The Rest of the Story

Mark Kleiman (I finally spelled it correctly) has a post on electronic medical records (EMR).  He points out that the VA hospital system has developed, at great cost, a system that actually works quite well.  My wife used to work for the VA, and she thought it was a great system.  There were some glitches connecting to the main hospital computer from satellite clinics, but there were workarounds that got the job done.

Apparently, there is an effort now to develop national standards for EMR.  Mark wonders why the government does not simply adopt the standards of the VA system for the national standards.  He points out that the system is in the public domain, so it shouldn't be difficult to do this.  I guess no trade secrets would be revealed. 

So why, in the scramble to develop a set of standards for national adoption, isn't there active consideration of simply making the VA system the national standard?

Of course, the private corporations that develop EMR systems would object, since they all have a vested interest in the standards. 

I could see a few non-financial problems with this, although the problems are minor.  For one, most medical offices are governed by state law, whereas the VA is not.  Legislation pertaining to medical records may vary from state to state.  This is especially true for prescriptions for controlled substances.  (EMR systems generally include prescription-writing software.)  Another problem has to do with privacy.  The VA system allows any employee to look up any record, including those of their fellow employees.  (Many VA employees are vets themselves, and thus patients at the same hospital where they work.)  The system keeps track of who looks at what, and the logs actually are checked by a human.  Anyone who accesses a record, without a clinical need to have that access, is subject to serious consequences.  That works in the VA, but I doubt it would be acceptable or workable in the private sector: patients are very concerned about privacy.

I think those are problems that could be corrected fairly easily.  Potentially more serious is the fact that EMR systems have not matured fully.  There is some evidence that EMR may actually increase the rate of certain types of error:

The researchers found that the CPOE [Computerized physician order entry] system they studied facilitated 22 types of medication error risks. Examples include fragmented CPOE displays that prevent a coherent view of patients' medications, pharmacy inventory displays mistaken for dosage guidelines, ignored antibiotic renewal notices placed on paper charts rather than in the CPOE system, separation of functions that facilitate double dosing and incompatible orders, and inflexible ordering formats generating wrong orders. Three-quarters of the house staff reported observing each of these errors risks, indicating that they occur weekly or more often. Use of multiple qualitative and survey methods identified and quantified error risks not previously considered, offering many opportunities for error reduction.

In practice, these systems result in a net reduction of errors, that is, they prevent more errors than they cause.  Still, when it comes to the development of national standards, it would be good to be sure the technology has matured sufficiently so that we have confidence that the standards are good standards.

Counterbalancing this argument is the fact that national standards could themselves result in a reduction of errors.  If a set of standards for the user interface is developed, doctors would only have to learn the one interface.  They would not have to use different keystrokes going from home, to office, to hospital, to ER; it wouldn't matter if they were using a desktop, laptop, or palm device.  Well, the Palm Pilot would be different, because usually one uses a stylus, not a keyboard; even so, there could be similarities in the interface. 

The open source movement is already developing interface standards.  For example, the Gnome desktop suite and development platform has a set of guidelines for the development of interfaces.  The problem with such standards is that they make it easier for people to switch from one system to another, which makes it harder for any one company to establish and maintain a dominant market position.  Companies would have to adopt a new business model.  That new model is not really new; Red Hat, Mandriva et. al. have been doing it for years.  They provide the software for a nominal fee, then they focus on service and support.  In the case of EMR, there also is opportunity in the area of content delivery.  Integrating databases such as drug information, consensus treatment guidelines, Medline, electronic textbooks, etc., would make an EMR system much more appealing to a physician.  Having all that available, with one consistent interface, would be really useful. 

For example: a physician who is writing a new prescription might decide to check the latest treatment guidelines for dosage recommendations.  Currently, that involves opening another application, and searching for the pertinent information.  With a fully integrated EMR and content management system, it would be possible to select a menu item that would take the name of the drug -- reading it right from the prescription that is being typed -- feed it into the various databases, and more or less instantly deliver the pertinent information.  That could be a tremendous growth industry, since the information content becomes obsolete quickly. 

Another growth area would be in the area of data abstraction and analysis.  A perfect EMR would be able to pull information from the records, and prepare periodic reports that would help the physician understand his or her own practice patterns.  For example, they could see how often they prescribe controlled substances, compared to national or regional norms.  If a new practice guideline comes out that says, say, that all patients with a new diagnosis of X should be seen for follow up in Y weeks, they could see if that is what they are actually doing.  This would make it much easier to set up internal quality improvement procedures.  Doctors don't have the time to learn SQL and set up the correct reports, but they might select one system over another, if it has this value-added service.  This would be a monthly subscription, rather than a one-time capital expense. 

So the problem with releasing the VA system standards on a national level is twofold: private companies wouldn't like it, because it would interfere with their business model; furthermore, they are using the wrong business model.