1. Start with the constraints, not the ratings
Three things narrow a list faster than any rating: whether a bed is actually available when it is needed, whether the facility accepts the paying programme involved, and how far it is from the people who will visit.
Distance deserves more weight than families usually give it. Someone who visits often notices changes early, asks questions, and is known to staff. A facility twenty minutes away that gets visited three times a week is a different situation from one an hour away that gets visited monthly, and no star rating captures that difference.
Ask about availability before researching quality in depth. It is the constraint most likely to remove a facility from consideration, and it costs one phone call.
2. Build the shortlist from the certified set
Search by state, county, city, or ZIP to see the CMS-certified nursing homes in an area. Every facility here is drawn from the same public CMS data, so the shortlist starts from a complete certified set rather than from whoever advertises.
One rule governs everything after this: compare facilities within the same state. CMS grades the health-inspection component relative to other facilities in that state, so a four-star facility in one state and a four-star facility in another have not cleared the same bar. Cross-state comparison is the most common way a careful reader still reaches a wrong conclusion.
3. Read the record in the order that matters
Start with the date of the most recent inspection, not the citation count. A facility with no recent citations may have had a strong survey — or may not have been surveyed lately, and the two records look alike. A facility inspected three months ago with two minor findings tells you more than one with none whose last survey was two years back.
Then read the inspection history itself, since CMS anchors the overall rating on it and adjusts from there. Look at the severity letters and whether findings reached residents, and note whether a citation came from a routine survey or a complaint investigation.
Then look at the components separately. When staffing and inspection disagree, that disagreement is information — it usually tells you which part of the record the overall star is averaging away.
4. Match the record to the stay you expect
The same building commonly serves two populations: people in short-term rehabilitation who expect to go home, and long-stay residents who live there. CMS reports several quality measures separately for each precisely because their needs differ.
So read the measures for the population you belong to. If the expected stay is a few weeks of rehabilitation after surgery, the short-stay measures and the facility's therapy programme are the relevant record; the long-stay measures describe a different group of people. A facility can be genuinely strong for one and unremarkable for the other, and the overall star will not separate them for you.
5. Visit, and know what you are checking
A visit is not a formality. It is where you check the things no dataset records: whether call lights get answered, whether residents are up and dressed and engaged, whether staff know residents by name, how the building smells, and whether the person showing you round can answer a direct question about staffing on a weekend night.
Go at an unglamorous hour if you can — a weekday evening or a weekend rather than mid-morning on a Tuesday. Ask specifically about turnover among nursing staff and about who covers nights, because continuity of staff is closer to day-to-day experience than any published figure.
6. Verify before you commit
Public data is a snapshot and can lag current conditions. Before deciding, confirm the details that matter on Medicare's official Care Compare and directly with the facility — the CMS Certification Number on each page here is the reliable way to look a facility up, since names are similar and change with ownership.
This site organizes public CMS data. It does not recommend facilities and does not provide medical, legal, or care-placement advice; those decisions belong with the discharge planner, physician, or care manager involved.