Dementia Care in Nursing Homes: What Good Actually Looks Like

Dementia is the most common reason a nursing home stay turns permanent, and it is where the gap between a good home and a bad one is widest. On a tour, you can see most of that gap in twenty minutes if you know what to look for. This article is about what to look for, what to ask, and which answers should stop you from signing.

One thing up front: a memory care unit is not a federal category. Medicare and Medicaid certification applies to the building. "Memory care" describes a unit inside it, and it means whatever that operator decides it means. You have to verify it in the building you are standing in.

Why dementia is the hardest thing to buy well

With rehab after a hip replacement, you can measure progress. Someone walks further, transfers faster, goes home. Dementia care does not work that way. The resident may not improve. The staff cannot be judged on a recovery curve, so the quality of what they do is much easier to hide, and much easier to fake on a brochure.

That is why the quality spread is so wide here. Two buildings can have the same star rating, the same carpet, the same dining room, and completely different days inside. One has residents folding towels and watering plants, and the other has a row of recliners facing a television that nobody is watching. Both bill the same way.

The second reason is staffing. The national average for total nurse staffing is 3.86 hours per resident per day, and registered nurse staffing averages 0.69 hours. Weekend nurse staffing averages 3.42 hours, so the building is thinner on Saturdays than on Tuesdays. National average nursing staff turnover is 46%. Nearly half the workforce turning over in a year is a dementia problem before it is anything else, because dementia care runs on familiarity. A resident who cannot form new memories cannot learn a new aide's face. They just experience a stranger.

You cannot fix any of this from the outside. You can only choose better and then watch.

What good actually looks like on a tour

Forget the activities calendar taped to the wall. Watch the room.

• People are doing something. Not everyone, and not all at once, but across the unit you should see residents occupied with something, even something small. Folding, sorting, holding, walking with a purpose.
• Staff redirect instead of argue. If you hear a staff member telling a resident that their mother is dead, that is a red flag. Good dementia staff enter the resident's reality when it is harmless, and steer around it when it is not.
• There is a secured outdoor space. Not just a locked door. A courtyard the residents actually use, ideally one you can see being used while you are there.
• The same faces appear. Ask whether this unit has consistent assignment, meaning the same aides work with the same residents most days rather than rotating. Consistency is one of the few things that reliably helps, and it costs a building nothing but discipline to arrange.
• Meals accommodate people who cannot manage cutlery. Finger foods, adapted utensils, staff sitting down to help rather than standing over people and rushing them.
• There is a written approach to behaviour. Ask to see it. It should start with finding a cause: pain, infection, needing the toilet, noise, boredom, a change in routine. Medication should be the last step, not the first.

Here is the part people miss. The locked door is not the problem. Every dementia unit has to control exits, and you should expect that. The question is whether the building has given residents any reason to want to stay inside it. A unit with a locked door and nothing to do is a different place from a unit with a locked door and a garden, a kitchen table, a laundry basket, and a cat.

Say, illustratively, you tour two buildings on the same afternoon. In the first, you pass a memory unit at 2pm and eight residents are seated in a semicircle facing a muted television. Staff are at the nurses' station charting. In the second, four residents are at a table with a staff member, sorting a basket of towels, and one is asleep in a chair nearby, which is fine, because it is 2pm. Same staffing on paper, probably. Different building in practice. That contrast is what you are shopping for.

The staffing questions that separate marketing from substance

Every building will tell you they provide "individualised care" and "person-centred programming." Those phrases mean nothing. These questions do.

1. How many hours of dementia-specific training do staff receive, and how often? Onboarding training alone is not enough. Ask what happens annually, and whether it applies to aides, not just nurses.
2. Is the memory unit staffed at a different ratio than the rest of the building? If the answer is no, ask why not, because the work is harder.
3. What actually happens when a resident becomes agitated at three in the morning? Ask them to walk you through the sequence, step by step. If the sequence starts with a medication, you have your answer.
4. How many residents on this unit are on antipsychotic medication, and why? This is the question most families never ask and it is the one that most reliably sorts buildings out.
5. Who covers this unit on nights and weekends, and how many of them are regulars versus agency staff?

That fourth question has a published answer for most buildings. CMS reports the percentage of long-stay residents receiving antipsychotic medication, and that figure is on the record for the building. A home well above the national picture deserves a direct question, not an assumption. It is not proof of bad care on its own, because some residents genuinely need these drugs and the diagnosis is real. But a number far above the average, in a building that cannot explain why, tells you something about how behaviour gets managed there.

You can pull that number, along with inspection history and staffing, from our search of all Medicare-certified nursing homes. If you want a shortlist to start from, the best-graded homes page is a reasonable first filter, and comparing homes side by side puts the antipsychotic percentage next to the staffing hours where it belongs.

Memory care is a label, not a certification

This trips up more families than almost anything else. Medicare and Medicaid certification is a federal status. A building either has it or does not, and it is tied to inspections, staffing requirements and the ability to bill those programmes. "Memory care" has no equivalent federal meaning for nursing homes. It describes a unit or a wing, and the operator defines it.

