Home Care

The hospital's newest wing has a front door

Patients and health systems want more acute care at home. Programs can’t scale on manual coordination. The gap now is operational, and that’s the one we can close.

The demand is real. The question is whether we can meet it.

Hospital at home is no longer a pilot. Demand is growing faster than most programs can scale, and what holds them back now is how they run. Clinical viability and policy are no longer the main constraints.

This week, a WVLT / Aging Untold report told the story of Colleen Callahan, 75, treated for bacterial pneumonia in both lungs through OSF HealthCare’s OnCall Digital Hospital in Peoria. Three days into her admission, her team offered to move her care home. She hesitated, worried she wasn’t well enough. She said yes, spent five days recovering in her own house, and says she would do it again. OSF has now cared for more than 30,000 patients this way.

The policy picture has caught up too. After years of short-term extensions, Congress extended the CMS Acute Hospital Care at Home waiver through September 2030. Health systems finally have the runway to invest.

I joined MedArrive after leading product organizations at brands including Allscripts, TytoCare and Alto Pharmacy. Across EHRs, virtual care and pharmacy, I saw the same pattern: good care decisions stall in the handoffs between them. Now I lead a team that used its technology expertise to solve a problem it lived firsthand, delivering home-based care in the field, and proved along the way that acute care can be delivered safely at home. What we haven’t yet proven, as an industry, is that we can deliver it at scale.

Why is hospital-at-home demand outpacing capacity?

The footprint has grown quickly. As of August 2025, 413 hospitals across 146 health systems in 39 states were approved to deliver hospital-level care at home. But approval does not mean volume. Chartis reports that many programs still run at a single-digit average daily census, well short of the roughly 15 patients it describes as a minimum viable census.

That gap between interest and volume is the story. Hospitals are full, and every bed freed at home makes room for a sicker patient inside the walls. OSF’s Melissa Meier made that point in the WVLT piece. Patients like the model too: Moving Health Home reports that 85% of recipients would recommend it.

So why do so many programs stall at a handful of patients? In my experience, it rarely comes down to clinical capability. It comes down to the operating system behind the care.

Hospital at home, defined: hospital at home is the delivery of acute, inpatient-level care in a patient’s home, with in-person nursing visits, virtual physician oversight, remote monitoring and a clear escalation path back to the hospital.

Where do hospital-at-home programs break down?

Three gaps limit growth, and each one compounds the next.

1. Identification happens too late, and by memory. Colleen was offered home care on day three. That’s typical. In most programs, eligible patients are found when a hospitalist remembers to ask, or a coordinator works through a census list by hand. Every patient found late is a bed day the hospital didn’t need to spend, or a patient who never gets offered home at all. Chartis names provider hesitancy and workflow breakdowns as core limits on identification and enrollment.

2. The transition is a dozen handoffs, not one decision. Saying yes to home is the decision. Getting the patient there is the operating process: eligibility, consent, caregiver readiness, equipment, medications, nursing coverage and the first visit, all lined up across teams and partners who don’t share a system. The broader post-acute data shows how often this breaks. In a JAMA Network Open study of Medicare patients referred to home health at discharge, only 54% received a visit within 14 days. And ONC data shows just 43% of acute-care hospitals routinely send, receive, find and integrate outside information.

3. Delivery runs on spreadsheets and phone calls. Twice-daily nursing visits within a 30-minute radius sound simple. At a census of 5 they are. At 25, with variable acuity, staffing gaps, traffic and a patient who needs an unplanned visit at 4 p.m., they are a live logistics problem. Most programs solve it with heroic coordinators. That works until it doesn’t, and it caps the census at what a few people can hold in their heads.

None of these is a clinical problem. They are operational problems. They are also the reason a program that works beautifully for 8 patients cannot get to 30.

Why won’t more point solutions close the gap?

The instinct is to look for a tool for each break: a predictive model here, a referral portal there, a scheduling app for the field team. Each one helps a little. Together they add more logins, more handoffs and more places for a patient to fall through.

