Bureaucrats Will Never Achieve EHR Interoperability

November 9, 2015

The kind of EHR interoperability envisioned by vendors and regulators will never be achieved and is not practical for healthcare.

Interoperability (interop) means different things to different people depending on their technical knowledge, background, and objectives. The most common definitions relate to the (relatively) instantaneous transfers between geographically separated locations. I’ll call this spatial interop. Depending on your circumstances, spatial interop may be critical or irrelevant. It may, for example, be more important to ensure continued access to information you have gathered as time passes. I’ll call this temporal interop. To meet my needs as a physician, I find spatial interop to be trivial. The real challenges are posed by the continuous and rapid changes to the underlying technology that have the effect of making today’s information inaccessible in the future.

First, let’s consider spatial interop. I want access to records of care the patient received in the past with minimum time and effort. In the old days (before EHR), I knew how to do this. It involved using the phone, the fax machine, and sometimes, the mail. I have even gotten the access I needed by driving to another facility and reviewing the records.

Then there is temporal Interop. Apparently many physicians would like to switch products. Will they be able to migrate the contents of an existing EHR to a new one? This should be the first question you ask an EHR vendor but it is rarely asked until it’s too late. In my experience, physicians that have expected to preserve their records during a migration have been seriously disappointed. Some have ended up losing everything.  If you don’t think this a problem, try to open a Word or PowerPoint document from the early 90s in the latest Microsoft apps. Neither Microsoft, Apple nor your EHR vendor have any real interest in you and your data. Their interest is to lock you into their ecosystem. Just look at the ELUAs (End User License Agreement), e.g. “ITUNES TERMS AND CONDITIONS: The Graphic Novel” instead [http://itunestandc.tumblr.com/].

Practitioners trust their vendors to provide both types of interop, but that trust is misplaced. Vendors have even less incentive to facilitate an exodus from their product than ensuring it interoperates spatially (with other vendor’s products). Not only don't practitioners and patient care really matter to vendors, they don’t really matter to federal regulators either. The regulatory definition of interop is that it takes place between machines, not between people.

When interop works, look below the surface you will discover a simple standard. For  example:

• All U.S. drivers drive on the right.

• Every car that uses unleaded is built with a filler opening that matches the diameter of the pump nozzle. Leaded gas and diesel nozzles don’t fit.

• In the U.S., 15 amp 110 volt AC outlets all use the same pattern and appliances that draw 15 amps or less use a matching plug. There are other patterns for other voltages and higher currents.

• Virtually all email messages conform to RFC 2822, the Internet Message Format.

• PCs have expanded because IBM made a simple electrical “bus,” an interconnect scheme that allows add-on cards to be plugged in, part of its original design. The bus has a simple specification. When IBM tried a second time with the Micro-Channel bus, it flopped; the specification was too complicated and expensive to implement.

What do these successful standards have in common that correlates with their success? They define a simple structure (like drawing a line down the middle of the road or the diameter of a fuel nozzle). The Internet Message Format specifies how to delineate those portions of the mail message that carry the sender, recipient, subject, etc. They don’t specify who the recipient is, only how to flag it so that the email server can find the address. These standards work because they are “content agnostic.” They can applied widely and readily accommodate new content and, new content doesn’t invalidate existing content so no “conversion” is required every time new is introduced.

Simple structure-based standards can be defined, agreed, put to use, and then “put to bed” so to speak. When this has been accomplished, the foundation for interop is laid. You can bring a British car into the US and it’s your problem how to figure out how to drive it on the right.

Healthcare standards concentrate on defining specific content, creating more problems than it solves:

• Not everyone needs or cares about the same elements.

• Even if they do, their system may not contain them.

• Even if they do, the meaning is not guaranteed to be consistent.

• The process of adding content items to a standard is so laborious that all of the “essential” items have not yet (after 25 years) been addressed.

• Had they been, many would be obsolete and irrelevant by now.

• Tomorrow will bring with it new ideas, changed agendas, and the need for elements that were never dreamed of before.

The job specifying the content can NEVER be finished. The kind of interop envisioned by bureaucrats, finance types, and regulators that depends on specific data element will NEVER be achieved. Their fixation on data elements is interfering with achieving the kind of practical interop that physicians need - the ability to access the patient’s medical records. I’ve said this before and I’ll keep saying it until people get the message.