Patient Appointment Reminders: When to Send, What to Count

Text each patient a few days before the visit asking them to confirm or reschedule, send a short reminder the day before, and call anyone who has not replied. Text reminders raised attendance in trials at clinics and hospitals; the timing is yours to test. Whether it works in your practice shows in one number, the no-show rate for each visit type, counted before you change anything and again two or three months later.

Quick answer

1. Count last month's no-show rate for each visit type: no-shows divided by arrivals plus no-shows.
2. Text a confirm-or-reschedule request two to three business days before each visit and a short reminder the day before, as a starting point to test, not a rule.
3. Call every patient still unconfirmed the afternoon before, and offer freed slots to a waitlist.
4. Keep each text to the practice name, address, day, time and phone number.
5. Count the same rate again two or three months later, changing one thing at a time.

How do you work out your no-show rate?

Use one formula and keep it:

No-show rate = no-shows ÷ (patients who arrived + no-shows) × 100

Leave out appointments canceled in time to rebook. Put late cancellations, those inside your own cutoff such as 24 hours, in a column of their own. They empty a slot too, and a reminder can turn a no-show into one. The formula fits any appointment business; the HIPAA and Medicare parts below are for health care.

Pull last month's appointment report, with every status, from your practice management system, the software that holds your schedule. Past appointments still marked booked or confirmed have no outcome, so have the front desk mark them first. Then split the count by visit type and by new versus existing patients, because one practice-wide rate hides where the empty chairs come from.

Illustrative: every number below is made up for an imaginary two-dentist practice, and none comes from a client.

The practice-wide 8.7 percent hides two different problems. Recall visits produce most of the misses, 24 of 39. New patients miss at the highest rate, 15.0 percent against 7.7 percent for existing patients (30 of 390), so the first changes go to recall visits and new-patient exams.

Do reminders actually reduce no-shows?

Yes, in the trials that tested them. A 2013 Cochrane review covered eight randomized trials with 6,615 participants, studies in which chance decides who gets a text, another kind of reminder or none. Across them, attendance was 67.8 percent with no reminder and 78.6 percent with a text, and the reviewers rated the evidence that texts beat no reminder as moderate quality, their grade for how far to trust it (Cochrane review on PubMed, checked 2026-10-04). Overall, they called the trials varied and the evidence low to moderate in quality.

A 2016 review in BMJ Open pooled 21 randomized studies with 16,076 patients. In the 16 studies that counted no-shows, 15 percent of patients who got text notifications missed, against 21 percent of those who got none. More than one notification raised attendance more than a single one did, but did not cut no-shows significantly more than a single one (BMJ Open review on PMC, checked 2026-10-04).

The reviews differ on calls, and each based its finding on three studies. Cochrane found texts and calls about equal, and in two studies a text cost 55 and 65 percent less per attended visit. The 2016 review found voice reminders appeared slightly more effective. Texting everyone and calling those who don't confirm uses both.

Of the 2016 review's 21 studies, 16 ran in Europe or Asia and one in the US. The Cochrane authors asked future trials to report message content and timing. So treat the timing below as something to test in your own practice.

What should you send, and when?

Treat this as a schedule to start from, and test it against your own counts:

The confirm-or-reschedule text goes out early enough that a freed slot can still be filled, and the two-week reminder covers visits booked months ahead, such as six-month hygiene recalls.

Keep each text short. HHS treats appointment reminders as part of treatment, so they need no signed HIPAA authorization (HHS, checked 2026-10-04). For answering-machine messages, HHS suggests leaving only the practice's name and number and what is needed to confirm the appointment (HHS, checked 2026-10-04). A text can sit on a lock screen, so I would hold it to the same limit: no visit type, provider specialty or reason for the visit.

The same HHS answer says a patient's reasonable request to be reached another way, such as calls at work instead of at home, must be accommodated. Check that your reminder system can store one.

What happens to the slots nobody confirms?

They become the front desk's afternoon calls. A two-way reminder is one the patient can answer, and the answer should land on the appointment where the desk sees it. That splits tomorrow into confirmed and unconfirmed. The desk calls the unconfirmed list at a set time and notes what each call found. Don't cancel an unconfirmed visit on your own, because no reply is not a cancellation.

When a patient cancels or moves, offer the slot to the waitlist, the patients who asked to come in sooner, and count the slots you refill. A reminder that turns a no-show into a cancellation two days ahead pays only when someone takes the slot.

Can you charge a fee for a missed appointment?

For Medicare patients, CMS allows it on two conditions: the policy applies to all patients, Medicare or not, and the amount is the same for everyone. Medicare pays nothing toward the fee, and it should not be billed to Medicare (Medicare Claims Processing Manual, chapter 1, section 30.3.13, checked 2026-10-04).

That is Medicare's rule only. For Medicaid patients ask your state Medicaid program, for insured patients read your contracts with insurers, and for state law ask a health care attorney. If you adopt a fee, put it in writing, tell patients when they book, and decide whether a late cancellation counts.

What should you ask a reminder vendor?

Start with the reminder feature in the practice management system you already use, since two systems texting one patient can double up. Then ask any vendor:

• Does a patient's reply update the appointment in the practice management system, or does someone retype it?
• Can timing and wording differ by visit type, provider and location?
• Will you sign a business associate agreement, the written contract in which a vendor holding patient information agrees to safeguard it? HHS lists an app vendor handling patient messaging for a provider among its examples of business associates (HHS, checked 2026-10-04).
• What shows on a lock screen, and can visit type and provider be left out?
• How do you record consent to texts and handle STOP replies? The FCC, which enforces the Telephone Consumer Protection Act, says texts sent to a cell phone with an autodialer need the owner's prior consent, and that anyone may opt out in any reasonable manner (FCC, checked 2026-10-04). Ask counsel how that applies to your reminders.
• Can the export list each appointment with its reminder status and its outcome?
• Does it offer a freed slot to the waitlist on its own?

If you are choosing scheduling software at the same time, the questions for a scheduling demo overlap with these.

How do you know the reminders worked?

Count the same rate for the same visit types for two or three full months after the change, and compare with the months before, or the same months last year if your schedule is seasonal. Change one thing at a time, or you won't know what moved the number.

Two more counts show where a change came from. Compare the no-show rates of confirmed and unconfirmed patients to see whether the misses sit on the list the afternoon calls work through. Then set late cancellations against refilled slots to see whether no-shows just became empty late cancellations.

Small counts move by chance. In the made-up practice, 7 new-patient no-shows instead of 9 would read 11.7 percent instead of 15.0, a change two patients can make on their own. Even three months at 11.7 percent after three months at 15.0 would not pass a standard statistical test for chance; the rate would have to fall below 8 percent. So judge a small line on several months together, and treat only a large drop as a result.

Why do no-shows make your ad reports look better than they are?

If your ads count a booking as the conversion, Google's word for a counted result, a patient who books from an ad and never comes still counts. In the made-up month, 13 new-patient bookings ended without a visit (9 no-shows, 4 late cancellations), and an account that counts bookings includes each one that came through an ad.

Google accepts a correction after a no-show, but after a set number of days it fixes the report without changing how the ads bid (how that clock works). The steadier fix is to count kept appointments, read from the appointment status in your own system (how that status gets out). When I set up patient conversion tracking for an NYC medical practice, I had Google Ads count the patient marked arrived, not the booking.

If reminders link to your online booking page, make sure a reschedule there does not fire the same conversion as a new booking, or the ad account may count one patient twice.