HIStalk Interviews George Schwend, CEO, Health Language

George Schwend is president and CEO of Health Language.

Give me some background about yourself and about Health Language.

I started in healthcare after college in pharmaceutical sales, which was boring. I moved to clinical laboratory sales, my first introduction to IT. This was back in the 80s when there was Cerner and Sunquest and a company called LabForce that I got involved with and developed an IT system. It just fascinated me. Then I expanded into doing blood banking systems. 

My real excitement came when we got out of the application side and got involved with tools. I was the founder of the company that delivered an integration engine called Cloverleaf. I don’t know if you’re familiar with it.

I am.

We founded that company and delivered it. Cloverleaf has probably had more owners than any other product in healthcare IT over the years [laughs], but it’s a very successful product. It changed the way people integrated systems. 

Healthcare IT started out as islands of information. Everybody bought best-of-breed systems, and then somebody realized one day that they were doing an awful lot of data entry. They started doing point-to-point interfaces, which were ugly and expensive. We came up with the same concept at the same time that STC did with DataGate. We were competing products and that product has gone on.

We went public, went on to another company name. I did what you’re supposed to do at that point — I built a new house on a golf course on a lake [laughs].

As I was realizing I was never going to be a great golfer and I couldn’t catch all the fish, I started thinking about the next horizon. The next horizon came from thinking about what we accomplished with an integration engine and then what the next big problem was going to be. In the final analysis, we had integrated systems and we allowed data to move around, but we never integrated information.

What Health Language is all about is true interoperability. Integration engines move data around. Language Engine, which is our flagship product, actually integrates the data so it’s usable across the enterprise, wherever it goes, wherever it’s needed. Health Language has been a pretty big success and we are growing rapidly.

Tell me what the Language Engine does.

We manage and handle standards and vocabulary across applications. We use consumer-friendly terms. We have physician-friendly terms. 

The problem with communications in healthcare is everybody speaks their own healthcare dialect. Machines to people, to nurses, to doctors, to laboratorians — a lot of stuff can get lost in the translation. If you create a large set of content that embodies all the standards — the financial standards, the clinical standards, the specialty standards — and you put that in a large database and then create some technology that will serve that up to whatever user or to whatever event is happening and tag everything, you get very usable data across the entire enterprise.

How important is terminology to where Meaningful Use is going?

I think it’s hugely important. Technology has been there for a while. We’re doing really neat stuff with technology. The problem is that we haven’t done a whole to improve the data or the information. 

For Meaningful Use, we make the data far more usable. Even in the case of triggering alerts, if the wrong terms might be used and there’s not a database that says those two things are the same, you’re not going to trigger that alert. The efficiency or the effectiveness of the computerization of healthcare is not maximized unless the data is close to 100% readable by everybody that relies on it.

We see ourselves in a very horizontal technology — horizontal across healthcare. If anybody generates or stores or reads data, does statistical or outcomes analysis, clinical trials, or is in the payer sector … if they’re using healthcare data, we can make their current tools better and make the data they’re looking at much more usable.

Are you seeing more interest because interoperability means providers have to talk to each other now, just like systems needing to talk to each other created the need for the integration engine?

A tremendous amount. Interoperability, I’ve kiddingly said, is our middle name. If you take HL7 and what it does — and it’s a very critical piece of the pie — and then take SNOMED and the ICD and all of those and put those and put them in a mapped environment where all of those things are usable, you’ve got your interoperability. You’ve got portable records that can be read anywhere and understood anywhere. I think we play a very significant role in Meaningful Use.

You also offer an alternative to providers for looking up SNOMED and ICD-9 codes to create problem lists and documentation. What’s important about that other than provider satisfaction?

Let me explain our business. We have a large technology group at Health Language. We also have a large medically competent content team. That’s doctors, nurses, laboratorians, all the different regiments within healthcare specialty. They build content and our technology people build tools that automate that and make it easily accessible either sitting on a network or embedded by an application vendor.

In the case of ICD-9 or ICD-10, or in the case of Meaningful Use or Accountable Care Organizations — which, by the way, in my opinion is a new name for an HMO, isn’t it? [laughs] — that content group keeps everything current. It makes problem lists almost a standard product. We allow individual users to create and add to those problem lists or make their own unique problem list. We serve all that up with the Language Engine, making it available to all users across the spectrum.

Who are your competitors?

A number of folks are trying to do a really good job with delivering content, vocabulary, and standards. What differentiates HLI is that we are truly an IT company of equal stature to any IT company out there, as well as a content purveyor. I don’t think anybody comes close to the tools that we provide and a true engine technology to make the delivery and the maintenance reasonably manageable.

Content takes a long time to build, so I would think being the first mover gives you advantage.

We believe it does. We believe that’s why we’ve attracted so many new partnerships from the system integrators out there. There is probably not a major system integrator that we’re not partnering with right now. They make proposals to the insurance side of the industry for handling the transition from ICD-9 to ICD-10. A key component in their service package is that engine itself and then the content we can provide with it.

It was a big deal when government licensed SNOMED for all users and, more recently, Kaiser donated its Convergent Medical Terminology to HHS. How important were those developments?

Kaiser’s a major player. Kaiser was probably the stimulus for our company. A bunch of people from IBM’s Watson Research Center and Kaiser started to attack this terminology communications and vocabulary issue in healthcare. That became some of the core basis for what Kaiser ended up doing. Kaiser Rocky Mountain won the Davies Award for what they did.

