Cardiology Scheduling for Device Checks and Rhythm Clinics

Cardiology scheduling rarely fails because the practice ran out of slots. It fails because the slot that was open was not the right kind of slot, and nobody found out until the patient arrived for a device check that a technician was not staffed to run.

A cardiology schedule carries more distinct appointment types than almost any other outpatient specialty of the same size.

Device clinic visits run on a cadence the practice sets and are staffed differently from an office visit. Rhythm and anticoagulation clinics have their own templates and their own eligible providers. Stress testing and advanced imaging occupy rooms with prep and turnover times that do not match clinic hours. New consults, post-procedure follow-ups, and established visits all look like an appointment to a patient and like four different objects to the schedule.

The person on the phone has to resolve which one applies from a sentence like I need to come in for my pacemaker. Get it wrong and the correction costs two phone calls, a room that sat empty, and a patient who drove in for nothing.

Access is the stated priority, and scheduling accuracy is most of it

Practice leaders already know where the pressure is.

Asked about their top patient access focus for 2026, 27% of 236 respondents in a Dec. 9, 2025, MGMA Stat poll chose no-shows, ahead of online scheduling at 24% and phone access at 22%, with wait times at 21%. Those four are usually discussed as separate initiatives. In a specialty practice they are the same initiative viewed from different sides.

A patient booked into the wrong appointment type becomes a no-show statistic when they do not arrive, a wait-time statistic when the visit has to be rebooked, and a phone-access statistic because the correction takes two more calls. Online scheduling makes all of it worse if the rules behind it cannot express what the practice actually does.

So the work that pays off most in cardiology scheduling is not adding capacity. It is making sure the capacity that exists gets matched to the right request the first time.

The complication: a generic slot is not a specific appointment type

This is the mechanic that defeats most scheduling automation, and it is worth stating precisely.

Practices build templates out of generic openings, an Any 15 or an Any 30, so that experienced schedulers can apply judgment about what fits. When you search availability, the system returns those generic openings. It does not tell you whether that particular opening, for that provider, in that department, is genuinely eligible for a device interrogation appointment that needs a technician, or for a new consult that needs an hour.

Resolving it means holding an explicit map: which specific appointment types each generic slot can absorb, per provider, per department. Reading GET /appointmenttypes to enumerate what the practice actually offers, then checking GET /appointments/open against that map, is what turns the map into something an automation can act on rather than knowledge held by two senior schedulers.

The catalog also moves. One practice retired an entire category of appointment types in the system overnight and folded them into a single fifteen-minute follow-up type, which cannot hold a forty-five minute service. Any rule set hardcoded against type names starts producing unservable bookings the same day a change like that lands. The map has to be re-derived from live configuration rather than from a spreadsheet written at go-live.

There is a second-order trap on duration. When staff manually convert a slot from one type to another and the duration changes, they have to freeze or delete the neighbouring slots to make room. An automation that changes a slot type without accounting for its neighbours creates a double-booking that nobody sees until the morning huddle.

Linked visits are where cardiology loses the most

The single most expensive scheduling failure in this specialty is booking one half of a pair.

A large share of specialty visits are two appointments that must happen together with specific spacing: a study and then the provider visit to go over it. At one vascular group roughly 90% of visits followed that pattern. Nothing in the scheduling template enforced the pairing. It existed only as convention, and patients called asking to see the doctor without knowing the first appointment existed.

The failure mode is real and quiet. On a live test at that practice, the booking for the study and the provider visit completed, but the follow-up task that should have been created was not. The second half of the workflow silently did not exist until staff caught it in review.

Native reminders make it worse. The system’s own reminder fires on the chronologically first appointment of the day, so a patient with a 9:30 study and a 10:00 provider visit gets reminded about one of them. Practices with paired visits routinely ask to replace the native reminder for exactly this reason.

The automation has two jobs here. Book the pair as a pair, with the spacing the practice requires, and confirm both legs in one message to the patient rather than two. Then hand the exceptions to a scheduler: cases where only one leg has availability inside the required window, where the pair spans a plan year, or where the patient’s description could map to two different pairings. Those go to a person with the candidate slots already assembled.

Recurring device visits are a different scheduling object

Device clinic follow-ups behave less like appointments and more like a series, and that changes what good scheduling looks like.

The cadence is set by the practice and the patient stays on it for years. That means the schedule is populated far in advance, and the cost of a missed visit is not one empty slot but a break in a series that someone has to notice and rebuild. It also means these are the visits most exposed to attrition, because a reminder for something booked eight months ago competes with everything else in a patient’s life.

MGMA’s reporting on missed appointments describes no-show rates returning toward pre-pandemic levels, with the single-specialty aggregate rate rising to 6.81% in 2023 after several lower years, and a Jan. 7, 2025, MGMA Stat poll of 622 applicable responses found 42% of medical group leaders report their practices using a no-show fee against 58% that do not. Fees are a blunt instrument for a recurring-series problem. Recovering the slot is better than charging for it.

The operational version: when a device visit cancels, the automation should know the series it belongs to, offer the vacated slot to patients waiting for that same appointment type rather than any patient, and rebook the cancelling patient inside the window the practice specified rather than at the next convenient opening. Whether the interval can be extended at all is a decision for the clinical team, and the automation books what they have specified rather than deciding it.

Measure whether the schedule is getting more accurate

Most scheduling programs report volume. Volume is the least useful number available.

Track the correction rate: appointments changed to a different appointment type within some window of booking. That single measure tells you whether requests are being resolved correctly at the point of contact, and it moves quickly when the type map improves.

Track pair completion for linked visits, expressed as the share of studies that have their matching provider visit booked. Track the fill rate on vacated device clinic slots separately from general no-show recovery, because those slots need a specific patient rather than any patient.

Wait time is the outcome those feed. A July 14, 2026, MGMA Stat poll of 197 applicable responses found 46% of medical groups reported new-patient appointment wait times unchanged year to date, while 28% said they were longer and 22% said shorter. A practice that is not moving on that number and has not measured its correction rate does not yet know whether its problem is capacity or matching.

Run the comparison before adding providers. Rebooked appointments and half-booked pairs consume capacity that already exists, and they are considerably cheaper to fix.

Key Takeaways

Cardiology scheduling fails on appointment type resolution far more often than on raw availability. Measure the correction rate before adding capacity.
Build an explicit map of which specific appointment types each generic template slot can absorb, per provider and per department, and re-derive it from live configuration.
Treat an appointment type catalog change as a scheduling incident. Rules hardcoded against type names start producing unservable bookings the day the catalog moves.
Book linked study and provider visits as a pair with the required spacing, and confirm both legs in one message rather than two.
Replace native reminders for paired-visit days. The system reminder fires on the first appointment only, so patients miss the other one.
Treat recurring device clinic visits as a series. A cancellation breaks a sequence someone has to rebuild, not just an empty slot.
Offer a vacated device clinic slot to patients waiting for that same appointment type, not to a general waitlist. Those slots need a specific patient.
Route single-leg availability, pairs that span a plan year, and ambiguous requests to a scheduler with the candidate slots already assembled.

The reason cardiology scheduling resists automation is that the rules are real, specific, and mostly undocumented. They live in the heads of schedulers who know that this provider’s Any 30 can hold a device check and that one cannot, that a study and a provider visit travel together, and that the catalog changed last month. An AI team working the scheduling surface inside athenaOne can hold that map per department, resolve the caller’s request to a specific appointment type before offering a slot, book paired visits as pairs, and hand the ambiguous cases to a scheduler with the options already worked out.

Sources