Composed of NewYork-Presbyterian Hospital/Columbia University Medical Center and NewYork-Presbyterian Hospital/Weill Cornell Medical Center, NewYork Presbyterian Healthcare System is one of more renowned organizations in the country. The system is also affiliated with two Ivy League medical institutions, Columbia University College of Physicians & Surgeons and Weill Medical College of Cornell University. Aurelia Boyer — senior vice president and CIO, NewYork-Presbyterian Hospital & NewYork-Presbyterian Healthcare System — is in charge of making sure the massive amounts of data collected and produced by the organization can be put to good use by patients, physicians and researchers. Anthony Guerra, HCI editor-in-chief, recently had a chance to talk with Boyer about her work at NYP.
AG: Portals are turning into much more than bill-pay screens. Tell me about the evolution of your portal strategy over at NYP.
AB: That’s interesting you put it that way, because we did first put up the ability to pay online and to get a lot of information, but as we were deciding what to do around patient access to the medical records, we had to consider the larger IT environment we were operating in.
First, a little background so things are more clear. We’re an interesting organization in that the hospital resulted from the merger of The New York Hospital and the Columbia Presbyterian Hospital, and we have two separate medical schools. We do not have any single governance structure across the three organizations. So, where at Hopkins you have a CIO that supports both the school and the hospital, here our governance is separate. So we have a physician organization on the Cornell campus and a different physician organization up on the Columbia campus. As the Cornell practice organization implemented a separate EMR for the physicians, not the same one we were using on the inpatient side…
AG: Can you tell me what vendor they use, the physician group?
AB: The downtown physician group has Epic.
AG: And who did you have on the inpatient side?
AB: Eclipsys. At the time we made that decision, Epic wasn’t doing much on the inpatient side and Eclipsys was having a hard time getting started on the ambulatory side. So if you wanted the best of breed, you really had to split the baby.
AG: And when was this approximately?
AB: I bet it’s at least six years ago. And so with that environment in mind, we had to figure out how we were going to have the hospital inpatient record reflect the ambulatory record. (The practice is) going through Epic’s My Chart at this point in a pilot phase; it’s not fully implemented. So, the notion of the personally controlled health record came into play in that, even within our organization, the notion of sending patients to multiple portals to pull together their information didn’t seem like the best approach for us.
Second of all, we’re looking at how the RHIO is going to unfold, and will that be a success or not, and how long would that take even if it is successful? When we saw the Microsoft approach to HealthVault, I looked at how the industry was evolving. We wanted to act in accordance with our major belief, which is that it’s the patient’s data; we felt that an approach like HealthVault is the right approach. Now, it doesn’t have to HealthVault. It could be Google, it could have been Dossia, but it’s that approach which was going to make things easier for the patients, and the patients really were the ones who own their data, not us.
AG: Tell me when you started using Amalga and why you started using that, and how Amalga and HealthVault work together.
AB: I think after we started talking to Microsoft about the HealthVault opportunity, they brought Amalga to the table. It was interesting. When we started, Microsoft saw them as separate products, and I said, “No guys. If I put the data in Amalga, why don’t you connect Amalga to HealthVault?” We did initially start with Amalga for its analytic potential, not for HealthVault, but we moved to that approach very quickly after beginning to work on both projects.
AG: Tell me a little bit more about the different vendor applications and databases you have in your environment. You mentioned one physician group has Epic while the hospital uses Eclipsys. Are we talking about multiple hospitals?
AB: Yes, for New York Presbyterian, proper, we have five inpatient hospital settings, one of which is psychiatric, and they also use Eclipsys and are fully automated, all their documentation as well as portal entry.
AG: Are all five of those hospitals on Eclipsys?
AB: Yes they are.
AG: Ok, so you’ve got five hospitals on Eclipsys. What about the other physician group?
AB: The Columbia Physician Organization is just in the initial stages of implementing Allscripts for their physician offices, and the hospital does use Eclipsys in ambulatory-based clinics that are not private physicians.
AG: It sounds like one physician group implemented Epic, and recently the second group went with Allscripts.
AG: Did you attempt to talk them into either Eclipsys or Epic to simplify things?
AB: We did not attempt to talk them into Epic. They did look at Eclipsys and consider Eclipsys as an option, but felt that Allscripts did the physician practice better.
AG: So, does Amalga help you move data between Epic, Eclipsys and Allscripts?
AB: The data that NYP has been uploading to Amalga includes text-based exam reports of different types (cardiac catheterizations, EKGs, Echocardiograms for example), Eclipsys CIS data (discharge medication lists, discharge notes both nursing and physician, chest tube drainage for example), and other departmental systems (operating room, blood bank, laboratory for example). Amalga provides functionality beyond being able to view this data to provide care to a patient — we have other systems optimized for that function. Amalga extends analytic capabilities to the data and, most powerfully, it allows us to pick out data elements for analytics that were previously stored only as part of a final report for storing and viewing.
A specific example is to find through Amalga all the patients who had an ejection fraction of less than 35 percent (a finding in a cardiac catheterization report which is not stored as a separate data element in a results reporting system the way a single laboratory value is stored) and who had a QRS interval over 1.1 msec (which requires calculating the different lengths of the EKG wave components reflected in an EKG report). We can now search the database to retrieve the patients that meet both criteria. This is not possible with a database developed for results reporting, or most data warehouses where the text-based result is stored as a single element. The ability extends beyond finding the data element; it also allows the calculation of the values that are relevant to the analytic in a single step. This was not possible with our current data repositories.
AG: What is the traditional method for inquires like that? Does it filter up to the CMO and then somehow get over to IT in terms of a request for specific medical data or a specific measure? Is there a formal process?
AB: The formal process around most of that would be through the IRB method here because, generally, if you’re asking a question like that, you’re doing it for a research purpose. Then, the other way we send requests for data is through the quality department, and so we have a physician that’s the chief quality officer, and so that would be the other formal mechanism of how a request like that would come to us.
AG: So Amalga helps extract the right data?
AB: Quite honestly, Amalga has been faster for us than our homegrown repositories were. I don’t want to say it’s always there, but in this case we had done a lot of work around cardiology, and so it was there. Now, we’re going to move through other areas, but, once again, we’re pretty far through some of the basic clinical data. Obviously, we may not have every neuro piece of data or other things, and so that data flow really is a formal research process for the most part.
AG: Let’s talk a little bit more about www.mynyp.org. When patients log in, are they in the HealthVault application?
Part II Coming Soon