When families search for a nursing home — often under pressure, often in crisis — the CMS Five-Star Quality Rating System is almost always the first thing they encounter. One to five stars. Simple, intuitive, authoritative-looking. A 5-star facility must be better than a 1-star facility. That much seems obvious.
It is sometimes true. It is far from reliably true. And understanding why requires understanding what the stars actually measure, how the measurement system has been systematically gamed for over a decade, and what the rating obscures entirely.
This is not an argument against using star ratings. It is an argument for using them correctly — which means understanding them well enough to know when they are telling you something real and when they are telling you something a facility wants you to believe.
Where the stars come from
CMS introduced the Five-Star Quality Rating System in December 2008, attached to its Nursing Home Compare website. The goal was laudable: give families a simple, standardized way to compare nursing home quality. The system is built from three separate domain ratings that are combined into an overall score.
The first domain is health inspections. This measures the number, severity, and scope of deficiencies cited during state survey agency inspections — the unannounced annual visits where inspectors spend days walking a facility, reviewing records, observing care, and interviewing residents and staff. Inspections also capture complaint investigations triggered when residents, families, or staff report concerns. This is the most independently verified component of the rating: the numbers come from surveyors, not the facility.
The second domain is staffing. This measures registered nurse hours per resident day and total nursing hours per resident day. Since 2016, these figures have been drawn from the Payroll-Based Journal (PBJ), which requires facilities to submit actual payroll data rather than self-reported estimates. This was a meaningful improvement — prior to PBJ, facilities could and did report staffing levels that bore limited resemblance to reality.
The third domain is quality measures: 15 clinical metrics including pressure ulcer rates, fall rates with major injury, antipsychotic medication use, pain prevalence, urinary tract infection rates, catheter use, unplanned weight loss, and others. These are drawn from the Minimum Data Set — clinical assessments that facility staff enter into the federal system during routine care. No independent auditor reviews whether the MDS accurately reflects what is happening to residents. The facility tells CMS what is happening to residents, and CMS reports it.
That third domain is where most of the gaming lives.
The inflation that appeared in plain sight
When CMS introduced quality measure public reporting, facilities faced a straightforward incentive: improve the numbers that CMS measures, whether by improving the underlying care or by improving the documentation. For facilities where the underlying care is difficult or expensive to improve, documentation offers the path of least resistance.
The results were visible almost immediately. In the first four years of Five-Star public reporting, the share of facilities receiving 4- or 5-star ratings on their self-reported quality measures rose from 10.1% to 34.8% — a tripling in just four years. Simultaneously, rates on independently inspected domains moved far more modestly. The gap between where facilities said they were performing and where independent surveyors found them performing steadily widened.
By the time researchers published systematic analyses, the pattern was unambiguous. The Abt Associates evaluation found the quality measure rating inflation was driven primarily by documentation changes, not care improvements. The Center for Medicare Advocacy analyzed facilities that held 4- or 5-star quality measure ratings alongside 1- or 2-star inspection ratings — a profile that should be nearly impossible if both measurements were accurate — and found them to be common. A facility can be routinely cited for serious care failures during independent inspections while simultaneously reporting exemplary performance on every self-assessed clinical metric.
This divergence is not random noise. It is a coherent signal about which domain the facility has invested in improving.
The inspection domain is not clean either
Health inspections are the most reliable component of the Five-Star system, but they have their own documented limitations.
The most significant is state variation. Inspection ratings are graded on a curve — CMS adjusts scores relative to each state's average, so a facility is rated against other facilities in its state, not against a national standard. A facility that would receive a 2-star inspection rating in a state with strong survey enforcement might receive a 4-star rating in a state with historically weak enforcement. The star you see on a profile does not mean the same thing in every state.
Survey cycles introduce timing effects. Nursing homes receive their annual unannounced inspection somewhere in a 9- to 15-month window. Facilities that have experienced recent leadership, staffing, or ownership changes may be surveyed before the consequences of those changes become visible in resident outcomes. Conversely, a facility that had a bad survey year and made genuine improvements may still carry the rating legacy of its worst period.
Survey frequency declined significantly during and after COVID-19. The federal government suspended non-emergency surveys for much of 2020 and 2021, creating substantial gaps in the inspection record for a period when nursing home conditions were, in many facilities, at their worst. The inspection ratings that appear in current profiles for some facilities may reflect surveys conducted before and after the pandemic's most damaging period, with limited visibility into what happened during it.
Complaint investigations are reactive, not proactive. They depend on someone — a resident, a family member, a staff member — knowing how to file a complaint and being willing to do so. Residents with dementia or limited English proficiency, or residents who fear retaliation, are systematically less likely to generate complaint investigations regardless of what is happening to them.
What the rating does not measure at all
Even if you accept the Five-Star system at face value — even if you set aside the gaming and the state variation and the survey gaps — there is a substantial category of information that the system simply does not contain.
There is no patient experience or satisfaction component. The 15 quality measures are all clinical and functional metrics assessed by facility staff. There is no systematic measure of whether residents feel respected, whether they have meaningful choices about their daily lives, whether they are lonely, whether they feel safe, whether staff know their names and their stories. The Pioneer Network, Eden Alternative, and Green House models have spent decades arguing that these dimensions of experience are central to quality of life in long-term care. The Five-Star system does not measure any of them.
