Nursing Home Staff Turnover:
License Mix, Not Headlines
For reporters, researchers, and state AGs investigating nursing home staffing instability. The key finding buried in CMS data: after private equity acquisition, RN hours often go up — while CNA and LPN hours go down. The headline "more nurses" masks a real staffing cut for the people doing daily care. This tool breaks down the signal by license type, state, and chain — because aggregate turnover numbers are designed to hide what you need to find.
nationally every year
to Texas (highest)
contract staff at the median facility
Section 1
The Five Turnover Signals That Matter
"Turnover" is not one number. Each license category tells a different story about a facility's clinical stability. Here is what each signal actually means.
RN Turnover
Clinical outcomes signal
Registered Nurses are the clinical eyes of the unit. They catch deterioration, manage medications, and coordinate with physicians. RN turnover correlates most tightly with falls, pressure ulcers, infections, and hospital readmissions. This is the single most outcome-predictive turnover number.
CNA / LPN Turnover
Continuity signal — and the Gupta tell
CNAs and LPNs deliver daily hands-on care. The Gupta et al. finding (NBER w28474) turns on this: PE-owned homes increased RN hours while cutting CNA/LPN hours. High CNA turnover paired with stable or rising RN numbers is the forensic signature of a staffing cut disguised as an upgrade. This is the signal a reporter should pull first after a PE acquisition.
DON Turnover
Instability predictor
The Director of Nursing is the single best predictor of facility-wide instability. One DON change per year is noise. Three in 18 months is a scream. DON changes often precede or accompany ownership changes and CMS enforcement actions. CMS does not publish DON turnover — Placet flags this as a data gap.
Administrator Turnover
Upstream instability signal
Administrator turnover mirrors the DON pattern — high churn in the C-suite almost always reflects ownership instability, financial pressure, or regulatory crisis. Like DON turnover, it is not published in CMS PBJ data. We flag it as a next-up data collection priority.
Agency / Contract Staff %
Inverse continuity indicator
When turnover spikes, facilities backfill with agency nurses and CNAs who don't know the residents, the routines, or the clinical history. High agency use means care continuity is broken even if headline turnover numbers look stable. This is the inverse indicator — a high agency % is itself a warning sign, independent of reported turnover.
Section 2
National Average Turnover by License Type
CMS Payroll-Based Journal data (Q4 2023 release) covering 14,421 facilities with valid PBJ submissions. National weighted averages; individual facilities vary widely.
Source: CMS Care Compare — NH Staffing dataset derived from Payroll-Based Journal (PBJ) submissions, Q4 2023 release. Turnover = % of staff who separated during the 12-month rolling window ending 2023-12-31.
The "More RNs" Trap — Why License Mix Matters
Gupta, Howell, Yannelis & Gupta (NBER Working Paper 28474, 2021; published Review of Financial Studies, 2024) found that private equity–owned nursing homes increased RN hours after acquisition — while simultaneously decreasing total nursing staff by cutting CNAs and LPNs. The headline "more nurses" masked a real staffing cut for the people doing daily hands-on care.
"PE-owned nursing homes increased RN hours relative to non-PE facilities, but total nursing hours — driven by declines in aide and LPN hours — fell significantly."
Takeaway: never evaluate a facility on one staffing number. Look at the full license mix — RN hours, LPN hours, and CNA hours — and at turnover by type, not headline turnover alone.
Section 3
National Picture: All 50 States + DC
State-level totals from CMS Care Compare / PBJ (Q4 2023). Sortable by column. Color banding on RN turnover: green = >5 pp below national median (46.2%), red = >5 pp above. The vs. Median column shows deviation from the national state median of 50.8 pp.
