Part two of a two-part series (Access part one here):
Since almost everyone these days is giving advice to the Obama administration, I asked William Stead, co-editor of the report, Computational Technology for Effective Health Care: Immediate Steps and Strategic Directions, what advice he would give to the new president, who mentioned healthcare reform in his inaugural address and who has proposed billions of dollars of federal spending on health information technology.
One draft of a stimulus bill made public last week by the influential House Ways and Means Committee includes $2 billion for HHS Office of the National Coordinator for IT grant-making, so rather than make recommendations to President Barack Obama, Stead directed his recommendations to Congress and to Robert Kolodner, the physician who at this writing remains the holdover head of the ONC.
Stead, a physician who is associate vice chancellor for strategy/transformation and the chief information officer of Vanderbilt University Medical Center in Nashville, volunteered that he served on the congressionally mandated Commission on Systemic Interoperability of heath IT, which served up several slickly bound reports in 2005 that have been scarcely heard of since. Asked if the previous administration erred in pressing for interoperability in its healthcare IT promotional activities, Stead bluntly indicated yes, in macrocosm, but no, in microcosm.
I think the pressure to get true, systemic interoperability for a large part of the medical record is unachievable, Stead said. One of the things in that report that Id still do today, if I were Rob Kolodner, I would work with the labs, medical-device and drug vendors to tag all of their products with some sort of electronically computable information.
There is no reason that every lab instrument couldnt include with the result an industry standard tag about what that result is about, he said. Thats doable. Thats something Id legislate and I think in this environment, it could be done. Youve got to have this standard label. It would not be expensive, but it would be something we could do today.
The second thing I would do is give money to the states to work with pharmacies and pharmacy benefit managers and require them to create a transaction every time they dispensed a medicine for an individual. I think it would be easier to work at the level of the states. I would then aggregate that (data) and use it for two purposes. One would be to use the data to eliminate the replication of any number of databases set up by states for drug enforcement; the other would be to make the data available, via some secure manner, to prescribers at the time they write prescriptions for their patients. It would make available to healthcare providers their patients prescription histories, whether the provider had an electronic health record or stand-alone prescription writing tool. At the same time, Stead said he would work out an incentive payment scheme for pharmacies and primary-care physicians to boost the currently low medication compliance rates among those patients with chronic medical conditions.
We dont have to do this complicated, end-to-end e-prescribing to create that system, he said.
The third thing I would do is to wrap any grant for information technology in a requirement that there be a site-specific improvement plan. That plan would tell you what the improvement is they want to achieve, what the change in work flow would be required, how the measurement (of the improvement) will occur and how their system will adapt to iterative change. Frankly, I think that would be much better than certification of IT systems.
In sum, Stead proposes using whatever technology is readily available to take little steps. Rather than trying to get the system right end-to-end, lets try to do a few things that improve things dramatically. I do think we can achieve semantic interoperability for an ever-increasing set of core information. If we simply aggregate the rest (of the data) without trying to homogenize it and use data-mining tools to paint the right pictures from it, youve got the right combination.
In building its VistA clinical computing system, the Veterans Affairs Department used an iterative software development method in which clinicians worked in close proximity with the software code writers building more than 100 modules addressing specific problems, a method that could be replicated, even with commercial vendors of comprehensive, enterprise hospital IT systems, Stead said.
The VA did one of our other recommendations, focus on the improvement and let the IT follow, he said. What our report says, that was actually the right model to do this stuff. Following that example, youll mix and match IT to get the data that you need to get the improvement you want.
So, I asked Stead if he is a fan of the best of breed approach to building a hospital IT system over deploying a fully integrated enterprise system from a single vendor? My question missed the point.
Im actually not advocating either of those scenarios, Stead said. What Im advocating and what Im saying explicitly in that report is to get those vendors to separate their data from their application.
In fact, the report called for government IT grants for bandwidth improvement, hardware and foundational IT systems, such as clinical data repositories, but not clinical applications.
Those vendors right now, whether they are best-of-breed or enterprise system vendors, their data is at the center of their systems and understands it better than any other system, and its very hard to use except in that system. Thats how they (the vendors) compete. What we recommend is to use those types of systems to automate processes within a facility and have them export the data to a repository that is independent of those systems that automate the processes. The repository could be an aggregate of several systems and it really doesnt have to be normalized, if you look at this idea of data-mining to recognize the key patterns you have in disparate data sets. You can do this if you break the historic industry tie between the systems that automate the process and those that house the data. If you export that data into a system-independent repository that can be accessed with simple data-mining tools and connectivity tools, you begin to see a better picture.
Stead said a piece of this vision has become a reality in a regional health information organization in Memphis, Tenn., that Vanderbilt helped create and still manages. It aggregates data from six competing hospitals organizations and it presents the information through an Internet browser that makes it available to 19 emergency departments and the hospitalists in all of the participating hospitals.
Regarding privacy, Stead said, The real power in IT in healthcare is to distill information from multiple sources, to make it easy to recognize patterns to make decisions. It would be possible to envision regulations that would put bars in the way of that aggregation and we wanted to say, please dont do that.