in our study as we compared FTR acquired during clinical routine to SR that were generated in a research setting without time constraints. It is possible that a study that compared FTR and SR that were both generated in a research setting would reveal better ratings for FTR. The effect size when comparing FTR to SR in clinical routine is likely to be smaller than the one in our study. Nevertheless, we believe that our study can generate preliminary evidence on the quality of such templated-based SR. Such findings are crucial to justify the introduction of SR in radiological departments to test the feasibility in daily routine.

Keeping in mind the retrospective design of our study, we did not test the feasibility of implementing our template in clinical routine reporting. This aspect could be addressed in a prospective study. Future studies should also evaluate if and to what extend the introduction of SR into the clinical workflow confirms the concerns about potentially increased reporting times and the risk of distraction by using an additional structured reporting tool [20, 21].

In conclusion, our study has shown that the generation of template-based SR resulted in higher completeness and overall quality of radiologic reports on X-ray exams of patients with atraumatic shoulder pain. When compared to conventional free text reports template-based structured reports on X-ray exams of patients with atraumatic shoulder pain have the potential to provide more complete reports, to facilitate information extraction and to lead to improved overall report quality. Thus, structured reporting appears to be a promising tool to enhance interdisciplinary communication and, thus, even might improve patient management. The clear preference of the referring surgeons for structured reports on X-ray exams of patients with atraumatic shoulder pain further supports the growing evidence that SR can improve current reporting practices.