Methodology

How the numbers are built.

Financial figures are compiled from public HCRIS Medicare cost reports (Worksheets G and G-3) and standardized by Astrelis into a consistent multi-year layer. Operating margin is computed on a cost-report basis. Days cash on hand for hospitals uses the corrected series: cash, temporary investments, and board-designated and funded reserves ÷ daily operating expense excluding depreciation. The older cash-plus-short-term-only convention understated liquidity for roughly 40% of hospitals. Facility-level days cash can still understate hospitals whose cash is swept to a parent system; it is context, never a standalone risk signal.

Peer comparison is calculated within comparable hospital groups only: critical access hospitals are compared against other critical access hospitals, never against urban acute-care systems.

Every statement line carries a tie-out check against the source filing. Lines are shown only where that reconciliation passes; anything that does not tie is labeled “pending tie-out” rather than presented as a number. Missing values are shown as missing, never as zero, never imputed.

County demographic and economic context is drawn from public federal sources: Census ACS 5-year estimates (2019–2023), CDC PLACES (2024 release), BEA regional accounts (2022), HRSA HPSA designations, and USDA ERS Rural-Urban Continuum Codes (2023). County tiles carry two comparators: the US figure is a population-weighted national mean across all counties; the rural figure is the median across nonmetro (RUCC 4–9) counties. Detailed source-line mapping and QA are maintained privately for audit. Every figure on a profile page traces to a public filing.

Method, by product family

generated from the build · metric definitions AD-1096.1 · 2026-08-02

Ratified definitions. AD-1218.1 · ratified 2026-07-31

Full year and period length. A full year is a reporting period of 355 to 380 days inclusive. The exact number of days in each filing is stored and shown. Periods longer than 380 days are labeled long-period and are excluded from comparisons; periods shorter than 355 days are labeled stub. Nothing is annualized — a short filing is never scaled up to look like a year.
Per-day measures. Per-day measures divide by the inclusive length of the reporting period (end date minus begin date plus one day), not by a fixed 365.
Multi-component measures. A measure built from several reported figures publishes only when every one of those figures was actually reported. A filed zero counts as reported; a line the provider did not file does not become a zero. When any component is missing the measure is shown as not reported, and the missing components are named.
Long-term debt. Debt ratios use interest-bearing long-term debt: Worksheet G line 46 (mortgage, bonds and notes payable) plus line 47 (capitalized lease obligations). Non-interest-bearing long-term liabilities are excluded. Both lines are summed across all four fund columns as filed.
Occupancy basis. Occupancy is patient days divided by bed-days available on BEDS IN OPERATION as filed — not licensed beds, not certified beds. A facility that staffs surge capacity beyond its bed count can genuinely exceed 100%; that is reported, not corrected.
Case mix and wage index. Case mix index and wage index are pinned to a single published vintage — CMI to the FY2026 IPPS Final Rule impact file, geographic wage index to the FY2025 hospice wage index file (geographic) — and applied as a disclosed cross-sectional convention across every reporting year rather than re-derived per year. Critical access hospitals and other non-IPPS providers have no published CMI; their values are shown UNADJUSTED and carry the label "no published CMI for this provider type — value shown unadjusted". Nulling the value out for non-IPPS providers was considered and rejected: an unadjusted number that says it is unadjusted is more useful than an absence.
Peer pools. Every peer pool and benchmark is fiscal-year pure: a facility is only ever compared with other facilities' filings for the SAME fiscal year. Every peer key and benchmark key carries its fiscal year. A key without a fiscal year is a defect.

Peer matching and pool floors. Peer pools are built per provider type (hospitals additionally per subtype cohort: CAH, rural PPS, urban PPS, psychiatric, children's, LTACH, REH), per fiscal year. A fiscal-year pool is valid at n >= 25 AND >= 50% of in-distribution members filed; every rendered percentile compares a metric against the pool at that metric's own fiscal year (same-year or absent, AD-1165). The peers section requires >= 8 same-year filers; a dot distribution requires >= 25, with 8–24 rendering median prose only.

Filing selection (preliminary vs settled). One row per facility per fiscal year: full-year filings (>= 335 days) beat stubs; then settled beats amended beats as-filed (status rank 3>2>5>4>1); then latest process date; then highest report number. Preliminary/as-filed rows display with their status labeled and are superseded automatically when a settled filing lands.

