Back to Articles|Published on November 21, 2026|27 min read

Docket Article

PBJ Contract Hours and Private Equity Staffing Screens

How to turn CMS Payroll-Based Journal contract hours into a cautious labor-dependency screen for nursing-home operators, with a facility-period worksheet, 2026Q2 release lag and supplier-evidence gates.

By Docket Research

PBJ Contract Hours and Private Equity Staffing Screens

Summary

  1. 01PBJ public files report paid employee and contract hours by facility, day and role. Contract hours include individually contracted workers and staff supplied through agencies.
  2. 02Calculate contract share using contract hours divided by employee plus contract hours for the same roles and dates. Calculate hours per resident-day using summed hours and summed census on the same dates with nonzero census.
  3. 03Record the release quarter, publication date and dictionary edition. Match each facility to its operator for the dates in scope and split periods at ownership changes, even when the provider number stays the same.
  4. 04Check for missing quarters, zero-census days and discrepancies between total hours and their components. Compare matched calendars to see whether contract hours recur across most weeks.
  5. 05Persistent contract use can justify a diligence question. Naming a supplier requires independent, dated evidence of the relationship and its scope. The named relationship alone leaves price and contract terms unknown.
Inside this article
  1. 01Executive Summary
  2. 02Introduction and Background
  3. 03Key Changes Affecting How the Screen Is Read
  4. 04The Contract-Hours Screen: Units, Fields and Formulas
  5. 05Implementation Considerations and Process Changes
  6. 06Data Analysis and Evidence
  7. 07Case Studies and Real-World Examples
  8. 08Implications and Future Directions
  9. 09Frequently Asked Questions (FAQs)
  10. 10Conclusion

Executive Summary

The Payroll-Based Journal (PBJ) public files give a private equity origination analyst one reliable observation about a nursing-home operator's labor model: how many paid hours each facility reported on each calendar day, split between employees and contract (including agency) staff, by job category. The nurse and non-nurse files are published as one row per facility per work day, and the nurse dictionary carries a resident census from the Minimum Data Set (MDS) plus total, employee and contract hours fields such as the one for the RN Director of Nursing [1]. The same dictionary contains no supplier, agency-identifier or wage-rate field [1]. That boundary defines what the screen can decide: whether reliance on contract labor deserves a diligence question. It cannot decide which staffing company supplied the hours, at what price, or under what contract.

The recommended worksheet computes contract share as summed contract hours divided by summed employee plus contract hours for exactly the same role and date set, and hours per resident-day as a ratio of sums, never an average of daily ratios. Published research shows why the choices matter. Agency staff supplied about 3 percent of direct care nursing hours in 2018 and 11 percent in 2022 [2], and the 2022 state-level range ran from 28.1 percent in Vermont to 2.9 percent in Arkansas [2]. A separate HRSA-funded analysis found contract certified nursing assistant (CNA) hours rising from 2 percent to 11 percent of CNA hours between 2017 and 2022 [3]. These are historical, category-level figures, not a benchmark for any target.

Release timing needs version control. CMS lists the nurse file as updated four times a year, so the release in hand should be checked against the current CMS catalog record before each pull. CMS moved PBJ submission into the Internet Quality Improvement and Evaluation System (iQIES) from August 17, 2026 [4], and stated that its September 11 update did not change PBJ policies [4]. A submission-system move is not evidence of a public-file schema change.

Operator matching and supplier attribution are separate gates. CMS's ownership files carry association and effective dates but hold only currently active enrollments, so contemporaneous operator membership must be built deliberately. Supplier attribution requires independent dated evidence of a named relationship and its scope, and even then price, exclusivity, margin and renewal remain unknown. Healthcare staffing is a roughly $39.0 billion US market in 2026 by one industry forecaster [5], so a facility's demand signal is not an addressable-revenue estimate for any one agency. The dispositions are observed dependency, uncertain due to coverage, supplier-supported for a stated period, or supplier-unresolved.

