Despite Deadline Reprieve, Imaging Informatics Teams Feel Sense of Urgency About Clinical Decision Support

Jan. 31, 2018
Although implementation of a law to require clinical decision support use involving imaging orders was delayed, imaging informatics executives still feel pressure to accelerate efforts to deploy a CDS system.
Although implementation of a law to require clinical decision support (CDS) use involving diagnostic imaging orders was pushed back from 2018 to 2020, imaging informatics executives still feel pressure to accelerate efforts to deploy a CDS system and integrate it with their EHR.  
In 2014 Congress passed a law that will require referring providers to consult appropriate use criteria (AUC) prior to ordering advanced diagnostic imaging services for Medicare patients. If the ordering provider does not consult AUC rules via clinical decision support before the order is placed, CMS will not pay the for the imaging. 
Much to the relief of imaging informatics executives, the effective date for the Protecting Access to Medicare Act (PAMA) has been pushed back from January 2018. 
In a recent Society for Imaging Informatics Management webinar, Kevin McEnery, M.D., a professor and director of innovation imaging informatics at University of Texas MD Anderson Cancer Center, explained that CMS would use the next year or so to upgrade its claims systems. Then 2020 will be an education and testing period, and the decision support penalties will kick in in 2021.
“For the people implementing systems, this timeline provides you the framework you need to align your project plans,” McEnery said. “You need to have CDS up and running by 2020 in order to take full advantage of the education and testing period. I would consider that an optimization period.” 
“The deadlines have changed, but the requirements have not,” McEnery said.  The extra time offers informatics leaders an opportunity to convince skeptical clinicians that this process can be beneficial to them, he added. “With providers, it is the extra clicks and alert fatigue that caused the primary pushback to the initial implementations of clinical decision support,” he said. Another issue is that clinicians using CDS can provide histories that are accurate for CDS but do not truly communicate to the radiologist the actual reason for the examination and the entirety of the patient’s clinical presentation.
As an aside, he mentioned that for those institutions that have implemented CDS or are close, it is worth 20 out of 40 points in the process improvement, category for MACRA, “so for those early adopters there is a benefit to continue to adopt.”
Speaking about early implementations, which he referred to as Version 1.0, McEnery said they involved putting a system in as an appendage to the EHR, and then having that system operate alongside the EHR but not integrated into it. “That is one of the reasons for some of the pushback from physicians. What you are seeing now is the idea that not only do you need to implement a CDS platform but you also need to look at ways to optimize your EHR to leverage that platform. By optimizing it, you could potentially have better acceptance by providers.”
With a January 2018 deadline, you would really have to put a Version 1.0 product in and your scope would have to be limited to achieve your goals, McEnery continued. “With the new deadline, you can look at your platform and optimize your EHR to fully leverage the capability of the platform for clinicians and radiologists.”
One imaging executive who has had considerable experience with implementing CDS is Keith Hentel, M.D., chief of the Division of Emergency/Musculoskeletal Radiology and executive vice chairman in the Department of Radiology at NewYork-Presbyterian Hospital-Weill Cornell Medical Center. 
During the webinar Hentel spoke about the urgency of planning for CDS implementation and some lessons learned from Cornell. “Everyone realizes that even though the program has been delayed, they need to be planning for it,” he said. 
Hentel said Cornell started working on CDS for imaging around 2005 when radiologists became targets for the reported increased utilization of imaging and the fear that this was going to break the Medicare program. “We thought CDS was a good idea — doing the best imaging at the right time,” he said. Cornell was selected to be part of the Brigham and Women’s convener group in a Medicare demonstration project, which was and is the largest demonstration project of clinical decision support for imaging ever done.  
That research didn’t show a huge impact on utilization or appropriateness of the imaging performed, he said. “One of the major contributing factors was that our work flow was very cumbersome. There was a lot of physician dissatisfaction in our enterprise when we turned on this system. In fact, we had initially thought that what was good for our Medicare patients was good for all our patients, so we did a very broad go-live, which eventually got pared back due to clinician complaints.” 
One of the reasons the Medicare imaging demonstration project may not have been as successful as it could, Hentel added, is that when you look at the interactions that took place, only a very small percent, less than 5 percent, were actionable alerts to the clinician or had any potential to improve care. “So this ended up with a lot of alert fatigue and physicians just ignoring what was on the screen, not to mention the fact that some of the nationally available appropriate use criteria conflicted with what we consider local best practice at Cornell,” he said. “That created a problem for some of our clinicians who didn’t quite know what to do. The final problem we saw was that there were no consequences of not using the CDS.”
When Cornell sought to do CDS again, it tried a different approach and mentality. “We went from a shotgun method to a very surgically precise method,” Hentel said. “What I mean by that is we didn't’ target CPT codes anymore. We targeted specific clinical presentations and specific imaging exams that we knew were ordered incorrectly.”
They also changed from a more global view, where they used nationally available appropriate use criteria, to what they considered local best practice. They also changed from strictly physician-entered orders to both clinician and automatic checks of AUC with information pulled from the EHR to lessen the burden on order providers. 
Cornell also more narrowly defined its goals. “Now every intervention we put in place, we have a well-defined goal, Hentel said.  “That goal may be decreasing inappropriate utilization; it may be making sure the ordering provider orders the most appropriate exam. We define a goal for every intervention and that has been very helpful.”
“With PAMA, we have had to retool again. We decided to put in our application to be a ‘provider-led entity’ and define our own AUC,” he added. “We are doing that now and it involves multiple physician groups. We have very complicated governance that involve subspecialty radiologists and physicians of all kinds that go over AUC and approve every implementation of a rule we put into our program.”