When those people were done with that work, they were cut loose. We were just finishing up with Healthcare Communications, which was the Cloverleaf company. I started to think about where we should go next. I hired a whole bunch of those folks and many of them are still with us today. They had a real core expertise on managing vocabulary and standards.

We’ve gone light years from there, but to answer your question specifically, when anyone makes available a good quality content mix like Kaiser had, we applaud it. We were the first to put out a press release to commend them for making that available. Our tools can adopt any standard, any set of content, and manage those in concert with all the others. It just enhances our position and offering.

What are your thoughts on the technical specifications about nomenclature in Meaningful Use?

They’re all practical. They’re saying, “If you’re going to claim this Meaningful Use investment in IT, it’s got to do certain things,” and those things are all logical. I don’t see a problem with any of them and I don’t see why anyone else should. I think they just make medicine and healthcare better.

What’s the state of readiness for the conversion to ICD-10?

I won’t be the first to say I think we’re all behind the curve on it, but a lot of people are gearing up to try to make a very hard run. The sane thing to do would be run parallels on ICD-10 and ICD-9 for literally a year — through all the seasons, through all the types of things that are recorded and charged for, etc. — to make sure you’ve got it right.

We’ve done some really interesting things in putting our technology on a Web portal and allowing customers to manage their own mappings and use our mappings and modify them. Those things are of improving the speed to market of getting ready.

One criticism of ICD-10 is that it’s a huge list of codes that are more granular, but hard to pick from. Do you see that as an improved market for your product?

Yes, definitely. The granularity is a good thing for healthcare. Fifteen thousand to maybe 150,000 — that’s a major paradigm shift. That’s why I believe the tools are absolutely essential right now. And as you know, ICD-11 is not that far behind.

We’ve been doing international ICD-10 business with customers around the world for almost ten years now, so we’re very ICD-10 astute.

And that creates new product opportunities?

Yes. The more you have to juggle and plow through to get it right, the better it is to have tools that help work you through those, that give you logical maps, that make you think through, “Am I making the right connection here? Am I using the right codes"?” And hopefully getting no one in trouble with fraud and abuse issues down the road, almost by accident, because it’s such a complex problem.

Do you see new terminology developments that will be needed for interoperability, such as for genomics?

I don’t see anything on the immediate horizon. I think the real issue is that all of the standards are in constant change and flux. They’re constantly being changed. If you could envision a pile of pick-up sticks and every stick had a different standard on it … we’re managing something close to 180 different standards worldwide right now in our content base. Everybody isn’t using all of them, but people use varying numbers of how many standards they juggle in a given day and those standards can change daily, weekly, monthly, annually. 

Keeping it all mapped together and keeping it all usable so nothing breaks is an art form. That’s the real problem if you don’t approach the situation from an, “I need a solution that will take me long into the future and not a quick fix on how do I get from ICD-9 to ICD-10.” The demand for all kinds of vocabulary requirements and all kinds of different standards is going to get greater, not lesser.

What’s the current state of text analytics and do you think there are additional opportunities to leverage that?

We do a lot of natural language processing-like services. We are talking with a lot of the text, the voice, the natural language processing folks. We have a lot of projects going on. Nobody is, I think, 100% comfortable with where all of that is without some human intervention, but it’s getting closer all the time.

Do you think the PCAST report’s recommendation to turn existing data into discrete document data by tagging it could work?

The problems always come when what’s written is “pneumonia” vs. “no sign of pneumonia,” catching all those little innuendos and not getting into trouble and not just grabbing “pneumonia.” Do I think it’s possible? Yes, I do. I think it’s just a matter of time. I just don’t think we’re there yet.

There are two ways to solve the problem of everybody wanting coded data — either make the providers code the information going in or to try to code it on the back end. There’s not really any easy way to do it except maybe to use products like yours to make it more palatable on the front end.

That’s true, but we also do it on the back end, too. We’ve run historical databases through our Language Engine and gotten a very high turn — not 100%, so I’m always reluctant to hold up some false Holy Grail when somebody will ask me to make it work for them and we just can’t get to 100% — but we can do an awful lot of that today.

For aggregated studies or for public health type uses, it’s probably plenty good. You might not want to make an individual treatment decision from it, but I would assume that if you just had lots of data to plow through, you could make some general inferences that you’d be comfortable with.

I agree with you. That’s an excellent way to state it. You’re right.

Where do you take the business from here?

We see a tremendous amount of growth over the next couple of years. We believe that the need will continue to grow and we’ll continue to be intimately involved with all kinds of standard delivery across the entire spectrum.

We have some development projects that we hope to announce down the road. Right now, we’re very busy taking care of the high demand for ICD-9 to ICD-10 and Meaningful Use standards. 

We work with clinical application vendors. We work with system integrators. We work with individual hospitals like Partners in Boston and Ascension Healthcare. The more the government continues to try to move us into a more common world, the more of a need and the greater the need is going to be for what we do right now.

Any final thoughts?

Healthcare and the initiatives currently going on are the next quantum major step for healthcare and healthcare IT. Everything that’s being done today, everything that’s being required and requested, will move healthcare into truly the next generation. It’s an exciting time to be involved and I’m just glad we’re part of it.