Telemedicine: The Requirements That Are Not About Video

Telemedicine products are frequently scoped around the video call, which is the part that is already solved. Managed services handle the connection, the encryption, the network traversal and the mobile clients, and a working consultation is achievable in days.

What makes these products difficult is everything the call sits inside: who is permitted to treat whom, what has to be recorded, what can be prescribed, and what happens when the patient should not be on a video call at all.

Licensing follows the patient, not the clinician

A clinician is generally licensed to practise in a specific jurisdiction, and the relevant jurisdiction is usually where the patient is located at the time of consultation, not where the clinician sits.

For a platform, that turns into routing logic. You need to know where the patient currently is, know where each clinician is licensed, and prevent bookings that would place a clinician outside their permitted scope. Patients travel, which means a booking valid when made can be invalid at the appointment.

Getting this wrong exposes the clinician personally, which makes it one of the requirements they care most about. It is also frequently discovered late, because it is invisible in a single-region pilot.

Documentation is the clinical record

A consultation produces a medical record, with the retention, amendment and access rules that come with it. That is not application logging, and it cannot be a free-text note in a database with no history.

Practically: notes need to be attributable and timestamped, amendments need to preserve the original rather than overwrite it, and access needs an audit trail. If the record is to be shared with the patient's regular provider — which is usually the clinically correct outcome — it needs to be exportable in a form their system accepts, which returns you to the integration problem that dominates healthcare software.

Prescribing is heavily constrained and varies

Remote prescribing rules differ substantially by jurisdiction and by drug class. Some medications cannot be prescribed without an in-person examination. Controlled substances carry additional identity verification and reporting requirements. Some jurisdictions require a pre-existing clinical relationship.

Building prescribing means encoding rules that change, so they belong in configuration with an owner rather than in code. It also means integrating with pharmacy systems and, in many places, a prescription monitoring programme with its own interface and its own reporting obligations.

Products that treat prescribing as a later phase often find it reshapes the earlier phases, because identity verification requirements flow back into onboarding.

Triage is the safety-critical part

The most important thing a telemedicine product does is recognise when a patient should not be using it. Chest pain, stroke symptoms, severe breathing difficulty and several other presentations need emergency care, not a video appointment in forty minutes.

That makes intake a clinical safety feature rather than a form. It needs escalation paths that are obvious and fast, and it needs to fail safe — when the answers are ambiguous, escalate rather than proceed.

It also needs to work for someone who is unwell, anxious and possibly using a borrowed device. Intake designed for the median user will fail the users it matters most for, which is an argument for simplicity over comprehensive data capture at the front door.

Connectivity is a clinical constraint

Patients in rural areas, in poor housing, or on limited mobile data are disproportionately likely to need remote care and least likely to have a good connection. A product that only works on good bandwidth serves the people who need it least.

The design responses are known: audio-only fallback that is a first-class path rather than a degraded error state, asynchronous consultation for cases that do not need real time, and clear guidance when quality drops. Audio-only in particular is worth treating as a legitimate consultation mode, since it works nearly everywhere and is clinically sufficient for a large share of cases.

Identity verification cuts both ways

Patients need confidence that the clinician is who they claim and is qualified. Clinicians need confidence about who they are treating, both for record accuracy and for prescribing.

Credential verification against professional registers, ideally re-checked periodically rather than once at onboarding, addresses the first. The second is a balance: heavy verification suppresses access for exactly the populations that benefit most from remote care, while none at all is unacceptable for prescribing. Most workable products verify proportionally, with more required as the clinical stakes rise.

Scheduling is harder than a calendar

Telemedicine scheduling has to reconcile clinician availability, licensing by patient location, appointment type, duration that varies by consultation reason, and buffer time for notes — while handling cancellations and no-shows, which run higher than in-person because the friction of not attending is lower.

Overbooking policies used in physical clinics do not transfer cleanly, because a video appointment cannot absorb a wait in the same way. What works better is short-notice fill from a waiting list, which requires a notification path and a patient willing to attend within the hour.

Timezone handling deserves specific attention. A patient booking while travelling, a clinician in another region, and daylight saving transitions between booking and appointment together produce a class of bugs that are embarrassing and entirely avoidable if times are stored with offsets from the start.

Payment and coverage vary by jurisdiction

Whether a remote consultation is reimbursed, at what rate, and under which conditions differs by payer and jurisdiction, and has changed repeatedly in recent years. Some require specific documentation of the consultation modality; some require the patient to be at a defined originating site.

That makes coverage rules configuration with an owner rather than logic in code, and it makes the claim record dependent on details — modality, duration, patient location — that must be captured during the consultation rather than reconstructed afterwards.

Where to start

Establish the jurisdictions and their prescribing rules before designing, because they determine the routing and the identity requirements. Build the clinical record properly, since retrofitting immutability and audit is painful. Treat triage as safety-critical and design it for someone who feels unwell. Make audio-only a real path.

The video call, which is where the product demo focuses, will be the least of the work.