SIIM keynote: It's the end of our world as we knew it

In his provocatively titled SIIM15 keynote, "The Next Imaging Evolution: A World Without PACS as We Know it," Donald K. Dennison, consultant, walked attendees through the market forces that are reconfiguring the world of imaging informatics.

While Dennison acknowledged that reports of the demise of PACS—like Mark Twain's—are premature, his message was that the image management strategy that will take providers into the future is vastly different than the ones in place in most institutions today.

Three market forces are driving the change, beginning with medicine's changing business model and the shift from transaction- to results-based healthcare. As risk shifts from payor to provider, "this is going to be very interesting in how it affects radiology,” he said.

Widespread EMR adoption is the second driver, resulting in fewer interfaces and more consistency. "That matters to us in radiology because instead of getting information from multiple sources we are getting one big aggregated feed," he said.

Consolidation is the third big trend that is literally reshaping healthcare, as large healthcare providers buy or affiliate with smaller hospitals and care facilities. "The general tactic is to have one EMR within the entire enterprise, and the same consolidation is happening with imaging," Dennison said.

Just as vendors folded RIS into PACS, a similar system consolidation is happening with the hospital information system—or EMR—and the RIS.   "What you see today is the RIS being decommissioned in lieu of a module operating within the EMR," Dennison said.

Among the value propositions of system consolidation now under consideration by CIOs is the consolidation of all PACS archives across a health system into a vendor neutral archive (VNA), but Dennison suggests that the greater opportunity in a value-based world may lie in having one system to manage the DICOM and non-DICOM images being generated by all departments.

The era of enterprise imaging

Dennison created a schematic to provide a complete picture of the EMR and the three levels of activity associated with the data acquired in an episode of care: acquire, manage and view.

The EMR was divided into three content categories: the structured patient information and clinical data in the EMR, the scanned documents in the enterprise content management system and the DICOM images generated by radiology and other specialties in the vendor neutral archive (VNA).

"The reason I show the VNA together with the EMR is because it has more to do with the enterprise than the department,'" Dennison said. 

At the viewing level, users can access these three categories of data via the EMR using a document viewer for scanned documents and an enterprise viewer for DICOM objects and non-DICOM images.

"Can enterprise viewers provide full diagnostic reading, in other words, do I have to have two separate systems?" Dennison asked. "It's cheaper and easier to have one system. Most users do not need the advanced tools or complexity of the PACS and they benefit from viewing patients images with the data in EMR."

He acknowledges that there are competing design needs from different categories  of users  that must be addressed, but suggests that these tools, eventually, must migrate to the EMR.

What is missing from the current EMR are the non-DICOM images generated on smart phones and non-DICOM modalities nearly everywhere in the enterprise. Dennison sees this as an emerging area of opportunity for radiology to leverage its imaging informatics expertise for the benefit of the enterprise.

New Problems, Challenges

System consolidation, both of provider healthcare systems and of imaging informatics systems, is resulting in a number of new problems for the people who are managing imaging, including the integration of patient identity domains across facilities, the integration and deployment of IT staff and the absorption and migration of data.

  • Dennison reported that SIIM has three teams working on the issues related to enterprise imaging:
  • definition of enterprise imaging, including the best techniques and methods of patient identification,
  • image acquisition, workflow and acquisition-related data governance, and
  • image sharing.

One tip he offered is to take off the departmental blinders and reach out to the other teams involved in managing the data in the EMR.

"Right now, the scanned acquisition and management of documents is well under control at the enterprise level, so if you show up and say we have a solution for managing this, they are going to say we already have that plan," Dennison said. "If you are not meeting regularly with scanned document people and medical record people, you are not doing your job."

Data governance is another issue of grave importance (one that received a good deal of attention at this year's SIIM meeting) as institutions embark on the challenge of completing the image piece of the EMR by incorporating nonDICOM objects into the VNA. "You need governance on where you are going to put it, how it will be identified and how it will be presented in context, " Dennison said. "You have to go the last mile, you can't have seven different solutions for that."

Another tremendous opportunity for radiologists is making that aggregated feed from the EMR available at the point of interpretation. Dennison cited work done by Cree Gaskin, MD, at the University of Virginia, that explores how these images, as well as data elements generated by other specialties such as pathology, are influencing diagnosis.

"In Europe we are seeing radiologists training to become pathologists," he said.

The EMR and the Shared Enterprise

Dennison described another series of steps that are leading radiology further into what he called the shared enterprise. For instance, clinical decision support (CDS) is mandated and likely to most often be embedded in the EMR.

To create a shared enterprise approach to imaging, first standardize results collection and presentation. Next, standardize orders and procedures. Third, standardize acquisition and reading workflows, with some local configuration required.

Once that work has been done, radiologists will be challenged to create a multi-facility workflow, perhaps sharing a worklist across facilities to function as one large team. "Load balancing will be critical if we a going to achieve quality,” i.e. the right study read by right subspecialty, he said.

The next step is the aforementioned task of dealing with patient IDs of affiliated facilities.

One system that will likely remain a departmental system is reporting, Dennison said. The report, however, must change by moving toward tighter integration with the EMR, populated with multimedia and other enhancements, with discrete findings and traceable recommendations.

He describe a multimedia enhanced report as one that provides one view of all context and information the reader needs to understand the findings. “It should be structured in a way preferred by consumer,” he added.

Dennison referenced an ACR project that provides a unique tracking ID that can be followed up based on the following parameters: item, action, time, strength, dependencies.

"I would argue that if we do not start tracking at this level, we will be left behind," he said. "Sending out a text report and hoping people do the right thing is not going to work in a value-based system."

Ode to PACS

Dennison concluded with both an acknowledgement of the contributions of the PACS vendors and a prediction that PACS will persist in many environments. “Their death announcement may be a little premature in a lot of environments,” he said.

"I think we need to pay respect to some of the great problems the PACS vendors have solved," Dennison continued. In addition to coming up with more modular systems, they are solving many of the critical problems, such as multi-facility worklists and handling multiple patient IDs.

Nonetheless, radiology departments will need to answer to higher powers on the questions of why they need the system (and its separate maintenance agreement). In an environment of increasingly constrained resources, they also will need to be ready to provide Tier 2 support. "You are going to have to be doing a lot of that," he said.

Think hard about which parts of your PACS can be replaced by an enterprise system, Dennison suggested. Recognize the trend toward system consolidation and be ready to defend multiple systems if necessary.