Clinical Efficiency in Radiology

Inefficiencies in the radiologist’s clinical workflow can collect and sprout so it’s important to identify gaps, Stephen Willis, chief technology officer of Canopy Partners, an IT and business solution vendor, said at the 2016 Radiology Summit from the RBMA.

The first place Willis looks in identifying inefficiencies is the order. He wants to make sure it has good data in it, including the patient’s name and kind of exam. The radiologist needs more, though.

“One of the places we first identified inefficiencies, which sort of permeate throughout the system, is when we have an order for pain or something along those lines,” Willis said. “Technically, I’m sure the patient is in pain but unfortunately, the radiologist can’t get a very meaningful result based on that, and additionally, we can rarely get paid based on that.”

Willis uses clinical decision support on the front end, which sometimes requires a manual process.

“The radiologist, as an example, catches an order that says ‘pain’ or ‘MVA’, that’s a sign to the radiologist that they won’t get paid for that test and they should push back,” he said. “Beyond not getting paid, which is obviously a motivational factor, radiologists can very rarely give a meaningful result if they don’t truly know patient history.”

Willis urges radiologists he works with to push back and ask the technologist or ordering personnel to get more information from the patient. It might not sound more efficient, he said, but at the end of the day it results in better patient care.

He also noted that technology can be especially useful in giving the radiologist more data. A sidecar, which Willis refers to as a data mining system that works with different EMRs or data application platforms and then presents that information to the radiologist in real time as the test comes up, is especially helpful.

“You can customize it to show the data the radiologists care about and we are seeing radiologists demand this type of data more and more,” he said. It keeps the radiologist from having to call an ordering physician to get more information or log into a separate EMR system and dig around for information.

Another area ready for improvement is the worklist, Willis said.

“Smaller groups, especially, read out of the same bucket or just a few buckets, which is certainly tenable if volumes don’t continue to increase while reimbursements goes down,” he said. “But the facts of life are that the volume is increasing while reimbursement is going down.”

Worklist/workflow software gives radiologists more meaningful data regarding what study to read next.

“For some radiologists, as much as 35% of their day is spent trying to figure out what test they should read next,” he said. “It’s not very efficient.”[[{"type":"media","view_mode":"media_crop","fid":"48106","attributes":{"alt":"Stephen WIllis","class":"media-image media-image-right","id":"media_crop_4346989346470","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"5727","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"float: right;","title":"Stephen WIllis","typeof":"foaf:Image"}}]]

Willis installs a worklist software that not only makes lists subspecialized, they also identify the best exams for specific radiologists to read. If everything is running smoothly and the patient load isn’t too high and everyone is meeting their turnaround times, such software might not be necessary. But most practices have SLAs to meet and workload isn’t always manageable.

“What we’ll do is we’ll put in SLAs around ER exams and when ER exams start to age over that 20-25 minute barrier that makes ER physicians very nervous, those tests start appearing on top of everyone’s list,” Willis said.

In some practices, Willis has done away with the worklist altogether and the software automatically feeds the radiologist his or her next exam.

“There was a little pushback [to removing worklists] in the beginning,” he said. “But we got a lot of resounding efficiencies out of that.”

Often, radiologists don’t know when looking at a worklist whether a patient is inpatient or outpatient, and sometimes that chest X-ray is the last test needed to discharge an inpatient, while the outpatients have been gone from the office for hours.

“There is just not a lot of value in that,” he said. “Worklist software that is intelligent and understands SLAs and priorities is very meaningful.”

The VR platform, if configured properly, can also save time. The VR should populate standard data: reason for study, proper exam name, technique if applicable, and any prepopulated relevant exams should already be listed, Willis said. Normal findings for all of the different pieces of the exam should also be included so the radiologist can go through and edit what needs to be edited and say what needs to be said, rather than dictate minutes’ worth of normals for an exam.

Natural language processing has matured VR with the ability to identify errors in real time.

“We want to be notified if we dictate a cervix for a male while we are dictating,” Willis said. “Things like that are a lot of help.”

On the bleeding edge of NLP and VR, however, is knowing who the dictating radiologist is and identifying what exams they are comfortable reading. If reading an exam they are less comfortable with, software can provide real time assistance in the form of documents that guide the radiologist on specific things to look for, while they’re dictating, Willis said.