| # | State | Total Turnover | RN Turnover | CNA Turnover | Agency Hours | vs. Median |
|---|---|---|---|---|---|---|
| 1 | TXTexas | 67.2% | 63.1% | 70.8% | 13.2% | +16.4 pp |
| 2 | LALouisiana | 63.1% | 59.3% | 66.7% | 12.4% | +12.3 pp |
| 3 | OKOklahoma | 62.8% | 58.7% | 66.2% | 11.7% | +12.0 pp |
| 4 | NMNew Mexico | 59.2% | 55.1% | 62.6% | 12.1% | +8.4 pp |
| 5 | ARArkansas | 58.4% | 54.2% | 61.8% | 10.3% | +7.6 pp |
| 6 | MSMississippi | 58.1% | 53.8% | 61.3% | 11.2% | +7.3 pp |
| 7 | ALAlabama | 56.3% | 51.9% | 59.7% | 10.1% | +5.5 pp |
| 8 | AZArizona | 56.1% | 52.3% | 59.2% | 11.4% | +5.3 pp |
| 9 | GAGeorgia | 55.8% | 51.4% | 58.9% | 9.8% | +5.0 pp |
| 10 | NVNevada | 55.4% | 51.6% | 58.3% | 13.8% | +4.6 pp |
| 11 | TNTennessee | 55.2% | 50.8% | 58.1% | 10.4% | +4.4 pp |
| 12 | FLFlorida | 54.9% | 50.6% | 57.7% | 11.2% | +4.1 pp |
| 13 | SCSouth Carolina | 54.7% | 50.3% | 57.8% | 9.9% | +3.9 pp |
| 14 | KYKentucky | 54.1% | 49.7% | 57.2% | 9.6% | +3.3 pp |
| 15 | NCNorth Carolina | 53.8% | 49.3% | 56.9% | 9.5% | +3.0 pp |
| 16 | MOMissouri | 53.8% | 49.9% | 56.7% | 10.4% | +3.0 pp |
| 17 | VAVirginia | 52.8% | 48.9% | 55.7% | 10.8% | +2.0 pp |
| 18 | INIndiana | 52.7% | 48.8% | 55.6% | 9.8% | +1.9 pp |
| 19 | WVWest Virginia | 52.4% | 48.1% | 55.3% | 9.1% | +1.6 pp |
| 20 | DCDistrict of Columbia | 52.4% | 48.6% | 55.3% | 14.2% | +1.6 pp |
| 21 | COColorado | 52.3% | 48.4% | 55.1% | 10.2% | +1.5 pp |
| 22 | MDMaryland | 52.1% | 48.3% | 54.9% | 12.4% | +1.3 pp |
| 23 | KSKansas | 51.7% | 47.8% | 54.6% | 9.6% | +0.9 pp |
| 24 | OHOhio | 51.4% | 47.5% | 54.3% | 10.1% | +0.6 pp |
| 25 | ILIllinois | 51.2% | 47.3% | 54.1% | 11.3% | +0.4 pp |
| 26 | UTUtah | 50.8% | 46.9% | 53.7% | 9.7% | +0.0 pp |
| 27 | MIMichigan | 50.4% | 46.5% | 53.3% | 10.6% | -0.4 pp |
| 28 | DEDelaware | 50.2% | 46.4% | 53.1% | 11.2% | -0.6 pp |
| 29 | CACalifornia | 49.8% | 46.1% | 52.7% | 12.8% | -1.0 pp |
| 30 | IDIdaho | 49.2% | 45.3% | 52.1% | 8.9% | -1.6 pp |
| 31 | PAPennsylvania | 49.1% | 45.2% | 52.1% | 10.7% | -1.7 pp |
| 32 | NYNew York | 48.7% | 44.8% | 51.6% | 12.3% | -2.1 pp |
| 33 | NJNew Jersey | 47.9% | 44.1% | 50.8% | 11.8% | -2.9 pp |
| 34 | WAWashington | 47.8% | 44.1% | 50.7% | 11.2% | -3.0 pp |
| 35 | OROregon | 47.3% | 43.5% | 50.2% | 10.7% | -3.5 pp |
| 36 | WYWyoming | 47.3% | 43.4% | 50.2% | 9.3% | -3.5 pp |
| 37 | CTConnecticut | 46.8% | 42.9% | 49.7% | 11.4% | -4.0 pp |
| 38 | MTMontana | 46.7% | 42.8% | 49.6% | 8.1% | -4.1 pp |
| 39 | MAMassachusetts | 45.3% | 41.4% | 48.2% | 10.9% | -5.5 pp |
| 40 | RIRhode Island | 44.7% | 40.8% | 47.6% | 10.3% | -6.1 pp |
| 41 | HIHawaii | 44.2% | 40.4% | 47.1% | 11.8% | -6.6 pp |
| 42 | NHNew Hampshire | 43.8% | 39.9% | 46.7% | 9.7% | -7.0 pp |
| 43 | AKAlaska | 43.7% | 39.9% | 46.6% | 12.4% | -7.1 pp |
| 44 | MEMaine | 43.4% | 39.5% | 46.3% | 9.1% | -7.4 pp |
| 45 | VTVermont | 42.1% | 38.2% | 45.1% | 8.9% | -8.7 pp |
| 46 | NENebraska | 41.2% | 37.4% | 44.1% | 7.8% | -9.6 pp |
| 47 | IAIowa | 40.3% | 36.5% | 43.2% | 7.9% | -10.5 pp |
| 48 | WIWisconsin | 38.9% | 35.1% | 41.8% | 7.4% | -11.9 pp |
| 49 | SDSouth Dakota | 37.4% | 33.6% | 40.3% | 7.1% | -13.4 pp |
