Michael Trambert, MD, had long suspected the workflow reporting automation made possible by the tightly integrated PACS/RIS/VR solution used by the Santa Barbara Radiology Medical Group (SBRMC) was saving him about an hour each day. A two-pronged study 1 presented at the recent meeting of the RSNA confirmed that the conjectured efficiencies were even greater than he thought—adding up to 75 minutes per day.
For the most part, Trambert attributes those reporting efficiencies to the ability to pre-populate templated reports with a lengthy list of exam and patient data that he, his partners and radiology residents would otherwise be dictating into the report. This functionality also prevents the inevitable errors that occur in the manual transfer of data from one location to another.
Trambert and his partners in the 13-radiologist SBRMG cover both the 483-bed Cottage Hospital and the Sansum Clinic, a 150-physician outpatient practice, in Santa Barbara. Both enterprises are on DR Systems Unity enterprise imaging platform along with the city’s other independent radiology practice, interconnecting all three enterprises and maximizing the potential for efficiency within the care community.
As the practice’s chief technology officer, Trambert has enjoyed the reporting benefits of Unity since a transition eight years ago to 100% voice recognition. His 12 partners and the 15 radiology residents have made the same transition in the past four years.
“You benefit from the fact that the system knows everything about the case and the exam, and everything that you’ve done to the case, the exam and that patient within the DR Unity PACS platform,” Trambert explains. As he opens any comparison exam when reading a current exam, all desired comparison exam information, including modality, body part, contrast material usage, imaging site, date of exam, is automatically logged into the current report, whether he looks at one comparison or—as could be the case with an oncological exam—six or more prior exams.
“That’s just one example of how this automation saves a huge amount of time each day,” Trambert says. “As you click around comparison exams, the PACS automatically injects that data into the report.”
A study is conceived
In early 2014, several radiology residents at Cottage Hospital sought a project that might earn them a spot on the program at the annual RSNA meeting. Trambert suggested that they attempt to quantify the workflow benefits of the tightly integrated radiology reporting solution.
A two-part study was conceived, consisting of a survey followed by a timing experiment. Thirteen staff radiologists and nine radiology residents at Cottage Hospital completed an anonymous 12-question five-point survey
For the timing segment, eight exams were selected: two MRs, two CTs, two ultrasound and two conventional radiographs. The investigators tried to not bias the case mix towards overly complex exams and limited the comparison exams to one or two priors per study.
Nine readers participated in the timing experiment. The automation functionality normally enjoyed by users of the DR Systems Unity solution was disabled.
Notes that listed the specific information that had to be speech-recognized were loaded into the digital requisition for each exam: history, comparison exam information, dose (if CT), type and amount of contrast material used and copy-to-physician information. Readers were asked to audit and correct any SR errors.
Maximilian Cho, MD, lead resident, sat beside the readers and timed each subject. As the subject went through the cases, Cho counted how many errors occurred and then how many errors were corrected as the radiologist reviewed the report prior to moving on to the next exam.
Assessing speed, accuracy
The survey segment of the study confirmed that respondents shared the belief that auto-population of report information improved their speed and accuracy: 95% believed that the functionality saved them time; 91% responded that report automation improved accuracy of dictations; and 82% thought report automation decreased their fatigue.
Cho and his stopwatch quantified the time saved, which exceeded Trambert’s expectation. The average time to dictate the information that would have otherwise automatically appeared in the pre-populated fields was 51 seconds per study. When 51 seconds per study was extrapolated across 100 studies, the savings added up to an average daily savings of 75 minutes.
Average error rate per