Skip to main content

Guide

GPhC expectations for online prescribing.

What the safeguards mean in software terms

The regulator's guidance for pharmacies providing services at a distance is not a technical specification, and that is exactly why it is hard to implement. It describes outcomes: the patient is identified, the supply is clinically appropriate, the prescriber can decline, and the whole thing is evidenced. Below is each expectation with the thing a platform actually has to store to show it happened. This is our reading, not a substitute for the guidance itself or for your own regulatory advice.

Know who the patient is

Identity has to be checked, not assumed from an email address, and the answer has to be part of the record. In practice that means an identity check per patient, configurable by product, with the result, the provider and the timestamp stored against the order. Just as important is the failure path: an inconclusive or failed check should hold the order and put it in front of a clinician with the reason, rather than silently blocking or silently passing.

Ask enough, and act on the answers

A questionnaire that cannot stop an order is decoration. The assessment needs branching, so that a concerning answer changes what happens next, and hard stops for the answers that should end the consultation. Store the questions as they were asked at the time, not just the answers, because flows change and an order has to be reviewable against the version the patient saw.

Higher risk categories get more scrutiny

Weight management, and other categories where the incentive to misreport is high, have been a repeated focus. Expect to evidence objective checks rather than self-reported numbers alone: photographic or measured evidence where appropriate, cross-checks against previous orders, and a documented reason when a prescriber proceeds despite a flag. Our note on the regulator's themed review covers this in more detail: GPhC calls on pharmacies to strengthen safeguards in weight management services.

A named prescriber decides, every time

Automation can triage, flag and summarise. It cannot prescribe. Every supply should carry the identity of the person who approved it, the time they did it, what they saw, and their rationale where they overrode a flag. Signing should be tied to the exact document content, so a later change is detectable.

Declining has to be a real option

A service where the commercial pressure makes refusal awkward will eventually refuse too little. Rejections should be as easy to record as approvals, with a reason, a message to the patient, and no financial trap that makes a prescriber hesitate. Refunds and cancelled orders need to flow back through the same record.

Duplicate and multi-account behaviour

Distance selling makes it trivial to try again with a different email. Reasonable safeguards include flagging reused devices, addresses and payment details, spotting repeat submissions for the same condition inside a short window, and showing the prescriber the prior submissions rather than making them search.

Involve the wider care team

Where it is appropriate, the patient's GP should be told, and the record should show what was sent and whether it arrived. Where a patient declines, that decision belongs in the record too, along with what the prescriber did about it.

Patients with no regular GP, or who decline sharing

The updated guidance is explicit that this cannot be handled with a blanket yes or a blanket no. Where a patient has no regular GP, or will not consent to their information being shared, the prescriber is expected to make an individual, risk-based judgement: weighing the risk of treating without a full picture against the risk that the patient is already being prescribed something elsewhere. If the balance does not fall in favour of treating, the patient should be directed elsewhere. In software terms that means the consent answer is a recorded field, not a checkbox that disappears, the prescriber sees it at the point of decision, and the rationale for proceeding or declining is captured against the order.

Choosing the medicine before the consultation

A patient can say what they would prefer before they are assessed. What cannot happen is that preference deciding the outcome. The regulator's position is that the treatment decision is made jointly by prescriber and patient during the consultation, so a shop that pre-selects a product has to be built so the assessment can change or refuse it. Practically: the product a patient picked is treated as an indication rather than an instruction, the prescriber can switch or decline it without leaving the flow, and any change is reflected in the order and the refund without a manual workaround.

Evidence it, for years

Everything above is only worth what you can produce on request. That means an append-only trail of clinically significant actions, written by the system rather than the application, tied to a named user, and retained for years. Role-based access, session timeouts and multi-factor authentication for clinical staff sit alongside it.

How this maps to Sync-RX

Each of these is a feature rather than a policy document: configurable assessments with hard stops, identity checks with a clinician-visible failure path, risk flags on submissions, prescriber review and signing with rationale capture, GP notification, and a seven year audit trail. The FAQ covers the specifics, and we are happy to walk through it against your own clinical model.

Sources

Repeats and cancellation

Ongoing supply is where most services come unstuck. See repeat supply, reassessment and cancellation for the reassessment gate, the record each cycle needs, and how to handle failed payments and cancellations.

Back to guides

Book a Demo