You know how the medication review goes, because you do it a dozen times a day. The medical assistant walks the list during rooming. You open the chart, glance at what they flagged, and click through. Somewhere in the next fifteen minutes you also need to address the reason the patient actually came in.
What the published research adds is a measurement of how little room that step actually gets — and how much the outcome depends on conditions nobody in the room controls.
Two minutes, three medications out of seven
A mixed-methods study that video-recorded ambulatory visits in five academic clinics found the medication review had a median duration of 2.1 minutes — roughly 10% of visit time. In that window, clinicians addressed a median of 3 of the patient’s 7 relevant medications. Nearly half of that time involved multitasking EHR use; focused clinician–patient talk occupied about a quarter of it (Matta et al., JMIR Med Inform 2018).
An independent ethnographic study across 15 primary care clinics observed essentially the same number: an average medication history of two minutes (Gionfriddo et al., PLOS One 2021).
Two minutes is not a failure of diligence. It is what remains after everything else the visit is required to contain.
How it actually moves through the EHR
The PLOS One study is the most detailed published account of the mechanics, combining 170 observed encounters, 48 staff interviews, and 616 staff survey responses inside one integrated system. Its central finding was not error — it was inconsistency:
- Adherence to best-possible-medication-history behaviors averaged 53%, ranging from 2% to 100% across observations.
- Only 38% of staff believed a standardized process was in place — and prescribers were far less likely to think so than non-prescribers (28% vs. 47%).
- 70% of staff reported never receiving formal medication reconciliation training from their organization.
- Over-the-counter medications were asked about in just 36% of observed visits.
- 24% of staff reported difficulty entering patient-reported medications into the EHR at least half the time — and prescribers reported this roughly twice as often as non-prescribers (35% vs. 17%).
The handoff itself has no fixed shape. Nurses in that study communicated what they’d found by speaking to the physician directly, leaving a note on the door, writing a nursing note, or using a comment field on the list — depending on the clinic and the individual provider’s preference. One physician leader described the reconciliation function in their own EHR candidly: they avoided clicking it because they weren’t sure what it would open, what they’d be reconciling, or what a care plan might silently lose.
And a category of change never reaches the record at all. A dose adjusted by phone or portal message — cut the Lasix in half, you’re feeling dizzy — is frequently never entered. On paper the patient is still taking 40 mg.
Why the burden lands differently on prescribers
Time pressure was near-universal in interviews — 96% of interviewed staff raised it. But it was not evenly felt. Among survey respondents, 44% said time affected their ability to reconcile at least half the time, and prescribers endorsed this barrier more than twice as often as non-prescribers (67% vs. 26%).
Comfort splits along the same seam. Prescribers were less comfortable with their role in reconciliation overall than non-prescribers (76% vs. 91%), and roughly a third of staff cited discomfort changing a medication ordered by someone else or outside their expertise — a barrier prescribers endorsed more often than non-prescribers (41% vs. 26%). A pharmacist in the same study described the standoff plainly: primary care doesn’t want to stop a specialist’s medication, the specialist doesn’t want to stop primary care’s, and the item stays on the list even when the patient says they’ve stopped taking it.
What the controlled trials found
The first randomized trial of ambulatory medication reconciliation tested two intuitive fixes across 20 primary care physician/MA pairs, randomizing 440 patients (367 analyzed): give the patient a printed copy of their list at check-in, and open the review with an open-ended question rather than reading the list aloud (Wolff et al., J Am Board Fam Med 2014).
- Printed list alone: 66.7% agreement, versus 67.4% with no intervention — no change.
- Open-ended question alone: 58.1% — agreement went down.
- Both together: 75.6%, with a statistically significant interaction (P = .04; P = .01 adjusted).
- Agreement across the 20 MD/MA pairs ranged from 33% to 91%.
Two details in that trial deserve attention. The in-visit review lasted 1.3 to 1.5 minutes regardless of arm. The pharmacist-led telephone reconciliation used to establish ground truth averaged 10.3 minutes (range 5–35). The authors were direct that a review of that depth is not feasible in routine primary care.
Neither intervention worked alone. Handing a patient a list without inviting them to speak changes nothing; inviting them to speak without giving them something to react to made things worse.
A more recent cluster-randomized trial of a pre-visit portal tool across six primary care clinics found no improvement in list accuracy over usual care — 7.5% of patients in the intervention arm versus 7.6% in usual care had zero discrepancies (OR 1.01, 95% CI 0.38–2.72, P = .98). The trial also noted that only about a third of patients have portal accessibility. The most common discrepancy in both arms was a patient reporting they were no longer taking a listed medication (Gionfriddo et al., JMIR Form Res 2022).
Where the lists are hardest to hold
Discrepancy risk rises with the number of medications, the number of prescribers, patient age, specialist involvement, and the patient’s own unfamiliarity with the regimen (PMC8453145).
- Nephrology. In one outpatient clinic, 225 ambulatory reconciliations across 115 advanced-CKD patients identified 180 discrepancies — most often wrong drug, then wrong dose, wrong frequency, and omissions. Nearly a quarter of patients faced potential moderate discomfort or clinical deterioration as a result (PubMed 29218971).
- Cardiology and heart failure. Cardiovascular medications appear repeatedly among those most involved in unintentional discrepancies, in a population where polypharmacy is standard care.
- Internal and family medicine carry the same structural load wherever long lists meet short visits.
The part that changed
None of the above is new to anyone who practices. What is new is that the drift is no longer private. Federal transparency rules gave patients direct access to their records, and when they read them, medications are among the categories they most often flag as serious.
That turns a documentation problem into a trust problem — and the evidence above suggests trust problems in this domain are best addressed by asking first, under conditions where the patient can actually answer: at home, unhurried, with the bottles in front of them.
ConfirmMyMeds is a working application. It reads your medication list through your EHR’s standardized, federally required data-exchange interfaces, asks the patient to confirm it before the appointment outside the portal, and returns what they confirmed and what they flagged — timestamped and attributed. It never writes to your record. We have no outcome data, and we’re not going to tell you it saves time or reduces errors, because we haven’t measured that.
What we’re looking for is a small number of clinics willing to use it and tell us where it fails.