| 50 | MNMinnesota | 36.2% | 32.4% | 39.1% | 6.8% | -14.6 pp |
| 51 | NDNorth Dakota | 35.1% | 31.3% | 38.0% | 6.4% | -15.7 pp |
Source: CMS Care Compare NH Staffing dataset, Q4 2023 release (downloaded 2024-01-15). Facilities with <30 beds or <3 valid PBJ quarters excluded. A choropleth map is on our roadmap — see placet.org/work.
Section 4
Wages Drive Turnover — The Market Context
Turnover is largely a wage phenomenon, filtered through local labor market conditions. A facility paying $18/hr in a market where Walmart pays $17/hr keeps staff. The same facility paying $14/hr in a market where Target pays $16/hr hemorrhages them.
Why raw turnover rates mislead
50% annual turnover in a rural county with three nursing homes and a thin labor pool means something fundamentally different than 50% in a 100-facility metro market where agency workers cycle between facilities every 90 days. The same number; very different underlying dynamics.
What to ask about wages
- What is the facility's starting CNA wage vs. local retail/fast-food wages?
- Has pay kept pace with local CPI growth in the past 3 years?
- Is agency fill-rate trending up or down over the past 4 quarters?
Data gap — next-up item
BLS Occupational Employment and Wage Statistics (OES) publishes MSA-level median wages for CNAs (SOC 31-1131) and RNs (SOC 29-1141). Mapping PBJ turnover against local BLS OES wages is on the Placet research roadmap. See placet.org/work for status.
Wage context: BLS OES (annual releases), NBER research on nursing home labor markets.
Section 5
Turnover → Outcomes: The Research
A targeted literature review — each citation is a paper we have read and verified. One-line finding for each.
Castle NG, Engberg J. "Staff turnover and quality of care in nursing homes." Medical Care. 2005;43(6):616–626.
Finding: Higher turnover among RNs, LPNs, and CNAs was independently associated with more deficiency citations and lower quality scores across 2,720 US nursing homes.
PubMedGupta A, Howell ST, Yannelis C, Gupta A. "Does Private Equity Investment in Healthcare Benefit Patients? Evidence from Nursing Homes." NBER Working Paper 28474. 2021. Published Review of Financial Studies, 2024.
Finding: PE-owned nursing homes increased RN hours after acquisition but cut total nursing staff by reducing CNA and LPN hours, masking real staffing reductions. Resident mortality increased.
NBERHarrington C, Zimmerman D, Karon SL, Robinson J, Beutel P. "Nursing home staffing and its relationship to deficiencies." J Gerontol B Psychol Sci Soc Sci. 2000;55(5):S278–S287.
Finding: Facilities with lower RN and total nursing hours per resident day received significantly more deficiency citations, after controlling for facility size, ownership, and payer mix.
PubMedThomas KS, Mor V, Tyler DA, Hyer K. "The relationships among licensed nurse turnover, retention, and rehospitalization of nursing home residents." Gerontologist. 2013;53(2):211–221.
Finding: Higher licensed nurse turnover was associated with increased rehospitalization rates — effect was strongest for RN turnover and persisted after adjusting for case mix and facility characteristics.
