OB Panel Scheduling Across an Entire Pregnancy

OB panel scheduling breaks in a specific way. The first few visits get booked, the rest are supposed to be booked later, and later never arrives on a schedule anybody owns.

Prenatal care is not a repeating appointment. The intervals tighten as the pregnancy progresses, some visits are tied to a window rather than a date, the provider class can change partway through, and a single change to the estimated due date shifts everything that follows. Practices handle this with a scheduler, a paper protocol, and a lot of remembering.

A pregnancy is a dependency graph, not an appointment

The mental model matters here, because the wrong one produces the wrong tooling.

Obstetrics is a dependency graph rather than a series of independent bookings. There is an intake, then interval visits that get closer together as the pregnancy advances, and often a switch in provider class late on. Procedures can require a clearance visit first. So when a patient says she wants to book an appointment, what she actually needs is several linked bookings whose spacing is defined by where she is in the pregnancy.

Most scheduling templates cannot express that. They can hold a recurring series at fixed intervals, which is not what prenatal care is, or they can hold single appointments, which is what practices fall back on. The result is that the schedule holds the next visit and the protocol holds the rest, and the protocol is enforced by whoever remembers it.

The visible symptom is a patient who has been seen four times and has nothing on the books, discovered when someone runs a report or when she calls. By then the window for the next visit may already be closing.

The episode is the object worth reading

What makes this tractable is that the obstetric episode already exists as a record. It does not have to be inferred from appointment history.

Reading the episode through GET /chart/{patientid}/obepisodes/{obepisodeid} gives the anchor: where the pregnancy is, which visits belong to it, and what has already happened. Comparing that against open availability through GET /appointments/open is what turns a protocol into a bookable plan.

That comparison is the whole mechanic. For a patient at a given point in her pregnancy, the practice’s protocol defines which visits should exist and roughly when. The episode says which of those already exist. The difference is a work list, and it can be generated every day rather than discovered by accident.

The work list is also the right unit for outreach. Rather than a general reminder campaign, the practice can call the specific patients whose next visit window opens soon and who have nothing booked inside it. That is a much smaller and much more productive call list than everyone with an active pregnancy.

What the automation does not do is decide the intervals or which visits a particular patient needs. The protocol comes from the clinicians and is configured once. When a patient’s plan departs from the protocol, and some do, the difference is a flag for staff rather than something the automation resolves.

The complication: a due date change rewrites the rest of the schedule

This is the failure that costs the most rework and it originates in a completely legitimate clinical event.

An estimated due date can be revised. When it moves, every visit downstream of it that was placed relative to the old date is now placed relative to nothing. Some of those appointments are still fine. Some are now too early, too late, or on the wrong side of a threshold that changes which provider can conduct them.

Nothing in the schedule announces this. The appointments sit there looking normal, exactly as they did before, and the practice discovers the misalignment one visit at a time as patients arrive.

Handling it correctly means treating a due date change as an event that triggers a review of the remaining series, not as a data update. The automation re-reads the episode, compares the booked visits against where the protocol now says they should sit, and produces the list of appointments that no longer fit. Then it offers to move the ones that are straightforward and routes the rest.

The handoff line is firm. The automation reschedules an administrative visit to a compliant slot when the protocol and availability both allow. It does not decide that a visit is no longer needed, does not compress a schedule because slots are tight, and does not tell a patient anything about what the date change means for her pregnancy. Those go to clinical staff, and a patient who asks about the change is transferred rather than answered.

Provider class changes partway through, and slots have to follow

The second structural complication is that the eligible provider set does not stay constant across the episode.

Many practices route new patients and symptom visits to nurse-midwives and nurse practitioners so physicians can concentrate on procedures and higher-risk patients later in pregnancy. Others express it as a rule that visits past a threshold are physician-only. Either way, appointments booked months in advance under one rule can fall on the wrong side of the threshold by the time they happen.

This interacts badly with the due date problem. Move the date and a visit that was correctly assigned to one provider class may now belong to another, which means the slot is wrong even though the time is fine.

A scheduling automation that reads the episode can apply the threshold as a rule rather than as a convention. When it books forward, it checks provider eligibility at the projected date rather than at today’s date. That single detail removes a large share of the rebooking that practices currently do by hand in the third trimester.

