Based on his experience working as chief science officer at the Office of the National Coordinator, Douglas Fridsma, M.D., Ph.D., says there are three things that he believes could make the biggest difference in health IT going forward: development of more consistent common data elements; better ways for patients to get a full extract of their medical record data; and interoperability testing requirements that include both sending and receiving data.
Speaking at the Johns Hopkins University’s Division of Health Science Informatics in Baltimore on Oct. 2, Fridsma, who is now president and CEO of the American Medical Informatics Association, started out by describing several lessons learned at ONC during the meaningful use roll-out:
• Framing matters. Fridsma spoke about the challenges involved in planning for an ultra-large-scale system. You can’t take a great health IT system at one institution such as Johns Hopkins, and supersize it. “These things don't scale,” he said. The fundamental thing to remember, he said, is that it is not about architecture; it is about city planning. “It is not about creating a gigantic building; it is about the building codes and zoning that make cities thrive and grow.” An ultra-large-scale system is naturally decentralized, so the ONC’s role is orchestration rather than command and control. Progress will have to be iterative and incremental.
• Interoperability is the road, not the destination. The ONC health IT interoperability roadmap does not contain many specifics. Fridsma said that is because interoperability is a road toward functionality rather than a destination. Starting with the IEEE definition of interoperability as the ability to exchange information and use it, he said interoperability is not separate from the thing you want to do. “When people complain about systems not being interoperable, I ask them what they are trying to do” he said. Interoperability is about making incremental progress toward better functionality.
• Frame solutions in terms of what matters to people. Health IT projects should focus on some aspect of the “triple aim” of improving the patient experience of care; improving the health of populations; and reducing the per capita cost of health care, Fridsma said. “If they don’t focus on those, people won’t care about your solutions. If efforts are about something else, and don’t address those, it is hard to advocate for those.”
• Four scales of engagement. To make the larger goal of a learning health system work, it has to be meaningful at many different levels of the healthcare system. “The learning health system is about patients, practices, populations and the public,” he said. We have to think about a learning loop and aggregate data, he said, to see if things can be generalized and learn from it and flow it back.
Moving on to what could make the biggest difference going forward, Fridsma said one problem has been that the EHR certification program doesn’t really test interoperability. “That is kind of a dirty little secret. It tests conformance on send, but it doesn’t test the receive end of things.” The way it works in a standards development organization, he explained, is that you get a series of interested parties, companies A, B and C. Company A says we do all our exchange using this kind of format. Company B says we do it all using another format. Company C says we use a slightly different variant of that. They say, OK, let’s make the standard A or B or C. They all vote yes. Then the government says we are going to test to see if you conform to the standard. And A keeps sending A, and B keeps sending B. And C keeps sending C, Fridsma said. And that is when you hear the reports that say this system can’t receive the standard. They are both certified products, but they only receive A, B, or C. “What ends up happening is if you don’t hold folks accountable for both ends of the transaction, you get conformance and incredible variability in your standards, and optionality all over the place,” he said. Whereas if you required people to test both ends of the equation, companies would have an economic choice of whether it is worth it to have two systems that can receive A and B, or should we just all agree that we are not going to be able to afford to support that variability. “So one of the things we have to do is fix the incentives in the standards development world so we test both ends of the equation, because that is how we are going to get to interoperability.”
Fridsma said a common core of data elements also would help. “We have 400 data elements that are part of the meaningful use data set,” he said. When he worked for ONC, he met with thoracic surgeons, radiologists, and other specialists and each one said they had 30 of the most important data elements, or cardiologists would say it you collect these elements around blood pressure and hypertension, we can make a big difference in heart disease. “You see what happens: you end up with thousands of data elements required to support all of these things,” he said.