Your Post-Discharge Program Might Be Measuring the Wrong Number

I watched a Stripe dashboard cross a million dollars in a single day once, back in my DTC days. Genuinely thrilling for about an hour.

Then you remember the product underneath wasn't actually that good, and scale just meant more people were about to find that out, faster than any of us could fix it.

A study just published in the Journal of Medical Internet Research handed health system leaders almost the identical lesson, except the currency wasn't revenue. It was patient engagement, and the dashboard everyone was celebrating was quietly hiding the same problem mine did.

What did the study actually find?

Researchers ran a secondary analysis on patients enrolled in an automated SMS follow-up program after hospital discharge, focused specifically on the ones who ended up back in the hospital within 30 days. Fewer than half of those patients had engaged with the mHealth program at all before their revisit. The program had reach. It didn't have engagement, and those turned out to be very different numbers.

Why does a gap like that matter if enrollment still looks strong on paper?

Because enrollment is the number that goes in a board deck, and engagement is the number that actually determines whether an intervention does anything. The parent trial behind this study, MORE-PC, found no significant reduction in 30-day readmissions from its texting intervention despite reasonable participation in an earlier pilot phase. You can hit your enrollment target and still be running a program that isn't reaching the patients about to come back through the ED. That's the version of my Stripe dashboard I keep thinking about. Growth without a good product underneath it isn't success, it's just a bigger number to be disappointed by later.

Is this specific to text messaging, or does it point at something bigger?

It generalizes, and this is the part health system leaders should sit with. mHealth users in the study skewed younger and more likely to have commercial insurance, which means the patients least likely to engage with a text-based program were often the ones with the fewest other resources. A follow-up strategy that only works for people with a smartphone, decent literacy, and time to read and respond to messages isn't a follow-up strategy. It's a filter, and it filters out exactly the patients who need contact the most.

What's actually missing from most post-discharge digital programs right now?

Presence. A text sits in a queue until someone gets around to it, if they ever do. A phone call happens in real time and asks for a response right there, which is a completely different psychological ask. My daughter is ten, and she has this line she uses on me when I'm distracted at dinner: "I work for you, not the other way around." She means the phone. It applies just as well to healthcare technology. A follow-up channel exists to serve the patient's actual behavior, not the other way around, and for a huge portion of the discharged population, a phone call is simply the channel that matches how people actually want to be reached.

How is a voice AI agent different from another app, portal, or SMS thread?

It doesn't require a download, a login, or literacy in a specific written format, and implementation doesn't mean building a new app or portal either, you can see how the integration actually works. Across more than a million patient interactions in five countries and three languages, we've seen 85% weekly engagement against the 15 to 20% baseline that most digital health programs settle for, with zero critical adverse events across that entire population. That gap isn't a marketing number. It's the difference between a program that reaches the patients who were never going to open the app and one that quietly excludes them while still looking successful in a quarterly report.

What should a health system actually change about how it measures these programs?

Stop measuring enrollment as if it were engagement. Start asking what percentage of discharged patients had an actual two-way interaction, not just a message delivered, and segment that number by the populations most likely to fall through. Case studies from health systems already doing this show the real work is in that reporting shift as much as it is in the technology itself. If the engagement number would embarrass you in front of your own board, that's the number to fix first, regardless of how good the enrollment slide looks.

Key takeaways

Reach and engagement are not the same metric, and the JMIR analysis of the MORE-PC trial makes that painfully specific: fewer than half of enrolled patients engaged with an SMS follow-up program before the exact moment that engagement mattered most. Programs that look successful on an enrollment dashboard can still be filtering out the patients with the fewest other resources, which turns a follow-up program into an equity problem wearing an efficiency costume. Voice closes that gap in a way text structurally can't, because it asks for presence instead of a click, and presence is what turns a follow-up program from a checkbox into an actual intervention.

FAQ

Isn't automated texting cheaper than a voice AI program? It can look cheaper on a line-item basis, but if fewer than half of enrolled patients ever engage, the effective cost per patient actually reached is often higher than a channel with genuine engagement. Cheap outreach nobody responds to isn't actually cheap, and the real math tends to land closer to a 31:1 return once you count what a prevented readmission is worth.

Does this study mean digital health follow-up doesn't work? No. It means one specific channel, asynchronous SMS, has a structural engagement ceiling for a meaningful share of patients. Voice-based follow-up solves a different part of that problem because it requires real-time presence rather than an optional reply.

How do we know if our own follow-up program has this same gap? Look at engagement, not enrollment. If you can't easily answer what percentage of discharged patients had a genuine two-way interaction within the follow-up window, and how that number breaks down by age and insurance type, that's usually the first sign the gap exists in your own data.

If your team wants to see what engagement actually looks like on your own discharge population, book a discovery call and we'll go through the numbers together.