Why EHRs are Data Systems, Not Information Systems

May 20, 2013

Physicians have medical information in their mind, but the EHR cannot access it, so the unhappy task of extracting data from information falls on providers.

If your house is burning, you want the fire department ASAP. If you called the number for the fire department and got street sweeping or business licenses instead, you would be in big trouble. Words matter.

Names matter. They precondition and orient thought. When words like "information" and "data" are confused, thinking is confused. Most "definitions" either equate information and data in some nebulous way or are circular. The whole field of computers is simply too new. The language has yet to catch up. I find it necessary to draw distinction so that I don't confuse myself when thinking about the subject.

Wikipedia says: "The English word [information] was apparently derived from the Latin …verb "informare" (to inform) in the sense of 'to give form to the mind' …Inform [therefore] means to give form, or to form an idea of [something]." Information can be anything that affects the dynamic state of your brain and its thoughts. When you communicate to those around you (spatial) or with the future by writing (temporal), you choose the substance of your utterances in the hope that what you say will accurately convey information that has formed in your mind (your thoughts and mental images) to others.

Data are different. Data have no general, universal meaning. They are always for something and are only meaningful in a specific context. Data are discrete, direct observations and measurements of phenomena, or metrics derived from them, either by transformation (e.g. calculation, coding) or aggregation (e.g. statistics). The choice of measuring technique, as well as the choice of the transform or calculation that will be applied, depend on the context. The weight, not the BMI, of your suitcase matters if you are traveling by air. In fact, the number representing the weight of your suitcase, say 67, only means something in the context that it was weight that was measured. Sixty-seven could be your pulse rate, your end-tidal CO2, or the conversion rate between dollars and Djibouti francs (66.47 today). As you can see, the datum (67) is only meaningful for a specific purpose; it has no meaning in the abstract, stripped of its context.

Through the process of interviewing and examining a patient, the physician informs him/herself about the patient's condition and problems. Although EHRs allow physicians to create narrative notes that may prove informative to other practitioners, now and in the future, the primary function of the typical EHR is to do things: order, bill, criticize by applying rigid rules, etc. The computer applications that are intended to perform these functions require specific data elements as input. Physicians have the information in their minds but the system cannot read minds, nor can it, being mindless, reliably extract data from narrative. Unhappily, the task of extracting data from information and entering it into the computer falls to physicians. The quality of the resulting data depends heavily on how the computer requests it and whether physicians fully understand the use to which the data will be put.

This leads to the following conclusions:

1.) Information is about thought and is for people.

2.) Data is for specific, context-dependent purposes and is for computers.

3.) It is a mistake to make "data collection" the overriding objective of an EHR.

4.) The typical EHR can only increase the physician's workload. There are now two tasks: to record the information that they have about the patient and to extract from that information, data that the computer will accept.

5.) The majority of the value in an EHR is trapped in the narrative. The more accessible it is to physicians, the greater its value.

6.) The majority of the effort and expense associated with EHR arises from data-based applications, the value of which decays like radioactivity, in a geometric fashion, giving it a relatively short half-life.

7.) Calling the typical EHR an information system is a mistake. Like most other computer systems, they are data systems not information systems.

8.) Creating an information technology department to oversee and control EHR is a bigger mistake, much as it is a mistake to have the blind lead the deaf.

9.) The biggest mistake is to see "information system" and be tricked into believing that your data system is more than it really is.