Epic and Cerner (and just about everyone else) Must Go

How medical systems should be written

The Problem

The entire way the Electronic Medical Records (EMR) industry is set up is just wrong.  The entire industry is set up around specialties, well over 100 of them, making an EMR nearly impossible to write.  This means that the EMR vendors are literally writing over a hundred different programs to cover medicine.  Maybe these programs are all in one place, but they are still designed and written by disparate teams, have different looks and feels, different training and different ‘codes’ to translate (you don’t need codes at all)

The Consequences

Because each specialty has its own software, we don’t see any reuse of effort or code in the EMR, the whole thing must be written by hand, from scratch. This is a daunting effort by itself and probably why Epic and Cerner are virtually the only hospital systems.  More on Epic and Cerner later.  Just as bad, you have to have specific EMR training in your specialty and that training doesn’t translate to other specialties.  I think last time I checked it takes six to eight weeks to learn how to use Epic’s silly program per module.  The bad news is that all this is wasted time and effort.

You Don’t Need Specialties

In medicine, like in everything else, most practitioners specialize.  Even a generalist is a kind of specialist.  This is kind of like a Ferrari mechanic being familiar with Maseratis but not able to just jump in and repair one.  Ferraris and Maseratis work exactly the same way and are even produced within a few miles of each other but work slightly differently.  Why then do we assert that there are no specialties?

Commonalities

Even if we are designing or repairing Ferraris or actually practicing medicine, the shared knowledge between the specialties is large.  If we aren’t actually practicing, we don’t even need to know the differences.  Almost all people work the same way with the same parts in the same locations.  What we are interested in is managing the process of delivering medical care, not actually performing procedures.  We have doctors for that. 

Managing Processes

Unless we are building or repairing Ferraris or Maseratis or actually practicing medicine, what we are really interested in is saving time and money.  That means managing processes.  Managing processes involves streamlining, consolidating and removing wasted effort.  This has nothing to do with specialties.  Even in documenting the encounter, we need to know what was discovered that led to test being run, what the results of those tests are and what plan we have to solve the problem, that is, procedures performed.  For example, Ferraris and Maseratis both have water pumps and while the procedure for replacing one varies between the two, the replacement procedure is documented and billed exactly the same way.

Conclusion on Specialties

Let the specialists worry about the specialty.  In fact, if we write our software correctly, we can use the same piece for documenting a medical patient encounter or taking your Ferrari in for repairs.  The process is the same.  Intake, observation, testing, diagnosis and treatment.  From that perspective we can let the smart people make the decisions about what they want to do, we just need to document it.  

Problems with Epic

Epic is written in MUMPS.  This was probably fairly revolutionary in the 60s, but is wholly inadequate today.  If Epic weren’t the world’s largest and richest medical software vendor already, they would have failed long ago.  This antiquated language used for nothing else, makes their program unmaintainable and therefore almost unfixable and nearly impossible to extend.  There aren’t many people who speak this language since it isn’t used anywhere else.

I personally have asked Epic for an HL7 feed and they state they can’t provide one.  They will send me a comma delimited text file (CSV) and Portable Document File (PDF) via SSH File Transfer Protocol.  The 90s called they want their “data exchange” system back.  That means I have to write a program to go pick up their files read them, import the CSV data into my database and then show the PDFs to a HUMAN BEING to type into our system.  You read that correctly.  They send PDFs that have to be opened, read and re-entered in the year 2024. The mind boggles.

Epic costs hundreds of millions of dollars per hospital installation.

Just as bad, is that some Epic installations just fail.  After spending the hundreds of millions and spending a couple of months in training some hospitals have to revert to paper.  You heard that correctly as well.  In 2024 we have mission-critical, lives-depending-on-it systems that fail and people have to use paper.

Problems with Cerner

Cerner is kind of the opposite of Epic, it is written in everything but MUMPS.  That means that nobody really has any idea what the whole thing does nor how it does it.  Like Epic though, this makes their program unmaintainable and therefore almost unfixable and nearly impossible to extend.

I don’t know how long Cerner training takes; I suspect it is a similar six to eight weeks that Epic requires.

A Cerner installation costs hundreds of millions of dollars.

Sometimes Cerner installations just fail.

Codes

Epic and Cerner and everyone else as far as I know, use a proprietary code set.  Insurance requires ICD10 and CPT codes but these aren’t sufficient to document a patient encounter.  This means that if you work at an Epic facility, you can’t just go apply your knowledge at a Cerner facility, or any other EMR at all.  Again, this is wasted time and effort.

The Solution

Codes

You don’t need codes or coding.  Our friends at CAP and NIH are smart and came up with a code set called SNOMED_CT that contains over 350,000 medical concepts and is suitable for documenting a patient encounter.  Even better, the National Institutes of Health (NIH) combined SNOMED_CT, RxNorm, the national drug database, and the Value Set Authority Center (VSAC) into the Unified Medical Language System (UMLS) that not only combines these three but translates them into a self-referencing and self-documenting schema where the same programmatic techniques can access any of a large number of medical nomenclatures.  The UMLS has over 14 million concepts.  That means we can translate back and forth from ICD and CPT as well.

With SNOMED we can easily search for the concept we need and it is adequate to document the entire patient encounter.  If we assign dollar values to procedures (and eventually pay for wellness instead of procedures) then we can pay for work done in real time.  We don’t ever even need know the ‘codes’ or have someone translate our documentation (a medical coder) into machine readable nomenclature. They are just part of the encounter documentation and are saved at the time with the language the doctor chooses.

One Solution to Rule them All

First, we need one, simple program for all specialties.  We know now that we shouldn’t have specialties, and we know that there is a universal code set, but what do we do with that knowledge? We know we need vitals, and we need them in a way that can be written from an automated status monitor.  These are things like blood pressure, blood oxygen, heart rate etc., and with the UMLS we can document everything else as well. 

Surveys

There are things that ARE specialty specific like questions that are asked and data that is captured and recorded.  These things are not really structured data they just need to be recorded for recall at a later date.  Maybe it is even just a checklist that documents a patient’s general health before a procedure in any number of categories.  These aren’t diagnoses, they are anecdotal things either we observe, or the patient tells us.  They do however need to be documented in a system that is designed and configured for that specialty but the people that perform it daily. These are simply questionnaires, or surveys.

Installation and Configuration

One word: Software-as-a-Service.  You don’t need installations.  You go to the site and log in.  Make everything configurable by the user.  Like the surveys above, we don’t care what questions you ask, you could require the color of the patient’s underwear for all we care, but you get to choose the questions.  Everything is configured that way, to be user configurable out of the box and easy to use and maintain with no training required.

We have shown why you can’t use Epic and Cerner et al., and what to do instead.  WE are working on this very system right now and are deploying the survey portion today.  In the next month or two we will have the non-specialty EMR up and running and ready for deployment and use.

Take a look at this article as a video here.

Be sure to subscribe to your bi-weekly newsletter as well.