There is no measure of how the facility performs when something goes wrong. Every nursing home will eventually face a medical emergency, a behavioral crisis, a fall, a medication error, a norovirus outbreak. How a facility responds — whether it communicates transparently with families, whether it escalates appropriately, whether it learns and adjusts — is perhaps the most important dimension of quality for a family placing a loved one in the facility's care. It is unmeasured.
There is no measure of financial structure or ownership stability. As covered elsewhere on NursingHomeIQ, ownership type — specifically for-profit versus nonprofit, and the presence of private equity ownership — is one of the most reliable predictors of care quality across the research literature. A facility can hold 4 stars while being owned by a leveraged buyout fund that has reduced RN staffing to minimum compliance levels and is extracting 15% of revenue through related-party management and real estate fees. None of that appears in the rating.
There is no measure of what happens at night, on weekends, and on holidays. Staffing is reported as an average over time. A facility can achieve an adequate average by staffing heavily on weekday daytimes — when state surveyors are most likely to be present — while operating with skeletal crews on Saturday nights. Weekend and overnight staffing patterns are where the gap between adequate and inadequate care is widest, and they are the least visible in any published metric.
There is no measure of care for specific populations. A facility might perform very well for short-stay post-acute rehabilitation patients and very poorly for long-stay dementia residents, or vice versa. The aggregate rating blurs these distinctions entirely.
The Special Focus Facility problem
Perhaps the sharpest illustration of the Five-Star system's limitations is what happens when you look at the facilities CMS itself has identified as its worst performers.
CMS maintains a Special Focus Facility (SFF) list — a designation for nursing homes with a history of serious quality problems, persistent deficiencies, and failure to correct cited violations. Being on the SFF list is the closest thing the federal system has to a failing designation. These are the facilities CMS has determined require intensive oversight and scrutiny.
A substantial share of SFF facilities hold 4- or 5-star overall ratings. Not 1-star. Not even 2-star. Four and five stars. The reason is structural: the overall rating is a composite, and a facility that has learned to generate excellent self-reported quality measure scores can achieve a high overall rating even while failing repeated independent inspections. The composite obscures the most important signal.
This is not an edge case. The Center for Medicare Advocacy has documented this pattern repeatedly. It means a family could use the Five-Star rating system exactly as CMS intends — search for high-rated facilities near them, tour a 4-star facility, feel reassured — and end up placing a loved one in a facility that CMS's own enforcement division has flagged as a persistent problem.
How to use star ratings without being misled by them
None of this means the Five-Star system is worthless. Used correctly, it contains real information.
The health inspection rating is the most meaningful single number in the system. It is the component least susceptible to facility manipulation, because the surveyors generating the data are not facility employees. A 1-star inspection rating is a serious warning sign. A consistent pattern of serious deficiency citations — infections, medication errors, abuse, resident harm — is a coherent signal even accounting for state-level variation. The inspection domain deserves more weight than the overall composite suggests.
The staffing numbers, since the shift to PBJ payroll data, are more reliable than the quality measures. Specifically, RN hours per resident day is the metric with the strongest empirical connection to clinical outcomes across decades of research. Look at it directly rather than accepting the staffing star as a summary. The difference between 0.4 RN hours per resident day and 0.9 RN hours per resident day is not a star difference. It is a fundamentally different model of care.
Staff turnover, now reported by CMS as a supplementary metric, is arguably more predictive than any single domain in the Five-Star system. Annual CNA turnover of 78% versus 35% — a difference invisible in the star rating — predicts abuse citations, inspection outcomes, and continuity of care far better than self-reported quality measures. Look for it explicitly.
The self-reported quality measures should be read skeptically, particularly in isolation. What makes them useful is comparison with inspection results. A facility with excellent self-reported quality measures and poor inspection ratings is a facility that has learned to report rather than improve. A facility with modest quality measure ratings but strong inspection results may be the more honest signal — surveyors found less to cite because there is genuinely less to cite.
Ownership type and chain affiliation, which do not appear in the Five-Star system at all, should inform how much you trust the reported numbers. For-profit chain-affiliated facilities have strong structural incentives to invest in documentation improvement. Nonprofit and government facilities have less incentive to game metrics and, on average, fewer deficiencies and better staffing relative to what their star ratings suggest. The same 3-star rating means something different depending on who owns the building.
The rating system CMS built and the one families need
The Five-Star system was designed to be simple. Simple is politically achievable — it produces a deliverable, a website, a metric that Congress and the press can point to. Simple is also easily gamed by an industry that is large, well-organized, financially motivated, and far more sophisticated about the measurement system than the typical family conducting a hurried search under difficult circumstances.
The fundamental problem is that the system asks facilities to report their own quality, attaches public consequences to what they report, and then presents the self-reports alongside independent assessments as though they carry equal weight. They do not. A pressure ulcer rate that a facility enters into the Minimum Data Set is not the same kind of data as a deficiency citation that a state surveyor issues after observing a resident with a Stage 3 wound.
Families deserve to know the difference. They deserve to understand that the overall star rating on a nursing home profile is a composite of one independently verified domain, one mostly-verified domain, and one systematically gamed domain — and that the composite can tell a story quite different from any of its parts.
The star is a starting point. What it hides is often more important than what it shows.
NursingHomeIQ displays health inspection ratings, staffing hours, and staff turnover alongside the overall CMS composite — so you can see each signal separately, not just the number CMS averages them into. Search any facility to see the full picture.