Just the other day I was at a local Vegas Tech event and happened to run into a government contractor that worked in IT. As we got talking I told him about my work with EMR and EHR. Once he heard those terms he started to recount his experience evaluating a contract position where he was to work at connecting the VA system with another government entity. He then said, did you know that the VA software runs on something called MUMPS?
Of course I’ve heard all about MUMPS and so I told him how a huge portion of healthcare IT is run on the back of MUMPS (My understanding is that Epic uses MUMPS as well). Obviously, MUMPS has its benefits since it’s gotten us this far. I even remember some past threads where people have argued some of the advantages of MUMPS over newer database technology. However, I still stand in the camp that wonders how we’re going to get off MUMPS so we can enjoy the benefits of some newer, more innovative technology.
Something called the Axial Project basically asked this same question back in March 2011 when they posted about how to Architect Vista for 2011 (which is possible since Vista is open source). They provided a really insightful look into why MUMPS has done well in healthcare and what current technologies could replace it. Here’s that section:
So if I were starting a Healthcare IT company would I invest in building on Mumps/M? No. There might be some business in supporting legacy applications, but very little innovation. I am not attacking Mumps/M from a technical perspective, I am trying to be pragmatic as a business person. So we need find an alternative. So you probably think I am going to say MS SQL Server or Oracle thinking I want that 100/hr price tag. Thanks, but no thanks. So I am not in it for the money, I must go the other way. PostgreSQL or MySQL. Intriguing, but still a no go. I have learned over the past 18 months that Healthcare data has very little integrity. One of the reasons I believe Mumps/M has excelled. Storing objects vs Storing relationships in normalized structures is not valuable to this market. Too many views of the data are required depending on your role you play in the system. I would try to use a NoSQL database like MongoDB, Cassandra, or CouchDB. My preference would be MongoDB because there are drivers for Ruby, Java, .NET, and Python. Also, these systems are truly data entry/reporting tools at their core. I need strong query support which MongoDB has through it’s BSON data structures without a ton of map/reduce requirements. So let’s go back to finding some resources that can help.
The part that struck me was when it said, “I have learned over the past 18 months that Healthcare data has very little integrity.” That makes a lot of sense and explains why a NoSQL solution could work well.
Turns out, Axial Exchange has brought on the previous COO of RedHat, Joanne Rohde, to work on the project. Check out Axial Exchange’s presentation at Mogenthaler’s DC to VC 2011:
Looks like Axial has shifted from redesigning Vista, but they’re working on some interesting stuff.
What do you see as the future of MUMPS in healthcare?