Is Hospital at Home Safe? What the Evidence Shows, and What Falls to You

Everyone asks the same question in the same words, and it is never really about statistics. It is about three in the morning, your mother’s house, and nobody there but you.

So is hospital at home safe? For the kind of patient hospitals select for it, yes — the outcomes are as good as a ward and in some respects better. Deaths during the stay and intensive care use both run lower, and far more patients finish the episode at home instead of in a rehab facility. But that sentence carries a load-bearing qualifier. Hospitals choose who gets offered this, and they choose carefully. The honest risk in hospital at home is not clinical. It is the twenty-two hours a day when no clinician is in the building, and who is expected to cover them.

What the evidence actually shows

The best current comparison is a 2026 analysis in JAMA Network Open that matched Medicare hospital-at-home patients against similar inpatients.

There is also older evidence on confusion. The original Johns Hopkins work found new delirium in 9 percent of patients treated at home against 24 percent in hospital, with less use of sedatives and restraints. That study is from 2005, was not randomized, and delivered far more nursing hours than any program running today — so treat it as a plausible mechanism rather than a current number. The mechanism itself is not mysterious. A person who knows which door is the bathroom is less likely to become disoriented at night.

The correction that table needs

Read the first row again: 0.4 percent against 3.6 percent. That looks like home is nine times safer. It is not, and anyone selling you that number is not being straight with you.

Hospitals screen for this program. A patient who is unstable, who needs intensive care or telemetry, who needs surgery or dialysis or imaging the house cannot provide, stays in the building. So the inpatient comparison group contains sicker people by construction, and matching on diagnosis codes does not fully undo that.

The defensible conclusion is narrower and still worth a lot: for a patient a hospital has judged suitable, treating her at home does not appear to add clinical risk. That is the claim the evidence supports. “Home is safer” is not.

Two more things belong in an honest account. Readmission rates are no better — 11.7 against 11.0 percent is a wash. And Medicare’s own advisory commission has noted that the safety and quality data participating hospitals report have not been validated against claims data. We are partly taking the programs’ word for it.

What the critics say

National Nurses United, the largest nurses’ union in the country, has argued against the model on two grounds: that remote monitoring and twice-daily visits cannot substitute for around-the-clock nursing presence, and that the model shifts unpaid labor onto families, disproportionately women.

It is worth noticing which half of that the neutral evidence supports. On clinical outcomes for selected patients, the data do not back the union’s concern. On the labor shift, they do — and the rest of this page is about that, because it is the part families are least prepared for.

What happens at three in the morning

This is the actual question, so here is the actual answer.

Her vital signs stream to a command center all night, staffed by clinicians. If something moves the wrong way, or if you press the emergency button, someone reaches you immediately by audio or video. From there the ladder is:

1. A clinician on the line, within seconds to minutes, who can see her vitals and talk to you.
2. A nurse or paramedic dispatched to the house. The federal standard is that the program must be able to reach a deteriorating patient within 30 minutes.
3. EMS, if it is faster or more appropriate.
4. Transfer back into the hospital. This happens in roughly 6 to 7 percent of stays. It is the same admission continuing — no second deductible, no new paperwork — and programs treat it as a normal route rather than a failure.

In a true emergency the instruction is the same as anywhere else: call 911. One family profiled by a large program had their six-year-old practice making the call.

Two caveats you should hear. In a rural area, “within 30 minutes” may mean the full thirty minutes. And the ladder only works if somebody notices something is wrong — the monitoring catches what monitoring catches, and the rest depends on a person in the house.

What you will actually be asked to do

Start with what should never be asked, because it is unambiguous. CMS’s then-chief medical officer stated that hospitals “are not to use family members, support persons or caregivers to provide care” that staff would otherwise give during an inpatient admission. Mass General Brigham tells families plainly that they “aren’t expected to take on medical responsibilities.” If anyone asks you to manage an IV line, adjust a pump, change a dressing or alter an oxygen setting without training you on that specific task, that is a problem worth raising on the spot.

Now the part the brochures leave out. Massachusetts’ Health Policy Commission looked at who actually provides personal care during these stays and found that help with “dressing, eating, toileting” is “typically provided by family or privately hired aides, not hospital staff.” One wife, caring for her husband during a home admission while working twelve-hour shifts, summarized it to reporters in four words: “Now it’s on you.”

So the realistic hospital at home caregiver role, assembled from what programs themselves publish, looks like this:

• Be reachable, and arrange for someone to be at the house for scheduled visits.
• Help with food, fluids, dressing, toileting and moving around between visits.
• Keep the equipment working — devices charged, wristband on, connection live.
• Watch and report. You are not being asked to assess her. If something looks wrong, press the button rather than wait for the next visit.
• Never touch IV lines, pumps or oxygen settings unless a nurse has trained you on that exact task.

