Part II: Medication Reconciliation – A Field of Onions
In Part I ( link) of this blog, I presented some background pertaining to medication reconciliation (MedRec), and a series of assumptions about this topic I gleaned from attending the AMIA conference. As promised, here in Part 2, I’m presenting my thoughts for moving forward.
My conclusions, based on what I learned at AMIA and captured in the list of assumptions, are there exist several concrete and useful ways to approach medication reconciliation that are not in common practice today:
A. Separate out and professionally design the user interface:
Usability: There were multiple examples provided at the conference demonstrating highly usable user interfaces. The opposite was also true; awful displays of what not to do. Hire professionals to do this; analysts, engineers, and end-users are often weak in usability methods, practice and evaluation skills.
Mobility: In addition, it was clear that the time is now to design the user interface to work concurrently on mobile devices, including tablets and Smartphones. That means the ability to drag and drop a medication list, and to be able to simply press on a trashcan or device appropriate method to remove an item from a list. On an iPhone for example, that may mean sliding your finger over a drug name from left to right to bring up a large, red, “delete” button. An iPhone user would expect that to be available and work accordingly.
Data-to-Ink Ratio: Aggressively work to achieve a high data-to-ink ratio. We need to develop ways to remove redundant text so the displays are simpler, clearer, and require less time to absorb wherever possible.
Order and Sequence: The order and grouping of medications on a list is critical. It should be possible to group drugs together in ways that help clinicians do their work.
B. Separate out the medication semantics issues:
Clarity: Brands and generic names, clinical indications, and so forth should be available for display and shown consistently. Drug information can be confusing; design should seek to reduce the chance of errors. One method of doing this is, for example, TALLMAN lettering such as "predniSONE" and "prednisoLONE" (http://en.wikipedia.org/wiki/Tall_Man_lettering ).
Decision Support: There were multiple examples provided that made it clear identification of therapeutic duplication requires clinical decision support systems. For example, if two different drugs both contain morphine or act as a beta blocker, this needs to be apparent to a busy clinician by the design of the user experience.
NLP: Natural Language Processing will clearly be required since medications, even codified ones coming through an interoperable Continuity of Care Document (CCD), will need to be semantically reasoned over to be presented effectively to a clinician. So, for example, if a patient’s visit note references a drug that doesn’t appear on the medication list, the clinician should be given the opportunity to add it or record that it is being discontinued.
C. Separate out ambulatory medication reconciliation from that of inpatient hospitalization:
- Ambulatory MedRec can often be accomplished with a two list model. It tends to be concerned only with ePrescribing.
- Ambulatory MedRec is far less concerned with acute issues like response to newly initiated intravenous medications.
- Medication reconciliation for inpatient hospitalization is a different animal in many, many respects. Probably the biggest difference is the profound and tight integration requirements with hospital processes such as clinical assessment and medication administration.
- Unlike ambulatory medication reconciliation, it is unrealistic to expect being able to decouple the reconciliation process from ordering, ePrescribing, or documentation for inpatients.
- Inpatient MedRec requires a different sort of dynamic conversation with the patient and any home care providers who may be involved, both when being admitted, as well as when being discharged. Poor hand-offs at admission and during discharge are significant problems that the MedRec process needs to address.
In summation, medication reconciliation is like working in a field of onions. There is enough variation that developing a one size fits all model won't work. Peeling back the layers in MedRec is critical.