Rethinking radiology workflow to improve patient care—today

The pace of innovation has quickened considerably during the past 18 months at vRad, ever since Ben Strong MD, chief medical officer, found a willing collaborator in CIO Shannon Werb.

Werb was hired at the end of 2013, and since then the two have reviewed each essential element of vRad’s radiology workflow. The endgame is optimizing performance for its more than 350 radiologists, who read 5 million patient radiology studies annually for more than 2,100 client hospitals, health systems and radiology group facilities.

With a scope such as that, the organization has an unmatched ability to impact the practice patterns of radiology and make lasting improvements to patient care. The challenge is both an inspiration and a motivation that is driving workflow innovation at vRad through the application of information technology.

Inspiration and perspiration

The vRad team looked at the entire arc of the interpretation process, including:

• order submission from the client site;
• study distribution to a vRad radiologist;
• radiologist interaction with the images;
• creation of the report; and
• communication of the findings and critical results to the referring physician who placed the original order at the client site.

“There are a number of preparation steps that happen before the radiologist’s eyes are ever on the diagnostic image,” Strong says, “but they’re very important in terms of giving us enough patient information, and getting that study to the most appropriately trained radiologist who must also read it efficiently.”

Sophisticated—and patented—computer algorithms for telemedicine workflow were designed to more accurately match examinations to radiologists of the appropriate specialty, while still managing to maintain turnaround times (TAT).

A prioritization algorithm for workload balancing runs concurrently, looking at the TAT countdown, the specialty of the examination and the age of the patient, as well as evaluating which radiologists are licensed to read in that state and credentialed in that facility. An overlay illustrates each radiologist’s current workload, and an assignment algorithm interprets the likelihood of that study being read in a timely manner if placed on a specific worklist.

“Workload balance is extremely important, and we have all kinds of algorithms that prevents studies from being assigned to long queues,” Strong adds. “The assignment algorithm is constantly redistributing studies according to volume and will preferentially fill the worklist of radiologists that have relatively fewer studies available to them.”

There when you need it

All of this happens before a radiologist ever sees a study. When they do, the optimization team made sure that all relevant patient data are presented to the radiologist in a uniform, interpretable format.

“It’s a great frustration for a radiologist to open a case and be missing prior studies, reports, important clinical information or the reason the study was ordered,” Strong says. “This is all objective information that we must aggregate and present to the radiologist, and incorporate it into the initial dictation.”

To automate this process, vRad integrated systems and programs so that the relevant patient data is pulled from all available entry fields—from HL-7 feeds to associated faxes or direct technologist entries into the system—scraped and mined for objective study information and populated in the appropriate dictation fields so that it appears when the radiologist opens the study. This rescues the radiologist from the mind-numbing repetition of objective data that is recited at the beginning of every dictation.

“It was my wish,” Strong explains, “that every radiologist open a case, assimilate the patient information and the prior study information in an interpretable interface and immediately begin dictating the findings of the report.”

This was no small feat, considering that vRad also had to tease the objective patient information and history from more than 2,100 customer sites that each operate using their own set of protocols and means of communication.

For instance, while some facilities have robust HL-7 interfaces that provide vRad with copious information electronically, other facilities do not utilize electronic health records, offering vRad zero opportunity for bidirectional electronic communication.

“The lack of digital sophistication across healthcare is just another example that it, compared to other industries, is shockingly far behind in terms of adopting technology, digital systems and digital information,” Strong says. “Many clients are still receiving our reports via fax, while others are receiving their reporting directly into their EHR. We are willing to help our clients move into this digital age, and encourage them to move for the benefit of their clinical staffs and their patients.”

Eyes on the prize

Once the right radiologist receives the study, Strong has two guiding principles he stresses when training new vRad radiologists:

1. always have your eyes on the diagnostic images, and
2. never touch the keyboard, so that your hands are always on a computer input device, and you are actively engaged in the interaction with the images you are interpreting.

Creating a report in a real-time fashion without distractions is key, and integrating voice recognition with reporting was a significant step in that direction. vRad recently took that one step further by implementing a structured reporting program that optimizes the formatting of all reports. The program mines the content of the reports in the system and, for example, if a study has been tagged as a trauma study, the program searches the report to determine if the appropriate terms are used to describe trauma injuries.

Referring physicians consistently report preferences for structured rather than free-text reports, and vRad has achieved 100% adoption by its 350-plus radiologists.

“If the program determines that positive findings are present,” Strong says, “it pops up a warning to the radiologist to remember to apply standardized injury grading for viscus lacerations and vessel injuries. Then it provides a link to the grading systems, which are kept in our radiologist portal. The automated system is helping to improve our TATs but also helps us to maintain our accuracy and quality standards to deliver the best patient care.”

Call automation

Any report that is negative for a routine study will be delivered to the site in whatever format they prefer; however, any critical finding must be communicated verbally, directly to the referring clinician. To make that happen, a vRad radiologist doesn’t even have to pick up the phone, another example of removing nonclinical tasks from radiologist workflow.

“If a radiologist must report a clinical finding, two mouse clicks automatically generate a call to the facility,” Strong explains. “The call is routed to a U.S.-based operations staff member who verifies the physician to whom the report will be given and—upon verification— transfers the call to that radiologist.”

Due to a highly automated workflow that is supported by information technology, vRad can track usage and analyze the performance of its operations center at every segment in the workflow process. Through a feedback loop, radiologists also can track a variety of metrics, such as failures of communication, comments on protocols or radiation dose levels. All information is logged and any issues immediately addressed with the corresponding facility or radiologist.

The year of 2014 will go on record as one of great innovation for vRad, one in which a close collaboration between clinical and IT leadership resulted in demonstrable improvements in quality and efficiency.

“I’m really grateful to our CIO,” Strong says. “Since Shannon’s arrival in 2013, we have been energized, and accomplished things at a pace and level of performance I never thought possible. With the new energy and activity, we’re way out ahead.”