Friday, May 16, 2008

Cognitive errors in medicine

You don’t have to be a subscriber to the American Journal of Cardiology to read J. Willis Hurst’s editorial on cognitive errors in medicine---the full text is free. I highly recommend it.

The popular approach to limiting cognitive error was recently articulated in Jerome Groopman’s marvelous book How Doctors Think. I call it the negative approach to cognitive error because it emphasizes pitfalls to avoid. Hurst’s essay suggests a more positive approach. It outlines things to do in the systematic collection and processing of clinical data. These approaches are complementary and both ultimately address the same types of error.

Hurst’s approach centers around the medical record as a tool for teaching and learning (thus addressing gaps in the clinician’s knowledge), defining data that need to be collected, analyzing the data and applying the information to the patient’s problems. He advocates the problem oriented medical record (AKA the Weed system) of which he has been a champion for many years.

Most of us would say we use the problem oriented medical record as the ever present SOAP notes and problem lists attest. But a careful reading of Hurst’s essay suggests it’s merely a pretense. The problem oriented record as originally conceived is systematic and rigorous. While it seems cumbersome, Hurst notes that it can be utilized quickly and efficiently once proper habits are developed.

The problem list must designate each item at the level of resolution achieved, from symptoms and laboratory abnormalities at the low end to definitive diagnoses at the highest level. Moreover, diagnoses must reflect the most up to date disease classification and terminology.

That raises an important question. Is it possible to generate and maintain a true problem oriented medical record in today’s environment where the record is hijacked by coders, core measure police and others with competing agendas? Just one example of this problem is the basing of diagnostic terminology on the decades obsolete (created in 1977 and replaced in 1992) ICD-9 classification. In paper based records if you use precise and up to date terminology your patient’s chart is likely to “bounce back” from the coding department and ultimately find itself on the delinquent list. In the electronic world where all diagnoses are inseparably linked to ICD-9 codes the best terms may not be available.

An effective medical record is more than an accurate problem list. Hurst writes (italics mine):

The medical record should reveal the thoughts and actions of the physician in charge. The challenge for the physician is to make the record simultaneously brief and complete. The record should be easily understood by another physician. Each page of the record should be uncluttered, and important information should be displayed prominently so that it can be retrieved easily.

I wonder what Hurst thinks of today’s template generated charting.

There’s much more.

No comments: