With T&E trimmed, just as it is in all organizations, HCI editors rarely get a chance to interview sources in person. But when our source turns out to be a fellow New Yorker, it’s time to take a cab across town for some real one-on-one time. I was lucky enough to do this recently, and have one of my best interviews of the year as a result, when I linked up with NYU Medical Center CIO Paul Conocenti. The result? An engaging look at a major medical center overhauling its IT environment to completely integrate the continuum of care. — Anthony Guerra
GUERRA: I always thought if you wanted to get everything from one vendor, GE and Siemens would be particularly attractive because they also have modalities, but you went with Epic, is it because PACS and modalities are easily plugged in, and you don’t need them really integrated?
CONOCENTI: My thoughts are that the big guys like GE and Siemens are just that, they’re really big guys and somehow they have not been able to build software that can compete with these other major vendors. They’re trying, and Siemens is trying to build software from the ground up and there’s nothing better than German engineering, except it takes a hell of a long time to get there. I think it’s wonderful, but it takes a long time.
GE just feels they can buy everybody and piece the stuff together, and that hasn’t worked, and they, in my opinion, Six Sigma-ed themselves into being paralyzed. Now, these are great institutions when it comes to medical equipment; but the reality is that there are adaptors and plug-ins for these devices, if you will. At the end of the day, they’re devices that are streaming data. And they are not as integral to the workflow for taking care of patients as these other systems.
So, I could plug in GE, I could plug in Siemens, there’s standard adaptors, and in fact, Epic and Eclipsys and many of the other vendors use the same third-party software that does the adapting, so that if you’re in the ICU, you’ve got electronic charting right off the modalities in a PACS environment. Now with 3D PACS coming onboard, it’s more important these days, so right now there is a different strategy. Siemens’ strategy is that they’re going to tether the PACS with their RIS system, but what’s happening is that the PACS technology has advanced so much that the worklists that are normally associated with a RIS are now embedded in the PACS, and so you don’t need as much power in the RIS because you’ve got it in the PACS.
So, now with an Epic, which is really a fine radiology information system but is not a PACS system, plugging into a PACS you can now have a choice — you can go with AGFA, Siemens, what have you, you can pick the best 3D PACS because those things are now embedded.
These modalities, whether it’s in cardiology, or in radiology are, at the end of the day, the device plug-ins to the clinical information system. The thing that physicians work on while taking care of patients which allows them to document a 15-minute encounter while looking the patient in the eye requires different sorts of design.
And so, I think the Siemens and the GEs, although they were positioned five years ago to take the market — I mean, they had the game, okay, they made a couple of poor purchasing decisions in my mind and, in some cases, took too long delivering product, that would be the Soarian (Siemens) thing — have left the market open. They’ve delivered some good stuff, but not enough, not fast enough. So, we can’t wait.
And now you go into the home and Microsoft, with its HealthVault, has gotten some really nice adaptors into the home appliances, whether it’s a scale or insulin measurement or what have you, they’ve got a lot of device adaptors. And they’re adaptors that can plug into things. And so it really becomes a device that you’re integrating, and there are so many different devices, which is fine, as long as you have the adaptors.
The clinical information systems that house the data are really where you’re getting the efficiency and you’re getting all the decision support rules. A rule might say, “This patient has this allergy and is of this particular heritage; and you’re trying to do this drug and that’s not going to work because of this research.” That’s embedded in the clinical information system, not the modalities, which are just capturing data at the end of the day. That’s the heart. The difficulty in interoperating that data over interfaces is that each of these systems and each of the implementations of those systems are different. I could have Eclipsys here, and I could have Eclipsys at Presby and I guarantee you, the data in our two systems are totally different.
GUERRA: The same product used differently.
CONOCENTI: Right. They code something X, we code it Y. It’s the same drug, we haven’t synchronized it. Are there any standards on problem lists or procedures and all of the notes and documentation that goes on? No, now I want to interchange data with Presbyterian, another Eclipsys system, people say, “Oh, let’s do Eclipsys, why not?” Totally different. So, that’s the challenge of interoperability because if you really want to use that data for decision making, and embedded in a rule that says, “If it’s this drug and if it’s this history, then you need to get that discrete data,” you need to then translate that data into some standard value, and that’s what these HITSP guys are all about, trying to do those standards.
That’s another work in progress, it’s good work; so LOINC has been around for a long, long time. You go to any of these systems out there, these lab systems, and you go dig into the system and see if they’re using LOINC codes internally in their database. They’re not. They’ll now be able to translate to LOINC outbound and probably take it inbound and convert backwards, but it’s a conversion process, that’s what’s going to happen with all these HITSP-type standards.
GUERRA: With HL7 and the CCD, etc.
CONOCENTI: HL7 is a transport mechanism, it’ll transport data in a specific format so that the system that gets it knows to take this data, put it here, put it there. But the data that it carries is not standardized.
GUERRA: At all? What about the CCD?
CONOCENTI: The CCD is document sharing with codified data in it, and it’s better because it carries some of the metadata associated with the description of what that drug is, so what you could do with that is, because it’s all XML-based, you’re able to view a document, even though they call this drug X and I call it Y, we’re going to put in the description that it’s insulin, and they’re going to put it as insulin XYZ, they could call it something different, but they carry the description with it. So, when I display it as a document, even though one code is X and the other code is Y, it still says insulin, so I can read it. That document now can be transferred as a note from one system to another, and voila, it’s like a fax, a document, I can read it.
But if I take that discrete data and push it into my system from theirs, then I suffer from the same problems that I do with an HL7 message, because now I have to know that their X is insulin and their Y is insulin and I have to do a conversion. Now, meanwhile Sinai’s calling it Z, and then you multiply that by 10,000 or however many providers there are, and you can see it’s unworkable. So, that’s what’s happening around these exchange centers, and these HIEs, these hubs. With NYCLIX and SHINNY, we are trying to get our arms around these member organizations by conforming to interoperability standards. We would have the ability to say, “Okay, well, let’s not start with the universe" — that’s this whole NHIN — "let’s just create a coordinated care zone around diabetes and within diabetes, let’s standardize some terminology.” All of a sudden, you can get your arms around that one, but on a national scale, it’s more challenging.
GUERRA: How much do you follow HITSP and try to stay in line with what they are doing? John Halamka is the head of HITSP, he’s also the co-chair of the Standards Committee, we have to assume that some of his work is going to find its way into codified law. How do you stay in touch with that?
CONOCENTI: I stay in touch by going to those conferences and being part of some of the committees. I was on the ONCHIT Security and Technology committee, and so I stay involved. I’m participating in the state collaboration and I know John, I know these folks, they’re good people, we follow their work.
So the CCD is the first thing that’s actually coming out that’s implementable, and that is a direct result of Halamka’s harmonization efforts of two, three years ago. And that was terrific work because we had HL7, and the ASCII — two national standardizing bodies doing the same thing, and not agreeing. That’s great (laughing), I have a national standard body, the problem is I have too many standards. John’s work was in harmonizing so that the ASCII organization came up with a CCR (Continuity of Care Record), and the idea is let’s just carry the data from one care setting to another that’s pertinent to the patient’s care. Let’s not worry about the universe of this.
Now HL7 has been working on reengineering this whole thing and they had the CDA (Clinical Data Architecture) and that was an all-encompassing thing, and a lot of years of work went into it, but it wasn’t specific to the continuity of care. There were very good things that were in the CDA work and there were very good things that were in the CCR work, and I think one of the brilliant things that John did was he came up with the CCD, and that CCD now is a hybrid of the HL7. As a result, we now have a document sharing standard and that’s going to enable interoperability, more than anything else in this country.