3 percentApproximate share of direct care nursing hours supplied by agency staff in 2018
11 percentShare of direct care nursing hours supplied by agency staff in 2022
49.1%Average share of nursing homes that used any direct care agency nursing staff in 2022
8.4%Decline in contract therapy staffing in the week after the 2019 payment change

Introduction and Background

This report examines one narrow question for private equity (PE) healthcare origination teams: when does a facility's use of contract labor, as reported in the CMS Payroll-Based Journal, justify a diligence question, and what additional evidence is needed before anyone names a staffing supplier? It is written for analysts assessing nursing-home operators and for those testing a healthcare-staffing or rehabilitation-services thesis. It is not guidance on submitting PBJ data, not a headcount estimator, not a quality verdict on any facility, and not a claim about any staffing agency's customers or revenue. It does not list facilities or suppliers.

PBJ exists because the Affordable Care Act, section 6106, requires nursing homes to submit direct care staffing information, including agency and contract staff, based on payroll and other auditable data [4]. Facilities submit quarterly, with submissions due by the end of the 45th calendar day after quarter end [4]. CMS's public use file (PUF) documentation says the public data are aggregated to the facility-day [6] and that facilities report hours for agency and contract staff. The nurse file and the non-nurse file are separate datasets.

For an origination analyst the appeal is plain. Contract hours are one of few labor-model variables that are public, facility-level and repeated by quarter. The risk is over-reading them. The public file is an hour count, not a ledger. CMS instructs facilities to report only paid hours [6], and its policy manual defines contract staff as individuals under contract as well as individuals who provide services through organizations that are under contract, such as an agency to provide nurses [7]. Hours from a directly contracted therapist and hours from a staffing company's nurse therefore sit in the same contract field.

Three further boundaries shape this report. First, evidence is lagged: the newest quarter lags the calendar by months, so the screen describes past operations. Second, the screen answers "how much" and "how persistent," not "who." Third, nothing here measures any facility; the worked examples are fictional and labeled as such. The PBJ screen is one input to origination work, not a conclusion.

Key Changes Affecting How the Screen Is Read

Confirm the newest public release before each pull

CMS lists the nurse file as updated four times a year; confirm the newest quarter and modification date on the CMS catalog record before each pull [8]. The release rhythm is quarterly, and publication lags the end of the period. Any worksheet therefore needs a "release pinned" column: which quarter, which publication date and which dictionary edition produced each row. Re-pulling later can change a facility's presence because CMS includes only data received by the deadline, 45 days after quarter end [4].

The iQIES migration changed submission, not the public schema

CMS launched PBJ functionality in iQIES as of August 17, 2026 [4]. The submission page also says the September 11, 2026 revisions of the policy manual and frequently asked questions include no changes to PBJ policies [4]. The iQIES release added an edit that prevents one System Employee ID from exceeding a daily hour limit, and the manual states that in no case should a single employee or system identifier have more than 22.5 hours per day [7]. These are submission-side controls. The opened public dictionary lists the same facility-day fields it listed before; the inspection supports only the narrow statement that the schema we read has no supplier or rate fields, not a claim about restricted PBJ files.

An analyst who sees a break in a facility's contract share across the third quarter of 2026 should therefore ask whether the break follows ingestion and edit changes or a real operating change. The public files cannot answer that. A submission-spec change does not establish a public-file schema change, and the reverse also holds.

The federal minimum staffing standards were repealed, shifting what an hours screen means

The 2024 federal minimum-staffing requirements are context for any hours-per-resident-day metric. A statute enacted in 2025 delays implementation, administration and enforcement of the specific provisions [9], and CMS then published an interim final rule rescinding the requirements for 24/7 registered-nurse coverage and the hours-per-resident-day minimums [10], effective February 2, 2026 [10]. The practical reading for a screen is that hours per resident-day is a descriptive staffing intensity measure, not a threshold to be passed. CMS itself states that the PBJ data cannot be used to indicate noncompliance [6]. Specialist legal review is needed for any consequential compliance interpretation.

