I recently wrote about what a treasure-trove the recent leadership conference turned out to be. One of the really significant discussions, to my geeky way of thinking, was Mark Siska's presentation on data stewardship. Data stewardship is like antibiotic stewardship; it is control of a process that yields more consistent and intelligent results, maximizing the appropriate use of a resource.
One of the issues we often see in the field is failure to insist on having database management be performed by someone who actually understands both the clinical and operational impacts of a database entry. If you think about it, the contents of a formulary record describe how the affected software will react to a particular drug product, whether we are talking about medication order entry, an automated dispensing cabinet, or a BCMA system. If the system is a robotic compounding device, this becomes even more important, because the robot may or may not produce a dose for which visual inspection provides any real value. Since most admixtures involve the injection of one clear liquid into another, it is pretty hard to tell whether that syringe that just came out of a robot is really a mixture of two ingredients or not, much less whether or not each ingredient was delivered in the appropriate volume. If the robotic device is also reconstituting the drug product, it gets even harder for the visual inspection to tell whether or not all the steps were done correctly.
It turns out, that the problem becomes even more acute when we are talking about the multiplicity of providers entering data into an electronic medical record. While these systems can (and often do) have input criteria to ensure that the information is at least within acceptable boundaries, it may still be possible for someone to enter a weight in pounds into a field intended to contain kilograms, to use a comment field to actually contain a medication order (or a substantial modification to that order), or to improperly encode transactions that make subsequent queries return apparently accurate, but completely misleading data.
Mark described three pillars that he considers the foundations for good data stewardship.
- Data standards - this is both syntactic (we know what is actually being said) and semantic (we know what the words actually mean). Without these standards, it can be amazing how many interpretations get associated with a word. This is about the words we use, the way they are stored (and retrievable), the rules that govern their input and behavior, and the reliability of the transport between systems.
- Data Quality - this means attention to, and enforcement of the standards as well as regular review of usage to be certain that the data remain retrievable
- Data Architecture and Management - the system of procedures, database architecture, data transport, and other related systems the enforces data integrity. One key feature of well-designed architecture is that each data element have a single source of truth.
This all falls under the more general umbrella of data governance which is a concept we don't often talk about in pharmacy. Indeed, our tendency to be rugged individualists makes us tend to want to resist such governance. However, without that governance, garbage goes in... and garbage comes out...
What are your thoughts?
#standards #Informaticists #dataquality