Viewpoint: Thought leaders banter on information vs. data

Clinical Innovation + Technology published a viewpoint on Feb. 5 by Ken Elek, MD, about information versus data. The following is a response submitted by Brian Martin, MD, founder and CMIO of Sedeo, a clinical analytics firm delivering strategic intelligence to the pharmaceutical industry, along with Eleck's rebuttal.

Martin's response

Let's move away from forcing physicians to adapt to EMR workflows, and instead require technology to adapt to the physician's practice of medicine.

Dr. Elek describes an EMR/CPOE system that fails to incorporate a patient's penicillin allergy into the system's medical errors detection and prevention algorithms, because the system fails to recognize that "PCN" in the patient's EMR-based allergy list means that the patient has an allergy to penicillins.

His proposed solution is to change the practice of medicine by requiring physicians to record PCN allergies in the format required by the EMR. While this may seem like a trivial change, the snowball effect of numerous EMR-initiated physician workflow changes fosters physician resentment of IT and ultimately undermines physician adoption of EMRs.

I suspect that physicians have far better and more rewarding uses for their personal and professional bandwidth than memorizing IT-initiated changes in their workflow.

As an alternative, why not let the technology adapt to the physician? The "PCN" example is a case in point.

Medical terminology mapping is and has been an active area of medical informatics. For example, the National Library of Medicine's Unified Medical Language System (UMLS) project, which started in the mid 1980s, has the ability to properly interpret "PCN allergy" as "Allergy to penicillin."

When I consulted our UMLS implementation, I noted that “PCN” mapped to over 3,000 phrases. Applying a technique drawn from the artificial intelligence domain, the 3,000 “PCN” terms reduced to the singular Allergy to penicillin term with an associated SNOMED CT code of 91936005.

An EMR system that properly incorporates a semantic network and clinical terminology mapping thesaurus would allow physicians to spend more of their time with patients and less of their time learning IT processes with built-in obsolescence.

Elek responds:

While Dr. Martin makes a very good point on his own, he may have missed the point I was trying to make. So let me see if I can clarify.

The intent is not to make anyone conform to anything other than ensuring the data you want recorded is correct. Going back to the PCN example, there was a nutritional supplement, since taken off the market called PCN-200, which started the issue I was asked to address. It was also not an “IT-initiated change in workflow”. Since PCN-200 is not penicillin, using that abbreviation for the allergy, since it did exist in the database at the time and was chosen by unsuspecting users, would not have indicated a penicillin allergy for that patient.

So, yes, in some respect you have to make sure that the recording truly matches your intent but it’s not necessarily in the way required by the EHR. It’s more a matter of ensuring that whatever shortcut, and PCN is a shortcut, communicates the full meaning. We run into this same problem when MR is used. Is that mitral regurgitation? Mental retardation? Medication Reconciliation? Motor retardation? The system needs to know exactly what I mean and it’s tough to “program in” the logic for every possibility when we each may have our own quirky abbreviations.

So if you can tell me that choosing PCN-200 from a medication allergy database that links to a particular product should be smart enough to know that I really mean penicillin, I’ll agree we’re talking about the same thing.

In the mean time, we all still need to be careful to make sure that the information we input translates to the correct data.