One study found a shift from short-term to more persistent agency use

Bowblis and coauthors, using PBJ facility-day data from 2018 to 2022, found that about 3 percent of direct care nursing hours came from agency staff in 2018 and 11 percent in 2022 [2]. In 2022 49.1 percent of nursing homes used some direct care agency nursing staff [2], and the study reports that only 19.4 percent of agency-using homes used agency staff for ninety days or fewer (its Discussion words the same figure as "less than ninety days") [2]. This single study's 2018-to-2022 comparison points from short bursts toward more sustained use, which is why a screen should examine persistence and not a single quarter's share. These are historical figures about a pandemic-era shock; they are not today's contract-labor benchmark.

The PBJ public files support a disciplined, limited conclusion: for a defined set of roles and dates, a facility reported a given share of paid hours as contract, and the pattern did or did not persist across matched periods. They do not show who supplied the hours, at what price, under what contract or with what margin.

The Contract-Hours Screen: Units, Fields and Formulas

The unit of analysis is the facility-day role-hour

The unit is a role-hour for one facility on one date, not a staffing company, worker, contract or transaction. The nurse dictionary identifies the facility with a Medicare provider number, a day for reported hours, a census, and for each role a total, an employee and a contract field [1]. Treat the provider number (PROVNUM) as a six-character string: CMS warns that spreadsheet software drops leading zeros when opening comma-separated files [11], and the State Operations Manual explains that the first two digits of the certification number identify the state and the last four the facility type [12].

Choose roles deliberately

The nurse file covers eight job categories, each with total, employee and contract hours. CMS's own Five-Star staffing measures group RN hours to include the RN Director of Nursing (job code 5) and registered nurses with administrative duties [13]. For a labor-dependency thesis, keep administrative and director roles distinct from direct RN, LPN and aide roles if the thesis depends on that distinction. A contract worker, in the language of one HRSA-funded analysis, is one engaged by the nursing home under contract or through an agency [3]. Never add a "total" field to its own employee and contract subfields; the worksheet sums only employee plus contract for the selected roles and reports the residual against the total field as a check.

Therapy and other non-nurse roles live in the non-nurse file, which carries contract fields for roles such as occupational therapist and speech/language pathologist, in addition to the nurse roles. MedPAC likewise describes PBJ as distinguishing employed from contract staff [14]. This matters for a rehabilitation-services thesis. A study of the 2019 payment change found that therapy staffing reductions were concentrated among contracted employees [15] and were larger in facilities with higher shares of Medicare-eligible short-stay residents [15], and another using the public files found that contract therapy staffing fell 8.4 percent in the week after the change, the largest decline of any subgroup [16]. Contract therapy hours behave as a flexible margin that responds to reimbursement rules, so a high share is not automatically a structural dependency.

The formulas

Contract share equals summed contract hours divided by summed (employee plus contract) hours, computed over exactly the same role set and date set. It is undefined when the denominator is zero. Hours per resident-day equals summed selected paid hours divided by summed MDS census on the same non-zero-census dates. Use a ratio of sums. For a fictional two-day illustration, day one has census 10 and 40 hours (4.0 per resident-day) and day two has census 90 and 270 hours (3.0); the ratio of sums is 310 divided by 100, or 3.10, while the unweighted mean of the daily ratios is 3.5, which overstates intensity because it gives the small day equal weight.

These are proposed worksheet calculations, not measured benchmark results. Census is itself an estimate: under the July 2023 PUF documentation, CMS derives it from MDS assessments, assuming a resident is no longer present after a discharge assessment or after 150 days with no assessment [6], and the manual notes that an over-estimate of census yields lower facility staffing per resident [7]. Because the same census sits in the denominator of every facility-day, it can move hours per resident-day without any change in contract share.

