Internal Medicine Refill Requests and the Three Queues Behind Them

Internal medicine refill requests arrive in a steady line all day, and they are only the loudest of three queues a chronic disease panel produces. The other two are results callbacks and open orders nobody has worked. All three land on the same small group of people.

None of these queues is dramatic. That is the difficulty. There is no crisis to escalate, just a volume of small administrative steps that has to happen every day and quietly does not, which is how a practice ends up with a hundred open orders and a refill line that runs two days behind. The staff absorbing it are usually the ones with clinical training, which is the most expensive possible way to move paperwork.

Three queues, one panel

An internal medicine panel is mostly people with ongoing conditions, and ongoing conditions generate administrative work on a schedule rather than in response to anything. Every prescription runs out. Every ordered test either happens or sits. Every result needs somebody to close the loop with the patient.

So the workload is predictable in aggregate and unpredictable in any given hour, which is the worst combination for staffing. You cannot hire to the peak and the trough still has to be covered.

The usual coping strategy is to let the queues age. Refills get worked when the phone stops, open orders get worked when someone has a slow afternoon, and results callbacks get worked because a patient called twice. Aging is a decision, it just never gets made out loud.

A refill request is a routing problem before it is anything else

Here is the line that matters, and it is worth being blunt about it. The automation never approves or denies a refill. It takes the request, confirms the identifying details, checks whether the medication and the patient match an existing record, and puts a complete request in the right staff queue inside athenaOne.

The hard part is the word “right.” Routing a refill to the correct person is not obvious, because the chart’s primary provider field is stale in most practices. Panels get reassigned, physicians leave, and nobody backfills the field. Routing off it sends requests to somebody who has not seen the patient in three years.

The workable fallback is behavioral rather than declared: route to whoever has actually seen the patient recently, and when that is ambiguous or the recent visits are split across providers, route to the practice’s defined default queue and say why. An automation that resolves this quietly and wrongly is worse than one that asks, because a misrouted refill looks handled.

Controlled substances are a separate track and should be configured that way from day one. Those requests go to staff, always, with no automated path that could be mistaken for one.

Results callbacks changed when patients started seeing results first

Federal rules give patients access to their own completed test reports, and laboratories may release completed test reports to the patient or their personal representative on request. In practice that means the portal often shows a result before anyone from the practice has spoken to the patient about it.

The operational consequence is a call volume shift, not a clinical one. Patients call in having already seen a number. The practice still owes them a conversation, and that conversation still belongs to a clinician.

What the automation handles is the scheduling of it. When a clinician marks that a patient needs a callback, the automation reaches the patient, finds a window that works, books it against the correct appointment type, and confirms it. When a patient calls in about a result, it captures that they are asking, notes what they are asking about in their own words, and routes it to the clinical staff queue without characterizing it.

That distinction is the whole design. The automation moves the appointment and the message. It does not tell the patient what the result means, and it does not decide how quickly someone needs to hear back.

Open orders are revenue sitting in a queue

The order and tickler queue is where internal medicine loses the most money with the least drama. A physician orders a test, the patient means to schedule it, and nothing happens. The order sits. The practice has already done the work of the visit that generated it.

Working that list is pure outbound calling against a list nobody has hours for, which makes it close to an ideal automation target. The automation calls the patient, explains that an ordered test is outstanding, offers slots that match the order, books it, and marks the ones that are unreachable or declined so the list actually shrinks instead of being re-dialed forever.

Recall date math belongs in the same queue and it is arithmetic, not opinion. Medicare’s annual wellness visit rules condition coverage on timing, and a subsequent visit is covered when 12 months have passed since the last one. That is a date a system can compute per patient and act on, which is exactly the kind of rule that a person tracking it manually will get wrong at scale.

The honest constraint: annual wellness visits compete with sick visits for the same slots. Automating the outreach without deciding how many wellness slots the schedule protects just moves the bottleneck. That template decision is the practice’s to make first.

Where it stops

Every one of these queues touches something a clinician owns, which is why the boundary has to be explicit rather than assumed.

The automation intakes requests, verifies identity and matching details, books and confirms appointments, works outbound lists, and puts complete, well-formed items in the correct queue. Those are logistics.

It does not approve refills, interpret results, decide who needs to be seen sooner, or answer a patient’s question about what something means. When a patient says something that sounds like it needs attention now, the automation’s job is to route it to a person immediately and not to evaluate it first.

Staff will test this boundary in week one, and they should. The practices that get value from this are the ones that write the escalation rules down before go-live rather than discovering them from a call recording afterward.

Key takeaways

Do not route refills off the chart’s primary provider field. It is stale in most practices. Route off who has actually seen the patient recently, with a defined default queue for the ambiguous cases.
Configure controlled substance requests as a separate track with no automated path, before go-live rather than after.
Treat results callbacks as a scheduling job. The clinician still owns the conversation, the automation owns finding the window and booking it.
Work the open order queue as outbound calling and let the list shrink. Marking a patient unreachable is worth almost as much as booking one, because it stops the re-dialing.
Decide how many wellness slots the template protects before you automate wellness outreach. Otherwise you move the bottleneck instead of clearing it.
Write the escalation rules down first. The value of the boundary comes from staff trusting it, and trust comes from it being explicit.

The three queues behind an internal medicine panel are not complicated work. They are relentless, they arrive every single day, and they get absorbed by the most clinically trained people in the building because those people are the ones who notice. Moving the routing and the outbound calling off them does not change who makes the decisions. It changes how much of the day is spent getting to them.

Sources: