Doctor Slot Configuration and Session Templates Explained

A session template describes a repeating block of a doctor's clinical time: which day, which hours, which location, which consult types are permitted, and how the block is divided. Configuring at the template level rather than the individual-slot level is what makes a schedule maintainable. When a consultant's Tuesday morning changes from three hours to two, you edit one template rather than reissuing hundreds of slots.

Templates should be versioned with an effective date rather than edited in place. If you change the Tuesday template today, the appointments already booked under the old shape need a defined fate — retained, moved, or cancelled with contact — and the historical record of what the schedule looked like last quarter needs to survive. Editing in place destroys that history and makes capacity analysis impossible after the fact.

A good template also encodes what may be booked into it, not just when. A consultant's Thursday session might permit new cases and follow-ups but exclude procedures; the pre-operative assessment session might permit only referred surgical cases. Encoding permitted consult types in the template stops the front desk from having to remember rules, and stops online booking from filling a specialised session with general cases.

A new case, a routine follow-up, a report review, and a procedure consultation are not the same length of work, and giving them the same slot is the root cause of most clinic overruns. A new case in general medicine involves history, examination, and a plan explained to someone who has not heard it before; a stable follow-up may be four minutes. Averaging them into a single ten-minute slot means every new case eats a follow-up's time and every short follow-up leaves the doctor idle without letting anyone else in.

Setting durations by type is straightforward to configure and immediately visible in the result. The harder part is getting honest numbers, because consultants under-report their own consultation time and coordinators over-report to protect the clinic. Use the actual timestamps your system already holds — consultation start to consultation end, by doctor and by consult type — and look at the distribution rather than the mean.

Where a department resists differentiated slots, a workable compromise is to keep one slot length but cap the number of new cases per session. It is less precise than type-based durations but it constrains the same variable, and it is easier to agree to because it does not require anyone to concede that their new cases take longer than their follow-ups.

The standard approach — measure the average consultation, divide the session by it, publish that many slots — fails because consultation time is not symmetrically distributed. It has a floor of a few minutes and a long tail: the breaking-bad-news conversation, the patient with six comorbidities, the case that turns out to need admission. A distribution with a long right tail has a mean that sits well below the cases that actually cause damage, so a schedule built on the mean is guaranteed to run late.

It also ignores that delay accumulates. If each slot is set at the average, roughly half of consultations overrun, and the overrun is not recovered by the short ones — because a short consultation only recovers time if the next patient is present and ready, which mid-session they often are not. Lateness ratchets in one direction. By the last hour, a clinic that was three minutes behind at the start can be an hour behind, and no individual consultation was unreasonable.

The better input is a higher percentile of the observed distribution for routine slots, with explicitly placed recovery time for the tail. As an illustrative example, if a consultant's follow-up consultations cluster between five and nine minutes with occasional twenty-minute cases, a nine-minute slot plus one catch-up slot per hour will hold far better than a seven-minute slot with no recovery, even though the seven-minute version promises more capacity on paper. Promised capacity that is never delivered is not capacity.

A buffer is unbooked time deliberately placed in a session so that accumulated delay has somewhere to go. There are two common patterns: a distributed buffer, where one slot per hour is left unbookable, and a terminal buffer, where the last twenty minutes of the session are held. Distributed buffers work better for clinics where delay accumulates steadily; terminal buffers only help if the clinic is not already so far behind that the buffer is consumed before it arrives.

Catch-up slots are the same idea with a different release rule: they are held unbookable until a defined point — often the morning of the session — then released to the front desk for urgent add-ons if the clinic is running on time, or absorbed as recovery time if it is not. This makes the buffer earn its keep rather than sitting idle in a clinic that happened to run smoothly.

The objection is always that buffers waste capacity. They do not, if the alternative is a schedule that runs so late the last two patients leave without being seen. The honest comparison is not slots offered but consultations actually completed within the session, and clinics with sensible buffers usually complete as many or more because they are not spending the last hour in triage-by-apology.

Every template needs an override mechanism, because clinical time is interrupted constantly. The categories are predictable: planned leave, operating-theatre days that pull a surgeon out of clinic, conference and academic absence, emergency unavailability, and locum or registrar cover. Each should be a distinct override type, because they have different notice periods and different patient-communication obligations.

The critical design question is what happens to appointments already booked into an overridden session. Cancelling silently is the worst outcome and the most common one when overrides are applied directly to the calendar. The override should generate a worklist: these patients are affected, here is the contact status of each, here are the alternative slots offered. A platform such as HealUDoc can drive that worklist from the override itself, so the cancellation and the patient contact are not two separate acts of memory.

Notice periods deserve a policy. A surgeon's OT list is usually known weeks ahead, so clinic sessions on OT days should never have been opened in the first place — that is a template problem, not an override problem. Emergency absence obviously cannot be planned, but if a department is generating frequent short-notice overrides for reasons that are actually foreseeable, the fix is upstream in how sessions are published.

How far ahead you publish slots is a real tradeoff. A long horizon lets patients book follow-ups at the time of their visit, which is when they are most likely to actually book; a short horizon reduces the disruption when a consultant's availability changes. Most hospitals settle between four and twelve weeks, longer for specialties with genuinely stable rosters and shorter for surgeons whose theatre commitments move.

Whatever the horizon, the change process should be logged. Who added slots, who removed them, when, and against which template version — this is the record that answers the recurring question of why a clinic that was supposed to have twenty slots ended up with twenty-eight. Without it, capacity drift is invisible until someone notices the clinic has been running an hour late for months.

Review templates on a fixed cycle rather than only when someone complains. A quarterly pass comparing configured slot duration against observed consultation duration, per doctor and consult type, will surface drift early and gives departments a factual basis for the conversation. The point of the review is not to police clinicians; it is to make the promised schedule match the one that actually happens.