Refill drop-off is the quiet revenue leak in most compounding-adjacent practices. A patient starts on a protocol, does well for the first two cycles, then something breaks in the handoff between prescriber, pharmacy, and patient. They miss a refill by ten days. Then twenty. By day thirty they've decided the results weren't worth the hassle, and you've lost not just the revenue but the clinical continuity that would have made month four the month they actually saw the change they came in for.

I've watched this pattern play out across GLP-1 programs, hormone protocols, peptide stacks, low-dose naltrexone, and just about every other compounded therapy that requires a monthly or bi-monthly cadence. The clinical case is usually solid. The operational case is where practices bleed. Here's how I think about fixing it.

Why compounding refills drop off at higher rates than commercial pharmacy

Commercial pharmacies have decades of muscle memory around auto-refill. The patient shows up, insurance runs, product is on the shelf. Compounded therapies break every one of those defaults. Nothing is auto-refilled. Nothing is on a shelf. The prescription is often patient-specific, cash-pay, requires a fresh clinical touchpoint every 30 to 90 days, and ships from a pharmacy the patient has no direct relationship with.

Every one of those differences is a friction point. In practice, we see drop-off cluster around five predictable moments:

  • The patient runs low and doesn't know who to contact (clinic? pharmacy? portal?).
  • A required check-in (labs, weight, symptom review) hasn't been scheduled and blocks the refill.
  • The prescriber needs to re-authorize but the request is sitting in a task queue behind 40 other messages.
  • Payment method on file has expired and no one has told the patient.
  • Shipping timing didn't match the patient's supply, so they had a gap and lost momentum.

Any one of these can kill a refill. Most practices lose patients to two or three of them stacked together.

The refill window is a workflow, not a date

The mental shift that matters most: stop thinking about the refill as a date on the calendar. Think of it as a window that opens 14 to 21 days before the patient runs out and closes when the medication is in their hands. Everything that needs to happen (clinical review, prescriber signoff, payment capture, pharmacy transmission, shipping) has to fit inside that window, in the right order, without any single step becoming a bottleneck.

When you draw it out, a typical GLP-1 refill has about nine discrete steps between "supply is running low" and "vial arrives at the door." If any one of those steps requires a human to remember to do something on a specific day, you will lose refills. Humans forget. Especially clinical staff carrying 200-plus active patients.

What automation should actually do

I want to be specific here, because "automated refill workflow" is one of those phrases that means everything and nothing. When I say automation, I mean a system that watches the state of every active patient and takes action on their behalf without a staff member having to notice.

Anticipate, don't react

The system should know each patient's dispense date, days supply, and titration schedule. Fourteen days out from projected run-out, it should already be doing work: checking whether a required lab is on file, whether the patient has a scheduled check-in, whether payment on file is valid, whether the prescriber has any pending questions about the last visit.

By the time a human touches the chart, the prep work is done. The prescriber sees a clean summary: "Patient X, month four of semaglutide, tolerating well per last check-in on the 12th, labs current, payment valid, ready for refill authorization." That's a 15-second decision instead of a 15-minute chart dig.

Talk to the patient in their preferred channel, at the right moment

The check-in prompt that goes out 10 days before run-out is the single highest-leverage message you send. It should be SMS for most patients, email for some, and it should ask only what you actually need: a weight, a side-effect check, confirmation that they want to continue. If they don't respond in 48 hours, a second nudge. If they still don't respond, the message escalates to a staff member with context about who's been silent and for how long.

The worst version of this is a generic "time to refill!" blast. The best version references the specific protocol, the specific dose they're on, and asks a question the patient can answer in one tap.

Route by exception, not by default

Staff time should be spent only on the patients who need a human. If the AI agent can confirm labs are current, payment is valid, and the patient reports no new symptoms, the chart should land in front of the prescriber pre-approved for signoff. If something is off (a new medication started, an unexplained weight change, an unpaid balance) the chart routes to the right person with the exception flagged at the top.

Most refill queues I've audited have the opposite pattern: every chart looks the same, staff have to open each one to figure out what it needs, and the urgent stuff hides in the middle of the pile.