Implementation Considerations and Process Changes

Pin the release, then match the operator separately

Record the release quarter, the publication date and the dictionary edition. Then, as a separate step, match the facility to its legal operator for the relevant dates. Do not join on display name. CMS describes its skilled nursing facility (SNF) enrollment and owner files as provider self-reported, and states that the files do not contain historical enrollment information [17]. Owner association dates identify when each owning or managing entity became associated with a facility, and the change-of-ownership (CHOW) file carries an effective-date field defined as the date the change became effective [18].

Two caveats make operator history harder than it looks. In a CHOW the provider number typically remains intact and passes with the provider agreement [19], consistent with the federal rule that the existing provider agreement is automatically assigned to the new owner [20], so a stable PROVNUM can hide a change of operator in the middle of a quarter. A single operator may also hold many certification numbers, and CMS's own "chain" grouping is a derived grouping of homes, not a legal entity. Group aggregation therefore requires supported contemporaneous operator membership and the same period. The ownership files refresh monthly or quarterly, so the match should be re-run with each release.

Apply a coverage gate before any share is read

Before interpreting a share, classify coverage. The coverage and exclusion rules below come from the July 2023 PUF documentation and should be re-checked against the current documentation for each release. A facility is in a quarterly public file only if it was active on the last day of the submission period [6], and CMS's exclusions are applied to whole quarters, so a facility excluded for aberrant weekend staffing is dropped for the quarter regardless of staffing on other days [6]. The consequence is important: a missing facility is not zero contract use, and a reported zero is a source value that cannot be distinguished from non-reporting in the public file. The coverage statuses used in the worksheet are:

  • Complete: the facility appears in every quarter of the chosen window, with no zero-census days and a total-check residual of zero.
  • Partial: one or more quarters absent, zero-census days present, or a nonzero residual; shares are shown but flagged.
  • Excluded or missing: the facility is absent from a release; no share is computed and no zero is assumed.
  • Identity change: a change of ownership or an operator effective date falls inside the window; periods before and after are computed separately.
  • Outlier days: daily values flagged for review, since CMS tells users to be cautious in interpreting extreme values.

Test persistence, not a single day

Daily series are noisy. Bowblis and coauthors used a 14-day moving average because daily variation can be significant [2], and noted that 2017 data had more outliers because of the transition year [2]. Weekends differ too: a conference analysis of 2017 and 2018 PBJ data found RN staffing 38 percent lower on weekends [21]. A persistent-dependency test therefore compares matched calendars (same quarter, same weekday mix) and asks whether contract hours appear across most weeks, not whether one day is high. Averages can also mask instability: one study of PBJ variation cautioned that measuring only average staffing levels may mask dangerously low levels on some days [22]. Do not manufacture a pass threshold from an industry average.

Match hours to the dollars where the thesis needs it

PBJ is hours. The Medicare SNF cost report is the public dollar-side complement. CMS's public use file dictionary includes a Contract Labor variable mapped to Worksheet S-3, Part II, Line 14 [23]. Cost reports cover fiscal years beginning on or after December 1, 2010 [24], and the dataset is refreshed only about annually, so they lag PBJ. A mismatch between hours-based dependency and dollar-based contract labor is itself a diligence question, but neither source identifies a supplier.

The screening procedure in eight steps

The checks above combine into one repeatable procedure for each facility-period:

  • Step 1, define the thesis roles: choose nursing roles, non-nurse roles or both, and record whether administrative and director roles are included.
  • Step 2, pin the source: record release quarter, publication date, dictionary edition and file name; keep PROVNUM as a string.
  • Step 3, match the operator: build a dated operator match from ownership and change-of-ownership snapshots, and split any period where an effective date falls inside it.
  • Step 4, compute the ratios: contract share and hours per resident-day, each as a ratio of sums over the same role and date set.
  • Step 5, check the residual: total minus employee minus contract, and investigate any nonzero value before reading the share.
  • Step 6, test persistence: compare matched calendars and look for contract hours across most weeks, not one unusual day.
  • Step 7, gate the supplier: require independent dated evidence of a named relationship and its scope; otherwise record "supplier-unresolved".
  • Step 8, assign the disposition: observed dependency, uncertain due to coverage, supplier-supported for a stated period, or supplier-unresolved.

