AMDIS: Reconsidering the uses of EHRs

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - question mark

BOSTON—Can we learn from all of the patients all the time instead of some of the patients all of the time or just a couple of patients not that often? That’s a question posed by Kenneth D. Mandl, MD, MPH, Harvard Medical School professor and director of the Intelligent Health Laboratory at Boston Children’s Hospital, at the AMDIS Fall Symposium.

“We don’t actually look at what we’ve done and the outcomes in any way, shape or form,” he said. Closer scrutiny presents opportunities to deliver care more efficiently.

Healthcare runs too many processes through EHRs which, “on the whole, are older technology developed in the 1980s and 1990s,” Mandl said. In the consumer world, we are willing to use multiple tools to get the job done, such as Google for searches, iTunes for music and Word for word processing. None of these companies has managed to produce a platform that can be used for all of our needs so why would it be any different for healthcare? He cited a large healthcare delivery system that uses its EHR for communication between clinicians. “Is that really the best technology for that use?”

Mandl turned these thoughts into a project called SMART (Substitutable Medical Applications, Reusable Technologies), a platform architecture to support a flexible health IT environment and promote innovation. The project was designed to study whether EMRs can behave like iPhones or Androids in that innovators can readily create and widely distribute apps across thousands of installs, he explained. “Apps compete with each other on the things that actually matter which is probably going to relate to usability, value and cost-effectiveness—things that are difficult in a noncompetitive market.”

Tonya Hongsermeier, MD, MBA, CMIO of Lahey Health based in Burlington, Mass., discussed the challenges of knowledge management using cloud computing. The process involves sending some kind of data packet through a firewall to a cloud-based decision support service provider, she explained. That provider might need to take that data and parse it and put it into some kind of semantic environment to classify and further reinterpret. They apply their logic interpreting that data and send back combinations of assessments or recommendations or other enablers that then get reinserted back into the workflow.

When externalizing the curation of decision support logic and content, the vendor is responsible for making it possible for you as a customer to send data to the service and send back information. The vendor also is responsible for making sure you insert it into the appropriate workflow context. People sending the data are accountable for informing the service supplier about any changes because the supplier might need to update its logic.

Hongsermeier discussed the Clinical Decision Support Consortium which was established in 2008. The large team sought to study the knowledge management lifecycle and create proposed approaches to specify knowledge to make it a shareable aspect with a repository structure that allows for collaboration across all participants.

The group’s specification model involved multiple layers. It was published and deployed among several partners who took active rules and externalized them to the service and turned them off in the native environment. After five years, they had multiple collaborators. Guidelines for conditions such as diabetes, coronary artery disease and hypertension were refined through multiple levels and published to the CDSC knowledge portal. “There was a lot of work we had to do to achieve semantic normalization and we worked through datasharing agreements to make it possible.”

Because there are multiple possibilities for user error, the group developed legal agreements to address liability points of failure, she said. Those agreements “basically asserted that the provider was responsible for determining that the guidance is appropriate because the provider always had the richest understanding of the patient context.” Guidance is continuously refined to meet the latest recommendations.

“It would be great if EHR vendors were invited to make it easier to import knowledge so I don’t have to do all this curation,” Hongsermeier said. “Many vendors are creating a model whereby customers are saddled with a lot of curation that could be commoditized.”

There are numerous opportunities and challenges to improving knowledge management, she said. “EHR vendors can’t expect their customers to curate all this knowledge inside their EHR walls. Some are more resistant than others and there are concerns about reliability.” She said providers need to help “vendors receive standards and get settled with those standards so they can go about the work of identifying the right architecture that makes the performance tractable. Few EHR clinical decision support systems can actually execute the kind of inferencing required for personalized medicine. We need to advocate for EHR vendors to move beyond ‘walled gardens of simple CDS.’”