William A. Hyman
Professor Emeritus, Biomedical Engineering
Texas A&M University, w-hyman@tamu.edu
Read other articles by this author
“Best Practice” (BP) is a popular descriptor for a variety of processes and activities. A recent search of HITECH Answers produced 13 items with Best Practice in the title. These included how to move a system from one point to the next as in adopting a new EHR, or how to better use the system you have as in maintaining cybersecurity. Occasionally best practice is also used to describe the development of a new product or system (see below). But what does best practice actually mean?
To be actually meaningful the term “best” should have an inherent component of comparison between one thing and one or more other things. That is a “best practice” should mean that the practice is superior to other practices in the same domain. This superiority should be demonstrable with respect to specific criteria, eg faster, safer, cheaper, etc. In this regard a particular practice that is being described as having worked out well for the users may be a reasonably good practice (for them), but it hardly meets the criterion of being best.
One available definition of “best practices” is “Methods and techniques that have consistently shown results superior than those achieved with other means, and which are used as benchmarks to strive for.” Note the inclusion here of having been shown to be superior, ie there should be some actual evidence. Another definition is “A best practice is a method or technique that has been generally accepted as superior to any alternatives because it produces results that are superior to those achieved by other means or because it has become a standard way of doing things.” This also has the concepts of superior to any alternatives based on comparative results. But yes, it also has the unfortunate phrase because it has become a standard way of doing things. This apparently means that if we all have agreed on how to do something, whether or not there is data to support the method compared to other ways of doing that thing, or a method has been dictated to us by some authority, then we shall undertake the self-delusion of calling it the best.
The eHealth Initiative occasionally posts a variety of things it calls “Best Practice”. Recently it covered, based on a meeting abstract, the development of a CDS system for predicting atrial fibrillation using “artificial intelligence”. The title of the abstract itself used the term “machine learning”, although AI was also mentioned. I note this because machine learning might be characterized as artificial pattern recognition which isn’t quite the same thing as intelligence. In the reported study a large dataset was used to “teach” the system to identify associations between various input parameters and the occurrence of a-fib. The system was then applied to a smaller data set of known results for which it had a level of predictive power that might be useful. The abstract then notes that future study will include a population health management intervention for actual patients. Let’s grant that this study was well executed and might have the potential to actually be useful. We might even applaud the effort based just on the abstract. But does this qualify for the adjective “best”? What was the work compared to?
In the absence of actually having some evidence that a practice is best we probably need different terminology so as to not over-hype general value and works in progress. I suggest “pretty good practice” (PGP), or “seemingly good practice” (SGP). Then we can reserve “best” for things that have actually been proven to deserve that attribution.