WASHINGTON, D.C.—The radiology profession can better serve the communication needs of referring physicians and patients by embracing structured reporting and implementing search tools, according to the June 3 session, "Orders, Results, and Beyond: Communication in Radiology," at the annual meeting of the Society for Imaging Informatics in Medicine (SIIM).
“Communication is not a one-way street,” Charles E. Kahn, Jr., MD, chief of radiology informatics at Medical College of Wisconsin in Milwaukee, reminded the audience. Overlooking that point can lead to problems, continued Kahn.
The radiology report, he offered, needs to provide information in context, with meaning and that can be used. Structured reporting provides a way to improve the process.
The nature of the radiology report, Kahn said, is evolving from a paper-based format to one that can incorporate multiple inputs, including images and multi-media data and knowledge resources such as decision support and reference materials. Data originating from the scanner, such as CT dose data, may also be included.
With complexity of reporting on the rise, Kahn noted that multiple studies have outlined the efficiency of structured reporting as well as clinicians’ preference for the model. Indeed, pathology and cardiology have embraced structured reporting.
Radiology is on board as well. For example, the RSNA Reporting Initiative recently formed 14 template development committees to create an online library of best practices radiology templates. “It’s a good starting point for practices that want to adopt structured reporting,” Kahn said. Although structured reporting is by its nature highly organized, there is a degree of flexibility as templates can be text based, speech macros or XML-based.
Although physicians typically consider the patient care objectives of reporting, Kahn explained that radiology reports can be used for other purposes such as practice administration, data mining and continuous quality improvement. “The power of structured reporting will come from our ability to mine reports.”
Data mining: Improved and improving
Data mining has come a long way from its lowly origins. Initial efforts at mining took weeks of work and required a database analyst. First-generation online tools required an unmanageable 120 seconds to return results. Current models have slashed that time to the millisecond range.
“There are many ways to create search engines: Microsoft Access, relational databases, server-side scripting language and Google desktop,” said Woojin Kim, MD, chief of radiography at University of Pennsylvania School of Medicine (Philadelphia).
Kim outlined PRESTO (Pathology Radiology Enterprise Search Tool), an internally developed, web-based search tool that allows cross-specialty search and supports clinical needs, operations, education and research.
“Radiologists are surprised at what you can do with a data mining tool,” stated Kim, who listed multiple options, including looking for the following data: percentage of positivity, recommendations for further studies and laterality errors. Other current uses include coding/billing checks and automated pathology/radiology follow-up.
Such applications can reduce errors, boost quality and patient safety and better the bottom line. However, text mining and search engines will continue to improve, predicted Kim, and offer increased simplicity such as timeline presentation of imaging findings in sequence. Indeed, next-generation tools may embed intuitiveness and provide the user with connections among data that weren’t requested and enable new directions for clinical research and departmental improvement.