Exclusions from peer statistics (the value always stays visible on its own page): CMS-POS-terminated (historical) records — labeled, out of pool membership, counts, and medians; each still RANKS against its own contemporaneous-year pool; non-panel (legacy-basis) records where a clean-room panel exists for the module — basis mismatch; filing-validity failures only: non-computable ratios (near-zero denominators), broken balance identities, and structurally impossible values — each renders as Astrelis calculation unavailable with its filed inputs shown.

The canonical report layer. Paid reports generate from the canonical clean-room panel layer: 06_Build_Outputs panels -> sha-manifested staging (data/source/cleanroom_v1) -> data/generated — the same layer that renders every page on this site. One golden report per coverage class is generated from data/generated and archived beside each release manifest.

Hospital financials (all subtypes)

CMS HCRIS 2552-10 cost reports, Worksheets G / G-3 / S-3, as filed

(operating income + other operating income) ÷ (net patient revenue + other operating income) — the AD-1096 Total Operating Revenue basis; NPR-only fallback where other operating income is not reported (never coerced to zero)

net income ÷ (net patient revenue + total other income)

(cash + investments) ÷ daily cash operating expense, where daily cash operating expense = (operating expense − depreciation) ÷ fiscal-period days — equivalently (cash + investments) × fiscal-period days ÷ annual cash operating expense; total-cash basis primary, liquid-only companion

CAH contribution margins & department economics

CMS HCRIS 2552-10, B/C/D-series worksheets

Medicare inpatient / outpatient contribution margin: department revenue less direct and stepped-down cost, per the CAH economics engine (same-fiscal-year national CAH pools; provider-labeled cost centers pinned by modal label, never hardcoded line numbers)

Performance benchmark gaps (SNF, home health, hospice)

HCRIS per-sector cost reports; peer cohorts matched on size/payer (and nursing-CMI acuity where the source supports it — SNF); conservative low band only

cost metrics: (facility rate − peer benchmark rate) × facility volume · revenue metrics: (peer benchmark rate − facility rate) × facility volume — an arithmetic benchmark difference against the same-year peer pool, conservative LOW band, acuity limits stated; missing acuity control demotes the claim to directional

SNF, home health, hospice, and dialysis measures

CMS-2540 / CMS-1728 / CMS-1984 / CMS-265 cost reports + CMS Care Compare quality files

net patient revenue ÷ treatments and total cost ÷ treatments (CMS-265); the teaser spread = (revenue per treatment − cost per treatment) × treatments — the facility's own filed revenue against its own filed cost, not a peer comparison

Rural health clinic economics

CMS-222 cost reports

total allowable cost ÷ total clinic visits (CMS-222 Worksheet B) — displayed beside both inputs; the filing's own Worksheet C adjusted rate renders separately as a reported value

Health center measures

HRSA UDS + CMS-224

Organization-level patients, visits, sites, payer mix, and clinical quality: HRSA UDS awardee data via the documented BHCMIS-CCN crosswalk, match status stated per profile (Verified / Probable / Not established); Medicare cost-report figures (CMS-224) render in a bounded appendix with no percentile and no peer framing

Commercial pricing files

Hospital machine-readable price files + CMS OPPS/IPPS reference rates, frozen and checksummed

Commercial negotiated rates as a percent of Medicare OPPS-equivalent, line-level volume-weighted outpatient basket (separately-payable non-drug services); drugs and comprehensive APCs are separate lanes and never blended in

Economic multipliers

NCRHW 2016 (IMPLAN Type II) + AHA national output ratio; rendered only inside the isolated illustrative-advocacy-estimate box

employment (direct jobs × 1.34) and labor income (direct compensation × 1.19) from the National Center for Rural Health Works 2016 study of Critical Access Hospital economic impact (IMPLAN Type II, U.S. rural county populations); economic activity (direct spending × 2.30) from the AHA national hospital economic activity report — a national hospital ratio, not a CAH or county figure. Applying any of these to an individual facility is an illustrative advocacy scenario, not a measurement: local capture depends on payroll residency, purchasing patterns, and county economic structure the sources do not observe.

Data vintages this build: 48 synced clean-room files · export v2.3.0 · source fingerprint 49af138d3587ad73…