Yes — agency and contract clinicians count toward your nursing home’s turnover rate, the same as employees. CMS puts it plainly: both are included if they work enough hours to be eligible. What decides it isn’t who signs their paycheck. It’s how many hours they worked and whether they came back.
That single fact changes how a staffing plan should be built, and almost nobody running a building has been told it directly. Here’s exactly how the calculation works, and where the leverage is.
How does CMS calculate nursing home staff turnover?
Not from termination dates. CMS never sees those. The turnover measures are built entirely from your Payroll-Based Journal (PBJ) submissions, using gaps in days worked and the employee system IDs you report. Someone is counted as having left when their ID stops appearing for long enough.
From the CMS Five-Star Technical Users’ Guide (July 2026 edition — check the cover date, because older copies of CMS methodology documents still circulate widely), the calculation requires six consecutive quarters of PBJ data and works in two halves:
| What it is | The rule | |
|---|---|---|
| Denominator | Who is eligible to count at all | Anyone who works at least 120 hours in a 90-day window, starting from their first observed workday |
| Numerator | Who counts as having left | Eligible people who then go 90 or more consecutive days without working at all |
Turnover is dated to the last shift worked before the gap began — not to the day you noticed they were gone.
Do agency and contract staff count toward nursing home turnover?
Yes. The guide is explicit: “both regular employees and agency staff are included in the turnover measure if they work sufficient hours to be eligible for the denominator.” There is no separate treatment, no exemption, and no way to file contract hours so they sit outside the calculation.
What does separate people is the 120-hour threshold. CMS built it specifically to exclude, in its words, “individuals who work infrequently (e.g., occasionally covering shifts at a nursing home).”
So the line runs like this:
- Under 120 hours in a 90-day window — roughly fewer than ten twelve-hour shifts — the clinician never enters the calculation. They can stop coming entirely and your turnover rate doesn’t move.
- At or above 120 hours, they’re in the denominator, counted exactly like a staff nurse or aide. If they then lapse for 90 days, that’s a separation on your public rating.
The counterintuitive part
Read those two bullets together and you get a result most administrators find backwards: your best contract clinicians are the only ones who can hurt your turnover rate.
The traveler who covered three shifts in March and vanished? Invisible. The per diem CNA who has been picking up two shifts a week for four months, who knows your residents by name, who your charge nurse requests — she is squarely in the denominator. If she stops coming and 90 days pass, she is recorded as turnover, and she costs you exactly what a departing employee costs you.
This is why “we only use agency for coverage, so it doesn’t affect our numbers” is wrong in a specific and expensive way. Occasional agency use genuinely doesn’t touch turnover. Reliable contract coverage does — which means the staffing arrangement you’re most satisfied with is the one carrying rating risk you probably haven’t been tracking.
What happens when a clinician comes back after a gap?
If they return before day 90, nothing happens. No gap of 90 consecutive days ever forms, so no turnover event is ever recorded. The measure simply doesn’t fire.
If they return after day 90, two things happen, and the second one surprises people. The separation has already been counted. And CMS treats the return as a fresh start: “individuals who return to the nursing home after a gap of more than 90 days can have multiple ’employment spells’ used in the turnover calculation… Essentially, they are treated as new employees.”
Because the formula divides spells that ended in turnover by total eligible spells, a clinician who lapses, returns, works enough to re-qualify, and lapses again can register as turnover more than once. A churning contract pool can charge you for the same person twice.
The practical reading: the 90-day mark is a real deadline with a real consequence, and it’s one of the very few things in the Five-Star system you can act on in advance. A clinician at day 60 without a shift is a preventable separation. At day 91 they’re a number on Care Compare for the next twelve months.
The employee ID trap
One more mechanic worth knowing, because it can quietly destroy continuity that you actually earned.
CMS tracks people by the employee system ID you report in PBJ. If those IDs change — a payroll system migration, a new numbering scheme, or a contract-hours process that assigns fresh IDs each quarter — every returning clinician looks like a brand new person, and everyone from the prior period looks like they left.
CMS knows this happens and handles it bluntly: nursing homes showing 100% daily turnover on any day in the six-quarter window (with at least five eligible nurse staff) are excluded from the total nurse and RN turnover measures entirely. Their words: this pattern “reflects the change in the employee IDs and not actual staff turnover.”
Being excluded isn’t a reprieve — it means the measure can’t be calculated for you, and it points at a reconciliation process that isn’t tracking the same human across quarters. If you use contract labor, ask whoever reconciles those hours one question: does the same clinician carry the same ID every quarter? If the answer is no, or nobody knows, your continuity is invisible in the only data CMS reads.
How much is turnover worth in the staffing rating?
