Industries / Health / Healthcare / Pharmacy
Advising patients
Advising patients is the pharmacist's work of explaining a medicine to the person taking it, checking that it is working and, in a growing number of countries, treating minor illness and prescribing.
7
Milestones achieved, from 1990 to 2024.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1990
Counseling becomes a duty
The US Omnibus Budget Reconciliation Act of 1990 required states that take federal Medicaid money to set standards under which pharmacists must offer to discuss each prescription with the Medicaid patient or caregiver, covering points such as the medicine's name, dose, special directions, common severe side effects, storage and what to do after a missed dose. The same year Charles Hepler and Linda Strand argued that dispensing the correct drug was not enough and that pharmacists should take responsibility for the results of drug treatment, a practice they called pharmaceutical care.
Credit
US Congress; C. D. Hepler and L. M. Strand
1997
Pharmacists paid to coach
From March 1997 two self-insured employers in Asheville, North Carolina, paid community pharmacists who had taken a diabetes course to meet regularly with covered employees who had diabetes. Over up to five years blood sugar control (measured as A1c) improved, and mean direct medical costs fell by $1,200 to $1,872 per patient per year against the starting level, by the evaluators' count, while spending on prescriptions rose.
Lessonthe study had no randomized control group, so the savings show a pattern over time, not proof of cause.
Credit
C. W. Cranor, B. A. Bunting and D. B. Christensen, the evaluators, with the Asheville employers and pharmacists
2006
Pharmacists may prescribe
From 1 May 2006, UK pharmacists with extra training and accreditation could prescribe any licensed medicine for any condition within their competence, except controlled drugs, which followed in 2012. By 30 September 2025, 19,615 of the 56,981 pharmacists registered in England (34.4%) were independent prescribers, by General Pharmaceutical Council figures.
Credit
the UK government, through the NHS independent prescribing regulations of 2006
2009
Vaccines at the pharmacy
The Washington State Pharmacists Association began formal vaccine training for pharmacists in 1994, and the American Pharmacists Association (APhA) a national certificate program on 1 November 1996; by 2004 about 15,000 pharmacists and students had been trained. In 2009, once Maine allowed it under emergency rules prompted by the H1N1 flu pandemic, pharmacists could give vaccines in all 50 states.
Credit
Washington State Pharmacists Association; APhA
2016
Check-ins on new medicines
In England's New Medicine Service a pharmacist follows up with a patient one to two weeks, and again two to three weeks later, after a new medicine for a long-term condition is started. In a randomized trial in 46 pharmacies with 504 patients, 71% of those who got the service were taking their medicine as prescribed at 10 weeks, against 61% with normal practice. At 26 weeks the difference (65.6% against 57.1%) was no longer statistically significant, which the authors put down partly to patients dropping out of the study.
Lessonan early gain in medicine-taking was hard to show months later.
Credit
R. A. Elliott and colleagues
2023
Medicine reviews without savings
Medicare drug plans must run medication therapy management (MTM) programs, in which pharmacists or other providers review a patient's medicines. From 2017 to 2021 Medicare's Innovation Center tested an enhanced version, with 6 plan sponsors running 22 plans in 5 regions and given extra payments and freedom to target services. The final evaluation found no significant effect on Medicare spending and no improvement in measures of medication use.
Lessonpaying plans to do more reviews did not by itself change results.
Credit
CMS Innovation Center and its evaluators
2024
Treatment at the counter
Since 31 January 2024 England's Pharmacy First service has let community pharmacists assess people with seven common conditions, such as shingles, and supply NHS-funded prescription-only treatment where clinically needed. They supply it under a Patient Group Direction (PGD), a written instruction that lets named staff supply a medicine without a prescription. People can walk in or be referred electronically by NHS 111 or their general practice.
Credit
NHS England and the Department of Health and Social Care, with Community Pharmacy England
Watch for
AI drafts, pharmacist checks
In a cluster randomized trial in 16 community pharmacies in Jordan with 136 patients, pharmacists used ChatGPT to draft counseling and then checked and edited the drafts. Correctness scored 82.4 against 64.9 with usual counseling, and 30-day adherence was better; 76.5% of the AI drafts needed editing or were rejected, and the pharmacists caught four high-severity near misses.
Achieved whenthe results appear in a peer-reviewed journal and another group repeats the trial elsewhere.