Modality rules stack on top and are worth configuring explicitly. Some visit types can be conducted virtually and some cannot, regardless of whether a telehealth slot is open. An automation that treats an open slot as permission will book visits that have to be converted back.

What this costs in phone time today

The work described here is currently done, and it is done on the telephone by people.

In a March 10, 2026, MGMA Stat poll of practice leaders, scheduling accounted for 31% of the most time-intensive phone tasks, behind eligibility and prior authorization at 45%, and ahead of intake at 9% and prescription refills at 6%. The poll had 294 applicable responses.

Obstetrics concentrates that scheduling burden into repeat contact with the same patients. A single pregnancy generates a dozen or more scheduling touches, and each one is an opportunity for the series to drift. The practices that stay on top of it usually do so because one experienced scheduler holds the protocol in her head and works the list personally.

That arrangement works until she takes a week off. Making the protocol a configured rule rather than a person’s knowledge is the actual deliverable here, and the reduction in phone time is a consequence of it rather than the point.

Coverage moves while the series is still running

A visit series that spans most of a year is a series during which the patient’s coverage can change, and the schedule has no way of knowing.

Medicaid is the largest single payer of pregnancy-related services, financing 41% of births nationally in 2023 and nearly half of births in rural communities. For a large share of obstetric patients, the coverage verified at intake is a fact about January that the practice is still relying on in July.

The useful move is to reuse the machinery you already built. The same daily work list that checks each active episode for missing appointments is the natural place to re-check coverage, because it already iterates over exactly the right patient set on exactly the right cadence.

What comes out of it is a short exception list rather than a full re-verification. Patients whose coverage no longer matches what is on file get a call before their next visit, and the call is handled the same way any coverage discrepancy should be: state what is on file, ask whether anything has changed, and route the answer to the staff who can fix it. The automation never tells a patient she has lost coverage, and a coverage question never cancels a prenatal visit.

Measure the series, not the appointment

Standard scheduling metrics do not detect this problem. Fill rate and no-show rate can both look healthy while half the active pregnancies have nothing booked beyond next month.

The measure that matches the workflow is the share of active obstetric episodes whose next protocol visit is booked inside its window. Run daily, that number is a live read on whether the series is holding, and it moves immediately when something breaks.

Two others are worth having. Count appointments rebooked following a due date change, which tells you whether the review is happening systematically or being discovered patient by patient. And count visits conducted by a provider class the protocol did not intend for that point in the pregnancy, which is the clearest signal that forward booking is being checked against today’s rules rather than the rules that will apply on the date of the visit.

Key Takeaways

Model prenatal care as a dependency graph, not a recurring series. The intervals tighten, provider class can change, and some visits depend on others happening first.
Read the obstetric episode rather than inferring the plan from appointment history. The episode is an existing record and it is the right anchor for the series.
Generate the gap list daily: for each active episode, which protocol visits should exist and which are missing. That list is a far better call list than every pregnant patient.
Treat a due date change as an event that triggers a review of every remaining visit, not as a field update. Misaligned appointments look completely normal on the schedule.
Check provider eligibility at the projected appointment date, not today’s date. A visit booked months ahead can land on the wrong side of a provider-class threshold.
Configure modality rules explicitly. An open telehealth slot is not permission to conduct a visit type that has to be in person.
Keep the protocol with the clinicians. The automation books against it and flags departures; it never decides intervals or drops a visit because slots are tight.
Measure the share of active episodes whose next protocol visit is booked inside its window. Fill rate and no-show rate will look fine while the series quietly falls apart.

The reason OB scheduling stays manual is not complexity for its own sake. It is that the rules are real, they are specific to the practice, and they change partway through a nine-month relationship with the patient. A scheduler who knows those rules can hold the series together, and most practices depend on exactly one person doing that. An AI team working inside athenaOne can hold the same rules as configuration. It reads the episode and compares booked visits against where the protocol says they belong. It works the gap list before windows close and re-checks the whole series when a due date moves. It books forward against the provider rules that will apply on the day of the visit, and sends every clinical question to the people who own it.

Sources