About a third of it. Of roughly 380 points available in the staffing domain, the three stability measures carry 130:
| Measure | Who’s counted (PBJ job codes) | Max points |
|---|---|---|
| Total nurse staffing turnover | RNs, LPNs/LVNs, CNAs, aides in training, med aides/techs | 50 |
| RN turnover | RN director of nursing, RNs with administrative duties, RNs | 50 |
| Administrator turnover | Nursing home administrators | 30 / 25 / 10 |
Turnover measures are scored on deciles against the national distribution — the lowest-turnover decile earns the full 50 points, the highest earns 5. And if you fail to submit staffing data for any quarter in the calculation, you receive the lowest possible score on these measures automatically.
Two exclusions are worth knowing: buildings with fewer than five eligible nurses in the denominator are excluded from the nurse turnover measures, and the administrator measure needs only one eligible administrator to apply.
How much turnover does CMS actually see?
Less than there is. Because of the 120-hour threshold, a large share of real churn never appears in the published numbers. Research in Health Affairs Scholar found that in 2022–2023, roughly 46% of total nursing hires and 41% of RN hires worked fewer than 120 hours in their first 90 days — and were therefore excluded from CMS turnover reporting entirely.
That cuts two ways. Your published turnover rate is probably better than your lived experience of the building, which is worth knowing before you conclude the number is wrong. But it also means the measure is specifically tracking your committed workforce — the people who showed up enough to matter. Losing them is exactly what the measure is designed to catch.
What you can actually do about it
Turnover is the slowest lever in the staffing rating — the measures look back twelve months, so nothing you do this quarter shows up this quarter. That’s an argument for starting now, not for deprioritizing it. Four things are within reach:
- Know who is in your denominator. Any clinician, employed or contract, who has worked 120+ hours in a 90-day window is on the list. Most buildings have never pulled this list. It is knowable from your own PBJ data.
- Track days since last shift for that group. The 90-day gap is the only Five-Star mechanic with a countdown you can see coming. A clinician at day 60 is a scheduling conversation. At day 91 they’re a rating event.
- Fix ID continuity before it becomes an exclusion. One clinician, one ID, every quarter — including contract hours.
- Choose coverage that returns. Whether a shift gets filled by the same person or a different one every time makes no difference to your hours measures, and all the difference to your stability measures.
That last point is the structural one. A building can post perfectly respectable hours per resident day while the turnover column erodes underneath it, because filling shifts and keeping people are measured separately and only one of them is about volume.
It’s also where continuity stops being a nice idea and becomes an operating requirement. On the facility side, Switch is built around clinicians returning to buildings they already know — arriving fully credentialed and current, with shifts worked at a 96% work rate, because accountability runs both ways under the strictest cancellation policy in the category. A clinician who keeps coming back is a clinician who never opens a 90-day gap.
If you want to put a dollar figure on what churn is already costing you, our CNA turnover cost calculator does the math for your building, and The Hidden Cost of CNA Turnover breaks down where the money goes. For how these measures roll up into the public rating, see how the CMS Five-Star staffing score actually works.
Frequently asked questions
Yes. CMS states that both regular employees and agency staff are included in the turnover measure if they work sufficient hours to be eligible for the denominator — at least 120 hours in a 90-day window. There is no separate treatment for contract labor.
From Payroll-Based Journal data, using gaps in days worked rather than termination dates. It requires six consecutive quarters of PBJ data, and divides employment spells that ended in turnover by total eligible employment spells. The maximum rate is 100%.
At least 120 hours during a 90-day window starting from their first observed workday — roughly ten twelve-hour shifts. CMS set this threshold to exclude individuals who work infrequently, such as occasionally covering a shift.
Ninety or more consecutive days with no hours worked at that nursing home. The turnover is dated to the last shift worked before the gap began. CMS raised this from 60 to 90 days in July 2024 so that full FMLA parental leave would not count as turnover.
The separation has already been counted, and CMS treats the return as a new employment spell — essentially a new employee. If they re-qualify and lapse again, the same person can register as turnover more than once.
About a third. Total nurse turnover and RN turnover carry 50 points each and administrator turnover carries up to 30, for roughly 130 of the 380 points available in the staffing domain.
Yes, badly. CMS tracks people by the employee system ID reported in PBJ, so changed IDs make returning staff look new. Facilities showing 100% daily turnover on any day in the six-quarter window are excluded from the nurse turnover measures entirely.
The bottom line
Your turnover rate isn’t a measure of your employees. It’s a measure of everyone who worked enough to matter and then stopped coming — employed, contract, or per diem, with no distinction between them. The threshold is 120 hours. The deadline is 90 days. Both are visible in data you already submit, and one of them is a countdown you can still act on.
See how Switch fills shifts with clinicians who come back — and what that does to the third of your staffing score that isn’t about hours.
Sign up. Download. Start today.
Switch is the easiest way to get the shifts you want at the rates you deserve—in the best facilities in your area.