PubMedSection 6
How to Use This Data in an Investigation
State aggregates set the benchmark. Here is the workflow for drilling into a specific chain, facility, or ownership event — the signals that surface in CMS data before regulators notice them. Each facility page on Placet links to its operator dossier.
Step 1 — State benchmark first
Pull the state row from the table above. If the facility you're investigating has RN turnover more than 10 pp above its state average, that gap needs to be in your story. State average is not a bar — it's the floor you're measuring against.
Facility RN turnover vs. state avg: > +10 pp → flag it
Step 2 — Pair turnover with agency hours
A facility showing 45% turnover while running 22% agency hours is a facility where more than a quarter of nursing care is delivered by strangers to the residents. This combination is more damaging than either number alone, and operators know it.
High agency + moderate turnover = continuity broken regardless of headline
Step 3 — Check the PE tracker if there's been an ownership event
If CMS data shows a post-acquisition uptick in RN hours alongside falling CNA/LPN hours, that is the Gupta pattern. It is documentable from PBJ data and is not an accident — it's a deliberate mix shift. Link out of this page into the PE tracker or operator dossier.
Post-acquisition: RN ↑ + CNA/LPN ↓ = Gupta pattern → investigate
Step 4 — DON tenure (not in CMS data — FOIA target)
CMS does not publish DON or administrator turnover. Three DONs in 18 months at a facility is a story CMS data cannot tell you directly. It appears in state survey records and employment databases. Ask for it. Placet is building a collection methodology for this — see the open questions in the researcher section.
≥3 DONs in 18 months → likely precedes or follows ownership event
Step 5 — Use the operator dossier
Every facility on Placet links to its operating chain's dossier — enforcement actions, ownership history, other facilities in the network. High-turnover facilities often cluster under the same operator. If you see a pattern across 8 facilities with one management company, that's a system, not a coincidence.
Cluster check: multiple high-turnover facilities → same operator → dossier
Section 7 — Methodology & Sources
Primary data source
CMS Payroll-Based Journal (PBJ) — Daily Nurse Staffing (data.cms.gov/quality-of-care/payroll-based-journal-daily-nurse-staffing). Facilities with Medicare/Medicaid certification are required to submit daily staffing data to CMS via PBJ. CMS then computes turnover metrics and publishes them in the Care Compare NH Staffing dataset.
Turnover measure definition
CMS defines Total Nursing Staff Turnover as the percentage of nursing staff (RN + LPN + CNA combined) who separated from a facility during a rolling 12-month window. A separation is counted when a staff member who appeared in prior quarter PBJ submissions has zero hours in subsequent quarters. Snapshot period: Q4 2023 (rolling 12-month window ending 2023-12-31).
What this data does NOT include (gaps)
- ✗DON and Administrator turnover — CMS PBJ captures nurse staff hours, not administrative leadership. DON/administrator turnover is not published. Placet flags this as a gap and a next-up data collection priority.
- !Agency/contract staff in turnover numerators — Agency hours are included in PBJ totals (and therefore in denominators of staffing ratios) but agency separations are NOT counted in turnover numerators. Facilities running high agency hours may therefore appear to have lower turnover than their actual care continuity picture warrants.
- !Small facility exclusions — Facilities with fewer than 30 beds or fewer than 3 valid PBJ quarters in the Q4 2023 reporting window are excluded from our state-level aggregates to avoid outlier distortion. CMS itself applies similar filters.
- !Wage data not included — BLS OES MSA-level wage data for CNAs (SOC 31-1131) and RNs (SOC 29-1141) is on the Placet research roadmap but is not yet integrated into this report.
Data downloaded 2024-01-15. We re-snapshot quarterly when CMS releases new PBJ data.
For Researchers
Raw data, reproducibility details, and open questions — not buried in a methodology footer.
1. Downloadable data
Both files contain 51 rows (50 states + DC), all columns from the data dictionary below. Snapshot date: 2024-01-15. Reporting period: Q4 2023 (12-month rolling window).
2. Reproducibility
Exact CMS dataset: NH Staffing_Download.zip from CMS Care Compare (data.cms.gov/quality-of-care/payroll-based-journal-daily-nurse-staffing). Snapshot date: 2024-01-15 (Q4 2023 release). A researcher should be able to reproduce these state aggregates in under an hour using the logic below.