What the screen cannot determine is equally explicit:

  • Supplier identity: the public dictionary has no agency field.
  • Price and margin: no wage or billing-rate field exists in the schema inspected.
  • Contract durability: hours show use, not term, renewal or exclusivity.
  • Allocation across agencies: one facility may use several suppliers in the same role.
  • Delivered care: paid hours are not therapy or nursing care delivered.
  • Legal compliance: low or high hours are not a finding about any requirement.

The supplier-attribution gate

The PBJ public files cannot name a counterparty. The policy manual ties reported contract hours to auditable records such as payroll, invoices or a contract [7], but the public dictionary has no field linking hours to an agency [1]. To state a named vendor relationship for a facility, require independent, dated evidence of the relationship and its scope: a facility or operator document, a regulatory filing, a public contract record, or authorized diligence material. Even a supported name supplies neither price, exclusivity, margin, renewal terms nor the allocation of hours across multiple agencies. Public staffing companies describe billing as hours worked multiplied by a bill rate [25] [26], which is exactly the rate information the PBJ file lacks.

Figure 01
Screening a facility for persistent contract use
  1. 01Choose roles and pin the release

    Choose nursing roles, non-nurse roles or both. Specify whether administrative and director roles count. Record the release quarter, publication date, dictionary edition and file name. Keep PROVNUM as a string.

  2. 02Match the operator by date

    Use ownership and change-of-ownership snapshots to match the operator. Split the period if an effective date falls within it.

  3. 03Calculate ratios and check totals

    Compute contract share and hours per resident-day as ratios of sums. Subtract employee and contract hours from total hours and investigate any nonzero residual before interpreting the share.

  4. 04Test persistence

    Compare periods with the same quarter and weekday mix. Check whether contract hours appear across most weeks.

  5. 05Check supplier evidence

    Require independent, dated evidence of a named relationship and its scope. If evidence is missing, record supplier-unresolved.

  6. 06Record the disposition

    Record observed dependency, uncertain due to coverage, supplier-supported for a stated period, or supplier-unresolved.

Data Analysis and Evidence

Table 1 below lists the proposed facility-period worksheet. It is a blank template, and every column has a stated source or a stated gap.

ColumnWhat goes in itSource or gap
Facility key (PROVNUM)Six-character string, never numericPBJ dictionary; keep leading zeros
Operator match and effective dateLegal operator and CHOW or association date for the periodCMS ownership files; no history, so build by snapshot
Release and quarterQuarter, publication date, dictionary editionPinned per row
Role setExact roles chosen (for example CNA only, or RN plus LPN)Analyst choice, recorded
Included datesDates used, after removing zero-census daysMDS census
Employee hoursSum of employee fields for the role setPBJ file
Contract hoursSum of contract fields for the role setPBJ file
Total-check residualTotal minus employee minus contractShould be zero
Resident-daysSum of census on included datesMDS-derived estimate
Excluded or zero-census daysCount and datesPBJ file
Coverage statusComplete, partial, excluded, identity change, outlierWorksheet rule
Corroborating supplier sourceIndependent document naming a vendorGap by default
Supplier scope and dateWhat the document covers and whenGap by default
Review dispositionOne of four outcomesAnalyst, with specialist review where needed

The worksheet separates what the public file shows (hours and census) from what only outside evidence can show (supplier, scope, date). The default for the last three columns is an empty gap, which is the honest state for most facilities.

Table 2 summarizes the historical reference points available from primary and peer-reviewed sources. They calibrate magnitudes; they are not thresholds, and they do not apply to any facility today.