So the label can mean a dedicated unit with trained staff, controlled access, and programming designed around dementia. Or it can mean a hallway with a keypad on the door and a different rate. Both get called memory care.

What you can verify:

• Is the unit physically separate, with its own dining and activity space, or is it a hallway designation?
• Are staff assigned to it permanently, or do they rotate through the whole building?
• What is the resident-to-staff ratio on that unit compared with the rest of the building?
• What does the building do when a resident's needs exceed what the unit can manage?
• Does the state license memory care units separately? In some states it does. In others it does not. Ask the building and ask your state's long-term care ombudsman what applies where you live.

If a building cannot answer these plainly, that is your answer.

The medication question, and why it matters more than it sounds

Antipsychotics are sometimes the right call. A resident with a serious mental illness, or with dementia and severe, persistent psychosis, may be better on them than off. The problem is not the drug. The problem is the drug used as the default response to behaviour that has a cause nobody looked for.

A resident who becomes agitated every afternoon at four may be hungry, may be in pain, may have a urinary tract infection, may be sundowning in a room with too much noise and no light discipline, or may simply be bored. Each of those has a different fix, and none of them is a sedative. A building with a written behaviour approach that starts with causes will find some of them. A building without one will medicate, and the resident will get quieter and less themselves, and the family will be told this is "the disease progressing."

Two things follow from that. First, ask the antipsychotic question before you sign, and ask it again three months in. Second, when a behaviour changes suddenly, treat it as a medical question first. "Sudden" is the key word. Dementia declines gradually. A sharp change in a week usually means something acute: an infection, pain, constipation, a medication side effect, a fall nobody witnessed.

What a good answer sounds like

A good answer names residents and situations. "We had a resident who was up all night and it turned out to be a UTI, and we treated it and it resolved." A weak answer is a policy sentence: "We follow all state and federal guidelines regarding psychotropic medications." Everyone follows the guidelines. That is the floor, not the standard.

Ask to see the record on the building, not the brochure

Marketing materials are written by people who know what families want to hear. The federal record is not. Start with the inspection history, because that is where behaviour management problems surface, usually as citations about unnecessary drugs, dignity, or accident hazards.

Our database holds 419,479 health-inspection citations across the homes we cover. Of those, 22,914 were scored at actual harm or immediate jeopardy, which is scope and severity G or above. That is the top of the scale, and it is where you want to look first in any building you are seriously considering. There is a guide to reading an inspection report that walks through what each citation actually means, because the letters and codes are not self-explanatory.

Staffing numbers need the same treatment. "24-hour nursing" is true of every certified building and tells you nothing. What matters is hours per resident per day, the mix of RN versus aide time, and how the weekend compares to the weekday. Our guide to staffing numbers explains how to read them, and the explanation of how our grade is calculated shows what we weight and why. If you want to know where the data comes from before you trust it, that is on our data sources page.

One more filter worth knowing about: 1,428 facilities currently carry a CMS abuse icon, and 526 are in or candidates for the Special Focus Facility programme, which is CMS's list of the poorest performers. Those are small numbers against 14,696 certified homes, and they are worth checking before you tour. The watchlist of flagged homes is where we keep them.

Ownership and staffing, because they set the ceiling

You cannot tell much about dementia care from a building's tax status. You can tell something from how it is run. Of the homes we cover, 10,882 are for-profit and 2,869 are non-profit, and 10,119 belong to a named chain. Chain ownership is not automatically worse, and non-profit is not automatically better. What matters is whether the operator invests in direct care hours and keeps staff, or runs thin and turns over.

With national turnover at 46%, a building that holds aides for years is doing something deliberate. Ask how long the unit's day-shift aides have been there. If the answer is measured in months across the board, the residents are meeting strangers constantly, and no amount of programming compensates for that.

Ownership structure is public. Our ownership tracker shows who is behind a building, and there is a separate guide to who owns nursing homes if you want to understand what a management company or a real estate investment trust actually does. The short version: the name on the door is often not the entity making staffing decisions.

Paying for it, and what changes when you cross into long-term care

Most dementia care in a nursing home ends up funded by Medicaid, not Medicare. Medicare covers skilled nursing for a limited period after a qualifying hospital stay, and it is not designed to cover custodial care, which is what dementia care mostly is. When the skilled need ends and the resident is not safe at home, the payer changes.

Of the homes we cover, 14,156 are certified to accept Medicaid. That is nearly all of them, but "certified" does not mean "has a bed right now," and it does not mean the memory unit has one. Medicaid bed availability is a real constraint in many markets, and it varies by state, because the programme is state-administered.

The practical sequence: find out whether the building accepts Medicaid for long-term care, how many Medicaid beds it holds, and whether the memory unit is included. Ask whether a resident who enters as a private payer can stay once their funds are spent. The answer is usually yes if the building is Medicaid-certified, but you want it in writing, and you want to know whether the bed itself is dually certified.