Arithmetic makes the point better than argument. Two visits of roughly 45 minutes each, plus a video round, is about two hours of professional presence out of twenty-four. The other twenty-two hours are her, monitored — and whoever is in the house.

The question that decides it

Does this program include home health aide hours, and if not, who is covering nights and bathroom trips?

Some programs fund aide shifts. Many do not. If the answer is “the family,” then saying yes is agreeing to roughly a week of hands-on personal care, and you are entitled to weigh that before you answer rather than discovering it on night two. If you would need to hire that help privately, go in knowing that Medicare does not pay for home care of that kind and what in-home care actually costs.

If you are already past empty before this starts, that is not a character flaw and it is not a reason to feel guilty about declining. When you can’t do it anymore is written for exactly that moment.

When it is not the safe choice

There are situations where the honest answer is no, and screening does not always catch them.

• She lives alone and cannot get to the bathroom unaided. Monitoring does not pick someone up off the floor.
• Significant dementia. Programs require her consent or a proxy physically present, and an unfamiliar routine with strangers arriving twice a day tends to worsen confusion rather than ease it.
• Nobody can be there, and no aide hours are included. This is the single most common reason the model fails a family.
• The house does not work — no reliable power, heat or cooling, running water, or cell signal.
• You are rural and the response time is genuinely thirty minutes. For some conditions that is fine. For others it is not, and the program should tell you which.
• You are the only support and you are spent. A caregiver who cannot function is itself a safety problem.

None of these make you difficult. Between 10 and 62 percent of eligible patients decline, most often because they want medical help physically present — a legitimate preference, not a failure of nerve.

Where to go from here

If you are weighing the decision tonight, the full argument on both sides is on should you say yes to hospital at home. If you have already said yes, what to expect day by day walks through the week. For the plain explanation of the program, start at what hospital at home is, and to find out whether a hospital near you runs one, see how to find a program or, in this state, the sixteen approved New Jersey hospitals.

Questions families ask

Is hospital at home safe?

For patients hospitals select for it, the evidence shows no added clinical risk: deaths during the stay and intensive care use are both lower than for matched inpatients, and far more patients finish the episode at home rather than in a rehab facility. The honest caveats are that hospitals screen for healthier patients, readmission rates are no better, stays run about a day longer, and the safety data hospitals report to Medicare are self-reported and unvalidated.

What happens if my parent gets worse in the middle of the night?

Monitoring runs around the clock from a staffed command center, and pressing the emergency button reaches a clinician immediately. The program can dispatch a nurse or paramedic to the house — the federal standard is the ability to reach a deteriorating patient within 30 minutes — send EMS, or transfer her back into the hospital. About 6 to 7 percent of stays end in a transfer back, which counts as the same admission with no second deductible.

Does someone have to stay with the patient 24 hours a day?

No program requires a live-in caregiver on paper. In practice, a patient who cannot manage the bathroom, meals and moving around between visits will usually be screened out unless someone is there to help. Ask the program directly whether home health aide hours are included, because that single answer determines how much falls to the family.

Will I be asked to give medical care?

You should not be. Federal policy states that hospitals are not to use family members or caregivers to provide care that staff would give during an inpatient admission, and programs tell families they are not expected to take on medical responsibilities. Never handle IV lines, pumps or oxygen settings unless a nurse has trained you on that specific task.

What is the family actually responsible for?

Everything underneath the medical care: being reachable, letting the team in, helping with food, fluids, dressing and getting to the bathroom between visits, keeping the monitoring equipment charged and connected, and calling the command center when something changes. Massachusetts’ health regulator found this personal care is typically provided by family or privately hired aides rather than hospital staff.

Is hospital at home safe for someone with dementia?

Usually not, if the dementia is significant and she lives alone. Programs require either her own consent or a health care proxy physically present, and an unfamiliar routine with clinicians arriving twice a day can increase confusion rather than reduce it. Mild memory problems with a family member staying is a different situation and may be reasonable.

What do nurses’ unions say about hospital at home?

National Nurses United argues that remote monitoring and twice-daily visits cannot substitute for around-the-clock nursing presence, and that the model shifts unpaid work onto families, disproportionately women. On clinical outcomes for selected patients the available evidence does not support the first concern; on the shift of personal care onto families, the neutral evidence does support the second.

Last verified: 1 November 2026 · Next review: May 2027

This page is educational and is not medical, legal or insurance advice. Mark Duda is not a physician, a nurse, or an attorney. Whether this is safe for a particular person is a clinical judgment only her own care team can make. See our disclaimers.