Reference pointFigure and periodDenominator and caution
Agency share of direct care nursing hoursAbout 3% in 2018 and 11% in 2022 [2]Direct care nursing hours, PBJ facility-day; historical
Homes using any agency nursing staff49.1% on average in 2022Prevalence, not intensity
Contract share of CNA hours2% in 2017 to 11% in 2022 [3]CNA role only; among users it reached 18% [3]
Contract therapy response to payment changeContract therapy staffing down 8.4% in the week after the 2019 changeTherapy only; policy-elastic, and paid minutes do not equal therapy delivered
Share of agency users with short episodes19.4% used agency staff for 90 days or fewer in 2022Persistence indicator
US healthcare staffing marketForecast $39.0 billion in 2026 [5]Vendor-side revenue forecast; not nursing-home specific
Agency use trend in nursing facilitiesFell about 44% from Q4 2022 levels [27]Industry assessment; the metric is not defined on the page
Travel nurse firm gross marginMedian 19.9% in 2025 [5]Travel segment; not nursing-home agency margin

Two readings follow. First, the national figures show that 11 percent of direct care hours or 11 percent of CNA hours is the latest year in these studies (2022), not a norm. Separately, an industry assessment reports that agency use fell about 44 percent from Q4 2022 levels, but that page does not define the metric, so it should not be read as a change in PBJ contract-hour share [27]. A facility screening at 11 percent today therefore cannot be called "typical" or "high" without a matched, current comparison group built from the same release. Second, the margin and market-size figures are supplier-side industry data. They show that staffing revenue pools are large and margins are moderate, but they cannot be converted to a facility-level revenue estimate: the market includes hospitals and other settings, and the largest firms, 91 companies with $33 billion of 2025 revenue and a five-firm share of 40.7 percent [28], have their own customer mix.

For scale, KFF counts 14,742 CMS-certified nursing facilities and about 1.24 million residents as of July 2025 [29]. KFF also cautions that CMS's private equity ownership flags may be incomplete because facilities began self-reporting only recently [29]. The owner flag is therefore a lead, not a classifier. The Bureau of Labor Statistics reports May 2025 median annual wages of $42,260 for nursing assistants [30], $64,400 for licensed practical nurses [31] and $97,550 for registered nurses [32]; these anchor labor cost per hour only loosely, since agency bill rates are not published in PBJ.

On the supply side, one public company's annual report describes revenue that is hours worked times a bill rate [25], a model with the risk that costs can change faster than bill rates under active contracts [25], typically on travel assignments of about 13 weeks [25]. The Bureau of Labor Statistics also publishes occupational employment and wage estimates for nursing care facilities (NAICS 623100) [33], which helps size the labor pool but not the supplier mix. Another public company defines billable hours and bill rates as its volume and price measures [26]. A third disclosure shows that large staffing companies themselves report customer concentration, with one company stating that no customer exceeded 10 percent of revenue [25] and another reporting a single client at about 22 percent [26]. The American Staffing Association puts the all-sector total at nearly 2.0 million temporary and contract employees working in an average week of 2025 [34]. These are descriptive disclosures about how staffing revenue is recorded. None says anything about nursing-home facility-level contracts.

Figure 02
2022 state range for agency share of direct care nursing hourspercent of direct care nursing hours
Source: Bowblis and coauthors

Case Studies and Real-World Examples

Fictional Worked Contrast (Hypothetical Example)

All facilities, operators and figures in this section are fictional and invented for illustration. They are not drawn from any real facility, operator or supplier. The quarter is April 1 to June 30, 2026 (91 days).

Table 3 shows three hypothetical facilities. Alpha and Beta have an identical contract share but different roles and census. Gamma changes operator on May 16, 2026. Delta is a reconciliation failure.

