1032 — Clinicians’ strategies for organizing and reconciling medications from multiple VA and non-VA sources: Card-sorting study
Lead/Presenter: Himalaya Patel,
COIN - Indianapolis
All Authors: Patel H (Center for Health Information and Communication, Richard L. Roudebush VA Medical Center), Savoy A (Center for Health Information and Communication, Richard L. Roudebush VA Medical Center; Indiana University-Purdue University Indianapolis; Regenstrief Institute, Inc.) Nguyen KA (University of Florida) Saleem JJ (University of Louisville) Sanchez SL (Center for Health Information and Communication, Richard L. Roudebush VA Medical Center) Adjei EA (Center for Health Information and Communication, Richard L. Roudebush VA Medical Center) Traylor M (Center for Health Information and Communication, Richard L. Roudebush VA Medical Center) Militello LG (Applied Decision Science, LLC) Diiulio J (Applied Decision Science, LLC) Weiner M (Center for Health Information and Communication, Richard L. Roudebush VA Medical Center; Indiana University; Regenstrief Institute, Inc.) Fan VS (VA Puget Sound Health Care System; University of Washington) Gibson B (Center for Informatics, Decision-Enhancement and Analytic Sciences, George E. Wahlen VA Medical Center; University of Utah) Mixon AS (Geriatric Research, Education, and Clinical Center, VA Tennessee Valley Healthcare System; Vanderbilt University Medical Center) Russ-Jara AL (Center for Health Information and Communication, Richard L. Roudebush VA Medical Center; Purdue University; Regenstrief Institute, Inc.)
It is unclear how to design health information exchange (HIE) technologies to best support cliniciansâ€™ information needs and priorities. We aimed to describe the sorting criteria used by a sample of clinicians during medication reconciliation with multiple electronic data sources.
We conducted an online card-sorting study with clinicians from four VA medical centers in geographically dispersed VISNs. We developed a sequence of two clinical scenarios for one hypothetical patient. Scenario A had 12 medication cards, each including up to 6 supporting details (e.g., prescriber location). Scenario B had 13 medication cards, each including up to 9 details. To simulate potential effects of HIE data, we created safety probes by inserting 14 discrepancies in medication information among the cards. Participants were scheduled individually. Scenarios were presented in the same order, together lasting approximately 30 minutes. In both scenarios, using the provided cards, the participant reconciled the patientâ€™s medications as if preparing for a clinical encounter with the patient. The participant grouped cards and named their card groups, using any approach supporting their reconciliation, while thinking aloud. After data collection, two analysts inductively labeled card groups and identified cliniciansâ€™ primary sorting criteria.
Altogether, 33 clinicians participated (24 prescribers, 9 pharmacists), representing both outpatient and inpatient practice settings. Across scenarios, we identified six distinct criteria clinicians used to sort medication data, along with representative group names. * Prescriber Site: other VA, local VA, and non-VA (36% of scenarios) * Drug Indication: e.g., diabetes and cardiovascular diseases (35% of scenarios) * Unverified: over-the-counter and duplicate (14% of scenarios) * Prescription Currency: current versus outdated (8% of scenarios) * Alphabetical: sorted by drug name (6% of scenarios) * Prescriber Specialty: aligned with, or distinct from, participantâ€™s specialty (2% of scenarios) Further analysis by clinician type (prescriber versus pharmacist) showed that drug indication was the most frequently used criterion among prescribers (40% of prescribersâ€™ scenarios), whereas prescriber location was most frequently used among pharmacists (56% of pharmacistsâ€™ scenarios). The unverified criterion, used only by prescribers, encompassed reasons to discuss medications with the patient. Twenty-one participants (64%) used the same primary sorting criterion in both scenarios. From scenario A to B, cliniciansâ€™ sorting by indication increased from 10 uses to 13 (+30%); sorting by unverified decreased from 6 uses to 3 (-50%).
Our sampleâ€™s clinicians relied on a small set of sorting criteria during medication reconciliation, which remained mostly stable even after adding information. Based on our studyâ€™s findings, electronic tools for HIE-aided medication reconciliation should facilitate cliniciansâ€™ organizing of medications using these criteria, at a minimum: prescriber site, drug indications, and prescriptions needing further review. Of these three, only prescriber site is a sortable attribute in JLV.
Our results identify limitations of current approaches to HIE-aided medication reconciliation, as well as specific data-sorting needs to address. Sorting criteria identified in this study may also provide insight into differences in cliniciansâ€™ motivations and cognitive processes for conducting medication reconciliation. For Veteran patients, potential long-term benefits include reduced risk of adverse drug events, especially at non-VA care transitions.