Among the 65% of clinicians who ordered imaging exams during the Medicare Imaging Demonstration (MID) and gave no specific reason for placing the order, how many could have used clinical decision support (CDS) but chose not to? Was their rationale for bypassing the system reasonable? And how many were non-doctor proxies ordering on behalf of physicians?
Safwan Halabi, MD, director of imaging informatics and a senior staff radiologist at six-hospital Henry Ford Health System in Michigan, one of five MID conveners, briefly addressed those issues with RadiologyBusiness.com.
RadiologyBusiness.com: Some observers were taken aback to see that 65% of orders in the demonstration could not be evaluated for appropriateness—these were placed in the “not covered by guidelines” bucket. Did you notice any patterns driving orders into that category?
Halabi: There’s a huge gap in a lot of follow-up studies for, say, cancer patients. A lot of these patients already had diagnoses of certain elements; they weren’t coming in with a clinical scenario such as, say, shortness of breath or abdominal pain. These were people who have had a known metastatic disease or things like ringing in the ears. So it was really those subspecialty-type things in oncology and in other specialties—those were the things that we got complaints about from our physicians, saying “I can’t find the disease or scenario that I’m looking for.”
RadiologyBusiness.com: Given that a relatively high percentage of orders could not be evaluated, was the demonstration project’s credibility somewhat compromised? Or will it still carry weight going forward?
Halabi: I think it will [carry weight], because what the sampling exposed was not only the holes in guidance but also the holes in implementation. That will help in determining what education and what change-management process needs to occur for this to be successful. A lot of the conveners were not prepared to implement this on a wide scale. It was delayed from the initial start date, but I think that the biggest lesson learned is what you need to have in place to really make this successful education-wise, training-wise and also technically. The user interface has to be something that fits within the workflow but doesn’t exacerbate the inefficiencies that are already there.
RadiologyBusiness.com: In an earlier discussion, you talked about problems with non-physician clinicians doing a lot of the ordering. Did the demonstration shine a light on the pros and cons of that common practice?
Halabi: We knew a lot about proxy ordering from the start. And we had a lot of trouble deciding whether or not to give ordering providers that “out” from selecting guidelines. But the thing is, the guidelines and knowledge base were anemic. There were many holes there. It was almost impossible to fill in those gaps before implementation. So I think of this project as sort of the 1.0 version of decision support. It was very bare-bones.
One of the big issues is that even physicians, and even radiologists, don’t know all the guidelines or are not well versed in them. And I can’t imagine having a nurse practitioner or a physician assistant understanding not only what guidelines exist but also what to search for. That’s asking a lot, especially if you did not have an intense education in this. We did train people in how to use the system, but we didn’t go to the step of saying, “this is how you should order” or “this is the methodology” or “this is the way that you select guidance and search for things.” And the other aspect of that is that we didn’t have a lot of synonyms to the nomenclature that people would use on a day-to-day basis. If you look in the Yellow Pages, you might look under rubbish removers or garbage collectors to find the same service. We didn’t have that.
RadiologyBusiness.com: Will observations you made about decision support during the demonstration change anything you do at Henry Ford?
Halabi: Yes, absolutely. We are currently in that process. At the end of this demonstration, we actually went to a different electronic health record. We went from a homegrown system to the Epic system. And when we went to the Epic system, we decided not to continue decision support at that time, just because we felt it was not ready for prime time, full integration with our CPOE system. Epic has really shaped up to integrate fully with decision support so that physicians can change their order on the fly and get the guidance that they