BTE: Can you tell me a bit about the Memphis health information exchange?
Mark Frisse: First and foremost, Memphis is 100 miles away from Nashville. I came to Memphis in 2004. At the time, Medicaid was in a catostrophic state in Tennessee, far worse than any other state. Governor Phil Bredeson in a previous life happened to be the founder of the largest HMO in the US as well as a Harvard physics major and computer programmer, so was no fool to technology.
He came in and said he was going to fix Medicaid and he did, which is pretty impressive considering we don’t pay any state income taxes, as opposed to Illinois where both you and my family live.
The city of Memphis has the highest concentration of Medicaid patients in the state, and there is a hospital there called the Regional Medical Center (The Med) which was a sinking black hole.
The Governor immediately recognized we needed to fix The Med because if it went under everyone would be swamped. Central to that rehab plan was information technology, so the central message I am trying to convey is we had strong leadership who truly understands technology, delivered the keynote at HIMSS. He is not really a nerd but he understands this stuff, particularly what technology could and could not do. We needed that strong unifying leadership to bring all of these hospitals together because, just like everywhere else, hospitals are intensely competitive and don’t play nice.
When I took the project on I was working in Minneapolis as a consultant and I jumped at the opportunity because Memphis is a very poor city and if you read Clayton Christenson’s The Innovator’s Dilemma you will understand its easier to innovate in less wealthy areas because there are no other options, but you could never accomplish these things in Boston or even Chicago because there is too much money in the game.
In our paper we showed approximately $2M net societal savings, but if its a 100 percent payer environment, somebody is losing $2M, so that is the real barrier here in our current reimbursement system – a net societal savings is coming at somebody’s expense.
The notion of knowing what everybody else is doing is a very powerful one. If, for instance, I am at University of Chicago hospitals and somebody leaves my hospital and is readmitted somewhere else it still counts against me. If a patient goes to multiple hospitals and gets a CT scan at both simply because neither knew the other one had already conducted the same test, one is going to get screwed because the payer group is going to see that and say “we aren’t going to pay for this twice”, since in their eyes it’s like paying for the same drug twice in one month– it doesn’t happen.
In 2004, prior to HITECH, Vanderbilt Medical Center had this architecture that could take any data in any format dirt cheap and wrap it in an identifying label, such as blood sample, and a program would then show researchers how to make the data look like everyone else’s data and then we were able to simply use a secure web browser to bring it up, which made all data essentially uniform even though it’s not.
So what we did was built a vault that each institution would use as essentially their off-site storage and each group could technically and legally flush that vault of all data at any time, so each felt very safe and didn’t feel like their data was all mixed up with everyone else’s even though it was in the same centralized computer facility. Then when somebody hit the emergency department, an alert went out across the system combing through all of the data in the vault and pulling that particular patient’s information out using a web browser, an admittedly not so eloquent manner, but useful, and the doctors could then look and see what else is going on.
Most of the time they looked at discharge summaries, x-ray reports and labs and the cost to the hospital was about $50,000/year. We didn’t say “you have to conform to this particular standard or that standard”, we said “what do you have, we will transform it into something useful”. So the total cost for each participating hospital is $50K, and the total cost of the entire project was approximately $880,000 during the year we did the study. We could operate the system for basically $1 per patient per year, which is really not that much money when you consider that capital expenditures around here is closer to $8,000 per patient. So it is actually very cheap to operate.
We worked very hard on data sharing agreements, we are ahead of the national curve on that issue using stuff from the Markle Foundation, which was about a 9-month process. We took the project out of the c-suite and got middle level individuals involved as well as people on the ground. We got the lawyers in late in the game because we knew what the issues were and we worked those up. It was a real laborious effort in its early days.
So the technology was developed at Vanderbilt, it’s a derivative of what we have been using here for fifteen years, and was licensed off to a third party Informatics Corporation of America (ICA) and ultimately we did move all of the data over to ICA and are currently under their independent management. Vanderbilt and myself have nothing to do with the Memphis health information exchange anymore. We built it, established its ROI and built an operational model, we showed it could save lives and then we spun it off.
It takes a lot of failures, 2-3 at least, before a market can determine what its worth. HIE is an immature economic model currently, but so was email and pretty much any computer system in the early days of its development, so the important thing to remember is the idea succeeds and its usually the same people behind that idea simply moving from one organizational form to another until one finally breaks through and finds a unique Michael Porter-like value proposition and they really nail it. We put this system in front of docs with a separate web browser and just let them use it at their will.
We saw the system used in about 7-8% of all ED visits, which makes sense because you aren’t going to use the system when somebody has a knife in their back, or if there is a sore throat, but you are going to use it if a patient looks ill and as a doctor think ‘I wonder if this patient has been here before’ or ‘I wonder if they have done this CT scan before’. We just said ‘use it if it seems right’ and doctors did the right thing using it for 7-8% of the visits, and just with that little bit of use in emergency departments our savings were about $2 million per year, lets say net savings of about $1 million per year.
And we didn’t really expect to see anything happen because most of the action in health care is in the coordination of care, the transition such as discharge from the hospital, it’s when the ambulatory guys are hooked up, but we didn’t expect to see significant cost savings from use of this technology in the ED because the ED is not a major driver of costs generally.
Another point I want to make that is very important is that these are not claims data, these are real clinical data, and its for everybody. Every patient had the chance to opt out and say ‘I don’t want information from this hospital to go there anymore’ and between 1-3% depending on the hospital did just that.
So you could say no, and here is a great example, Steve Jobs had his liver transplant in Memphis while this was going on. So I called my privacy officer, as our people are not allowed to go on and just do queries in the system, and asked him to go to the software guys and find out if there was anyone in their late-40s to early-50s with a liver transplant in our database in the last month. There was nobody there because each hospital could block data to keep it from being in the database at all, or if it was in the database we could put a flag on it so nobody but the software people could see it, and in this case Steve Jobs was not in the database at all.
When we set up the system people kept saying things like ‘its not cost effective’ or ‘if I share patient data with other groups it’s going to hurt my business and it’s going to help my competitor’, and this is coming from the senior executives of all the hospitals. But what we learned over time was it doesn’t hurt hospitals from a competitive standpoint to make this data available to all hospitals in an exchange and it sure as hell does help the patients and gives doctors a chance to do the right thing. So believe it or not, even something that obvious was like a revelation. It was difficult for hospital executives to understand that the actual differentiating factor that determines if a person goes to this place or that is how easy it is to park.
We are not in the economics business, we are in the trust business, and all of this stuff is about trust because you can’t develop these new emerging business models if you don’t trust the people. So this project was all about trust, and all about return on investment, and the only way to really get return on your investment is to keep investment low. I can point you to a HIMSS address given by Governor Bredeson which I live by in which he made three key points. The first was build version 1.0 first, the simplest version possible, and as the Governor put it to the HIMSS audience, “I am afraid they are asking you to build version 6.0 before you build version 1.0.” The second thing was do one or two things really well to show you can make a difference, so we knew the common pain point, where all of the hospitals had a common interest was in the emergency departments with indigent care, because not only did the EDs know who their frequent flyers were, they also discovered that the frequent fliers were the same people at all of the EDs.
About ten years from now talking about this is going to be as dumb as talking about the Internet. Kids are going to be like, ‘duh, of course you make this data available!’
[Be sure to check tomorrow for Part 2 of this interview]