Why Did a 19-Hospital Remote Monitoring Trial Fail to Keep Patients Home?
I built a mental health app for people with bipolar disorder. Mood tracking, sleep logs, medication reminders, a lovely onboarding flow I was genuinely proud of. Fifteen percent of patients used it weekly.
Fifteen.
I kept telling myself the number would climb once we fixed activation. It didn't. So I threw the app in the bin and had an AI call those patients on the phone instead, and weekly engagement went to 85%, and it stayed there, and I've spent every year since trying to understand why something so much dumber worked so much better. That wasn't a product insight. It was a slap.
I thought about that app again in June, when a randomized trial across 19 hospitals reported that remote patient monitoring didn't keep people home after sepsis. Not "modest benefit." Not "needs more study." For patients 65 and older, it made things worse.
What did the sepsis remote monitoring trial actually find?
It found nothing, and that's the finding. Researchers randomized 1,286 post-discharge patients with sepsis or lower respiratory tract infection into four remote monitoring arms and usual care, then measured days spent alive at home over 90 days. Every arm landed at a median of 90 days. No arm beat usual care. Among patients 65 and older, the monitoring arms produced fewer home days than doing nothing at all, with an inferiority probability of 99.6%.
One number nobody's quoting. Of 887 patients assigned to remote monitoring, 529 actually enrolled. That's 59.6%. Four in ten never started.
Why does engagement collapse the moment you hand someone a device?
Because you moved the work onto the sickest person in the room. That's it. That's the whole mechanism.
A patient discharged after sepsis is exhausted, frightened, and holding a medication list they don't fully understand. Now hand them a tablet, a Bluetooth pairing flow, a daily questionnaire and a password. You've built a system whose first requirement is initiative from someone who has none left. The trial's own criteria required a smartphone or internet-connected device and no cognitive impairment, and it still lost four in ten at the door.
My bipolar patients weren't lazy. They were unwell. Pull-based engagement asks the person with the least capacity to do the most work, and then we act surprised when the graph goes flat.
What actually moved readmissions in the same window?
Contact that goes to the patient instead of waiting for the patient to come to it. A multi-site virtual nursing study in npj Digital Medicine looked at nine hospitals and matched 4,662 virtual-nurse-assisted discharges against 4,662 traditional ones. Thirty-day emergency department readmissions came in at 3.7% versus 13.3%.
Same staffing crisis. Same population. Wildly different result.
The difference wasn't sophistication, it was direction. At HANA we've run over a million patient interactions on that principle and we see 85% weekly engagement against an industry baseline of 15 to 20%. Not because the AI is clever. Because the phone rings and a patient answers it, which is a thing humans have known how to do since 1876. Our outcomes research is public if you want the breakdown.
So is remote monitoring a waste of money?
No, and I want to be careful, because this industry swings its pendulum hard. RPM works when it's aimed properly. Michigan Medicine's Patient Monitoring at Home program cut average hospital admissions by 59% across 1,139 encounters, with nurses monitoring seven days a week and a video visit inside three days of enrolment.
Look at what that program had that the sepsis trial didn't. Humans reaching outward, on a schedule, whether or not the patient ever touched the tablet. Adherence in that program climbed from 59% to 75% precisely because nurses started calling patients directly.
The device was never the intervention. The contact was.
What should a clinic actually change this year?
Stop measuring enrolment and start measuring reach. Enrolment tells you how many patients accepted a device. Reach tells you how many you actually spoke to inside 72 hours of discharge, which is the only window that predicts anything.
Then make outreach a property of the system rather than an item on somebody's list. Your coordinator is not going to call 400 discharged patients this month. Nobody is. That isn't a discipline problem, it's arithmetic. Automated voice follow-up absorbs the routine volume and escalates the calls that matter, which is the pattern behind our clinical use cases and the deployments in our case studies.
And pick one cohort. Heart failure, post-op, whatever your readmission penalty punishes hardest. Prove it there, then widen. The integration docs show how the plumbing actually connects.
Key Takeaways
The trial that failed and the programs that worked differ on one axis, and it isn't technical sophistication. Monitoring that requires patients to initiate loses roughly four in ten before day one, and the loss concentrates in exactly the elderly, comorbid population you were trying to protect. Programs that push contact outward, whether by virtual nurse or automated voice, reach nearly everybody and move readmissions hard.
If your engagement strategy depends on a sick person opening an app, you don't have an engagement strategy. You have a filter. And it's filtering out the people who need you most.
FAQ
Does remote patient monitoring reduce hospital readmissions? Sometimes, and the deciding factor is enrolment plus human follow-up rather than the devices themselves. Programs with proactive nurse contact show large reductions, while device-first programs with passive monitoring showed no benefit in a 2026 randomized trial and measurable harm in patients over 65.
Why do older patients do worse with remote monitoring? The likeliest explanation is that setup burden filters out the least tech-comfortable patients while monitoring routes the remaining ones back into acute care rather than resolving problems at home. The trial authors call for tailoring these programs to patient need instead of expanding them uniformly.
What engagement rate should a clinic expect from automated follow-up calls? The baseline for post-discharge phone follow-up sits around 15 to 20% weekly. Well-designed automated voice outreach reaches 85%, largely because it removes the requirement for the patient to initiate anything at all.
If you're carrying a readmission penalty and a follow-up list nobody has time to work, book twenty minutes with me and we'll do the arithmetic on your real volumes.