Template-based structured reports can significantly shorten reporting times without compromising report quality
Our findings that SRs exhibit better completeness and result in higher satisfaction of referring physicians are consistent with and extend those from prior reports [6–13, 15, 16]. However, one common concern has been that structured reporting might be more time-consuming and complex than free-text reporting and thus impede productivity [18, 41].
Here we provide evidence that SRs may not only lead to better completeness and higher satisfaction of referring physicians but also save time, at least for highly standardized examination types such as DXA. Notably, the more experienced the subject was, the smaller the difference between SRs and FTRs became in terms of reporting time. This observation could be explained by the fact that the medical students typed their FTRs unlike the residents who used the free-speech dictation software. As typing takes more time than free-speech dictation our study might overestimate the difference between the reporting times for medical students and residents. Another explanation could be that residents were more efficient in generating FTRs as they had a more advanced level of experience.
One major advantage of structured reporting is that the user is less prone to make careless mistakes. In this study, for instance, the template not only indicated the reference ranges of the t-score in the decision tree (see Fig. 2), but also performed automatic calculations of the change of bone mineral density over time (in %).
All in all, these findings have major economic implications for the entire field of radiology. In the light of the current demographic transition, the workload in radiology is rapidly increasing and structured reporting could make a relevant contribution to improve the efficiency of radiologic workflow. This is particularly true for DXA exams as the prevalence of osteoporosis is increasing in many countries [42–45].
Structured reporting as an educational tool for residents and medical students
The template used in this study reveals that structured reporting can further be a powerful educational tool. Using info boxes within the user interface of the software, relevant background information and exemplary images and reports were displayed. In general, the info boxes might be particularly beneficial to illustrate anatomical images, classifications and up-to-date guidelines. This feature is highly useful to train inexperienced students and residents in a self-guided way. This theory is supported by our finding that the medical students preferred learning DXA reporting with a template and were able to generate very good SRs quickly. However, the medical students also reported that structured reporting might lead to a superficial evaluation due to just clicking through the template, which is a common concern about structured reporting [46]. In the present study, all four readers could easily utilize the software after a short initial training which indicates that only a minimum level of adaptation is required to switch from free-text reporting to structured reporting.
Limitations
Despite the benefits of structured reporting over free-text reporting highlighted in this study, several limitations need to be acknowledged.
First, due to the retrospective nature of our study, our subjects created the reports in a study setting and not in actual clinical practice. Thus, findings will need to be validated using the template during routine clinical reporting.
Second, in clinical practice there is a broad spectrum of reports (FTR as well as SR) being created. For example, a survey of 265 radiologists in the United States found that only 51% used structured reporting for at least half of their reports [47]. Another survey in Italy found that 56% of radiologists never used structured reporting [17]. When it comes to bone density measurements, many centers are still creating FTRs while others have adopted templates like the ISCD’s which is essentially a form containing headings and sentences with blanks for the individual BMD, Z and T-Scores, among others [48]. Furthermore, fully structured online templates with more flexibility, like the ones used in this study have been developed. These templates generate sentences automatically in standardized language depending on user entries. Other centers are even attempting to generate their DXA reports fully automatically, although they currently still require revision by radiologists [49]. Given these different approaches to structured reporting of bone density measurements with varying degrees of automation, it might also be beneficial to further evaluate and compare these different types of structured reporting.
Third, the number of subjects who created reports and their experience were limited. A prospective study including a more varied sample of reporting individuals is likely to provide further important insights into the potential of a widespread use of this template. One interesting hypothesis that could be tested is whether the time difference between FTRs and SRs decreases further with increasing experience, although our study indicates that the time required for SRs varies less with experience compared to FTRs.
Additionally, one may argue that the evaluation of report quality is rather subjective and is largely influenced by the evaluating clinician, potentially limiting the generalizability of our findings. Evaluations of the report quality by a larger, more diverse group of clinicians may be beneficial. However, due to the highly standardized nature of DXA exams, we believe that the quality ratings are likely to be consistent even among many referring physicians.
Finally, the observations of the present study cannot be generalized to other radiology examinations and their reports. Reporting times might be less likely to be improved by structured reporting in less standardized, highly variable exams, since a much more complex template structure would be required. But at the same time, reports of highly complex exams might especially benefit from the guidance of a template, since SRs were shown to exhibit higher completeness and allow better extraction of information [6–13]. Importantly, the extent to which structured reporting can improve reporting efficiency and quality strongly depends on the technical features of the utilized software. For instance, an automatic insertion of technical details into the radiology report was found to significantly improve report accuracy [50]. In a similar manner, features such as an automatic insertion of references to previous reports or an automatic identification of certain types of information could create added value, even for less standardized exam.
Further evaluation of different types of structured reporting templates in prospective (multicenter) studies with readers at various levels of experience and a larger number of evaluating physicians is likely to provide a broader impression.