The pharmacy handoff is where things get sloppy

Once the prescriber authorizes, the prescription still has to reach the compounding pharmacy, get compounded, and ship on a timeline that matches the patient's supply. This handoff is where a lot of clinic-side automation stops caring, and it's a mistake.

Your system should track the prescription past the point of transmission. Did the pharmacy acknowledge receipt? Has it entered production? Has it shipped? What's the tracking number? If the pharmacy is delayed, the patient needs to know before they run out, not after. A patient who gets a text saying "your refill is running two days behind, here's what to expect" stays on protocol. A patient who runs out on Tuesday and has no idea where their medication is starts googling alternatives on Wednesday.

Practices that operate their own compounding pharmacy or have a tight partnership with one have a real advantage here, but only if the data flows both directions. A pharmacy status change should update the patient chart automatically, and a clinical change should update the pharmacy queue automatically. Manual re-keying between systems is where these workflows die.

What to measure

If you're going to fix refill drop-off, you need to see it. The metrics I care about:

  • Refill continuation rate by month. Of patients who started in January, what percent got a refill in month two, three, four, six, twelve? Segment by protocol.
  • Median days from run-out projection to medication delivered. If this number is creeping up, something in your workflow is slowing down.
  • Percentage of refills processed without staff intervention. A healthy automated workflow should be handling 60 to 80 percent of routine refills exception-free.
  • Reasons for drop-off, coded. When a patient discontinues, capture why. Clinical? Financial? Life circumstance? Dissatisfaction? Without this you're guessing.

The practices I see growing sustainably in this space are obsessive about the second and third numbers. They know exactly how many hours of staff time each refill consumes and they push that number down every quarter.

A practical starting point

If your refill workflow is currently held together by a shared spreadsheet and someone's memory, don't try to automate everything at once. Start with the anticipation piece: a system that flags every active patient 14 days before projected run-out with a checklist of what's needed. Even that alone will recover refills you're currently losing.

Then layer in patient communication. Then payment verification. Then prescriber pre-approval routing. Each layer compounds the last.

If you want to see how this looks when the EMR, the AI agents, and the pharmacy communication are all built to work together from day one, talk to our team at Qintara. We built the platform specifically for practices running compounded protocols, and the refill workflow is one of the first things we tune with a new client.

Frequently Asked Questions

How far in advance should we start the refill workflow?

For most monthly compounded protocols, 14 to 21 days before projected run-out is the sweet spot. Earlier and patients aren't yet thinking about it. Later and you don't have time to resolve exceptions like expired payment methods or missing labs before they cause a supply gap.

What's a realistic refill continuation rate for a compounding-focused practice?

It depends heavily on protocol and patient population, but well-run GLP-1 programs I've seen hold 70 to 80 percent of patients at month three and 50 to 60 percent at month six. Hormone protocols tend to run higher because the patient population is more committed at the outset. If your month-three number is below 50 percent, workflow is almost certainly a bigger factor than clinical outcome.

Do automated patient messages create compliance risk?

Automated messages are fine when they're built with the same privacy and consent guardrails as any other patient communication. Get explicit consent for SMS at intake, keep PHI out of message bodies where possible, and make sure your vendor's messaging infrastructure is covered under a BAA. Confirm the specifics with your own counsel and compliance officer, since the details vary by state and by the exact content of your messages.

How do we handle patients who go silent?

Build an escalation ladder. First contact by their preferred channel. Second contact 48 hours later. Third contact from a human staff member who has full context on what's been tried. After that, a documented outreach attempt from the prescriber. Patients who still don't respond get moved to an inactive status with a clear re-engagement pathway if they come back later. The goal is to avoid both hounding people and quietly losing them.

Can we run this kind of workflow on a general-purpose EMR?

You can, but you'll spend a lot of engineering effort bolting on the pieces a general EMR wasn't built for: cash-pay billing, compounding pharmacy integration, protocol-based patient tracking, cadence-driven automation. Practices tend to hit a ceiling around a few hundred active patients where the workarounds start breaking. Purpose-built infrastructure gets you further with less staff overhead.