Vascular Surgery Follow-Up Tasks and the Missing Second Visit

Vascular surgery follow-up tasks look simple in a queue and are not simple in practice. One line item can mean two appointments on two different calendars with specific spacing between them, and the patient on the phone has no idea any of that is true. They called to see the doctor.

A vascular booking history can show that many visits are an ultrasound study with the technologist plus an office visit with the physician, booked as a linked pair. Nothing in the scheduling template requires the pairing. It exists as convention, held in the heads of the people who have worked there longest, and it is visible in the data only because the two appointments always appear together.

That gap between convention and configuration is where the revenue leaks. A new scheduler books the visit the patient asked for. The second half of the workflow never gets created. Nobody notices until the physician opens a chart and the study is not there, and by then the appointment is tomorrow.

The same environment runs a second version of the problem on the reminder side, a third in the open order queue nobody is calling from, and a fourth on surveillance recalls that come due a year out. All of them are administrative, all of them are invisible in a monthly report, and all of them cost real visits.

The pairing rule is in the history, not the template

The way to find rules like this is to stop asking and start looking. A useful opening move on a new account is to pull ninety days of appointments, departments, providers, and appointment types, rank them by volume, and see what the practice actually does rather than what the written rules say it does.

Booking history often shows that a meaningful share of the logic a practice hands a vendor is out of date, and the history is what corrects it. When the study and provider visit consistently appear together with the required spacing, that pattern can become a booking rule instead of a tradition.

Once it is a rule, the automation can enforce it. A patient calling for an appointment with the physician gets both legs offered, in the right order, with the spacing the practice requires, and neither one is bookable without the other. Enforcing it consistently is what separates a booking tool from genuine schedule optimization.

A real failure: one leg booked, the other silently missing

This is not hypothetical. On a live test call, the agent can complete the booking for the study and the provider visit for one body region while the follow-up task that should have been created is not, so the second half of the workflow silently does not exist. Staff catch it during call review, which is exactly what call review is for.

The lesson we took from it is that a linked booking is not done when the appointments exist. It is done when the appointments exist and the task state matches. Whether completing a booking satisfies and closes the underlying tickler is a per-practice policy, and closing an order that has an authorization linked to it behaves differently again.

So the workflow verifies its own work. It confirms both appointments landed, confirms the task or order reached the state the practice defined, and raises the exception to a person when it did not. AHRQ’s ambulatory safety work makes the same point in a different vocabulary: the steps between visits are where things fall through, and only a reliable process closes them. The verification step is only possible when the automation can see task state directly, which is the practical argument for working inside athena.

The open order queue is an outbound call list nobody owns

An order sitting unscheduled is a decision the practice already made and has not collected on. In a vascular practice the order is usually a prerequisite, so an unworked order does not just delay revenue, it wastes the consult slot booked behind it. The surgeon opens the chart before the visit, the study is not there, and the appointment becomes a conversation about scheduling instead of a decision about a procedure.

The scale of this is documented well outside vascular. An analysis of primary care referrals to specialists in a large health system found the documented rate of closing the referral loop was under 35%. Loop closure fails the same way everywhere: no owner, no trigger, and a patient who has to be reached at a time nobody is calling.

Working the queue is outbound calling in volume, spread across evenings and weekends when patients actually answer. The automation takes the athena order and tickler queue as its work list, offers a real slot against the appointment type the order specifies, books it, and marks the entries where the patient has moved, changed insurance, or already had the study elsewhere. Whether the order is still appropriate is not its call. Anything that has turned into a question goes back to staff.

Native reminders only mention the first appointment

Here is the one that surprises administrators most. When a patient has a 9:30 study and a 10:00 provider visit, the EHR’s native reminder sends the chronologically first appointment, and the self-check-in reminder launches only for the provider visit. The patient hears about one appointment and shows up for one appointment.

Teams often ask whether native reminders can be replaced for exactly this reason. On a multi-appointment day the native behavior is not just unhelpful, it actively teaches the patient the wrong thing. They arrive at 10:00 having missed the technologist entirely, which burns a study slot, delays the visit, and generates a rebooking call. The practice counts that as a no show. It is a notification design problem with a no show attached.

It matters because no shows are where practice leaders say their attention is. Asked where to concentrate patient access effort, they split across no shows at 27%, online scheduling at 24%, phone access at 22%, and wait times at 21%. A meaningful share of those no shows are linked appointments where only one leg was ever confirmed.

Replacing that means the reminder is built from the day, not from a single appointment record. One message, both appointments, correct arrival time for the first one, and a confirmation reply that writes back so staff can see who has acknowledged what.

Parsing what the task actually says

Open follow-up tasks in a procedural practice carry free-text clinician shorthand. Internally we call it doctor speak: an abbreviation for the specific procedure, a note that there is no sedation, which clinic, which provider, and the authorization window it has to land inside.

To place that appointment, the outreach has to resolve the shorthand into an appointment type and a provider, check the authorization dates, and offer only slots inside the window. Get that wrong and the visit happens and the claim does not. Rescheduling adds its own trap, because a rescheduled appointment is not automatically linked to the existing authorization, and an appointment whose authorization has not come back can only be pushed later, never pulled earlier.

The boundary here is firm. The automation reads the task to work out what to book, calls the patient, and places the appointment. It does not decide what the patient needs, it does not act on anything in the record beyond scheduling, and anything ambiguous goes to the clinical staff queue with the original note attached.

Surveillance recalls and the year-long gap

Vascular practices carry a long tail of surveillance imaging with recall intervals measured in months or years. Patients fall out of the loop because a year is long enough to move, change plans, or simply forget, and because nobody has time to work a recall list that will not affect this week’s schedule.

Staffing is why it stays that way. Only 24% of medical group leaders reported patient access improved in a year when 38% said it stayed the same and another 38% said it got worse. Recall is margin work, and the margin is where a shortage lands first. Recall also has a discovery problem: a real fraction of any long list is no longer reachable or no longer the practice’s patient, and finding that out is what makes the work feel unrewarding. Automating the top of the funnel changes the economics, because the automation absorbs the dead entries and hands staff the live ones.

Order-driven outbound outreach handles this well, since it is the same mechanic as the linked-visit booking with a longer clock. The order exists in athena, the recall date is known, and an automated call can reach the patient, confirm their information is current, and put the study on the calendar.

Care coordination research keeps landing on the same conclusion: the reliable version of this is a process that runs on its own, not a reminder to staff who are already fully booked with the phones in front of them. Procedural practices run the same pattern on their case calendars, as in surgical scheduling automation.

The results callback is scheduling, not interpreting

After a study is read, somebody has to talk to the patient. In most vascular practices that is a call a clinician or nurse makes, and the bottleneck is not the conversation. It is arranging it: rounds of phone tag, a patient at work, a callback window nobody agreed to.

Separating the two halves is what makes this automatable. Finding a time both parties can commit to is scheduling. What gets said on the call is clinical and stays with the person qualified to say it. The automation reaches the patient, confirms a window, puts it on the right staff member’s schedule in athena, and reminds both sides. Asked what the result was, it says a member of the care team will go over it on the call, and goes no further. That line is the design, not a limitation to work around.

The measurable effect is time to callback, a number most practices have never had because the work lived in phone tag instead of on a schedule.

Key Takeaways

Infer linked-visit rules from ninety days of booking history, then confirm them with the practice. Expect roughly a fifth of the written rules to be out of date.
A linked booking is not complete until both appointments exist and the task or order reached the state your practice defines. Verify it, do not assume it.
Check what your native reminders send on a multi-appointment day. Patients told about one of two appointments will miss the other.
Free-text task notes have to be resolved into an appointment type, a provider, and an authorization window before anything gets booked.
Rescheduling breaks the authorization link. Treat a reschedule as a new alignment problem, not a date change.
Treat the athena order and tickler queue as an outbound call list with an owner, not a report somebody reviews when there is time.
Run surveillance recalls as automated order-driven outreach. A year-out recall list will never get worked by hand.
Book the results callback as a real appointment and measure time to callback. Phone tag hides the delay.

The pattern in vascular practices is that the hard part is never the first appointment. It is the second one, the task behind it, and the reminder that only mentions half the day. Those are administrative problems with administrative fixes, and they live in the athena order queue, appointment types, and reminder configuration you already have.

Sources