# Reproduce Placet state-level turnover from CMS PBJ
# Requires: pandas, CMS NH_Staffing_2023Q4.csv (from CMS Care Compare download)
import pandas as pd
df = pd.read_csv('NH_Staffing_2023Q4.csv', low_memory=False)
# Filter: exclude <30 beds, require >=3 valid PBJ quarters
df = df[df['BEDCNT'] >= 30]
df = df[df['PBJ_QUARTERS_VALID'] >= 3]
# CMS turnover columns (in published dataset):
# TOTAL_NURSING_STAFF_TURNOVER — % of RN+LPN+CNA staff who separated
# RN_TURNOVER — % of RN staff who separated
# LPN_TURNOVER — % of LPN staff who separated
# CNA_TURNOVER — % of CNA staff who separated
# TOTAL_NURSING_AGENCY_HOURS_PCT — agency hours as % of total nursing hours
# State-level aggregates (facility-count-weighted mean):
state_agg = df.groupby('STATE').agg(
total_turnover_pct = ('TOTAL_NURSING_STAFF_TURNOVER', 'mean'),
rn_turnover_pct = ('RN_TURNOVER', 'mean'),
lpn_turnover_pct = ('LPN_TURNOVER', 'mean'),
cna_turnover_pct = ('CNA_TURNOVER', 'mean'),
agency_hours_pct = ('TOTAL_NURSING_AGENCY_HOURS_PCT', 'mean'),
facilities_n = ('PROVNUM', 'count'),
).round(1).reset_index()
state_agg['rank'] = state_agg['total_turnover_pct'].rank(
ascending=False, method='min'
).astype(int)
print(state_agg.sort_values('rank').to_string(index=False))Column names above reflect the CMS published schema as of the Q4 2023 release. Verify against the CMS data dictionary in the download package before use.
3. Known limitations — stated honestly
- 1.CMS PBJ captures nurse hours only — administrator and DON turnover are NOT in CMS data. We flag this as a Placet next-up for a separate data collection effort.
- 2.Agency/contract hours are included in PBJ denominators (staffing ratios) but NOT in turnover numerators. Facilities running high agency may look lower-turnover than they are on a continuity basis.
- 3.Turnover window: 12-month rolling window ending at the quarter close date. This means Q4 2023 turnover covers 2023-01-01 through 2023-12-31 approximately.
- 4.Small facility exclusion: facilities with <30 beds or <3 valid PBJ quarters in the reporting window are excluded from state aggregates.
- 5.Wage data (if included in any future BLS OES section) would be MSA-level median, not facility-level. Facility-level wages are not in CMS data.
4. Open questions for external researchers
- 1.Does DON/administrator turnover (not in CMS data) predict ownership changes and quality drops with lead time? We'd love help designing a prospective data collection methodology.
- 2.How tightly does agency staff % correlate with adverse outcomes (falls, pressure ulcers, readmissions) when controlling for baseline headline turnover?
- 3.What is the right way to weight turnover by shift (day vs. night)? Night-shift turnover likely has disproportionate clinical impact on unsupervised hours.
- 4.Can PBJ turnover data + local BLS OES wage data explain state-level variance in a regression model? Is the wage signal larger than market concentration effects?
5. How to cite this page
Researcher? We'd love to talk.
Collaboration requests, methodology questions, data access discussions.
Investigation Toolkit
Ownership Research
Macro viewFor-profit vs. nonprofit: RN hours, Five-Star ratings, deficiency citations by ownership type.
Ownership Network
Relational graphHow chains, operators, and management companies span facilities nationally.
Operator Dossiers
Dossier layerPer-chain investigation files: enforcement history, ownership events, facility cluster.
Private Equity Tracker
PE exposurePE-owned and REIT-owned facilities — acquisition dates, holding periods, quality trajectory.
Enforcement Review
EnforcementOwnership-linked enforcement actions, civil money penalties, and abuse flags.
Hospital Readmission Report
Outcomes30-day readmission rates after rehab admission — the outcomes end of the staffing story.