Medicare’s five-year hospital-at-home extension won’t fix the integration problem

To use a blood pressure cuff shipped to their house, hospital-at-home patients must strap it on, take the reading, and type the numbers into a tablet by hand. Dr. David Levine has watched 80- and 90-year-olds try to do this on their own. “That is a tall order,” he said in a recent webinar, “yet that is a predominant system in many programs.”

Levine, a Harvard Medical School physician who directs research and development for Mass General Brigham’s home hospital program, sees the cuff as a fair stand-in for the overall state of hospital at home right now. The clinical case is settled, but the technology underneath it is not.

What’s changed? The money, of course. For years, the federal payment that makes hospital at home viable came in short extensions, and nobody wanted to spend real capital on infrastructure that might be abandoned. Levine compared this situation to building a service line while knowing reimbursement might disappear in six months. Despite the risk, roughly 400 hospitals built one anyway, which he reads as a sign of how much the hospital at home model is worth clinically.

The five-year extension Congress enacted in February 2026, part of the Consolidated Appropriations Act, runs the hospital-at-home waiver through 2030. With funding no longer at risk of vanishing, hospitals can shift from asking if they should build a program to how to build the most effective version. The focus now is on the unglamorous parts of hospital at home that were easy to shortchange when the life of the program was measured in months.

Integration is the least visible part of hospital at home, but the most likely to quietly wreck it

When asked what slows hospital-at-home programs down, Levine goes straight to the data. A patient’s vitals have to travel from a device in their home into the chart the care team is watching, quickly enough to act on, and making that happen is a grind in practice.

Hospital systems don’t speak a common language, either. The modern standard for this, SMART on FHIR, is meant to let systems talk to each other cleanly, but plenty of programs don’t use it. They run older, custom HL7 interfaces instead, the kind that have to be built and tested one connection at a time, with each needing its own security review before it goes near live patient data. Levine put the timeline at six to 12 months to bring one of these interfaces online.

That’s the version where somebody is doing the work, but Levine said many programs still don’t pipe their hospital-at-home data into the EHR at all. It’s judged too difficult and labor intensive, so it doesn’t happen. “That is a sad, sad fact,” he said, “but it’s kind of the state of our healthcare systems.” This leaves patients admitted and receiving hospital-level care, yet the record of that care lives somewhere other than their chart. For programs that are now supposed to be permanent, this is a technical debt that gets more expensive the longer a program grows on top of it.

The usability problem is really a safety problem

Levine splits data capture into two types, active and passive. Active means the patient does the work, reads the device, enters the number. Passive means a wearable sends the data along and the patient does nothing. Passive is what the model needs when you’re running it for dozens or hundreds of patients at once. The trouble, in Levine’s telling, is that the passive systems that actually work, with interfaces an older patient can handle, mostly don’t exist yet at the quality needed.

This isn’t just a comfort issue. If the data you act on depends on a frail patient entering numbers correctly on a tablet, you have to question how far you can trust it, and that limits which patients you can safely enroll at all.

Don’t buy one platform for three different jobs

According to Levine, there’s a reflex, whenever a health system builds digital capability, to find one vendor to cover everything. He warned against this flatly.

Hospital at home, routine ambulatory video visits, and long-term remote monitoring for chronic outpatients can look like the same thing from the executive level. Yet, Levine says organizations treating them as one procurement “get almost none of them right, because they’re trying to find a one size fits all.” Infrastructure matters, but you can only get it right by admitting these jobs require different tooling.

There’s some good news here. Once the core is built, it scales without much drama. “Once you have your EHR system in place, it doesn’t care if there’s 10, 20, 100, 1,000 patients being cared for,” Levine said. The ceiling isn’t infrastructure, it’s finding enough clinicians who can be the broad generalists a hospital-at-home program needs.

Internet access and analytics must be counted as infrastructure

Many patients don’t have reliable internet connectivity at home, and Levine frames the router and the device as an equity question. If hospital at home runs on streaming vitals and video, leaving connectivity out of the equation quietly excludes the patients who most need it.

Analytics, meanwhile, matters before a single patient is enrolled. Levine described hot spotting, mapping where patients live and layering in acuity and transfer data to size the program realistically. That work shapes the clinical build itself, rather than just reporting on it.

There’s not much time to get hospital at home right

None of Levine’s advice is exotic. He says that hospitals must fit the program to meet a real need and treat integration and connectivity as long builds with security baked in. He advocates for tooling that meets a program’s specific requirements. Scaling, he notes, is easy on the technology side but limited by the workforce using it.

He stresses that getting the “boring” parts right is urgent based on what’s coming next. By 2028, Levine suspects that the country will be near a tipping point where patients come to expect care at home. Programs that spend the five-year runway building something solid will be the ones that can handle this flip and won’t keep their patients waiting.

Go deeper at the 2026 Hospital at Home Technology Summit

Hospital at home is moving past its waiver-era roots into a permanent pillar of acute care, and the technology stack behind it is changing just as fast. On Tuesday, September 22, from 12:00 to 4:00 PM EST, Digital Health Insights hosts a free virtual summit on the systems that make Hospital at Home programs scale, from virtual command centers and continuous remote monitoring to agentic AI that prioritizes risk. Speakers from Stanford Health Care, Mass General Brigham Healthcare at Home, NewYork-Presbyterian, Tampa General, OSF OnCall and UMass Memorial Health will cover virtual nursing, care at home models outside the CMS framework, tailoring programs to Medicaid populations and building a sustainable business model. Register here.