Bill Gardner at the Incidental Economist laments the state of electronic medical records today and contrasts it to a vertically integrated data exchange system that he used last week.
the car has been worked on at several dealerships in several cities. And of course, I have no clear memory of what I’ve gotten done when.
No problem. The car has a VIN number. Dealerships are independent enterprises, but Toyota has an international database that has all the service records from all the dealerships. The service manager and I are looking at the records in seconds and we quickly make a decision about the needed service.
This is interesting only because I can’t do this for my body….
The engineering and legal work required to make electronic health records interconnect is harder than the engineering required to connect Toyota dealerships. Nevertheless, it is seriously stupid
I was able to speak with a medical informatics person for a while about this post and what she said was that right now most electronic medical record (EMR)systems are pretty good within their own silo. If you, as a patient, stay within one EMR ecosystem, the results are pretty decent. Providers can see trends, they can see case notes, they can see allergies, and they can see any idiosyncracies about your medical history fairly easily. Too many of the systems are still mainly focused on data management and not provider-patient interaction or even provider-computer interactions so too many appointments have the doc looking at a screen too much. However, the back-end of any individual system on the market that has more than a couple major hospital groups using it is reasonably solid.
The problem arises when you as a patient switch between environments. If your primary care provider uses System 1 and your cardiologist just got bought out by a hospital group that uses its own in-house EMR system, the odds that those two systems will talk well with each other is slightly better than a Swahili speaking customer getting a perfect coq au vin in Seoul. It is possible, just highly unlikely, that the righ set of crosswalks and translation protocol will be in place between the two systems. Some of the larger systems will have rough translation protocols in place between systems, and there are developmental efforts to put a translation layer above a variety of EMR bases, but those systems are not widely deployed.
The big problem is simple for Toyota. They keep everything vertically integrated and they make it a requirement that all the dealers in their network buy the same piece of software with the same protocols, the same data dictionaries, and the same agreed upon sets of meaning. From here, it is just a nasty database management problem, and realistically, it is not too nasty as there is a unique key (the VIN) that already has external meaning. EMRs in this country don’t have shared meaning, they don’t have shared data dictionaries, they don’t have shared formatting (for example,how do you write a date, as I can think of at least seven valid ways, three of which produce major screwups in other format systems), and they don’t share the same protocol.
The VA gets around this problem because they are effectively in the same boat as Toyota. Kaiser gets around this problem as they are a mini-NHS segregated from most of the US medical system. Big players that own their own hospitals, doctors and insurance companies get around this problem as they can keep most of their patients in the same environment. But crossing networks and getting the systems to talk to each other is an ugly problem.