Exploring Whether EHRs Have a Scientific Basis

January 6, 2014

Lacking validated science, EHR builders, like the builders of medieval cathedrals, use proprietary heuristics to guide them, and achieve inconsistent results.

In "The Pillars of the Earth," Ken Follett describes the process of designing and building a medieval cathedral. Cathedrals were some of the most complex and expensive systems of that age. There was no scientific basis for favoring one design over another. The job was in the hands of a master builder who combined his experience, artistic sense, and craftsmanship with what he learned as an apprentice into a set of heuristics - rules of thumb - that he used and some of which he taught to his workers. It was a rare event when a bishop or nobleman decided, or could be convinced, to undertake a new cathedral. There was intense competition for the post of master builder; it meant, in all likelihood, a lifetime sinecure. Master builders guarded their most important heuristics jealously. They were valuable and proprietary trade secrets.

Master builders had a working knowledge of geometry and perhaps a bit of trig. The lack of a formal understanding of structural engineering, moduli of elasticity, tensile strength, and the compressive strength of stone became painfully apparent every time a structure failed (collapsed). The only, and natural, response was to fall back on designs that were less daring and innovative. No two cathedrals were alike. The later phases of construction of a particular cathedral used techniques that represented refinements based on prior negative results.

If you were to substitute the word EHR for cathedral in the above paragraphs, you would have an accurate description of how most EHRs are, and have been, built. Lacking widely available (public) and thoroughly vetted scientifically-derived principles upon which to base a design, master EHR builders use a treasured set of proprietary heuristics to guide them. Lacking scientific knowledge about EHR, physicians are deprived of the ability to apply their evidence-based reasoning skills to EHR.

One way to establish a scientific basis would be to subject an EHR to randomized, controlled trials but this is not really possible. 

Developers won't reveal their proprietary techniques. There is no way to conduct controlled trials because after configuration and customization, every installation is unique. It is not possible to do EHR research the way we study physiology and pharmacology.

Another approach would be to conduct what engineers call Failure Modes and Effects Analysis (FMEA) of EHRs after a flaw or fault has been encountered. An example of a FMEA can be found in the Rogers Commission Report on the Space Shuttle Challenger disaster

Hospital QI efforts include what they cause "root cause analysis" but are usually forced to treat the EHR as a "black box" since vendors are loath to expose the internal workings of their system.

The remaining avenue for research on EHR is to first theorize and build prototypes with different characteristics, explore their strengths and weaknesses, and formulate hypotheses and potential theories. Second, to test them against one's own experience and what one can observe from afar. Just as it is possible to diagnose advanced syphilis by observing only the characteristic shuffling gait, one with sufficient experience can infer the inner workings of an EHR by merely looking at the screens, listening to user's complaints, and noting the nature of the errors and flaws that are manifest.

While awaiting the science, people should use today's EHRs intelligently. Here are some heuristics:

1. Patient care comes first. If the computer is interfering with patient care, turn your attention to the patient. If you ignore the patients and they choose to go elsewhere, you won't get paid.

2. Pay as little as possible for EHR. It's only a tool, not the end itself.

3. The benefit of EHR depends on how much effort you invest on understanding and configuring it, not on how much it cost.

4. EHR is an expense, not an investment. Don't mortgage the future.

5. You should be in charge, not the vendor. Choose your own goals and objectives.

6. Never proceed directly to full scale implementation or trust the vendor's "best practices." Do a pilot. Understand how your decisions about roles, rules, order-sets, charge-masters, etc., will affect the work flow. Adjust the configuration until the work flow makes sense and doesn't interfere with patient care. Remember that an EHR that compromises your productivity may result in a financial penalty greater than that which the government will impose for failure to adopt EHR.

7. EHR, like all technology, is transitory; plan on changing every 5 years to 15 years. Buy only what you need. Don't over-commit.

8. Before choosing your next EHR, identify an exit strategy from it (this is easier than it sounds). Then identify an exit strategy from whatever you are doing now. Only choose a product that can act as a bridge between now and the future. Your survival is more important than the vendor's.

9. Do not allow the information in your EHR to become the property, or the hostage, of any vendor.