Hypothetical facilityRole setEmployee hoursContract hoursTotal fieldResidualContract shareResident-daysHours per resident-day
Alpha (census 80 daily)CNA only18,7003,30022,000015.0%7,2803.02
Beta (census 40 daily)RN direct only1,3602401,600015.0%3,6400.44
Gamma, before May 16 (45 days)CNA only7,2003007,50004.0%4,5001.67
Gamma, from May 16 (46 days)CNA only5,9801,6107,590021.2%4,6001.65
Gamma, blended quarterCNA only13,1801,91015,090012.7%9,1001.66
DeltaLPN only4,2008005,10010016.0% (15.7% on total)not shownnot shown

Alpha and Beta both show 15.0 percent, yet they raise different questions. Alpha's 3,300 contract CNA hours equal about 412 eight-hour shifts over the quarter, a sustained daily presence that merits a question about how the operator staffs the floor. Beta's 240 contract RN hours equal 30 eight-hour shifts; the hours-per-resident-day figure of 0.44 shows a small RN base where a handful of shifts moves the share. A screen that reports only "15%" would treat them as equivalent.

Gamma shows why operator matching comes first. The blended quarter reads 12.7 percent. Split at the May 16 effective date, the first operator period reads 4.0 percent and the second 21.2 percent. Attributing the blended figure to either operator would be wrong for both. Group aggregation across a portfolio would require the same split for every facility. Whether the change reflects a real operating shift or an ingestion and reporting change cannot be settled from the public file.

Delta shows the reconciliation check. Total less employee less contract is 100 hours, so something is missing: a role outside the chosen set, a rounding or aggregation issue, or an omitted subfield. The share is 16.0 percent on employee plus contract (800 of 5,000) and 15.7 percent on the total field (800 of 5,100). The worksheet uses employee plus contract as the denominator, flags the residual and holds Delta at "uncertain due to coverage" until it is resolved.

A matched-period sensitivity run on Alpha shows how fragile a headline share can be. Suppose a single anomalous day carried 900 contract hours and 100 employee hours. Removing that one day cuts the quarter to 2,400 contract hours of 21,000 total, an 11.4 percent share, versus 15.0 percent with the day included: one day moves the share by 3.6 points. If the same quarter a year earlier showed 2,200 contract hours of 22,000 (10.0 percent, fictional), the matched comparison reads 15.0 versus 10.0 including the anomaly, and 11.4 versus 10.0 without it. The disposition depends on whether the anomaly is accepted as real, which is an analyst decision to document, not an automatic output.

Figure 03
Same contract share, different staffing questions
Alpha (CNA only)15.0% contract share
  • 3,300 contract CNA hours, equivalent to about 412 eight-hour shifts over the quarter.
  • 3.02 hours per resident-day, with a daily census of 80.
Beta (direct RN only)15.0% contract share
  • 240 contract RN hours, equivalent to 30 eight-hour shifts.
  • 0.44 hours per resident-day, with a daily census of 40.

In these fictional examples, Alpha's sustained CNA use raises a question about floor staffing. A handful of shifts can move Beta's share.

Implications and Future Directions

For origination, the practical output of the screen is a ranked list of diligence questions, not a ranking of targets. A facility or operator with persistent, matched-calendar contract dependence in the same roles across several quarters warrants questions about turnover, wage competitiveness, recruiting channels and how the operator procures agency hours. Prior work cautions that the relationship runs both ways: a study of high agency users cautioned that it does not address reverse causality, since turnover and agency use may feed each other [35]. The screen should therefore not be read as a cause or an effect.

For a healthcare-staffing or rehabilitation-services thesis, the demand-side signal is useful but incomplete. An industry forecaster notes that long-term care demand is shifting toward higher-paid RNs as facilities care for higher-acuity residents [5], and that per diem nursing revenue is forecast to decline 7 percent in 2026 [5]. Much staffing revenue also flows through managed-service or vendor-management programs, at a median 86.7 percent for travel nurse revenue [27], which means a facility may buy through an intermediary and the visible supplier may not be the staffing employer. Such structure is another reason hours cannot be translated to an agency's addressable revenue.

