How CMS counts turnover, and who it leaves out

Where the number comes from

Nursing homes send CMS their staffing hours every quarter through the Payroll-Based Journal, or PBJ. The Affordable Care Act requires it, and the hours have to come from payroll and other records that can be audited, including agency and contract staff.

CMS turns those hours into three turnover measures and publishes them on Care Compare: the share of nursing staff who left over a 12-month period, the share of RNs who left, and the number of administrators who left. Nursing staff here means RNs, LPNs, certified nurse aides, aides in training and medication aides.

The same three measures count toward your staffing star rating. We explain how in Turnover and your star rating.

What counts as leaving

CMS doesn’t use the termination dates you report. It looks at the days each person worked. Someone counts as having left once they go 90 days in a row without working a shift.

That way every building is measured the same way, whatever its own records say. Someone who comes back after more than 90 days away is counted as a new employee.

Who counts at all

Only people who work at least 120 hours in their first 90 days count at all. That’s about three weeks of full-time work.

CMS says the rule leaves out people who work infrequently, like someone who occasionally covers a shift. Agency staff count if they reach 120 hours.

A new hire who leaves before reaching 120 hours never shows up in your published rate.

Who it leaves out

That leaves out a lot of people. A 2026 study in Health Affairs Scholar, using CMS’s own payroll records, found that by 2022-2023, 45% of nursing hires didn’t reach 120 hours in their first 90 days. Counting everyone, turnover ran about 19 percentage points higher than under CMS’s definition, and 16 points higher for RNs.

Counting everyone also changed how buildings compared. Only 30% of nursing homes landed in the same tenth of the national ranking under both definitions, and buildings that relied more on contract staff were more likely to rank higher for turnover once everyone was counted.

The published rate still tells you something: the study found both versions of turnover were linked to quality measures in nearly the same way. It just doesn’t see everyone.

Why it matters for your building

Those early departures still cost recruiting, screening and orientation. So a building can look steady on Care Compare while it hires for the same jobs again and again.

New hires also carry a real share of the work. In an average week, about 15% of nursing hours are worked by someone hired in the last 90 days.

And the published rate always looks back. One year of turnover takes six quarters of payroll data to calculate, including the quarter after the year ends, and each quarter’s data are due 45 days after it closes.

How to see what your rate misses

Your own payroll and HR records hold what CMS leaves out. A few counts, by building and by role, show it:

• How many people you hired in the last year, and how many are still with you.
• How many left in their first few weeks, before they’d have reached 120 hours.
• How many of your nursing hours come from agency and contract staff.

Look at them every month, next to your Care Compare rate. Where your own count runs higher, look first at the people who left in their first weeks.

Then talk with new hires early, while there’s still time to help. Our guide to stay conversations has questions to start with.