RSNA feature: Tales from the cryptwhen structured reporting goes awry
CHICAGO—Adhering to best practices and engaging end users in a project implementation does not guarantee compliance or a successful adoption of structured reporting templates, according to the scientific poster presented Nov. 27, at the 97th Scientific Assembly and Annual Meeting of the Radiological Society of North America (RSNA).
Jackson Memorial Hospital, a 1,550-bed hospital in Miami, implemented voice recognition and structured reporting simultaneously. “We felt implementing both systems simultaneously would enable the easiest transition to structured reporting,” Gary H. Danton, MD, PhD, medical director of radiology, said in an interview. The rationale was straightforward: faculty had to learn a new system of dictating with voice recognition and could adjust to structured reporting at the same time.
Danton and colleagues used RSNA structured reporting recommendations as a framework for the project and introduced team members to RadLex as they developed templates for the reports. The entire department from junior residents to faculty and administration was involved in the development process to ensure buy-in.
The development team selected an itemized report model with report headings and descriptions in bulleted format. “A busy clinician could look at the report and immediately go to the organs of interest,” said Danton.
Although the end product was designed to be clinician-friendly, the development process proved to be arduous. “It takes a long time to build templates,” observed Danton. In fact, the hospital delayed its voice recognition implementation by one month in an attempt to complete the templates and ultimately finished normal templates before go-live.
Despite the additional time, the end product left staff dissatisfied. One of the major points of contention, said Danton, is that the attending physician must sign a report that doesn’t look exactly like the final report. That is, it includes extra options and text that doesn’t disappear until after signing.
Another complaint was the plethora of recommendations. Without dropdown menus, the report style shows all-or-nothing as far as recommendations, a look that is too cluttered for both the radiologist and clinician.
In addition, the order of various sections in the report can force physicians into an unfamiliar interpretation style. A radiologist may review a study in the order gallbladder, spleen, pancreas, liver; but if a template is ordered differently, it may slow the reader.
Finally, although earlier data had suggested that templates improved the efficiency of voice recognition, Danton found that the accuracy of voice recognition has improved to make it faster than a structured template.
“Because of these issues some radiologists have decided not to use structured reporting templates,” he said.
Despite the missteps, the department has had some small victories. The current template model does work well for complex studies that require a significant amount of information such as an abdominal ultrasound, said Danton.
Another success stemming from the project is the itemized report model. “We’ve been somewhat successful in getting more physicians to use an itemized format rather than one long paragraph. This is helpful for clinicians, but not as structured as we wanted.”
Over the next year, the radiology department will optimize its templates and redesign them to make them more user-friendly and ensure that they include information necessary for billing and coding. Danton and colleagues also are surveying clinicians to better understand their needs. He predicted that the survey will show clinicians want itemized structured reports, which, in turn, could spur radiologists to embrace the format.
Danton offered some advice for other sites considering a structured report model, which included:
The poster was titled, “Implementation of Structured Reporting at a Large Teaching Hospital: Where We Went Wrong.”
Jackson Memorial Hospital, a 1,550-bed hospital in Miami, implemented voice recognition and structured reporting simultaneously. “We felt implementing both systems simultaneously would enable the easiest transition to structured reporting,” Gary H. Danton, MD, PhD, medical director of radiology, said in an interview. The rationale was straightforward: faculty had to learn a new system of dictating with voice recognition and could adjust to structured reporting at the same time.
Danton and colleagues used RSNA structured reporting recommendations as a framework for the project and introduced team members to RadLex as they developed templates for the reports. The entire department from junior residents to faculty and administration was involved in the development process to ensure buy-in.
The development team selected an itemized report model with report headings and descriptions in bulleted format. “A busy clinician could look at the report and immediately go to the organs of interest,” said Danton.
Although the end product was designed to be clinician-friendly, the development process proved to be arduous. “It takes a long time to build templates,” observed Danton. In fact, the hospital delayed its voice recognition implementation by one month in an attempt to complete the templates and ultimately finished normal templates before go-live.
Despite the additional time, the end product left staff dissatisfied. One of the major points of contention, said Danton, is that the attending physician must sign a report that doesn’t look exactly like the final report. That is, it includes extra options and text that doesn’t disappear until after signing.
Another complaint was the plethora of recommendations. Without dropdown menus, the report style shows all-or-nothing as far as recommendations, a look that is too cluttered for both the radiologist and clinician.
In addition, the order of various sections in the report can force physicians into an unfamiliar interpretation style. A radiologist may review a study in the order gallbladder, spleen, pancreas, liver; but if a template is ordered differently, it may slow the reader.
Finally, although earlier data had suggested that templates improved the efficiency of voice recognition, Danton found that the accuracy of voice recognition has improved to make it faster than a structured template.
“Because of these issues some radiologists have decided not to use structured reporting templates,” he said.
Despite the missteps, the department has had some small victories. The current template model does work well for complex studies that require a significant amount of information such as an abdominal ultrasound, said Danton.
Another success stemming from the project is the itemized report model. “We’ve been somewhat successful in getting more physicians to use an itemized format rather than one long paragraph. This is helpful for clinicians, but not as structured as we wanted.”
Over the next year, the radiology department will optimize its templates and redesign them to make them more user-friendly and ensure that they include information necessary for billing and coding. Danton and colleagues also are surveying clinicians to better understand their needs. He predicted that the survey will show clinicians want itemized structured reports, which, in turn, could spur radiologists to embrace the format.
Danton offered some advice for other sites considering a structured report model, which included:
- Consider an implementation process rather than a quick jump to structured reporting. Start small and don’t structure all reports from day one.
- Involve end users early in the process. Include coders, who can review templates for mistakes prior to implementation.
- Understand the limitations of voice recognition and structured reporting software and how they will impact the template building and reporting processes.
The poster was titled, “Implementation of Structured Reporting at a Large Teaching Hospital: Where We Went Wrong.”