Medication Reconciliation Assistance: AI Support for One of Healthcare's Highest-Risk Processes

The Joint Commission identifies medication reconciliation errors as a leading root cause of sentinel events. At care transitions — hospital admission, discharge, transfer — medication lists from multiple sources must be compared and discrepancies resolved. This process is time-intensive, prone to information gaps, and high-stakes. AI can accelerate the comparison and flagging step, but the pharmacist or clinician must confirm every resolution.

Reconciliation support prompt

1. 1High-alert medications deserve explicit flagging — errors with these cause the most severe harm
2. 2Dose discrepancies are as dangerous as omissions — a 10mg vs. 100mg difference can be fatal
3. 3Interaction checking requires knowing what both lists contain concurrently
4. 4AI-generated reconciliation output is a first-pass only — pharmacist review is required
5. 5Document the reconciliation source and the clinician who confirmed each decision

The home medication problem

Home medication lists are notoriously inaccurate — patients take OTC drugs, supplements, and medications from multiple providers that never appear in the EHR. AI cannot fix the underlying data quality problem. When using AI for reconciliation, always ask patients directly about supplements, vitamins, OTC medications, and medications from other providers, then add these to the input.

The big idea: AI flags discrepancies fast. The pharmacist resolves them with clinical judgment. Speed matters; accuracy is what the patient needs.