Sources: 3 · Verified 2026-10-27
A practice switches from phone reminders to text and the no-show rate falls by two points. The celebration lasts until someone notices that the switch happened in the same month as a new confirmation call for new patients. Reminder channel comparisons are easy to run and hard to interpret; the channel is only one of several things that changed.
Comparison at a glance
| Factor | Details |
|---|---|
| Unit | One appointment with a recorded reminder attempt |
| Groups | Defined by consent and channel availability |
| Outcome | Attendance, not message delivery |
| Limit | Observational unless assignment is controlled |
Why channel comparisons mislead
Patients who consent to text are not a random sample of the practice. They may be younger, more mobile, or more engaged, and those traits also affect attendance. A practice that compares text recipients with phone-only patients is comparing two different populations. Add a simultaneous change in staffing or confirmation calls and the comparison becomes a story about the month rather than the channel.
Comparison checkpoints
| Category | Specific Tasks | Planning Estimate / Note |
|---|---|---|
| Define |
| Before the cycle |
| Assign |
| Per appointment |
| Observe |
| Per appointment |
| Analyze |
| Monthly |
- Category
- Define
- Specific Tasks
- Name the reminder policies
- Record consent status
- Set the observation window
- Planning Estimate / Note
- Before the cycle
- Category
- Assign
- Specific Tasks
- Describe how patients reached a channel
- Check for confounders
- Document exclusions
- Planning Estimate / Note
- Per appointment
- Category
- Observe
- Specific Tasks
- Record delivery
- Record attendance
- Separate cancellations
- Planning Estimate / Note
- Per appointment
- Category
- Analyze
- Specific Tasks
- Compare within strata
- Report uncertainty
- State the design limit
- Planning Estimate / Note
- Monthly
Comparison designs compared
| Decision factor | Before and after | Stratified comparison |
|---|---|---|
| Assignment | Time period | Channel eligibility |
| Confounders | Uncontrolled | Measured and reported |
| Outcome | Delivery rate | Attendance |
| Claim | Causal | Associational |
Assignment
- Before and after
- Time period
- Stratified comparison
- Channel eligibility
Confounders
- Before and after
- Uncontrolled
- Stratified comparison
- Measured and reported
Outcome
- Before and after
- Delivery rate
- Stratified comparison
- Attendance
Claim
- Before and after
- Causal
- Stratified comparison
- Associational
What a stratified comparison changes
The analysis separates patients by consent and channel eligibility, then compares attendance within visit type and lead-time bands. Delivery is reported as a process measure, not an outcome. The write-up states plainly that the design is observational and that the estimate is associational. The practice can still learn which channel reaches which patients, without claiming that the channel caused the difference.
A quarterly reminder review
The analyst pulls three months of appointments and groups them by reminder policy: text where consented, phone where text was not consented, and both where the patient agreed. Within new-patient visits booked more than two weeks ahead, attendance is similar across the text and phone groups, but the both-channel group is smaller and noisier. The report recommends a confirmation call at day seven for the long-lead segment and labels the channel finding as inconclusive.
Controls that keep the comparison honest
| Success Factor | How To Do It | Results You Get |
|---|---|---|
| Consent strata | Separate patients by channel consent | Comparable groups |
| Lead time | Report by days between booking and visit | No mix effect |
| Attendance outcome | Use arrived, cancelled, no-show | A meaningful endpoint |
| Design statement | Say observational when it is | No overclaim |
- Success Factor
- Consent strata
- How To Do It
- Separate patients by channel consent
- Results You Get
- Comparable groups
- Success Factor
- Lead time
- How To Do It
- Report by days between booking and visit
- Results You Get
- No mix effect
- Success Factor
- Attendance outcome
- How To Do It
- Use arrived, cancelled, no-show
- Results You Get
- A meaningful endpoint
- Success Factor
- Design statement
- How To Do It
- Say observational when it is
- Results You Get
- No overclaim
Mistakes that turn a comparison into a claim
Comparing patients who consented to text with patients who did not. Measuring delivery and calling it attendance. Ignoring a simultaneous change in staffing or template. Reporting a percentage without the group size. Using the word caused without a controlled design. Each mistake produces a confident slide that does not survive a second question.
Why the comparison belongs to scheduling
Marketing owns the message and the platform. Scheduling owns the appointment and the attendance outcome. A channel comparison run by marketing alone tends to stop at delivery; one run by scheduling alone tends to ignore consent and platform constraints. The useful version joins the two records and describes the limits.
Research methodology
This paper specifies a consent-stratified observational comparison of reminder channels, with attendance as the outcome and delivery as a process measure. It does not report a channel effect.
This is a comparison design. It does not establish that one channel causes better attendance.
Data sources and methodology
This brief reports published findings as stated by each source. It does not combine study populations into a new benchmark; local operators should treat the figures as context and measure their own workflow.
- FCC, Stop Unwanted Robocalls and Texts: Consent and message-type context
- AHRQ, CAHPS Surveys: Access and communication measures
- NIST Privacy Framework: Consent and data minimization structure
Frequently asked questions
Can we randomly assign reminder channels?
What outcome should we use?
How long should we observe?
Report the design before the result
State who could receive which channel and what changed at the same time. That paragraph prevents most of the overclaiming that follows.
Book a Free Consultation →
