Texting Patients After Discharge Doesn't Cut Readmissions. Calling Them Does.
Health systems keep buying texting platforms to fix readmissions. The patients keep getting readmitted anyway. The 2026 evidence finally explains why, and it's uncomfortable reading for anyone who just signed a three-year SMS contract.
I say this as someone who once watched a Stripe dashboard cross $1M in a single day for a product I knew was bad. Scale hides problems. A texting program that reaches 100% of patients and helps none of them looks fantastic on a dashboard. Reach isn't outcome.
What does the evidence actually say about post-discharge texting?
The big one is the MORE-PC randomized trial published in JAMA Network Open: a 30-day automated texting program across 30 primary care practices, properly randomized, and it found no significant reduction in acute care revisits. Engagement was high. Satisfaction was high. Readmissions didn't move. A 2026 secondary analysis in JMIR went further: among texting-program patients who bounced back to hospital, fewer than half had engaged with the platform at all before returning.
Patients answered the texts. The texts just didn't change anything.
Why do phone calls work where texts fail?
Because a conversation captures what a survey can't. The Fraser Health quasi-randomized trial tested one nurse-led call 48 hours after discharge: 28% relative reduction in 7-day ED visits, 12% at 30 days. One call. A text asks "are you OK?" and accepts whatever box gets ticked. A call hears the hesitation before "fine," asks about the medication that didn't get picked up, catches the swollen leg the patient didn't think was worth mentioning. Structure plus voice plus follow-up questions. That's the active ingredient.
We rebuilt our whole company on this distinction. My first product was an app, and 15% of patients used it. We threw it out and started calling patients with AI instead. Engagement went to 85%. Same patients, same clinical content, different channel. The detail is in our research.
If calls work, why doesn't every health system just call everyone?
Nurses. You don't have enough of them, and the ones you have are exhausted. The Fraser Health model needed registered nurses working from daily discharge lists, and it worked, but it caps at whatever your staffing survives. That's exactly the gap voice AI fills: the agent makes the 24-hour, 72-hour and 7-day calls, asks the structured questions, and escalates red flags to your team. Capture and escalate, never advise. We've run more than 1M patient interactions on that principle with zero critical adverse events, across these use cases.
The nurse stops being the dialer. She becomes the person who only handles the calls that matter.
How do the 2026 RPM reimbursement changes fit into this?
They reward exactly this kind of lighter-touch monitoring. The 2026 CPT updates covered by MedCity News now reimburse shorter monitoring windows, 2 to 15 days within a 30-day period, and lower clinical-management thresholds. Translation: a structured post-discharge calling program isn't just clinically supported now, it's billable in configurations that used to fall through the reimbursement cracks. And a March 2026 meta-analysis of 160,857 patients found remote biometric monitoring cut all-cause readmission risk by 25%. Pair the sensor data with a voice follow-up and you have a transition-of-care program that actually closes the loop. The economics work too; we walk through the math on our pricing page.
What should a health system do with this evidence?
Stop measuring reach. Start measuring caught deterioration. Pick your highest-readmission service line, usually heart failure or post-surgical, and run structured calls at 24 hours, 72 hours and 7 days with explicit escalation paths. Keep the texting for appointment logistics, it's fine for that. Then count: ED visits, 7-day and 30-day, against a matched cohort.
Basically, fund the channel the evidence supports. Not the one with the prettiest engagement dashboard.
Key takeaways: randomized evidence now shows automated texting does not reduce post-discharge acute care use, while structured calls cut early ED visits by double digits. The constraint on calling programs is staffing, which is precisely what voice AI removes. The 2026 RPM reimbursement changes make lighter-touch follow-up billable. The systems that win this year will be the ones that measure outcomes, not opens.
FAQ
Should we cancel our post-discharge texting program?
Not necessarily. Texting is cheap and patients like it for logistics, reminders and scheduling. The evidence says don't expect it to move readmissions. Use it as the supporting channel and put the clinical weight on structured calls.
Can AI calls really substitute for nurse calls after discharge?
For the structured screening layer, yes: the AI asks the protocol questions, captures answers, and escalates red flags to clinicians immediately. The clinical judgment stays human. The AI just guarantees every patient actually gets the call, in their language, on schedule. Ours run in 5 countries and 3 languages, fully open-source and self-hosted, details at docs.hana.health.
What outcomes should we track in a post-discharge calling pilot?
Seven-day and 30-day ED visits and readmissions against a matched control, escalation volume and time-to-escalation, and call completion rates. If you want a pilot design for your service line, book a discovery call and I'll share the template we use.