Advisory commentary on post-acute transactions makes the same practical point from the buy side: it argues that a post-acute acquisition is largely a labor obligation [36] and that agency spending that cannot be measured by shift cannot be reduced [36]. It is first-party commentary from an advisory firm and is cited as perspective, not evidence. A PBJ screen supplies hours by day, which is the precondition for shift-level analysis, but the pricing, rates and supplier contracts will come only from authorized diligence.

Several developments could change the screen. A future dictionary revision could add fields. CMS could change exclusion rules. The Medicare data used for operator matching is refreshed monthly or quarterly, so the match must be re-run with each release. Analysts should also expect continuing research on measurement: the literature already cautions that paid hours are not the same as delivered care [16] and that PBJ may not reflect all hours worked [14]. Any consequential legal, accounting or regulatory interpretation requires a qualified specialist.

Frequently Asked Questions (FAQs)

What are PBJ contract staff hours?

They are paid hours that facilities report for staff who are not direct employees, including individuals under their own contract and those provided by an agency. The public files report them by role and day, next to employee hours.

Can PBJ staffing data identify a nursing home's staffing agency?

No. The opened dictionary has no supplier, agency or rate field. Identifying a supplier needs independent dated evidence, and even then price and exclusivity remain unknown.

How should an analyst measure nursing home contract labor percentage?

Divide summed contract hours by summed employee plus contract hours for the same roles and dates, and compute hours per resident-day as a ratio of sums. CMS's own Five-Star guide combines employee and contract hours in its staffing measures [13], and the guide does not describe a separate contract-share measure; the worksheet adds one.

Does nursing home agency staffing dependency indicate a staffing agency's revenue?

No. The hours cannot be multiplied by an average bill rate to estimate a supplier's revenue. Public staffing companies bill hours times a rate, but the rate, margin and contract terms for any specific facility are not in the public file. Industry margins describe a segment only.

How do PBJ and the Medicare cost report differ for acquisition screening?

PBJ gives hours by day and role; the cost report gives annual contract labor dollars on Worksheet S-3 [23]. Use PBJ for persistence and the cost report for dollars, and expect the cost report to lag.

Is the employee-level research method usable with the public file?

No. One analysis built unique-worker counts from daily employee-level records [37], a different grain from the public facility-day file, which aggregates away worker identity. Do not replicate unique-employee calculations from summed hours, and do not equate full-time equivalents with unique employees. The authors themselves caution that their results should not be interpreted as causal [37].

Conclusion

The PBJ public files support a disciplined, limited conclusion: for a defined set of roles and dates, a facility reported a given share of paid hours as contract, and the pattern did or did not persist across matched periods. They do not show who supplied the hours, at what price, under what contract or with what margin. The pinned release, the separate operator match, the coverage gate and the supplier-attribution gate keep those layers apart.

The four dispositions are the output. Observed dependency means reported contract hours are material and persistent in the chosen roles, with complete coverage. Uncertain due to coverage means missing quarters, zero-census days, residuals, identity changes or outliers prevent a reading. Supplier-supported for a stated period means independent dated evidence names a vendor and a scope for that period, nothing more. Supplier-unresolved is the default for most facilities. Use the worksheet to decide whether a diligence question is warranted, keep the source lag in view, and send legal, accounting and regulatory interpretations to specialists. No facility benchmark or financial outcome was measured in this report.

External Sources (37)

Published by

Docket Research

Docket provides deal-origination research software and managed research for private equity teams, screening companies against a mandate and showing the evidence behind each finding.

Our platform · Evidence standards · Contact

Get your first 10 targets free

Editorial note

Researched and written by Docket Research with AI-assisted research tools; sources are cited inline and checked before publication. Report an error: https://docket.capital/evidence-standards#corrections.