The problem isn’t that any one step is missing software. It’s that the steps aren’t connected. A prediction that a patient is home-ready is worth nothing if it doesn’t trigger an eligibility review. An accepted referral is worth nothing if no nurse can reach the house by evening. Healthcare has invested heavily in clinical intelligence. What hospital at home needs now is operational intelligence: the ability to turn what we know about a patient into coordinated action across people, systems and partners.

What should AI actually do for hospital at home?

This is where AI earns its place. It shouldn’t make the clinical call; it should make sure the right call gets made in time and then actually happens. Practically, that means:

  • Surfacing home-ready patients earlier from near-real-time EHR data, so the conversation Colleen had on day three can happen on day one.
  • Orchestrating the transition, moving each patient through eligibility, acceptance and service activation inside the EHR and referral tools teams already use, without adding a new system to check.
  • Matching capacity to demand in real time, so scheduling, routing and day-of changes are planned around clinician skills, patient acuity and geography rather than a whiteboard.
  • Learning from every episode, so outcomes and delays feed back into who gets identified next and how.

Clinicians stay in charge of clinical judgment. AI takes on the coordination work that currently consumes their time and caps how many patients a program can serve.

What does scale look like when the loop is closed?

We built MedArrive around a simple idea our team learned the hard way, running care in the field: a care decision is only as good as the care that follows it. That means connecting identification, transition and delivery in one closed loop, not three disconnected projects.

When those pieces work together, the economics of hospital at home change. In our initial deployment supporting a major health system’s hospital-at-home program, the system’s finance team validated more than 10x first-year ROI, 27% more patients served with the same staff, more than 14% cost reduction and 13% less administrative time. Those results reflect that deployment and are not guaranteed outcomes. But they point to where the leverage sits. The clinicians didn’t change. The operations around them did.

That’s the shift I’d ask every health system leader to make. Stop measuring hospital at home by how many patients the program can safely treat. Start measuring it by how many eligible patients it can find, move and serve without adding headcount at the same rate.

What should health system leaders do now?

The 2030 waiver extension gives us four years to build programs that last. Here’s where I’d start:

  1. Treat hospital at home as an enterprise capacity strategy, not a service line pilot. Its value shows up in beds, throughput and access across the system.
  2. Map the transition end to end. Count the handoffs between “this patient could go home” and “the first nurse is at the door.” Each one is a place to lose time or a patient.
  3. Build on the environment you have. New tools should work inside your EHR and existing referral workflows, built on standards such as HL7v2 and FHIR, not alongside them.
  4. Plan for census, not launch. Design staffing, routing and capacity visibility for 25 patients, even if you start with 5.
  5. Keep caregivers in the frame. Families carry real work when care moves home, and too many programs plan around them rather than with them. Better coordination is how we take that burden off families instead of shifting it to them.

Patients like Colleen are ready to go home. Congress has given us the runway. What’s left is the work that has always mattered most in care delivery: making sure what should happen next actually happens.

The right care decision only creates value when it becomes care delivered.

Frequently asked questions

Is hospital at home safe? CMS’s 2024 study of the Acute Hospital Care at Home initiative found generally lower mortality than comparable inpatients and lower rates across all six hospital-acquired conditions it evaluated, though those differences were not statistically significant.

How long can hospitals offer hospital at home under Medicare? The CMS waiver has been extended through September 30, 2030, giving health systems multi-year certainty to invest.

What limits hospital-at-home growth? Mostly operations: finding eligible patients early, coordinating the transition across teams and partners, and managing nursing capacity and routing as census grows.

Does AI replace clinical judgment in hospital at home? No. AI helps identify patients, coordinate handoffs and plan capacity. Clinicians make the clinical decisions, with human oversight throughout.

The bottom line

Hospital-at-home demand is no longer the constraint; execution is. MedArrive provides operational intelligence for the transition home, connecting the people, data, workflows and partners required to identify patients earlier, coordinate cleaner transitions and enable care delivery at home.

Sources

‍

Want to see what this looks like for your population? Book a walkthrough with our team, or browse more field notes on the MedArrive blog.