What is medication reconciliation?

Medication reconciliation is the process of comparing the medicines a patient should be using and is actually using with new orders, then resolving discrepancies. It is more than marking a list reviewed. The record should show what evidence was used, what differed, what decision was made, and who confirmed it.

What should a clinic ask during medication reconciliation?

Ask about prescription medicines, over-the-counter products, supplements, dose, route, schedule, last dose when relevant, recent starts or stops, allergies or reactions, and the patient's actual use. Confirm the list against available records and pharmacy or caregiver evidence under the clinic's approved workflow.

Who can resolve a medication discrepancy?

The clinic should define roles by licensure, scope, and policy. Staff may collect and compare information, while a licensed clinician makes clinical decisions about starting, stopping, or changing therapy. The record should preserve the reported difference and the responsible clinician's decision rather than silently replacing the old value.

How often should an outpatient medication list be reconciled?

Set the cadence around the care setting and risk. Reconcile when the regimen may change and at transitions of care, and use a change-detection process for repeated visits where the same course of care continues. High-risk or unclear lists may need more frequent clinician review under local policy.

Can AI complete medication reconciliation automatically?

AI can help structure a patient report or flag that two sources differ. It should not decide that a medicine is active, discontinued, duplicated, or safe. A licensed clinician must review the evidence, resolve clinically meaningful discrepancies, and sign the resulting list under the organization's policy.

Medication Reconciliation Checklist for Clinics

The EHR says lisinopril 20 mg. The patient says, “the small white blood-pressure pill, half a tablet.” A hospital discharge sheet says 10 mg. Copying any one source into the final list would be fast. It would not be reconciliation.

At a glance

Keep the patient report separate from the clinician-approved list.

Record the source behind each discrepancy.

Preserve the decision and its owner, not just the final value.

AI may organize evidence; it must not make a medication decision.

This checklist is a documentation aid. Licensed clinicians and pharmacy professionals should adapt it to the setting, patient population, and local policy.

A copyable clinic checklist

MEDICATION RECONCILIATION

Patient and encounter: ___________________________________
Reconciliation reason: routine visit / transition / change / other
Sources checked: patient / caregiver / EHR / discharge list /
                 pharmacy record / bottles or photos / other

FOR EACH MEDICINE OR PRODUCT
Name: ____________________________________________________
Strength, dose, route, and schedule: ______________________
Patient says they take it this way: _______________________
Last dose or recent change, when relevant: _______________
Prescriber or source: ____________________________________
Discrepancy found: none / dose / schedule / duplicate /
                   omission / stopped / unclear / other
Evidence and follow-up: __________________________________
Clinician decision: ______________________________________
Decision owner and time: __________________________________

FINAL CHECK
Allergies and reactions reviewed: yes / no / not in scope
Unresolved items and owner: _______________________________
Updated list shared with: patient / caregiver / next provider
Reconciled by and time: ___________________________________

The Joint Commission’s current outpatient interpretation describes reconciliation as comparing what the patient should be using and actually uses with new orders, then resolving discrepancies. It also recognizes a change-detection process across repeated visits in the same course of care. Joint Commission reorganised programme requirements for 2026, so confirm the current chapter for the clinic’s accreditation programme instead of relying on a prior-year PDF.

Four layers keep the record honest

Layer	Example	Why it stays separate
Patient report	”I take half at night”	captures actual use without treating it as an order
Outside evidence	discharge list says 10 mg daily	shows where the conflicting value came from
Clinician decision	continue 10 mg daily	identifies the clinical resolution
Shared final list	updated list sent to patient	records communication after the decision

A single mutable list hides the journey. When someone later asks why a dose changed, the final state cannot answer. The four-layer record can.

AHRQ’s Medication Management Strategy includes tools for creating a medication list with patients and families. Use that participation deliberately. “Medication list reviewed” is weak evidence if nobody asked what the patient actually takes.

Collection and decision need different permissions

Medical assistants can often collect bottles, patient statements, or outside lists under clinic policy. Clinical decisions belong to the licensed professional whose scope permits them. Configure the workflow so a collected discrepancy creates an owned task rather than silently changing the active order.

Try these test cases:

The brand and generic names both appear.
The patient splits a tablet, but the list shows a whole tablet.
A short course has ended, yet it remains active.
A hospital list and clinic list disagree after discharge.
The patient cannot name the medicine and has no outside evidence.

The last case needs an unresolved state. “Unknown” with an owner is safer documentation than a tidy guess.

AI belongs at the evidence boundary

An ambient note may capture, “Cardiology stopped the evening dose last week.” That sentence can become a reconciliation prompt. It should not become an automatic discontinuation. The product must keep a human review step and make the source visible.

This is the same boundary described in the prescription-draft safety guide and the medical-record correction guide. A draft is not an order. A correction is not a hidden overwrite.

Patient Square is an AI clinical platform. Practice Copilot brings the whole practice under one AI copilot: an ambient AI Medical Scribe that hands back a structured SOAP note, ICD-10 suggestions, and a prescription draft minutes after the visit, plus a bundled AI EHR, scheduling, and messaging as you move up the plan. Hospitals get Hospital Copilot. Review the current handling terms on the security page and test whether synthetic discrepancies retain their source, clinician decision, and final shared list before using the workflow in care.