Our guide to who pays for nursing home care covers the mechanics, and Medicaid by state shows how the programme is administered where you live. For what buildings actually charge, see what care costs. We do not publish dollar figures here because they change annually and vary by market, and a stale number is worse than no number. Get current figures from the building and from your state Medicaid agency.

Red flags that should stop the tour

Some things are negotiable. These are not.

• Residents parked in front of a television with no staff interaction. One resident watching TV is fine. A row of them, with staff elsewhere, is the whole problem in one image.
• An unanswered call light during your visit. Time it. If it runs past several minutes while you are standing there, it runs longer when nobody is watching.
• Smells that follow you. Occasional odour happens in any building. A persistent smell across a unit means toileting is not getting done on schedule, and in dementia care that is both a dignity issue and a health one.
• Staff who cannot tell you what a specific resident likes. Ask an aide what a particular resident enjoys. If they cannot answer, consistent assignment is not real.
• No written behaviour policy, or one that starts with medication.
• Vague answers about antipsychotic use on the unit. "We follow physician orders" is not an answer.
• Pressure to sign today. A building with a waiting list does not need to close you on the spot.

We keep a fuller list in the guide to red flags on a tour, which is worth reading before you go rather than after.

Questions to ask, in the order you should ask them

Bring these on paper. Write the answers down in the room, because buildings blur together by the drive home.

1. What percentage of long-stay residents here receive antipsychotic medication, and how does that compare with the national figure?
2. How many hours of dementia-specific training do aides get at hire, and how many per year after that?
3. Is this unit staffed at a different ratio than the rest of the building?
4. Do aides here have consistent assignment, or do they rotate?
5. Walk me through what happens when a resident is agitated at 3am.
6. What is your written approach to behaviour, and can I see it?
7. Who covers nights and weekends, and how many of those staff are regular employees?
8. What is the average tenure of the day-shift aides on this unit?
9. Can residents go outside, and is the outdoor space secured and actually used?
10. How do meals work for residents who cannot use utensils?
11. Does this building accept Medicaid for long-term care, and is the memory unit included?
12. What happens when a resident's needs exceed what this unit can manage? Do you discharge, or do you adjust?

That last one matters and families rarely ask it. Some buildings will move a resident out when behaviour becomes difficult, and "move out" for someone with advanced dementia usually means a hospital stay and a scramble. Ask it before you sign, not after.

The mistake almost everyone makes

The most common mistake is choosing on the strength of the tour and the brochure, then going quiet once the person moves in. It costs you the thing you were trying to buy.

Here is how it plays out, illustratively. A family picks a building with a nice memory unit, good first impression, close to home. The parent moves in. The family visits on Sundays, sees their mother in the activity room, and assumes the other six days look similar. Three months later she is sleeping most of the day, eating less, and has been started on a medication for "agitation." Nobody called. The family finds out at the quarterly care conference, and by then the change has been normalised inside the building. The window to ask why is largely gone.

The fix is unglamorous. Visit at different times, including one evening and one weekend, because weekend staffing averages 3.42 hours nationally against 3.86 on weekdays. Ask at every care conference for the current antipsychotic list and the reason each resident is on it. Ask what has changed in the last ninety days. If your parent has a sudden behaviour change, ask for a medical workup before you accept a medication change, and ask in writing.

You have rights here. The guide to residents' rights lays out what you can demand and what the building has to give you, including access to records and participation in care planning. Use them. A building that knows the family is paying attention behaves differently from one that does not. That is not cynicism, it is how every understaffed system works.

When the nursing home is the wrong setting

Sometimes the honest answer is that a building is not the right level of care yet, or no longer is. If your parent is still safe at home with support, home health agencies can bring care to them, and if the diagnosis is advanced and the goal has shifted, hospice providers work in homes and in nursing homes both. Moving to a nursing home is not a one-way door, and neither is staying in one.

What you should not do is choose a building purely on proximity if the record is poor. A twenty-minute drive to a building with a clean inspection history and stable staff beats a five-minute drive to one on the watchlist. You will visit less often over time, not more, and the building has to be good on the days you are not there. That is the entire point.

The bottom line

Dementia care is the hardest thing to buy in this industry because there is no recovery to measure and no federal memory care standard to hold a building to. That means the burden of judgement falls on you, and it is a burden you can carry if you stop shopping on atmosphere and start shopping on four things: how behaviour is handled before medication, how much dementia-specific training staff actually get, whether the same faces show up on the same unit, and what the federal record says about antipsychotic use, inspections and staffing. A beautiful building with vague answers to those questions is a worse choice than a plain one with specific answers. Look at the record first, tour second, and keep asking after the move, because the first ninety days tell you more about the next five years than any tour ever will.

Where these figures come from

Every number in this article that describes American nursing homes comes from the CMS Provider Data Catalog, the same public federal files behind Medicare's Care Compare. You can see exactly which datasets we use, and how current they are, on our data sources page. Nothing here is medical, legal or financial advice, rules differ by state and change over time, so verify anything that affects a decision with the facility, your state survey agency, your Long-Term Care Ombudsman, or a qualified professional.