Working Paper No. 01 · October 2024
The staffing week has a shape that quarterly averages erase
The number families use to compare nursing homes is a quarterly average that treats all seven days equally. The daily data reveals a staffing week with a distinct shape — and four of its days are not what the published figure describes.
Preprint — not peer-reviewedAbstract
Using daily-resolution CMS Payroll-Based Journal data covering 10.8 million facility-days across 15,829 nursing homes (2017Q1–2020Q2), this paper shows that nurse staffing follows a consistent weekly shape invisible in published quarterly averages: coverage peaks Tuesday through Thursday, declines on Monday and Friday, and drops sharply on weekends. The quarterly average obscures a four-day reduced-staffing period affecting more than half the week. Forty-three percent of facilities meeting the 3.48 HPRD federal minimum on their quarterly average fall below it every weekend, representing approximately 391,000 residents. California, the only large state with mandatory minimums, shows a 10.9% weekend drop versus 16.7% nationally, sustained across all ten quarters.
1. The shape of the staffing week
Coverage peaks on Tuesday and Wednesday at 3.96 nurse hours per resident. Thursday holds near that level. Friday drops 3.4% from peak. Saturday drops another 14%. Sunday reaches 3.24 — an 18.2% decline from Wednesday. Three days carry full coverage. Four days do not. The quarterly average treats all seven as identical.
Median HPRD by day of week — grand average across 10 quarters
10.8 million facility-days. Every bar is real data.
For a typical facility with 78 residents, the Wednesday-to-Sunday difference is 56 nurse hours — six eight-hour shifts gone from the building. Every week, at nearly every facility in the country.
2. The gap does not close
Across 10 consecutive quarters, the weekday-weekend gap held within a 1.5 percentage point band. CMS has had this daily data since 2016. The gap did not narrow.
Weekday vs weekend HPRD — national medians, 2017–2020
Dashed line: 3.48 HPRD proposed federal minimum
3. The pattern is universal
Sixty-one percent of facility-quarters show a weekend drop between 10% and 20%. Only 0.4% show weekends staffed equal to or above weekdays. The gap appears in for-profit and nonprofit facilities, in 1-star and 5-star facilities, in large chains and independents.
Distribution of weekend staffing drops — 146,241 facility-quarters
Share of facility-quarters in each drop range
4. What the federal minimum looks like measured by day
CMS finalized a 3.48 HPRD minimum in April 2024. Measured against daily data rather than the quarterly average, the compliance picture changes completely — and the change is stable across every quarter in the dataset.
Facilities meeting the 3.48 HPRD standard: weekday vs weekend
Gap holds near 38 percentage points across all 10 quarters
5. Everyone cuts weekends. The floor is what differs.
All five CMS staffing star tiers show nearly identical percentage drops on weekends (~16%). But a 16% cut from a 1-star baseline produces a weekend floor of 2.91 HPRD. From a 5-star baseline it produces 3.96. The bottom 10% of 1-star facilities drop to 2.45 on weekends.
Weekday HPRD, weekend HPRD, and bottom-10% weekend floor by star rating
The percentage drop is uniform. The absolute floor is not.
6. What a mandate does to the weekend floor
California's weekend drop is 10.9% versus 16.7% for every other state — nearly six percentage points, sustained across all 10 quarters. California is the only large state with mandatory minimum staffing ratios. For a 78-resident facility, the California-to-Alabama difference is approximately 22 nurse hours every weekend day.
California vs all other states: weekend HPRD by quarter
California's floor stays consistently above the national figure for three straight years
7. The geographic pattern
Weekend drops range from 10.9% (California) to 20.2% (Alabama and Mississippi). Hover over a state to see its figure.
Median weekend staffing drop by state
Darker = larger drop. States with fewer than 50 facilities excluded.
8. The metric that improved during the crisis
In 2020 Q2, national HPRD rose 4% as COVID emptied facilities. Resident census fell 9.1%. Same staff, fewer residents, better-looking ratio — during the worst quarter nursing homes had faced in decades.
HPRD rising as census falls: the COVID Q2 2020 artifact
The staffing metric improved as conditions worsened
What this analysis does not tell you
Whether the weekend drop causes harm. This paper measures a staffing pattern. Connecting it to patient outcomes requires inspection deficiency and Medicare claims data.
Why the drop exists. Scheduling norms, pay differentials, weekend labor supply, and ownership structure are all plausible drivers. This data cannot separate them.
Whether the pattern held after mid-2020. This dataset ends in 2020 Q2. Post-COVID staffing dynamics may have changed the picture.
Whether the California effect is causal. Consistent with mandatory minimums compressing the weekend reduction — but observational data cannot establish causation.
Source. CMS Payroll-Based Journal Public Use Files, nurse staffing, quarterly. Ten quarters: 2017Q1–Q4, 2018Q4, 2019Q1–Q4, 2020Q2. Retrieved October 2024.
Coverage. 10,832,137 facility-days, 15,829 unique facilities, all states plus Puerto Rico.
HPRD. Total nurse hours (RN, LPN, CNA, Med Aide, Nurse Aide in Training) divided by MDS census for each day. Median across all facilities per day of week per quarter. Grand averages across all 10 quarters.
Limitation. PBJ captures only paid hours. 22 quarterly files whose URLs returned 404 errors at time of retrieval are excluded.
All code, the dated dataset snapshot, and the Python environment are in the repository. Reproduces from committed data without re-downloading from CMS.
zatara-research-data/01-weekend-staffing-cliff