Claimed by
researchers at the University of Petra, June 2026
Pharmacists prescribe at the counter
From 29 October 2026, pharmacist prescribers in approved English pharmacies may prescribe within Pharmacy First in place of PGDs, and may treat five more conditions, among them acute otitis externa (outer ear infection) and episodic migraine, only by prescribing. From 2026 nearly all new UK pharmacists are independent prescribers when they register.
Achieved whenNHS prescribing data show pharmacists' prescriptions under the new pathways in routine numbers and an outside evaluation reports how they are used.
Claimed by
Community Pharmacy England and NHS England, October 2026
Open
01
A shared counseling test
A public set of at least 300 scripted counseling scenarios for new prescriptions, each with a made-up patient, a medicine and a checklist of what the pharmacist should cover, drawn from the points US Medicaid law lists (name, dose and route, special directions, severe side effects, self-monitoring, storage, refills, missed doses), plus an AI-played patient for each script. AI models and pharmacy students are scored against it. Only actors or software play patients.
The briefCredit · unclaimed02
One-page leaflets, tested
One-page patient leaflets for 50 common medicines in the format of the US Food and Drug Administration's (FDA) proposed Patient Medication Information (PMI), each drafted by an AI model from the FDA-approved label, fact-checked line by line by two pharmacists, and user-tested with 20 lay volunteers by the EU's readability method. The volunteers read about medicines they are not taking, and nobody is advised.
The briefCredit · unclaimed
Open 01 · the brief
A shared counseling test
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public set of at least 300 scripted counseling scenarios for new prescriptions, each with a made-up patient, a medicine and a checklist of what the pharmacist should cover, drawn from the points US Medicaid law lists (name, dose and route, special directions, severe side effects, self-monitoring, storage, refills, missed doses), plus an AI-played patient for each script. AI models and pharmacy students are scored against it. Only actors or software play patients.
Why it is still open
The Jordan trial scored counseling from its own blinded transcript analysis, which is not public. A preprint from China that used DeepSeek-R1 as a virtual patient for trainee clinical pharmacists covered 21 trainees at one institution, and found that its automatic scorer marked trainees down for facts the virtual patient had volunteered. Nearby benchmarks test other conversations: DischargeSim (September 2025) scores AI models as discharge educators across six hospital discharge topics, and RxSafeBench (November 2025) asks multiple-choice questions about medication risks planted in 2,443 simulated diagnosis consultations. RxBench (November 2025) scores prescription review. None scores the counseling a pharmacist gives when handing over a new prescription. A search of Hugging Face datasets for "pharmacy counseling" returned nothing.
Built with what exists
general AI models to play patients from the scripts; open-weight models for runs on local machines (see the Note "A trillion open parameters"); speech-to-text such as ElevenLabs Scribe v3 to transcribe spoken role-plays (see the Note "Fewer word errors"); the counseling points in 42 U.S.C. 1396r-8(g)(2)(A)(ii).
First step
Write 30 scenarios for common new medicines, such as an inhaler, a blood thinner and a diabetes tablet, and have two pharmacists score the same 10 role-plays to measure their agreement.
Done looks like
300 scenarios published with a rubric, two pharmacists' scores agreeing at a Cohen's kappa (a measure of agreement beyond chance) of 0.7 or more on 50 transcripts, and published scores for at least three AI models and one class of pharmacy students.
Open 02 · the brief
One-page leaflets, tested
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
One-page patient leaflets for 50 common medicines in the format of the US Food and Drug Administration's (FDA) proposed Patient Medication Information (PMI), each drafted by an AI model from the FDA-approved label, fact-checked line by line by two pharmacists, and user-tested with 20 lay volunteers by the EU's readability method. The volunteers read about medicines they are not taking, and nobody is advised.
Why it is still open
FDA proposed the one-page PMI on 31 May 2023, and the Federal Register lists no final rule since, so no such leaflets exist. A 2026 Brazilian test of medicine instructions written by three AI models found that health professionals judged about 23% of each model's instructions likely to lead to incorrect use; the instructions were rated by 62 professionals, not tested on lay readers.
Built with what exists
FDA labels from DailyMed or openFDA; a general AI model given the label as its only source; the European Commission's 2009 readability guideline, under which a leaflet passes when 16 of 20 participants can find and understand each key point (readability guideline).
First step
Draft leaflets for five medicines in the proposed rule's format and run a 10-person pilot of the readability test.
Done looks like
50 leaflets published with their fact-check logs and test results, and a count of how many pass the 16-of-20 test in the first round and after one revision.