SMS Reminders and No-Shows: What the Trials Found

Appointment-based businesses lose money in a specific, measurable way: a booked slot that nobody attends. Staff are paid, the room sits empty, and the revenue does not arrive.

This is one of the few operational problems in a small business with genuine randomised controlled trial evidence behind the fix. It is worth knowing exactly what that evidence says, because it supports a cheaper solution than most vendors will sell you.

The evidence base

The relevant source is a Cochrane systematic review: Gurol-Urganci, I., de Jongh, T., Vodopivec-Jamsek, V., Atun, R., and Car, J., "Mobile phone messaging reminders for attendance at healthcare appointments," Cochrane Database of Systematic Reviews, 2013, CD007458. It updates an earlier review published in July 2012.

Cochrane reviews pool randomised trials and grade the certainty of the result. This one included eight randomised controlled trials covering 6,615 participants.

What it found

Comparison	Result	Evidence quality
Text reminder vs no reminder	RR 1.14 (95% CI 1.03 to 1.26), 7 studies, 5,841 participants	Moderate
Text reminder vs phone call reminder	RR 0.99 (95% CI 0.95 to 1.02), 3 studies, 2,509 participants	Moderate
Text plus postal vs postal alone	RR 1.10 (95% CI 1.02 to 1.19), 1 study, 291 participants	Low

Absolute attendance rates across the review: 67.8% with no reminder, 78.6% with text message reminders, 80.3% with phone call reminders.

Two results deserve more attention than they usually get.

Text reminders performed about the same as phone calls. A risk ratio of 0.99 with a confidence interval from 0.95 to 1.02 means the review could not distinguish between them. Staff ringing every patient the day before produced roughly the same attendance as an automated message.

The cost difference was large. Two of the included studies reported cost per text message per attendance at 55% and 65% lower than the cost per phone call reminder.

Together those two findings are the business case. You are not choosing between a reminder system and nothing. For most appointment businesses, you are choosing between paying a person to make calls and paying a few cents per message for a similar result.

What the review is careful about

The authors are more restrained than the marketing built on top of their work, and their caveats are worth repeating.

The included studies were heterogeneous. Statistical heterogeneity in the main comparison was high, which means the trials did not all find the same size of effect. The certainty of evidence was graded low to moderate, not high.

The review reports no data on health outcomes, and only one study commented on adverse effects at all. Specific risks such as loss of privacy, misinterpretation of a message, or delivery failure were not reported in detail anywhere.

The authors' own conclusion states plainly that the evidence "still remains insufficient to conclusively inform policy decisions," and calls for trials that measure cost-effectiveness rather than attendance alone.

Finally, the searches were run in 2012. Messaging behaviour, delivery rates, and consent expectations have all moved since. The direction of the finding is robust. The exact percentages are a decade old.

Applying it outside healthcare

Every trial in this review took place in a healthcare setting. Nobody has run equivalent trials on hairdressers, mobile mechanics, or trades quoting appointments.

The mechanism is the same one the review names: forgetfulness. Where your no-shows are caused by people forgetting, the evidence transfers reasonably. Where they are caused by price shock, low intent, or a competitor arriving first, a reminder will not help. Those are different problems with different fixes.

How to run it properly

The intervention is simple enough that the implementation quality decides the result.

Send at a useful interval. The review notes that trials often failed to report message timing relative to the appointment, which is itself a gap in the evidence. Practically, a reminder needs to arrive early enough for the customer to reschedule rather than simply not turn up.

Make rescheduling possible in the reply. A reminder that cannot be acted on converts a no-show into a no-show that was warned about. A reply that reaches a human or a booking link converts it into a rebooked slot.

Get consent right. In Australia, commercial electronic messages fall under the Spam Act 2003 and require consent, sender identification, and a functional unsubscribe. Transactional appointment reminders sit differently to marketing, but the safest position is explicit consent captured at booking.

Do not stack channels without reason. The review found text plus postal beat postal alone, on one small study. It did not establish that more channels are always better. Three reminders for one appointment is a good way to have people mute you.

Measure it as a test

The effect size here is modest. A risk ratio of 1.14 is real and worth having, but it is not the transformation a vendor deck will promise. That makes measurement necessary rather than optional.

Record your no-show rate for a defined period before you start. Then either stage the rollout by location or by appointment type, or hold back a small control group, and compare. If your baseline no-show rate is 4%, the achievable saving is small and you should spend your attention elsewhere. If it is 20%, this is likely the highest-return automation available to you, and it costs almost nothing.

The honest summary: text reminders work, work about as well as calling people, cost substantially less, and will not fix no-shows that were never caused by forgetting.

Sources

• Gurol-Urganci, I., de Jongh, T., Vodopivec-Jamsek, V., Atun, R., and Car, J. "Mobile phone messaging reminders for attendance at healthcare appointments." Cochrane Database of Systematic Reviews, 2013, Issue 12, CD007458.
• Spam Act 2003 (Cth), and Australian Communications and Media Authority guidance on commercial electronic messages.