Radiologist decision support may cut unnecessary studies

So far, clinical decision support has been promoted as a tool for referring physicians, but what if it's used by radiologists? A decision-support tool that's readily available for radiologists when reading medical images can help them order more-appropriate follow-up exams, according to researchers from NYU Langone Medical Center.

After adopting the tool -- and integrating it into the institution's dictation software -- the research team found that radiologists' adherence to clinical guidelines improved from 50% to 80%, a statistically significant difference.

"Decision support available to the radiologist at the point of care and integrated directly into the dictation software may encourage use and help improve adherence to follow-up recommendations," said Dr. Ankur Doshi. He presented the group's experience during a scientific session at the recent Society for Imaging Informatics in Medicine (SIIM) meeting in National Harbor, MD.


It's been estimated that as many as half of all advanced medical imaging exams may be unnecessary, and that 31% of all abdominal CT exams contain recommendations for additional imaging.

Incidental ovarian cysts are frequently detected on imaging, the overwhelming majority of which are benign. Managing these patients too aggressively and ordering too many follow-up tests can generate anxiety and costs that could be avoided, Doshi said.

Clinical guidelines -- such as the Society of Radiologists in Ultrasound (SRU) 2010 guidance for ovarian cysts on ultrasound and the 2013 American College of Radiology (ACR) white paper for cysts found on CT or MRI -- can help reduce these unnecessary examinations, he said. But despite published guidelines, there is still significant variability in follow-up recommendations made by radiologists.

Radiologists might stray from clinical guidelines due to variations in local practice patterns, their personal experience, having to go through a cumbersome process to access the guidance documents, and institutional Internet restrictions that prevent radiologists from accessing Web-based guidelines, Doshi said.

As a result, the researchers sought to develop collaborative recommendations for managing incidental, asymptomatic ovarian cysts detected on ultrasound, CT, and MRI. A meeting between four gynecologic oncologists and two radiologists produced the recommendations, which were based on collective experience and local practice patterns, as well as the 2010 SRU guidelines and the 2013 ACR white paper on incidental findings.

These recommendations for managing asymptomatic ovarian and adnexal cysts were then converted into Web-based flow charts using Adobe Illustrator and Dreamweaver software. Next, the researchers used their dictation software's administrator control panel to launch the decision-support tool during image interpretation.

To assess the use of the software, the group included tags in the link to collect statistics; the data were then sent to an internally developed radiology informatics server and logged into an Oracle database, Doshi said.

Radiologists get the message

To determine the effect of the radiology decision-support tool, the researchers conducted a retrospective review of ultrasound exams performed between January 2012 and November 2014, a period that spanned 26 months before the launch of the tool and nine months after it was launched.

They performed a computerized search of the radiology report impression for mentions of ovarian and adnexal cysts and calculated the degree to which radiologists adhered to guidelines for follow-up imaging, before and after the tool was implemented. They also reviewed the electronic medical records to determine what follow-up was chosen by the clinician.

The radiologist decision-support tool was accessed 406 times between April 1 and December 31, 2014, for an average of 2.1 clicks per day. There were 33 unique users, including abdominal, nuclear medicine, and musculoskeletal specialists.

The results indicate that radiologists were more likely to adhere to clinical guidelines after adoption of the tool, compared with before the tool was available. Also, the number of cases classified as "overmanaged" decreased.

Effect of radiologist decision support on guideline adherence
Status SRU recommendations (before adoption of decision-support tool) Institutional recommendations (after adoption of decision-support tool) p-value
Adherent 98/197 (50%) 111/139 (80%) < 0.0001
Undermanaged 32/197 (16%) 14/139 (10%) 0.155
Overmanaged 67/197 (34%) 14/139 (10%) < 0.0001

While the likelihood that clinicians would choose the follow-up exam recommended by radiologists did improve from 36 (46%) of 73 cases before adoption of the tool to 27 (57%) of 47 afterward, this change did not achieve statistical significance (p = 0.5034).

The researchers acknowledged several limitations of their study, including the possibility that other dictation clients available in the market may not offer such customization. In addition, the postintervention data analysis was performed in the immediate nine months following the launch of the tool.

"It's possible that usage may decrease over time," he said.

In addition, the radiology reports were retrospectively searched, and it's possible that some radiologists may not have mentioned benign small cysts or follicles in their reports, Doshi said.

Page 1 of 603
Next Page