Expert in the Field: Dr. Cree Gaskin, UVA Health System

Implementation of ACR Select® led to significant improvement in the appropriateness scores of ordered imaging tests in a recent JACR published study.

 

Dr. Cree Gaskin

Professor of Radiology, Associate Chief Medical Information Officer, Vice-Chair of Informatics for the Department of Radiology and Medical Imaging, Division Director of Musculoskeletal Imaging and Intervention, UVA Health System

In addition to his executive positions at UVA Health System, Dr. Gaskin co-authored the recent JACR published study, Impact of a Commercially Available Clinical Decision Support Program on Provider Ordering Habits, revealing implementation of a commercially available Clinical Decision Support (CDS) tool integrated into the EHR provided a significant improvement in imaging study appropriateness scores.

In the JACR study, ACR Select® was integrated into UVA’s EHR, without displaying appropriateness scores for 6 months. Then, appropriateness feedback was “turned on” at order entry for adult patients in the emergency and inpatient settings for 24 months. The appropriateness scores of imaging tests before and after displaying feedback at order entry were compared and evaluated by modality and attending versus trainee status. After implementation of a commercially available Clinical Decision Support (CDS) tool integrated into the EHR, there was a significant improvement in imaging study appropriateness scores, more pronounced in studies ordered by trainees.

Electronic order entry became standard, there were existing appropriateness criteria, and providers sometimes ordered the wrong imaging tests. Thus, even before the Protecting Access to Medicare Act (PAMA) of 2014 mandated decision support, it made sense that this should all be put together to ensure patients get the best care.

 

Tell us more about your role at UVA.

In the current context, I lead UVA’s Clinical Decision Support program for the ordering of imaging tests. I work with technical, clinical, and administrative staff to optimize our implementation, hopefully for the benefit of our patients and referring providers.

 

What motivated UVA’s initial adoption of ACR Select?

The radiology department at Massachusetts General Hospital published their success with home-grown decision support integrated with imaging order entry. We were impressed by this idea and in 2010 we thought it was a logical part of the future for everyone. Electronic order entry became standard, there were existing appropriateness criteria, and providers sometimes ordered the wrong imaging tests. Thus, even before the Protecting Access to Medicare Act (PAMA) of 2014 mandated decision support, it made sense that this should all be put together to ensure patients get the best care.

We waited for a product that we felt could be successful at our institution, and we decided to move forward with ACR Select. It was important to us that the clinical content had long been curated by the American College of Radiology in the form of its Appropriateness Criteria. It was also important to us to have the right technical knowledge regarding how to integrate with our electronic health record Epic. After all, the program could only be successful if the ordering experience was acceptable to ordering providers.

 

What was the impetus for the study, and who was involved?

We were an early adopter of commercially available decision support. While we believed in the concept, we were also aware that an early generation effort could fail. We knew that our providers had largely accepted the user experience, but we wanted to know if positive impacts had actually occurred. Specifically, we sought to determine if our efforts had led to improvements in the appropriateness of imaging orders. Our study team involved radiologists, a database analyst and a statistician.

 

Please summarize key learnings from the study.

Our implementation of a commercially available decision support program was associated with a significant improvement in the appropriateness scores of ordered imaging tests. This was true for MRI, CT, and ultrasound. Among provider groups, trainees exhibited a greater response, but faculty also showed improvements.

We did not study outpatient orders, because our early implementation involved only the inpatient and emergency department settings. I’d also like to emphasize that our implementation did not involve any incentive or penalty to comply for ordering providers.

 

 

Figure. The percentage of low utility studies ordered by trainees decreased from 10.8%  to 4.8%  and the percentage of indicated studies rose from 65.6%  to 83.7%. 

UVA has recently expanded its installation to CareSelect™ Lab, how will these findings inform your future decision support strategy?

Clinical Decision Support has potential to inform, and positively impact, many types of orders. It is our hope that thoughtful implementation can support our providers by informing their care decisions in a helpful manner. If we continue to see benefits, we will continue to expand decision support.