A New “Model” for Electronic Medical Record Systems
As a physician formally trained in computer science, I have the opportunity to look at today’s computerized medical record systems both from the perspective of a end user and as a software designer. It is perhaps because of this that I have been so persistently disappointed with the current state of clinical record software.
I am disappointed because despite all the fancy hardware and expensive software, our clinical records systems aren’t that much better than paper. We would think that a patient could go to any doctor and present their medical records the doctor could read them, but they can’t. We would think that it would be easy for me to get a CT scan report that was done at an outside hospital, but no. It actually has to be printed out and faxed, requiring not only human intervention and time, but if reentered into the receiving provider’s system actually converts a digitally stored report into a picture of a piece of paper, completely breaking the idea of an electronic record system. While information can be digital in one system, if it ever is passed on to someone working in another system, it becomes just another piece of digital paper. The sad truth is that despite our incredible investment in EMR systems, we have only created a massive collection of information silos, and have almost no way to transfer information between them – a system little better than the paper charts we sought to eliminate. And sadly, because these silos are hard coded and massive, innovation is stifled.
There is a very specific reason why our system operates like this, and it is that EMRs as a whole lack a common way to represent information. Each system represents medical records in its own proprietary format, and thus lack the ability to speak to each other. An thus no matter how wonderfully a EMR system represents information to its users, if information has to get out of the system, it can only be through pictures of pieces of paper.
So is there a solution to these problems? I would argue yes. But it requires a fundamental change in our paradigm – a change to a common “Model” for representing data.
To those that lack a programming background, when I say “Model” I mean something very specific. A Model is part of a computer programming paradigm called Model-View-Controller. This paradigm, pioneered by Steve Jobs and his team at NeXT, and continued in XCode at Apple, is based on the idea that any piece of software can be broken down into a Model, a Controller, and a View.
A Model is the part of the software that represents the data. All it knows how to do is take data in, store it, and serve it up when asked. By design, it has no idea who is doing the asking or who is stuffing data into it, and by design it doesn’t care.
A Controller is a piece of software that takes data from the Model and from the View, and then tells the View what to show to the User. Most people would consider the Controller to be the “brain” of the software. Like the Model, a Controller can potentially talk to multiple Views, and in a complex piece of software often does.
A View is the part of the software that shows the interface to the user, and organizes how user input will be presented to the Controller.
In the paradigm of a clinical record system, the View is all the windows you see and how you interact with them, the Controller is the brain of how that data is collected and how information is passed between different systems, and the Model is how it gets stored on a hard drive somewhere.
What I am proposing is that we create a single Model for representing clinical information that would be accepted across industry as the only way to represent medical information. Every vendor would be free to represent that data on screen and interact with the user any way they like, but when they store it there is only one way – because there is only one Model. Vendors could still fight for the best design to attract customers, or even create a wonderful custom system for only particular customer – all without destroying the portability of the data.
A move to a common model has tremendous advantages that we lack in our current systems. The single biggest difference is that it allows a complete shift of paradigm from a system where individual providers or hospitals keep isolated medical records to a system where a single patient’s entire medical record from every provider is kept in one place. Instead of storing records, hospitals would access the patient’s file, edit it and add to it, and then put it back where it is stored. If the patient went to another doctor or hospital, they would be able to access those new records. This alone would be a massive improvement in our healthcare system. Millions if not billions of dollars are wasted every year because records cannot be easily transferred. It is quite common for doctors to re-order expensive imaging tests or labs because they need information that already exist in another medical record system. With a common Model, this would be eliminated.
Another advantage to a common Model is that it would foster an incredible surge in creativity among software designers. Right now, there are only a few big players in EMR software, and it is almost impossible for a small player to get a foothold. Hospital systems pay huge sums of money to have Cerner, GE, or EPIC manage their EMR system, and thus store their patient’s medical information in the Model defined by one of those companies. While functional, none of these companies products have particularly great designs, all being relational databases designed by arguably unimaginative software engineers. The sad thing about this is that there could be an wonderful young designer out there with an incredible idea on how to represent medical records, but with the way the current system is he or she would have no possibility of breaking into the market. The current players are far too established, and the cost to switch to a new system too great.
But with a common Model, this problem is eliminated. As all players would agree to represent medical records the same way, any number of new interfaces, or Controller/View combinations per the MVC paradigm, could penetrate the market. If someone came up with a new system for viewing and editing records, it could be integrated into a hospital’s workflow at a low level, perhaps by only a few doctors. If it were liked, it could spread organically. This is not unlike the way that new web browsers have spread out and been adopted. They all work on the Model called HTML, and thus they each can be tried out and adopted or rejected by each potential user. With a common Model for representing medical records, it would be entirely possible for different physicians at the same hospital to use different medical record systems to view the same records. They could also use different hardware in different environments, such as a computer while on the ward and an iPad while walking around the hospital or in the operating room. The advantages are tremendous, which makes the lack of a common Model painful to use all.
With a common Model, hospitals no longer permanently store medical records, but rather access records that were stored elsewhere. Where that would be would be a question to answer, but I think the answer is that they will be stored in many places, an in most cases in multiple places simultaneously. It is quite easy for an individual to carry enough storage on a USB stick to carry their entire life’s medical record. That information could be mirrored to a cloud service that kept it backed up and available anywhere. Hospitals could download a copy when a patient is admitted and intermittently back it up to the cloud. There are of course technical issues, but they would be solved in time. Perhaps we will even evolve as a society to the point that we would accept the idea of storage being implanted under our skin, to carry our medical records for our entire lives. I’m just Star Trek enough to believe in such an idea.
I have discussed this idea with people in the industry, and many claim that acceptance of a common Model at this point is impossible, that there is just too much momentum to overcome. I would argue that there are countless examples of where others have succeeded in exactly this, and there is no reason why it cannot happen in medicine as well. For example, DVD became an accepted Model, as did MP3. In many cases, these models were started by one company and managed to spread throughout the industry, like Sony’s creation of the DVD standard, or Apple’s creation of the FireWire standard. The same thing can be done for medical records. It has to be done.
The current system we have, in its many forms, is really just a digitization of a paper chart. This just isn’t good enough. While this was perhaps the logical first step, it cannot be the last. We must move forward. In order to create a truly great record systems, we have to throw out the old ideas of how medical information is stored and represented, and ask ourselves how we would do it if we had no restrictions at all. Information must be patient centric, and interfaces must be agnostic to the Models they read and write to. Everyone could have the system they wanted, yet everyone could still communicate. It is my true belief that if this could be accomplished it would be one of the greatest medical technology advancement in history, and perhaps contribute more to the health of humans than any new drug or surgical technique.
It is my hope that true industry players will read this and consider what their role could be in this potential healthcare revolution. If I had to single out a single player that should take the lead, I would choose Apple. Apple has had incredible success and sponsoring and developing new Models, and has the commitment to design to create something great.
The only problem is that whatever industry heavyweight takes on this task doesn’t have all the people they are going to need. They have brilliant Engineers… but the lack the healthcare professionals that they will need to help them design what is truly needed.
I’ll be waiting for their email.